diff options
Diffstat (limited to 'drivers/gpu/drm/amd')
505 files changed, 32772 insertions, 4197 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 1acfed2f92ef..7f515be5185d 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -43,14 +43,16 @@ config DRM_AMDGPU_SI bool "Enable amdgpu support for SI parts" depends on DRM_AMDGPU help - Choose this option if you want to enable experimental support + Choose this option if you want to enable support for SI (Southern Islands) asics. - SI is already supported in radeon. Experimental support for SI - in amdgpu will be disabled by default and is still provided by - radeon. Use module options to override this: + SI (Southern Islands) are first generation GCN GPUs, + supported by both drivers: radeon (old) and amdgpu (new). + By default, SI dedicated GPUs are supported by amdgpu. - radeon.si_support=0 amdgpu.si_support=1 + Use module options to override this: + To use radeon for SI, + radeon.si_support=1 amdgpu.si_support=0 config DRM_AMDGPU_CIK bool "Enable amdgpu support for CIK parts" @@ -59,11 +61,17 @@ config DRM_AMDGPU_CIK Choose this option if you want to enable support for CIK (Sea Islands) asics. - CIK is already supported in radeon. Support for CIK in amdgpu - will be disabled by default and is still provided by radeon. - Use module options to override this: + CIK (Sea Islands) are second generation GCN GPUs, + supported by both drivers: radeon (old) and amdgpu (new). + By default, + CIK dedicated GPUs are supported by amdgpu + CIK APUs are supported by radeon + Use module options to override this: + To use amdgpu for CIK, radeon.cik_support=0 amdgpu.cik_support=1 + To use radeon for CIK, + radeon.cik_support=1 amdgpu.cik_support=0 config DRM_AMDGPU_USERPTR bool "Always enable userptr write support" diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile index 64e7acff8f18..c88760fb52ea 100644 --- a/drivers/gpu/drm/amd/amdgpu/Makefile +++ b/drivers/gpu/drm/amd/amdgpu/Makefile @@ -37,7 +37,8 @@ ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \ -I$(FULL_AMD_DISPLAY_PATH)/modules/inc \ -I$(FULL_AMD_DISPLAY_PATH)/dc \ -I$(FULL_AMD_DISPLAY_PATH)/amdgpu_dm \ - -I$(FULL_AMD_PATH)/amdkfd + -I$(FULL_AMD_PATH)/amdkfd \ + -I$(FULL_AMD_PATH)/ras/ras_mgr # Locally disable W=1 warnings enabled in drm subsystem Makefile subdir-ccflags-y += -Wno-override-init @@ -77,7 +78,7 @@ amdgpu-$(CONFIG_DRM_AMDGPU_CIK)+= cik.o cik_ih.o \ dce_v8_0.o gfx_v7_0.o cik_sdma.o uvd_v4_2.o vce_v2_0.o amdgpu-$(CONFIG_DRM_AMDGPU_SI)+= si.o gmc_v6_0.o gfx_v6_0.o si_ih.o si_dma.o dce_v6_0.o \ - uvd_v3_1.o + uvd_v3_1.o vce_v1_0.o amdgpu-y += \ vi.o mxgpu_vi.o nbio_v6_1.o soc15.o emu_soc.o mxgpu_ai.o nbio_v7_0.o vega10_reg_init.o \ @@ -324,4 +325,9 @@ amdgpu-y += \ isp_v4_1_1.o endif +AMD_GPU_RAS_PATH := ../ras +AMD_GPU_RAS_FULL_PATH := $(FULL_AMD_PATH)/ras +include $(AMD_GPU_RAS_FULL_PATH)/Makefile +amdgpu-y += $(AMD_GPU_RAS_FILES) + obj-$(CONFIG_DRM_AMDGPU)+= amdgpu.o diff --git a/drivers/gpu/drm/amd/amdgpu/aldebaran.c b/drivers/gpu/drm/amd/amdgpu/aldebaran.c index 9569dc16dd3d..daa7b23bc775 100644 --- a/drivers/gpu/drm/amd/amdgpu/aldebaran.c +++ b/drivers/gpu/drm/amd/amdgpu/aldebaran.c @@ -88,6 +88,10 @@ static int aldebaran_mode2_suspend_ip(struct amdgpu_device *adev) uint32_t ip_block; int r, i; + /* Skip suspend of SDMA IP versions >= 4.4.2. They are multi-aid */ + if (adev->aid_mask) + ip_block_mask &= ~BIT(AMD_IP_BLOCK_TYPE_SDMA); + amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 2a0df4cabb99..9f9774f58ce1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -372,13 +372,15 @@ void amdgpu_device_ip_get_clockgating_state(struct amdgpu_device *adev, u64 *flags); int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev, enum amd_ip_block_type block_type); +bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev, + enum amd_ip_block_type block_type); bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev, enum amd_ip_block_type block_type); int amdgpu_ip_block_suspend(struct amdgpu_ip_block *ip_block); int amdgpu_ip_block_resume(struct amdgpu_ip_block *ip_block); -#define AMDGPU_MAX_IP_NUM 16 +#define AMDGPU_MAX_IP_NUM AMD_IP_BLOCK_TYPE_NUM struct amdgpu_ip_block_status { bool valid; @@ -839,8 +841,6 @@ struct amd_powerplay { const struct amd_pm_funcs *pp_funcs; }; -struct ip_discovery_top; - /* polaris10 kickers */ #define ASICID_IS_P20(did, rid) (((did == 0x67DF) && \ ((rid == 0xE3) || \ @@ -972,8 +972,7 @@ struct amdgpu_device { struct notifier_block acpi_nb; struct notifier_block pm_nb; struct amdgpu_i2c_chan *i2c_bus[AMDGPU_MAX_I2C_BUS]; - struct debugfs_blob_wrapper debugfs_vbios_blob; - struct debugfs_blob_wrapper debugfs_discovery_blob; + struct debugfs_blob_wrapper debugfs_vbios_blob; struct mutex srbm_mutex; /* GRBM index mutex. Protects concurrent access to GRBM index */ struct mutex grbm_idx_mutex; @@ -1063,6 +1062,9 @@ struct amdgpu_device { u32 log2_max_MBps; } mm_stats; + /* discovery*/ + struct amdgpu_discovery_info discovery; + /* display */ bool enable_virtual_display; struct amdgpu_vkms_output *amdgpu_vkms_output; @@ -1174,6 +1176,12 @@ struct amdgpu_device { * queue fence. */ struct xarray userq_xa; + /** + * @userq_doorbell_xa: Global user queue map (doorbell index → queue) + * Key: doorbell_index (unique global identifier for the queue) + * Value: struct amdgpu_usermode_queue + */ + struct xarray userq_doorbell_xa; /* df */ struct amdgpu_df df; @@ -1265,8 +1273,6 @@ struct amdgpu_device { struct list_head ras_list; - struct ip_discovery_top *ip_top; - struct amdgpu_reset_domain *reset_domain; struct mutex benchmark_mutex; @@ -1290,6 +1296,7 @@ struct amdgpu_device { bool debug_disable_gpu_ring_reset; bool debug_vm_userptr; bool debug_disable_ce_logs; + bool debug_enable_ce_cs; /* Protection for the following isolation structure */ struct mutex enforce_isolation_mutex; @@ -1308,9 +1315,8 @@ struct amdgpu_device { */ bool apu_prefer_gtt; - struct list_head userq_mgr_list; - struct mutex userq_mutex; bool userq_halt_for_enforce_isolation; + struct work_struct userq_reset_work; struct amdgpu_uid *uid_info; /* KFD @@ -1534,11 +1540,6 @@ int emu_soc_asic_init(struct amdgpu_device *adev); #define amdgpu_asic_read_bios_from_rom(adev, b, l) (adev)->asic_funcs->read_bios_from_rom((adev), (b), (l)) #define amdgpu_asic_read_register(adev, se, sh, offset, v)((adev)->asic_funcs->read_register((adev), (se), (sh), (offset), (v))) #define amdgpu_asic_get_config_memsize(adev) (adev)->asic_funcs->get_config_memsize((adev)) -#define amdgpu_asic_flush_hdp(adev, r) \ - ((adev)->asic_funcs->flush_hdp ? (adev)->asic_funcs->flush_hdp((adev), (r)) : (adev)->hdp.funcs->flush_hdp((adev), (r))) -#define amdgpu_asic_invalidate_hdp(adev, r) \ - ((adev)->asic_funcs->invalidate_hdp ? (adev)->asic_funcs->invalidate_hdp((adev), (r)) : \ - ((adev)->hdp.funcs->invalidate_hdp ? (adev)->hdp.funcs->invalidate_hdp((adev), (r)) : (void)0)) #define amdgpu_asic_need_full_reset(adev) (adev)->asic_funcs->need_full_reset((adev)) #define amdgpu_asic_init_doorbell_index(adev) (adev)->asic_funcs->init_doorbell_index((adev)) #define amdgpu_asic_get_pcie_usage(adev, cnt0, cnt1) ((adev)->asic_funcs->get_pcie_usage((adev), (cnt0), (cnt1))) @@ -1637,7 +1638,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, struct drm_file *file_priv); void amdgpu_driver_release_kms(struct drm_device *dev); -int amdgpu_device_ip_suspend(struct amdgpu_device *adev); int amdgpu_device_prepare(struct drm_device *dev); void amdgpu_device_complete(struct drm_device *dev); int amdgpu_device_suspend(struct drm_device *dev, bool fbcon); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c index 4926996f94da..381ef205b0df 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c @@ -302,17 +302,19 @@ static int acp_hw_init(struct amdgpu_ip_block *ip_block) adev->acp.acp_res[2].end = adev->acp.acp_res[2].start; adev->acp.acp_cell[0].name = "acp_audio_dma"; + adev->acp.acp_cell[0].id = 0; adev->acp.acp_cell[0].num_resources = 3; adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0]; adev->acp.acp_cell[0].platform_data = &adev->asic_type; adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type); adev->acp.acp_cell[1].name = "designware-i2s"; + adev->acp.acp_cell[1].id = 1; adev->acp.acp_cell[1].num_resources = 1; adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1]; adev->acp.acp_cell[1].platform_data = &i2s_pdata[0]; adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data); - r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, 2); + r = mfd_add_devices(adev->acp.parent, 0, adev->acp.acp_cell, 2, NULL, 0, NULL); if (r) goto failure; r = device_for_each_child(adev->acp.parent, &adev->acp.acp_genpd->gpd, @@ -410,30 +412,34 @@ static int acp_hw_init(struct amdgpu_ip_block *ip_block) adev->acp.acp_res[4].end = adev->acp.acp_res[4].start; adev->acp.acp_cell[0].name = "acp_audio_dma"; + adev->acp.acp_cell[0].id = 0; adev->acp.acp_cell[0].num_resources = 5; adev->acp.acp_cell[0].resources = &adev->acp.acp_res[0]; adev->acp.acp_cell[0].platform_data = &adev->asic_type; adev->acp.acp_cell[0].pdata_size = sizeof(adev->asic_type); adev->acp.acp_cell[1].name = "designware-i2s"; + adev->acp.acp_cell[1].id = 1; adev->acp.acp_cell[1].num_resources = 1; adev->acp.acp_cell[1].resources = &adev->acp.acp_res[1]; adev->acp.acp_cell[1].platform_data = &i2s_pdata[0]; adev->acp.acp_cell[1].pdata_size = sizeof(struct i2s_platform_data); adev->acp.acp_cell[2].name = "designware-i2s"; + adev->acp.acp_cell[2].id = 2; adev->acp.acp_cell[2].num_resources = 1; adev->acp.acp_cell[2].resources = &adev->acp.acp_res[2]; adev->acp.acp_cell[2].platform_data = &i2s_pdata[1]; adev->acp.acp_cell[2].pdata_size = sizeof(struct i2s_platform_data); adev->acp.acp_cell[3].name = "designware-i2s"; + adev->acp.acp_cell[3].id = 3; adev->acp.acp_cell[3].num_resources = 1; adev->acp.acp_cell[3].resources = &adev->acp.acp_res[3]; adev->acp.acp_cell[3].platform_data = &i2s_pdata[2]; adev->acp.acp_cell[3].pdata_size = sizeof(struct i2s_platform_data); - r = mfd_add_hotplug_devices(adev->acp.parent, adev->acp.acp_cell, ACP_DEVS); + r = mfd_add_devices(adev->acp.parent, 0, adev->acp.acp_cell, ACP_DEVS, NULL, 0, NULL); if (r) goto failure; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 6c62e27b9800..d31460a9e958 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -507,7 +507,6 @@ static int amdgpu_atif_handler(struct amdgpu_device *adev, pm_runtime_get_sync(adev_to_drm(adev)->dev); /* Just fire off a uevent and let userspace tell us what to do */ drm_helper_hpd_irq_event(adev_to_drm(adev)); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 9e120c934cc1..8bdfcde2029b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -71,7 +71,7 @@ struct kgd_mem { struct mutex lock; struct amdgpu_bo *bo; struct dma_buf *dmabuf; - struct hmm_range *range; + struct amdgpu_hmm_range *range; struct list_head attachments; /* protected by amdkfd_process_info.lock */ struct list_head validate_list; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 83020963dfde..b1c24c8fa686 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1057,7 +1057,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, struct amdkfd_process_info *process_info = mem->process_info; struct amdgpu_bo *bo = mem->bo; struct ttm_operation_ctx ctx = { true, false }; - struct hmm_range *range; + struct amdgpu_hmm_range *range; int ret = 0; mutex_lock(&process_info->lock); @@ -1089,8 +1089,15 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, return 0; } - ret = amdgpu_ttm_tt_get_user_pages(bo, &range); + range = amdgpu_hmm_range_alloc(NULL); + if (unlikely(!range)) { + ret = -ENOMEM; + goto unregister_out; + } + + ret = amdgpu_ttm_tt_get_user_pages(bo, range); if (ret) { + amdgpu_hmm_range_free(range); if (ret == -EAGAIN) pr_debug("Failed to get user pages, try again\n"); else @@ -1113,7 +1120,7 @@ static int init_user_pages(struct kgd_mem *mem, uint64_t user_addr, amdgpu_bo_unreserve(bo); release_out: - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range); + amdgpu_hmm_range_free(range); unregister_out: if (ret) amdgpu_hmm_unregister(bo); @@ -1267,6 +1274,10 @@ static int unmap_bo_from_gpuvm(struct kgd_mem *mem, (void)amdgpu_vm_bo_unmap(adev, bo_va, entry->va); + /* VM entity stopped if process killed, don't clear freed pt bo */ + if (!amdgpu_vm_ready(vm)) + return 0; + (void)amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update); (void)amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL); @@ -1916,7 +1927,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( if (amdgpu_ttm_tt_get_usermm(mem->bo->tbo.ttm)) { amdgpu_hmm_unregister(mem->bo); mutex_lock(&process_info->notifier_lock); - amdgpu_ttm_tt_discard_user_pages(mem->bo->tbo.ttm, mem->range); + amdgpu_hmm_range_free(mem->range); mutex_unlock(&process_info->notifier_lock); } @@ -1954,9 +1965,7 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu( */ if (size) { if (!is_imported && - (mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_VRAM || - (adev->apu_prefer_gtt && - mem->bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT))) + mem->alloc_flags & KFD_IOC_ALLOC_MEM_FLAGS_VRAM) *size = bo_size; else *size = 0; @@ -2329,10 +2338,9 @@ void amdgpu_amdkfd_gpuvm_unmap_gtt_bo_from_kernel(struct kgd_mem *mem) int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct amdgpu_device *adev, struct kfd_vm_fault_info *mem) { - if (atomic_read(&adev->gmc.vm_fault_info_updated) == 1) { + if (atomic_read_acquire(&adev->gmc.vm_fault_info_updated) == 1) { *mem = *adev->gmc.vm_fault_info; - mb(); /* make sure read happened */ - atomic_set(&adev->gmc.vm_fault_info_updated, 0); + atomic_set_release(&adev->gmc.vm_fault_info_updated, 0); } return 0; } @@ -2543,7 +2551,7 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info, bo = mem->bo; - amdgpu_ttm_tt_discard_user_pages(bo->tbo.ttm, mem->range); + amdgpu_hmm_range_free(mem->range); mem->range = NULL; /* BO reservations and getting user pages (hmm_range_fault) @@ -2567,9 +2575,14 @@ static int update_invalid_user_pages(struct amdkfd_process_info *process_info, } } + mem->range = amdgpu_hmm_range_alloc(NULL); + if (unlikely(!mem->range)) + return -ENOMEM; /* Get updated user pages */ - ret = amdgpu_ttm_tt_get_user_pages(bo, &mem->range); + ret = amdgpu_ttm_tt_get_user_pages(bo, mem->range); if (ret) { + amdgpu_hmm_range_free(mem->range); + mem->range = NULL; pr_debug("Failed %d to get user pages\n", ret); /* Return -EFAULT bad address error as success. It will @@ -2742,8 +2755,8 @@ static int confirm_valid_user_pages_locked(struct amdkfd_process_info *process_i continue; /* Only check mem with hmm range associated */ - valid = amdgpu_ttm_tt_get_user_pages_done( - mem->bo->tbo.ttm, mem->range); + valid = amdgpu_hmm_range_valid(mem->range); + amdgpu_hmm_range_free(mem->range); mem->range = NULL; if (!valid) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c index c7d32fb216e4..636385c80f64 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_atomfirmware.c @@ -181,19 +181,22 @@ int amdgpu_atomfirmware_allocate_fb_scratch(struct amdgpu_device *adev) u8 frev, crev; int usage_bytes = 0; - if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) { - if (frev == 2 && crev == 1) { - fw_usage_v2_1 = - (struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset); - amdgpu_atomfirmware_allocate_fb_v2_1(adev, - fw_usage_v2_1, - &usage_bytes); - } else if (frev >= 2 && crev >= 2) { - fw_usage_v2_2 = - (struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset); - amdgpu_atomfirmware_allocate_fb_v2_2(adev, - fw_usage_v2_2, - &usage_bytes); + /* Skip atomfirmware allocation for SRIOV VFs when dynamic crit regn is enabled */ + if (!(amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled)) { + if (amdgpu_atom_parse_data_header(ctx, index, NULL, &frev, &crev, &data_offset)) { + if (frev == 2 && crev == 1) { + fw_usage_v2_1 = + (struct vram_usagebyfirmware_v2_1 *)(ctx->bios + data_offset); + amdgpu_atomfirmware_allocate_fb_v2_1(adev, + fw_usage_v2_1, + &usage_bytes); + } else if (frev >= 2 && crev >= 2) { + fw_usage_v2_2 = + (struct vram_usagebyfirmware_v2_2 *)(ctx->bios + data_offset); + amdgpu_atomfirmware_allocate_fb_v2_2(adev, + fw_usage_v2_2, + &usage_bytes); + } } } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c index 00e96419fcda..35d04e69aec0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bios.c @@ -96,13 +96,14 @@ void amdgpu_bios_release(struct amdgpu_device *adev) * part of the system bios. On boot, the system bios puts a * copy of the igp rom at the start of vram if a discrete card is * present. - * For SR-IOV, the vbios image is also put in VRAM in the VF. + * For SR-IOV, if dynamic critical region is not enabled, + * the vbios image is also put at the start of VRAM in the VF. */ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev) { - uint8_t __iomem *bios; + uint8_t __iomem *bios = NULL; resource_size_t vram_base; - resource_size_t size = 256 * 1024; /* ??? */ + u32 size = 256U * 1024U; /* ??? */ if (!(adev->flags & AMD_IS_APU)) if (amdgpu_device_need_post(adev)) @@ -114,18 +115,33 @@ static bool amdgpu_read_bios_from_vram(struct amdgpu_device *adev) adev->bios = NULL; vram_base = pci_resource_start(adev->pdev, 0); - bios = ioremap_wc(vram_base, size); - if (!bios) - return false; adev->bios = kmalloc(size, GFP_KERNEL); - if (!adev->bios) { - iounmap(bios); + if (!adev->bios) return false; + + /* For SRIOV with dynamic critical region is enabled, + * the vbios image is put at a dynamic offset of VRAM in the VF. + * If dynamic critical region is disabled, follow the existing logic as on baremetal. + */ + if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) { + if (amdgpu_virt_get_dynamic_data_info(adev, + AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, adev->bios, &size)) { + amdgpu_bios_release(adev); + return false; + } + } else { + bios = ioremap_wc(vram_base, size); + if (!bios) { + amdgpu_bios_release(adev); + return false; + } + + memcpy_fromio(adev->bios, bios, size); + iounmap(bios); } + adev->bios_size = size; - memcpy_fromio(adev->bios, bios, size); - iounmap(bios); if (!check_atom_bios(adev, size)) { amdgpu_bios_release(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h index a716c9886c74..2b5e7c46a39d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.h @@ -38,7 +38,7 @@ struct amdgpu_bo_list_entry { struct amdgpu_bo *bo; struct amdgpu_bo_va *bo_va; uint32_t priority; - struct hmm_range *range; + struct amdgpu_hmm_range *range; bool user_invalidated; }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c index 47e9bfba0642..9f96d568acf2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c @@ -734,10 +734,8 @@ amdgpu_connector_lvds_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -919,10 +917,8 @@ amdgpu_connector_vga_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1146,10 +1142,8 @@ out: amdgpu_connector_update_scratch_regs(connector, ret); exit: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } return ret; } @@ -1486,10 +1480,8 @@ amdgpu_connector_dp_detect(struct drm_connector *connector, bool force) amdgpu_connector_update_scratch_regs(connector, ret); out: - if (!drm_kms_helper_is_poll_worker()) { - pm_runtime_mark_last_busy(connector->dev->dev); + if (!drm_kms_helper_is_poll_worker()) pm_runtime_put_autosuspend(connector->dev->dev); - } if (connector->connector_type == DRM_MODE_CONNECTOR_DisplayPort || connector->connector_type == DRM_MODE_CONNECTOR_eDP) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c index ef996493115f..425a3e564360 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.c @@ -1,4 +1,4 @@ -// SPDX-License-Identifier: GPL-2.0 +// SPDX-License-Identifier: MIT /* * Copyright 2025 Advanced Micro Devices, Inc. * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h index bcb97d245673..353421807387 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cper.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2025 Advanced Micro Devices, Inc. * diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 9cd7741d2254..ecdfe6cb36cc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -29,7 +29,6 @@ #include <linux/pagemap.h> #include <linux/sync_file.h> #include <linux/dma-buf.h> -#include <linux/hmm.h> #include <drm/amdgpu_drm.h> #include <drm/drm_syncobj.h> @@ -41,6 +40,7 @@ #include "amdgpu_gmc.h" #include "amdgpu_gem.h" #include "amdgpu_ras.h" +#include "amdgpu_hmm.h" static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p, struct amdgpu_device *adev, @@ -364,6 +364,12 @@ static int amdgpu_cs_p2_ib(struct amdgpu_cs_parser *p, if (p->uf_bo && ring->funcs->no_user_fence) return -EINVAL; + if (!p->adev->debug_enable_ce_cs && + chunk_ib->flags & AMDGPU_IB_FLAG_CE) { + dev_err_ratelimited(p->adev->dev, "CE CS is blocked, use debug=0x400 to override\n"); + return -EINVAL; + } + if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX && chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) { if (chunk_ib->flags & AMDGPU_IB_FLAG_CE) @@ -702,7 +708,7 @@ static void amdgpu_cs_get_threshold_for_moves(struct amdgpu_device *adev, */ const s64 us_upper_bound = 200000; - if (!adev->mm_stats.log2_max_MBps) { + if ((!adev->mm_stats.log2_max_MBps) || !ttm_resource_manager_used(&adev->mman.vram_mgr.manager)) { *max_bytes = 0; *max_vis_bytes = 0; return; @@ -885,12 +891,17 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, bool userpage_invalidated = false; struct amdgpu_bo *bo = e->bo; - r = amdgpu_ttm_tt_get_user_pages(bo, &e->range); + e->range = amdgpu_hmm_range_alloc(NULL); + if (unlikely(!e->range)) + return -ENOMEM; + + r = amdgpu_ttm_tt_get_user_pages(bo, e->range); if (r) goto out_free_user_pages; for (i = 0; i < bo->tbo.ttm->num_pages; i++) { - if (bo->tbo.ttm->pages[i] != hmm_pfn_to_page(e->range->hmm_pfns[i])) { + if (bo->tbo.ttm->pages[i] != + hmm_pfn_to_page(e->range->hmm_range.hmm_pfns[i])) { userpage_invalidated = true; break; } @@ -984,9 +995,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p, out_free_user_pages: amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) { - struct amdgpu_bo *bo = e->bo; - - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, e->range); + amdgpu_hmm_range_free(e->range); e->range = NULL; } mutex_unlock(&p->bo_list->bo_list_mutex); @@ -1317,8 +1326,8 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p, */ r = 0; amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) { - r |= !amdgpu_ttm_tt_get_user_pages_done(e->bo->tbo.ttm, - e->range); + r |= !amdgpu_hmm_range_valid(e->range); + amdgpu_hmm_range_free(e->range); e->range = NULL; } if (r) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index f5d5c45ddc0d..afedea02188d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -236,7 +236,7 @@ static int amdgpu_ctx_init_entity(struct amdgpu_ctx *ctx, u32 hw_ip, r = amdgpu_xcp_select_scheds(adev, hw_ip, hw_prio, fpriv, &num_scheds, &scheds); if (r) - goto cleanup_entity; + goto error_free_entity; } /* disable load balance if the hw engine retains context among dependent jobs */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c index a70651050acf..62d43b8cbe58 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_debugfs.c @@ -129,7 +129,6 @@ static int amdgpu_debugfs_process_reg_op(bool read, struct file *f, if (use_bank) { if ((sh_bank != 0xFFFFFFFF && sh_bank >= adev->gfx.config.max_sh_per_se) || (se_bank != 0xFFFFFFFF && se_bank >= adev->gfx.config.max_shader_engines)) { - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return -EINVAL; @@ -179,7 +178,6 @@ end: if (pm_pg_lock) mutex_unlock(&adev->pm.mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); @@ -255,7 +253,6 @@ static ssize_t amdgpu_debugfs_regs2_op(struct file *f, char __user *buf, u32 off if (rd->id.use_grbm) { if ((rd->id.grbm.sh != 0xFFFFFFFF && rd->id.grbm.sh >= adev->gfx.config.max_sh_per_se) || (rd->id.grbm.se != 0xFFFFFFFF && rd->id.grbm.se >= adev->gfx.config.max_shader_engines)) { - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); mutex_unlock(&rd->lock); @@ -310,7 +307,6 @@ end: mutex_unlock(&rd->lock); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); @@ -446,7 +442,6 @@ static ssize_t amdgpu_debugfs_gprwave_read(struct file *f, char __user *buf, siz amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, rd->id.xcc_id); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (!x) { @@ -557,7 +552,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -617,7 +611,6 @@ static ssize_t amdgpu_debugfs_regs_pcie_write(struct file *f, const char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -676,7 +669,6 @@ static ssize_t amdgpu_debugfs_regs_didt_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -736,7 +728,6 @@ static ssize_t amdgpu_debugfs_regs_didt_write(struct file *f, const char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -795,7 +786,6 @@ static ssize_t amdgpu_debugfs_regs_smc_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -855,7 +845,6 @@ static ssize_t amdgpu_debugfs_regs_smc_write(struct file *f, const char __user * r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); amdgpu_virt_disable_access_debugfs(adev); return r; @@ -1003,7 +992,6 @@ static ssize_t amdgpu_debugfs_sensor_read(struct file *f, char __user *buf, r = amdgpu_dpm_read_sensor(adev, idx, &values[0], &valuesize); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (r) { @@ -1094,7 +1082,6 @@ static ssize_t amdgpu_debugfs_wave_read(struct file *f, char __user *buf, amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); if (!x) { @@ -1192,7 +1179,6 @@ static ssize_t amdgpu_debugfs_gpr_read(struct file *f, char __user *buf, amdgpu_gfx_select_se_sh(adev, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF, 0); mutex_unlock(&adev->grbm_idx_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); while (size) { @@ -1266,7 +1252,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_read(struct file *f, char __user r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1315,7 +1300,6 @@ static ssize_t amdgpu_debugfs_gfxoff_residency_write(struct file *f, const char r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1365,7 +1349,6 @@ static ssize_t amdgpu_debugfs_gfxoff_count_read(struct file *f, char __user *buf r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1414,7 +1397,6 @@ static ssize_t amdgpu_debugfs_gfxoff_write(struct file *f, const char __user *bu r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1460,7 +1442,6 @@ static ssize_t amdgpu_debugfs_gfxoff_read(struct file *f, char __user *buf, r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1501,7 +1482,6 @@ static ssize_t amdgpu_debugfs_gfxoff_status_read(struct file *f, char __user *bu r = result; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -1701,7 +1681,6 @@ static int amdgpu_debugfs_test_ib_show(struct seq_file *m, void *unused) up_write(&adev->reset_domain->sem); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1721,7 +1700,6 @@ static int amdgpu_debugfs_evict_vram(void *data, u64 *val) *val = amdgpu_ttm_evict_resources(adev, TTM_PL_VRAM); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1742,7 +1720,6 @@ static int amdgpu_debugfs_evict_gtt(void *data, u64 *val) *val = amdgpu_ttm_evict_resources(adev, TTM_PL_TT); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; @@ -1762,7 +1739,6 @@ static int amdgpu_debugfs_benchmark(void *data, u64 val) r = amdgpu_benchmark(adev, val); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return r; @@ -1902,7 +1878,7 @@ no_preempt: continue; } job = to_amdgpu_job(s_job); - if (preempted && (&job->hw_fence.base) == fence) + if (preempted && (&job->hw_fence->base) == fence) /* mark the job as preempted */ job->preemption_status |= AMDGPU_IB_PREEMPTED; } @@ -2014,7 +1990,6 @@ static int amdgpu_debugfs_sclk_set(void *data, u64 val) ret = -EINVAL; out: - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return ret; @@ -2123,10 +2098,9 @@ int amdgpu_debugfs_init(struct amdgpu_device *adev) debugfs_create_blob("amdgpu_vbios", 0444, root, &adev->debugfs_vbios_blob); - adev->debugfs_discovery_blob.data = adev->mman.discovery_bin; - adev->debugfs_discovery_blob.size = adev->mman.discovery_tmr_size; - debugfs_create_blob("amdgpu_discovery", 0444, root, - &adev->debugfs_discovery_blob); + if (adev->discovery.debugfs_blob.size) + debugfs_create_blob("amdgpu_discovery", 0444, root, + &adev->discovery.debugfs_blob); return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c index 8a026bc9ea44..4e2fe6674db8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dev_coredump.c @@ -217,8 +217,7 @@ amdgpu_devcoredump_read(char *buffer, loff_t offset, size_t count, drm_printf(&p, "version: " AMDGPU_COREDUMP_VERSION "\n"); drm_printf(&p, "kernel: " UTS_RELEASE "\n"); drm_printf(&p, "module: " KBUILD_MODNAME "\n"); - drm_printf(&p, "time: %lld.%09ld\n", coredump->reset_time.tv_sec, - coredump->reset_time.tv_nsec); + drm_printf(&p, "time: %ptSp\n", &coredump->reset_time); if (coredump->reset_task_info.task.pid) drm_printf(&p, "process_name: %s PID: %d\n", diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 7a899fb4de29..a1817b4b5173 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -71,6 +71,7 @@ #include "amdgpu_xgmi.h" #include "amdgpu_ras.h" +#include "amdgpu_ras_mgr.h" #include "amdgpu_pmu.h" #include "amdgpu_fru_eeprom.h" #include "amdgpu_reset.h" @@ -179,6 +180,10 @@ struct amdgpu_init_level amdgpu_init_minimal_xgmi = { BIT(AMD_IP_BLOCK_TYPE_PSP) }; +static int amdgpu_device_ip_resume_phase1(struct amdgpu_device *adev); +static int amdgpu_device_ip_resume_phase2(struct amdgpu_device *adev); +static int amdgpu_device_ip_resume_phase3(struct amdgpu_device *adev); + static void amdgpu_device_load_switch_state(struct amdgpu_device *adev); static inline bool amdgpu_ip_member_of_hwini(struct amdgpu_device *adev, @@ -1882,6 +1887,13 @@ static bool amdgpu_device_pcie_dynamic_switching_supported(struct amdgpu_device static bool amdgpu_device_aspm_support_quirk(struct amdgpu_device *adev) { + /* Enabling ASPM causes randoms hangs on Tahiti and Oland on Zen4. + * It's unclear if this is a platform-specific or GPU-specific issue. + * Disable ASPM on SI for the time being. + */ + if (adev->family == AMDGPU_FAMILY_SI) + return true; + #if IS_ENABLED(CONFIG_X86) struct cpuinfo_x86 *c = &cpu_data(0); @@ -2380,7 +2392,7 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev, } /** - * amdgpu_device_ip_is_valid - is the hardware IP enabled + * amdgpu_device_ip_is_hw - is the hardware IP enabled * * @adev: amdgpu_device pointer * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.) @@ -2388,6 +2400,27 @@ int amdgpu_device_ip_wait_for_idle(struct amdgpu_device *adev, * Check if the hardware IP is enable or not. * Returns true if it the IP is enable, false if not. */ +bool amdgpu_device_ip_is_hw(struct amdgpu_device *adev, + enum amd_ip_block_type block_type) +{ + int i; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (adev->ip_blocks[i].version->type == block_type) + return adev->ip_blocks[i].status.hw; + } + return false; +} + +/** + * amdgpu_device_ip_is_valid - is the hardware IP valid + * + * @adev: amdgpu_device pointer + * @block_type: Type of hardware IP (SMU, GFX, UVD, etc.) + * + * Check if the hardware IP is valid or not. + * Returns true if it the IP is valid, false if not. + */ bool amdgpu_device_ip_is_valid(struct amdgpu_device *adev, enum amd_ip_block_type block_type) { @@ -2466,6 +2499,7 @@ static const char *ip_block_names[] = { [AMD_IP_BLOCK_TYPE_VPE] = "vpe", [AMD_IP_BLOCK_TYPE_UMSCH_MM] = "umsch_mm", [AMD_IP_BLOCK_TYPE_ISP] = "isp", + [AMD_IP_BLOCK_TYPE_RAS] = "ras", }; static const char *ip_block_name(struct amdgpu_device *adev, enum amd_ip_block_type type) @@ -2626,11 +2660,13 @@ static int amdgpu_device_parse_gpu_info_fw(struct amdgpu_device *adev) chip_name = "arcturus"; break; case CHIP_NAVI12: - if (adev->mman.discovery_bin) + if (adev->discovery.bin) return 0; chip_name = "navi12"; break; case CHIP_CYAN_SKILLFISH: + if (adev->discovery.bin) + return 0; chip_name = "cyan_skillfish"; break; } @@ -2754,6 +2790,10 @@ static int amdgpu_device_ip_early_init(struct amdgpu_device *adev) r = amdgpu_virt_request_full_gpu(adev, true); if (r) return r; + + r = amdgpu_virt_init_critical_region(adev); + if (r) + return r; } switch (adev->asic_type) { @@ -3407,10 +3447,11 @@ int amdgpu_device_set_pg_state(struct amdgpu_device *adev, (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_GFX || adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_SDMA)) continue; - /* skip CG for VCE/UVD, it's handled specially */ + /* skip CG for VCE/UVD/VPE, it's handled specially */ if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_UVD && adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCE && adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VCN && + adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_VPE && adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_JPEG && adev->ip_blocks[i].version->funcs->set_powergating_state) { /* enable powergating to save power */ @@ -3642,6 +3683,20 @@ static int amdgpu_device_ip_fini_early(struct amdgpu_device *adev) "failed to release exclusive mode on fini\n"); } + /* + * Driver reload on the APU can fail due to firmware validation because + * the PSP is always running, as it is shared across the whole SoC. + * This same issue does not occur on dGPU because it has a mechanism + * that checks whether the PSP is running. A solution for those issues + * in the APU is to trigger a GPU reset, but this should be done during + * the unload phase to avoid adding boot latency and screen flicker. + */ + if ((adev->flags & AMD_IS_APU) && !adev->gmc.is_app_apu) { + r = amdgpu_asic_reset(adev); + if (r) + dev_err(adev->dev, "asic reset on %s failed\n", __func__); + } + return 0; } @@ -3752,7 +3807,7 @@ static void amdgpu_device_delay_enable_gfx_off(struct work_struct *work) */ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) { - int i, r; + int i, r, rec; amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE); amdgpu_device_set_cg_state(adev, AMD_CG_STATE_UNGATE); @@ -3773,13 +3828,25 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) if (adev->ip_blocks[i].version->type != AMD_IP_BLOCK_TYPE_DCE) continue; - /* XXX handle errors */ r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); if (r) - return r; + goto unwind; } return 0; +unwind: + rec = amdgpu_device_ip_resume_phase3(adev); + if (rec) + dev_err(adev->dev, + "amdgpu_device_ip_resume_phase3 failed during unwind: %d\n", + rec); + + amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW); + + amdgpu_device_set_pg_state(adev, AMD_PG_STATE_GATE); + amdgpu_device_set_cg_state(adev, AMD_CG_STATE_GATE); + + return r; } /** @@ -3795,7 +3862,7 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) */ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) { - int i, r; + int i, r, rec; if (adev->in_s0ix) amdgpu_dpm_gfx_state_change(adev, sGpuChangeState_D3Entry); @@ -3856,9 +3923,9 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_PSP) continue; - /* XXX handle errors */ r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); - adev->ip_blocks[i].status.hw = false; + if (r) + goto unwind; /* handle putting the SMC in the appropriate state */ if (!amdgpu_sriov_vf(adev)) { @@ -3868,13 +3935,40 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) dev_err(adev->dev, "SMC failed to set mp1 state %d, %d\n", adev->mp1_state, r); - return r; + goto unwind; } } } } return 0; +unwind: + /* suspend phase 2 = resume phase 1 + resume phase 2 */ + rec = amdgpu_device_ip_resume_phase1(adev); + if (rec) { + dev_err(adev->dev, + "amdgpu_device_ip_resume_phase1 failed during unwind: %d\n", + rec); + return r; + } + + rec = amdgpu_device_fw_loading(adev); + if (rec) { + dev_err(adev->dev, + "amdgpu_device_fw_loading failed during unwind: %d\n", + rec); + return r; + } + + rec = amdgpu_device_ip_resume_phase2(adev); + if (rec) { + dev_err(adev->dev, + "amdgpu_device_ip_resume_phase2 failed during unwind: %d\n", + rec); + return r; + } + + return r; } /** @@ -3888,7 +3982,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) * in each IP into a state suitable for suspend. * Returns 0 on success, negative error code on failure. */ -int amdgpu_device_ip_suspend(struct amdgpu_device *adev) +static int amdgpu_device_ip_suspend(struct amdgpu_device *adev) { int r; @@ -4172,25 +4266,13 @@ bool amdgpu_device_asic_has_dc_support(struct pci_dev *pdev, case CHIP_PITCAIRN: case CHIP_VERDE: case CHIP_OLAND: - /* - * We have systems in the wild with these ASICs that require - * LVDS and VGA support which is not supported with DC. - * - * Fallback to the non-DC driver here by default so as not to - * cause regressions. - */ -#if defined(CONFIG_DRM_AMD_DC_SI) - return amdgpu_dc > 0; -#else - return false; -#endif - case CHIP_BONAIRE: + return amdgpu_dc != 0 && IS_ENABLED(CONFIG_DRM_AMD_DC_SI); case CHIP_KAVERI: case CHIP_KABINI: case CHIP_MULLINS: /* * We have systems in the wild with these ASICs that require - * VGA support which is not supported with DC. + * TRAVIS and NUTMEG support which is not supported with DC. * * Fallback to the non-DC driver here by default so as not to * cause regressions. @@ -4278,58 +4360,53 @@ static int amdgpu_device_get_job_timeout_settings(struct amdgpu_device *adev) long timeout; int ret = 0; - /* - * By default timeout for jobs is 10 sec - */ - adev->compute_timeout = adev->gfx_timeout = msecs_to_jiffies(10000); - adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout; + /* By default timeout for all queues is 2 sec */ + adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout = + adev->video_timeout = msecs_to_jiffies(2000); - if (strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) { - while ((timeout_setting = strsep(&input, ",")) && - strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) { - ret = kstrtol(timeout_setting, 0, &timeout); - if (ret) - return ret; + if (!strnlen(input, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) + return 0; - if (timeout == 0) { - index++; - continue; - } else if (timeout < 0) { - timeout = MAX_SCHEDULE_TIMEOUT; - dev_warn(adev->dev, "lockup timeout disabled"); - add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK); - } else { - timeout = msecs_to_jiffies(timeout); - } + while ((timeout_setting = strsep(&input, ",")) && + strnlen(timeout_setting, AMDGPU_MAX_TIMEOUT_PARAM_LENGTH)) { + ret = kstrtol(timeout_setting, 0, &timeout); + if (ret) + return ret; - switch (index++) { - case 0: - adev->gfx_timeout = timeout; - break; - case 1: - adev->compute_timeout = timeout; - break; - case 2: - adev->sdma_timeout = timeout; - break; - case 3: - adev->video_timeout = timeout; - break; - default: - break; - } + if (timeout == 0) { + index++; + continue; + } else if (timeout < 0) { + timeout = MAX_SCHEDULE_TIMEOUT; + dev_warn(adev->dev, "lockup timeout disabled"); + add_taint(TAINT_SOFTLOCKUP, LOCKDEP_STILL_OK); + } else { + timeout = msecs_to_jiffies(timeout); } - /* - * There is only one value specified and - * it should apply to all non-compute jobs. - */ - if (index == 1) { - adev->sdma_timeout = adev->video_timeout = adev->gfx_timeout; - if (amdgpu_sriov_vf(adev) || amdgpu_passthrough(adev)) - adev->compute_timeout = adev->gfx_timeout; + + switch (index++) { + case 0: + adev->gfx_timeout = timeout; + break; + case 1: + adev->compute_timeout = timeout; + break; + case 2: + adev->sdma_timeout = timeout; + break; + case 3: + adev->video_timeout = timeout; + break; + default: + break; } } + /* When only one value specified apply it to all queues. */ + if (index == 1) + adev->gfx_timeout = adev->compute_timeout = adev->sdma_timeout = + adev->video_timeout = timeout; + return ret; } @@ -4384,6 +4461,55 @@ static void amdgpu_device_set_mcbp(struct amdgpu_device *adev) dev_info(adev->dev, "MCBP is enabled\n"); } +static int amdgpu_device_sys_interface_init(struct amdgpu_device *adev) +{ + int r; + + r = amdgpu_atombios_sysfs_init(adev); + if (r) + drm_err(&adev->ddev, + "registering atombios sysfs failed (%d).\n", r); + + r = amdgpu_pm_sysfs_init(adev); + if (r) + dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r); + + r = amdgpu_ucode_sysfs_init(adev); + if (r) { + adev->ucode_sysfs_en = false; + dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r); + } else + adev->ucode_sysfs_en = true; + + r = amdgpu_device_attr_sysfs_init(adev); + if (r) + dev_err(adev->dev, "Could not create amdgpu device attr\n"); + + r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group); + if (r) + dev_err(adev->dev, + "Could not create amdgpu board attributes\n"); + + amdgpu_fru_sysfs_init(adev); + amdgpu_reg_state_sysfs_init(adev); + amdgpu_xcp_sysfs_init(adev); + + return r; +} + +static void amdgpu_device_sys_interface_fini(struct amdgpu_device *adev) +{ + if (adev->pm.sysfs_initialized) + amdgpu_pm_sysfs_fini(adev); + if (adev->ucode_sysfs_en) + amdgpu_ucode_sysfs_fini(adev); + amdgpu_device_attr_sysfs_fini(adev); + amdgpu_fru_sysfs_fini(adev); + + amdgpu_reg_state_sysfs_fini(adev); + amdgpu_xcp_sysfs_fini(adev); +} + /** * amdgpu_device_init - initialize the driver * @@ -4483,7 +4609,6 @@ int amdgpu_device_init(struct amdgpu_device *adev, mutex_init(&adev->gfx.userq_sch_mutex); mutex_init(&adev->gfx.workload_profile_mutex); mutex_init(&adev->vcn.workload_profile_mutex); - mutex_init(&adev->userq_mutex); amdgpu_device_init_apu_flags(adev); @@ -4511,7 +4636,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, INIT_LIST_HEAD(&adev->pm.od_kobj_list); - INIT_LIST_HEAD(&adev->userq_mgr_list); + xa_init(&adev->userq_doorbell_xa); INIT_DELAYED_WORK(&adev->delayed_init_work, amdgpu_device_delayed_init_work_handler); @@ -4534,6 +4659,7 @@ int amdgpu_device_init(struct amdgpu_device *adev, } INIT_WORK(&adev->xgmi_reset_work, amdgpu_device_xgmi_reset_func); + INIT_WORK(&adev->userq_reset_work, amdgpu_userq_reset_work); adev->gfx.gfx_off_req_count = 1; adev->gfx.gfx_off_residency = 0; @@ -4807,39 +4933,14 @@ fence_driver_init: flush_delayed_work(&adev->delayed_init_work); } + if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI) + amdgpu_xgmi_reset_on_init(adev); /* * Place those sysfs registering after `late_init`. As some of those * operations performed in `late_init` might affect the sysfs * interfaces creating. */ - r = amdgpu_atombios_sysfs_init(adev); - if (r) - drm_err(&adev->ddev, - "registering atombios sysfs failed (%d).\n", r); - - r = amdgpu_pm_sysfs_init(adev); - if (r) - dev_err(adev->dev, "registering pm sysfs failed (%d).\n", r); - - r = amdgpu_ucode_sysfs_init(adev); - if (r) { - adev->ucode_sysfs_en = false; - dev_err(adev->dev, "Creating firmware sysfs failed (%d).\n", r); - } else - adev->ucode_sysfs_en = true; - - r = amdgpu_device_attr_sysfs_init(adev); - if (r) - dev_err(adev->dev, "Could not create amdgpu device attr\n"); - - r = devm_device_add_group(adev->dev, &amdgpu_board_attrs_group); - if (r) - dev_err(adev->dev, - "Could not create amdgpu board attributes\n"); - - amdgpu_fru_sysfs_init(adev); - amdgpu_reg_state_sysfs_init(adev); - amdgpu_xcp_sysfs_init(adev); + r = amdgpu_device_sys_interface_init(adev); if (IS_ENABLED(CONFIG_PERF_EVENTS)) r = amdgpu_pmu_init(adev); @@ -4867,9 +4968,6 @@ fence_driver_init: if (px) vga_switcheroo_init_domain_pm_ops(adev->dev, &adev->vga_pm_domain); - if (adev->init_lvl->level == AMDGPU_INIT_LEVEL_MINIMAL_XGMI) - amdgpu_xgmi_reset_on_init(adev); - amdgpu_device_check_iommu_direct_map(adev); adev->pm_nb.notifier_call = amdgpu_device_pm_notifier; @@ -4961,15 +5059,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) } amdgpu_fence_driver_hw_fini(adev); - if (adev->pm.sysfs_initialized) - amdgpu_pm_sysfs_fini(adev); - if (adev->ucode_sysfs_en) - amdgpu_ucode_sysfs_fini(adev); - amdgpu_device_attr_sysfs_fini(adev); - amdgpu_fru_sysfs_fini(adev); - - amdgpu_reg_state_sysfs_fini(adev); - amdgpu_xcp_sysfs_fini(adev); + amdgpu_device_sys_interface_fini(adev); /* disable ras feature must before hw fini */ amdgpu_ras_pre_fini(adev); @@ -5044,7 +5134,7 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev) if (IS_ENABLED(CONFIG_PERF_EVENTS)) amdgpu_pmu_fini(adev); - if (adev->mman.discovery_bin) + if (adev->discovery.bin) amdgpu_discovery_fini(adev); amdgpu_reset_put_reset_domain(adev->reset_domain); @@ -5192,7 +5282,7 @@ void amdgpu_device_complete(struct drm_device *dev) int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) { struct amdgpu_device *adev = drm_to_adev(dev); - int r = 0; + int r, rec; if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) return 0; @@ -5208,39 +5298,92 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) return r; } - if (amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D3)) - dev_warn(adev->dev, "smart shift update failed\n"); + r = amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D3); + if (r) + goto unwind_sriov; if (notify_clients) - drm_client_dev_suspend(adev_to_drm(adev), false); + drm_client_dev_suspend(adev_to_drm(adev)); cancel_delayed_work_sync(&adev->delayed_init_work); amdgpu_ras_suspend(adev); - amdgpu_device_ip_suspend_phase1(adev); + r = amdgpu_device_ip_suspend_phase1(adev); + if (r) + goto unwind_smartshift; amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm); - amdgpu_userq_suspend(adev); + r = amdgpu_userq_suspend(adev); + if (r) + goto unwind_ip_phase1; r = amdgpu_device_evict_resources(adev); if (r) - return r; + goto unwind_userq; amdgpu_ttm_set_buffer_funcs_status(adev, false); amdgpu_fence_driver_hw_fini(adev); - amdgpu_device_ip_suspend_phase2(adev); + r = amdgpu_device_ip_suspend_phase2(adev); + if (r) + goto unwind_evict; if (amdgpu_sriov_vf(adev)) amdgpu_virt_release_full_gpu(adev, false); - r = amdgpu_dpm_notify_rlc_state(adev, false); - if (r) + return 0; + +unwind_evict: + if (adev->mman.buffer_funcs_ring->sched.ready) + amdgpu_ttm_set_buffer_funcs_status(adev, true); + amdgpu_fence_driver_hw_init(adev); + +unwind_userq: + rec = amdgpu_userq_resume(adev); + if (rec) { + dev_warn(adev->dev, "failed to re-initialize user queues: %d\n", rec); + return r; + } + rec = amdgpu_amdkfd_resume(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm); + if (rec) { + dev_warn(adev->dev, "failed to re-initialize kfd: %d\n", rec); return r; + } - return 0; +unwind_ip_phase1: + /* suspend phase 1 = resume phase 3 */ + rec = amdgpu_device_ip_resume_phase3(adev); + if (rec) { + dev_warn(adev->dev, "failed to re-initialize IPs phase1: %d\n", rec); + return r; + } + +unwind_smartshift: + rec = amdgpu_acpi_smart_shift_update(adev, AMDGPU_SS_DEV_D0); + if (rec) { + dev_warn(adev->dev, "failed to re-update smart shift: %d\n", rec); + return r; + } + + if (notify_clients) + drm_client_dev_resume(adev_to_drm(adev)); + + amdgpu_ras_resume(adev); + +unwind_sriov: + if (amdgpu_sriov_vf(adev)) { + rec = amdgpu_virt_request_full_gpu(adev, true); + if (rec) { + dev_warn(adev->dev, "failed to reinitialize sriov: %d\n", rec); + return r; + } + } + + adev->in_suspend = adev->in_s0ix = adev->in_s3 = false; + + return r; } static inline int amdgpu_virt_resume(struct amdgpu_device *adev) @@ -5346,7 +5489,7 @@ exit: flush_delayed_work(&adev->delayed_init_work); if (notify_clients) - drm_client_dev_resume(adev_to_drm(adev), false); + drm_client_dev_resume(adev_to_drm(adev)); amdgpu_ras_resume(adev); @@ -5802,11 +5945,6 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, if (!amdgpu_ring_sched_ready(ring)) continue; - /* Clear job fence from fence drv to avoid force_completion - * leave NULL and vm flush fence in fence drv - */ - amdgpu_fence_driver_clear_job_fences(ring); - /* after all hw jobs are reset, hw fence is meaningless, so force_completion */ amdgpu_fence_driver_force_completion(ring); } @@ -5951,7 +6089,11 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context) if (r) goto out; - drm_client_dev_resume(adev_to_drm(tmp_adev), false); + r = amdgpu_userq_post_reset(tmp_adev, vram_lost); + if (r) + goto out; + + drm_client_dev_resume(adev_to_drm(tmp_adev)); /* * The GPU enters bad state once faulty pages @@ -6173,6 +6315,7 @@ static inline void amdgpu_device_stop_pending_resets(struct amdgpu_device *adev) if (!amdgpu_sriov_vf(adev)) cancel_work(&adev->reset_work); #endif + cancel_work(&adev->userq_reset_work); if (adev->kfd.dev) cancel_work(&adev->kfd.reset_work); @@ -6286,13 +6429,15 @@ static void amdgpu_device_halt_activities(struct amdgpu_device *adev, */ amdgpu_unregister_gpu_instance(tmp_adev); - drm_client_dev_suspend(adev_to_drm(tmp_adev), false); + drm_client_dev_suspend(adev_to_drm(tmp_adev)); /* disable ras on ALL IPs */ if (!need_emergency_restart && !amdgpu_reset_in_dpc(adev) && amdgpu_device_ip_need_full_reset(tmp_adev)) amdgpu_ras_suspend(tmp_adev); + amdgpu_userq_pre_reset(tmp_adev); + for (i = 0; i < AMDGPU_MAX_RINGS; ++i) { struct amdgpu_ring *ring = tmp_adev->rings[i]; @@ -6522,6 +6667,9 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, goto end_reset; } + /* Cannot be called after locking reset domain */ + amdgpu_ras_pre_reset(adev, &device_list); + /* We need to lock reset domain only once both for XGMI and single device */ amdgpu_device_recovery_get_reset_lock(adev, &device_list); @@ -6535,7 +6683,7 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev, * * job->base holds a reference to parent fence */ - if (job && dma_fence_is_signaled(&job->hw_fence.base)) { + if (job && dma_fence_is_signaled(&job->hw_fence->base)) { job_signaled = true; dev_info(adev->dev, "Guilty job already signaled, skipping HW reset"); goto skip_hw_reset; @@ -6552,6 +6700,7 @@ skip_sched_resume: amdgpu_device_gpu_resume(adev, &device_list, need_emergency_restart); reset_unlock: amdgpu_device_recovery_put_reset_lock(adev, &device_list); + amdgpu_ras_post_reset(adev, &device_list); end_reset: if (hive) { mutex_unlock(&hive->hive_lock); @@ -7279,10 +7428,17 @@ void amdgpu_device_flush_hdp(struct amdgpu_device *adev, if (adev->gmc.xgmi.connected_to_cpu) return; - if (ring && ring->funcs->emit_hdp_flush) + if (ring && ring->funcs->emit_hdp_flush) { amdgpu_ring_emit_hdp_flush(ring); - else - amdgpu_asic_flush_hdp(adev, ring); + return; + } + + if (!ring && amdgpu_sriov_runtime(adev)) { + if (!amdgpu_kiq_hdp_flush(adev)) + return; + } + + amdgpu_hdp_flush(adev, ring); } void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev, @@ -7295,7 +7451,7 @@ void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev, if (adev->gmc.xgmi.connected_to_cpu) return; - amdgpu_asic_invalidate_hdp(adev, ring); + amdgpu_hdp_invalidate(adev, ring); } int amdgpu_in_reset(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 73401f0aeb34..fa2a22dfa048 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -107,6 +107,7 @@ #include "vcn_v5_0_1.h" #include "jpeg_v5_0_0.h" #include "jpeg_v5_0_1.h" +#include "amdgpu_ras_mgr.h" #include "amdgpu_vpe.h" #if defined(CONFIG_DRM_AMD_ISP) @@ -254,9 +255,9 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, pos = tmr_offset + tmr_size - DISCOVERY_TMR_OFFSET; /* This region is read-only and reserved from system use */ - discv_regn = memremap(pos, adev->mman.discovery_tmr_size, MEMREMAP_WC); + discv_regn = memremap(pos, adev->discovery.size, MEMREMAP_WC); if (discv_regn) { - memcpy(binary, discv_regn, adev->mman.discovery_tmr_size); + memcpy(binary, discv_regn, adev->discovery.size); memunmap(discv_regn); return 0; } @@ -298,10 +299,31 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, else vram_size <<= 20; + /* + * If in VRAM, discovery TMR is marked for reservation. If it is in system mem, + * then it is not required to be reserved. + */ if (sz_valid) { - uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET; - amdgpu_device_vram_access(adev, pos, (uint32_t *)binary, - adev->mman.discovery_tmr_size, false); + if (amdgpu_sriov_vf(adev) && adev->virt.is_dynamic_crit_regn_enabled) { + /* For SRIOV VFs with dynamic critical region enabled, + * we will get the IPD binary via below call. + * If dynamic critical is disabled, fall through to normal seq. + */ + if (amdgpu_virt_get_dynamic_data_info(adev, + AMD_SRIOV_MSG_IPD_TABLE_ID, binary, + &adev->discovery.size)) { + dev_err(adev->dev, + "failed to read discovery info from dynamic critical region."); + ret = -EINVAL; + goto exit; + } + } else { + uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET; + + amdgpu_device_vram_access(adev, pos, (uint32_t *)binary, + adev->discovery.size, false); + adev->discovery.reserve_tmr = true; + } } else { ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary); } @@ -310,7 +332,7 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, dev_err(adev->dev, "failed to read discovery info from memory, vram size read: %llx", vram_size); - +exit: return ret; } @@ -389,6 +411,7 @@ static void amdgpu_discovery_harvest_config_quirk(struct amdgpu_device *adev) static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, struct binary_header *bhdr) { + uint8_t *discovery_bin = adev->discovery.bin; struct table_info *info; uint16_t checksum; uint16_t offset; @@ -398,14 +421,14 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, checksum = le16_to_cpu(info->checksum); struct nps_info_header *nhdr = - (struct nps_info_header *)(adev->mman.discovery_bin + offset); + (struct nps_info_header *)(discovery_bin + offset); if (le32_to_cpu(nhdr->table_id) != NPS_INFO_TABLE_ID) { dev_dbg(adev->dev, "invalid ip discovery nps info table id\n"); return -EINVAL; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, le32_to_cpu(nhdr->size_bytes), checksum)) { dev_dbg(adev->dev, "invalid nps info data table checksum\n"); @@ -417,8 +440,11 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev) { - if (amdgpu_discovery == 2) + if (amdgpu_discovery == 2) { + /* Assume there is valid discovery TMR in VRAM even if binary is sideloaded */ + adev->discovery.reserve_tmr = true; return "amdgpu/ip_discovery.bin"; + } switch (adev->asic_type) { case CHIP_VEGA10: @@ -447,49 +473,53 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) { struct table_info *info; struct binary_header *bhdr; + uint8_t *discovery_bin; const char *fw_name; uint16_t offset; uint16_t size; uint16_t checksum; int r; - adev->mman.discovery_tmr_size = DISCOVERY_TMR_SIZE; - adev->mman.discovery_bin = kzalloc(adev->mman.discovery_tmr_size, GFP_KERNEL); - if (!adev->mman.discovery_bin) + adev->discovery.bin = kzalloc(DISCOVERY_TMR_SIZE, GFP_KERNEL); + if (!adev->discovery.bin) return -ENOMEM; + adev->discovery.size = DISCOVERY_TMR_SIZE; + adev->discovery.debugfs_blob.data = adev->discovery.bin; + adev->discovery.debugfs_blob.size = adev->discovery.size; + discovery_bin = adev->discovery.bin; /* Read from file if it is the preferred option */ fw_name = amdgpu_discovery_get_fw_name(adev); if (fw_name != NULL) { drm_dbg(&adev->ddev, "use ip discovery information from file"); - r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin, fw_name); + r = amdgpu_discovery_read_binary_from_file(adev, discovery_bin, + fw_name); if (r) goto out; } else { drm_dbg(&adev->ddev, "use ip discovery information from memory"); - r = amdgpu_discovery_read_binary_from_mem( - adev, adev->mman.discovery_bin); + r = amdgpu_discovery_read_binary_from_mem(adev, discovery_bin); if (r) goto out; } /* check the ip discovery binary signature */ - if (!amdgpu_discovery_verify_binary_signature(adev->mman.discovery_bin)) { + if (!amdgpu_discovery_verify_binary_signature(discovery_bin)) { dev_err(adev->dev, "get invalid ip discovery binary signature\n"); r = -EINVAL; goto out; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = offsetof(struct binary_header, binary_checksum) + sizeof(bhdr->binary_checksum); size = le16_to_cpu(bhdr->binary_size) - offset; checksum = le16_to_cpu(bhdr->binary_checksum); - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - size, checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, size, + checksum)) { dev_err(adev->dev, "invalid ip discovery binary checksum\n"); r = -EINVAL; goto out; @@ -501,15 +531,16 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct ip_discovery_header *ihdr = - (struct ip_discovery_header *)(adev->mman.discovery_bin + offset); + (struct ip_discovery_header *)(discovery_bin + offset); if (le32_to_cpu(ihdr->signature) != DISCOVERY_TABLE_SIGNATURE) { dev_err(adev->dev, "invalid ip discovery data table signature\n"); r = -EINVAL; goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le16_to_cpu(ihdr->size), checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, + le16_to_cpu(ihdr->size), + checksum)) { dev_err(adev->dev, "invalid ip discovery data table checksum\n"); r = -EINVAL; goto out; @@ -522,7 +553,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct gpu_info_header *ghdr = - (struct gpu_info_header *)(adev->mman.discovery_bin + offset); + (struct gpu_info_header *)(discovery_bin + offset); if (le32_to_cpu(ghdr->table_id) != GC_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery gc table id\n"); @@ -530,8 +561,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(ghdr->size), checksum)) { + if (!amdgpu_discovery_verify_checksum(discovery_bin + offset, + le32_to_cpu(ghdr->size), + checksum)) { dev_err(adev->dev, "invalid gc data table checksum\n"); r = -EINVAL; goto out; @@ -544,7 +576,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct harvest_info_header *hhdr = - (struct harvest_info_header *)(adev->mman.discovery_bin + offset); + (struct harvest_info_header *)(discovery_bin + offset); if (le32_to_cpu(hhdr->signature) != HARVEST_TABLE_SIGNATURE) { dev_err(adev->dev, "invalid ip discovery harvest table signature\n"); @@ -552,8 +584,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - sizeof(struct harvest_table), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + sizeof(struct harvest_table), checksum)) { dev_err(adev->dev, "invalid harvest data table checksum\n"); r = -EINVAL; goto out; @@ -566,7 +599,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (offset) { struct vcn_info_header *vhdr = - (struct vcn_info_header *)(adev->mman.discovery_bin + offset); + (struct vcn_info_header *)(discovery_bin + offset); if (le32_to_cpu(vhdr->table_id) != VCN_INFO_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery vcn table id\n"); @@ -574,8 +607,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(vhdr->size_bytes), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + le32_to_cpu(vhdr->size_bytes), checksum)) { dev_err(adev->dev, "invalid vcn data table checksum\n"); r = -EINVAL; goto out; @@ -588,7 +622,7 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) if (0 && offset) { struct mall_info_header *mhdr = - (struct mall_info_header *)(adev->mman.discovery_bin + offset); + (struct mall_info_header *)(discovery_bin + offset); if (le32_to_cpu(mhdr->table_id) != MALL_INFO_TABLE_ID) { dev_err(adev->dev, "invalid ip discovery mall table id\n"); @@ -596,8 +630,9 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) goto out; } - if (!amdgpu_discovery_verify_checksum(adev->mman.discovery_bin + offset, - le32_to_cpu(mhdr->size_bytes), checksum)) { + if (!amdgpu_discovery_verify_checksum( + discovery_bin + offset, + le32_to_cpu(mhdr->size_bytes), checksum)) { dev_err(adev->dev, "invalid mall data table checksum\n"); r = -EINVAL; goto out; @@ -607,8 +642,8 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) return 0; out: - kfree(adev->mman.discovery_bin); - adev->mman.discovery_bin = NULL; + kfree(adev->discovery.bin); + adev->discovery.bin = NULL; if ((amdgpu_discovery != 2) && (RREG32(mmIP_DISCOVERY_VERSION) == 4)) amdgpu_ras_query_boot_status(adev, 4); @@ -620,8 +655,8 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev); void amdgpu_discovery_fini(struct amdgpu_device *adev) { amdgpu_discovery_sysfs_fini(adev); - kfree(adev->mman.discovery_bin); - adev->mman.discovery_bin = NULL; + kfree(adev->discovery.bin); + adev->discovery.bin = NULL; } static int amdgpu_discovery_validate_ip(struct amdgpu_device *adev, @@ -646,6 +681,7 @@ static int amdgpu_discovery_validate_ip(struct amdgpu_device *adev, static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, uint32_t *vcn_harvest_count) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; @@ -655,21 +691,21 @@ static void amdgpu_discovery_read_harvest_bit_per_ip(struct amdgpu_device *adev, uint8_t inst; int i, j; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); /* scan harvest bit of all IP data structures */ for (i = 0; i < num_dies; i++) { die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); for (j = 0; j < num_ips; j++) { - ip = (struct ip *)(adev->mman.discovery_bin + - ip_offset); + ip = (struct ip *)(discovery_bin + ip_offset); inst = ip->number_instance; hw_id = le16_to_cpu(ip->hw_id); if (amdgpu_discovery_validate_ip(adev, inst, hw_id)) @@ -711,13 +747,14 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, uint32_t *vcn_harvest_count, uint32_t *umc_harvest_count) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct harvest_table *harvest_info; u16 offset; int i; uint32_t umc_harvest_config = 0; - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[HARVEST_INFO].offset); if (!offset) { @@ -725,7 +762,7 @@ static void amdgpu_discovery_read_from_harvest_table(struct amdgpu_device *adev, return; } - harvest_info = (struct harvest_table *)(adev->mman.discovery_bin + offset); + harvest_info = (struct harvest_table *)(discovery_bin + offset); for (i = 0; i < 32; i++) { if (le16_to_cpu(harvest_info->list[i].hw_id) == 0) @@ -1021,8 +1058,8 @@ static void ip_disc_release(struct kobject *kobj) kobj); struct amdgpu_device *adev = ip_top->adev; - adev->ip_top = NULL; kfree(ip_top); + adev->discovery.ip_top = NULL; } static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev, @@ -1033,7 +1070,9 @@ static uint8_t amdgpu_discovery_get_harvest_info(struct amdgpu_device *adev, /* Until a uniform way is figured, get mask based on hwid */ switch (hw_id) { case VCN_HWID: - harvest = ((1 << inst) & adev->vcn.inst_mask) == 0; + /* VCN vs UVD+VCE */ + if (!amdgpu_ip_version(adev, VCE_HWIP, 0)) + harvest = ((1 << inst) & adev->vcn.inst_mask) == 0; break; case DMU_HWID: if (adev->harvest_ip_mask & AMD_HARVEST_IP_DMU_MASK) @@ -1060,6 +1099,7 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev, const size_t _ip_offset, const int num_ips, bool reg_base_64) { + uint8_t *discovery_bin = adev->discovery.bin; int ii, jj, kk, res; uint16_t hw_id; uint8_t inst; @@ -1077,7 +1117,7 @@ static int amdgpu_discovery_sysfs_ips(struct amdgpu_device *adev, struct ip_v4 *ip; struct ip_hw_instance *ip_hw_instance; - ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset); + ip = (struct ip_v4 *)(discovery_bin + ip_offset); inst = ip->instance_number; hw_id = le16_to_cpu(ip->hw_id); if (amdgpu_discovery_validate_ip(adev, inst, hw_id) || @@ -1164,17 +1204,20 @@ next_ip: static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) { + struct ip_discovery_top *ip_top = adev->discovery.ip_top; + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; - struct kset *die_kset = &adev->ip_top->die_kset; + struct kset *die_kset = &ip_top->die_kset; u16 num_dies, die_offset, num_ips; size_t ip_offset; int ii, res; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); DRM_DEBUG("number of dies: %d\n", num_dies); @@ -1183,7 +1226,7 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) struct ip_die_entry *ip_die_entry; die_offset = le16_to_cpu(ihdr->die_info[ii].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); @@ -1217,30 +1260,32 @@ static int amdgpu_discovery_sysfs_recurse(struct amdgpu_device *adev) static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; + struct ip_discovery_top *ip_top; struct kset *die_kset; int res, ii; - if (!adev->mman.discovery_bin) + if (!discovery_bin) return -EINVAL; - adev->ip_top = kzalloc(sizeof(*adev->ip_top), GFP_KERNEL); - if (!adev->ip_top) + ip_top = kzalloc(sizeof(*ip_top), GFP_KERNEL); + if (!ip_top) return -ENOMEM; - adev->ip_top->adev = adev; - - res = kobject_init_and_add(&adev->ip_top->kobj, &ip_discovery_ktype, + ip_top->adev = adev; + adev->discovery.ip_top = ip_top; + res = kobject_init_and_add(&ip_top->kobj, &ip_discovery_ktype, &adev->dev->kobj, "ip_discovery"); if (res) { DRM_ERROR("Couldn't init and add ip_discovery/"); goto Err; } - die_kset = &adev->ip_top->die_kset; + die_kset = &ip_top->die_kset; kobject_set_name(&die_kset->kobj, "%s", "die"); - die_kset->kobj.parent = &adev->ip_top->kobj; + die_kset->kobj.parent = &ip_top->kobj; die_kset->kobj.ktype = &die_kobj_ktype; - res = kset_register(&adev->ip_top->die_kset); + res = kset_register(&ip_top->die_kset); if (res) { DRM_ERROR("Couldn't register die_kset"); goto Err; @@ -1254,7 +1299,7 @@ static int amdgpu_discovery_sysfs_init(struct amdgpu_device *adev) return res; Err: - kobject_put(&adev->ip_top->kobj); + kobject_put(&ip_top->kobj); return res; } @@ -1299,10 +1344,11 @@ static void amdgpu_discovery_sysfs_die_free(struct ip_die_entry *ip_die_entry) static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev) { + struct ip_discovery_top *ip_top = adev->discovery.ip_top; struct list_head *el, *tmp; struct kset *die_kset; - die_kset = &adev->ip_top->die_kset; + die_kset = &ip_top->die_kset; spin_lock(&die_kset->list_lock); list_for_each_prev_safe(el, tmp, &die_kset->list) { list_del_init(el); @@ -1311,8 +1357,8 @@ static void amdgpu_discovery_sysfs_fini(struct amdgpu_device *adev) spin_lock(&die_kset->list_lock); } spin_unlock(&die_kset->list_lock); - kobject_put(&adev->ip_top->die_kset.kobj); - kobject_put(&adev->ip_top->kobj); + kobject_put(&ip_top->die_kset.kobj); + kobject_put(&ip_top->kobj); } /* ================================================== */ @@ -1323,6 +1369,7 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) struct binary_header *bhdr; struct ip_discovery_header *ihdr; struct die_header *dhdr; + uint8_t *discovery_bin; struct ip_v4 *ip; uint16_t die_offset; uint16_t ip_offset; @@ -1338,22 +1385,23 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) r = amdgpu_discovery_init(adev); if (r) return r; - + discovery_bin = adev->discovery.bin; wafl_ver = 0; adev->gfx.xcc_mask = 0; adev->sdma.sdma_mask = 0; adev->vcn.inst_mask = 0; adev->jpeg.inst_mask = 0; - bhdr = (struct binary_header *)adev->mman.discovery_bin; - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); + bhdr = (struct binary_header *)discovery_bin; + ihdr = (struct ip_discovery_header + *)(discovery_bin + + le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset)); num_dies = le16_to_cpu(ihdr->num_dies); DRM_DEBUG("number of dies: %d\n", num_dies); for (i = 0; i < num_dies; i++) { die_offset = le16_to_cpu(ihdr->die_info[i].die_offset); - dhdr = (struct die_header *)(adev->mman.discovery_bin + die_offset); + dhdr = (struct die_header *)(discovery_bin + die_offset); num_ips = le16_to_cpu(dhdr->num_ips); ip_offset = die_offset + sizeof(*dhdr); @@ -1367,7 +1415,7 @@ static int amdgpu_discovery_reg_base_init(struct amdgpu_device *adev) le16_to_cpu(dhdr->die_id), num_ips); for (j = 0; j < num_ips; j++) { - ip = (struct ip_v4 *)(adev->mman.discovery_bin + ip_offset); + ip = (struct ip_v4 *)(discovery_bin + ip_offset); inst = ip->instance_number; hw_id = le16_to_cpu(ip->hw_id); @@ -1517,16 +1565,16 @@ next_ip: static void amdgpu_discovery_harvest_ip(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct ip_discovery_header *ihdr; struct binary_header *bhdr; int vcn_harvest_count = 0; int umc_harvest_count = 0; uint16_t offset, ihdr_ver; - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[IP_DISCOVERY].offset); - ihdr = (struct ip_discovery_header *)(adev->mman.discovery_bin + - offset); + ihdr = (struct ip_discovery_header *)(discovery_bin + offset); ihdr_ver = le16_to_cpu(ihdr->version); /* * Harvest table does not fit Navi1x and legacy GPUs, @@ -1573,22 +1621,23 @@ union gc_info { static int amdgpu_discovery_get_gfx_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union gc_info *gc_info; u16 offset; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[GC].offset); if (!offset) return 0; - gc_info = (union gc_info *)(adev->mman.discovery_bin + offset); + gc_info = (union gc_info *)(discovery_bin + offset); switch (le16_to_cpu(gc_info->v1.header.version_major)) { case 1: @@ -1681,24 +1730,25 @@ union mall_info { static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union mall_info *mall_info; u32 u, mall_size_per_umc, m_s_present, half_use; u64 mall_size; u16 offset; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[MALL_INFO].offset); if (!offset) return 0; - mall_info = (union mall_info *)(adev->mman.discovery_bin + offset); + mall_info = (union mall_info *)(discovery_bin + offset); switch (le16_to_cpu(mall_info->v1.header.version_major)) { case 1: @@ -1737,12 +1787,13 @@ union vcn_info { static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) { + uint8_t *discovery_bin = adev->discovery.bin; struct binary_header *bhdr; union vcn_info *vcn_info; u16 offset; int v; - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { DRM_ERROR("ip discovery uninitialized\n"); return -EINVAL; } @@ -1757,13 +1808,13 @@ static int amdgpu_discovery_get_vcn_info(struct amdgpu_device *adev) return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[VCN_INFO].offset); if (!offset) return 0; - vcn_info = (union vcn_info *)(adev->mman.discovery_bin + offset); + vcn_info = (union vcn_info *)(discovery_bin + offset); switch (le16_to_cpu(vcn_info->v1.header.version_major)) { case 1: @@ -1823,6 +1874,7 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev, struct amdgpu_gmc_memrange **ranges, int *range_cnt, bool refresh) { + uint8_t *discovery_bin = adev->discovery.bin; struct amdgpu_gmc_memrange *mem_ranges; struct binary_header *bhdr; union nps_info *nps_info; @@ -1839,13 +1891,13 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev, return r; nps_info = &nps_data; } else { - if (!adev->mman.discovery_bin) { + if (!discovery_bin) { dev_err(adev->dev, "fetch mem range failed, ip discovery uninitialized\n"); return -EINVAL; } - bhdr = (struct binary_header *)adev->mman.discovery_bin; + bhdr = (struct binary_header *)discovery_bin; offset = le16_to_cpu(bhdr->table_list[NPS_INFO].offset); if (!offset) @@ -1855,8 +1907,7 @@ int amdgpu_discovery_get_nps_info(struct amdgpu_device *adev, if (amdgpu_discovery_verify_npsinfo(adev, bhdr)) return -ENOENT; - nps_info = - (union nps_info *)(adev->mman.discovery_bin + offset); + nps_info = (union nps_info *)(discovery_bin + offset); } switch (le16_to_cpu(nps_info->v1.header.version_major)) { @@ -2359,6 +2410,21 @@ static int amdgpu_discovery_set_sdma_ip_blocks(struct amdgpu_device *adev) amdgpu_ip_version(adev, SDMA0_HWIP, 0)); return -EINVAL; } + + return 0; +} + +static int amdgpu_discovery_set_ras_ip_blocks(struct amdgpu_device *adev) +{ + switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 12): + case IP_VERSION(13, 0, 14): + amdgpu_device_ip_block_add(adev, &ras_v1_0_ip_block); + break; + default: + break; + } return 0; } @@ -2565,7 +2631,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 0, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 0, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 0); @@ -2592,7 +2660,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 3, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 3, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 0, 1); @@ -2619,8 +2689,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); vega10_reg_base_init(adev); adev->sdma.num_instances = 1; + adev->sdma.sdma_mask = 1; adev->vcn.num_vcn_inst = 1; adev->gmc.num_umc = 2; + adev->gfx.xcc_mask = 1; if (adev->apu_flags & AMD_APU_IS_RAVEN2) { adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 2, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 2, 0); @@ -2665,7 +2737,9 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); vega20_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; adev->gmc.num_umc = 8; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 0); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 0); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 0); @@ -2693,8 +2767,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); arct_reg_base_init(adev); adev->sdma.num_instances = 8; + adev->sdma.sdma_mask = 0xff; adev->vcn.num_vcn_inst = 2; adev->gmc.num_umc = 8; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 1); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 1); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 2, 1); @@ -2726,8 +2802,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) amdgpu_discovery_init(adev); aldebaran_reg_base_init(adev); adev->sdma.num_instances = 5; + adev->sdma.sdma_mask = 0x1f; adev->vcn.num_vcn_inst = 2; adev->gmc.num_umc = 4; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(9, 4, 2); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(9, 4, 2); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(4, 4, 0); @@ -2762,6 +2840,8 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) } else { cyan_skillfish_reg_base_init(adev); adev->sdma.num_instances = 2; + adev->sdma.sdma_mask = 3; + adev->gfx.xcc_mask = 1; adev->ip_versions[MMHUB_HWIP][0] = IP_VERSION(2, 0, 3); adev->ip_versions[ATHUB_HWIP][0] = IP_VERSION(2, 0, 3); adev->ip_versions[OSSSYS_HWIP][0] = IP_VERSION(5, 0, 1); @@ -3125,6 +3205,10 @@ int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev) if (r) return r; + r = amdgpu_discovery_set_ras_ip_blocks(adev); + if (r) + return r; + if ((adev->firmware.load_type == AMDGPU_FW_LOAD_DIRECT && !amdgpu_sriov_vf(adev)) || (adev->firmware.load_type == AMDGPU_FW_LOAD_RLC_BACKDOOR_AUTO && amdgpu_dpm == 1)) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h index b44d56465c5b..4ce04486cc31 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.h @@ -24,9 +24,21 @@ #ifndef __AMDGPU_DISCOVERY__ #define __AMDGPU_DISCOVERY__ +#include <linux/debugfs.h> + #define DISCOVERY_TMR_SIZE (10 << 10) #define DISCOVERY_TMR_OFFSET (64 << 10) +struct ip_discovery_top; + +struct amdgpu_discovery_info { + struct debugfs_blob_wrapper debugfs_blob; + struct ip_discovery_top *ip_top; + uint32_t size; + uint8_t *bin; + bool reserve_tmr; +}; + void amdgpu_discovery_fini(struct amdgpu_device *adev); int amdgpu_discovery_set_ip_blocks(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c index 51bab32fd8c6..b5d34797d606 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c @@ -332,8 +332,6 @@ int amdgpu_display_crtc_set_config(struct drm_mode_set *set, if (crtc->enabled) active = true; - pm_runtime_mark_last_busy(dev->dev); - adev = drm_to_adev(dev); /* if we have active crtcs and we don't have a power ref, * take the current one @@ -1365,6 +1363,64 @@ static const struct drm_prop_enum_list amdgpu_dither_enum_list[] = { { AMDGPU_FMT_DITHER_ENABLE, "on" }, }; +/** + * DOC: property for adaptive backlight modulation + * + * The 'adaptive backlight modulation' property is used for the compositor to + * directly control the adaptive backlight modulation power savings feature + * that is part of DCN hardware. + * + * The property will be attached specifically to eDP panels that support it. + * + * The property is by default set to 'sysfs' to allow the sysfs file 'panel_power_savings' + * to be able to control it. + * If set to 'off' the compositor will ensure it stays off. + * The other values 'min', 'bias min', 'bias max', and 'max' will control the + * intensity of the power savings. + * + * Modifying this value can have implications on color accuracy, so tread + * carefully. + */ +static int amdgpu_display_setup_abm_prop(struct amdgpu_device *adev) +{ + const struct drm_prop_enum_list props[] = { + { ABM_SYSFS_CONTROL, "sysfs" }, + { ABM_LEVEL_OFF, "off" }, + { ABM_LEVEL_MIN, "min" }, + { ABM_LEVEL_BIAS_MIN, "bias min" }, + { ABM_LEVEL_BIAS_MAX, "bias max" }, + { ABM_LEVEL_MAX, "max" }, + }; + struct drm_property *prop; + int i; + + if (!adev->dc_enabled) + return 0; + + prop = drm_property_create(adev_to_drm(adev), DRM_MODE_PROP_ENUM, + "adaptive backlight modulation", + 6); + if (!prop) + return -ENOMEM; + + for (i = 0; i < ARRAY_SIZE(props); i++) { + int ret; + + ret = drm_property_add_enum(prop, props[i].type, + props[i].name); + + if (ret) { + drm_property_destroy(adev_to_drm(adev), prop); + + return ret; + } + } + + adev->mode_info.abm_level_property = prop; + + return 0; +} + int amdgpu_display_modeset_create_props(struct amdgpu_device *adev) { int sz; @@ -1411,7 +1467,7 @@ int amdgpu_display_modeset_create_props(struct amdgpu_device *adev) "dither", amdgpu_dither_enum_list, sz); - return 0; + return amdgpu_display_setup_abm_prop(adev); } void amdgpu_display_update_priority(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h index 930c171473b4..49a29bf47a37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.h @@ -55,4 +55,11 @@ int amdgpu_display_resume_helper(struct amdgpu_device *adev); int amdgpu_display_get_scanout_buffer(struct drm_plane *plane, struct drm_scanout_buffer *sb); +#define ABM_SYSFS_CONTROL -1 +#define ABM_LEVEL_OFF 0 +#define ABM_LEVEL_MIN 1 +#define ABM_LEVEL_BIAS_MIN 2 +#define ABM_LEVEL_BIAS_MAX 3 +#define ABM_LEVEL_MAX 4 + #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 8561ad7f6180..e22cfa7c6d32 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -81,13 +81,44 @@ static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf, struct drm_gem_object *obj = dmabuf->priv; struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); + int r; + + /* + * Disable peer-to-peer access for DCC-enabled VRAM surfaces on GFX12+. + * Such buffers cannot be safely accessed over P2P due to device-local + * compression metadata. Fallback to system-memory path instead. + * Device supports GFX12 (GC 12.x or newer) + * BO was created with the AMDGPU_GEM_CREATE_GFX12_DCC flag + * + */ + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0) && + bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC) + attach->peer2peer = false; + + /* + * Disable peer-to-peer access for DCC-enabled VRAM surfaces on GFX12+. + * Such buffers cannot be safely accessed over P2P due to device-local + * compression metadata. Fallback to system-memory path instead. + * Device supports GFX12 (GC 12.x or newer) + * BO was created with the AMDGPU_GEM_CREATE_GFX12_DCC flag + * + */ + if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0) && + bo->flags & AMDGPU_GEM_CREATE_GFX12_DCC) + attach->peer2peer = false; if (!amdgpu_dmabuf_is_xgmi_accessible(attach_adev, bo) && pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0) attach->peer2peer = false; + r = dma_resv_lock(bo->tbo.base.resv, NULL); + if (r) + return r; + amdgpu_vm_bo_update_shared(bo); + dma_resv_unlock(bo->tbo.base.resv); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index bff25ef3e2d0..2dfbddcef9ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -144,7 +144,8 @@ enum AMDGPU_DEBUG_MASK { AMDGPU_DEBUG_DISABLE_GPU_RING_RESET = BIT(6), AMDGPU_DEBUG_SMU_POOL = BIT(7), AMDGPU_DEBUG_VM_USERPTR = BIT(8), - AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9) + AMDGPU_DEBUG_DISABLE_RAS_CE_LOG = BIT(9), + AMDGPU_DEBUG_ENABLE_CE_CS = BIT(10) }; unsigned int amdgpu_vram_limit = UINT_MAX; @@ -311,7 +312,7 @@ module_param_named(moverate, amdgpu_moverate, int, 0600); * DOC: audio (int) * Set HDMI/DPAudio. Only affects non-DC display handling. The default is -1 (Enabled), set 0 to disabled it. */ -MODULE_PARM_DESC(audio, "Audio enable (-1 = auto, 0 = disable, 1 = enable)"); +MODULE_PARM_DESC(audio, "HDMI/DP Audio enable for non DC displays (-1 = auto, 0 = disable, 1 = enable)"); module_param_named(audio, amdgpu_audio, int, 0444); /** @@ -353,22 +354,16 @@ module_param_named(svm_default_granularity, amdgpu_svm_default_granularity, uint * DOC: lockup_timeout (string) * Set GPU scheduler timeout value in ms. * - * The format can be [Non-Compute] or [GFX,Compute,SDMA,Video]. That is there can be one or - * multiple values specified. 0 and negative values are invalidated. They will be adjusted - * to the default timeout. + * The format can be [single value] for setting all timeouts at once or + * [GFX,Compute,SDMA,Video] to set individual timeouts. + * Negative values mean infinity. * - * - With one value specified, the setting will apply to all non-compute jobs. - * - With multiple values specified, the first one will be for GFX. - * The second one is for Compute. The third and fourth ones are - * for SDMA and Video. - * - * By default(with no lockup_timeout settings), the timeout for all jobs is 10000. + * By default(with no lockup_timeout settings), the timeout for all queues is 2000. */ MODULE_PARM_DESC(lockup_timeout, - "GPU lockup timeout in ms (default: 10000 for all jobs. " - "0: keep default value. negative: infinity timeout), format: for bare metal [Non-Compute] or [GFX,Compute,SDMA,Video]; " - "for passthrough or sriov [all jobs] or [GFX,Compute,SDMA,Video]."); -module_param_string(lockup_timeout, amdgpu_lockup_timeout, sizeof(amdgpu_lockup_timeout), 0444); + "GPU lockup timeout in ms (default: 2000. 0: keep default value. negative: infinity timeout), format: [single value for all] or [GFX,Compute,SDMA,Video]."); +module_param_string(lockup_timeout, amdgpu_lockup_timeout, + sizeof(amdgpu_lockup_timeout), 0444); /** * DOC: dpm (int) @@ -623,39 +618,39 @@ module_param_named(timeout_period, amdgpu_watchdog_timer.period, uint, 0644); /** * DOC: si_support (int) - * Set SI support driver. This parameter works after set config CONFIG_DRM_AMDGPU_SI. For SI asic, when radeon driver is enabled, - * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available, - * otherwise using amdgpu driver. - */ + * 1 = enabled, 0 = disabled, -1 = default + * + * SI (Southern Islands) are first generation GCN GPUs, supported by both + * drivers: radeon (old) and amdgpu (new). This parameter controls whether + * amdgpu should support SI. + * By default, SI dedicated GPUs are supported by amdgpu. + * Only relevant when CONFIG_DRM_AMDGPU_SI is enabled to build SI support in amdgpu. + * See also radeon.si_support which should be disabled when amdgpu.si_support is + * enabled, and vice versa. + */ +int amdgpu_si_support = -1; #ifdef CONFIG_DRM_AMDGPU_SI - -#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE) -int amdgpu_si_support; -MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled (default))"); -#else -int amdgpu_si_support = 1; -MODULE_PARM_DESC(si_support, "SI support (1 = enabled (default), 0 = disabled)"); -#endif - +MODULE_PARM_DESC(si_support, "SI support (1 = enabled, 0 = disabled, -1 = default)"); module_param_named(si_support, amdgpu_si_support, int, 0444); #endif /** * DOC: cik_support (int) - * Set CIK support driver. This parameter works after set config CONFIG_DRM_AMDGPU_CIK. For CIK asic, when radeon driver is enabled, - * set value 0 to use radeon driver, while set value 1 to use amdgpu driver. The default is using radeon driver when it available, - * otherwise using amdgpu driver. - */ + * 1 = enabled, 0 = disabled, -1 = default + * + * CIK (Sea Islands) are second generation GCN GPUs, supported by both + * drivers: radeon (old) and amdgpu (new). This parameter controls whether + * amdgpu should support CIK. + * By default: + * - CIK dedicated GPUs are supported by amdgpu. + * - CIK APUs are supported by radeon (except when radeon is not built). + * Only relevant when CONFIG_DRM_AMDGPU_CIK is enabled to build CIK support in amdgpu. + * See also radeon.cik_support which should be disabled when amdgpu.cik_support is + * enabled, and vice versa. + */ +int amdgpu_cik_support = -1; #ifdef CONFIG_DRM_AMDGPU_CIK - -#if IS_ENABLED(CONFIG_DRM_RADEON) || IS_ENABLED(CONFIG_DRM_RADEON_MODULE) -int amdgpu_cik_support; -MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled (default))"); -#else -int amdgpu_cik_support = 1; -MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled (default), 0 = disabled)"); -#endif - +MODULE_PARM_DESC(cik_support, "CIK support (1 = enabled, 0 = disabled, -1 = default)"); module_param_named(cik_support, amdgpu_cik_support, int, 0444); #endif @@ -2233,7 +2228,6 @@ static void amdgpu_get_secondary_funcs(struct amdgpu_device *adev) adev->pdev->bus->number, i); if (p) { pm_runtime_get_sync(&p->dev); - pm_runtime_mark_last_busy(&p->dev); pm_runtime_put_autosuspend(&p->dev); pci_dev_put(p); } @@ -2289,6 +2283,11 @@ static void amdgpu_init_debug_options(struct amdgpu_device *adev) pr_info("debug: disable kernel logs of correctable errors\n"); adev->debug_disable_ce_logs = true; } + + if (amdgpu_debug_mask & AMDGPU_DEBUG_ENABLE_CE_CS) { + pr_info("debug: allowing command submission to CE engine\n"); + adev->debug_enable_ce_cs = true; + } } static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long flags) @@ -2307,6 +2306,72 @@ static unsigned long amdgpu_fix_asic_type(struct pci_dev *pdev, unsigned long fl return flags; } +static bool amdgpu_support_enabled(struct device *dev, + const enum amd_asic_type family) +{ + const char *gen; + const char *param; + int module_param = -1; + bool radeon_support_built = IS_ENABLED(CONFIG_DRM_RADEON); + bool amdgpu_support_built = false; + bool support_by_default = false; + + switch (family) { + case CHIP_TAHITI: + case CHIP_PITCAIRN: + case CHIP_VERDE: + case CHIP_OLAND: + case CHIP_HAINAN: + gen = "SI"; + param = "si_support"; + module_param = amdgpu_si_support; + amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_SI); + support_by_default = true; + break; + + case CHIP_BONAIRE: + case CHIP_HAWAII: + support_by_default = true; + fallthrough; + case CHIP_KAVERI: + case CHIP_KABINI: + case CHIP_MULLINS: + gen = "CIK"; + param = "cik_support"; + module_param = amdgpu_cik_support; + amdgpu_support_built = IS_ENABLED(CONFIG_DRM_AMDGPU_CIK); + break; + + default: + /* All other chips are supported by amdgpu only */ + return true; + } + + if (!amdgpu_support_built) { + dev_info(dev, "amdgpu built without %s support\n", gen); + return false; + } + + if ((module_param == -1 && (support_by_default || !radeon_support_built)) || + module_param == 1) { + if (radeon_support_built) + dev_info(dev, "%s support provided by amdgpu.\n" + "Use radeon.%s=1 amdgpu.%s=0 to override.\n", + gen, param, param); + + return true; + } + + if (radeon_support_built) + dev_info(dev, "%s support provided by radeon.\n" + "Use radeon.%s=0 amdgpu.%s=1 to override.\n", + gen, param, param); + else if (module_param == 0) + dev_info(dev, "%s support disabled by module param\n", gen); + + return false; +} + static int amdgpu_pci_probe(struct pci_dev *pdev, const struct pci_device_id *ent) { @@ -2354,48 +2419,8 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, return -ENOTSUPP; } - switch (flags & AMD_ASIC_MASK) { - case CHIP_TAHITI: - case CHIP_PITCAIRN: - case CHIP_VERDE: - case CHIP_OLAND: - case CHIP_HAINAN: -#ifdef CONFIG_DRM_AMDGPU_SI - if (!amdgpu_si_support) { - dev_info(&pdev->dev, - "SI support provided by radeon.\n"); - dev_info(&pdev->dev, - "Use radeon.si_support=0 amdgpu.si_support=1 to override.\n" - ); - return -ENODEV; - } - break; -#else - dev_info(&pdev->dev, "amdgpu is built without SI support.\n"); - return -ENODEV; -#endif - case CHIP_KAVERI: - case CHIP_BONAIRE: - case CHIP_HAWAII: - case CHIP_KABINI: - case CHIP_MULLINS: -#ifdef CONFIG_DRM_AMDGPU_CIK - if (!amdgpu_cik_support) { - dev_info(&pdev->dev, - "CIK support provided by radeon.\n"); - dev_info(&pdev->dev, - "Use radeon.cik_support=0 amdgpu.cik_support=1 to override.\n" - ); - return -ENODEV; - } - break; -#else - dev_info(&pdev->dev, "amdgpu is built without CIK support.\n"); + if (!amdgpu_support_enabled(&pdev->dev, flags & AMD_ASIC_MASK)) return -ENODEV; -#endif - default: - break; - } adev = devm_drm_dev_alloc(&pdev->dev, &amdgpu_kms_driver, typeof(*adev), ddev); if (IS_ERR(adev)) @@ -2474,7 +2499,6 @@ retry_init: pm_runtime_allow(ddev->dev); - pm_runtime_mark_last_busy(ddev->dev); pm_runtime_put_autosuspend(ddev->dev); pci_wake_from_d3(pdev, TRUE); @@ -2558,7 +2582,8 @@ amdgpu_pci_shutdown(struct pci_dev *pdev) */ if (!amdgpu_passthrough(adev)) adev->mp1_state = PP_MP1_STATE_UNLOAD; - amdgpu_device_ip_suspend(adev); + amdgpu_device_prepare(dev); + amdgpu_device_suspend(dev, true); adev->mp1_state = PP_MP1_STATE_NONE; } @@ -2626,9 +2651,14 @@ static int amdgpu_pmops_suspend_noirq(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); struct amdgpu_device *adev = drm_to_adev(drm_dev); + int r; - if (amdgpu_acpi_should_gpu_reset(adev)) - return amdgpu_asic_reset(adev); + if (amdgpu_acpi_should_gpu_reset(adev)) { + amdgpu_device_lock_reset_domain(adev->reset_domain); + r = amdgpu_asic_reset(adev); + amdgpu_device_unlock_reset_domain(adev->reset_domain); + return r; + } return 0; } @@ -2771,22 +2801,8 @@ static int amdgpu_runtime_idle_check_userq(struct device *dev) struct pci_dev *pdev = to_pci_dev(dev); struct drm_device *drm_dev = pci_get_drvdata(pdev); struct amdgpu_device *adev = drm_to_adev(drm_dev); - struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; - int ret = 0; - - mutex_lock(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - ret = -EBUSY; - goto done; - } - } -done: - mutex_unlock(&adev->userq_mutex); - return ret; + return xa_empty(&adev->userq_doorbell_xa) ? 0 : -EBUSY; } static int amdgpu_pmops_runtime_suspend(struct device *dev) @@ -2933,7 +2949,6 @@ static int amdgpu_pmops_runtime_idle(struct device *dev) ret = amdgpu_runtime_idle_check_userq(dev); done: - pm_runtime_mark_last_busy(dev); pm_runtime_autosuspend(dev); return ret; } @@ -2969,7 +2984,6 @@ long amdgpu_drm_ioctl(struct file *filp, ret = drm_ioctl(filp, cmd, arg); - pm_runtime_mark_last_busy(dev->dev); out: pm_runtime_put_autosuspend(dev->dev); return ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c index fd8cca241da6..c7843e336310 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c @@ -45,16 +45,11 @@ * Cast helper */ static const struct dma_fence_ops amdgpu_fence_ops; -static const struct dma_fence_ops amdgpu_job_fence_ops; static inline struct amdgpu_fence *to_amdgpu_fence(struct dma_fence *f) { struct amdgpu_fence *__f = container_of(f, struct amdgpu_fence, base); - if (__f->base.ops == &amdgpu_fence_ops || - __f->base.ops == &amdgpu_job_fence_ops) - return __f; - - return NULL; + return __f; } /** @@ -98,51 +93,32 @@ static u32 amdgpu_fence_read(struct amdgpu_ring *ring) * amdgpu_fence_emit - emit a fence on the requested ring * * @ring: ring the fence is associated with - * @f: resulting fence object * @af: amdgpu fence input * @flags: flags to pass into the subordinate .emit_fence() call * * Emits a fence command on the requested ring (all asics). * Returns 0 on success, -ENOMEM on failure. */ -int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, - struct amdgpu_fence *af, unsigned int flags) +int amdgpu_fence_emit(struct amdgpu_ring *ring, struct amdgpu_fence *af, + unsigned int flags) { struct amdgpu_device *adev = ring->adev; struct dma_fence *fence; - struct amdgpu_fence *am_fence; struct dma_fence __rcu **ptr; uint32_t seq; int r; - if (!af) { - /* create a separate hw fence */ - am_fence = kzalloc(sizeof(*am_fence), GFP_KERNEL); - if (!am_fence) - return -ENOMEM; - } else { - am_fence = af; - } - fence = &am_fence->base; - am_fence->ring = ring; + fence = &af->base; + af->ring = ring; seq = ++ring->fence_drv.sync_seq; - am_fence->seq = seq; - if (af) { - dma_fence_init(fence, &amdgpu_job_fence_ops, - &ring->fence_drv.lock, - adev->fence_context + ring->idx, seq); - /* Against remove in amdgpu_job_{free, free_cb} */ - dma_fence_get(fence); - } else { - dma_fence_init(fence, &amdgpu_fence_ops, - &ring->fence_drv.lock, - adev->fence_context + ring->idx, seq); - } + dma_fence_init(fence, &amdgpu_fence_ops, + &ring->fence_drv.lock, + adev->fence_context + ring->idx, seq); amdgpu_ring_emit_fence(ring, ring->fence_drv.gpu_addr, seq, flags | AMDGPU_FENCE_FLAG_INT); - amdgpu_fence_save_wptr(fence); + amdgpu_fence_save_wptr(af); pm_runtime_get_noresume(adev_to_drm(adev)->dev); ptr = &ring->fence_drv.fences[seq & ring->fence_drv.num_fences_mask]; if (unlikely(rcu_dereference_protected(*ptr, 1))) { @@ -167,8 +143,6 @@ int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, */ rcu_assign_pointer(*ptr, dma_fence_get(fence)); - *f = fence; - return 0; } @@ -276,7 +250,6 @@ bool amdgpu_fence_process(struct amdgpu_ring *ring) drv->signalled_wptr = am_fence->wptr; dma_fence_signal(fence); dma_fence_put(fence); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); } while (last_seq != seq); @@ -670,36 +643,6 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev) } /** - * amdgpu_fence_driver_clear_job_fences - clear job embedded fences of ring - * - * @ring: fence of the ring to be cleared - * - */ -void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring) -{ - int i; - struct dma_fence *old, **ptr; - - for (i = 0; i <= ring->fence_drv.num_fences_mask; i++) { - ptr = &ring->fence_drv.fences[i]; - old = rcu_dereference_protected(*ptr, 1); - if (old && old->ops == &amdgpu_job_fence_ops) { - struct amdgpu_job *job; - - /* For non-scheduler bad job, i.e. failed ib test, we need to signal - * it right here or we won't be able to track them in fence_drv - * and they will remain unsignaled during sa_bo free. - */ - job = container_of(old, struct amdgpu_job, hw_fence.base); - if (!job->base.s_fence && !dma_fence_is_signaled(old)) - dma_fence_signal(old); - RCU_INIT_POINTER(*ptr, NULL); - dma_fence_put(old); - } - } -} - -/** * amdgpu_fence_driver_set_error - set error code on fences * @ring: the ring which contains the fences * @error: the error code to set @@ -755,21 +698,50 @@ void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring) /** * amdgpu_fence_driver_guilty_force_completion - force signal of specified sequence * - * @fence: fence of the ring to signal + * @af: fence of the ring to signal * */ -void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence) +void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *af) { - dma_fence_set_error(&fence->base, -ETIME); - amdgpu_fence_write(fence->ring, fence->seq); - amdgpu_fence_process(fence->ring); + struct dma_fence *unprocessed; + struct dma_fence __rcu **ptr; + struct amdgpu_fence *fence; + struct amdgpu_ring *ring = af->ring; + unsigned long flags; + u32 seq, last_seq; + + last_seq = amdgpu_fence_read(ring) & ring->fence_drv.num_fences_mask; + seq = ring->fence_drv.sync_seq & ring->fence_drv.num_fences_mask; + + /* mark all fences from the guilty context with an error */ + spin_lock_irqsave(&ring->fence_drv.lock, flags); + do { + last_seq++; + last_seq &= ring->fence_drv.num_fences_mask; + + ptr = &ring->fence_drv.fences[last_seq]; + rcu_read_lock(); + unprocessed = rcu_dereference(*ptr); + + if (unprocessed && !dma_fence_is_signaled_locked(unprocessed)) { + fence = container_of(unprocessed, struct amdgpu_fence, base); + + if (fence == af) + dma_fence_set_error(&fence->base, -ETIME); + else if (fence->context == af->context) + dma_fence_set_error(&fence->base, -ECANCELED); + } + rcu_read_unlock(); + } while (last_seq != seq); + spin_unlock_irqrestore(&ring->fence_drv.lock, flags); + /* signal the guilty fence */ + amdgpu_fence_write(ring, (u32)af->base.seqno); + amdgpu_fence_process(ring); } -void amdgpu_fence_save_wptr(struct dma_fence *fence) +void amdgpu_fence_save_wptr(struct amdgpu_fence *af) { - struct amdgpu_fence *am_fence = container_of(fence, struct amdgpu_fence, base); - - am_fence->wptr = am_fence->ring->wptr; + af->wptr = af->ring->wptr; } static void amdgpu_ring_backup_unprocessed_command(struct amdgpu_ring *ring, @@ -790,14 +762,19 @@ void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring, struct dma_fence *unprocessed; struct dma_fence __rcu **ptr; struct amdgpu_fence *fence; - u64 wptr, i, seqno; + u64 wptr; + u32 seq, last_seq; - seqno = amdgpu_fence_read(ring); + last_seq = amdgpu_fence_read(ring) & ring->fence_drv.num_fences_mask; + seq = ring->fence_drv.sync_seq & ring->fence_drv.num_fences_mask; wptr = ring->fence_drv.signalled_wptr; ring->ring_backup_entries_to_copy = 0; - for (i = seqno + 1; i <= ring->fence_drv.sync_seq; ++i) { - ptr = &ring->fence_drv.fences[i & ring->fence_drv.num_fences_mask]; + do { + last_seq++; + last_seq &= ring->fence_drv.num_fences_mask; + + ptr = &ring->fence_drv.fences[last_seq]; rcu_read_lock(); unprocessed = rcu_dereference(*ptr); @@ -813,7 +790,7 @@ void amdgpu_ring_backup_unprocessed_commands(struct amdgpu_ring *ring, wptr = fence->wptr; } rcu_read_unlock(); - } + } while (last_seq != seq); } /* @@ -830,13 +807,6 @@ static const char *amdgpu_fence_get_timeline_name(struct dma_fence *f) return (const char *)to_amdgpu_fence(f)->ring->name; } -static const char *amdgpu_job_fence_get_timeline_name(struct dma_fence *f) -{ - struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base); - - return (const char *)to_amdgpu_ring(job->base.sched)->name; -} - /** * amdgpu_fence_enable_signaling - enable signalling on fence * @f: fence @@ -854,23 +824,6 @@ static bool amdgpu_fence_enable_signaling(struct dma_fence *f) } /** - * amdgpu_job_fence_enable_signaling - enable signalling on job fence - * @f: fence - * - * This is the simliar function with amdgpu_fence_enable_signaling above, it - * only handles the job embedded fence. - */ -static bool amdgpu_job_fence_enable_signaling(struct dma_fence *f) -{ - struct amdgpu_job *job = container_of(f, struct amdgpu_job, hw_fence.base); - - if (!timer_pending(&to_amdgpu_ring(job->base.sched)->fence_drv.fallback_timer)) - amdgpu_fence_schedule_fallback(to_amdgpu_ring(job->base.sched)); - - return true; -} - -/** * amdgpu_fence_free - free up the fence memory * * @rcu: RCU callback head @@ -886,21 +839,6 @@ static void amdgpu_fence_free(struct rcu_head *rcu) } /** - * amdgpu_job_fence_free - free up the job with embedded fence - * - * @rcu: RCU callback head - * - * Free up the job with embedded fence after the RCU grace period. - */ -static void amdgpu_job_fence_free(struct rcu_head *rcu) -{ - struct dma_fence *f = container_of(rcu, struct dma_fence, rcu); - - /* free job if fence has a parent job */ - kfree(container_of(f, struct amdgpu_job, hw_fence.base)); -} - -/** * amdgpu_fence_release - callback that fence can be freed * * @f: fence @@ -913,19 +851,6 @@ static void amdgpu_fence_release(struct dma_fence *f) call_rcu(&f->rcu, amdgpu_fence_free); } -/** - * amdgpu_job_fence_release - callback that job embedded fence can be freed - * - * @f: fence - * - * This is the simliar function with amdgpu_fence_release above, it - * only handles the job embedded fence. - */ -static void amdgpu_job_fence_release(struct dma_fence *f) -{ - call_rcu(&f->rcu, amdgpu_job_fence_free); -} - static const struct dma_fence_ops amdgpu_fence_ops = { .get_driver_name = amdgpu_fence_get_driver_name, .get_timeline_name = amdgpu_fence_get_timeline_name, @@ -933,13 +858,6 @@ static const struct dma_fence_ops amdgpu_fence_ops = { .release = amdgpu_fence_release, }; -static const struct dma_fence_ops amdgpu_job_fence_ops = { - .get_driver_name = amdgpu_fence_get_driver_name, - .get_timeline_name = amdgpu_job_fence_get_timeline_name, - .enable_signaling = amdgpu_job_fence_enable_signaling, - .release = amdgpu_job_fence_release, -}; - /* * Fence debugfs */ @@ -1009,7 +927,6 @@ static int gpu_recover_get(void *data, u64 *val) *val = atomic_read(&adev->reset_domain->reset_res); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c index b2033f8352f5..d2237ce9da70 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.c @@ -302,7 +302,6 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, int pages) { unsigned t; - unsigned p; int i, j; u64 page_base; /* Starting from VEGA10, system bit must be 0 to mean invalid. */ @@ -316,8 +315,7 @@ void amdgpu_gart_unbind(struct amdgpu_device *adev, uint64_t offset, return; t = offset / AMDGPU_GPU_PAGE_SIZE; - p = t / AMDGPU_GPU_PAGES_IN_CPU_PAGE; - for (i = 0; i < pages; i++, p++) { + for (i = 0; i < pages; i++) { page_base = adev->dummy_page_addr; if (!adev->gart.ptr) continue; @@ -370,6 +368,42 @@ void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, } /** + * amdgpu_gart_map_vram_range - map VRAM pages into the GART page table + * + * @adev: amdgpu_device pointer + * @pa: physical address of the first page to be mapped + * @start_page: first page to map in the GART aperture + * @num_pages: number of pages to be mapped + * @flags: page table entry flags + * @dst: CPU address of the GART table + * + * Binds a BO that is allocated in VRAM to the GART page table + * (all ASICs). + * + * Useful when a kernel BO is located in VRAM but + * needs to be accessed from the GART address space. + */ +void amdgpu_gart_map_vram_range(struct amdgpu_device *adev, uint64_t pa, + uint64_t start_page, uint64_t num_pages, + uint64_t flags, void *dst) +{ + u32 i, idx; + + /* The SYSTEM flag indicates the pages aren't in VRAM. */ + WARN_ON_ONCE(flags & AMDGPU_PTE_SYSTEM); + + if (!drm_dev_enter(adev_to_drm(adev), &idx)) + return; + + for (i = 0; i < num_pages; ++i) { + amdgpu_gmc_set_pte_pde(adev, adev->gart.ptr, + start_page + i, pa + AMDGPU_GPU_PAGE_SIZE * i, flags); + } + + drm_dev_exit(idx); +} + +/** * amdgpu_gart_bind - bind pages into the gart page table * * @adev: amdgpu_device pointer diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h index 7cc980bf4725..d3118275ddae 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gart.h @@ -64,5 +64,8 @@ void amdgpu_gart_map(struct amdgpu_device *adev, uint64_t offset, void *dst); void amdgpu_gart_bind(struct amdgpu_device *adev, uint64_t offset, int pages, dma_addr_t *dma_addr, uint64_t flags); +void amdgpu_gart_map_vram_range(struct amdgpu_device *adev, uint64_t pa, + uint64_t start_page, uint64_t num_pages, + uint64_t flags, void *dst); void amdgpu_gart_invalidate_tlb(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index b7ebae289bea..3e38c5db2987 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -198,7 +198,7 @@ static void amdgpu_gem_object_free(struct drm_gem_object *gobj) struct amdgpu_bo *aobj = gem_to_amdgpu_bo(gobj); amdgpu_hmm_unregister(aobj); - ttm_bo_put(&aobj->tbo); + ttm_bo_fini(&aobj->tbo); } int amdgpu_gem_object_create(struct amdgpu_device *adev, unsigned long size, @@ -531,7 +531,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, struct drm_amdgpu_gem_userptr *args = data; struct amdgpu_fpriv *fpriv = filp->driver_priv; struct drm_gem_object *gobj; - struct hmm_range *range; + struct amdgpu_hmm_range *range; struct amdgpu_bo *bo; uint32_t handle; int r; @@ -572,10 +572,14 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, goto release_object; if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) { - r = amdgpu_ttm_tt_get_user_pages(bo, &range); - if (r) + range = amdgpu_hmm_range_alloc(NULL); + if (unlikely(!range)) + return -ENOMEM; + r = amdgpu_ttm_tt_get_user_pages(bo, range); + if (r) { + amdgpu_hmm_range_free(range); goto release_object; - + } r = amdgpu_bo_reserve(bo, true); if (r) goto user_pages_done; @@ -597,8 +601,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, user_pages_done: if (args->flags & AMDGPU_GEM_USERPTR_VALIDATE) - amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm, range); - + amdgpu_hmm_range_free(range); release_object: drm_gem_object_put(gobj); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index ebe2b4c68b0f..8b118c53f351 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -33,6 +33,7 @@ #include "amdgpu_reset.h" #include "amdgpu_xcp.h" #include "amdgpu_xgmi.h" +#include "amdgpu_mes.h" #include "nvd.h" /* delay 0.1 second to enable gfx off feature */ @@ -1194,6 +1195,75 @@ failed_kiq_write: dev_err(adev->dev, "failed to write reg:%x\n", reg); } +int amdgpu_kiq_hdp_flush(struct amdgpu_device *adev) +{ + signed long r, cnt = 0; + unsigned long flags; + uint32_t seq; + struct amdgpu_kiq *kiq = &adev->gfx.kiq[0]; + struct amdgpu_ring *ring = &kiq->ring; + + if (amdgpu_device_skip_hw_access(adev)) + return 0; + + if (adev->enable_mes_kiq && adev->mes.ring[0].sched.ready) + return amdgpu_mes_hdp_flush(adev); + + if (!ring->funcs->emit_hdp_flush) { + return -EOPNOTSUPP; + } + + spin_lock_irqsave(&kiq->ring_lock, flags); + r = amdgpu_ring_alloc(ring, 32); + if (r) + goto failed_unlock; + + amdgpu_ring_emit_hdp_flush(ring); + r = amdgpu_fence_emit_polling(ring, &seq, MAX_KIQ_REG_WAIT); + if (r) + goto failed_undo; + + amdgpu_ring_commit(ring); + spin_unlock_irqrestore(&kiq->ring_lock, flags); + + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + + /* don't wait anymore for gpu reset case because this way may + * block gpu_recover() routine forever, e.g. this virt_kiq_rreg + * is triggered in TTM and ttm_bo_lock_delayed_workqueue() will + * never return if we keep waiting in virt_kiq_rreg, which cause + * gpu_recover() hang there. + * + * also don't wait anymore for IRQ context + * */ + if (r < 1 && (amdgpu_in_reset(adev) || in_interrupt())) + goto failed_kiq_hdp_flush; + + might_sleep(); + while (r < 1 && cnt++ < MAX_KIQ_REG_TRY) { + if (amdgpu_in_reset(adev)) + goto failed_kiq_hdp_flush; + + msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + } + + if (cnt > MAX_KIQ_REG_TRY) { + dev_err(adev->dev, "failed to flush HDP via KIQ timeout\n"); + return -ETIMEDOUT; + } + + return 0; + +failed_undo: + amdgpu_ring_undo(ring); +failed_unlock: + spin_unlock_irqrestore(&kiq->ring_lock, flags); +failed_kiq_hdp_flush: + dev_err(adev->dev, "failed to flush HDP via KIQ\n"); + return r < 0 ? r : -EIO; +} + int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev) { if (amdgpu_num_kcq == -1) { @@ -1600,7 +1670,6 @@ static ssize_t amdgpu_gfx_set_run_cleaner_shader(struct device *dev, ret = amdgpu_gfx_run_cleaner_shader(adev, value); - pm_runtime_mark_last_busy(ddev->dev); pm_runtime_put_autosuspend(ddev->dev); if (ret) @@ -2485,3 +2554,4 @@ void amdgpu_debugfs_compute_sched_mask_init(struct amdgpu_device *adev) &amdgpu_debugfs_compute_sched_mask_fops); #endif } + diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h index fb5f7a0ee029..efd61a1ccc66 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h @@ -615,6 +615,7 @@ int amdgpu_gfx_cp_ecc_error_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry); uint32_t amdgpu_kiq_rreg(struct amdgpu_device *adev, uint32_t reg, uint32_t xcc_id); void amdgpu_kiq_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t v, uint32_t xcc_id); +int amdgpu_kiq_hdp_flush(struct amdgpu_device *adev); int amdgpu_gfx_get_num_kcq(struct amdgpu_device *adev); void amdgpu_gfx_cp_init_microcode(struct amdgpu_device *adev, uint32_t ucode_id); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 9dcf51991b5b..869bceb0fe2c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -597,6 +597,9 @@ int amdgpu_gmc_allocate_vm_inv_eng(struct amdgpu_device *adev) /* reserve engine 5 for firmware */ if (adev->enable_mes) vm_inv_engs[i] &= ~(1 << 5); + /* reserve engine 6 for uni mes */ + if (adev->enable_uni_mes) + vm_inv_engs[i] &= ~(1 << 6); /* reserve mmhub engine 3 for firmware */ if (adev->enable_umsch_mm) vm_inv_engs[i] &= ~(1 << 3); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h index 55097ca10738..727342689d4b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h @@ -86,6 +86,11 @@ enum amdgpu_memory_partition { #define AMDGPU_MAX_MEM_RANGES 8 +#define AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY 0x80 +#define AMDGPU_GMC9_FAULT_SOURCE_DATA_READ 0x40 +#define AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE 0x20 +#define AMDGPU_GMC9_FAULT_SOURCE_DATA_EXE 0x10 + /* * GMC page fault information */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c index 0760e70402ec..895c1e4c6747 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gtt_mgr.c @@ -284,6 +284,7 @@ int amdgpu_gtt_mgr_init(struct amdgpu_device *adev, uint64_t gtt_size) ttm_resource_manager_init(man, &adev->mman.bdev, gtt_size); start = AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS; + start += amdgpu_vce_required_gart_pages(adev); size = (adev->gmc.gart_size >> PAGE_SHIFT) - start; drm_mm_init(&mgr->mm, start, size); spin_lock_init(&mgr->lock); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c index 6e02fb9ac2f6..5a60d69a3e1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.c @@ -66,3 +66,19 @@ void amdgpu_hdp_generic_flush(struct amdgpu_device *adev, 0); } } + +void amdgpu_hdp_invalidate(struct amdgpu_device *adev, struct amdgpu_ring *ring) +{ + if (adev->asic_funcs && adev->asic_funcs->invalidate_hdp) + adev->asic_funcs->invalidate_hdp(adev, ring); + else if (adev->hdp.funcs && adev->hdp.funcs->invalidate_hdp) + adev->hdp.funcs->invalidate_hdp(adev, ring); +} + +void amdgpu_hdp_flush(struct amdgpu_device *adev, struct amdgpu_ring *ring) +{ + if (adev->asic_funcs && adev->asic_funcs->flush_hdp) + adev->asic_funcs->flush_hdp(adev, ring); + else if (adev->hdp.funcs && adev->hdp.funcs->flush_hdp) + adev->hdp.funcs->flush_hdp(adev, ring); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h index 4cfd932b7e91..d9f488fa76b9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hdp.h @@ -46,4 +46,8 @@ struct amdgpu_hdp { int amdgpu_hdp_ras_sw_init(struct amdgpu_device *adev); void amdgpu_hdp_generic_flush(struct amdgpu_device *adev, struct amdgpu_ring *ring); +void amdgpu_hdp_invalidate(struct amdgpu_device *adev, + struct amdgpu_ring *ring); +void amdgpu_hdp_flush(struct amdgpu_device *adev, + struct amdgpu_ring *ring); #endif /* __AMDGPU_HDP_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c index 2c6a6b858112..90d26d820bac 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c @@ -168,17 +168,13 @@ void amdgpu_hmm_unregister(struct amdgpu_bo *bo) int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier, uint64_t start, uint64_t npages, bool readonly, void *owner, - struct hmm_range **phmm_range) + struct amdgpu_hmm_range *range) { - struct hmm_range *hmm_range; unsigned long end; unsigned long timeout; unsigned long *pfns; int r = 0; - - hmm_range = kzalloc(sizeof(*hmm_range), GFP_KERNEL); - if (unlikely(!hmm_range)) - return -ENOMEM; + struct hmm_range *hmm_range = &range->hmm_range; pfns = kvmalloc_array(npages, sizeof(*pfns), GFP_KERNEL); if (unlikely(!pfns)) { @@ -221,28 +217,77 @@ retry: hmm_range->start = start; hmm_range->hmm_pfns = pfns; - *phmm_range = hmm_range; - return 0; out_free_pfns: kvfree(pfns); + hmm_range->hmm_pfns = NULL; out_free_range: - kfree(hmm_range); - if (r == -EBUSY) r = -EAGAIN; return r; } -bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range) +/** + * amdgpu_hmm_range_valid - check if an HMM range is still valid + * @range: pointer to the &struct amdgpu_hmm_range to validate + * + * Determines whether the given HMM range @range is still valid by + * checking for invalidations via the MMU notifier sequence. This is + * typically used to verify that the range has not been invalidated + * by concurrent address space updates before it is accessed. + * + * Return: + * * true if @range is valid and can be used safely + * * false if @range is NULL or has been invalidated + */ +bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range) { - bool r; + if (!range) + return false; - r = mmu_interval_read_retry(hmm_range->notifier, - hmm_range->notifier_seq); - kvfree(hmm_range->hmm_pfns); - kfree(hmm_range); + return !mmu_interval_read_retry(range->hmm_range.notifier, + range->hmm_range.notifier_seq); +} - return r; +/** + * amdgpu_hmm_range_alloc - allocate and initialize an AMDGPU HMM range + * @bo: optional buffer object to associate with this HMM range + * + * Allocates memory for amdgpu_hmm_range and associates it with the @bo passed. + * The reference count of the @bo is incremented. + * + * Return: + * Pointer to a newly allocated struct amdgpu_hmm_range on success, + * or NULL if memory allocation fails. + */ +struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo) +{ + struct amdgpu_hmm_range *range; + + range = kzalloc(sizeof(*range), GFP_KERNEL); + if (!range) + return NULL; + + range->bo = amdgpu_bo_ref(bo); + return range; +} + +/** + * amdgpu_hmm_range_free - release an AMDGPU HMM range + * @range: pointer to the range object to free + * + * Releases all resources held by @range, including the associated + * hmm_pfns and the dropping reference of associated bo if any. + * + * Return: void + */ +void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range) +{ + if (!range) + return; + + kvfree(range->hmm_range.hmm_pfns); + amdgpu_bo_unref(&range->bo); + kfree(range); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h index 953e1d06de20..140bc9cd57b4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.h @@ -31,13 +31,20 @@ #include <linux/interval_tree.h> #include <linux/mmu_notifier.h> +struct amdgpu_hmm_range { + struct hmm_range hmm_range; + struct amdgpu_bo *bo; +}; + int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier, uint64_t start, uint64_t npages, bool readonly, void *owner, - struct hmm_range **phmm_range); -bool amdgpu_hmm_range_get_pages_done(struct hmm_range *hmm_range); + struct amdgpu_hmm_range *range); #if defined(CONFIG_HMM_MIRROR) +bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range); +struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo); +void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range); int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr); void amdgpu_hmm_unregister(struct amdgpu_bo *bo); #else @@ -47,7 +54,20 @@ static inline int amdgpu_hmm_register(struct amdgpu_bo *bo, unsigned long addr) "add CONFIG_ZONE_DEVICE=y in config file to fix this\n"); return -ENODEV; } + static inline void amdgpu_hmm_unregister(struct amdgpu_bo *bo) {} + +static inline bool amdgpu_hmm_range_valid(struct amdgpu_hmm_range *range) +{ + return false; +} + +static inline struct amdgpu_hmm_range *amdgpu_hmm_range_alloc(struct amdgpu_bo *bo) +{ + return NULL; +} + +static inline void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range) {} #endif #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 7d9bcb72e8dd..586a58facca1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -149,17 +149,19 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, if (job) { vm = job->vm; fence_ctx = job->base.s_fence ? - job->base.s_fence->scheduled.context : 0; + job->base.s_fence->finished.context : 0; shadow_va = job->shadow_va; csa_va = job->csa_va; gds_va = job->gds_va; init_shadow = job->init_shadow; - af = &job->hw_fence; + af = job->hw_fence; /* Save the context of the job for reset handling. * The driver needs this so it can skip the ring * contents for guilty contexts. */ - af->context = job->base.s_fence ? job->base.s_fence->finished.context : 0; + af->context = fence_ctx; + /* the vm fence is also part of the job's context */ + job->hw_vm_fence->context = fence_ctx; } else { vm = NULL; fence_ctx = 0; @@ -167,23 +169,28 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, csa_va = 0; gds_va = 0; init_shadow = false; - af = NULL; + af = kzalloc(sizeof(*af), GFP_ATOMIC); + if (!af) + return -ENOMEM; } if (!ring->sched.ready) { dev_err(adev->dev, "couldn't schedule ib on ring <%s>\n", ring->name); - return -EINVAL; + r = -EINVAL; + goto free_fence; } if (vm && !job->vmid) { dev_err(adev->dev, "VM IB without ID\n"); - return -EINVAL; + r = -EINVAL; + goto free_fence; } if ((ib->flags & AMDGPU_IB_FLAGS_SECURE) && (!ring->funcs->secure_submission_supported)) { dev_err(adev->dev, "secure submissions not supported on ring <%s>\n", ring->name); - return -EINVAL; + r = -EINVAL; + goto free_fence; } alloc_size = ring->funcs->emit_frame_size + num_ibs * @@ -192,7 +199,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, r = amdgpu_ring_alloc(ring, alloc_size); if (r) { dev_err(adev->dev, "scheduling IB failed (%d).\n", r); - return r; + goto free_fence; } need_ctx_switch = ring->current_ctx != fence_ctx; @@ -289,7 +296,7 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, amdgpu_ring_init_cond_exec(ring, ring->cond_exe_gpu_addr); } - r = amdgpu_fence_emit(ring, f, af, fence_flags); + r = amdgpu_fence_emit(ring, af, fence_flags); if (r) { dev_err(adev->dev, "failed to emit fence (%d)\n", r); if (job && job->vmid) @@ -297,6 +304,10 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, amdgpu_ring_undo(ring); return r; } + *f = &af->base; + /* get a ref for the job */ + if (job) + dma_fence_get(*f); if (ring->funcs->insert_end) ring->funcs->insert_end(ring); @@ -317,12 +328,17 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, * fence so we know what rings contents to backup * after we reset the queue. */ - amdgpu_fence_save_wptr(*f); + amdgpu_fence_save_wptr(af); amdgpu_ring_ib_end(ring); amdgpu_ring_commit(ring); return 0; + +free_fence: + if (!job) + kfree(af); + return r; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c index 3ef5bc95642c..9cab36322c16 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c @@ -201,58 +201,34 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_ring *ring, struct amdgpu_device *adev = ring->adev; unsigned vmhub = ring->vm_hub; struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub]; - struct dma_fence **fences; - unsigned i; + /* If anybody is waiting for a VMID let everybody wait for fairness */ if (!dma_fence_is_signaled(ring->vmid_wait)) { *fence = dma_fence_get(ring->vmid_wait); return 0; } - fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_NOWAIT); - if (!fences) - return -ENOMEM; - /* Check if we have an idle VMID */ - i = 0; - list_for_each_entry((*idle), &id_mgr->ids_lru, list) { + list_for_each_entry_reverse((*idle), &id_mgr->ids_lru, list) { /* Don't use per engine and per process VMID at the same time */ struct amdgpu_ring *r = adev->vm_manager.concurrent_flush ? NULL : ring; - fences[i] = amdgpu_sync_peek_fence(&(*idle)->active, r); - if (!fences[i]) - break; - ++i; + *fence = amdgpu_sync_peek_fence(&(*idle)->active, r); + if (!(*fence)) + return 0; } - /* If we can't find a idle VMID to use, wait till one becomes available */ - if (&(*idle)->list == &id_mgr->ids_lru) { - u64 fence_context = adev->vm_manager.fence_context + ring->idx; - unsigned seqno = ++adev->vm_manager.seqno[ring->idx]; - struct dma_fence_array *array; - unsigned j; - - *idle = NULL; - for (j = 0; j < i; ++j) - dma_fence_get(fences[j]); - - array = dma_fence_array_create(i, fences, fence_context, - seqno, true); - if (!array) { - for (j = 0; j < i; ++j) - dma_fence_put(fences[j]); - kfree(fences); - return -ENOMEM; - } - - *fence = dma_fence_get(&array->base); - dma_fence_put(ring->vmid_wait); - ring->vmid_wait = &array->base; - return 0; - } - kfree(fences); + /* + * If we can't find a idle VMID to use, wait on a fence from the least + * recently used in the hope that it will be available soon. + */ + *idle = NULL; + dma_fence_put(ring->vmid_wait); + ring->vmid_wait = dma_fence_get(*fence); + /* This is the reference we return */ + dma_fence_get(*fence); return 0; } @@ -313,7 +289,7 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm, * user of the VMID. */ r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished, - GFP_NOWAIT); + GFP_ATOMIC); if (r) return r; @@ -373,7 +349,7 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm, */ r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished, - GFP_NOWAIT); + GFP_ATOMIC); if (r) return r; @@ -426,7 +402,7 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring, /* Remember this submission as user of the VMID */ r = amdgpu_sync_fence(&id->active, &job->base.s_fence->finished, - GFP_NOWAIT); + GFP_ATOMIC); if (r) goto error; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c index 9cddbf50442a..37270c4dab8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_isp.c @@ -280,6 +280,8 @@ int isp_kernel_buffer_alloc(struct device *dev, u64 size, if (ret) return ret; + /* Ensure *bo is NULL so a new BO will be created */ + *bo = NULL; ret = amdgpu_bo_create_kernel(adev, size, ISP_MC_ADDR_ALIGN, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index d020a890a0ea..0a0dcbf0798d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -130,14 +130,12 @@ static enum drm_gpu_sched_stat amdgpu_job_timedout(struct drm_sched_job *s_job) } /* attempt a per ring reset */ - if (unlikely(adev->debug_disable_gpu_ring_reset)) { - dev_err(adev->dev, "Ring reset disabled by debug mask\n"); - } else if (amdgpu_gpu_recovery && - amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_PER_QUEUE) && - ring->funcs->reset) { + if (amdgpu_gpu_recovery && + amdgpu_ring_is_reset_type_supported(ring, AMDGPU_RESET_TYPE_PER_QUEUE) && + ring->funcs->reset) { dev_err(adev->dev, "Starting %s ring reset\n", s_job->sched->name); - r = amdgpu_ring_reset(ring, job->vmid, &job->hw_fence); + r = amdgpu_ring_reset(ring, job->vmid, job->hw_fence); if (!r) { atomic_inc(&ring->adev->gpu_reset_counter); dev_err(adev->dev, "Ring %s reset succeeded\n", @@ -186,6 +184,9 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm, unsigned int num_ibs, struct amdgpu_job **job, u64 drm_client_id) { + struct amdgpu_fence *af; + int r; + if (num_ibs == 0) return -EINVAL; @@ -193,6 +194,20 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm, if (!*job) return -ENOMEM; + af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL); + if (!af) { + r = -ENOMEM; + goto err_job; + } + (*job)->hw_fence = af; + + af = kzalloc(sizeof(struct amdgpu_fence), GFP_KERNEL); + if (!af) { + r = -ENOMEM; + goto err_fence; + } + (*job)->hw_vm_fence = af; + (*job)->vm = vm; amdgpu_sync_create(&(*job)->explicit_sync); @@ -204,6 +219,14 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm, return drm_sched_job_init(&(*job)->base, entity, 1, owner, drm_client_id); + +err_fence: + kfree((*job)->hw_fence); +err_job: + kfree(*job); + *job = NULL; + + return r; } int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, @@ -223,7 +246,10 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, if (r) { if (entity) drm_sched_job_cleanup(&(*job)->base); + kfree((*job)->hw_vm_fence); + kfree((*job)->hw_fence); kfree(*job); + *job = NULL; } return r; @@ -251,11 +277,11 @@ void amdgpu_job_free_resources(struct amdgpu_job *job) struct dma_fence *f; unsigned i; - /* Check if any fences where initialized */ + /* Check if any fences were initialized */ if (job->base.s_fence && job->base.s_fence->finished.ops) f = &job->base.s_fence->finished; - else if (job->hw_fence.base.ops) - f = &job->hw_fence.base; + else if (job->hw_fence && job->hw_fence->base.ops) + f = &job->hw_fence->base; else f = NULL; @@ -271,11 +297,16 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job) amdgpu_sync_free(&job->explicit_sync); - /* only put the hw fence if has embedded fence */ - if (!job->hw_fence.base.ops) - kfree(job); + if (job->hw_fence->base.ops) + dma_fence_put(&job->hw_fence->base); + else + kfree(job->hw_fence); + if (job->hw_vm_fence->base.ops) + dma_fence_put(&job->hw_vm_fence->base); else - dma_fence_put(&job->hw_fence.base); + kfree(job->hw_vm_fence); + + kfree(job); } void amdgpu_job_set_gang_leader(struct amdgpu_job *job, @@ -304,10 +335,16 @@ void amdgpu_job_free(struct amdgpu_job *job) if (job->gang_submit != &job->base.s_fence->scheduled) dma_fence_put(job->gang_submit); - if (!job->hw_fence.base.ops) - kfree(job); + if (job->hw_fence->base.ops) + dma_fence_put(&job->hw_fence->base); + else + kfree(job->hw_fence); + if (job->hw_vm_fence->base.ops) + dma_fence_put(&job->hw_vm_fence->base); else - dma_fence_put(&job->hw_fence.base); + kfree(job->hw_vm_fence); + + kfree(job); } struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h index 4a6487eb6cb5..7abf069d17d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h @@ -64,7 +64,8 @@ struct amdgpu_job { struct drm_sched_job base; struct amdgpu_vm *vm; struct amdgpu_sync explicit_sync; - struct amdgpu_fence hw_fence; + struct amdgpu_fence *hw_fence; + struct amdgpu_fence *hw_vm_fence; struct dma_fence *gang_submit; uint32_t preamble_status; uint32_t preemption_status; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c index 6b7d66b6d4cc..63ee6ba6a931 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c @@ -371,7 +371,7 @@ static int amdgpu_debugfs_jpeg_sched_mask_set(void *data, u64 val) for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { for (j = 0; j < adev->jpeg.num_jpeg_rings; ++j) { ring = &adev->jpeg.inst[i].ring_dec[j]; - if (val & (BIT_ULL(1) << ((i * adev->jpeg.num_jpeg_rings) + j))) + if (val & (BIT_ULL((i * adev->jpeg.num_jpeg_rings) + j))) ring->sched.ready = true; else ring->sched.ready = false; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index a9327472c651..6ee77f431d56 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -758,7 +758,8 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) ui64 = atomic64_read(&adev->num_vram_cpu_page_faults); return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0; case AMDGPU_INFO_VRAM_USAGE: - ui64 = ttm_resource_manager_usage(&adev->mman.vram_mgr.manager); + ui64 = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ? + ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) : 0; return copy_to_user(out, &ui64, min(size, 8u)) ? -EFAULT : 0; case AMDGPU_INFO_VIS_VRAM_USAGE: ui64 = amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr); @@ -804,8 +805,8 @@ int amdgpu_info_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) mem.vram.usable_heap_size = adev->gmc.real_vram_size - atomic64_read(&adev->vram_pin_size) - AMDGPU_VM_RESERVED_VRAM; - mem.vram.heap_usage = - ttm_resource_manager_usage(vram_man); + mem.vram.heap_usage = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ? + ttm_resource_manager_usage(vram_man) : 0; mem.vram.max_allocation = mem.vram.usable_heap_size * 3 / 4; mem.cpu_accessible_vram.total_heap_size = @@ -1470,7 +1471,6 @@ error_pasid: kfree(fpriv); out_suspend: - pm_runtime_mark_last_busy(dev->dev); pm_put: pm_runtime_put_autosuspend(dev->dev); @@ -1538,7 +1538,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, kfree(fpriv); file_priv->driver_priv = NULL; - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index 5bf9be073cdd..9c182ce501af 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -105,8 +105,8 @@ int amdgpu_mes_init(struct amdgpu_device *adev) spin_lock_init(&adev->mes.ring_lock[i]); adev->mes.total_max_queue = AMDGPU_FENCE_MES_QUEUE_ID_MASK; - adev->mes.vmid_mask_mmhub = 0xffffff00; - adev->mes.vmid_mask_gfxhub = adev->gfx.disable_kq ? 0xfffffffe : 0xffffff00; + adev->mes.vmid_mask_mmhub = 0xFF00; + adev->mes.vmid_mask_gfxhub = adev->gfx.disable_kq ? 0xFFFE : 0xFF00; num_pipes = adev->gfx.me.num_pipe_per_me * adev->gfx.me.num_me; if (num_pipes > AMDGPU_MES_MAX_GFX_PIPES) @@ -409,7 +409,7 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev, return -EINVAL; /* Clear the doorbell array before detection */ - memset(adev->mes.hung_queue_db_array_cpu_addr, 0, + memset(adev->mes.hung_queue_db_array_cpu_addr, AMDGPU_MES_INVALID_DB_OFFSET, adev->mes.hung_queue_db_array_size * sizeof(u32)); input.queue_type = queue_type; input.detect_only = detect_only; @@ -420,12 +420,17 @@ int amdgpu_mes_detect_and_reset_hung_queues(struct amdgpu_device *adev, dev_err(adev->dev, "failed to detect and reset\n"); } else { *hung_db_num = 0; - for (i = 0; i < adev->mes.hung_queue_db_array_size; i++) { + for (i = 0; i < adev->mes.hung_queue_hqd_info_offset; i++) { if (db_array[i] != AMDGPU_MES_INVALID_DB_OFFSET) { hung_db_array[i] = db_array[i]; *hung_db_num += 1; } } + + /* + * TODO: return HQD info for MES scheduled user compute queue reset cases + * stored in hung_db_array hqd info offset to full array size + */ } return r; @@ -523,6 +528,18 @@ error: return r; } +int amdgpu_mes_hdp_flush(struct amdgpu_device *adev) +{ + uint32_t hdp_flush_req_offset, hdp_flush_done_offset, ref_and_mask; + + hdp_flush_req_offset = adev->nbio.funcs->get_hdp_flush_req_offset(adev); + hdp_flush_done_offset = adev->nbio.funcs->get_hdp_flush_done_offset(adev); + ref_and_mask = adev->nbio.hdp_flush_reg->ref_and_mask_cp0; + + return amdgpu_mes_reg_write_reg_wait(adev, hdp_flush_req_offset, hdp_flush_done_offset, + ref_and_mask, ref_and_mask); +} + int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev, uint64_t process_context_addr, uint32_t spi_gdbg_per_vmid_cntl, @@ -686,14 +703,11 @@ out: bool amdgpu_mes_suspend_resume_all_supported(struct amdgpu_device *adev) { uint32_t mes_rev = adev->mes.sched_version & AMDGPU_MES_VERSION_MASK; - bool is_supported = false; - - if (amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) && - amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0) && - mes_rev >= 0x63) - is_supported = true; - return is_supported; + return ((amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(11, 0, 0) && + amdgpu_ip_version(adev, GC_HWIP, 0) < IP_VERSION(12, 0, 0) && + mes_rev >= 0x63) || + amdgpu_ip_version(adev, GC_HWIP, 0) >= IP_VERSION(12, 0, 0)); } /* Fix me -- node_id is used to identify the correct MES instances in the future */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h index 6b506fc72f58..e989225b354b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h @@ -149,6 +149,7 @@ struct amdgpu_mes { void *resource_1_addr[AMDGPU_MAX_MES_PIPES]; int hung_queue_db_array_size; + int hung_queue_hqd_info_offset; struct amdgpu_bo *hung_queue_db_array_gpu_obj; uint64_t hung_queue_db_array_gpu_addr; void *hung_queue_db_array_cpu_addr; @@ -238,6 +239,7 @@ struct mes_add_queue_input { struct mes_remove_queue_input { uint32_t doorbell_offset; uint64_t gang_context_addr; + bool remove_queue_after_reset; }; struct mes_map_legacy_queue_input { @@ -427,6 +429,7 @@ int amdgpu_mes_wreg(struct amdgpu_device *adev, int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev, uint32_t reg0, uint32_t reg1, uint32_t ref, uint32_t mask); +int amdgpu_mes_hdp_flush(struct amdgpu_device *adev); int amdgpu_mes_set_shader_debugger(struct amdgpu_device *adev, uint64_t process_context_addr, uint32_t spi_gdbg_per_vmid_cntl, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h index 20460cfd09bc..dc8d2f52c7d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mode.h @@ -326,6 +326,8 @@ struct amdgpu_mode_info { struct drm_property *audio_property; /* FMT dithering */ struct drm_property *dither_property; + /* Adaptive Backlight Modulation (power feature) */ + struct drm_property *abm_level_property; /* hardcoded DFP edid from BIOS */ const struct drm_edid *bios_hardcoded_edid; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h index 656b8a931dae..52c2d1731aab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h @@ -96,6 +96,7 @@ struct amdgpu_bo_va { * if non-zero, cannot unmap from GPU because user queues may still access it */ unsigned int queue_refcount; + atomic_t userq_va_mapped; }; struct amdgpu_bo { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 8c0e5d03de50..0b10497d487c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -1539,6 +1539,7 @@ static void psp_xgmi_reflect_topology_info(struct psp_context *psp, uint64_t src_node_id = psp->adev->gmc.xgmi.node_id; uint64_t dst_node_id = node_info.node_id; uint8_t dst_num_hops = node_info.num_hops; + uint8_t dst_is_sharing_enabled = node_info.is_sharing_enabled; uint8_t dst_num_links = node_info.num_links; hive = amdgpu_get_xgmi_hive(psp->adev); @@ -1558,13 +1559,20 @@ static void psp_xgmi_reflect_topology_info(struct psp_context *psp, continue; mirror_top_info->nodes[j].num_hops = dst_num_hops; - /* - * prevent 0 num_links value re-reflection since reflection + mirror_top_info->nodes[j].is_sharing_enabled = dst_is_sharing_enabled; + /* prevent 0 num_links value re-reflection since reflection * criteria is based on num_hops (direct or indirect). - * */ - if (dst_num_links) + if (dst_num_links) { mirror_top_info->nodes[j].num_links = dst_num_links; + /* swap src and dst due to frame of reference */ + for (int k = 0; k < dst_num_links; k++) { + mirror_top_info->nodes[j].port_num[k].src_xgmi_port_num = + node_info.port_num[k].dst_xgmi_port_num; + mirror_top_info->nodes[j].port_num[k].dst_xgmi_port_num = + node_info.port_num[k].src_xgmi_port_num; + } + } break; } @@ -1639,9 +1647,10 @@ int psp_xgmi_get_topology_info(struct psp_context *psp, amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == IP_VERSION(13, 0, 6) || amdgpu_ip_version(psp->adev, MP0_HWIP, 0) == - IP_VERSION(13, 0, 14); - bool ta_port_num_support = amdgpu_sriov_vf(psp->adev) ? 0 : - psp->xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG; + IP_VERSION(13, 0, 14) || + amdgpu_sriov_vf(psp->adev); + bool ta_port_num_support = psp->xgmi_context.xgmi_ta_caps & EXTEND_PEER_LINK_INFO_CMD_FLAG || + amdgpu_sriov_xgmi_ta_ext_peer_link_en(psp->adev); /* popluate the shared output buffer rather than the cmd input buffer * with node_ids as the input for GET_PEER_LINKS command execution. @@ -2355,8 +2364,11 @@ static int psp_securedisplay_initialize(struct psp_context *psp) if (!ret && !psp->securedisplay_context.context.resp_status) { psp->securedisplay_context.context.initialized = true; mutex_init(&psp->securedisplay_context.mutex); - } else + } else { + /* don't try again */ + psp->securedisplay_context.context.bin_desc.size_bytes = 0; return ret; + } mutex_lock(&psp->securedisplay_context.mutex); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c index 123bcf5c2bb1..bacf888735db 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_rap.c @@ -101,7 +101,6 @@ static ssize_t amdgpu_rap_debugfs_write(struct file *f, const char __user *buf, } amdgpu_gfx_off_ctrl(adev, true); - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index e0ee21150860..2a6cf7963dde 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -41,6 +41,7 @@ #include "atom.h" #include "amdgpu_reset.h" #include "amdgpu_psp.h" +#include "amdgpu_ras_mgr.h" #ifdef CONFIG_X86_MCE_AMD #include <asm/mce.h> @@ -149,6 +150,8 @@ static void amdgpu_ras_critical_region_fini(struct amdgpu_device *adev); #ifdef CONFIG_X86_MCE_AMD static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev); +static void +amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev); struct mce_notifier_adev_list { struct amdgpu_device *devs[MAX_GPU_INSTANCE]; int num_gpu; @@ -611,6 +614,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f, return size; } +static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev); + /** * DOC: AMDGPU RAS debugfs EEPROM table reset interface * @@ -635,6 +640,11 @@ static ssize_t amdgpu_ras_debugfs_eeprom_write(struct file *f, (struct amdgpu_device *)file_inode(f)->i_private; int ret; + if (amdgpu_uniras_enabled(adev)) { + ret = amdgpu_uniras_clear_badpages_info(adev); + return ret ? ret : size; + } + ret = amdgpu_ras_eeprom_reset_table( &(amdgpu_ras_get_context(adev)->eeprom_control)); @@ -1542,9 +1552,51 @@ out_fini_err_data: return ret; } +static int amdgpu_uniras_clear_badpages_info(struct amdgpu_device *adev) +{ + struct ras_cmd_dev_handle req = {0}; + int ret; + + ret = amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__CLEAR_BAD_PAGE_INFO, + &req, sizeof(req), NULL, 0); + if (ret) { + dev_err(adev->dev, "Failed to clear bad pages info, ret: %d\n", ret); + return ret; + } + + return 0; +} + +static int amdgpu_uniras_query_block_ecc(struct amdgpu_device *adev, + struct ras_query_if *info) +{ + struct ras_cmd_block_ecc_info_req req = {0}; + struct ras_cmd_block_ecc_info_rsp rsp = {0}; + int ret; + + if (!info) + return -EINVAL; + + req.block_id = info->head.block; + req.subblock_id = info->head.sub_block_index; + + ret = amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__GET_BLOCK_ECC_STATUS, + &req, sizeof(req), &rsp, sizeof(rsp)); + if (!ret) { + info->ce_count = rsp.ce_count; + info->ue_count = rsp.ue_count; + info->de_count = rsp.de_count; + } + + return ret; +} + int amdgpu_ras_query_error_status(struct amdgpu_device *adev, struct ras_query_if *info) { - return amdgpu_ras_query_error_status_with_event(adev, info, RAS_EVENT_TYPE_INVALID); + if (amdgpu_uniras_enabled(adev)) + return amdgpu_uniras_query_block_ecc(adev, info); + else + return amdgpu_ras_query_error_status_with_event(adev, info, RAS_EVENT_TYPE_INVALID); } int amdgpu_ras_reset_error_count(struct amdgpu_device *adev, @@ -1596,6 +1648,27 @@ int amdgpu_ras_reset_error_status(struct amdgpu_device *adev, return 0; } +static int amdgpu_uniras_error_inject(struct amdgpu_device *adev, + struct ras_inject_if *info) +{ + struct ras_cmd_inject_error_req inject_req; + struct ras_cmd_inject_error_rsp rsp; + + if (!info) + return -EINVAL; + + memset(&inject_req, 0, sizeof(inject_req)); + inject_req.block_id = info->head.block; + inject_req.subblock_id = info->head.sub_block_index; + inject_req.address = info->address; + inject_req.error_type = info->head.type; + inject_req.instance_mask = info->instance_mask; + inject_req.method = info->value; + + return amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__INJECT_ERROR, + &inject_req, sizeof(inject_req), &rsp, sizeof(rsp)); +} + /* wrapper of psp_ras_trigger_error */ int amdgpu_ras_error_inject(struct amdgpu_device *adev, struct ras_inject_if *info) @@ -1613,6 +1686,9 @@ int amdgpu_ras_error_inject(struct amdgpu_device *adev, info->head.block, info->head.sub_block_index); + if (amdgpu_uniras_enabled(adev)) + return amdgpu_uniras_error_inject(adev, info); + /* inject on guest isn't allowed, return success directly */ if (amdgpu_sriov_vf(adev)) return 0; @@ -1757,7 +1833,9 @@ int amdgpu_ras_query_error_count(struct amdgpu_device *adev, /* sysfs begin */ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev, - struct ras_badpage **bps, unsigned int *count); + struct ras_badpage *bps, uint32_t count, uint32_t start); +static int amdgpu_uniras_badpages_read(struct amdgpu_device *adev, + struct ras_badpage *bps, uint32_t count, uint32_t start); static char *amdgpu_ras_badpage_flags_str(unsigned int flags) { @@ -1815,19 +1893,50 @@ static ssize_t amdgpu_ras_sysfs_badpages_read(struct file *f, unsigned int end = div64_ul(ppos + count - 1, element_size); ssize_t s = 0; struct ras_badpage *bps = NULL; - unsigned int bps_count = 0; + int bps_count = 0, i, status; + uint64_t address; memset(buf, 0, count); - if (amdgpu_ras_badpages_read(adev, &bps, &bps_count)) + bps_count = end - start; + bps = kmalloc_array(bps_count, sizeof(*bps), GFP_KERNEL); + if (!bps) + return 0; + + memset(bps, 0, sizeof(*bps) * bps_count); + + if (amdgpu_uniras_enabled(adev)) + bps_count = amdgpu_uniras_badpages_read(adev, bps, bps_count, start); + else + bps_count = amdgpu_ras_badpages_read(adev, bps, bps_count, start); + + if (bps_count <= 0) { + kfree(bps); return 0; + } + + for (i = 0; i < bps_count; i++) { + address = ((uint64_t)bps[i].bp) << AMDGPU_GPU_PAGE_SHIFT; + if (amdgpu_ras_check_critical_address(adev, address)) + continue; + + bps[i].size = AMDGPU_GPU_PAGE_SIZE; + + status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr, + address); + if (status == -EBUSY) + bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING; + else if (status == -ENOENT) + bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT; + else + bps[i].flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED; - for (; start < end && start < bps_count; start++) s += scnprintf(&buf[s], element_size + 1, "0x%08x : 0x%08x : %1s\n", - bps[start].bp, - bps[start].size, - amdgpu_ras_badpage_flags_str(bps[start].flags)); + bps[i].bp, + bps[i].size, + amdgpu_ras_badpage_flags_str(bps[i].flags)); + } kfree(bps); @@ -1843,12 +1952,42 @@ static ssize_t amdgpu_ras_sysfs_features_read(struct device *dev, return sysfs_emit(buf, "feature mask: 0x%x\n", con->features); } +static bool amdgpu_ras_get_version_info(struct amdgpu_device *adev, u32 *major, + u32 *minor, u32 *rev) +{ + int i; + + if (!adev || !major || !minor || !rev || !amdgpu_uniras_enabled(adev)) + return false; + + for (i = 0; i < adev->num_ip_blocks; i++) { + if (adev->ip_blocks[i].version->type == AMD_IP_BLOCK_TYPE_RAS) { + *major = adev->ip_blocks[i].version->major; + *minor = adev->ip_blocks[i].version->minor; + *rev = adev->ip_blocks[i].version->rev; + return true; + } + } + + return false; +} + static ssize_t amdgpu_ras_sysfs_version_show(struct device *dev, struct device_attribute *attr, char *buf) { struct amdgpu_ras *con = container_of(attr, struct amdgpu_ras, version_attr); - return sysfs_emit(buf, "table version: 0x%x\n", con->eeprom_control.tbl_hdr.version); + u32 major, minor, rev; + ssize_t size = 0; + + size += sysfs_emit_at(buf, size, "table version: 0x%x\n", + con->eeprom_control.tbl_hdr.version); + + if (amdgpu_ras_get_version_info(con->adev, &major, &minor, &rev)) + size += sysfs_emit_at(buf, size, "ras version: %u.%u.%u\n", + major, minor, rev); + + return size; } static ssize_t amdgpu_ras_sysfs_schema_show(struct device *dev, @@ -2241,6 +2380,11 @@ void amdgpu_ras_interrupt_fatal_error_handler(struct amdgpu_device *adev) amdgpu_ras_is_err_state(adev, AMDGPU_RAS_BLOCK__ANY)) return; + if (amdgpu_uniras_enabled(adev)) { + amdgpu_ras_mgr_handle_fatal_interrupt(adev, NULL); + return; + } + if (adev->nbio.ras && adev->nbio.ras->handle_ras_controller_intr_no_bifring) adev->nbio.ras->handle_ras_controller_intr_no_bifring(adev); @@ -2411,6 +2555,16 @@ int amdgpu_ras_interrupt_dispatch(struct amdgpu_device *adev, struct ras_manager *obj; struct ras_ih_data *data; + if (amdgpu_uniras_enabled(adev)) { + struct ras_ih_info ih_info; + + memset(&ih_info, 0, sizeof(ih_info)); + ih_info.block = info->head.block; + memcpy(&ih_info.iv_entry, info->entry, sizeof(struct amdgpu_iv_entry)); + + return amdgpu_ras_mgr_handle_controller_interrupt(adev, &ih_info); + } + obj = amdgpu_ras_find_obj(adev, &info->head); if (!obj) return -EINVAL; @@ -2605,62 +2759,83 @@ static void amdgpu_ras_query_err_status(struct amdgpu_device *adev) } } -/* recovery begin */ - -/* return 0 on success. - * caller need free bps. - */ static int amdgpu_ras_badpages_read(struct amdgpu_device *adev, - struct ras_badpage **bps, unsigned int *count) + struct ras_badpage *bps, uint32_t count, uint32_t start) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); struct ras_err_handler_data *data; - int i = 0; - int ret = 0, status; + int r = 0; + uint32_t i; if (!con || !con->eh_data || !bps || !count) return -EINVAL; mutex_lock(&con->recovery_lock); data = con->eh_data; - if (!data || data->count == 0) { - *bps = NULL; - ret = -EINVAL; - goto out; + if (start < data->count) { + for (i = start; i < data->count; i++) { + if (!data->bps[i].ts) + continue; + + bps[r].bp = data->bps[i].retired_page; + r++; + if (r >= count) + break; + } } + mutex_unlock(&con->recovery_lock); - *bps = kmalloc_array(data->count, sizeof(struct ras_badpage), GFP_KERNEL); - if (!*bps) { - ret = -ENOMEM; - goto out; - } + return r; +} - for (; i < data->count; i++) { - if (!data->bps[i].ts) - continue; +static int amdgpu_uniras_badpages_read(struct amdgpu_device *adev, + struct ras_badpage *bps, uint32_t count, uint32_t start) +{ + struct ras_cmd_bad_pages_info_req cmd_input; + struct ras_cmd_bad_pages_info_rsp *output; + uint32_t group, start_group, end_group; + uint32_t pos, pos_in_group; + int r = 0, i; - (*bps)[i] = (struct ras_badpage){ - .bp = data->bps[i].retired_page, - .size = AMDGPU_GPU_PAGE_SIZE, - .flags = AMDGPU_RAS_RETIRE_PAGE_RESERVED, - }; + if (!bps || !count) + return -EINVAL; - if (amdgpu_ras_check_critical_address(adev, - data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT)) - continue; + output = kmalloc(sizeof(*output), GFP_KERNEL); + if (!output) + return -ENOMEM; - status = amdgpu_vram_mgr_query_page_status(&adev->mman.vram_mgr, - data->bps[i].retired_page << AMDGPU_GPU_PAGE_SHIFT); - if (status == -EBUSY) - (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_PENDING; - else if (status == -ENOENT) - (*bps)[i].flags = AMDGPU_RAS_RETIRE_PAGE_FAULT; + memset(&cmd_input, 0, sizeof(cmd_input)); + + start_group = start / RAS_CMD_MAX_BAD_PAGES_PER_GROUP; + end_group = (start + count + RAS_CMD_MAX_BAD_PAGES_PER_GROUP - 1) / + RAS_CMD_MAX_BAD_PAGES_PER_GROUP; + + pos = start; + for (group = start_group; group < end_group; group++) { + memset(output, 0, sizeof(*output)); + cmd_input.group_index = group; + if (amdgpu_ras_mgr_handle_ras_cmd(adev, RAS_CMD__GET_BAD_PAGES, + &cmd_input, sizeof(cmd_input), output, sizeof(*output))) + goto out; + + if (pos >= output->bp_total_cnt) + goto out; + + pos_in_group = pos - group * RAS_CMD_MAX_BAD_PAGES_PER_GROUP; + for (i = pos_in_group; i < output->bp_in_group; i++, pos++) { + if (!output->records[i].ts) + continue; + + bps[r].bp = output->records[i].retired_page; + r++; + if (r >= count) + goto out; + } } - *count = con->bad_page_num; out: - mutex_unlock(&con->recovery_lock); - return ret; + kfree(output); + return r; } static void amdgpu_ras_set_fed_all(struct amdgpu_device *adev, @@ -2748,8 +2923,12 @@ static void amdgpu_ras_do_recovery(struct work_struct *work) type = amdgpu_ras_get_fatal_error_event(adev); list_for_each_entry(remote_adev, device_list_handle, gmc.xgmi.head) { - amdgpu_ras_query_err_status(remote_adev); - amdgpu_ras_log_on_err_counter(remote_adev, type); + if (amdgpu_uniras_enabled(remote_adev)) { + amdgpu_ras_mgr_update_ras_ecc(remote_adev); + } else { + amdgpu_ras_query_err_status(remote_adev); + amdgpu_ras_log_on_err_counter(remote_adev, type); + } } } @@ -2837,8 +3016,13 @@ static int amdgpu_ras_mca2pa_by_idx(struct amdgpu_device *adev, addr_in.ma.err_addr = bps->address; addr_in.ma.socket_id = socket; addr_in.ma.ch_inst = bps->mem_channel; - /* tell RAS TA the node instance is not used */ - addr_in.ma.node_inst = TA_RAS_INV_NODE; + if (!amdgpu_ras_smu_eeprom_supported(adev)) { + /* tell RAS TA the node instance is not used */ + addr_in.ma.node_inst = TA_RAS_INV_NODE; + } else { + addr_in.ma.umc_inst = bps->mcumc_id; + addr_in.ma.node_inst = bps->cu; + } if (adev->umc.ras && adev->umc.ras->convert_ras_err_addr) ret = adev->umc.ras->convert_ras_err_addr(adev, err_data, @@ -2981,8 +3165,16 @@ static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev, int i = 0; enum amdgpu_memory_partition save_nps; - save_nps = (bps->retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK; - bps->retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT); + if (!amdgpu_ras_smu_eeprom_supported(adev)) { + save_nps = (bps->retired_page >> UMC_NPS_SHIFT) & UMC_NPS_MASK; + bps->retired_page &= ~(UMC_NPS_MASK << UMC_NPS_SHIFT); + } else { + /* if pmfw manages eeprom, save_nps is not stored on eeprom, + * we should always convert mca address into physical address, + * make save_nps different from nps + */ + save_nps = nps + 1; + } if (save_nps == nps) { if (amdgpu_umc_pages_in_a_row(adev, err_data, @@ -3048,7 +3240,8 @@ int amdgpu_ras_add_bad_pages(struct amdgpu_device *adev, if (from_rom) { /* there is no pa recs in V3, so skip pa recs processing */ - if (control->tbl_hdr.version < RAS_TABLE_VER_V3) { + if ((control->tbl_hdr.version < RAS_TABLE_VER_V3) && + !amdgpu_ras_smu_eeprom_supported(adev)) { for (i = 0; i < pages; i++) { if (control->ras_num_recs - i >= adev->umc.retire_unit) { if ((bps[i].address == bps[i + 1].address) && @@ -3118,7 +3311,13 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev, mutex_lock(&con->recovery_lock); control = &con->eeprom_control; data = con->eh_data; - unit_num = data->count / adev->umc.retire_unit - control->ras_num_recs; + if (amdgpu_ras_smu_eeprom_supported(adev)) + unit_num = control->ras_num_recs - + control->ras_num_recs_old; + else + unit_num = data->count / adev->umc.retire_unit - + control->ras_num_recs; + save_count = con->bad_page_num - control->ras_num_bad_pages; mutex_unlock(&con->recovery_lock); @@ -3126,7 +3325,7 @@ int amdgpu_ras_save_bad_pages(struct amdgpu_device *adev, *new_cnt = unit_num; /* only new entries are saved */ - if (unit_num > 0) { + if (unit_num && save_count) { /*old asics only save pa to eeprom like before*/ if (IP_VERSION_MAJ(amdgpu_ip_version(adev, UMC_HWIP, 0)) < 12) { if (amdgpu_ras_eeprom_append(control, @@ -3179,7 +3378,8 @@ static int amdgpu_ras_load_bad_pages(struct amdgpu_device *adev) /*In V3, there is no pa recs, and some cases(when address==0) may be parsed as pa recs, so add verion check to avoid it. */ - if (control->tbl_hdr.version < RAS_TABLE_VER_V3) { + if ((control->tbl_hdr.version < RAS_TABLE_VER_V3) && + !amdgpu_ras_smu_eeprom_supported(adev)) { for (i = 0; i < control->ras_num_recs; i++) { if ((control->ras_num_recs - i) >= adev->umc.retire_unit) { if ((bps[i].address == bps[i + 1].address) && @@ -3590,7 +3790,12 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *adev) if (!con || amdgpu_sriov_vf(adev)) return 0; + if (amdgpu_uniras_enabled(adev)) + return 0; + control = &con->eeprom_control; + con->ras_smu_drv = amdgpu_dpm_get_ras_smu_driver(adev); + ret = amdgpu_ras_eeprom_init(control); control->is_eeprom_valid = !ret; @@ -3751,7 +3956,9 @@ static int amdgpu_ras_recovery_fini(struct amdgpu_device *adev) mutex_unlock(&con->recovery_lock); amdgpu_ras_critical_region_init(adev); - +#ifdef CONFIG_X86_MCE_AMD + amdgpu_unregister_bad_pages_mca_notifier(adev); +#endif return 0; } /* recovery end */ @@ -3975,7 +4182,6 @@ static void amdgpu_ras_counte_dw(struct work_struct *work) atomic_set(&con->ras_ue_count, ue_count); } - pm_runtime_mark_last_busy(dev->dev); Out: pm_runtime_put_autosuspend(dev->dev); } @@ -4584,6 +4790,9 @@ int amdgpu_ras_mark_ras_event_caller(struct amdgpu_device *adev, enum ras_event_ struct ras_event_state *event_state; int ret = 0; + if (amdgpu_uniras_enabled(adev)) + return 0; + if (type >= RAS_EVENT_TYPE_COUNT) { ret = -EINVAL; goto out; @@ -4634,20 +4843,18 @@ u64 amdgpu_ras_acquire_event_id(struct amdgpu_device *adev, enum ras_event_type return id; } -void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) +int amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) { if (atomic_cmpxchg(&amdgpu_ras_in_intr, 0, 1) == 0) { struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); enum ras_event_type type = RAS_EVENT_TYPE_FATAL; - u64 event_id; + u64 event_id = RAS_EVENT_INVALID_ID; - if (amdgpu_ras_mark_ras_event(adev, type)) { - dev_err(adev->dev, - "uncorrectable hardware error (ERREVENT_ATHUB_INTERRUPT) detected!\n"); - return; - } + if (amdgpu_uniras_enabled(adev)) + return 0; - event_id = amdgpu_ras_acquire_event_id(adev, type); + if (!amdgpu_ras_mark_ras_event(adev, type)) + event_id = amdgpu_ras_acquire_event_id(adev, type); RAS_EVENT_LOG(adev, event_id, "uncorrectable hardware error" "(ERREVENT_ATHUB_INTERRUPT) detected!\n"); @@ -4656,6 +4863,8 @@ void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev) ras->gpu_reset_flags |= AMDGPU_RAS_GPU_RESET_MODE1_RESET; amdgpu_ras_reset_gpu(adev); } + + return -EBUSY; } bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev) @@ -4783,6 +4992,28 @@ static void amdgpu_register_bad_pages_mca_notifier(struct amdgpu_device *adev) notifier_registered = true; } } +static void amdgpu_unregister_bad_pages_mca_notifier(struct amdgpu_device *adev) +{ + int i, j; + + if (!notifier_registered && !mce_adev_list.num_gpu) + return; + for (i = 0, j = 0; i < mce_adev_list.num_gpu; i++) { + if (mce_adev_list.devs[i] == adev) + mce_adev_list.devs[i] = NULL; + if (!mce_adev_list.devs[i]) + ++j; + } + + if (j == mce_adev_list.num_gpu) { + mce_adev_list.num_gpu = 0; + /* Unregister x86 notifier with MCE subsystem. */ + if (notifier_registered) { + mce_unregister_decode_chain(&amdgpu_bad_page_nb); + notifier_registered = false; + } + } +} #endif struct amdgpu_ras *amdgpu_ras_get_context(struct amdgpu_device *adev) @@ -5408,6 +5639,9 @@ bool amdgpu_ras_is_rma(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + if (amdgpu_uniras_enabled(adev)) + return amdgpu_ras_mgr_is_rma(adev); + if (!con) return false; @@ -5490,3 +5724,25 @@ bool amdgpu_ras_check_critical_address(struct amdgpu_device *adev, uint64_t addr return ret; } + +void amdgpu_ras_pre_reset(struct amdgpu_device *adev, + struct list_head *device_list) +{ + struct amdgpu_device *tmp_adev = NULL; + + list_for_each_entry(tmp_adev, device_list, reset_list) { + if (amdgpu_uniras_enabled(tmp_adev)) + amdgpu_ras_mgr_pre_reset(tmp_adev); + } +} + +void amdgpu_ras_post_reset(struct amdgpu_device *adev, + struct list_head *device_list) +{ + struct amdgpu_device *tmp_adev = NULL; + + list_for_each_entry(tmp_adev, device_list, reset_list) { + if (amdgpu_uniras_enabled(tmp_adev)) + amdgpu_ras_mgr_post_reset(tmp_adev); + } +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h index 6cf0dfd38be8..ff44190d7d98 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -503,7 +503,34 @@ struct ras_critical_region { uint64_t size; }; +struct ras_eeprom_table_version { + uint32_t minor : 16; + uint32_t major : 16; +}; + +struct ras_eeprom_smu_funcs { + int (*get_ras_table_version)(struct amdgpu_device *adev, + uint32_t *table_version); + int (*get_badpage_count)(struct amdgpu_device *adev, uint32_t *count, uint32_t timeout); + int (*get_badpage_mca_addr)(struct amdgpu_device *adev, uint16_t index, uint64_t *mca_addr); + int (*set_timestamp)(struct amdgpu_device *adev, uint64_t timestamp); + int (*get_timestamp)(struct amdgpu_device *adev, + uint16_t index, uint64_t *timestamp); + int (*get_badpage_ipid)(struct amdgpu_device *adev, uint16_t index, uint64_t *ipid); + int (*erase_ras_table)(struct amdgpu_device *adev, uint32_t *result); +}; + +enum ras_smu_feature_flags { + RAS_SMU_FEATURE_BIT__RAS_EEPROM = BIT_ULL(0), +}; + +struct ras_smu_drv { + const struct ras_eeprom_smu_funcs *smu_eeprom_funcs; + void (*ras_smu_feature_flags)(struct amdgpu_device *adev, uint64_t *flags); +}; + struct amdgpu_ras { + void *ras_mgr; /* ras infrastructure */ /* for ras itself. */ uint32_t features; @@ -590,6 +617,10 @@ struct amdgpu_ras { /* Protect poison injection */ struct mutex poison_lock; + + /* Disable/Enable uniras switch */ + bool uniras_enabled; + const struct ras_smu_drv *ras_smu_drv; }; struct ras_fs_data { @@ -909,7 +940,7 @@ static inline void amdgpu_ras_intr_cleared(void) atomic_set(&amdgpu_ras_in_intr, 0); } -void amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); +int amdgpu_ras_global_ras_isr(struct amdgpu_device *adev); void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready); @@ -1008,4 +1039,9 @@ void amdgpu_ras_event_log_print(struct amdgpu_device *adev, u64 event_id, const char *fmt, ...); bool amdgpu_ras_is_rma(struct amdgpu_device *adev); + +void amdgpu_ras_pre_reset(struct amdgpu_device *adev, + struct list_head *device_list); +void amdgpu_ras_post_reset(struct amdgpu_device *adev, + struct list_head *device_list); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 3eb3fb55ccb0..64dd7a81bff5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -32,6 +32,7 @@ #include <linux/uaccess.h> #include "amdgpu_reset.h" +#include "amdgpu_ras_mgr.h" /* These are memory addresses as would be seen by one or more EEPROM * chips strung on the I2C bus, usually by manipulating pins 1-3 of a @@ -123,6 +124,8 @@ RAS_TABLE_V2_1_INFO_SIZE) \ / RAS_TABLE_RECORD_SIZE) +#define RAS_SMU_MESSAGE_TIMEOUT_MS 1000 /* 1s */ + /* Given a zero-based index of an EEPROM RAS record, yields the EEPROM * offset off of RAS_TABLE_START. That is, this is something you can * add to control->i2c_address, and then tell I2C layer to read @@ -443,40 +446,57 @@ int amdgpu_ras_eeprom_reset_table(struct amdgpu_ras_eeprom_control *control) struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr; struct amdgpu_ras_eeprom_table_ras_info *rai = &control->tbl_rai; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + u32 erase_res = 0; u8 csum; int res; mutex_lock(&control->ras_tbl_mutex); - hdr->header = RAS_TABLE_HDR_VAL; - amdgpu_ras_set_eeprom_table_version(control); + if (!amdgpu_ras_smu_eeprom_supported(adev)) { + hdr->header = RAS_TABLE_HDR_VAL; + amdgpu_ras_set_eeprom_table_version(control); - if (hdr->version >= RAS_TABLE_VER_V2_1) { - hdr->first_rec_offset = RAS_RECORD_START_V2_1; - hdr->tbl_size = RAS_TABLE_HEADER_SIZE + - RAS_TABLE_V2_1_INFO_SIZE; - rai->rma_status = GPU_HEALTH_USABLE; - /** - * GPU health represented as a percentage. - * 0 means worst health, 100 means fully health. - */ - rai->health_percent = 100; - /* ecc_page_threshold = 0 means disable bad page retirement */ - rai->ecc_page_threshold = con->bad_page_cnt_threshold; + if (hdr->version >= RAS_TABLE_VER_V2_1) { + hdr->first_rec_offset = RAS_RECORD_START_V2_1; + hdr->tbl_size = RAS_TABLE_HEADER_SIZE + + RAS_TABLE_V2_1_INFO_SIZE; + rai->rma_status = GPU_HEALTH_USABLE; + + control->ras_record_offset = RAS_RECORD_START_V2_1; + control->ras_max_record_count = RAS_MAX_RECORD_COUNT_V2_1; + /** + * GPU health represented as a percentage. + * 0 means worst health, 100 means fully health. + */ + rai->health_percent = 100; + /* ecc_page_threshold = 0 means disable bad page retirement */ + rai->ecc_page_threshold = con->bad_page_cnt_threshold; + } else { + hdr->first_rec_offset = RAS_RECORD_START; + hdr->tbl_size = RAS_TABLE_HEADER_SIZE; + + control->ras_record_offset = RAS_RECORD_START; + control->ras_max_record_count = RAS_MAX_RECORD_COUNT; + } + + csum = __calc_hdr_byte_sum(control); + if (hdr->version >= RAS_TABLE_VER_V2_1) + csum += __calc_ras_info_byte_sum(control); + csum = -csum; + hdr->checksum = csum; + res = __write_table_header(control); + if (!res && hdr->version > RAS_TABLE_VER_V1) + res = __write_table_ras_info(control); } else { - hdr->first_rec_offset = RAS_RECORD_START; - hdr->tbl_size = RAS_TABLE_HEADER_SIZE; + res = amdgpu_ras_smu_erase_ras_table(adev, &erase_res); + if (res || erase_res) { + dev_warn(adev->dev, "RAS EEPROM reset failed, res:%d result:%d", + res, erase_res); + if (!res) + res = -EIO; + } } - csum = __calc_hdr_byte_sum(control); - if (hdr->version >= RAS_TABLE_VER_V2_1) - csum += __calc_ras_info_byte_sum(control); - csum = -csum; - hdr->checksum = csum; - res = __write_table_header(control); - if (!res && hdr->version > RAS_TABLE_VER_V1) - res = __write_table_ras_info(control); - control->ras_num_recs = 0; control->ras_num_bad_pages = 0; control->ras_num_mca_recs = 0; @@ -556,6 +576,9 @@ bool amdgpu_ras_eeprom_check_err_threshold(struct amdgpu_device *adev) { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + if (amdgpu_uniras_enabled(adev)) + return amdgpu_ras_mgr_check_eeprom_safety_watermark(adev); + if (!__is_ras_eeprom_supported(adev) || !amdgpu_bad_page_threshold) return false; @@ -766,7 +789,8 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) "Saved bad pages %d reaches threshold value %d\n", control->ras_num_bad_pages, ras->bad_page_cnt_threshold); - if (adev->cper.enabled && amdgpu_cper_generate_bp_threshold_record(adev)) + if (adev->cper.enabled && !amdgpu_uniras_enabled(adev) && + amdgpu_cper_generate_bp_threshold_record(adev)) dev_warn(adev->dev, "fail to generate bad page threshold cper records\n"); if ((amdgpu_bad_page_threshold != -1) && @@ -849,6 +873,71 @@ Out: return res; } +int amdgpu_ras_eeprom_update_record_num(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + int ret, retry = 20; + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return 0; + + control->ras_num_recs_old = control->ras_num_recs; + + do { + /* 1000ms timeout is long enough, smu_get_badpage_count won't + * return -EBUSY before timeout. + */ + ret = amdgpu_ras_smu_get_badpage_count(adev, + &(control->ras_num_recs), RAS_SMU_MESSAGE_TIMEOUT_MS); + if (!ret && + (control->ras_num_recs_old == control->ras_num_recs)) { + /* record number update in PMFW needs some time, + * smu_get_badpage_count may return immediately without + * count update, sleep for a while and retry again. + */ + msleep(50); + retry--; + } else { + break; + } + } while (retry); + + /* no update of record number is not a real failure, + * don't print warning here + */ + if (!ret && (control->ras_num_recs_old == control->ras_num_recs)) + ret = -EINVAL; + + return ret; +} + +static int amdgpu_ras_smu_eeprom_append(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev) || !con) + return 0; + + control->ras_num_bad_pages = con->bad_page_num; + + if (amdgpu_bad_page_threshold != 0 && + control->ras_num_bad_pages > con->bad_page_cnt_threshold) { + dev_warn(adev->dev, + "Saved bad pages %d reaches threshold value %d\n", + control->ras_num_bad_pages, con->bad_page_cnt_threshold); + + if (adev->cper.enabled && amdgpu_cper_generate_bp_threshold_record(adev)) + dev_warn(adev->dev, "fail to generate bad page threshold cper records\n"); + + if ((amdgpu_bad_page_threshold != -1) && + (amdgpu_bad_page_threshold != -2)) + con->is_rma = true; + } + + return 0; +} + /** * amdgpu_ras_eeprom_append -- append records to the EEPROM RAS table * @control: pointer to control structure @@ -873,6 +962,9 @@ int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control, if (!__is_ras_eeprom_supported(adev)) return 0; + if (amdgpu_ras_smu_eeprom_supported(adev)) + return amdgpu_ras_smu_eeprom_append(control); + if (num == 0) { dev_err(adev->dev, "will not append 0 records\n"); return -EINVAL; @@ -948,6 +1040,50 @@ static int __amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, return res; } +int amdgpu_ras_eeprom_read_idx(struct amdgpu_ras_eeprom_control *control, + struct eeprom_table_record *record, u32 rec_idx, + const u32 num) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + uint64_t ts, end_idx; + int i, ret; + u64 mca, ipid; + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return 0; + + if (!adev->umc.ras || !adev->umc.ras->mca_ipid_parse) + return -EOPNOTSUPP; + + end_idx = rec_idx + num; + for (i = rec_idx; i < end_idx; i++) { + ret = amdgpu_ras_smu_get_badpage_mca_addr(adev, i, &mca); + if (ret) + return ret; + + ret = amdgpu_ras_smu_get_badpage_ipid(adev, i, &ipid); + if (ret) + return ret; + + ret = amdgpu_ras_smu_get_timestamp(adev, i, &ts); + if (ret) + return ret; + + record[i - rec_idx].address = mca; + /* retired_page (pa) is unused now */ + record[i - rec_idx].retired_page = 0x1ULL; + record[i - rec_idx].ts = ts; + record[i - rec_idx].err_type = AMDGPU_RAS_EEPROM_ERR_NON_RECOVERABLE; + + adev->umc.ras->mca_ipid_parse(adev, ipid, + (uint32_t *)&(record[i - rec_idx].cu), + (uint32_t *)&(record[i - rec_idx].mem_channel), + (uint32_t *)&(record[i - rec_idx].mcumc_id), NULL); + } + + return 0; +} + /** * amdgpu_ras_eeprom_read -- read EEPROM * @control: pointer to control structure @@ -969,6 +1105,9 @@ int amdgpu_ras_eeprom_read(struct amdgpu_ras_eeprom_control *control, u8 *buf, *pp; u32 g0, g1; + if (amdgpu_ras_smu_eeprom_supported(adev)) + return amdgpu_ras_eeprom_read_idx(control, record, 0, num); + if (!__is_ras_eeprom_supported(adev)) return 0; @@ -1140,6 +1279,10 @@ static ssize_t amdgpu_ras_debugfs_table_read(struct file *f, char __user *buf, int res = -EFAULT; size_t data_len; + /* pmfw manages eeprom data by itself */ + if (amdgpu_ras_smu_eeprom_supported(adev)) + return 0; + mutex_lock(&control->ras_tbl_mutex); /* We want *pos - data_len > 0, which means there's @@ -1370,6 +1513,42 @@ Out: return res == RAS_TABLE_V2_1_INFO_SIZE ? 0 : res; } +static int amdgpu_ras_smu_eeprom_init(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras_eeprom_table_header *hdr = &control->tbl_hdr; + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + uint64_t local_time; + int res; + + ras->is_rma = false; + + if (!__is_ras_eeprom_supported(adev)) + return 0; + mutex_init(&control->ras_tbl_mutex); + + res = amdgpu_ras_smu_get_table_version(adev, &(hdr->version)); + if (res) + return res; + + res = amdgpu_ras_smu_get_badpage_count(adev, + &(control->ras_num_recs), 100); + if (res) + return res; + + local_time = (uint64_t)ktime_get_real_seconds(); + res = amdgpu_ras_smu_set_timestamp(adev, local_time); + if (res) + return res; + + control->ras_max_record_count = 4000; + + control->ras_num_mca_recs = 0; + control->ras_num_pa_recs = 0; + + return 0; +} + int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control) { struct amdgpu_device *adev = to_amdgpu_device(control); @@ -1378,6 +1557,9 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control) struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); int res; + if (amdgpu_ras_smu_eeprom_supported(adev)) + return amdgpu_ras_smu_eeprom_init(control); + ras->is_rma = false; if (!__is_ras_eeprom_supported(adev)) @@ -1444,6 +1626,47 @@ int amdgpu_ras_eeprom_init(struct amdgpu_ras_eeprom_control *control) return 0; } +static int amdgpu_ras_smu_eeprom_check(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + if (!__is_ras_eeprom_supported(adev)) + return 0; + + control->ras_num_bad_pages = ras->bad_page_num; + + if ((ras->bad_page_cnt_threshold < control->ras_num_bad_pages) && + amdgpu_bad_page_threshold != 0) { + dev_warn(adev->dev, + "RAS records:%d exceed threshold:%d\n", + control->ras_num_bad_pages, ras->bad_page_cnt_threshold); + if ((amdgpu_bad_page_threshold == -1) || + (amdgpu_bad_page_threshold == -2)) { + dev_warn(adev->dev, + "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n"); + } else { + ras->is_rma = true; + dev_warn(adev->dev, + "User defined threshold is set, runtime service will be halt when threshold is reached\n"); + } + + return 0; + } + + dev_dbg(adev->dev, + "Found existing EEPROM table with %d records", + control->ras_num_bad_pages); + + /* Warn if we are at 90% of the threshold or above + */ + if (10 * control->ras_num_bad_pages >= 9 * ras->bad_page_cnt_threshold) + dev_warn(adev->dev, "RAS records:%u exceeds 90%% of threshold:%d", + control->ras_num_bad_pages, + ras->bad_page_cnt_threshold); + return 0; +} + int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control) { struct amdgpu_device *adev = to_amdgpu_device(control); @@ -1451,6 +1674,9 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control) struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); int res = 0; + if (amdgpu_ras_smu_eeprom_supported(adev)) + return amdgpu_ras_smu_eeprom_check(control); + if (!__is_ras_eeprom_supported(adev)) return 0; @@ -1541,7 +1767,8 @@ void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev) struct amdgpu_ras_eeprom_control *control; int res; - if (!__is_ras_eeprom_supported(adev) || !ras) + if (!__is_ras_eeprom_supported(adev) || !ras || + amdgpu_ras_smu_eeprom_supported(adev)) return; control = &ras->eeprom_control; if (!control->is_eeprom_valid) @@ -1561,4 +1788,143 @@ void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev) control->is_eeprom_valid = false; } return; -}
\ No newline at end of file +} + +static const struct ras_smu_drv *amdgpu_ras_get_smu_ras_drv(struct amdgpu_device *adev) +{ + struct amdgpu_ras *ras = amdgpu_ras_get_context(adev); + + if (!ras) + return NULL; + + return ras->ras_smu_drv; +} + +static uint64_t amdgpu_ras_smu_get_feature_flags(struct amdgpu_device *adev) +{ + const struct ras_smu_drv *ras_smu_drv = amdgpu_ras_get_smu_ras_drv(adev); + uint64_t flags = 0ULL; + + if (!ras_smu_drv) + goto out; + + if (ras_smu_drv->ras_smu_feature_flags) + ras_smu_drv->ras_smu_feature_flags(adev, &flags); + +out: + return flags; +} + +bool amdgpu_ras_smu_eeprom_supported(struct amdgpu_device *adev) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + uint64_t flags = 0ULL; + + if (!__is_ras_eeprom_supported(adev) || !smu_ras_drv) + return false; + + if (!smu_ras_drv->smu_eeprom_funcs) + return false; + + flags = amdgpu_ras_smu_get_feature_flags(adev); + + return !!(flags & RAS_SMU_FEATURE_BIT__RAS_EEPROM); +} + +int amdgpu_ras_smu_get_table_version(struct amdgpu_device *adev, + uint32_t *table_version) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->get_ras_table_version) + return smu_ras_drv->smu_eeprom_funcs->get_ras_table_version(adev, + table_version); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_get_badpage_count(struct amdgpu_device *adev, + uint32_t *count, uint32_t timeout) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->get_badpage_count) + return smu_ras_drv->smu_eeprom_funcs->get_badpage_count(adev, + count, timeout); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_get_badpage_mca_addr(struct amdgpu_device *adev, + uint16_t index, uint64_t *mca_addr) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->get_badpage_mca_addr) + return smu_ras_drv->smu_eeprom_funcs->get_badpage_mca_addr(adev, + index, mca_addr); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_set_timestamp(struct amdgpu_device *adev, + uint64_t timestamp) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->set_timestamp) + return smu_ras_drv->smu_eeprom_funcs->set_timestamp(adev, + timestamp); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_get_timestamp(struct amdgpu_device *adev, + uint16_t index, uint64_t *timestamp) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->get_timestamp) + return smu_ras_drv->smu_eeprom_funcs->get_timestamp(adev, + index, timestamp); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_get_badpage_ipid(struct amdgpu_device *adev, + uint16_t index, uint64_t *ipid) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->get_badpage_ipid) + return smu_ras_drv->smu_eeprom_funcs->get_badpage_ipid(adev, + index, ipid); + return -EOPNOTSUPP; +} + +int amdgpu_ras_smu_erase_ras_table(struct amdgpu_device *adev, + uint32_t *result) +{ + const struct ras_smu_drv *smu_ras_drv = amdgpu_ras_get_smu_ras_drv(adev); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return -EOPNOTSUPP; + + if (smu_ras_drv->smu_eeprom_funcs->erase_ras_table) + return smu_ras_drv->smu_eeprom_funcs->erase_ras_table(adev, + result); + return -EOPNOTSUPP; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h index ebfca4cb5688..2e5d63957e71 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.h @@ -82,6 +82,7 @@ struct amdgpu_ras_eeprom_control { /* Number of records in the table. */ u32 ras_num_recs; + u32 ras_num_recs_old; /* the bad page number is ras_num_recs or * ras_num_recs * umc.retire_unit @@ -163,6 +164,35 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control); void amdgpu_ras_eeprom_check_and_recover(struct amdgpu_device *adev); +bool amdgpu_ras_smu_eeprom_supported(struct amdgpu_device *adev); + +int amdgpu_ras_smu_get_table_version(struct amdgpu_device *adev, + uint32_t *table_version); + +int amdgpu_ras_smu_get_badpage_count(struct amdgpu_device *adev, + uint32_t *count, uint32_t timeout); + +int amdgpu_ras_smu_get_badpage_mca_addr(struct amdgpu_device *adev, + uint16_t index, uint64_t *mca_addr); + +int amdgpu_ras_smu_set_timestamp(struct amdgpu_device *adev, + uint64_t timestamp); + +int amdgpu_ras_smu_get_timestamp(struct amdgpu_device *adev, + uint16_t index, uint64_t *timestamp); + +int amdgpu_ras_smu_get_badpage_ipid(struct amdgpu_device *adev, + uint16_t index, uint64_t *ipid); + +int amdgpu_ras_smu_erase_ras_table(struct amdgpu_device *adev, + uint32_t *result); + +int amdgpu_ras_eeprom_read_idx(struct amdgpu_ras_eeprom_control *control, + struct eeprom_table_record *record, u32 rec_idx, + const u32 num); + +int amdgpu_ras_eeprom_update_record_num(struct amdgpu_ras_eeprom_control *control); + extern const struct file_operations amdgpu_ras_debugfs_eeprom_size_ops; extern const struct file_operations amdgpu_ras_debugfs_eeprom_table_ops; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c index 8f6ce948c684..c596b6df2e2d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c @@ -33,6 +33,7 @@ #include <drm/amdgpu_drm.h> #include "amdgpu.h" +#include "amdgpu_ras_mgr.h" #include "atom.h" /* @@ -159,8 +160,16 @@ void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count) */ void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib) { - while (ib->length_dw & ring->funcs->align_mask) - ib->ptr[ib->length_dw++] = ring->funcs->nop; + u32 align_mask = ring->funcs->align_mask; + u32 count = ib->length_dw & align_mask; + + if (count) { + count = align_mask + 1 - count; + + memset32(&ib->ptr[ib->length_dw], ring->funcs->nop, count); + + ib->length_dw += count; + } } /** @@ -460,9 +469,6 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid, ktime_t deadline; bool ret; - if (unlikely(ring->adev->debug_disable_soft_recovery)) - return false; - deadline = ktime_add_us(ktime_get(), 10000); if (amdgpu_sriov_vf(ring->adev) || !ring->funcs->soft_recovery || !fence) @@ -490,6 +496,66 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid, */ #if defined(CONFIG_DEBUG_FS) +static ssize_t amdgpu_ras_cper_debugfs_read(struct file *f, char __user *buf, + size_t size, loff_t *offset) +{ + const uint8_t ring_header_size = 12; + struct amdgpu_ring *ring = file_inode(f)->i_private; + struct ras_cmd_cper_snapshot_req *snapshot_req __free(kfree) = + kzalloc(sizeof(struct ras_cmd_cper_snapshot_req), GFP_KERNEL); + struct ras_cmd_cper_snapshot_rsp *snapshot_rsp __free(kfree) = + kzalloc(sizeof(struct ras_cmd_cper_snapshot_rsp), GFP_KERNEL); + struct ras_cmd_cper_record_req *record_req __free(kfree) = + kzalloc(sizeof(struct ras_cmd_cper_record_req), GFP_KERNEL); + struct ras_cmd_cper_record_rsp *record_rsp __free(kfree) = + kzalloc(sizeof(struct ras_cmd_cper_record_rsp), GFP_KERNEL); + uint8_t *ring_header __free(kfree) = + kzalloc(ring_header_size, GFP_KERNEL); + uint32_t total_cper_num; + uint64_t start_cper_id; + int r; + + if (!snapshot_req || !snapshot_rsp || !record_req || !record_rsp || + !ring_header) + return -ENOMEM; + + if (!(*offset)) { + /* Need at least 12 bytes for the header on the first read */ + if (size < ring_header_size) + return -EINVAL; + + if (copy_to_user(buf, ring_header, ring_header_size)) + return -EFAULT; + buf += ring_header_size; + size -= ring_header_size; + } + + r = amdgpu_ras_mgr_handle_ras_cmd(ring->adev, + RAS_CMD__GET_CPER_SNAPSHOT, + snapshot_req, sizeof(struct ras_cmd_cper_snapshot_req), + snapshot_rsp, sizeof(struct ras_cmd_cper_snapshot_rsp)); + if (r || !snapshot_rsp->total_cper_num) + return r; + + start_cper_id = snapshot_rsp->start_cper_id; + total_cper_num = snapshot_rsp->total_cper_num; + + record_req->buf_ptr = (uint64_t)(uintptr_t)buf; + record_req->buf_size = size; + record_req->cper_start_id = start_cper_id + *offset; + record_req->cper_num = total_cper_num; + r = amdgpu_ras_mgr_handle_ras_cmd(ring->adev, RAS_CMD__GET_CPER_RECORD, + record_req, sizeof(struct ras_cmd_cper_record_req), + record_rsp, sizeof(struct ras_cmd_cper_record_rsp)); + if (r) + return r; + + r = *offset ? record_rsp->real_data_size : record_rsp->real_data_size + ring_header_size; + (*offset) += record_rsp->real_cper_num; + + return r; +} + /* Layout of file is 12 bytes consisting of * - rptr * - wptr @@ -506,6 +572,9 @@ static ssize_t amdgpu_debugfs_ring_read(struct file *f, char __user *buf, loff_t i; int r; + if (ring->funcs->type == AMDGPU_RING_TYPE_CPER && amdgpu_uniras_enabled(ring->adev)) + return amdgpu_ras_cper_debugfs_read(f, buf, size, pos); + if (*pos & 3 || size & 3) return -EINVAL; @@ -811,7 +880,7 @@ int amdgpu_ring_reset_helper_end(struct amdgpu_ring *ring, if (r) return r; - /* signal the fence of the bad job */ + /* signal the guilty fence and set an error on all fences from the context */ if (guilty_fence) amdgpu_fence_driver_guilty_force_completion(guilty_fence); /* Re-emit the non-guilty commands */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index b6b649179776..7a27c6c4bb44 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h @@ -83,6 +83,7 @@ enum amdgpu_ring_type { AMDGPU_RING_TYPE_MES, AMDGPU_RING_TYPE_UMSCH_MM, AMDGPU_RING_TYPE_CPER, + AMDGPU_RING_TYPE_MAX, }; enum amdgpu_ib_pool_type { @@ -147,16 +148,14 @@ struct amdgpu_fence { u64 wptr; /* fence context for resets */ u64 context; - uint32_t seq; }; extern const struct drm_sched_backend_ops amdgpu_sched_ops; -void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring); void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error); void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring); -void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *fence); -void amdgpu_fence_save_wptr(struct dma_fence *fence); +void amdgpu_fence_driver_guilty_force_completion(struct amdgpu_fence *af); +void amdgpu_fence_save_wptr(struct amdgpu_fence *af); int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring); int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring, @@ -166,8 +165,8 @@ void amdgpu_fence_driver_hw_init(struct amdgpu_device *adev); void amdgpu_fence_driver_hw_fini(struct amdgpu_device *adev); int amdgpu_fence_driver_sw_init(struct amdgpu_device *adev); void amdgpu_fence_driver_sw_fini(struct amdgpu_device *adev); -int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **f, - struct amdgpu_fence *af, unsigned int flags); +int amdgpu_fence_emit(struct amdgpu_ring *ring, struct amdgpu_fence *af, + unsigned int flags); int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s, uint32_t timeout); bool amdgpu_fence_process(struct amdgpu_ring *ring); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c index 41ebe690eeff..3739be1b71e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_securedisplay.c @@ -159,7 +159,6 @@ static ssize_t amdgpu_securedisplay_debugfs_write(struct file *f, const char __u dev_err(adev->dev, "Invalid input: %s\n", str); } - pm_runtime_mark_last_busy(dev->dev); pm_runtime_put_autosuspend(dev->dev); return size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c index aa9ee5dffa45..2b931e855abd 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c @@ -188,7 +188,6 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, struct amdgpu_job *job; void *cpu_addr; uint64_t flags; - unsigned int i; int r; BUG_ON(adev->mman.buffer_funcs->copy_max_bytes < @@ -255,16 +254,9 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, dma_addr = &bo->ttm->dma_address[mm_cur->start >> PAGE_SHIFT]; amdgpu_gart_map(adev, 0, num_pages, dma_addr, flags, cpu_addr); } else { - dma_addr_t dma_address; - - dma_address = mm_cur->start; - dma_address += adev->vm_manager.vram_base_offset; + u64 pa = mm_cur->start + adev->vm_manager.vram_base_offset; - for (i = 0; i < num_pages; ++i) { - amdgpu_gart_map(adev, i << PAGE_SHIFT, 1, &dma_address, - flags, cpu_addr); - dma_address += PAGE_SIZE; - } + amdgpu_gart_map_vram_range(adev, pa, 0, num_pages, flags, cpu_addr); } dma_fence_put(amdgpu_job_submit(job)); @@ -286,12 +278,13 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo, * move and different for a BO to BO copy. * */ -int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, - const struct amdgpu_copy_mem *src, - const struct amdgpu_copy_mem *dst, - uint64_t size, bool tmz, - struct dma_resv *resv, - struct dma_fence **f) +__attribute__((nonnull)) +static int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, + const struct amdgpu_copy_mem *src, + const struct amdgpu_copy_mem *dst, + uint64_t size, bool tmz, + struct dma_resv *resv, + struct dma_fence **f) { struct amdgpu_ring *ring = adev->mman.buffer_funcs_ring; struct amdgpu_res_cursor src_mm, dst_mm; @@ -365,9 +358,7 @@ int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, } error: mutex_unlock(&adev->mman.gtt_window_lock); - if (f) - *f = dma_fence_get(fence); - dma_fence_put(fence); + *f = fence; return r; } @@ -706,10 +697,11 @@ struct amdgpu_ttm_tt { * memory and start HMM tracking CPU page table update * * Calling function must call amdgpu_ttm_tt_userptr_range_done() once and only - * once afterwards to stop HMM tracking + * once afterwards to stop HMM tracking. Its the caller responsibility to ensure + * that range is a valid memory and it is freed too. */ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, - struct hmm_range **range) + struct amdgpu_hmm_range *range) { struct ttm_tt *ttm = bo->tbo.ttm; struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm); @@ -719,9 +711,6 @@ int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, bool readonly; int r = 0; - /* Make sure get_user_pages_done() can cleanup gracefully */ - *range = NULL; - mm = bo->notifier.mm; if (unlikely(!mm)) { DRM_DEBUG_DRIVER("BO is not registered?\n"); @@ -756,38 +745,6 @@ out_unlock: return r; } -/* amdgpu_ttm_tt_discard_user_pages - Discard range and pfn array allocations - */ -void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm, - struct hmm_range *range) -{ - struct amdgpu_ttm_tt *gtt = (void *)ttm; - - if (gtt && gtt->userptr && range) - amdgpu_hmm_range_get_pages_done(range); -} - -/* - * amdgpu_ttm_tt_get_user_pages_done - stop HMM track the CPU page table change - * Check if the pages backing this ttm range have been invalidated - * - * Returns: true if pages are still valid - */ -bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, - struct hmm_range *range) -{ - struct amdgpu_ttm_tt *gtt = ttm_to_amdgpu_ttm_tt(ttm); - - if (!gtt || !gtt->userptr || !range) - return false; - - DRM_DEBUG_DRIVER("user_pages_done 0x%llx pages 0x%x\n", - gtt->userptr, ttm->num_pages); - - WARN_ONCE(!range->hmm_pfns, "No user pages to check\n"); - - return !amdgpu_hmm_range_get_pages_done(range); -} #endif /* @@ -797,12 +754,12 @@ bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, * that backs user memory and will ultimately be mapped into the device * address space. */ -void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct hmm_range *range) +void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct amdgpu_hmm_range *range) { unsigned long i; for (i = 0; i < ttm->num_pages; ++i) - ttm->pages[i] = range ? hmm_pfn_to_page(range->hmm_pfns[i]) : NULL; + ttm->pages[i] = range ? hmm_pfn_to_page(range->hmm_range.hmm_pfns[i]) : NULL; } /* @@ -1372,7 +1329,7 @@ uint64_t amdgpu_ttm_tt_pde_flags(struct ttm_tt *ttm, struct ttm_resource *mem) mem->mem_type == AMDGPU_PL_MMIO_REMAP)) { flags |= AMDGPU_PTE_SYSTEM; - if (ttm->caching == ttm_cached) + if (ttm && ttm->caching == ttm_cached) flags |= AMDGPU_PTE_SNOOPED; } @@ -1529,6 +1486,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, if (r) goto out; + mutex_lock(&adev->mman.gtt_window_lock); amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm); src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) + src_mm.start; @@ -1543,6 +1501,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo, WARN_ON(job->ibs[0].length_dw > num_dw); fence = amdgpu_job_submit(job); + mutex_unlock(&adev->mman.gtt_window_lock); if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout)) r = -ETIMEDOUT; @@ -1804,18 +1763,14 @@ static int amdgpu_ttm_reserve_tmr(struct amdgpu_device *adev) ctx->init = PSP_MEM_TRAIN_RESERVE_SUCCESS; } - if (!adev->gmc.is_app_apu) { - ret = amdgpu_bo_create_kernel_at( - adev, adev->gmc.real_vram_size - reserve_size, - reserve_size, &adev->mman.fw_reserved_memory, NULL); - if (ret) { - dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret); - amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, - NULL, NULL); - return ret; - } - } else { - DRM_DEBUG_DRIVER("backdoor fw loading path for PSP TMR, no reservation needed\n"); + ret = amdgpu_bo_create_kernel_at( + adev, adev->gmc.real_vram_size - reserve_size, reserve_size, + &adev->mman.fw_reserved_memory, NULL); + if (ret) { + dev_err(adev->dev, "alloc tmr failed(%d)!\n", ret); + amdgpu_bo_free_kernel(&adev->mman.fw_reserved_memory, NULL, + NULL); + return ret; } return 0; @@ -1837,7 +1792,7 @@ static int amdgpu_ttm_pools_init(struct amdgpu_device *adev) for (i = 0; i < adev->gmc.num_mem_partitions; i++) { ttm_pool_init(&adev->mman.ttm_pools[i], adev->dev, adev->gmc.mem_partitions[i].numa.node, - false, false); + TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M))); } return 0; } @@ -1930,8 +1885,11 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) r = ttm_device_init(&adev->mman.bdev, &amdgpu_bo_driver, adev->dev, adev_to_drm(adev)->anon_inode->i_mapping, adev_to_drm(adev)->vma_offset_manager, - adev->need_swiotlb, - dma_addressing_limited(adev->dev)); + (adev->need_swiotlb ? + TTM_ALLOCATION_POOL_USE_DMA_ALLOC : 0) | + (dma_addressing_limited(adev->dev) ? + TTM_ALLOCATION_POOL_USE_DMA32 : 0) | + TTM_ALLOCATION_POOL_BENEFICIAL_ORDER(get_order(SZ_2M))); if (r) { dev_err(adev->dev, "failed initializing buffer object driver(%d).\n", r); @@ -1980,19 +1938,19 @@ int amdgpu_ttm_init(struct amdgpu_device *adev) return r; /* - *The reserved vram for driver must be pinned to the specified - *place on the VRAM, so reserve it early. + * The reserved VRAM for the driver must be pinned to a specific + * location in VRAM, so reserve it early. */ r = amdgpu_ttm_drv_reserve_vram_init(adev); if (r) return r; /* - * only NAVI10 and onwards ASIC support for IP discovery. - * If IP discovery enabled, a block of memory should be - * reserved for IP discovey. + * only NAVI10 and later ASICs support IP discovery. + * If IP discovery is enabled, a block of memory should be + * reserved for it. */ - if (adev->mman.discovery_bin) { + if (adev->discovery.reserve_tmr) { r = amdgpu_ttm_reserve_tmr(adev); if (r) return r; @@ -2229,8 +2187,10 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable) } else { drm_sched_entity_destroy(&adev->mman.high_pr); drm_sched_entity_destroy(&adev->mman.low_pr); - dma_fence_put(man->move); - man->move = NULL; + /* Drop all the old fences since re-creating the scheduler entities + * will allocate new contexts. + */ + ttm_resource_manager_cleanup(man); } /* this just adjusts TTM size idea, which sets lpfn to the correct value */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h index 0be2728aa872..577ee04ce0bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h @@ -28,6 +28,7 @@ #include <drm/gpu_scheduler.h> #include <drm/ttm/ttm_placement.h> #include "amdgpu_vram_mgr.h" +#include "amdgpu_hmm.h" #define AMDGPU_PL_GDS (TTM_PL_PRIV + 0) #define AMDGPU_PL_GWS (TTM_PL_PRIV + 1) @@ -82,9 +83,6 @@ struct amdgpu_mman { uint64_t stolen_reserved_offset; uint64_t stolen_reserved_size; - /* discovery */ - uint8_t *discovery_bin; - uint32_t discovery_tmr_size; /* fw reserved memory */ struct amdgpu_bo *fw_reserved_memory; struct amdgpu_bo *fw_reserved_memory_extend; @@ -170,12 +168,6 @@ int amdgpu_copy_buffer(struct amdgpu_ring *ring, uint64_t src_offset, struct dma_resv *resv, struct dma_fence **fence, bool direct_submit, bool vm_needs_flush, uint32_t copy_flags); -int amdgpu_ttm_copy_mem_to_mem(struct amdgpu_device *adev, - const struct amdgpu_copy_mem *src, - const struct amdgpu_copy_mem *dst, - uint64_t size, bool tmz, - struct dma_resv *resv, - struct dma_fence **f); int amdgpu_ttm_clear_buffer(struct amdgpu_bo *bo, struct dma_resv *resv, struct dma_fence **fence); @@ -192,29 +184,16 @@ uint64_t amdgpu_ttm_domain_start(struct amdgpu_device *adev, uint32_t type); #if IS_ENABLED(CONFIG_DRM_AMDGPU_USERPTR) int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, - struct hmm_range **range); -void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm, - struct hmm_range *range); -bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, - struct hmm_range *range); + struct amdgpu_hmm_range *range); #else static inline int amdgpu_ttm_tt_get_user_pages(struct amdgpu_bo *bo, - struct hmm_range **range) + struct amdgpu_hmm_range *range) { return -EPERM; } -static inline void amdgpu_ttm_tt_discard_user_pages(struct ttm_tt *ttm, - struct hmm_range *range) -{ -} -static inline bool amdgpu_ttm_tt_get_user_pages_done(struct ttm_tt *ttm, - struct hmm_range *range) -{ - return false; -} #endif -void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct hmm_range *range); +void amdgpu_ttm_tt_set_user_pages(struct ttm_tt *ttm, struct amdgpu_hmm_range *range); int amdgpu_ttm_tt_get_userptr(const struct ttm_buffer_object *tbo, uint64_t *user_addr); int amdgpu_ttm_tt_set_userptr(struct ttm_buffer_object *bo, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 2e039fb778ea..3f0b0e9af4f3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -24,6 +24,7 @@ #include <linux/sort.h> #include "amdgpu.h" #include "umc_v6_7.h" +#include "amdgpu_ras_mgr.h" #define MAX_UMC_POISON_POLLING_TIME_SYNC 20 //ms #define MAX_UMC_HASH_STRING_SIZE 256 @@ -96,67 +97,96 @@ void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, { struct ras_err_data *err_data = (struct ras_err_data *)ras_error_status; struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_ras_eeprom_control *control = &con->eeprom_control; unsigned int error_query_mode; int ret = 0; unsigned long err_count; amdgpu_ras_get_error_query_mode(adev, &error_query_mode); + err_data->err_addr = + kcalloc(adev->umc.max_ras_err_cnt_per_query, + sizeof(struct eeprom_table_record), GFP_KERNEL); + + /* still call query_ras_error_address to clear error status + * even NOMEM error is encountered + */ + if (!err_data->err_addr) + dev_warn(adev->dev, + "Failed to alloc memory for umc error address record!\n"); + else + err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query; + mutex_lock(&con->page_retirement_lock); - ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc)); - if (ret == -EOPNOTSUPP && - error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) { - if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && - adev->umc.ras->ras_block.hw_ops->query_ras_error_count) - adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, ras_error_status); - - if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && - adev->umc.ras->ras_block.hw_ops->query_ras_error_address && - adev->umc.max_ras_err_cnt_per_query) { - err_data->err_addr = - kcalloc(adev->umc.max_ras_err_cnt_per_query, - sizeof(struct eeprom_table_record), GFP_KERNEL); - - /* still call query_ras_error_address to clear error status - * even NOMEM error is encountered - */ - if(!err_data->err_addr) - dev_warn(adev->dev, "Failed to alloc memory for " - "umc error address record!\n"); - else - err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query; - - /* umc query_ras_error_address is also responsible for clearing - * error status - */ - adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, ras_error_status); + if (!amdgpu_ras_smu_eeprom_supported(adev)) { + ret = amdgpu_dpm_get_ecc_info(adev, (void *)&(con->umc_ecc)); + if (ret == -EOPNOTSUPP && + error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) { + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_count) + adev->umc.ras->ras_block.hw_ops->query_ras_error_count(adev, + ras_error_status); + + if (adev->umc.ras && adev->umc.ras->ras_block.hw_ops && + adev->umc.ras->ras_block.hw_ops->query_ras_error_address && + adev->umc.max_ras_err_cnt_per_query) { + err_data->err_addr = + kcalloc(adev->umc.max_ras_err_cnt_per_query, + sizeof(struct eeprom_table_record), GFP_KERNEL); + + /* still call query_ras_error_address to clear error status + * even NOMEM error is encountered + */ + if (!err_data->err_addr) + dev_warn(adev->dev, + "Failed to alloc memory for umc error address record!\n"); + else + err_data->err_addr_len = + adev->umc.max_ras_err_cnt_per_query; + + /* umc query_ras_error_address is also responsible for clearing + * error status + */ + adev->umc.ras->ras_block.hw_ops->query_ras_error_address(adev, + ras_error_status); + } + } else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY || + (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) { + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_count) + adev->umc.ras->ecc_info_query_ras_error_count(adev, + ras_error_status); + + if (adev->umc.ras && + adev->umc.ras->ecc_info_query_ras_error_address && + adev->umc.max_ras_err_cnt_per_query) { + err_data->err_addr = + kcalloc(adev->umc.max_ras_err_cnt_per_query, + sizeof(struct eeprom_table_record), GFP_KERNEL); + + /* still call query_ras_error_address to clear error status + * even NOMEM error is encountered + */ + if (!err_data->err_addr) + dev_warn(adev->dev, + "Failed to alloc memory for umc error address record!\n"); + else + err_data->err_addr_len = + adev->umc.max_ras_err_cnt_per_query; + + /* umc query_ras_error_address is also responsible for clearing + * error status + */ + adev->umc.ras->ecc_info_query_ras_error_address(adev, + ras_error_status); + } } - } else if (error_query_mode == AMDGPU_RAS_FIRMWARE_ERROR_QUERY || - (!ret && error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY)) { - if (adev->umc.ras && - adev->umc.ras->ecc_info_query_ras_error_count) - adev->umc.ras->ecc_info_query_ras_error_count(adev, ras_error_status); - - if (adev->umc.ras && - adev->umc.ras->ecc_info_query_ras_error_address && - adev->umc.max_ras_err_cnt_per_query) { - err_data->err_addr = - kcalloc(adev->umc.max_ras_err_cnt_per_query, - sizeof(struct eeprom_table_record), GFP_KERNEL); - - /* still call query_ras_error_address to clear error status - * even NOMEM error is encountered - */ - if(!err_data->err_addr) - dev_warn(adev->dev, "Failed to alloc memory for " - "umc error address record!\n"); - else - err_data->err_addr_len = adev->umc.max_ras_err_cnt_per_query; - - /* umc query_ras_error_address is also responsible for clearing - * error status - */ - adev->umc.ras->ecc_info_query_ras_error_address(adev, ras_error_status); + } else { + if (!amdgpu_ras_eeprom_update_record_num(control)) { + err_data->err_addr_cnt = err_data->de_count = + control->ras_num_recs - control->ras_num_recs_old; + amdgpu_ras_eeprom_read_idx(control, err_data->err_addr, + control->ras_num_recs_old, err_data->de_count); } } @@ -166,7 +196,7 @@ void amdgpu_umc_handle_bad_pages(struct amdgpu_device *adev, if ((amdgpu_bad_page_threshold != 0) && err_data->err_addr_cnt) { amdgpu_ras_add_bad_pages(adev, err_data->err_addr, - err_data->err_addr_cnt, false); + err_data->err_addr_cnt, amdgpu_ras_smu_eeprom_supported(adev)); amdgpu_ras_save_bad_pages(adev, &err_count); amdgpu_dpm_send_hbm_bad_pages_num(adev, @@ -244,6 +274,15 @@ int amdgpu_umc_pasid_poison_handler(struct amdgpu_device *adev, } amdgpu_ras_error_data_fini(&err_data); + } else if (amdgpu_uniras_enabled(adev)) { + struct ras_ih_info ih_info = {0}; + + ih_info.block = block; + ih_info.pasid = pasid; + ih_info.reset = reset; + ih_info.pasid_fn = pasid_fn; + ih_info.data = data; + amdgpu_ras_mgr_handle_consumer_interrupt(adev, &ih_info); } else { struct amdgpu_ras *con = amdgpu_ras_get_context(adev); int ret; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index ec203f9e5ffa..28dff750c47e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -113,6 +113,8 @@ struct amdgpu_umc_ras { uint32_t (*get_die_id_from_pa)(struct amdgpu_device *adev, uint64_t mca_addr, uint64_t retired_page); void (*get_retire_flip_bits)(struct amdgpu_device *adev); + void (*mca_ipid_parse)(struct amdgpu_device *adev, uint64_t ipid, + uint32_t *did, uint32_t *ch, uint32_t *umc_inst, uint32_t *sid); }; struct amdgpu_umc_funcs { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c index 1add21160d21..9a969175900e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -25,10 +25,13 @@ #include <drm/drm_auth.h> #include <drm/drm_exec.h> #include <linux/pm_runtime.h> +#include <drm/drm_drv.h> #include "amdgpu.h" +#include "amdgpu_reset.h" #include "amdgpu_vm.h" #include "amdgpu_userq.h" +#include "amdgpu_hmm.h" #include "amdgpu_userq_fence.h" u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev) @@ -44,10 +47,130 @@ u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev) return userq_ip_mask; } -int amdgpu_userq_input_va_validate(struct amdgpu_vm *vm, u64 addr, - u64 expected_size) +static bool amdgpu_userq_is_reset_type_supported(struct amdgpu_device *adev, + enum amdgpu_ring_type ring_type, int reset_type) +{ + + if (ring_type < 0 || ring_type >= AMDGPU_RING_TYPE_MAX) + return false; + + switch (ring_type) { + case AMDGPU_RING_TYPE_GFX: + if (adev->gfx.gfx_supported_reset & reset_type) + return true; + break; + case AMDGPU_RING_TYPE_COMPUTE: + if (adev->gfx.compute_supported_reset & reset_type) + return true; + break; + case AMDGPU_RING_TYPE_SDMA: + if (adev->sdma.supported_reset & reset_type) + return true; + break; + case AMDGPU_RING_TYPE_VCN_DEC: + case AMDGPU_RING_TYPE_VCN_ENC: + if (adev->vcn.supported_reset & reset_type) + return true; + break; + case AMDGPU_RING_TYPE_VCN_JPEG: + if (adev->jpeg.supported_reset & reset_type) + return true; + break; + default: + break; + } + return false; +} + +static void amdgpu_userq_gpu_reset(struct amdgpu_device *adev) +{ + if (amdgpu_device_should_recover_gpu(adev)) { + amdgpu_reset_domain_schedule(adev->reset_domain, + &adev->userq_reset_work); + /* Wait for the reset job to complete */ + flush_work(&adev->userq_reset_work); + } +} + +static int +amdgpu_userq_detect_and_reset_queues(struct amdgpu_userq_mgr *uq_mgr) +{ + struct amdgpu_device *adev = uq_mgr->adev; + const int queue_types[] = { + AMDGPU_RING_TYPE_COMPUTE, + AMDGPU_RING_TYPE_GFX, + AMDGPU_RING_TYPE_SDMA + }; + const int num_queue_types = ARRAY_SIZE(queue_types); + bool gpu_reset = false; + int r = 0; + int i; + + /* Warning if current process mutex is not held */ + WARN_ON(!mutex_is_locked(&uq_mgr->userq_mutex)); + + if (unlikely(adev->debug_disable_gpu_ring_reset)) { + dev_err(adev->dev, "userq reset disabled by debug mask\n"); + return 0; + } + + /* + * If GPU recovery feature is disabled system-wide, + * skip all reset detection logic + */ + if (!amdgpu_gpu_recovery) + return 0; + + /* + * Iterate through all queue types to detect and reset problematic queues + * Process each queue type in the defined order + */ + for (i = 0; i < num_queue_types; i++) { + int ring_type = queue_types[i]; + const struct amdgpu_userq_funcs *funcs = adev->userq_funcs[ring_type]; + + if (!amdgpu_userq_is_reset_type_supported(adev, ring_type, AMDGPU_RESET_TYPE_PER_QUEUE)) + continue; + + if (atomic_read(&uq_mgr->userq_count[ring_type]) > 0 && + funcs && funcs->detect_and_reset) { + r = funcs->detect_and_reset(adev, ring_type); + if (r) { + gpu_reset = true; + break; + } + } + } + + if (gpu_reset) + amdgpu_userq_gpu_reset(adev); + + return r; +} + +static int amdgpu_userq_buffer_va_list_add(struct amdgpu_usermode_queue *queue, + struct amdgpu_bo_va_mapping *va_map, u64 addr) +{ + struct amdgpu_userq_va_cursor *va_cursor; + struct userq_va_list; + + va_cursor = kzalloc(sizeof(*va_cursor), GFP_KERNEL); + if (!va_cursor) + return -ENOMEM; + + INIT_LIST_HEAD(&va_cursor->list); + va_cursor->gpu_addr = addr; + atomic_set(&va_map->bo_va->userq_va_mapped, 1); + list_add(&va_cursor->list, &queue->userq_va_list); + + return 0; +} + +int amdgpu_userq_input_va_validate(struct amdgpu_usermode_queue *queue, + u64 addr, u64 expected_size) { struct amdgpu_bo_va_mapping *va_map; + struct amdgpu_vm *vm = queue->vm; u64 user_addr; u64 size; int r = 0; @@ -67,6 +190,7 @@ int amdgpu_userq_input_va_validate(struct amdgpu_vm *vm, u64 addr, /* Only validate the userq whether resident in the VM mapping range */ if (user_addr >= va_map->start && va_map->last - user_addr + 1 >= size) { + amdgpu_userq_buffer_va_list_add(queue, va_map, user_addr); amdgpu_bo_unreserve(vm->root.bo); return 0; } @@ -77,6 +201,76 @@ out_err: return r; } +static bool amdgpu_userq_buffer_va_mapped(struct amdgpu_vm *vm, u64 addr) +{ + struct amdgpu_bo_va_mapping *mapping; + bool r; + + if (amdgpu_bo_reserve(vm->root.bo, false)) + return false; + + mapping = amdgpu_vm_bo_lookup_mapping(vm, addr); + if (!IS_ERR_OR_NULL(mapping) && atomic_read(&mapping->bo_va->userq_va_mapped)) + r = true; + else + r = false; + amdgpu_bo_unreserve(vm->root.bo); + + return r; +} + +static bool amdgpu_userq_buffer_vas_mapped(struct amdgpu_usermode_queue *queue) +{ + struct amdgpu_userq_va_cursor *va_cursor, *tmp; + int r = 0; + + list_for_each_entry_safe(va_cursor, tmp, &queue->userq_va_list, list) { + r += amdgpu_userq_buffer_va_mapped(queue->vm, va_cursor->gpu_addr); + dev_dbg(queue->userq_mgr->adev->dev, + "validate the userq mapping:%p va:%llx r:%d\n", + queue, va_cursor->gpu_addr, r); + } + + if (r != 0) + return true; + + return false; +} + +static void amdgpu_userq_buffer_va_list_del(struct amdgpu_bo_va_mapping *mapping, + struct amdgpu_userq_va_cursor *va_cursor) +{ + atomic_set(&mapping->bo_va->userq_va_mapped, 0); + list_del(&va_cursor->list); + kfree(va_cursor); +} + +static int amdgpu_userq_buffer_vas_list_cleanup(struct amdgpu_device *adev, + struct amdgpu_usermode_queue *queue) +{ + struct amdgpu_userq_va_cursor *va_cursor, *tmp; + struct amdgpu_bo_va_mapping *mapping; + int r; + + r = amdgpu_bo_reserve(queue->vm->root.bo, false); + if (r) + return r; + + list_for_each_entry_safe(va_cursor, tmp, &queue->userq_va_list, list) { + mapping = amdgpu_vm_bo_lookup_mapping(queue->vm, va_cursor->gpu_addr); + if (!mapping) { + r = -EINVAL; + goto err; + } + dev_dbg(adev->dev, "delete the userq:%p va:%llx\n", + queue, va_cursor->gpu_addr); + amdgpu_userq_buffer_va_list_del(mapping, va_cursor); + } +err: + amdgpu_bo_unreserve(queue->vm->root.bo); + return r; +} + static int amdgpu_userq_preempt_helper(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_usermode_queue *queue) @@ -84,17 +278,22 @@ amdgpu_userq_preempt_helper(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_device *adev = uq_mgr->adev; const struct amdgpu_userq_funcs *userq_funcs = adev->userq_funcs[queue->queue_type]; + bool found_hung_queue = false; int r = 0; if (queue->state == AMDGPU_USERQ_STATE_MAPPED) { r = userq_funcs->preempt(uq_mgr, queue); if (r) { queue->state = AMDGPU_USERQ_STATE_HUNG; + found_hung_queue = true; } else { queue->state = AMDGPU_USERQ_STATE_PREEMPTED; } } + if (found_hung_queue) + amdgpu_userq_detect_and_reset_queues(uq_mgr); + return r; } @@ -126,16 +325,23 @@ amdgpu_userq_unmap_helper(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_device *adev = uq_mgr->adev; const struct amdgpu_userq_funcs *userq_funcs = adev->userq_funcs[queue->queue_type]; + bool found_hung_queue = false; int r = 0; if ((queue->state == AMDGPU_USERQ_STATE_MAPPED) || (queue->state == AMDGPU_USERQ_STATE_PREEMPTED)) { r = userq_funcs->unmap(uq_mgr, queue); - if (r) + if (r) { queue->state = AMDGPU_USERQ_STATE_HUNG; - else + found_hung_queue = true; + } else { queue->state = AMDGPU_USERQ_STATE_UNMAPPED; + } } + + if (found_hung_queue) + amdgpu_userq_detect_and_reset_queues(uq_mgr); + return r; } @@ -152,26 +358,33 @@ amdgpu_userq_map_helper(struct amdgpu_userq_mgr *uq_mgr, r = userq_funcs->map(uq_mgr, queue); if (r) { queue->state = AMDGPU_USERQ_STATE_HUNG; + amdgpu_userq_detect_and_reset_queues(uq_mgr); } else { queue->state = AMDGPU_USERQ_STATE_MAPPED; } } + return r; } -static void +static int amdgpu_userq_wait_for_last_fence(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_usermode_queue *queue) { struct dma_fence *f = queue->last_fence; - int ret; + int ret = 0; if (f && !dma_fence_is_signaled(f)) { - ret = dma_fence_wait_timeout(f, true, msecs_to_jiffies(100)); - if (ret <= 0) + ret = dma_fence_wait_timeout(f, true, MAX_SCHEDULE_TIMEOUT); + if (ret <= 0) { drm_file_err(uq_mgr->file, "Timed out waiting for fence=%llu:%llu\n", f->context, f->seqno); + queue->state = AMDGPU_USERQ_STATE_HUNG; + return -ETIME; + } } + + return ret; } static void @@ -182,16 +395,27 @@ amdgpu_userq_cleanup(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_device *adev = uq_mgr->adev; const struct amdgpu_userq_funcs *uq_funcs = adev->userq_funcs[queue->queue_type]; + /* Wait for mode-1 reset to complete */ + down_read(&adev->reset_domain->sem); + + /* Drop the userq reference. */ + amdgpu_userq_buffer_vas_list_cleanup(adev, queue); uq_funcs->mqd_destroy(uq_mgr, queue); amdgpu_userq_fence_driver_free(queue); - idr_remove(&uq_mgr->userq_idr, queue_id); + /* Use interrupt-safe locking since IRQ handlers may access these XArrays */ + xa_erase_irq(&uq_mgr->userq_mgr_xa, (unsigned long)queue_id); + xa_erase_irq(&adev->userq_doorbell_xa, queue->doorbell_index); + queue->userq_mgr = NULL; + list_del(&queue->userq_va_list); kfree(queue); + + up_read(&adev->reset_domain->sem); } static struct amdgpu_usermode_queue * amdgpu_userq_find(struct amdgpu_userq_mgr *uq_mgr, int qid) { - return idr_find(&uq_mgr->userq_idr, qid); + return xa_load(&uq_mgr->userq_mgr_xa, qid); } void @@ -319,17 +543,6 @@ amdgpu_userq_get_doorbell_index(struct amdgpu_userq_mgr *uq_mgr, case AMDGPU_HW_IP_DMA: db_size = sizeof(u64); break; - - case AMDGPU_HW_IP_VCN_ENC: - db_size = sizeof(u32); - db_info->doorbell_offset += AMDGPU_NAVI10_DOORBELL64_VCN0_1 << 1; - break; - - case AMDGPU_HW_IP_VPE: - db_size = sizeof(u32); - db_info->doorbell_offset += AMDGPU_NAVI10_DOORBELL64_VPE << 1; - break; - default: drm_file_err(uq_mgr->file, "[Usermode queues] IP %d not support\n", db_info->queue_type); @@ -378,10 +591,11 @@ amdgpu_userq_destroy(struct drm_file *filp, int queue_id) amdgpu_bo_unreserve(queue->db_obj.obj); } amdgpu_bo_unref(&queue->db_obj.obj); - + atomic_dec(&uq_mgr->userq_count[queue->queue_type]); #if defined(CONFIG_DEBUG_FS) debugfs_remove_recursive(queue->debugfs_queue); #endif + amdgpu_userq_detect_and_reset_queues(uq_mgr); r = amdgpu_userq_unmap_helper(uq_mgr, queue); /*TODO: It requires a reset for userq hw unmap error*/ if (unlikely(r != AMDGPU_USERQ_STATE_UNMAPPED)) { @@ -391,7 +605,6 @@ amdgpu_userq_destroy(struct drm_file *filp, int queue_id) amdgpu_userq_cleanup(uq_mgr, queue, queue_id); mutex_unlock(&uq_mgr->userq_mutex); - pm_runtime_mark_last_busy(adev_to_drm(adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(adev)->dev); return r; @@ -463,8 +676,9 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) struct amdgpu_db_info db_info; char *queue_name; bool skip_map_queue; + u32 qid; uint64_t index; - int qid, r = 0; + int r = 0; int priority = (args->in.flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_MASK) >> AMDGPU_USERQ_CREATE_FLAGS_QUEUE_PRIORITY_SHIFT; @@ -487,7 +701,6 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) * * This will also make sure we have a valid eviction fence ready to be used. */ - mutex_lock(&adev->userq_mutex); amdgpu_userq_ensure_ev_fence(&fpriv->userq_mgr, &fpriv->evf_mgr); uq_funcs = adev->userq_funcs[args->in.ip_type]; @@ -505,14 +718,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) goto unlock; } - /* Validate the userq virtual address.*/ - if (amdgpu_userq_input_va_validate(&fpriv->vm, args->in.queue_va, args->in.queue_size) || - amdgpu_userq_input_va_validate(&fpriv->vm, args->in.rptr_va, AMDGPU_GPU_PAGE_SIZE) || - amdgpu_userq_input_va_validate(&fpriv->vm, args->in.wptr_va, AMDGPU_GPU_PAGE_SIZE)) { - r = -EINVAL; - kfree(queue); - goto unlock; - } + INIT_LIST_HEAD(&queue->userq_va_list); queue->doorbell_handle = args->in.doorbell_handle; queue->queue_type = args->in.ip_type; queue->vm = &fpriv->vm; @@ -523,6 +729,15 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) db_info.db_obj = &queue->db_obj; db_info.doorbell_offset = args->in.doorbell_offset; + /* Validate the userq virtual address.*/ + if (amdgpu_userq_input_va_validate(queue, args->in.queue_va, args->in.queue_size) || + amdgpu_userq_input_va_validate(queue, args->in.rptr_va, AMDGPU_GPU_PAGE_SIZE) || + amdgpu_userq_input_va_validate(queue, args->in.wptr_va, AMDGPU_GPU_PAGE_SIZE)) { + r = -EINVAL; + kfree(queue); + goto unlock; + } + /* Convert relative doorbell offset into absolute doorbell index */ index = amdgpu_userq_get_doorbell_index(uq_mgr, &db_info, filp); if (index == (uint64_t)-EINVAL) { @@ -548,16 +763,27 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) goto unlock; } + /* Wait for mode-1 reset to complete */ + down_read(&adev->reset_domain->sem); + r = xa_err(xa_store_irq(&adev->userq_doorbell_xa, index, queue, GFP_KERNEL)); + if (r) { + kfree(queue); + up_read(&adev->reset_domain->sem); + goto unlock; + } - qid = idr_alloc(&uq_mgr->userq_idr, queue, 1, AMDGPU_MAX_USERQ_COUNT, GFP_KERNEL); - if (qid < 0) { + r = xa_alloc(&uq_mgr->userq_mgr_xa, &qid, queue, XA_LIMIT(1, AMDGPU_MAX_USERQ_COUNT), GFP_KERNEL); + if (r) { drm_file_err(uq_mgr->file, "Failed to allocate a queue id\n"); amdgpu_userq_fence_driver_free(queue); uq_funcs->mqd_destroy(uq_mgr, queue); kfree(queue); r = -ENOMEM; + up_read(&adev->reset_domain->sem); goto unlock; } + up_read(&adev->reset_domain->sem); + queue->userq_mgr = uq_mgr; /* don't map the queue if scheduling is halted */ if (adev->userq_halt_for_enforce_isolation && @@ -570,7 +796,7 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) r = amdgpu_userq_map_helper(uq_mgr, queue); if (r) { drm_file_err(uq_mgr->file, "Failed to map Queue\n"); - idr_remove(&uq_mgr->userq_idr, qid); + xa_erase(&uq_mgr->userq_mgr_xa, qid); amdgpu_userq_fence_driver_free(queue); uq_funcs->mqd_destroy(uq_mgr, queue); kfree(queue); @@ -592,10 +818,10 @@ amdgpu_userq_create(struct drm_file *filp, union drm_amdgpu_userq *args) kfree(queue_name); args->out.queue_id = qid; + atomic_inc(&uq_mgr->userq_count[queue->queue_type]); unlock: mutex_unlock(&uq_mgr->userq_mutex); - mutex_unlock(&adev->userq_mutex); return r; } @@ -693,11 +919,19 @@ static int amdgpu_userq_restore_all(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id; + unsigned long queue_id; int ret = 0, r; /* Resume all the queues for this process */ - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { + + if (!amdgpu_userq_buffer_vas_mapped(queue)) { + drm_file_err(uq_mgr->file, + "trying restore queue without va mapping\n"); + queue->state = AMDGPU_USERQ_STATE_INVALID_VA; + continue; + } + r = amdgpu_userq_restore_helper(uq_mgr, queue); if (r) ret = r; @@ -760,12 +994,21 @@ static int amdgpu_userq_vm_validate(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr); + bool invalidated = false, new_addition = false; + struct ttm_operation_ctx ctx = { true, false }; struct amdgpu_device *adev = uq_mgr->adev; + struct amdgpu_hmm_range *range; struct amdgpu_vm *vm = &fpriv->vm; + unsigned long key, tmp_key; struct amdgpu_bo_va *bo_va; + struct amdgpu_bo *bo; struct drm_exec exec; + struct xarray xa; int ret; + xa_init(&xa); + +retry_lock: drm_exec_init(&exec, DRM_EXEC_IGNORE_DUPLICATES, 0); drm_exec_until_all_locked(&exec) { ret = amdgpu_vm_lock_pd(vm, &exec, 1); @@ -792,10 +1035,74 @@ amdgpu_userq_vm_validate(struct amdgpu_userq_mgr *uq_mgr) goto unlock_all; } + if (invalidated) { + xa_for_each(&xa, tmp_key, range) { + bo = range->bo; + amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_CPU); + ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx); + if (ret) + goto unlock_all; + + amdgpu_ttm_tt_set_user_pages(bo->tbo.ttm, range); + + amdgpu_bo_placement_from_domain(bo, AMDGPU_GEM_DOMAIN_GTT); + ret = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx); + if (ret) + goto unlock_all; + } + invalidated = false; + } + ret = amdgpu_vm_handle_moved(adev, vm, NULL); if (ret) goto unlock_all; + key = 0; + /* Validate User Ptr BOs */ + list_for_each_entry(bo_va, &vm->done, base.vm_status) { + bo = bo_va->base.bo; + if (!bo) + continue; + + if (!amdgpu_ttm_tt_is_userptr(bo->tbo.ttm)) + continue; + + range = xa_load(&xa, key); + if (range && range->bo != bo) { + xa_erase(&xa, key); + amdgpu_hmm_range_free(range); + range = NULL; + } + + if (!range) { + range = amdgpu_hmm_range_alloc(bo); + if (!range) { + ret = -ENOMEM; + goto unlock_all; + } + + xa_store(&xa, key, range, GFP_KERNEL); + new_addition = true; + } + key++; + } + + if (new_addition) { + drm_exec_fini(&exec); + xa_for_each(&xa, tmp_key, range) { + if (!range) + continue; + bo = range->bo; + ret = amdgpu_ttm_tt_get_user_pages(bo, range); + if (ret) + goto unlock_all; + } + + invalidated = true; + new_addition = false; + goto retry_lock; + } + ret = amdgpu_vm_update_pdes(adev, vm, false); if (ret) goto unlock_all; @@ -815,6 +1122,13 @@ amdgpu_userq_vm_validate(struct amdgpu_userq_mgr *uq_mgr) unlock_all: drm_exec_fini(&exec); + xa_for_each(&xa, tmp_key, range) { + if (!range) + continue; + bo = range->bo; + amdgpu_hmm_range_free(range); + } + xa_destroy(&xa); return ret; } @@ -848,11 +1162,12 @@ static int amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id; + unsigned long queue_id; int ret = 0, r; + amdgpu_userq_detect_and_reset_queues(uq_mgr); /* Try to unmap all the queues in this process ctx */ - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { r = amdgpu_userq_preempt_helper(uq_mgr, queue); if (r) ret = r; @@ -863,13 +1178,31 @@ amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr) return ret; } +void amdgpu_userq_reset_work(struct work_struct *work) +{ + struct amdgpu_device *adev = container_of(work, struct amdgpu_device, + userq_reset_work); + struct amdgpu_reset_context reset_context; + + memset(&reset_context, 0, sizeof(reset_context)); + + reset_context.method = AMD_RESET_METHOD_NONE; + reset_context.reset_req_dev = adev; + reset_context.src = AMDGPU_RESET_SRC_USERQ; + set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags); + /*set_bit(AMDGPU_SKIP_COREDUMP, &reset_context.flags);*/ + + amdgpu_device_gpu_recover(adev, NULL, &reset_context); +} + static int amdgpu_userq_wait_for_signal(struct amdgpu_userq_mgr *uq_mgr) { struct amdgpu_usermode_queue *queue; - int queue_id, ret; + unsigned long queue_id; + int ret; - idr_for_each_entry(&uq_mgr->userq_idr, queue, queue_id) { + xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { struct dma_fence *f = queue->last_fence; if (!f || dma_fence_is_signaled(f)) @@ -889,22 +1222,19 @@ void amdgpu_userq_evict(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_eviction_fence *ev_fence) { - int ret; struct amdgpu_fpriv *fpriv = uq_mgr_to_fpriv(uq_mgr); struct amdgpu_eviction_fence_mgr *evf_mgr = &fpriv->evf_mgr; + struct amdgpu_device *adev = uq_mgr->adev; + int ret; /* Wait for any pending userqueue fence work to finish */ ret = amdgpu_userq_wait_for_signal(uq_mgr); - if (ret) { - drm_file_err(uq_mgr->file, "Not evicting userqueue, timeout waiting for work\n"); - return; - } + if (ret) + dev_err(adev->dev, "Not evicting userqueue, timeout waiting for work\n"); ret = amdgpu_userq_evict_all(uq_mgr); - if (ret) { - drm_file_err(uq_mgr->file, "Failed to evict userqueue\n"); - return; - } + if (ret) + dev_err(adev->dev, "Failed to evict userqueue\n"); /* Signal current eviction fence */ amdgpu_eviction_fence_signal(evf_mgr, ev_fence); @@ -922,44 +1252,31 @@ int amdgpu_userq_mgr_init(struct amdgpu_userq_mgr *userq_mgr, struct drm_file *f struct amdgpu_device *adev) { mutex_init(&userq_mgr->userq_mutex); - idr_init_base(&userq_mgr->userq_idr, 1); + xa_init_flags(&userq_mgr->userq_mgr_xa, XA_FLAGS_ALLOC); userq_mgr->adev = adev; userq_mgr->file = file_priv; - mutex_lock(&adev->userq_mutex); - list_add(&userq_mgr->list, &adev->userq_mgr_list); - mutex_unlock(&adev->userq_mutex); - INIT_DELAYED_WORK(&userq_mgr->resume_work, amdgpu_userq_restore_worker); return 0; } void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr) { - struct amdgpu_device *adev = userq_mgr->adev; struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - uint32_t queue_id; + unsigned long queue_id; cancel_delayed_work_sync(&userq_mgr->resume_work); - mutex_lock(&adev->userq_mutex); mutex_lock(&userq_mgr->userq_mutex); - idr_for_each_entry(&userq_mgr->userq_idr, queue, queue_id) { + amdgpu_userq_detect_and_reset_queues(userq_mgr); + xa_for_each(&userq_mgr->userq_mgr_xa, queue_id, queue) { amdgpu_userq_wait_for_last_fence(userq_mgr, queue); amdgpu_userq_unmap_helper(userq_mgr, queue); amdgpu_userq_cleanup(userq_mgr, queue, queue_id); } - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - if (uqm == userq_mgr) { - list_del(&uqm->list); - break; - } - } - idr_destroy(&userq_mgr->userq_idr); + xa_destroy(&userq_mgr->userq_mgr_xa); mutex_unlock(&userq_mgr->userq_mutex); - mutex_unlock(&adev->userq_mutex); mutex_destroy(&userq_mgr->userq_mutex); } @@ -967,57 +1284,51 @@ int amdgpu_userq_suspend(struct amdgpu_device *adev) { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; - int ret = 0, r; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; + int r; if (!ip_mask) return 0; - mutex_lock(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; cancel_delayed_work_sync(&uqm->resume_work); - mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (adev->in_s0ix) - r = amdgpu_userq_preempt_helper(uqm, queue); - else - r = amdgpu_userq_unmap_helper(uqm, queue); - if (r) - ret = r; - } - mutex_unlock(&uqm->userq_mutex); + guard(mutex)(&uqm->userq_mutex); + amdgpu_userq_detect_and_reset_queues(uqm); + if (adev->in_s0ix) + r = amdgpu_userq_preempt_helper(uqm, queue); + else + r = amdgpu_userq_unmap_helper(uqm, queue); + if (r) + return r; } - mutex_unlock(&adev->userq_mutex); - return ret; + return 0; } int amdgpu_userq_resume(struct amdgpu_device *adev) { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; - int ret = 0, r; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; + int r; if (!ip_mask) return 0; - mutex_lock(&adev->userq_mutex); - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (adev->in_s0ix) - r = amdgpu_userq_restore_helper(uqm, queue); - else - r = amdgpu_userq_map_helper(uqm, queue); - if (r) - ret = r; - } - mutex_unlock(&uqm->userq_mutex); + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; + guard(mutex)(&uqm->userq_mutex); + if (adev->in_s0ix) + r = amdgpu_userq_restore_helper(uqm, queue); + else + r = amdgpu_userq_map_helper(uqm, queue); + if (r) + return r; } - mutex_unlock(&adev->userq_mutex); - return ret; + + return 0; } int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev, @@ -1025,33 +1336,32 @@ int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev, { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int ret = 0, r; /* only need to stop gfx/compute */ if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE)))) return 0; - mutex_lock(&adev->userq_mutex); if (adev->userq_halt_for_enforce_isolation) dev_warn(adev->dev, "userq scheduling already stopped!\n"); adev->userq_halt_for_enforce_isolation = true; - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; cancel_delayed_work_sync(&uqm->resume_work); mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (((queue->queue_type == AMDGPU_HW_IP_GFX) || - (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && - (queue->xcp_id == idx)) { - r = amdgpu_userq_preempt_helper(uqm, queue); - if (r) - ret = r; - } + if (((queue->queue_type == AMDGPU_HW_IP_GFX) || + (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && + (queue->xcp_id == idx)) { + amdgpu_userq_detect_and_reset_queues(uqm); + r = amdgpu_userq_preempt_helper(uqm, queue); + if (r) + ret = r; } mutex_unlock(&uqm->userq_mutex); } - mutex_unlock(&adev->userq_mutex); + return ret; } @@ -1060,21 +1370,20 @@ int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev, { u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; - int queue_id; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; int ret = 0, r; /* only need to stop gfx/compute */ if (!(ip_mask & ((1 << AMDGPU_HW_IP_GFX) | (1 << AMDGPU_HW_IP_COMPUTE)))) return 0; - mutex_lock(&adev->userq_mutex); if (!adev->userq_halt_for_enforce_isolation) dev_warn(adev->dev, "userq scheduling already started!\n"); adev->userq_halt_for_enforce_isolation = false; - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; mutex_lock(&uqm->userq_mutex); - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { if (((queue->queue_type == AMDGPU_HW_IP_GFX) || (queue->queue_type == AMDGPU_HW_IP_COMPUTE)) && (queue->xcp_id == idx)) { @@ -1082,9 +1391,92 @@ int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev, if (r) ret = r; } - } mutex_unlock(&uqm->userq_mutex); } - mutex_unlock(&adev->userq_mutex); + return ret; } + +int amdgpu_userq_gem_va_unmap_validate(struct amdgpu_device *adev, + struct amdgpu_bo_va_mapping *mapping, + uint64_t saddr) +{ + u32 ip_mask = amdgpu_userq_get_supported_ip_mask(adev); + struct amdgpu_bo_va *bo_va = mapping->bo_va; + struct dma_resv *resv = bo_va->base.bo->tbo.base.resv; + int ret = 0; + + if (!ip_mask) + return 0; + + dev_warn_once(adev->dev, "now unmapping a vital queue va:%llx\n", saddr); + /** + * The userq VA mapping reservation should include the eviction fence, + * if the eviction fence can't signal successfully during unmapping, + * then driver will warn to flag this improper unmap of the userq VA. + * Note: The eviction fence may be attached to different BOs, and this + * unmap is only for one kind of userq VAs, so at this point suppose + * the eviction fence is always unsignaled. + */ + if (!dma_resv_test_signaled(resv, DMA_RESV_USAGE_BOOKKEEP)) { + ret = dma_resv_wait_timeout(resv, DMA_RESV_USAGE_BOOKKEEP, true, + MAX_SCHEDULE_TIMEOUT); + if (ret <= 0) + return -EBUSY; + } + + return 0; +} + +void amdgpu_userq_pre_reset(struct amdgpu_device *adev) +{ + const struct amdgpu_userq_funcs *userq_funcs; + struct amdgpu_usermode_queue *queue; + struct amdgpu_userq_mgr *uqm; + unsigned long queue_id; + + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; + cancel_delayed_work_sync(&uqm->resume_work); + if (queue->state == AMDGPU_USERQ_STATE_MAPPED) { + amdgpu_userq_wait_for_last_fence(uqm, queue); + userq_funcs = adev->userq_funcs[queue->queue_type]; + userq_funcs->unmap(uqm, queue); + /* just mark all queues as hung at this point. + * if unmap succeeds, we could map again + * in amdgpu_userq_post_reset() if vram is not lost + */ + queue->state = AMDGPU_USERQ_STATE_HUNG; + amdgpu_userq_fence_driver_force_completion(queue); + } + } +} + +int amdgpu_userq_post_reset(struct amdgpu_device *adev, bool vram_lost) +{ + /* if any queue state is AMDGPU_USERQ_STATE_UNMAPPED + * at this point, we should be able to map it again + * and continue if vram is not lost. + */ + struct amdgpu_userq_mgr *uqm; + struct amdgpu_usermode_queue *queue; + const struct amdgpu_userq_funcs *userq_funcs; + unsigned long queue_id; + int r = 0; + + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + uqm = queue->userq_mgr; + if (queue->state == AMDGPU_USERQ_STATE_HUNG && !vram_lost) { + userq_funcs = adev->userq_funcs[queue->queue_type]; + /* Re-map queue */ + r = userq_funcs->map(uqm, queue); + if (r) { + dev_err(adev->dev, "Failed to remap queue %ld\n", queue_id); + continue; + } + queue->state = AMDGPU_USERQ_STATE_MAPPED; + } + } + + return r; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h index c027dd916672..c37444427a14 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h @@ -37,6 +37,7 @@ enum amdgpu_userq_state { AMDGPU_USERQ_STATE_MAPPED, AMDGPU_USERQ_STATE_PREEMPTED, AMDGPU_USERQ_STATE_HUNG, + AMDGPU_USERQ_STATE_INVALID_VA, }; struct amdgpu_mqd_prop; @@ -47,6 +48,11 @@ struct amdgpu_userq_obj { struct amdgpu_bo *obj; }; +struct amdgpu_userq_va_cursor { + u64 gpu_addr; + struct list_head list; +}; + struct amdgpu_usermode_queue { int queue_type; enum amdgpu_userq_state state; @@ -66,6 +72,8 @@ struct amdgpu_usermode_queue { u32 xcp_id; int priority; struct dentry *debugfs_queue; + + struct list_head userq_va_list; }; struct amdgpu_userq_funcs { @@ -88,12 +96,17 @@ struct amdgpu_userq_funcs { /* Usermode queues for gfx */ struct amdgpu_userq_mgr { - struct idr userq_idr; + /** + * @userq_mgr_xa: Per-process user queue map (queue ID → queue) + * Key: queue_id (unique ID within the process's userq manager) + * Value: struct amdgpu_usermode_queue + */ + struct xarray userq_mgr_xa; struct mutex userq_mutex; struct amdgpu_device *adev; struct delayed_work resume_work; - struct list_head list; struct drm_file *file; + atomic_t userq_count[AMDGPU_RING_TYPE_MAX]; }; struct amdgpu_db_info { @@ -136,7 +149,13 @@ int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev, u32 idx); int amdgpu_userq_start_sched_for_enforce_isolation(struct amdgpu_device *adev, u32 idx); - -int amdgpu_userq_input_va_validate(struct amdgpu_vm *vm, u64 addr, - u64 expected_size); +void amdgpu_userq_reset_work(struct work_struct *work); +void amdgpu_userq_pre_reset(struct amdgpu_device *adev); +int amdgpu_userq_post_reset(struct amdgpu_device *adev, bool vram_lost); + +int amdgpu_userq_input_va_validate(struct amdgpu_usermode_queue *queue, + u64 addr, u64 expected_size); +int amdgpu_userq_gem_va_unmap_validate(struct amdgpu_device *adev, + struct amdgpu_bo_va_mapping *mapping, + uint64_t saddr); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c index 761bad98da3e..eba9fb359047 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c @@ -151,15 +151,16 @@ void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_d { struct amdgpu_userq_fence *userq_fence, *tmp; struct dma_fence *fence; + unsigned long flags; u64 rptr; int i; if (!fence_drv) return; + spin_lock_irqsave(&fence_drv->fence_list_lock, flags); rptr = amdgpu_userq_fence_read(fence_drv); - spin_lock(&fence_drv->fence_list_lock); list_for_each_entry_safe(userq_fence, tmp, &fence_drv->fences, link) { fence = &userq_fence->base; @@ -174,7 +175,7 @@ void amdgpu_userq_fence_driver_process(struct amdgpu_userq_fence_driver *fence_d list_del(&userq_fence->link); dma_fence_put(fence); } - spin_unlock(&fence_drv->fence_list_lock); + spin_unlock_irqrestore(&fence_drv->fence_list_lock, flags); } void amdgpu_userq_fence_driver_destroy(struct kref *ref) @@ -386,6 +387,7 @@ static int amdgpu_userq_fence_read_wptr(struct amdgpu_usermode_queue *queue, amdgpu_bo_unreserve(queue->vm->root.bo); r = amdgpu_bo_reserve(bo, true); if (r) { + amdgpu_bo_unref(&bo); DRM_ERROR("Failed to reserve userqueue wptr bo"); return r; } @@ -537,7 +539,7 @@ int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data, } /* Retrieve the user queue */ - queue = idr_find(&userq_mgr->userq_idr, args->queue_id); + queue = xa_load(&userq_mgr->userq_mgr_xa, args->queue_id); if (!queue) { r = -ENOENT; goto put_gobj_write; @@ -899,7 +901,7 @@ int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data, */ num_fences = dma_fence_dedup_array(fences, num_fences); - waitq = idr_find(&userq_mgr->userq_idr, wait_info->waitq_id); + waitq = xa_load(&userq_mgr->userq_mgr_xa, wait_info->waitq_id); if (!waitq) { r = -EINVAL; goto free_fences; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c index ce318f5de047..a7d8f1ce6ac2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c @@ -41,6 +41,9 @@ #define VCE_IDLE_TIMEOUT msecs_to_jiffies(1000) /* Firmware Names */ +#ifdef CONFIG_DRM_AMDGPU_SI +#define FIRMWARE_VCE_V1_0 "amdgpu/vce_1_0_0.bin" +#endif #ifdef CONFIG_DRM_AMDGPU_CIK #define FIRMWARE_BONAIRE "amdgpu/bonaire_vce.bin" #define FIRMWARE_KABINI "amdgpu/kabini_vce.bin" @@ -61,6 +64,9 @@ #define FIRMWARE_VEGA12 "amdgpu/vega12_vce.bin" #define FIRMWARE_VEGA20 "amdgpu/vega20_vce.bin" +#ifdef CONFIG_DRM_AMDGPU_SI +MODULE_FIRMWARE(FIRMWARE_VCE_V1_0); +#endif #ifdef CONFIG_DRM_AMDGPU_CIK MODULE_FIRMWARE(FIRMWARE_BONAIRE); MODULE_FIRMWARE(FIRMWARE_KABINI); @@ -88,82 +94,93 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle, bool direct, struct dma_fence **fence); /** - * amdgpu_vce_sw_init - allocate memory, load vce firmware + * amdgpu_vce_firmware_name() - determine the firmware file name for VCE * * @adev: amdgpu_device pointer - * @size: size for the new BO * - * First step to get VCE online, allocate memory and load the firmware + * Each chip that has VCE IP may need a different firmware. + * This function returns the name of the VCE firmware file + * appropriate for the current chip. */ -int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size) +static const char *amdgpu_vce_firmware_name(struct amdgpu_device *adev) { - const char *fw_name; - const struct common_firmware_header *hdr; - unsigned int ucode_version, version_major, version_minor, binary_id; - int i, r; - switch (adev->asic_type) { +#ifdef CONFIG_DRM_AMDGPU_SI + case CHIP_PITCAIRN: + case CHIP_TAHITI: + case CHIP_VERDE: + return FIRMWARE_VCE_V1_0; +#endif #ifdef CONFIG_DRM_AMDGPU_CIK case CHIP_BONAIRE: - fw_name = FIRMWARE_BONAIRE; - break; + return FIRMWARE_BONAIRE; case CHIP_KAVERI: - fw_name = FIRMWARE_KAVERI; - break; + return FIRMWARE_KAVERI; case CHIP_KABINI: - fw_name = FIRMWARE_KABINI; - break; + return FIRMWARE_KABINI; case CHIP_HAWAII: - fw_name = FIRMWARE_HAWAII; - break; + return FIRMWARE_HAWAII; case CHIP_MULLINS: - fw_name = FIRMWARE_MULLINS; - break; + return FIRMWARE_MULLINS; #endif case CHIP_TONGA: - fw_name = FIRMWARE_TONGA; - break; + return FIRMWARE_TONGA; case CHIP_CARRIZO: - fw_name = FIRMWARE_CARRIZO; - break; + return FIRMWARE_CARRIZO; case CHIP_FIJI: - fw_name = FIRMWARE_FIJI; - break; + return FIRMWARE_FIJI; case CHIP_STONEY: - fw_name = FIRMWARE_STONEY; - break; + return FIRMWARE_STONEY; case CHIP_POLARIS10: - fw_name = FIRMWARE_POLARIS10; - break; + return FIRMWARE_POLARIS10; case CHIP_POLARIS11: - fw_name = FIRMWARE_POLARIS11; - break; + return FIRMWARE_POLARIS11; case CHIP_POLARIS12: - fw_name = FIRMWARE_POLARIS12; - break; + return FIRMWARE_POLARIS12; case CHIP_VEGAM: - fw_name = FIRMWARE_VEGAM; - break; + return FIRMWARE_VEGAM; case CHIP_VEGA10: - fw_name = FIRMWARE_VEGA10; - break; + return FIRMWARE_VEGA10; case CHIP_VEGA12: - fw_name = FIRMWARE_VEGA12; - break; + return FIRMWARE_VEGA12; case CHIP_VEGA20: - fw_name = FIRMWARE_VEGA20; - break; + return FIRMWARE_VEGA20; default: - return -EINVAL; + return NULL; } +} + +/** + * amdgpu_vce_early_init() - try to load VCE firmware + * + * @adev: amdgpu_device pointer + * + * Tries to load the VCE firmware. + * + * When not found, returns ENOENT so that the driver can + * still load and initialize the rest of the IP blocks. + * The GPU can function just fine without VCE, they will just + * not support video encoding. + */ +int amdgpu_vce_early_init(struct amdgpu_device *adev) +{ + const char *fw_name = amdgpu_vce_firmware_name(adev); + const struct common_firmware_header *hdr; + unsigned int ucode_version, version_major, version_minor, binary_id; + int r; + + if (!fw_name) + return -ENOENT; r = amdgpu_ucode_request(adev, &adev->vce.fw, AMDGPU_UCODE_REQUIRED, "%s", fw_name); if (r) { - dev_err(adev->dev, "amdgpu_vce: Can't validate firmware \"%s\"\n", - fw_name); + dev_err(adev->dev, + "amdgpu_vce: Firmware \"%s\" not found or failed to validate (%d)\n", + fw_name, r); + amdgpu_ucode_release(&adev->vce.fw); - return r; + return -ENOENT; } hdr = (const struct common_firmware_header *)adev->vce.fw->data; @@ -172,11 +189,35 @@ int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size) version_major = (ucode_version >> 20) & 0xfff; version_minor = (ucode_version >> 8) & 0xfff; binary_id = ucode_version & 0xff; - DRM_INFO("Found VCE firmware Version: %d.%d Binary ID: %d\n", + dev_info(adev->dev, "Found VCE firmware Version: %d.%d Binary ID: %d\n", version_major, version_minor, binary_id); adev->vce.fw_version = ((version_major << 24) | (version_minor << 16) | (binary_id << 8)); + return 0; +} + +/** + * amdgpu_vce_sw_init() - allocate memory for VCE BO + * + * @adev: amdgpu_device pointer + * @size: size for the new BO + * + * First step to get VCE online: allocate memory for VCE BO. + * The VCE firmware binary is copied into the VCE BO later, + * in amdgpu_vce_resume. The VCE executes its code from the + * VCE BO and also uses the space in this BO for its stack and data. + * + * Ideally this BO should be placed in VRAM for optimal performance, + * although technically it also runs from system RAM (albeit slowly). + */ +int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size) +{ + int i, r; + + if (!adev->vce.fw) + return -ENOENT; + r = amdgpu_bo_create_kernel(adev, size, PAGE_SIZE, AMDGPU_GEM_DOMAIN_VRAM | AMDGPU_GEM_DOMAIN_GTT, @@ -285,40 +326,23 @@ int amdgpu_vce_suspend(struct amdgpu_device *adev) */ int amdgpu_vce_resume(struct amdgpu_device *adev) { - void *cpu_addr; const struct common_firmware_header *hdr; unsigned int offset; - int r, idx; + int idx; if (adev->vce.vcpu_bo == NULL) return -EINVAL; - r = amdgpu_bo_reserve(adev->vce.vcpu_bo, false); - if (r) { - dev_err(adev->dev, "(%d) failed to reserve VCE bo\n", r); - return r; - } - - r = amdgpu_bo_kmap(adev->vce.vcpu_bo, &cpu_addr); - if (r) { - amdgpu_bo_unreserve(adev->vce.vcpu_bo); - dev_err(adev->dev, "(%d) VCE map failed\n", r); - return r; - } - hdr = (const struct common_firmware_header *)adev->vce.fw->data; offset = le32_to_cpu(hdr->ucode_array_offset_bytes); if (drm_dev_enter(adev_to_drm(adev), &idx)) { - memcpy_toio(cpu_addr, adev->vce.fw->data + offset, + memset_io(adev->vce.cpu_addr, 0, amdgpu_bo_size(adev->vce.vcpu_bo)); + memcpy_toio(adev->vce.cpu_addr, adev->vce.fw->data + offset, adev->vce.fw->size - offset); drm_dev_exit(idx); } - amdgpu_bo_kunmap(adev->vce.vcpu_bo); - - amdgpu_bo_unreserve(adev->vce.vcpu_bo); - return 0; } @@ -427,6 +451,24 @@ void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp) } /** + * amdgpu_vce_required_gart_pages() - gets number of GART pages required by VCE + * + * @adev: amdgpu_device pointer + * + * Returns how many GART pages we need before GTT for the VCE IP block. + * For VCE1, see vce_v1_0_ensure_vcpu_bo_32bit_addr for details. + * For VCE2+, this is not needed so return zero. + */ +u32 amdgpu_vce_required_gart_pages(struct amdgpu_device *adev) +{ + /* VCE IP block not added yet, so can't use amdgpu_ip_version */ + if (adev->family == AMDGPU_FAMILY_SI) + return 512; + + return 0; +} + +/** * amdgpu_vce_get_create_msg - generate a VCE create msg * * @ring: ring we should submit the msg to diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h index 6e53f872d084..1c3464ce5037 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vce.h @@ -51,14 +51,17 @@ struct amdgpu_vce { struct drm_sched_entity entity; uint32_t srbm_soft_reset; unsigned num_rings; + uint32_t keyselect; }; +int amdgpu_vce_early_init(struct amdgpu_device *adev); int amdgpu_vce_sw_init(struct amdgpu_device *adev, unsigned long size); int amdgpu_vce_sw_fini(struct amdgpu_device *adev); int amdgpu_vce_entity_init(struct amdgpu_device *adev, struct amdgpu_ring *ring); int amdgpu_vce_suspend(struct amdgpu_device *adev); int amdgpu_vce_resume(struct amdgpu_device *adev); void amdgpu_vce_free_handles(struct amdgpu_device *adev, struct drm_file *filp); +u32 amdgpu_vce_required_gart_pages(struct amdgpu_device *adev); int amdgpu_vce_ring_parse_cs(struct amdgpu_cs_parser *p, struct amdgpu_job *job, struct amdgpu_ib *ib); int amdgpu_vce_ring_parse_cs_vm(struct amdgpu_cs_parser *p, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index dc8a17bcc3c8..82624b44e661 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -100,7 +100,8 @@ #define SOC15_DPG_MODE_OFFSET(ip, inst_idx, reg) \ ({ \ - uint32_t internal_reg_offset, addr; \ + /* To avoid a -Wunused-but-set-variable warning. */ \ + uint32_t internal_reg_offset __maybe_unused, addr; \ bool video_range, video1_range, aon_range, aon1_range; \ \ addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg); \ @@ -161,7 +162,8 @@ #define SOC24_DPG_MODE_OFFSET(ip, inst_idx, reg) \ ({ \ - uint32_t internal_reg_offset, addr; \ + /* To avoid a -Wunused-but-set-variable warning. */ \ + uint32_t internal_reg_offset __maybe_unused, addr; \ bool video_range, video1_range, aon_range, aon1_range; \ \ addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg); \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 3328ab63376b..47a6ce4fdc74 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -44,6 +44,18 @@ vf2pf_info->ucode_info[ucode].version = ver; \ } while (0) +#define mmRCC_CONFIG_MEMSIZE 0xde3 + +const char *amdgpu_virt_dynamic_crit_table_name[] = { + "IP DISCOVERY", + "VBIOS IMG", + "RAS TELEMETRY", + "DATA EXCHANGE", + "BAD PAGE INFO", + "INIT HEADER", + "LAST", +}; + bool amdgpu_virt_mmio_blocked(struct amdgpu_device *adev) { /* By now all MMIO pages except mailbox are blocked */ @@ -150,9 +162,10 @@ void amdgpu_virt_request_init_data(struct amdgpu_device *adev) virt->ops->req_init_data(adev); if (adev->virt.req_init_data_ver > 0) - DRM_INFO("host supports REQ_INIT_DATA handshake\n"); + dev_info(adev->dev, "host supports REQ_INIT_DATA handshake of critical_region_version %d\n", + adev->virt.req_init_data_ver); else - DRM_WARN("host doesn't support REQ_INIT_DATA handshake\n"); + dev_warn(adev->dev, "host doesn't support REQ_INIT_DATA handshake\n"); } /** @@ -205,12 +218,12 @@ int amdgpu_virt_alloc_mm_table(struct amdgpu_device *adev) &adev->virt.mm_table.gpu_addr, (void *)&adev->virt.mm_table.cpu_addr); if (r) { - DRM_ERROR("failed to alloc mm table and error = %d.\n", r); + dev_err(adev->dev, "failed to alloc mm table and error = %d.\n", r); return r; } memset((void *)adev->virt.mm_table.cpu_addr, 0, PAGE_SIZE); - DRM_INFO("MM table gpu addr = 0x%llx, cpu addr = %p.\n", + dev_info(adev->dev, "MM table gpu addr = 0x%llx, cpu addr = %p.\n", adev->virt.mm_table.gpu_addr, adev->virt.mm_table.cpu_addr); return 0; @@ -390,7 +403,9 @@ static void amdgpu_virt_ras_reserve_bps(struct amdgpu_device *adev) if (amdgpu_bo_create_kernel_at(adev, bp << AMDGPU_GPU_PAGE_SHIFT, AMDGPU_GPU_PAGE_SIZE, &bo, NULL)) - DRM_DEBUG("RAS WARN: reserve vram for retired page %llx fail\n", bp); + dev_dbg(adev->dev, + "RAS WARN: reserve vram for retired page %llx fail\n", + bp); data->bps_bo[i] = bo; } data->last_reserved = i + 1; @@ -598,8 +613,8 @@ static int amdgpu_virt_write_vf2pf_data(struct amdgpu_device *adev) vf2pf_info->driver_cert = 0; vf2pf_info->os_info.all = 0; - vf2pf_info->fb_usage = - ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) >> 20; + vf2pf_info->fb_usage = ttm_resource_manager_used(&adev->mman.vram_mgr.manager) ? + ttm_resource_manager_usage(&adev->mman.vram_mgr.manager) >> 20 : 0; vf2pf_info->fb_vis_usage = amdgpu_vram_mgr_vis_usage(&adev->mman.vram_mgr) >> 20; vf2pf_info->fb_size = adev->gmc.real_vram_size >> 20; @@ -658,10 +673,34 @@ out: schedule_delayed_work(&(adev->virt.vf2pf_work), adev->virt.vf2pf_update_interval_ms); } +static int amdgpu_virt_read_exchange_data_from_mem(struct amdgpu_device *adev, uint32_t *pfvf_data) +{ + uint32_t dataexchange_offset = + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset; + uint32_t dataexchange_size = + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb << 10; + uint64_t pos = 0; + + dev_info(adev->dev, + "Got data exchange info from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n", + dataexchange_offset, dataexchange_size); + + if (!IS_ALIGNED(dataexchange_offset, 4) || !IS_ALIGNED(dataexchange_size, 4)) { + dev_err(adev->dev, "Data exchange data not aligned to 4 bytes\n"); + return -EINVAL; + } + + pos = (uint64_t)dataexchange_offset; + amdgpu_device_vram_access(adev, pos, pfvf_data, + dataexchange_size, false); + + return 0; +} + void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev) { if (adev->virt.vf2pf_update_interval_ms != 0) { - DRM_INFO("clean up the vf2pf work item\n"); + dev_info(adev->dev, "clean up the vf2pf work item\n"); cancel_delayed_work_sync(&adev->virt.vf2pf_work); adev->virt.vf2pf_update_interval_ms = 0; } @@ -669,13 +708,15 @@ void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev) void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) { + uint32_t *pfvf_data = NULL; + adev->virt.fw_reserve.p_pf2vf = NULL; adev->virt.fw_reserve.p_vf2pf = NULL; adev->virt.vf2pf_update_interval_ms = 0; adev->virt.vf2pf_update_retry_cnt = 0; if (adev->mman.fw_vram_usage_va && adev->mman.drv_vram_usage_va) { - DRM_WARN("Currently fw_vram and drv_vram should not have values at the same time!"); + dev_warn(adev->dev, "Currently fw_vram and drv_vram should not have values at the same time!"); } else if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) { /* go through this logic in ip_init and reset to init workqueue*/ amdgpu_virt_exchange_data(adev); @@ -684,11 +725,34 @@ void amdgpu_virt_init_data_exchange(struct amdgpu_device *adev) schedule_delayed_work(&(adev->virt.vf2pf_work), msecs_to_jiffies(adev->virt.vf2pf_update_interval_ms)); } else if (adev->bios != NULL) { /* got through this logic in early init stage to get necessary flags, e.g. rlcg_acc related*/ - adev->virt.fw_reserve.p_pf2vf = - (struct amd_sriov_msg_pf2vf_info_header *) - (adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10)); + if (adev->virt.req_init_data_ver == GPU_CRIT_REGION_V2) { + pfvf_data = + kzalloc(adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb << 10, + GFP_KERNEL); + if (!pfvf_data) { + dev_err(adev->dev, "Failed to allocate memory for pfvf_data\n"); + return; + } - amdgpu_virt_read_pf2vf_data(adev); + if (amdgpu_virt_read_exchange_data_from_mem(adev, pfvf_data)) + goto free_pfvf_data; + + adev->virt.fw_reserve.p_pf2vf = + (struct amd_sriov_msg_pf2vf_info_header *)pfvf_data; + + amdgpu_virt_read_pf2vf_data(adev); + +free_pfvf_data: + kfree(pfvf_data); + pfvf_data = NULL; + adev->virt.fw_reserve.p_pf2vf = NULL; + } else { + adev->virt.fw_reserve.p_pf2vf = + (struct amd_sriov_msg_pf2vf_info_header *) + (adev->bios + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10)); + + amdgpu_virt_read_pf2vf_data(adev); + } } } @@ -701,23 +765,38 @@ void amdgpu_virt_exchange_data(struct amdgpu_device *adev) if (adev->mman.fw_vram_usage_va || adev->mman.drv_vram_usage_va) { if (adev->mman.fw_vram_usage_va) { - adev->virt.fw_reserve.p_pf2vf = - (struct amd_sriov_msg_pf2vf_info_header *) - (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10)); - adev->virt.fw_reserve.p_vf2pf = - (struct amd_sriov_msg_vf2pf_info_header *) - (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10)); - adev->virt.fw_reserve.ras_telemetry = - (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10)); + if (adev->virt.req_init_data_ver == GPU_CRIT_REGION_V2) { + adev->virt.fw_reserve.p_pf2vf = + (struct amd_sriov_msg_pf2vf_info_header *) + (adev->mman.fw_vram_usage_va + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset); + adev->virt.fw_reserve.p_vf2pf = + (struct amd_sriov_msg_vf2pf_info_header *) + (adev->mman.fw_vram_usage_va + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset + + (AMD_SRIOV_MSG_SIZE_KB << 10)); + adev->virt.fw_reserve.ras_telemetry = + (adev->mman.fw_vram_usage_va + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].offset); + } else { + adev->virt.fw_reserve.p_pf2vf = + (struct amd_sriov_msg_pf2vf_info_header *) + (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10)); + adev->virt.fw_reserve.p_vf2pf = + (struct amd_sriov_msg_vf2pf_info_header *) + (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 << 10)); + adev->virt.fw_reserve.ras_telemetry = + (adev->mman.fw_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1 << 10)); + } } else if (adev->mman.drv_vram_usage_va) { adev->virt.fw_reserve.p_pf2vf = (struct amd_sriov_msg_pf2vf_info_header *) - (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB << 10)); + (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 << 10)); adev->virt.fw_reserve.p_vf2pf = (struct amd_sriov_msg_vf2pf_info_header *) - (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB << 10)); + (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 << 10)); adev->virt.fw_reserve.ras_telemetry = - (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB << 10)); + (adev->mman.drv_vram_usage_va + (AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1 << 10)); } amdgpu_virt_read_pf2vf_data(adev); @@ -816,7 +895,7 @@ static bool amdgpu_virt_init_req_data(struct amdgpu_device *adev, u32 reg) break; default: /* other chip doesn't support SRIOV */ is_sriov = false; - DRM_ERROR("Unknown asic type: %d!\n", adev->asic_type); + dev_err(adev->dev, "Unknown asic type: %d!\n", adev->asic_type); break; } } @@ -838,10 +917,220 @@ static void amdgpu_virt_init_ras(struct amdgpu_device *adev) RATELIMIT_MSG_ON_RELEASE); mutex_init(&adev->virt.ras.ras_telemetry_mutex); + mutex_init(&adev->virt.access_req_mutex); adev->virt.ras.cper_rptr = 0; } +static uint8_t amdgpu_virt_crit_region_calc_checksum(uint8_t *buf_start, uint8_t *buf_end) +{ + uint32_t sum = 0; + + if (buf_start >= buf_end) + return 0; + + for (; buf_start < buf_end; buf_start++) + sum += buf_start[0]; + + return 0xffffffff - sum; +} + +int amdgpu_virt_init_critical_region(struct amdgpu_device *adev) +{ + struct amd_sriov_msg_init_data_header *init_data_hdr = NULL; + u64 init_hdr_offset = adev->virt.init_data_header.offset; + u64 init_hdr_size = (u64)adev->virt.init_data_header.size_kb << 10; /* KB → bytes */ + u64 vram_size; + u64 end; + int r = 0; + uint8_t checksum = 0; + + /* Skip below init if critical region version != v2 */ + if (adev->virt.req_init_data_ver != GPU_CRIT_REGION_V2) + return 0; + + if (init_hdr_offset < 0) { + dev_err(adev->dev, "Invalid init header offset\n"); + return -EINVAL; + } + + vram_size = RREG32(mmRCC_CONFIG_MEMSIZE); + if (!vram_size || vram_size == U32_MAX) + return -EINVAL; + vram_size <<= 20; + + if (check_add_overflow(init_hdr_offset, init_hdr_size, &end) || end > vram_size) { + dev_err(adev->dev, "init_data_header exceeds VRAM size, exiting\n"); + return -EINVAL; + } + + /* Allocate for init_data_hdr */ + init_data_hdr = kzalloc(sizeof(struct amd_sriov_msg_init_data_header), GFP_KERNEL); + if (!init_data_hdr) + return -ENOMEM; + + amdgpu_device_vram_access(adev, (uint64_t)init_hdr_offset, (uint32_t *)init_data_hdr, + sizeof(struct amd_sriov_msg_init_data_header), false); + + /* Table validation */ + if (strncmp(init_data_hdr->signature, + AMDGPU_SRIOV_CRIT_DATA_SIGNATURE, + AMDGPU_SRIOV_CRIT_DATA_SIG_LEN) != 0) { + dev_err(adev->dev, "Invalid init data signature: %.4s\n", + init_data_hdr->signature); + r = -EINVAL; + goto out; + } + + checksum = amdgpu_virt_crit_region_calc_checksum( + (uint8_t *)&init_data_hdr->initdata_offset, + (uint8_t *)init_data_hdr + + sizeof(struct amd_sriov_msg_init_data_header)); + if (checksum != init_data_hdr->checksum) { + dev_err(adev->dev, "Found unmatching checksum from calculation 0x%x and init_data 0x%x\n", + checksum, init_data_hdr->checksum); + r = -EINVAL; + goto out; + } + + memset(&adev->virt.crit_regn, 0, sizeof(adev->virt.crit_regn)); + memset(adev->virt.crit_regn_tbl, 0, sizeof(adev->virt.crit_regn_tbl)); + + adev->virt.crit_regn.offset = init_data_hdr->initdata_offset; + adev->virt.crit_regn.size_kb = init_data_hdr->initdata_size_in_kb; + + /* Validation and initialization for each table entry */ + if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_IPD_TABLE_ID)) { + if (!init_data_hdr->ip_discovery_size_in_kb || + init_data_hdr->ip_discovery_size_in_kb > DISCOVERY_TMR_SIZE) { + dev_err(adev->dev, "Invalid %s size: 0x%x\n", + amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_IPD_TABLE_ID], + init_data_hdr->ip_discovery_size_in_kb); + r = -EINVAL; + goto out; + } + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].offset = + init_data_hdr->ip_discovery_offset; + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb = + init_data_hdr->ip_discovery_size_in_kb; + } + + if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID)) { + if (!init_data_hdr->vbios_img_size_in_kb) { + dev_err(adev->dev, "Invalid %s size: 0x%x\n", + amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID], + init_data_hdr->vbios_img_size_in_kb); + r = -EINVAL; + goto out; + } + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID].offset = + init_data_hdr->vbios_img_offset; + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID].size_kb = + init_data_hdr->vbios_img_size_in_kb; + } + + if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID)) { + if (!init_data_hdr->ras_tele_info_size_in_kb) { + dev_err(adev->dev, "Invalid %s size: 0x%x\n", + amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID], + init_data_hdr->ras_tele_info_size_in_kb); + r = -EINVAL; + goto out; + } + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].offset = + init_data_hdr->ras_tele_info_offset; + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID].size_kb = + init_data_hdr->ras_tele_info_size_in_kb; + } + + if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID)) { + if (!init_data_hdr->dataexchange_size_in_kb) { + dev_err(adev->dev, "Invalid %s size: 0x%x\n", + amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID], + init_data_hdr->dataexchange_size_in_kb); + r = -EINVAL; + goto out; + } + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].offset = + init_data_hdr->dataexchange_offset; + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID].size_kb = + init_data_hdr->dataexchange_size_in_kb; + } + + if (IS_SRIOV_CRIT_REGN_ENTRY_VALID(init_data_hdr, AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID)) { + if (!init_data_hdr->bad_page_size_in_kb) { + dev_err(adev->dev, "Invalid %s size: 0x%x\n", + amdgpu_virt_dynamic_crit_table_name[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID], + init_data_hdr->bad_page_size_in_kb); + r = -EINVAL; + goto out; + } + + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID].offset = + init_data_hdr->bad_page_info_offset; + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID].size_kb = + init_data_hdr->bad_page_size_in_kb; + } + + /* Validation for critical region info */ + if (adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb > DISCOVERY_TMR_SIZE) { + dev_err(adev->dev, "Invalid IP discovery size: 0x%x\n", + adev->virt.crit_regn_tbl[AMD_SRIOV_MSG_IPD_TABLE_ID].size_kb); + r = -EINVAL; + goto out; + } + + /* reserved memory starts from crit region base offset with the size of 5MB */ + adev->mman.fw_vram_usage_start_offset = adev->virt.crit_regn.offset; + adev->mman.fw_vram_usage_size = adev->virt.crit_regn.size_kb << 10; + dev_info(adev->dev, + "critical region v%d requested to reserve memory start at %08llx with %llu KB.\n", + init_data_hdr->version, + adev->mman.fw_vram_usage_start_offset, + adev->mman.fw_vram_usage_size >> 10); + + adev->virt.is_dynamic_crit_regn_enabled = true; + +out: + kfree(init_data_hdr); + init_data_hdr = NULL; + + return r; +} + +int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev, + int data_id, uint8_t *binary, u32 *size) +{ + uint32_t data_offset = 0; + uint32_t data_size = 0; + enum amd_sriov_msg_table_id_enum data_table_id = data_id; + + if (data_table_id >= AMD_SRIOV_MSG_MAX_TABLE_ID) + return -EINVAL; + + data_offset = adev->virt.crit_regn_tbl[data_table_id].offset; + data_size = adev->virt.crit_regn_tbl[data_table_id].size_kb << 10; + + /* Validate on input params */ + if (!binary || !size || *size < (uint64_t)data_size) + return -EINVAL; + + /* Proceed to copy the dynamic content */ + amdgpu_device_vram_access(adev, + (uint64_t)data_offset, (uint32_t *)binary, data_size, false); + *size = (uint64_t)data_size; + + dev_dbg(adev->dev, + "Got %s info from dynamic crit_region_table at offset 0x%x with size of 0x%x bytes.\n", + amdgpu_virt_dynamic_crit_table_name[data_id], data_offset, data_size); + + return 0; +} + void amdgpu_virt_init(struct amdgpu_device *adev) { bool is_sriov = false; @@ -1289,7 +1578,7 @@ amdgpu_ras_block_to_sriov(struct amdgpu_device *adev, enum amdgpu_ras_block bloc case AMDGPU_RAS_BLOCK__MPIO: return RAS_TELEMETRY_GPU_BLOCK_MPIO; default: - DRM_WARN_ONCE("Unsupported SRIOV RAS telemetry block 0x%x\n", + dev_warn(adev->dev, "Unsupported SRIOV RAS telemetry block 0x%x\n", block); return RAS_TELEMETRY_GPU_BLOCK_COUNT; } @@ -1304,7 +1593,7 @@ static int amdgpu_virt_cache_host_error_counts(struct amdgpu_device *adev, checksum = host_telemetry->header.checksum; used_size = host_telemetry->header.used_size; - if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10)) + if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10)) return 0; tmp = kmemdup(&host_telemetry->body.error_count, used_size, GFP_KERNEL); @@ -1383,7 +1672,7 @@ amdgpu_virt_write_cpers_to_ring(struct amdgpu_device *adev, checksum = host_telemetry->header.checksum; used_size = host_telemetry->header.used_size; - if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10)) + if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10)) return -EINVAL; cper_dump = kmemdup(&host_telemetry->body.cper_dump, used_size, GFP_KERNEL); @@ -1515,7 +1804,7 @@ static int amdgpu_virt_cache_chk_criti_hit(struct amdgpu_device *adev, checksum = host_telemetry->header.checksum; used_size = host_telemetry->header.used_size; - if (used_size > (AMD_SRIOV_RAS_TELEMETRY_SIZE_KB << 10)) + if (used_size > (AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 << 10)) return 0; tmp = kmemdup(&host_telemetry->body.chk_criti, used_size, GFP_KERNEL); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index d1172c8e58c4..01d5bca2dee1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -54,6 +54,12 @@ #define AMDGPU_VF2PF_UPDATE_MAX_RETRY_LIMIT 2 +/* Signature used to validate the SR-IOV dynamic critical region init data header ("INDA") */ +#define AMDGPU_SRIOV_CRIT_DATA_SIGNATURE "INDA" +#define AMDGPU_SRIOV_CRIT_DATA_SIG_LEN 4 + +#define IS_SRIOV_CRIT_REGN_ENTRY_VALID(hdr, id) ((hdr)->valid_tables & (1 << (id))) + enum amdgpu_sriov_vf_mode { SRIOV_VF_MODE_BARE_METAL = 0, SRIOV_VF_MODE_ONE_VF, @@ -144,6 +150,7 @@ enum AMDGIM_FEATURE_FLAG { AMDGIM_FEATURE_RAS_CAPS = (1 << 9), AMDGIM_FEATURE_RAS_TELEMETRY = (1 << 10), AMDGIM_FEATURE_RAS_CPER = (1 << 11), + AMDGIM_FEATURE_XGMI_TA_EXT_PEER_LINK = (1 << 12), }; enum AMDGIM_REG_ACCESS_FLAG { @@ -262,6 +269,11 @@ struct amdgpu_virt_ras { DECLARE_ATTR_CAP_CLASS(amdgpu_virt, AMDGPU_VIRT_CAPS_LIST); +struct amdgpu_virt_region { + uint32_t offset; + uint32_t size_kb; +}; + /* GPU virtualization */ struct amdgpu_virt { uint32_t caps; @@ -289,6 +301,12 @@ struct amdgpu_virt { bool ras_init_done; uint32_t reg_access; + /* dynamic(v2) critical regions */ + struct amdgpu_virt_region init_data_header; + struct amdgpu_virt_region crit_regn; + struct amdgpu_virt_region crit_regn_tbl[AMD_SRIOV_MSG_MAX_TABLE_ID]; + bool is_dynamic_crit_regn_enabled; + /* vf2pf message */ struct delayed_work vf2pf_work; uint32_t vf2pf_update_interval_ms; @@ -307,6 +325,8 @@ struct amdgpu_virt { /* Spinlock to protect access to the RLCG register interface */ spinlock_t rlcg_reg_lock; + struct mutex access_req_mutex; + union amd_sriov_ras_caps ras_en_caps; union amd_sriov_ras_caps ras_telemetry_en_caps; struct amdgpu_virt_ras ras; @@ -378,6 +398,9 @@ struct amdgpu_video_codec_info; #define amdgpu_sriov_ras_cper_en(adev) \ ((adev)->virt.gim_feature & AMDGIM_FEATURE_RAS_CPER) +#define amdgpu_sriov_xgmi_ta_ext_peer_link_en(adev) \ +((adev)->virt.gim_feature & AMDGIM_FEATURE_XGMI_TA_EXT_PEER_LINK) + static inline bool is_virtual_machine(void) { #if defined(CONFIG_X86) @@ -424,6 +447,10 @@ void amdgpu_virt_exchange_data(struct amdgpu_device *adev); void amdgpu_virt_fini_data_exchange(struct amdgpu_device *adev); void amdgpu_virt_init(struct amdgpu_device *adev); +int amdgpu_virt_init_critical_region(struct amdgpu_device *adev); +int amdgpu_virt_get_dynamic_data_info(struct amdgpu_device *adev, + int data_id, uint8_t *binary, u32 *size); + bool amdgpu_virt_can_access_debugfs(struct amdgpu_device *adev); int amdgpu_virt_enable_access_debugfs(struct amdgpu_device *adev); void amdgpu_virt_disable_access_debugfs(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index c1a801203949..a67285118c37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -484,15 +484,19 @@ int amdgpu_vm_lock_done_list(struct amdgpu_vm *vm, struct drm_exec *exec, spin_lock(&vm->status_lock); while (!list_is_head(prev->next, &vm->done)) { bo_va = list_entry(prev->next, typeof(*bo_va), base.vm_status); - spin_unlock(&vm->status_lock); bo = bo_va->base.bo; if (bo) { + amdgpu_bo_ref(bo); + spin_unlock(&vm->status_lock); + ret = drm_exec_prepare_obj(exec, &bo->tbo.base, 1); + amdgpu_bo_unref(&bo); if (unlikely(ret)) return ret; + + spin_lock(&vm->status_lock); } - spin_lock(&vm->status_lock); prev = prev->next; } spin_unlock(&vm->status_lock); @@ -779,7 +783,6 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool cleaner_shader_needed = false; bool pasid_mapping_needed = false; struct dma_fence *fence = NULL; - struct amdgpu_fence *af; unsigned int patch; int r; @@ -842,12 +845,12 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, } if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) { - r = amdgpu_fence_emit(ring, &fence, NULL, 0); + r = amdgpu_fence_emit(ring, job->hw_vm_fence, 0); if (r) return r; - /* this is part of the job's context */ - af = container_of(fence, struct amdgpu_fence, base); - af->context = job->base.s_fence ? job->base.s_fence->finished.context : 0; + fence = &job->hw_vm_fence->base; + /* get a ref for the job */ + dma_fence_get(fence); } if (vm_flush_needed) { @@ -1066,7 +1069,7 @@ amdgpu_vm_tlb_flush(struct amdgpu_vm_update_params *params, } /* Prepare a TLB flush fence to be attached to PTs */ - if (!params->unlocked && vm->is_compute_context) { + if (!params->unlocked) { amdgpu_vm_tlb_fence_create(params->adev, vm, fence); /* Makes sure no PD/PT is freed before the flush */ @@ -1952,6 +1955,7 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev, struct amdgpu_bo_va_mapping *mapping; struct amdgpu_vm *vm = bo_va->base.vm; bool valid = true; + int r; saddr /= AMDGPU_GPU_PAGE_SIZE; @@ -1972,6 +1976,17 @@ int amdgpu_vm_bo_unmap(struct amdgpu_device *adev, return -ENOENT; } + /* It's unlikely to happen that the mapping userq hasn't been idled + * during user requests GEM unmap IOCTL except for forcing the unmap + * from user space. + */ + if (unlikely(atomic_read(&bo_va->userq_va_mapped) > 0)) { + r = amdgpu_userq_gem_va_unmap_validate(adev, mapping, saddr); + if (unlikely(r == -EBUSY)) + dev_warn_once(adev->dev, + "Attempt to unmap an active userq buffer\n"); + } + list_del(&mapping->list); amdgpu_vm_it_remove(mapping, &vm->va); mapping->bo_va = NULL; @@ -2078,7 +2093,7 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, struct amdgpu_bo *bo = before->bo_va->base.bo; amdgpu_vm_it_insert(before, &vm->va); - if (before->flags & AMDGPU_PTE_PRT_FLAG(adev)) + if (before->flags & AMDGPU_VM_PAGE_PRT) amdgpu_vm_prt_get(adev); if (amdgpu_vm_is_bo_always_valid(vm, bo) && @@ -2093,7 +2108,7 @@ int amdgpu_vm_bo_clear_mappings(struct amdgpu_device *adev, struct amdgpu_bo *bo = after->bo_va->base.bo; amdgpu_vm_it_insert(after, &vm->va); - if (after->flags & AMDGPU_PTE_PRT_FLAG(adev)) + if (after->flags & AMDGPU_VM_PAGE_PRT) amdgpu_vm_prt_get(adev); if (amdgpu_vm_is_bo_always_valid(vm, bo) && @@ -2828,8 +2843,6 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm) */ void amdgpu_vm_manager_init(struct amdgpu_device *adev) { - unsigned i; - /* Concurrent flushes are only possible starting with Vega10 and * are broken on Navi10 and Navi14. */ @@ -2838,11 +2851,6 @@ void amdgpu_vm_manager_init(struct amdgpu_device *adev) adev->asic_type == CHIP_NAVI14); amdgpu_vmid_mgr_init(adev); - adev->vm_manager.fence_context = - dma_fence_context_alloc(AMDGPU_MAX_RINGS); - for (i = 0; i < AMDGPU_MAX_RINGS; ++i) - adev->vm_manager.seqno[i] = 0; - spin_lock_init(&adev->vm_manager.prt_lock); atomic_set(&adev->vm_manager.num_prt_users, 0); @@ -2908,8 +2916,7 @@ int amdgpu_vm_ioctl(struct drm_device *dev, void *data, struct drm_file *filp) switch (args->in.op) { case AMDGPU_VM_OP_RESERVE_VMID: /* We only have requirement to reserve vmid from gfxhub */ - amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0)); - break; + return amdgpu_vmid_alloc_reserved(adev, vm, AMDGPU_GFXHUB(0)); case AMDGPU_VM_OP_UNRESERVE_VMID: amdgpu_vmid_free_reserved(adev, vm, AMDGPU_GFXHUB(0)); break; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index cf0ec94e8a07..15d757c016cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -453,10 +453,6 @@ struct amdgpu_vm_manager { unsigned int first_kfd_vmid; bool concurrent_flush; - /* Handling of VM fences */ - u64 fence_context; - unsigned seqno[AMDGPU_MAX_RINGS]; - uint64_t max_pfn; uint32_t num_level; uint32_t block_size; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c index 474bfe36c0c2..aa78c2ee9e21 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vpe.c @@ -322,6 +322,26 @@ static int vpe_early_init(struct amdgpu_ip_block *ip_block) return 0; } +static bool vpe_need_dpm0_at_power_down(struct amdgpu_device *adev) +{ + switch (amdgpu_ip_version(adev, VPE_HWIP, 0)) { + case IP_VERSION(6, 1, 1): + return adev->pm.fw_version < 0x0a640500; + default: + return false; + } +} + +static int vpe_get_dpm_level(struct amdgpu_device *adev) +{ + struct amdgpu_vpe *vpe = &adev->vpe; + + if (!adev->pm.dpm_enabled) + return 0; + + return RREG32(vpe_get_reg_offset(vpe, 0, vpe->regs.dpm_request_lv)); +} + static void vpe_idle_work_handler(struct work_struct *work) { struct amdgpu_device *adev = @@ -329,11 +349,17 @@ static void vpe_idle_work_handler(struct work_struct *work) unsigned int fences = 0; fences += amdgpu_fence_count_emitted(&adev->vpe.ring); + if (fences) + goto reschedule; - if (fences == 0) - amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE); - else - schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT); + if (vpe_need_dpm0_at_power_down(adev) && vpe_get_dpm_level(adev) != 0) + goto reschedule; + + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VPE, AMD_PG_STATE_GATE); + return; + +reschedule: + schedule_delayed_work(&adev->vpe.idle_work, VPE_IDLE_TIMEOUT); } static int vpe_common_init(struct amdgpu_vpe *vpe) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index a5adb2ed9b3c..9d934c07fa6b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -234,6 +234,9 @@ static umode_t amdgpu_vram_attrs_is_visible(struct kobject *kobj, !adev->gmc.vram_vendor) return 0; + if (!ttm_resource_manager_used(&adev->mman.vram_mgr.manager)) + return 0; + return attr->mode; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index 1ede308a7c67..aad530c46a9f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -298,6 +298,9 @@ int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num) { int link_map_6_4_x[8] = { 0, 3, 1, 2, 7, 6, 4, 5 }; + if (adev->gmc.xgmi.num_physical_nodes <= 1) + return -EINVAL; + switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) { case IP_VERSION(6, 4, 0): case IP_VERSION(6, 4, 1): @@ -333,6 +336,10 @@ static u32 xgmi_v6_4_get_link_status(struct amdgpu_device *adev, int global_link } i = global_link_num / n; + + if (!(adev->aid_mask & BIT(i))) + return U32_MAX; + addr += adev->asic_funcs->encode_ext_smn_addressing(i); return RREG32_PCIE_EXT(addr); @@ -342,6 +349,9 @@ int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev, int global_link_num) { u32 xgmi_state_reg_val; + if (adev->gmc.xgmi.num_physical_nodes <= 1) + return -EINVAL; + switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) { case IP_VERSION(6, 4, 0): case IP_VERSION(6, 4, 1): @@ -958,28 +968,6 @@ static int amdgpu_xgmi_initialize_hive_get_data_partition(struct amdgpu_hive_inf return 0; } -static void amdgpu_xgmi_fill_topology_info(struct amdgpu_device *adev, - struct amdgpu_device *peer_adev) -{ - struct psp_xgmi_topology_info *top_info = &adev->psp.xgmi_context.top_info; - struct psp_xgmi_topology_info *peer_info = &peer_adev->psp.xgmi_context.top_info; - - for (int i = 0; i < peer_info->num_nodes; i++) { - if (peer_info->nodes[i].node_id == adev->gmc.xgmi.node_id) { - for (int j = 0; j < top_info->num_nodes; j++) { - if (top_info->nodes[j].node_id == peer_adev->gmc.xgmi.node_id) { - peer_info->nodes[i].num_hops = top_info->nodes[j].num_hops; - peer_info->nodes[i].is_sharing_enabled = - top_info->nodes[j].is_sharing_enabled; - peer_info->nodes[i].num_links = - top_info->nodes[j].num_links; - return; - } - } - } - } -} - int amdgpu_xgmi_add_device(struct amdgpu_device *adev) { struct psp_xgmi_topology_info *top_info; @@ -1065,11 +1053,6 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev) /* To do: continue with some node failed or disable the whole hive*/ goto exit_unlock; } - - /* fill the topology info for peers instead of getting from PSP */ - list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) { - amdgpu_xgmi_fill_topology_info(adev, tmp_adev); - } } else { /* get latest topology info for each device from psp */ list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h index 3a79ed7d8031..3cdb1e0eca37 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgv_sriovmsg.h @@ -23,26 +23,84 @@ #ifndef AMDGV_SRIOV_MSG__H_ #define AMDGV_SRIOV_MSG__H_ -/* unit in kilobytes */ -#define AMD_SRIOV_MSG_VBIOS_OFFSET 0 -#define AMD_SRIOV_MSG_VBIOS_SIZE_KB 64 -#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB AMD_SRIOV_MSG_VBIOS_SIZE_KB -#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB 4 -#define AMD_SRIOV_MSG_TMR_OFFSET_KB 2048 -#define AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB 2 -#define AMD_SRIOV_RAS_TELEMETRY_SIZE_KB 64 +#define AMD_SRIOV_MSG_SIZE_KB 1 + /* - * layout + * layout v1 * 0 64KB 65KB 66KB 68KB 132KB * | VBIOS | PF2VF | VF2PF | Bad Page | RAS Telemetry Region | ... * | 64KB | 1KB | 1KB | 2KB | 64KB | ... */ -#define AMD_SRIOV_MSG_SIZE_KB 1 -#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB -#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB (AMD_SRIOV_MSG_PF2VF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB) -#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB (AMD_SRIOV_MSG_VF2PF_OFFSET_KB + AMD_SRIOV_MSG_SIZE_KB) -#define AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB (AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB) +/* + * layout v2 (offsets are dynamically allocated and the offsets below are examples) + * 0 1KB 64KB 65KB 66KB 68KB 132KB + * | INITD_H | VBIOS | PF2VF | VF2PF | Bad Page | RAS Telemetry Region | ... + * | 1KB | 64KB | 1KB | 1KB | 2KB | 64KB | ... + * + * Note: PF2VF + VF2PF + Bad Page = DataExchange region (allocated contiguously) + */ + +/* v1 layout sizes */ +#define AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1 64 +#define AMD_SRIOV_MSG_PF2VF_SIZE_KB_V1 1 +#define AMD_SRIOV_MSG_VF2PF_SIZE_KB_V1 1 +#define AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1 2 +#define AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1 64 +#define AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB_V1 \ + (AMD_SRIOV_MSG_PF2VF_SIZE_KB_V1 + AMD_SRIOV_MSG_VF2PF_SIZE_KB_V1 + \ + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1) + +/* v1 offsets */ +#define AMD_SRIOV_MSG_VBIOS_OFFSET_V1 0 +#define AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB_V1 AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1 +#define AMD_SRIOV_MSG_TMR_OFFSET_KB 2048 +#define AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 AMD_SRIOV_MSG_DATAEXCHANGE_OFFSET_KB_V1 +#define AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 \ + (AMD_SRIOV_MSG_PF2VF_OFFSET_KB_V1 + AMD_SRIOV_MSG_SIZE_KB) +#define AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB_V1 \ + (AMD_SRIOV_MSG_VF2PF_OFFSET_KB_V1 + AMD_SRIOV_MSG_SIZE_KB) +#define AMD_SRIOV_MSG_RAS_TELEMETRY_OFFSET_KB_V1 \ + (AMD_SRIOV_MSG_BAD_PAGE_OFFSET_KB_V1 + AMD_SRIOV_MSG_BAD_PAGE_SIZE_KB_V1) +#define AMD_SRIOV_MSG_INIT_DATA_TOT_SIZE_KB_V1 \ + (AMD_SRIOV_MSG_VBIOS_SIZE_KB_V1 + AMD_SRIOV_MSG_DATAEXCHANGE_SIZE_KB_V1 + \ + AMD_SRIOV_MSG_RAS_TELEMETRY_SIZE_KB_V1) + +enum amd_sriov_crit_region_version { + GPU_CRIT_REGION_V1 = 1, + GPU_CRIT_REGION_V2 = 2, +}; + +/* v2 layout offset enum (in order of allocation) */ +enum amd_sriov_msg_table_id_enum { + AMD_SRIOV_MSG_IPD_TABLE_ID = 0, + AMD_SRIOV_MSG_VBIOS_IMG_TABLE_ID, + AMD_SRIOV_MSG_RAS_TELEMETRY_TABLE_ID, + AMD_SRIOV_MSG_DATAEXCHANGE_TABLE_ID, + AMD_SRIOV_MSG_BAD_PAGE_INFO_TABLE_ID, + AMD_SRIOV_MSG_INITD_H_TABLE_ID, + AMD_SRIOV_MSG_MAX_TABLE_ID, +}; + +struct amd_sriov_msg_init_data_header { + char signature[4]; /* "INDA" */ + uint32_t version; + uint32_t checksum; + uint32_t initdata_offset; /* 0 */ + uint32_t initdata_size_in_kb; /* 5MB */ + uint32_t valid_tables; + uint32_t vbios_img_offset; + uint32_t vbios_img_size_in_kb; + uint32_t dataexchange_offset; + uint32_t dataexchange_size_in_kb; + uint32_t ras_tele_info_offset; + uint32_t ras_tele_info_size_in_kb; + uint32_t ip_discovery_offset; + uint32_t ip_discovery_size_in_kb; + uint32_t bad_page_info_offset; + uint32_t bad_page_size_in_kb; + uint32_t reserved[8]; +}; /* * PF2VF history log: @@ -102,7 +160,8 @@ union amd_sriov_msg_feature_flags { uint32_t ras_caps : 1; uint32_t ras_telemetry : 1; uint32_t ras_cper : 1; - uint32_t reserved : 20; + uint32_t xgmi_ta_ext_peer_link : 1; + uint32_t reserved : 19; } flags; uint32_t all; }; @@ -140,8 +199,9 @@ union amd_sriov_ras_caps { uint64_t block_jpeg : 1; uint64_t block_ih : 1; uint64_t block_mpio : 1; + uint64_t block_mmsch : 1; uint64_t poison_propogation_mode : 1; - uint64_t reserved : 44; + uint64_t reserved : 43; } bits; uint64_t all; }; diff --git a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c index 811124ff88a8..f9e2edf5260b 100644 --- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c +++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram.c @@ -407,7 +407,8 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, return -EINVAL; } - if (adev->kfd.init_complete && !amdgpu_in_reset(adev)) + if (adev->kfd.init_complete && !amdgpu_in_reset(adev) && + !adev->in_suspend) flags |= AMDGPU_XCP_OPS_KFD; if (flags & AMDGPU_XCP_OPS_KFD) { diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c index 41f4705bdbbd..876a3256dba4 100644 --- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c @@ -156,6 +156,9 @@ static int cik_ih_irq_init(struct amdgpu_device *adev) /* enable irqs */ cik_ih_enable_interrupts(adev); + if (adev->irq.ih_soft.ring_size) + adev->irq.ih_soft.enabled = true; + return 0; } @@ -192,6 +195,9 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev, wptr = le32_to_cpu(*ih->wptr_cpu); + if (ih == &adev->irq.ih_soft) + goto out; + if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) { wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK; /* When a ring buffer overflow happen start parsing interrupt @@ -211,6 +217,8 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev, tmp &= ~IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK; WREG32(mmIH_RB_CNTL, tmp); } + +out: return (wptr & ih->ptr_mask); } @@ -306,6 +314,10 @@ static int cik_ih_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true); + if (r) + return r; + r = amdgpu_irq_init(adev); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c index 96616a865aac..ed1e25661706 100644 --- a/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c +++ b/drivers/gpu/drm/amd/amdgpu/cyan_skillfish_reg_init.c @@ -1,4 +1,4 @@ -// SPDX-License-Identifier: GPL-2.0 +// SPDX-License-Identifier: MIT /* * Copyright 2018 Advanced Micro Devices, Inc. * diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c index 2f891fb846d5..bc7a2e06ab5f 100644 --- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c @@ -157,6 +157,9 @@ static int cz_ih_irq_init(struct amdgpu_device *adev) /* enable interrupts */ cz_ih_enable_interrupts(adev); + if (adev->irq.ih_soft.ring_size) + adev->irq.ih_soft.enabled = true; + return 0; } @@ -194,6 +197,9 @@ static u32 cz_ih_get_wptr(struct amdgpu_device *adev, wptr = le32_to_cpu(*ih->wptr_cpu); + if (ih == &adev->irq.ih_soft) + goto out; + if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) goto out; @@ -297,6 +303,10 @@ static int cz_ih_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true); + if (r) + return r; + r = amdgpu_irq_init(adev); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 8841d7213de4..d75b9940f248 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -4956,7 +4956,8 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block) amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); adev->gfx.compute_supported_reset = amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); - if (!amdgpu_sriov_vf(adev)) { + if (!amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; } @@ -9951,6 +9952,7 @@ static const struct amdgpu_ring_funcs gfx_v10_0_ring_funcs_kiq = { .emit_wreg = gfx_v10_0_ring_emit_wreg, .emit_reg_wait = gfx_v10_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v10_0_ring_emit_reg_write_reg_wait, + .emit_hdp_flush = gfx_v10_0_ring_emit_hdp_flush, }; static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 66c47c466532..8a2ee2de390f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -1821,13 +1821,15 @@ static int gfx_v11_0_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(11, 0, 3): if ((adev->gfx.me_fw_version >= 2280) && (adev->gfx.mec_fw_version >= 2410) && - !amdgpu_sriov_vf(adev)) { + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; } break; default: - if (!amdgpu_sriov_vf(adev)) { + if (!amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; } @@ -2438,7 +2440,7 @@ static int gfx_v11_0_rlc_load_microcode(struct amdgpu_device *adev) if (version_minor == 3) gfx_v11_0_load_rlcp_rlcv_microcode(adev); } - + return 0; } @@ -3886,7 +3888,7 @@ static int gfx_v11_0_cp_compute_load_microcode(struct amdgpu_device *adev) } memcpy(fw, fw_data, fw_size); - + amdgpu_bo_kunmap(adev->gfx.mec.mec_fw_obj); amdgpu_bo_unreserve(adev->gfx.mec.mec_fw_obj); @@ -5862,8 +5864,6 @@ static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; - BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE); - header = PACKET3(PACKET3_INDIRECT_BUFFER, 2); control |= ib->length_dw | (vmid << 24); @@ -5874,9 +5874,9 @@ static void gfx_v11_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, if (flags & AMDGPU_IB_PREEMPTED) control |= INDIRECT_BUFFER_PRE_RESUME(1); - if (vmid) + if (vmid && !ring->adev->gfx.rs64_enable) gfx_v11_0_ring_emit_de_meta(ring, - (!amdgpu_sriov_vf(ring->adev) && flags & AMDGPU_IB_PREEMPTED) ? true : false); + !amdgpu_sriov_vf(ring->adev) && (flags & AMDGPU_IB_PREEMPTED)); } amdgpu_ring_write(ring, header); @@ -7320,6 +7320,7 @@ static const struct amdgpu_ring_funcs gfx_v11_0_ring_funcs_kiq = { .emit_wreg = gfx_v11_0_ring_emit_wreg, .emit_reg_wait = gfx_v11_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v11_0_ring_emit_reg_write_reg_wait, + .emit_hdp_flush = gfx_v11_0_ring_emit_hdp_flush, }; static void gfx_v11_0_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index 710ec9c34e43..d01d2712cf57 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -1548,7 +1548,8 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(12, 0, 1): if ((adev->gfx.me_fw_version >= 2660) && (adev->gfx.mec_fw_version >= 2920) && - !amdgpu_sriov_vf(adev)) { + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.gfx_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; } @@ -4419,8 +4420,6 @@ static void gfx_v12_0_ring_emit_ib_gfx(struct amdgpu_ring *ring, unsigned vmid = AMDGPU_JOB_GET_VMID(job); u32 header, control = 0; - BUG_ON(ib->flags & AMDGPU_IB_FLAG_CE); - header = PACKET3(PACKET3_INDIRECT_BUFFER, 2); control |= ib->length_dw | (vmid << 24); @@ -5597,6 +5596,7 @@ static const struct amdgpu_ring_funcs gfx_v12_0_ring_funcs_kiq = { .emit_wreg = gfx_v12_0_ring_emit_wreg, .emit_reg_wait = gfx_v12_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v12_0_ring_emit_reg_write_reg_wait, + .emit_hdp_flush = gfx_v12_0_ring_emit_hdp_flush, }; static void gfx_v12_0_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c index 7693b7953426..80565392313f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v6_0.c @@ -3102,6 +3102,11 @@ static int gfx_v6_0_sw_init(struct amdgpu_ip_block *ip_block) return r; } + adev->gfx.gfx_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); + adev->gfx.compute_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); + return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c index 5976ed55d9db..2b7aba22ecc1 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v7_0.c @@ -4399,6 +4399,11 @@ static int gfx_v7_0_sw_init(struct amdgpu_ip_block *ip_block) gfx_v7_0_gpu_early_init(adev); + adev->gfx.gfx_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); + adev->gfx.compute_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); + return r; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c index 0856ff65288c..1c87375e1dd5 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c @@ -2023,6 +2023,11 @@ static int gfx_v8_0_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + adev->gfx.gfx_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); + adev->gfx.compute_supported_reset = + amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); + return 0; } @@ -6939,6 +6944,7 @@ static const struct amdgpu_ring_funcs gfx_v8_0_ring_funcs_kiq = { .pad_ib = amdgpu_ring_generic_pad_ib, .emit_rreg = gfx_v8_0_ring_emit_rreg, .emit_wreg = gfx_v8_0_ring_emit_wreg, + .emit_hdp_flush = gfx_v8_0_ring_emit_hdp_flush, }; static void gfx_v8_0_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index dd19a97436db..0148d7ff34d9 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -2409,7 +2409,7 @@ static int gfx_v9_0_sw_init(struct amdgpu_ip_block *ip_block) amdgpu_get_soft_full_reset_mask(&adev->gfx.gfx_ring[0]); adev->gfx.compute_supported_reset = amdgpu_get_soft_full_reset_mask(&adev->gfx.compute_ring[0]); - if (!amdgpu_sriov_vf(adev)) + if (!amdgpu_sriov_vf(adev) && !adev->debug_disable_gpu_ring_reset) adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; r = amdgpu_gfx_kiq_init(adev, GFX9_MEC_HPD_SIZE, 0); @@ -7586,6 +7586,7 @@ static const struct amdgpu_ring_funcs gfx_v9_0_ring_funcs_kiq = { .emit_wreg = gfx_v9_0_ring_emit_wreg, .emit_reg_wait = gfx_v9_0_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v9_0_ring_emit_reg_write_reg_wait, + .emit_hdp_flush = gfx_v9_0_ring_emit_hdp_flush, }; static void gfx_v9_0_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index 77f9d5b9a556..cbb74ffc4792 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -1149,14 +1149,16 @@ static int gfx_v9_4_3_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(9, 4, 3): case IP_VERSION(9, 4, 4): if ((adev->gfx.mec_fw_version >= 155) && - !amdgpu_sriov_vf(adev)) { + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE; } break; case IP_VERSION(9, 5, 0): if ((adev->gfx.mec_fw_version >= 21) && - !amdgpu_sriov_vf(adev)) { + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) { adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; adev->gfx.compute_supported_reset |= AMDGPU_RESET_TYPE_PER_PIPE; } @@ -2152,7 +2154,8 @@ static int gfx_v9_4_3_xcc_kiq_init_queue(struct amdgpu_ring *ring, int xcc_id) return 0; } -static int gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, bool restore) +static void gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, + bool restore) { struct amdgpu_device *adev = ring->adev; struct v9_mqd *mqd = ring->mqd_ptr; @@ -2186,8 +2189,6 @@ static int gfx_v9_4_3_xcc_kcq_init_queue(struct amdgpu_ring *ring, int xcc_id, b atomic64_set((atomic64_t *)&adev->wb.wb[ring->wptr_offs], 0); amdgpu_ring_clear_ring(ring); } - - return 0; } static int gfx_v9_4_3_xcc_kcq_fini_register(struct amdgpu_device *adev, int xcc_id) @@ -2220,7 +2221,7 @@ static int gfx_v9_4_3_xcc_kiq_resume(struct amdgpu_device *adev, int xcc_id) static int gfx_v9_4_3_xcc_kcq_resume(struct amdgpu_device *adev, int xcc_id) { struct amdgpu_ring *ring; - int i, r; + int i; gfx_v9_4_3_xcc_cp_compute_enable(adev, true, xcc_id); @@ -2228,9 +2229,7 @@ static int gfx_v9_4_3_xcc_kcq_resume(struct amdgpu_device *adev, int xcc_id) ring = &adev->gfx.compute_ring[i + xcc_id * adev->gfx.num_compute_rings]; - r = gfx_v9_4_3_xcc_kcq_init_queue(ring, xcc_id, false); - if (r) - return r; + gfx_v9_4_3_xcc_kcq_init_queue(ring, xcc_id, false); } return amdgpu_gfx_enable_kcq(adev, xcc_id); @@ -2292,7 +2291,9 @@ static int gfx_v9_4_3_cp_resume(struct amdgpu_device *adev) r = amdgpu_xcp_init(adev->xcp_mgr, num_xcp, mode); } else { - if (amdgpu_xcp_query_partition_mode(adev->xcp_mgr, + if (adev->in_suspend) + amdgpu_xcp_restore_partition_mode(adev->xcp_mgr); + else if (amdgpu_xcp_query_partition_mode(adev->xcp_mgr, AMDGPU_XCP_FL_NONE) == AMDGPU_UNKNOWN_COMPUTE_PARTITION_MODE) r = amdgpu_xcp_switch_partition_mode( @@ -3605,11 +3606,8 @@ pipe_reset: return r; } - r = gfx_v9_4_3_xcc_kcq_init_queue(ring, ring->xcc_id, true); - if (r) { - dev_err(adev->dev, "fail to init kcq\n"); - return r; - } + gfx_v9_4_3_xcc_kcq_init_queue(ring, ring->xcc_id, true); + spin_lock_irqsave(&kiq->ring_lock, flags); r = amdgpu_ring_alloc(kiq_ring, kiq->pmf->map_queues_size); if (r) { @@ -4798,6 +4796,7 @@ static const struct amdgpu_ring_funcs gfx_v9_4_3_ring_funcs_kiq = { .emit_wreg = gfx_v9_4_3_ring_emit_wreg, .emit_reg_wait = gfx_v9_4_3_ring_emit_reg_wait, .emit_reg_write_reg_wait = gfx_v9_4_3_ring_emit_reg_write_reg_wait, + .emit_hdp_flush = gfx_v9_4_3_ring_emit_hdp_flush, }; static void gfx_v9_4_3_set_ring_funcs(struct amdgpu_device *adev) diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index d7499be8c4bf..ce6e04242c52 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -103,8 +103,10 @@ static int gmc_v10_0_process_interrupt(struct amdgpu_device *adev, uint32_t vmhub_index = entry->client_id == SOC15_IH_CLIENTID_VMC ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0); struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index]; - bool retry_fault = !!(entry->src_data[1] & 0x80); - bool write_fault = !!(entry->src_data[1] & 0x20); + bool retry_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY); + bool write_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE); struct amdgpu_task_info *task_info; uint32_t status = 0; u64 addr; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c index 7bc389d9f5c4..ba59ee8e398a 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c @@ -103,12 +103,41 @@ static int gmc_v11_0_process_interrupt(struct amdgpu_device *adev, uint32_t vmhub_index = entry->client_id == SOC21_IH_CLIENTID_VMC ? AMDGPU_MMHUB0(0) : AMDGPU_GFXHUB(0); struct amdgpu_vmhub *hub = &adev->vmhub[vmhub_index]; + bool retry_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY); + bool write_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE); uint32_t status = 0; u64 addr; addr = (u64)entry->src_data[0] << 12; addr |= ((u64)entry->src_data[1] & 0xf) << 44; + if (retry_fault) { + /* Returning 1 here also prevents sending the IV to the KFD */ + + /* Process it only if it's the first fault for this address */ + if (entry->ih != &adev->irq.ih_soft && + amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid, + entry->timestamp)) + return 1; + + /* Delegate it to a different ring if the hardware hasn't + * already done it. + */ + if (entry->ih == &adev->irq.ih) { + amdgpu_irq_delegate(adev, entry, 8); + return 1; + } + + /* Try to handle the recoverable page faults by filling page + * tables + */ + if (amdgpu_vm_handle_fault(adev, entry->pasid, 0, 0, addr, + entry->timestamp, write_fault)) + return 1; + } + if (!amdgpu_sriov_vf(adev)) { /* * Issue a dummy read to wait for the status register to diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c index f4a19357ccbc..7a9d6894e321 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c @@ -91,6 +91,10 @@ static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { struct amdgpu_vmhub *hub; + bool retry_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY); + bool write_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE); uint32_t status = 0; u64 addr; @@ -102,6 +106,31 @@ static int gmc_v12_0_process_interrupt(struct amdgpu_device *adev, else hub = &adev->vmhub[AMDGPU_GFXHUB(0)]; + if (retry_fault) { + /* Returning 1 here also prevents sending the IV to the KFD */ + + /* Process it only if it's the first fault for this address */ + if (entry->ih != &adev->irq.ih_soft && + amdgpu_gmc_filter_faults(adev, entry->ih, addr, entry->pasid, + entry->timestamp)) + return 1; + + /* Delegate it to a different ring if the hardware hasn't + * already done it. + */ + if (entry->ih == &adev->irq.ih) { + amdgpu_irq_delegate(adev, entry, 8); + return 1; + } + + /* Try to handle the recoverable page faults by filling page + * tables + */ + if (amdgpu_vm_handle_fault(adev, entry->pasid, 0, 0, addr, + entry->timestamp, write_fault)) + return 1; + } + if (!amdgpu_sriov_vf(adev)) { /* * Issue a dummy read to wait for the status register to @@ -312,9 +341,7 @@ static void gmc_v12_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, return; } - mutex_lock(&adev->mman.gtt_window_lock); gmc_v12_0_flush_vm_hub(adev, vmid, vmhub, 0); - mutex_unlock(&adev->mman.gtt_window_lock); return; } diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c index f6ad7911f1e6..a8ec95f42926 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v6_0.c @@ -213,7 +213,7 @@ static void gmc_v6_0_vram_gtt_location(struct amdgpu_device *adev, amdgpu_gmc_set_agp_default(adev, mc); amdgpu_gmc_vram_location(adev, mc, base); - amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_BEST_FIT); + amdgpu_gmc_gart_location(adev, mc, AMDGPU_GART_PLACEMENT_LOW); } static void gmc_v6_0_mc_program(struct amdgpu_device *adev) @@ -610,23 +610,21 @@ static void gmc_v6_0_gart_disable(struct amdgpu_device *adev) } static void gmc_v6_0_vm_decode_fault(struct amdgpu_device *adev, - u32 status, u32 addr, u32 mc_client) + u32 status, u32 addr) { u32 mc_id; u32 vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID); u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, PROTECTIONS); - char block[5] = { mc_client >> 24, (mc_client >> 16) & 0xff, - (mc_client >> 8) & 0xff, mc_client & 0xff, 0 }; mc_id = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, MEMORY_CLIENT_ID); - dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from '%s' (0x%08x) (%d)\n", + dev_err(adev->dev, "VM fault (0x%02x, vmid %d) at page %u, %s from %d\n", protections, vmid, addr, REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, MEMORY_CLIENT_RW) ? - "write" : "read", block, mc_client, mc_id); + "write" : "read", mc_id); } static const u32 mc_cg_registers[] = { @@ -1072,6 +1070,12 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev, { u32 addr, status; + /* Delegate to the soft IRQ handler ring */ + if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) { + amdgpu_irq_delegate(adev, entry, 4); + return 1; + } + addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR); status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS); WREG32_P(mmVM_CONTEXT1_CNTL2, 1, ~1); @@ -1079,6 +1083,10 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev, if (!addr && !status) return 0; + amdgpu_vm_update_fault_cache(adev, entry->pasid, + ((u64)addr) << AMDGPU_GPU_PAGE_SHIFT, + status, AMDGPU_GFXHUB(0)); + if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_FIRST) gmc_v6_0_set_fault_enable_default(adev, false); @@ -1089,7 +1097,7 @@ static int gmc_v6_0_process_interrupt(struct amdgpu_device *adev, addr); dev_err(adev->dev, " VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x%08X\n", status); - gmc_v6_0_vm_decode_fault(adev, status, addr, 0); + gmc_v6_0_vm_decode_fault(adev, status, addr); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c index 93d7ccb7d013..fbd0bf147f50 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v7_0.c @@ -1068,7 +1068,7 @@ static int gmc_v7_0_sw_init(struct amdgpu_ip_block *ip_block) GFP_KERNEL); if (!adev->gmc.vm_fault_info) return -ENOMEM; - atomic_set(&adev->gmc.vm_fault_info_updated, 0); + atomic_set_release(&adev->gmc.vm_fault_info_updated, 0); return 0; } @@ -1261,6 +1261,12 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev, { u32 addr, status, mc_client, vmid; + /* Delegate to the soft IRQ handler ring */ + if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) { + amdgpu_irq_delegate(adev, entry, 4); + return 1; + } + addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR); status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS); mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT); @@ -1290,7 +1296,7 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev, vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID); if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid) - && !atomic_read(&adev->gmc.vm_fault_info_updated)) { + && !atomic_read_acquire(&adev->gmc.vm_fault_info_updated)) { struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info; u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, @@ -1306,8 +1312,7 @@ static int gmc_v7_0_process_interrupt(struct amdgpu_device *adev, info->prot_read = protections & 0x8 ? true : false; info->prot_write = protections & 0x10 ? true : false; info->prot_exec = protections & 0x20 ? true : false; - mb(); - atomic_set(&adev->gmc.vm_fault_info_updated, 1); + atomic_set_release(&adev->gmc.vm_fault_info_updated, 1); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c index c5e2a2c41e06..6551b60f2584 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v8_0.c @@ -1183,7 +1183,7 @@ static int gmc_v8_0_sw_init(struct amdgpu_ip_block *ip_block) GFP_KERNEL); if (!adev->gmc.vm_fault_info) return -ENOMEM; - atomic_set(&adev->gmc.vm_fault_info_updated, 0); + atomic_set_release(&adev->gmc.vm_fault_info_updated, 0); return 0; } @@ -1439,6 +1439,12 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev, return 0; } + /* Delegate to the soft IRQ handler ring */ + if (adev->irq.ih_soft.enabled && entry->ih != &adev->irq.ih_soft) { + amdgpu_irq_delegate(adev, entry, 4); + return 1; + } + addr = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_ADDR); status = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_STATUS); mc_client = RREG32(mmVM_CONTEXT1_PROTECTION_FAULT_MCCLIENT); @@ -1478,7 +1484,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev, vmid = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, VMID); if (amdgpu_amdkfd_is_kfd_vmid(adev, vmid) - && !atomic_read(&adev->gmc.vm_fault_info_updated)) { + && !atomic_read_acquire(&adev->gmc.vm_fault_info_updated)) { struct kfd_vm_fault_info *info = adev->gmc.vm_fault_info; u32 protections = REG_GET_FIELD(status, VM_CONTEXT1_PROTECTION_FAULT_STATUS, @@ -1494,8 +1500,7 @@ static int gmc_v8_0_process_interrupt(struct amdgpu_device *adev, info->prot_read = protections & 0x8 ? true : false; info->prot_write = protections & 0x10 ? true : false; info->prot_exec = protections & 0x20 ? true : false; - mb(); - atomic_set(&adev->gmc.vm_fault_info_updated, 1); + atomic_set_release(&adev->gmc.vm_fault_info_updated, 1); } return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 0d1dd587db5f..8ad7519f7b58 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -544,8 +544,10 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) { - bool retry_fault = !!(entry->src_data[1] & 0x80); - bool write_fault = !!(entry->src_data[1] & 0x20); + bool retry_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_RETRY); + bool write_fault = !!(entry->src_data[1] & + AMDGPU_GMC9_FAULT_SOURCE_DATA_WRITE); uint32_t status = 0, cid = 0, rw = 0, fed = 0; struct amdgpu_task_info *task_info; struct amdgpu_vmhub *hub; @@ -1843,6 +1845,10 @@ static void gmc_v9_4_3_init_vram_info(struct amdgpu_device *adev) if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 5, 0)) adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E; + if (amdgpu_ip_version(adev, GC_HWIP, 0) == IP_VERSION(9, 4, 4) && + adev->rev_id == 0x3) + adev->gmc.vram_type = AMDGPU_VRAM_TYPE_HBM3E; + if (!(adev->flags & AMD_IS_APU) && !amdgpu_sriov_vf(adev)) { vram_info = RREG32(regBIF_BIOS_SCRATCH_4); adev->gmc.vram_vendor = vram_info & 0xF; diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c index 1317ede131b6..01cadf898c00 100644 --- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c @@ -157,6 +157,9 @@ static int iceland_ih_irq_init(struct amdgpu_device *adev) /* enable interrupts */ iceland_ih_enable_interrupts(adev); + if (adev->irq.ih_soft.ring_size) + adev->irq.ih_soft.enabled = true; + return 0; } @@ -194,6 +197,9 @@ static u32 iceland_ih_get_wptr(struct amdgpu_device *adev, wptr = le32_to_cpu(*ih->wptr_cpu); + if (ih == &adev->irq.ih_soft) + goto out; + if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) goto out; @@ -296,6 +302,10 @@ static int iceland_ih_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true); + if (r) + return r; + r = amdgpu_irq_init(adev); return r; diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c index baf097d2e1ac..ab0bf880d3d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c @@ -878,6 +878,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = { .get_rptr = jpeg_v5_0_1_dec_ring_get_rptr, .get_wptr = jpeg_v5_0_1_dec_ring_get_wptr, .set_wptr = jpeg_v5_0_1_dec_ring_set_wptr, + .parse_cs = amdgpu_jpeg_dec_parse_cs, .emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + diff --git a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c index 2db9b2c63693..64cae89357b6 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_userqueue.c @@ -205,13 +205,14 @@ static int mes_userq_detect_and_reset(struct amdgpu_device *adev, int db_array_size = amdgpu_mes_get_hung_queue_db_array_size(adev); struct mes_detect_and_reset_queue_input input; struct amdgpu_usermode_queue *queue; - struct amdgpu_userq_mgr *uqm, *tmp; unsigned int hung_db_num = 0; - int queue_id, r, i; - u32 db_array[4]; + unsigned long queue_id; + u32 db_array[8]; + bool found_hung_queue = false; + int r, i; - if (db_array_size > 4) { - dev_err(adev->dev, "DB array size (%d vs 4) too small\n", + if (db_array_size > 8) { + dev_err(adev->dev, "DB array size (%d vs 8) too small\n", db_array_size); return -EINVAL; } @@ -227,22 +228,26 @@ static int mes_userq_detect_and_reset(struct amdgpu_device *adev, if (r) { dev_err(adev->dev, "Failed to detect and reset queues, err (%d)\n", r); } else if (hung_db_num) { - list_for_each_entry_safe(uqm, tmp, &adev->userq_mgr_list, list) { - idr_for_each_entry(&uqm->userq_idr, queue, queue_id) { - if (queue->queue_type == queue_type) { - for (i = 0; i < hung_db_num; i++) { - if (queue->doorbell_index == db_array[i]) { - queue->state = AMDGPU_USERQ_STATE_HUNG; - atomic_inc(&adev->gpu_reset_counter); - amdgpu_userq_fence_driver_force_completion(queue); - drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL); - } + xa_for_each(&adev->userq_doorbell_xa, queue_id, queue) { + if (queue->queue_type == queue_type) { + for (i = 0; i < hung_db_num; i++) { + if (queue->doorbell_index == db_array[i]) { + queue->state = AMDGPU_USERQ_STATE_HUNG; + found_hung_queue = true; + atomic_inc(&adev->gpu_reset_counter); + amdgpu_userq_fence_driver_force_completion(queue); + drm_dev_wedged_event(adev_to_drm(adev), DRM_WEDGE_RECOVERY_NONE, NULL); } } } } } + if (found_hung_queue) { + /* Resume scheduling after hang recovery */ + r = amdgpu_mes_resume(adev); + } + return r; } @@ -254,7 +259,6 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, struct amdgpu_mqd *mqd_hw_default = &adev->mqds[queue->queue_type]; struct drm_amdgpu_userq_in *mqd_user = args_in; struct amdgpu_mqd_prop *userq_props; - struct amdgpu_gfx_shadow_info shadow_info; int r; /* Structure to initialize MQD for userqueue using generic MQD init function */ @@ -280,8 +284,6 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, userq_props->doorbell_index = queue->doorbell_index; userq_props->fence_address = queue->fence_drv->gpu_addr; - if (adev->gfx.funcs->get_gfx_shadow_info) - adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true); if (queue->queue_type == AMDGPU_HW_IP_COMPUTE) { struct drm_amdgpu_userq_mqd_compute_gfx11 *compute_mqd; @@ -298,8 +300,9 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, goto free_mqd; } - if (amdgpu_userq_input_va_validate(queue->vm, compute_mqd->eop_va, - max_t(u32, PAGE_SIZE, AMDGPU_GPU_PAGE_SIZE))) + r = amdgpu_userq_input_va_validate(queue, compute_mqd->eop_va, + 2048); + if (r) goto free_mqd; userq_props->eop_gpu_addr = compute_mqd->eop_va; @@ -311,6 +314,14 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, kfree(compute_mqd); } else if (queue->queue_type == AMDGPU_HW_IP_GFX) { struct drm_amdgpu_userq_mqd_gfx11 *mqd_gfx_v11; + struct amdgpu_gfx_shadow_info shadow_info; + + if (adev->gfx.funcs->get_gfx_shadow_info) { + adev->gfx.funcs->get_gfx_shadow_info(adev, &shadow_info, true); + } else { + r = -EINVAL; + goto free_mqd; + } if (mqd_user->mqd_size != sizeof(*mqd_gfx_v11) || !mqd_user->mqd) { DRM_ERROR("Invalid GFX MQD\n"); @@ -330,8 +341,13 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, userq_props->tmz_queue = mqd_user->flags & AMDGPU_USERQ_CREATE_FLAGS_QUEUE_SECURE; - if (amdgpu_userq_input_va_validate(queue->vm, mqd_gfx_v11->shadow_va, - shadow_info.shadow_size)) + r = amdgpu_userq_input_va_validate(queue, mqd_gfx_v11->shadow_va, + shadow_info.shadow_size); + if (r) + goto free_mqd; + r = amdgpu_userq_input_va_validate(queue, mqd_gfx_v11->csa_va, + shadow_info.csa_size); + if (r) goto free_mqd; kfree(mqd_gfx_v11); @@ -350,9 +366,9 @@ static int mes_userq_mqd_create(struct amdgpu_userq_mgr *uq_mgr, r = -ENOMEM; goto free_mqd; } - - if (amdgpu_userq_input_va_validate(queue->vm, mqd_sdma_v11->csa_va, - shadow_info.csa_size)) + r = amdgpu_userq_input_va_validate(queue, mqd_sdma_v11->csa_va, + 32); + if (r) goto free_mqd; userq_props->csa_addr = mqd_sdma_v11->csa_va; diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index e82188431f79..3a52754b5cad 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -66,7 +66,8 @@ static int mes_v11_0_kiq_hw_fini(struct amdgpu_device *adev); #define GFX_MES_DRAM_SIZE 0x80000 #define MES11_HW_RESOURCE_1_SIZE (128 * AMDGPU_GPU_PAGE_SIZE) -#define MES11_HUNG_DB_OFFSET_ARRAY_SIZE 4 +#define MES11_HUNG_DB_OFFSET_ARRAY_SIZE 8 /* [0:3] = db offset, [4:7] = hqd info */ +#define MES11_HUNG_HQD_INFO_OFFSET 4 static void mes_v11_0_ring_set_wptr(struct amdgpu_ring *ring) { @@ -368,6 +369,7 @@ static int mes_v11_0_remove_hw_queue(struct amdgpu_mes *mes, struct mes_remove_queue_input *input) { union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt; + uint32_t mes_rev = mes->sched_version & AMDGPU_MES_VERSION_MASK; memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt)); @@ -378,6 +380,9 @@ static int mes_v11_0_remove_hw_queue(struct amdgpu_mes *mes, mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset; mes_remove_queue_pkt.gang_context_addr = input->gang_context_addr; + if (mes_rev >= 0x60) + mes_remove_queue_pkt.remove_queue_after_reset = input->remove_queue_after_reset; + return mes_v11_0_submit_pkt_and_poll_completion(mes, &mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt), offsetof(union MESAPI__REMOVE_QUEUE, api_status)); @@ -1720,8 +1725,9 @@ static int mes_v11_0_early_init(struct amdgpu_ip_block *ip_block) struct amdgpu_device *adev = ip_block->adev; int pipe, r; - adev->mes.hung_queue_db_array_size = - MES11_HUNG_DB_OFFSET_ARRAY_SIZE; + adev->mes.hung_queue_db_array_size = MES11_HUNG_DB_OFFSET_ARRAY_SIZE; + adev->mes.hung_queue_hqd_info_offset = MES11_HUNG_HQD_INFO_OFFSET; + for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) { if (!adev->enable_mes_kiq && pipe == AMDGPU_MES_KIQ_PIPE) continue; diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c index aff06f06aeee..744e95d3984a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v12_0.c @@ -47,7 +47,8 @@ static int mes_v12_0_kiq_hw_fini(struct amdgpu_device *adev); #define MES_EOP_SIZE 2048 -#define MES12_HUNG_DB_OFFSET_ARRAY_SIZE 4 +#define MES12_HUNG_DB_OFFSET_ARRAY_SIZE 8 /* [0:3] = db offset [4:7] hqd info */ +#define MES12_HUNG_HQD_INFO_OFFSET 4 static void mes_v12_0_ring_set_wptr(struct amdgpu_ring *ring) { @@ -228,7 +229,12 @@ static int mes_v12_0_submit_pkt_and_poll_completion(struct amdgpu_mes *mes, pipe, x_pkt->header.opcode); r = amdgpu_fence_wait_polling(ring, seq, timeout); - if (r < 1 || !*status_ptr) { + + /* + * status_ptr[31:0] == 0 (fail) or status_ptr[63:0] == 1 (success). + * If status_ptr[31:0] == 0 then status_ptr[63:32] will have debug error information. + */ + if (r < 1 || !(lower_32_bits(*status_ptr))) { if (misc_op_str) dev_err(adev->dev, "MES(%d) failed to respond to msg=%s (%s)\n", @@ -355,6 +361,7 @@ static int mes_v12_0_remove_hw_queue(struct amdgpu_mes *mes, struct mes_remove_queue_input *input) { union MESAPI__REMOVE_QUEUE mes_remove_queue_pkt; + uint32_t mes_rev = mes->sched_version & AMDGPU_MES_VERSION_MASK; memset(&mes_remove_queue_pkt, 0, sizeof(mes_remove_queue_pkt)); @@ -365,6 +372,9 @@ static int mes_v12_0_remove_hw_queue(struct amdgpu_mes *mes, mes_remove_queue_pkt.doorbell_offset = input->doorbell_offset; mes_remove_queue_pkt.gang_context_addr = input->gang_context_addr; + if (mes_rev >= 0x5a) + mes_remove_queue_pkt.remove_queue_after_reset = input->remove_queue_after_reset; + return mes_v12_0_submit_pkt_and_poll_completion(mes, AMDGPU_MES_SCHED_PIPE, &mes_remove_queue_pkt, sizeof(mes_remove_queue_pkt), @@ -1899,8 +1909,9 @@ static int mes_v12_0_early_init(struct amdgpu_ip_block *ip_block) struct amdgpu_device *adev = ip_block->adev; int pipe, r; - adev->mes.hung_queue_db_array_size = - MES12_HUNG_DB_OFFSET_ARRAY_SIZE; + adev->mes.hung_queue_db_array_size = MES12_HUNG_DB_OFFSET_ARRAY_SIZE; + adev->mes.hung_queue_hqd_info_offset = MES12_HUNG_HQD_INFO_OFFSET; + for (pipe = 0; pipe < AMDGPU_MAX_MES_PIPES; pipe++) { r = amdgpu_mes_init_microcode(adev, pipe); if (r) diff --git a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c index e5282a5d05d9..e7cd07383d56 100644 --- a/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c +++ b/drivers/gpu/drm/amd/amdgpu/mxgpu_nv.c @@ -173,13 +173,17 @@ static void xgpu_nv_mailbox_trans_msg (struct amdgpu_device *adev, static int xgpu_nv_send_access_requests_with_param(struct amdgpu_device *adev, enum idh_request req, u32 data1, u32 data2, u32 data3) { - int r, retry = 1; + struct amdgpu_virt *virt = &adev->virt; + int r = 0, retry = 1; enum idh_event event = -1; + mutex_lock(&virt->access_req_mutex); send_request: - if (amdgpu_ras_is_rma(adev)) - return -ENODEV; + if (amdgpu_ras_is_rma(adev)) { + r = -ENODEV; + goto out; + } xgpu_nv_mailbox_trans_msg(adev, req, data1, data2, data3); @@ -217,17 +221,25 @@ send_request: if (req != IDH_REQ_GPU_INIT_DATA) { dev_err(adev->dev, "Doesn't get msg:%d from pf, error=%d\n", event, r); - return r; + goto out; } else /* host doesn't support REQ_GPU_INIT_DATA handshake */ adev->virt.req_init_data_ver = 0; } else { if (req == IDH_REQ_GPU_INIT_DATA) { - adev->virt.req_init_data_ver = - RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW1); - - /* assume V1 in case host doesn't set version number */ - if (adev->virt.req_init_data_ver < 1) - adev->virt.req_init_data_ver = 1; + switch (RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW1)) { + case GPU_CRIT_REGION_V2: + adev->virt.req_init_data_ver = GPU_CRIT_REGION_V2; + adev->virt.init_data_header.offset = + RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW2); + adev->virt.init_data_header.size_kb = + RREG32_NO_KIQ(mmMAILBOX_MSGBUF_RCV_DW3); + break; + default: + adev->virt.req_init_data_ver = GPU_CRIT_REGION_V1; + adev->virt.init_data_header.offset = -1; + adev->virt.init_data_header.size_kb = 0; + break; + } } } @@ -238,7 +250,10 @@ send_request: } } - return 0; +out: + mutex_unlock(&virt->access_req_mutex); + + return r; } static int xgpu_nv_send_access_requests(struct amdgpu_device *adev, @@ -285,7 +300,8 @@ static int xgpu_nv_release_full_gpu_access(struct amdgpu_device *adev, static int xgpu_nv_request_init_data(struct amdgpu_device *adev) { - return xgpu_nv_send_access_requests(adev, IDH_REQ_GPU_INIT_DATA); + return xgpu_nv_send_access_requests_with_param(adev, IDH_REQ_GPU_INIT_DATA, + 0, GPU_CRIT_REGION_V2, 0); } static int xgpu_nv_mailbox_ack_irq(struct amdgpu_device *adev, diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c index 1c22bc11c1f8..bdfd2917e3ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c @@ -41,19 +41,21 @@ static void nbio_v7_9_remap_hdp_registers(struct amdgpu_device *adev) static u32 nbio_v7_9_get_rev_id(struct amdgpu_device *adev) { - u32 tmp; - - tmp = IP_VERSION_SUBREV(amdgpu_ip_version_full(adev, NBIO_HWIP, 0)); - /* If it is VF or subrevision holds a non-zero value, that should be used */ - if (tmp || amdgpu_sriov_vf(adev)) - return tmp; + u32 rev_id; - /* If discovery subrev is not updated, use register version */ - tmp = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); - tmp = REG_GET_FIELD(tmp, RCC_STRAP0_RCC_DEV0_EPF0_STRAP0, - STRAP_ATI_REV_ID_DEV0_F0); + /* + * fetch the sub-revision field from the IP-discovery table + * (returns zero if the table entry is not populated). + */ + if (amdgpu_sriov_vf(adev)) { + rev_id = IP_VERSION_SUBREV(amdgpu_ip_version_full(adev, NBIO_HWIP, 0)); + } else { + rev_id = RREG32_SOC15(NBIO, 0, regRCC_STRAP0_RCC_DEV0_EPF0_STRAP0); + rev_id = REG_GET_FIELD(rev_id, RCC_STRAP0_RCC_DEV0_EPF0_STRAP0, + STRAP_ATI_REV_ID_DEV0_F0); + } - return tmp; + return rev_id; } static void nbio_v7_9_mc_access_enable(struct amdgpu_device *adev, bool enable) diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 64b240b51f1a..a9be7a505026 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -142,13 +142,37 @@ static int psp_v11_0_init_microcode(struct psp_context *psp) return err; } -static int psp_v11_0_wait_for_bootloader(struct psp_context *psp) +static int psp_v11_wait_for_tos_unload(struct psp_context *psp) { struct amdgpu_device *adev = psp->adev; + uint32_t sol_reg1, sol_reg2; + int retry_loop; + /* Wait for the TOS to be unloaded */ + for (retry_loop = 0; retry_loop < 20; retry_loop++) { + sol_reg1 = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81); + usleep_range(1000, 2000); + sol_reg2 = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81); + if (sol_reg1 == sol_reg2) + return 0; + } + dev_err(adev->dev, "TOS unload failed, C2PMSG_33: %x C2PMSG_81: %x", + RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_33), + RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81)); + + return -ETIME; +} + +static int psp_v11_0_wait_for_bootloader(struct psp_context *psp) +{ + struct amdgpu_device *adev = psp->adev; int ret; int retry_loop; + /* For a reset done at the end of S3, only wait for TOS to be unloaded */ + if (adev->in_s3 && !(adev->flags & AMD_IS_APU) && amdgpu_in_reset(adev)) + return psp_v11_wait_for_tos_unload(psp); + for (retry_loop = 0; retry_loop < 20; retry_loop++) { /* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */ diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index 36b1ca73c2ed..a1443990d5c6 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -2361,11 +2361,15 @@ static void sdma_v4_4_2_update_reset_mask(struct amdgpu_device *adev) switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(9, 4, 3): case IP_VERSION(9, 4, 4): - if ((adev->gfx.mec_fw_version >= 0xb0) && amdgpu_dpm_reset_sdma_is_supported(adev)) + if ((adev->gfx.mec_fw_version >= 0xb0) && + amdgpu_dpm_reset_sdma_is_supported(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; case IP_VERSION(9, 5, 0): - if ((adev->gfx.mec_fw_version >= 0xf) && amdgpu_dpm_reset_sdma_is_supported(adev)) + if ((adev->gfx.mec_fw_version >= 0xf) && + amdgpu_dpm_reset_sdma_is_supported(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 7dc67a22a7a0..8ddc4df06a1f 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -1429,7 +1429,8 @@ static int sdma_v5_0_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(5, 0, 2): case IP_VERSION(5, 0, 5): if ((adev->sdma.instance[0].fw_version >= 35) && - !amdgpu_sriov_vf(adev)) + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index 3bd44c24f692..51101b0aa2fa 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -342,7 +342,7 @@ static void sdma_v5_2_ring_emit_hdp_flush(struct amdgpu_ring *ring) const struct nbio_hdp_flush_reg *nbio_hf_reg = adev->nbio.hdp_flush_reg; if (ring->me > 1) { - amdgpu_asic_flush_hdp(adev, ring); + amdgpu_hdp_flush(adev, ring); } else { ref_and_mask = nbio_hf_reg->ref_and_mask_sdma0 << ring->me; @@ -1348,12 +1348,14 @@ static int sdma_v5_2_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(5, 2, 3): case IP_VERSION(5, 2, 4): if ((adev->sdma.instance[0].fw_version >= 76) && - !amdgpu_sriov_vf(adev)) + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; case IP_VERSION(5, 2, 5): if ((adev->sdma.instance[0].fw_version >= 34) && - !amdgpu_sriov_vf(adev)) + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; default: diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index db6e41967f12..217040044987 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -1356,7 +1356,8 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block) case IP_VERSION(6, 0, 2): case IP_VERSION(6, 0, 3): if ((adev->sdma.instance[0].fw_version >= 21) && - !amdgpu_sriov_vf(adev)) + !amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; default: @@ -1389,7 +1390,7 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block) adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; break; case IP_VERSION(6, 0, 3): - if ((adev->sdma.instance[0].fw_version >= 27) && !adev->sdma.disable_uq) + if (adev->sdma.instance[0].fw_version >= 29 && !adev->sdma.disable_uq) adev->userq_funcs[AMDGPU_HW_IP_DMA] = &userq_mes_funcs; break; case IP_VERSION(6, 1, 0): diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c index 326ecc8d37d2..2b81344dcd66 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c @@ -1337,7 +1337,8 @@ static int sdma_v7_0_sw_init(struct amdgpu_ip_block *ip_block) adev->sdma.supported_reset = amdgpu_get_soft_full_reset_mask(&adev->sdma.instance[0].ring); - if (!amdgpu_sriov_vf(adev)) + if (!amdgpu_sriov_vf(adev) && + !adev->debug_disable_gpu_ring_reset) adev->sdma.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; r = amdgpu_sdma_sysfs_reset_mask_init(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/si.c b/drivers/gpu/drm/amd/amdgpu/si.c index e0f139de7991..f7288372ee61 100644 --- a/drivers/gpu/drm/amd/amdgpu/si.c +++ b/drivers/gpu/drm/amd/amdgpu/si.c @@ -45,6 +45,7 @@ #include "dce_v6_0.h" #include "si.h" #include "uvd_v3_1.h" +#include "vce_v1_0.h" #include "uvd/uvd_4_0_d.h" @@ -921,8 +922,6 @@ static const u32 hainan_mgcg_cgcg_init[] = 0x3630, 0xfffffff0, 0x00000100, }; -/* XXX: update when we support VCE */ -#if 0 /* tahiti, pitcairn, verde */ static const struct amdgpu_video_codec_info tahiti_video_codecs_encode_array[] = { @@ -940,13 +939,7 @@ static const struct amdgpu_video_codecs tahiti_video_codecs_encode = .codec_count = ARRAY_SIZE(tahiti_video_codecs_encode_array), .codec_array = tahiti_video_codecs_encode_array, }; -#else -static const struct amdgpu_video_codecs tahiti_video_codecs_encode = -{ - .codec_count = 0, - .codec_array = NULL, -}; -#endif + /* oland and hainan don't support encode */ static const struct amdgpu_video_codecs hainan_video_codecs_encode = { @@ -1925,6 +1918,14 @@ static int si_set_vce_clocks(struct amdgpu_device *adev, u32 evclk, u32 ecclk) ~VCEPLL_BYPASS_EN_MASK); if (!evclk || !ecclk) { + /* + * On some chips, the PLL takes way too long to get out of + * sleep mode, causing a timeout waiting on CTLACK/CTLACK2. + * Leave the PLL running in bypass mode. + */ + if (adev->pdev->device == 0x6780) + return 0; + /* Keep the Bypass mode, put PLL to sleep */ WREG32_SMC_P(CG_VCEPLL_FUNC_CNTL, VCEPLL_SLEEP_MASK, ~VCEPLL_SLEEP_MASK); @@ -2717,7 +2718,7 @@ int si_set_ip_blocks(struct amdgpu_device *adev) else amdgpu_device_ip_block_add(adev, &dce_v6_0_ip_block); amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block); - /* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */ + amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); break; case CHIP_OLAND: amdgpu_device_ip_block_add(adev, &si_common_ip_block); @@ -2735,7 +2736,6 @@ int si_set_ip_blocks(struct amdgpu_device *adev) else amdgpu_device_ip_block_add(adev, &dce_v6_4_ip_block); amdgpu_device_ip_block_add(adev, &uvd_v3_1_ip_block); - /* amdgpu_device_ip_block_add(adev, &vce_v1_0_ip_block); */ break; case CHIP_HAINAN: amdgpu_device_ip_block_add(adev, &si_common_ip_block); diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c index 1df00f8a2406..66f650f87243 100644 --- a/drivers/gpu/drm/amd/amdgpu/si_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c @@ -96,6 +96,9 @@ static int si_ih_irq_init(struct amdgpu_device *adev) pci_set_master(adev->pdev); si_ih_enable_interrupts(adev); + if (adev->irq.ih_soft.ring_size) + adev->irq.ih_soft.enabled = true; + return 0; } @@ -112,6 +115,9 @@ static u32 si_ih_get_wptr(struct amdgpu_device *adev, wptr = le32_to_cpu(*ih->wptr_cpu); + if (ih == &adev->irq.ih_soft) + goto out; + if (wptr & IH_RB_WPTR__RB_OVERFLOW_MASK) { wptr &= ~IH_RB_WPTR__RB_OVERFLOW_MASK; dev_warn(adev->dev, "IH ring buffer overflow (0x%08X, 0x%08X, 0x%08X)\n", @@ -127,6 +133,8 @@ static u32 si_ih_get_wptr(struct amdgpu_device *adev, tmp &= ~IH_RB_CNTL__WPTR_OVERFLOW_CLEAR_MASK; WREG32(IH_RB_CNTL, tmp); } + +out: return (wptr & ih->ptr_mask); } @@ -175,6 +183,10 @@ static int si_ih_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true); + if (r) + return r; + return amdgpu_irq_init(adev); } diff --git a/drivers/gpu/drm/amd/amdgpu/sid.h b/drivers/gpu/drm/amd/amdgpu/sid.h index cbd4f8951cfa..561462a8332e 100644 --- a/drivers/gpu/drm/amd/amdgpu/sid.h +++ b/drivers/gpu/drm/amd/amdgpu/sid.h @@ -582,45 +582,6 @@ #define DMA_PACKET_NOP 0xf /* VCE */ -#define VCE_STATUS 0x20004 -#define VCE_VCPU_CNTL 0x20014 -#define VCE_CLK_EN (1 << 0) -#define VCE_VCPU_CACHE_OFFSET0 0x20024 -#define VCE_VCPU_CACHE_SIZE0 0x20028 -#define VCE_VCPU_CACHE_OFFSET1 0x2002c -#define VCE_VCPU_CACHE_SIZE1 0x20030 -#define VCE_VCPU_CACHE_OFFSET2 0x20034 -#define VCE_VCPU_CACHE_SIZE2 0x20038 -#define VCE_SOFT_RESET 0x20120 -#define VCE_ECPU_SOFT_RESET (1 << 0) -#define VCE_FME_SOFT_RESET (1 << 2) -#define VCE_RB_BASE_LO2 0x2016c -#define VCE_RB_BASE_HI2 0x20170 -#define VCE_RB_SIZE2 0x20174 -#define VCE_RB_RPTR2 0x20178 -#define VCE_RB_WPTR2 0x2017c -#define VCE_RB_BASE_LO 0x20180 -#define VCE_RB_BASE_HI 0x20184 -#define VCE_RB_SIZE 0x20188 -#define VCE_RB_RPTR 0x2018c -#define VCE_RB_WPTR 0x20190 -#define VCE_CLOCK_GATING_A 0x202f8 -#define VCE_CLOCK_GATING_B 0x202fc -#define VCE_UENC_CLOCK_GATING 0x205bc -#define VCE_UENC_REG_CLOCK_GATING 0x205c0 -#define VCE_FW_REG_STATUS 0x20e10 -# define VCE_FW_REG_STATUS_BUSY (1 << 0) -# define VCE_FW_REG_STATUS_PASS (1 << 3) -# define VCE_FW_REG_STATUS_DONE (1 << 11) -#define VCE_LMI_FW_START_KEYSEL 0x20e18 -#define VCE_LMI_FW_PERIODIC_CTRL 0x20e20 -#define VCE_LMI_CTRL2 0x20e74 -#define VCE_LMI_CTRL 0x20e98 -#define VCE_LMI_VM_CTRL 0x20ea0 -#define VCE_LMI_SWAP_CNTL 0x20eb4 -#define VCE_LMI_SWAP_CNTL1 0x20eb8 -#define VCE_LMI_CACHE_CTRL 0x20ef4 - #define VCE_CMD_NO_OP 0x00000000 #define VCE_CMD_END 0x00000001 #define VCE_CMD_IB 0x00000002 @@ -629,7 +590,6 @@ #define VCE_CMD_IB_AUTO 0x00000005 #define VCE_CMD_SEMAPHORE 0x00000006 - //#dce stupp /* display controller offsets used for crtc/cur/lut/grph/viewport/etc. */ #define CRTC0_REGISTER_OFFSET (0x1b7c - 0x1b7c) //(0x6df0 - 0x6df0)/4 diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index 9785fada4fa7..42f5d9c0e3af 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -853,10 +853,6 @@ static bool soc15_need_reset_on_init(struct amdgpu_device *adev) { u32 sol_reg; - /* CP hangs in IGT reloading test on RN, reset to WA */ - if (adev->asic_type == CHIP_RENOIR) - return true; - if (amdgpu_gmc_need_reset_on_init(adev)) return true; if (amdgpu_psp_tos_reload_needed(adev)) diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c index 7d17ae56f901..ee8038df17e3 100644 --- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c +++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c @@ -159,6 +159,9 @@ static int tonga_ih_irq_init(struct amdgpu_device *adev) /* enable interrupts */ tonga_ih_enable_interrupts(adev); + if (adev->irq.ih_soft.ring_size) + adev->irq.ih_soft.enabled = true; + return 0; } @@ -196,6 +199,9 @@ static u32 tonga_ih_get_wptr(struct amdgpu_device *adev, wptr = le32_to_cpu(*ih->wptr_cpu); + if (ih == &adev->irq.ih_soft) + goto out; + if (!REG_GET_FIELD(wptr, IH_RB_WPTR, RB_OVERFLOW)) goto out; @@ -306,6 +312,10 @@ static int tonga_ih_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + r = amdgpu_ih_ring_init(adev, &adev->irq.ih_soft, IH_SW_RING_SIZE, true); + if (r) + return r; + adev->irq.ih.use_doorbell = true; adev->irq.ih.doorbell_index = adev->doorbell_index.ih; diff --git a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c index 8dc32787d625..0f5b1719fda5 100644 --- a/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/umc_v12_0.c @@ -711,6 +711,19 @@ static uint32_t umc_v12_0_get_die_id(struct amdgpu_device *adev, return die; } +static void umc_v12_0_mca_ipid_parse(struct amdgpu_device *adev, uint64_t ipid, + uint32_t *did, uint32_t *ch, uint32_t *umc_inst, uint32_t *sid) +{ + if (did) + *did = MCA_IPID_2_DIE_ID(ipid); + if (ch) + *ch = MCA_IPID_2_UMC_CH(ipid); + if (umc_inst) + *umc_inst = MCA_IPID_2_UMC_INST(ipid); + if (sid) + *sid = MCA_IPID_2_SOCKET_ID(ipid); +} + struct amdgpu_umc_ras umc_v12_0_ras = { .ras_block = { .hw_ops = &umc_v12_0_ras_hw_ops, @@ -724,5 +737,6 @@ struct amdgpu_umc_ras umc_v12_0_ras = { .convert_ras_err_addr = umc_v12_0_convert_error_address, .get_die_id_from_pa = umc_v12_0_get_die_id, .get_retire_flip_bits = umc_v12_0_get_retire_flip_bits, + .mca_ipid_parse = umc_v12_0_mca_ipid_parse, }; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v1_0.c new file mode 100644 index 000000000000..9ae424618556 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vce_v1_0.c @@ -0,0 +1,839 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2013 Advanced Micro Devices, Inc. + * Copyright 2025 Valve Corporation + * Copyright 2025 Alexandre Demers + * All Rights Reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the + * "Software"), to deal in the Software without restriction, including + * without limitation the rights to use, copy, modify, merge, publish, + * distribute, sub license, and/or sell copies of the Software, and to + * permit persons to whom the Software is furnished to do so, subject to + * the following conditions: + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDERS, AUTHORS AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM, + * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR + * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE + * USE OR OTHER DEALINGS IN THE SOFTWARE. + * + * The above copyright notice and this permission notice (including the + * next paragraph) shall be included in all copies or substantial portions + * of the Software. + * + * Authors: Christian König <christian.koenig@amd.com> + * Timur Kristóf <timur.kristof@gmail.com> + * Alexandre Demers <alexandre.f.demers@gmail.com> + */ + +#include <linux/firmware.h> + +#include "amdgpu.h" +#include "amdgpu_vce.h" +#include "amdgpu_gart.h" +#include "sid.h" +#include "vce_v1_0.h" +#include "vce/vce_1_0_d.h" +#include "vce/vce_1_0_sh_mask.h" +#include "oss/oss_1_0_d.h" +#include "oss/oss_1_0_sh_mask.h" + +#define VCE_V1_0_FW_SIZE (256 * 1024) +#define VCE_V1_0_STACK_SIZE (64 * 1024) +#define VCE_V1_0_DATA_SIZE (7808 * (AMDGPU_MAX_VCE_HANDLES + 1)) +#define VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK 0x02 + +#define VCE_V1_0_GART_PAGE_START \ + (AMDGPU_GTT_MAX_TRANSFER_SIZE * AMDGPU_GTT_NUM_TRANSFER_WINDOWS) +#define VCE_V1_0_GART_ADDR_START \ + (VCE_V1_0_GART_PAGE_START * AMDGPU_GPU_PAGE_SIZE) + +static void vce_v1_0_set_ring_funcs(struct amdgpu_device *adev); +static void vce_v1_0_set_irq_funcs(struct amdgpu_device *adev); + +struct vce_v1_0_fw_signature { + int32_t offset; + uint32_t length; + int32_t number; + struct { + uint32_t chip_id; + uint32_t keyselect; + uint32_t nonce[4]; + uint32_t sigval[4]; + } val[8]; +}; + +/** + * vce_v1_0_ring_get_rptr - get read pointer + * + * @ring: amdgpu_ring pointer + * + * Returns the current hardware read pointer + */ +static uint64_t vce_v1_0_ring_get_rptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring->me == 0) + return RREG32(mmVCE_RB_RPTR); + else + return RREG32(mmVCE_RB_RPTR2); +} + +/** + * vce_v1_0_ring_get_wptr - get write pointer + * + * @ring: amdgpu_ring pointer + * + * Returns the current hardware write pointer + */ +static uint64_t vce_v1_0_ring_get_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring->me == 0) + return RREG32(mmVCE_RB_WPTR); + else + return RREG32(mmVCE_RB_WPTR2); +} + +/** + * vce_v1_0_ring_set_wptr - set write pointer + * + * @ring: amdgpu_ring pointer + * + * Commits the write pointer to the hardware + */ +static void vce_v1_0_ring_set_wptr(struct amdgpu_ring *ring) +{ + struct amdgpu_device *adev = ring->adev; + + if (ring->me == 0) + WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr)); + else + WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr)); +} + +static int vce_v1_0_lmi_clean(struct amdgpu_device *adev) +{ + int i, j; + + for (i = 0; i < 10; ++i) { + for (j = 0; j < 100; ++j) { + if (RREG32(mmVCE_LMI_STATUS) & 0x337f) + return 0; + + mdelay(10); + } + } + + return -ETIMEDOUT; +} + +static int vce_v1_0_firmware_loaded(struct amdgpu_device *adev) +{ + int i, j; + + for (i = 0; i < 10; ++i) { + for (j = 0; j < 100; ++j) { + if (RREG32(mmVCE_STATUS) & VCE_STATUS_VCPU_REPORT_FW_LOADED_MASK) + return 0; + mdelay(10); + } + + dev_err(adev->dev, "VCE not responding, trying to reset the ECPU\n"); + + WREG32_P(mmVCE_SOFT_RESET, + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + WREG32_P(mmVCE_SOFT_RESET, 0, + ~VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK); + mdelay(10); + } + + return -ETIMEDOUT; +} + +static void vce_v1_0_init_cg(struct amdgpu_device *adev) +{ + u32 tmp; + + tmp = RREG32(mmVCE_CLOCK_GATING_A); + tmp |= VCE_CLOCK_GATING_A__CGC_DYN_CLOCK_MODE_MASK; + WREG32(mmVCE_CLOCK_GATING_A, tmp); + + tmp = RREG32(mmVCE_CLOCK_GATING_B); + tmp |= 0x1e; + tmp &= ~0xe100e1; + WREG32(mmVCE_CLOCK_GATING_B, tmp); + + tmp = RREG32(mmVCE_UENC_CLOCK_GATING); + tmp &= ~0xff9ff000; + WREG32(mmVCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + tmp &= ~0x3ff; + WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp); +} + +/** + * vce_v1_0_load_fw_signature - load firmware signature into VCPU BO + * + * @adev: amdgpu_device pointer + * + * The VCE1 firmware validation mechanism needs a firmware signature. + * This function finds the signature appropriate for the current + * ASIC and writes that into the VCPU BO. + */ +static int vce_v1_0_load_fw_signature(struct amdgpu_device *adev) +{ + const struct common_firmware_header *hdr; + struct vce_v1_0_fw_signature *sign; + unsigned int ucode_offset; + uint32_t chip_id; + u32 *cpu_addr; + int i; + + hdr = (const struct common_firmware_header *)adev->vce.fw->data; + ucode_offset = le32_to_cpu(hdr->ucode_array_offset_bytes); + cpu_addr = adev->vce.cpu_addr; + + sign = (void *)adev->vce.fw->data + ucode_offset; + + switch (adev->asic_type) { + case CHIP_TAHITI: + chip_id = 0x01000014; + break; + case CHIP_VERDE: + chip_id = 0x01000015; + break; + case CHIP_PITCAIRN: + chip_id = 0x01000016; + break; + default: + dev_err(adev->dev, "asic_type %#010x was not found!", adev->asic_type); + return -EINVAL; + } + + for (i = 0; i < le32_to_cpu(sign->number); ++i) { + if (le32_to_cpu(sign->val[i].chip_id) == chip_id) + break; + } + + if (i == le32_to_cpu(sign->number)) { + dev_err(adev->dev, "chip_id 0x%x for %s was not found in VCE firmware", + chip_id, amdgpu_asic_name[adev->asic_type]); + return -EINVAL; + } + + cpu_addr += (256 - 64) / 4; + memcpy_toio(&cpu_addr[0], &sign->val[i].nonce[0], 16); + cpu_addr[4] = cpu_to_le32(le32_to_cpu(sign->length) + 64); + + memset_io(&cpu_addr[5], 0, 44); + memcpy_toio(&cpu_addr[16], &sign[1], hdr->ucode_size_bytes - sizeof(*sign)); + + cpu_addr += (le32_to_cpu(sign->length) + 64) / 4; + memcpy_toio(&cpu_addr[0], &sign->val[i].sigval[0], 16); + + adev->vce.keyselect = le32_to_cpu(sign->val[i].keyselect); + + return 0; +} + +static int vce_v1_0_wait_for_fw_validation(struct amdgpu_device *adev) +{ + int i; + + dev_dbg(adev->dev, "VCE keyselect: %d", adev->vce.keyselect); + WREG32(mmVCE_LMI_FW_START_KEYSEL, adev->vce.keyselect); + + for (i = 0; i < 10; ++i) { + mdelay(10); + if (RREG32(mmVCE_FW_REG_STATUS) & VCE_FW_REG_STATUS__DONE_MASK) + break; + } + + if (!(RREG32(mmVCE_FW_REG_STATUS) & VCE_FW_REG_STATUS__DONE_MASK)) { + dev_err(adev->dev, "VCE FW validation timeout\n"); + return -ETIMEDOUT; + } + + if (!(RREG32(mmVCE_FW_REG_STATUS) & VCE_FW_REG_STATUS__PASS_MASK)) { + dev_err(adev->dev, "VCE FW validation failed\n"); + return -EINVAL; + } + + for (i = 0; i < 10; ++i) { + mdelay(10); + if (!(RREG32(mmVCE_FW_REG_STATUS) & VCE_FW_REG_STATUS__BUSY_MASK)) + break; + } + + if (RREG32(mmVCE_FW_REG_STATUS) & VCE_FW_REG_STATUS__BUSY_MASK) { + dev_err(adev->dev, "VCE FW busy timeout\n"); + return -ETIMEDOUT; + } + + return 0; +} + +static int vce_v1_0_mc_resume(struct amdgpu_device *adev) +{ + uint32_t offset; + uint32_t size; + + /* + * When the keyselect is already set, don't perturb VCE FW. + * Validation seems to always fail the second time. + */ + if (RREG32(mmVCE_LMI_FW_START_KEYSEL)) { + dev_dbg(adev->dev, "keyselect already set: 0x%x (on CPU: 0x%x)\n", + RREG32(mmVCE_LMI_FW_START_KEYSEL), adev->vce.keyselect); + + WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100); + return 0; + } + + WREG32_P(mmVCE_CLOCK_GATING_A, 0, ~(1 << 16)); + WREG32_P(mmVCE_UENC_CLOCK_GATING, 0x1FF000, ~0xFF9FF000); + WREG32_P(mmVCE_UENC_REG_CLOCK_GATING, 0x3F, ~0x3F); + WREG32(mmVCE_CLOCK_GATING_B, 0); + + WREG32_P(mmVCE_LMI_FW_PERIODIC_CTRL, 0x4, ~0x4); + + WREG32(mmVCE_LMI_CTRL, 0x00398000); + + WREG32_P(mmVCE_LMI_CACHE_CTRL, 0x0, ~0x1); + WREG32(mmVCE_LMI_SWAP_CNTL, 0); + WREG32(mmVCE_LMI_SWAP_CNTL1, 0); + WREG32(mmVCE_LMI_VM_CTRL, 0); + + WREG32(mmVCE_VCPU_SCRATCH7, AMDGPU_MAX_VCE_HANDLES); + + offset = adev->vce.gpu_addr + AMDGPU_VCE_FIRMWARE_OFFSET; + size = VCE_V1_0_FW_SIZE; + WREG32(mmVCE_VCPU_CACHE_OFFSET0, offset & 0x7fffffff); + WREG32(mmVCE_VCPU_CACHE_SIZE0, size); + + offset += size; + size = VCE_V1_0_STACK_SIZE; + WREG32(mmVCE_VCPU_CACHE_OFFSET1, offset & 0x7fffffff); + WREG32(mmVCE_VCPU_CACHE_SIZE1, size); + + offset += size; + size = VCE_V1_0_DATA_SIZE; + WREG32(mmVCE_VCPU_CACHE_OFFSET2, offset & 0x7fffffff); + WREG32(mmVCE_VCPU_CACHE_SIZE2, size); + + WREG32_P(mmVCE_LMI_CTRL2, 0x0, ~0x100); + + return vce_v1_0_wait_for_fw_validation(adev); +} + +/** + * vce_v1_0_is_idle() - Check idle status of VCE1 IP block + * + * @ip_block: amdgpu_ip_block pointer + * + * Check whether VCE is busy according to VCE_STATUS. + * Also check whether the SRBM thinks VCE is busy, although + * SRBM_STATUS.VCE_BUSY seems to be bogus because it + * appears to mirror the VCE_STATUS.VCPU_REPORT_FW_LOADED bit. + */ +static bool vce_v1_0_is_idle(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + bool busy = + (RREG32(mmVCE_STATUS) & (VCE_STATUS__JOB_BUSY_MASK | VCE_STATUS__UENC_BUSY_MASK)) || + (RREG32(mmSRBM_STATUS2) & SRBM_STATUS2__VCE_BUSY_MASK); + + return !busy; +} + +static int vce_v1_0_wait_for_idle(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + unsigned int i; + + for (i = 0; i < adev->usec_timeout; i++) { + udelay(1); + if (vce_v1_0_is_idle(ip_block)) + return 0; + } + return -ETIMEDOUT; +} + +/** + * vce_v1_0_start - start VCE block + * + * @adev: amdgpu_device pointer + * + * Setup and start the VCE block + */ +static int vce_v1_0_start(struct amdgpu_device *adev) +{ + struct amdgpu_ring *ring; + int r; + + WREG32_P(mmVCE_STATUS, 1, ~1); + + r = vce_v1_0_mc_resume(adev); + if (r) + return r; + + ring = &adev->vce.ring[0]; + WREG32(mmVCE_RB_RPTR, lower_32_bits(ring->wptr)); + WREG32(mmVCE_RB_WPTR, lower_32_bits(ring->wptr)); + WREG32(mmVCE_RB_BASE_LO, lower_32_bits(ring->gpu_addr)); + WREG32(mmVCE_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); + WREG32(mmVCE_RB_SIZE, ring->ring_size / 4); + + ring = &adev->vce.ring[1]; + WREG32(mmVCE_RB_RPTR2, lower_32_bits(ring->wptr)); + WREG32(mmVCE_RB_WPTR2, lower_32_bits(ring->wptr)); + WREG32(mmVCE_RB_BASE_LO2, lower_32_bits(ring->gpu_addr)); + WREG32(mmVCE_RB_BASE_HI2, upper_32_bits(ring->gpu_addr)); + WREG32(mmVCE_RB_SIZE2, ring->ring_size / 4); + + WREG32_P(mmVCE_VCPU_CNTL, VCE_VCPU_CNTL__CLK_EN_MASK, + ~VCE_VCPU_CNTL__CLK_EN_MASK); + + WREG32_P(mmVCE_SOFT_RESET, + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK | + VCE_SOFT_RESET__FME_SOFT_RESET_MASK, + ~(VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK | + VCE_SOFT_RESET__FME_SOFT_RESET_MASK)); + + mdelay(100); + + WREG32_P(mmVCE_SOFT_RESET, 0, + ~(VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK | + VCE_SOFT_RESET__FME_SOFT_RESET_MASK)); + + r = vce_v1_0_firmware_loaded(adev); + + /* Clear VCE_STATUS, otherwise SRBM thinks VCE1 is busy. */ + WREG32(mmVCE_STATUS, 0); + + if (r) { + dev_err(adev->dev, "VCE not responding, giving up\n"); + return r; + } + + return 0; +} + +static int vce_v1_0_stop(struct amdgpu_device *adev) +{ + struct amdgpu_ip_block *ip_block; + int status; + int i; + + ip_block = amdgpu_device_ip_get_ip_block(adev, AMD_IP_BLOCK_TYPE_VCE); + if (!ip_block) + return -EINVAL; + + if (vce_v1_0_lmi_clean(adev)) + dev_warn(adev->dev, "VCE not idle\n"); + + if (vce_v1_0_wait_for_idle(ip_block)) + dev_warn(adev->dev, "VCE busy: VCE_STATUS=0x%x, SRBM_STATUS2=0x%x\n", + RREG32(mmVCE_STATUS), RREG32(mmSRBM_STATUS2)); + + /* Stall UMC and register bus before resetting VCPU */ + WREG32_P(mmVCE_LMI_CTRL2, 1 << 8, ~(1 << 8)); + + for (i = 0; i < 100; ++i) { + status = RREG32(mmVCE_LMI_STATUS); + if (status & 0x240) + break; + mdelay(1); + } + + WREG32_P(mmVCE_VCPU_CNTL, 0, ~VCE_VCPU_CNTL__CLK_EN_MASK); + + WREG32_P(mmVCE_SOFT_RESET, + VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK | + VCE_SOFT_RESET__FME_SOFT_RESET_MASK, + ~(VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK | + VCE_SOFT_RESET__FME_SOFT_RESET_MASK)); + + WREG32(mmVCE_STATUS, 0); + + return 0; +} + +static void vce_v1_0_enable_mgcg(struct amdgpu_device *adev, bool enable) +{ + u32 tmp; + + if (enable && (adev->cg_flags & AMD_CG_SUPPORT_VCE_MGCG)) { + tmp = RREG32(mmVCE_CLOCK_GATING_A); + tmp |= VCE_CLOCK_GATING_A__CGC_DYN_CLOCK_MODE_MASK; + WREG32(mmVCE_CLOCK_GATING_A, tmp); + + tmp = RREG32(mmVCE_UENC_CLOCK_GATING); + tmp &= ~0x1ff000; + tmp |= 0xff800000; + WREG32(mmVCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + tmp &= ~0x3ff; + WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp); + } else { + tmp = RREG32(mmVCE_CLOCK_GATING_A); + tmp &= ~VCE_CLOCK_GATING_A__CGC_DYN_CLOCK_MODE_MASK; + WREG32(mmVCE_CLOCK_GATING_A, tmp); + + tmp = RREG32(mmVCE_UENC_CLOCK_GATING); + tmp |= 0x1ff000; + tmp &= ~0xff800000; + WREG32(mmVCE_UENC_CLOCK_GATING, tmp); + + tmp = RREG32(mmVCE_UENC_REG_CLOCK_GATING); + tmp |= 0x3ff; + WREG32(mmVCE_UENC_REG_CLOCK_GATING, tmp); + } +} + +static int vce_v1_0_early_init(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + int r; + + r = amdgpu_vce_early_init(adev); + if (r) + return r; + + adev->vce.num_rings = 2; + + vce_v1_0_set_ring_funcs(adev); + vce_v1_0_set_irq_funcs(adev); + + return 0; +} + +/** + * vce_v1_0_ensure_vcpu_bo_32bit_addr() - ensure the VCPU BO has a 32-bit address + * + * @adev: amdgpu_device pointer + * + * Due to various hardware limitations, the VCE1 requires + * the VCPU BO to be in the low 32 bit address range. + * Ensure that the VCPU BO has a 32-bit GPU address, + * or return an error code when that isn't possible. + * + * To accomodate that, we put GART to the LOW address range + * and reserve some GART pages where we map the VCPU BO, + * so that it gets a 32-bit address. + */ +static int vce_v1_0_ensure_vcpu_bo_32bit_addr(struct amdgpu_device *adev) +{ + u64 gpu_addr = amdgpu_bo_gpu_offset(adev->vce.vcpu_bo); + u64 bo_size = amdgpu_bo_size(adev->vce.vcpu_bo); + u64 max_vcpu_bo_addr = 0xffffffff - bo_size; + u64 num_pages = ALIGN(bo_size, AMDGPU_GPU_PAGE_SIZE) / AMDGPU_GPU_PAGE_SIZE; + u64 pa = amdgpu_gmc_vram_pa(adev, adev->vce.vcpu_bo); + u64 flags = AMDGPU_PTE_READABLE | AMDGPU_PTE_WRITEABLE | AMDGPU_PTE_VALID; + + /* + * Check if the VCPU BO already has a 32-bit address. + * Eg. if MC is configured to put VRAM in the low address range. + */ + if (gpu_addr <= max_vcpu_bo_addr) + return 0; + + /* Check if we can map the VCPU BO in GART to a 32-bit address. */ + if (adev->gmc.gart_start + VCE_V1_0_GART_ADDR_START > max_vcpu_bo_addr) + return -EINVAL; + + amdgpu_gart_map_vram_range(adev, pa, VCE_V1_0_GART_PAGE_START, + num_pages, flags, adev->gart.ptr); + adev->vce.gpu_addr = adev->gmc.gart_start + VCE_V1_0_GART_ADDR_START; + if (adev->vce.gpu_addr > max_vcpu_bo_addr) + return -EINVAL; + + return 0; +} + +static int vce_v1_0_sw_init(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ring *ring; + int r, i; + + r = amdgpu_irq_add_id(adev, AMDGPU_IRQ_CLIENTID_LEGACY, 167, &adev->vce.irq); + if (r) + return r; + + r = amdgpu_vce_sw_init(adev, VCE_V1_0_FW_SIZE + + VCE_V1_0_STACK_SIZE + VCE_V1_0_DATA_SIZE); + if (r) + return r; + + r = amdgpu_vce_resume(adev); + if (r) + return r; + r = vce_v1_0_load_fw_signature(adev); + if (r) + return r; + r = vce_v1_0_ensure_vcpu_bo_32bit_addr(adev); + if (r) + return r; + + for (i = 0; i < adev->vce.num_rings; i++) { + enum amdgpu_ring_priority_level hw_prio = amdgpu_vce_get_ring_prio(i); + + ring = &adev->vce.ring[i]; + sprintf(ring->name, "vce%d", i); + r = amdgpu_ring_init(adev, ring, 512, &adev->vce.irq, 0, + hw_prio, NULL); + if (r) + return r; + } + + return r; +} + +static int vce_v1_0_sw_fini(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + int r; + + r = amdgpu_vce_suspend(adev); + if (r) + return r; + + return amdgpu_vce_sw_fini(adev); +} + +/** + * vce_v1_0_hw_init - start and test VCE block + * + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. + * + * Initialize the hardware, boot up the VCPU and do some testing + */ +static int vce_v1_0_hw_init(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + int i, r; + + if (adev->pm.dpm_enabled) + amdgpu_dpm_enable_vce(adev, true); + else + amdgpu_asic_set_vce_clocks(adev, 10000, 10000); + + for (i = 0; i < adev->vce.num_rings; i++) { + r = amdgpu_ring_test_helper(&adev->vce.ring[i]); + if (r) + return r; + } + + dev_info(adev->dev, "VCE initialized successfully.\n"); + + return 0; +} + +static int vce_v1_0_hw_fini(struct amdgpu_ip_block *ip_block) +{ + int r; + + r = vce_v1_0_stop(ip_block->adev); + if (r) + return r; + + cancel_delayed_work_sync(&ip_block->adev->vce.idle_work); + return 0; +} + +static int vce_v1_0_suspend(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + int r; + + /* + * Proper cleanups before halting the HW engine: + * - cancel the delayed idle work + * - enable powergating + * - enable clockgating + * - disable dpm + * + * TODO: to align with the VCN implementation, move the + * jobs for clockgating/powergating/dpm setting to + * ->set_powergating_state(). + */ + cancel_delayed_work_sync(&adev->vce.idle_work); + + if (adev->pm.dpm_enabled) { + amdgpu_dpm_enable_vce(adev, false); + } else { + amdgpu_asic_set_vce_clocks(adev, 0, 0); + amdgpu_device_ip_set_powergating_state(adev, AMD_IP_BLOCK_TYPE_VCE, + AMD_PG_STATE_GATE); + amdgpu_device_ip_set_clockgating_state(adev, AMD_IP_BLOCK_TYPE_VCE, + AMD_CG_STATE_GATE); + } + + r = vce_v1_0_hw_fini(ip_block); + if (r) { + dev_err(adev->dev, "vce_v1_0_hw_fini() failed with error %i", r); + return r; + } + + return amdgpu_vce_suspend(adev); +} + +static int vce_v1_0_resume(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + int r; + + r = amdgpu_vce_resume(adev); + if (r) + return r; + r = vce_v1_0_load_fw_signature(adev); + if (r) + return r; + r = vce_v1_0_ensure_vcpu_bo_32bit_addr(adev); + if (r) + return r; + + return vce_v1_0_hw_init(ip_block); +} + +static int vce_v1_0_set_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + uint32_t val = 0; + + if (state == AMDGPU_IRQ_STATE_ENABLE) + val |= VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK; + + WREG32_P(mmVCE_SYS_INT_EN, val, + ~VCE_SYS_INT_EN__VCE_SYS_INT_TRAP_INTERRUPT_EN_MASK); + return 0; +} + +static int vce_v1_0_process_interrupt(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + dev_dbg(adev->dev, "IH: VCE\n"); + switch (entry->src_data[0]) { + case 0: + case 1: + amdgpu_fence_process(&adev->vce.ring[entry->src_data[0]]); + break; + default: + dev_err(adev->dev, "Unhandled interrupt: %d %d\n", + entry->src_id, entry->src_data[0]); + break; + } + + return 0; +} + +static int vce_v1_0_set_clockgating_state(struct amdgpu_ip_block *ip_block, + enum amd_clockgating_state state) +{ + struct amdgpu_device *adev = ip_block->adev; + + vce_v1_0_init_cg(adev); + vce_v1_0_enable_mgcg(adev, state == AMD_CG_STATE_GATE); + + return 0; +} + +static int vce_v1_0_set_powergating_state(struct amdgpu_ip_block *ip_block, + enum amd_powergating_state state) +{ + struct amdgpu_device *adev = ip_block->adev; + + /* + * This doesn't actually powergate the VCE block. + * That's done in the dpm code via the SMC. This + * just re-inits the block as necessary. The actual + * gating still happens in the dpm code. We should + * revisit this when there is a cleaner line between + * the smc and the hw blocks + */ + if (state == AMD_PG_STATE_GATE) + return vce_v1_0_stop(adev); + else + return vce_v1_0_start(adev); +} + +static const struct amd_ip_funcs vce_v1_0_ip_funcs = { + .name = "vce_v1_0", + .early_init = vce_v1_0_early_init, + .sw_init = vce_v1_0_sw_init, + .sw_fini = vce_v1_0_sw_fini, + .hw_init = vce_v1_0_hw_init, + .hw_fini = vce_v1_0_hw_fini, + .suspend = vce_v1_0_suspend, + .resume = vce_v1_0_resume, + .is_idle = vce_v1_0_is_idle, + .wait_for_idle = vce_v1_0_wait_for_idle, + .set_clockgating_state = vce_v1_0_set_clockgating_state, + .set_powergating_state = vce_v1_0_set_powergating_state, +}; + +static const struct amdgpu_ring_funcs vce_v1_0_ring_funcs = { + .type = AMDGPU_RING_TYPE_VCE, + .align_mask = 0xf, + .nop = VCE_CMD_NO_OP, + .support_64bit_ptrs = false, + .no_user_fence = true, + .get_rptr = vce_v1_0_ring_get_rptr, + .get_wptr = vce_v1_0_ring_get_wptr, + .set_wptr = vce_v1_0_ring_set_wptr, + .parse_cs = amdgpu_vce_ring_parse_cs, + .emit_frame_size = 6, /* amdgpu_vce_ring_emit_fence x1 no user fence */ + .emit_ib_size = 4, /* amdgpu_vce_ring_emit_ib */ + .emit_ib = amdgpu_vce_ring_emit_ib, + .emit_fence = amdgpu_vce_ring_emit_fence, + .test_ring = amdgpu_vce_ring_test_ring, + .test_ib = amdgpu_vce_ring_test_ib, + .insert_nop = amdgpu_ring_insert_nop, + .pad_ib = amdgpu_ring_generic_pad_ib, + .begin_use = amdgpu_vce_ring_begin_use, + .end_use = amdgpu_vce_ring_end_use, +}; + +static void vce_v1_0_set_ring_funcs(struct amdgpu_device *adev) +{ + int i; + + for (i = 0; i < adev->vce.num_rings; i++) { + adev->vce.ring[i].funcs = &vce_v1_0_ring_funcs; + adev->vce.ring[i].me = i; + } +}; + +static const struct amdgpu_irq_src_funcs vce_v1_0_irq_funcs = { + .set = vce_v1_0_set_interrupt_state, + .process = vce_v1_0_process_interrupt, +}; + +static void vce_v1_0_set_irq_funcs(struct amdgpu_device *adev) +{ + adev->vce.irq.num_types = 1; + adev->vce.irq.funcs = &vce_v1_0_irq_funcs; +}; + +const struct amdgpu_ip_block_version vce_v1_0_ip_block = { + .type = AMD_IP_BLOCK_TYPE_VCE, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &vce_v1_0_ip_funcs, +}; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v1_0.h b/drivers/gpu/drm/amd/amdgpu/vce_v1_0.h new file mode 100644 index 000000000000..206e7bec897f --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/vce_v1_0.h @@ -0,0 +1,32 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * Copyright 2025 Valve Corporation + * Copyright 2025 Alexandre Demers + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __VCE_V1_0_H__ +#define __VCE_V1_0_H__ + +extern const struct amdgpu_ip_block_version vce_v1_0_ip_block; + +#endif diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c index bee3e904a6bc..8ea8a6193492 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v2_0.c @@ -407,6 +407,11 @@ static void vce_v2_0_enable_mgcg(struct amdgpu_device *adev, bool enable, static int vce_v2_0_early_init(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; + int r; + + r = amdgpu_vce_early_init(adev); + if (r) + return r; adev->vce.num_rings = 2; diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c index 708123899c41..719e9643c43d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v3_0.c @@ -399,6 +399,7 @@ static unsigned vce_v3_0_get_harvest_config(struct amdgpu_device *adev) static int vce_v3_0_early_init(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; + int r; adev->vce.harvest_config = vce_v3_0_get_harvest_config(adev); @@ -407,6 +408,10 @@ static int vce_v3_0_early_init(struct amdgpu_ip_block *ip_block) (AMDGPU_VCE_HARVEST_VCE0 | AMDGPU_VCE_HARVEST_VCE1)) return -ENOENT; + r = amdgpu_vce_early_init(adev); + if (r) + return r; + adev->vce.num_rings = 3; vce_v3_0_set_ring_funcs(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c index 335bda64ff5b..2d64002bed61 100644 --- a/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vce_v4_0.c @@ -410,6 +410,11 @@ static int vce_v4_0_stop(struct amdgpu_device *adev) static int vce_v4_0_early_init(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; + int r; + + r = amdgpu_vce_early_init(adev); + if (r) + return r; if (amdgpu_sriov_vf(adev)) /* currently only VCN0 support SRIOV */ adev->vce.num_rings = 1; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index eacf4e93ba2f..cb7123ec1a5d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -141,7 +141,7 @@ static int vcn_v4_0_3_late_init(struct amdgpu_ip_block *ip_block) adev->vcn.supported_reset = amdgpu_get_soft_full_reset_mask(&adev->vcn.inst[0].ring_enc[0]); - if (amdgpu_dpm_reset_vcn_is_supported(adev)) + if (amdgpu_dpm_reset_vcn_is_supported(adev) && !amdgpu_sriov_vf(adev)) adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; return 0; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c index 714350cabf2f..8bd457dea4cf 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c @@ -122,7 +122,9 @@ static int vcn_v5_0_1_late_init(struct amdgpu_ip_block *ip_block) switch (amdgpu_ip_version(adev, MP0_HWIP, 0)) { case IP_VERSION(13, 0, 12): - if ((adev->psp.sos.fw_version >= 0x00450025) && amdgpu_dpm_reset_vcn_is_supported(adev)) + if ((adev->psp.sos.fw_version >= 0x00450025) && + amdgpu_dpm_reset_vcn_is_supported(adev) && + !amdgpu_sriov_vf(adev)) adev->vcn.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; break; default: diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 0f0719528bcc..22925df6a791 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -2826,7 +2826,7 @@ retry: static int runtime_disable(struct kfd_process *p) { - int i = 0, ret; + int i = 0, ret = 0; bool was_enabled = p->runtime_info.runtime_state == DEBUG_RUNTIME_STATE_ENABLED; p->runtime_info.runtime_state = DEBUG_RUNTIME_STATE_DISABLED; @@ -2863,6 +2863,7 @@ static int runtime_disable(struct kfd_process *p) /* disable ttmp setup */ for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; + int last_err = 0; if (kfd_dbg_is_per_vmid_supported(pdd->dev)) { pdd->spi_dbg_override = @@ -2872,14 +2873,17 @@ static int runtime_disable(struct kfd_process *p) pdd->dev->vm_info.last_vmid_kfd); if (!pdd->dev->kfd->shared_resources.enable_mes) - debug_refresh_runlist(pdd->dev->dqm); + last_err = debug_refresh_runlist(pdd->dev->dqm); else - kfd_dbg_set_mes_debug_mode(pdd, + last_err = kfd_dbg_set_mes_debug_mode(pdd, !kfd_dbg_has_cwsr_workaround(pdd->dev)); + + if (last_err) + ret = last_err; } } - return 0; + return ret; } static int kfd_ioctl_runtime_enable(struct file *filep, struct kfd_process *p, void *data) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index 6c5c7c1bf5ed..d7a2e7178ea9 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -1209,6 +1209,15 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, pr_debug_ratelimited("Evicting process pid %d queues\n", pdd->process->lead_thread->pid); + if (dqm->dev->kfd->shared_resources.enable_mes) { + pdd->last_evict_timestamp = get_jiffies_64(); + retval = suspend_all_queues_mes(dqm); + if (retval) { + dev_err(dev, "Suspending all queues failed"); + goto out; + } + } + /* Mark all queues as evicted. Deactivate all active queues on * the qpd. */ @@ -1221,23 +1230,27 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, decrement_queue_count(dqm, qpd, q); if (dqm->dev->kfd->shared_resources.enable_mes) { - int err; - - err = remove_queue_mes(dqm, q, qpd); - if (err) { + retval = remove_queue_mes(dqm, q, qpd); + if (retval) { dev_err(dev, "Failed to evict queue %d\n", q->properties.queue_id); - retval = err; + goto out; } } } - pdd->last_evict_timestamp = get_jiffies_64(); - if (!dqm->dev->kfd->shared_resources.enable_mes) + + if (!dqm->dev->kfd->shared_resources.enable_mes) { + pdd->last_evict_timestamp = get_jiffies_64(); retval = execute_queues_cpsch(dqm, qpd->is_debug ? KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES : KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD); + } else { + retval = resume_all_queues_mes(dqm); + if (retval) + dev_err(dev, "Resuming all queues failed"); + } out: dqm_unlock(dqm); @@ -1884,6 +1897,8 @@ fail_packet_manager_init: static int stop_cpsch(struct device_queue_manager *dqm) { + int ret = 0; + dqm_lock(dqm); if (!dqm->sched_running) { dqm_unlock(dqm); @@ -1891,9 +1906,10 @@ static int stop_cpsch(struct device_queue_manager *dqm) } if (!dqm->dev->kfd->shared_resources.enable_mes) - unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, 0, USE_DEFAULT_GRACE_PERIOD, false); + ret = unmap_queues_cpsch(dqm, KFD_UNMAP_QUEUES_FILTER_ALL_QUEUES, + 0, USE_DEFAULT_GRACE_PERIOD, false); else - remove_all_kfd_queues_mes(dqm); + ret = remove_all_kfd_queues_mes(dqm); dqm->sched_running = false; @@ -1907,7 +1923,7 @@ static int stop_cpsch(struct device_queue_manager *dqm) dqm->detect_hang_info = NULL; dqm_unlock(dqm); - return 0; + return ret; } static int create_kernel_queue_cpsch(struct device_queue_manager *dqm, @@ -2078,7 +2094,8 @@ int amdkfd_fence_wait_timeout(struct device_queue_manager *dqm, while (*fence_addr != fence_value) { /* Fatal err detected, this response won't come */ - if (amdgpu_amdkfd_is_fed(dqm->dev->adev)) + if (amdgpu_amdkfd_is_fed(dqm->dev->adev) || + amdgpu_in_reset(dqm->dev->adev)) return -EIO; if (time_after(jiffies, end_jiffies)) { @@ -3098,61 +3115,17 @@ out: return ret; } -static int kfd_dqm_evict_pasid_mes(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) -{ - struct device *dev = dqm->dev->adev->dev; - int ret = 0; - - /* Check if process is already evicted */ - dqm_lock(dqm); - if (qpd->evicted) { - /* Increment the evicted count to make sure the - * process stays evicted before its terminated. - */ - qpd->evicted++; - dqm_unlock(dqm); - goto out; - } - dqm_unlock(dqm); - - ret = suspend_all_queues_mes(dqm); - if (ret) { - dev_err(dev, "Suspending all queues failed"); - goto out; - } - - ret = dqm->ops.evict_process_queues(dqm, qpd); - if (ret) { - dev_err(dev, "Evicting process queues failed"); - goto out; - } - - ret = resume_all_queues_mes(dqm); - if (ret) - dev_err(dev, "Resuming all queues failed"); - -out: - return ret; -} - int kfd_evict_process_device(struct kfd_process_device *pdd) { struct device_queue_manager *dqm; struct kfd_process *p; - int ret = 0; p = pdd->process; dqm = pdd->dev->dqm; WARN(debug_evictions, "Evicting pid %d", p->lead_thread->pid); - if (dqm->dev->kfd->shared_resources.enable_mes) - ret = kfd_dqm_evict_pasid_mes(dqm, &pdd->qpd); - else - ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd); - - return ret; + return dqm->ops.evict_process_queues(dqm, &pdd->qpd); } int reserve_debug_trap_vmid(struct device_queue_manager *dqm, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index 82905f3e54dd..5a190dd6be4e 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -748,16 +748,6 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id, uint64_t *slots = page_slots(p->signal_page); uint32_t id; - /* - * If id is valid but slot is not signaled, GPU may signal the same event twice - * before driver have chance to process the first interrupt, then signal slot is - * auto-reset after set_event wakeup the user space, just drop the second event as - * the application only need wakeup once. - */ - if ((valid_id_bits > 31 || (1U << valid_id_bits) >= KFD_SIGNAL_EVENT_LIMIT) && - partial_id < KFD_SIGNAL_EVENT_LIMIT && slots[partial_id] == UNSIGNALED_EVENT_SLOT) - goto out_unlock; - if (valid_id_bits) pr_debug_ratelimited("Partial ID invalid: %u (%u valid bits)\n", partial_id, valid_id_bits); @@ -786,7 +776,6 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id, } } -out_unlock: rcu_read_unlock(); kfd_unref_process(p); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c index 4ceb251312a6..d76fb61869c7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c @@ -28,6 +28,7 @@ #include "kfd_device_queue_manager.h" #include "kfd_smi_events.h" #include "amdgpu_ras.h" +#include "amdgpu_ras_mgr.h" /* * GFX9 SQ Interrupts @@ -228,7 +229,11 @@ static void event_interrupt_poison_consumption_v9(struct kfd_node *dev, kfd_signal_poison_consumed_event(dev, pasid); - event_id = amdgpu_ras_acquire_event_id(dev->adev, type); + if (amdgpu_uniras_enabled(dev->adev)) + event_id = amdgpu_ras_mgr_gen_ras_event_seqno(dev->adev, + RAS_SEQNO_TYPE_POISON_CONSUMPTION); + else + event_id = amdgpu_ras_acquire_event_id(dev->adev, type); RAS_EVENT_LOG(dev->adev, event_id, "poison is consumed by client %d, kick off gpu reset flow\n", client_id); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c index 59a5a3fea65d..46c84fc60af1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c @@ -21,7 +21,6 @@ * OTHER DEALINGS IN THE SOFTWARE. */ #include <linux/types.h> -#include <linux/hmm.h> #include <linux/dma-direction.h> #include <linux/dma-mapping.h> #include <linux/migrate.h> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h index 2eebf67f9c2c..2b7fd442d29c 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.h @@ -31,7 +31,6 @@ #include <linux/list.h> #include <linux/mutex.h> #include <linux/sched/mm.h> -#include <linux/hmm.h> #include "kfd_priv.h" #include "kfd_svm.h" diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index ddfe30c13e9d..a085faac9fe1 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -1083,7 +1083,6 @@ static void kfd_process_destroy_pdds(struct kfd_process *p) * for auto suspend */ if (pdd->runtime_inuse) { - pm_runtime_mark_last_busy(adev_to_drm(pdd->dev->adev)->dev); pm_runtime_put_autosuspend(adev_to_drm(pdd->dev->adev)->dev); pdd->runtime_inuse = false; } @@ -1162,9 +1161,6 @@ static void kfd_process_wq_release(struct work_struct *work) release_work); struct dma_fence *ef; - kfd_process_dequeue_from_all_devices(p); - pqm_uninit(&p->pqm); - /* * If GPU in reset, user queues may still running, wait for reset complete. */ @@ -1226,6 +1222,14 @@ static void kfd_process_notifier_release_internal(struct kfd_process *p) cancel_delayed_work_sync(&p->eviction_work); cancel_delayed_work_sync(&p->restore_work); + /* + * Dequeue and destroy user queues, it is not safe for GPU to access + * system memory after mmu release notifier callback returns because + * exit_mmap free process memory afterwards. + */ + kfd_process_dequeue_from_all_devices(p); + pqm_uninit(&p->pqm); + for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c index a65c67cf56ff..f1e7583650c4 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_queue.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_queue.c @@ -297,16 +297,16 @@ int kfd_queue_acquire_buffers(struct kfd_process_device *pdd, struct queue_prope goto out_err_unreserve; } - if (properties->ctx_save_restore_area_size != topo_dev->node_props.cwsr_size) { - pr_debug("queue cwsr size 0x%x not equal to node cwsr size 0x%x\n", + if (properties->ctx_save_restore_area_size < topo_dev->node_props.cwsr_size) { + pr_debug("queue cwsr size 0x%x not sufficient for node cwsr size 0x%x\n", properties->ctx_save_restore_area_size, topo_dev->node_props.cwsr_size); err = -EINVAL; goto out_err_unreserve; } - total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev->node_props.debug_memory_size) - * NUM_XCC(pdd->dev->xcc_mask); + total_cwsr_size = (properties->ctx_save_restore_area_size + + topo_dev->node_props.debug_memory_size) * NUM_XCC(pdd->dev->xcc_mask); total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE); err = kfd_queue_buffer_get(vm, (void *)properties->ctx_save_restore_area_address, @@ -352,8 +352,8 @@ int kfd_queue_release_buffers(struct kfd_process_device *pdd, struct queue_prope topo_dev = kfd_topology_device_by_id(pdd->dev->id); if (!topo_dev) return -EINVAL; - total_cwsr_size = (topo_dev->node_props.cwsr_size + topo_dev->node_props.debug_memory_size) - * NUM_XCC(pdd->dev->xcc_mask); + total_cwsr_size = (properties->ctx_save_restore_area_size + + topo_dev->node_props.debug_memory_size) * NUM_XCC(pdd->dev->xcc_mask); total_cwsr_size = ALIGN(total_cwsr_size, PAGE_SIZE); kfd_queue_buffer_svm_put(pdd, properties->ctx_save_restore_area_address, total_cwsr_size); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 9d72411c3379..97c2270f278f 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1698,7 +1698,7 @@ static int svm_range_validate_and_map(struct mm_struct *mm, start = map_start << PAGE_SHIFT; end = (map_last + 1) << PAGE_SHIFT; for (addr = start; !r && addr < end; ) { - struct hmm_range *hmm_range = NULL; + struct amdgpu_hmm_range *range = NULL; unsigned long map_start_vma; unsigned long map_last_vma; struct vm_area_struct *vma; @@ -1737,9 +1737,12 @@ static int svm_range_validate_and_map(struct mm_struct *mm, } WRITE_ONCE(p->svms.faulting_task, current); - r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, - readonly, owner, - &hmm_range); + range = amdgpu_hmm_range_alloc(NULL); + if (likely(range)) + r = amdgpu_hmm_range_get_pages(&prange->notifier, addr, npages, + readonly, owner, range); + else + r = -ENOMEM; WRITE_ONCE(p->svms.faulting_task, NULL); if (r) pr_debug("failed %d to get svm range pages\n", r); @@ -1750,7 +1753,7 @@ static int svm_range_validate_and_map(struct mm_struct *mm, if (!r) { offset = (addr >> PAGE_SHIFT) - prange->start; r = svm_range_dma_map(prange, ctx->bitmap, offset, npages, - hmm_range->hmm_pfns); + range->hmm_range.hmm_pfns); if (r) pr_debug("failed %d to dma map range\n", r); } @@ -1758,14 +1761,17 @@ static int svm_range_validate_and_map(struct mm_struct *mm, svm_range_lock(prange); /* Free backing memory of hmm_range if it was initialized - * Overrride return value to TRY AGAIN only if prior returns + * Override return value to TRY AGAIN only if prior returns * were successful */ - if (hmm_range && amdgpu_hmm_range_get_pages_done(hmm_range) && !r) { + if (range && !amdgpu_hmm_range_valid(range) && !r) { pr_debug("hmm update the range, need validate again\n"); r = -EAGAIN; } + /* Free the hmm range */ + amdgpu_hmm_range_free(range); + if (!r && !list_empty(&prange->child_list)) { pr_debug("range split by unmap in parallel, validate again\n"); r = -EAGAIN; @@ -3687,6 +3693,8 @@ svm_range_set_attr(struct kfd_process *p, struct mm_struct *mm, svm_range_apply_attrs(p, prange, nattr, attrs, &update_mapping); /* TODO: unmap ranges from GPU that lost access */ } + update_mapping |= !p->xnack_enabled && !list_empty(&remap_list); + list_for_each_entry_safe(prange, next, &remove_list, update_list) { pr_debug("unlink old 0x%p prange 0x%p [0x%lx 0x%lx]\n", prange->svms, prange, prange->start, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h index 01c7a4877904..a63dfc95b602 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h @@ -31,7 +31,6 @@ #include <linux/list.h> #include <linux/mutex.h> #include <linux/sched/mm.h> -#include <linux/hmm.h> #include "amdgpu.h" #include "kfd_priv.h" diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index 5c98746eb72d..811636af14ea 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -530,7 +530,9 @@ static ssize_t node_show(struct kobject *kobj, struct attribute *attr, sysfs_show_32bit_prop(buffer, offs, "sdma_fw_version", dev->gpu->kfd->sdma_fw_version); sysfs_show_64bit_prop(buffer, offs, "unique_id", - dev->gpu->xcp ? + dev->gpu->xcp && + (dev->gpu->xcp->xcp_mgr->mode != + AMDGPU_SPX_PARTITION_MODE) ? dev->gpu->xcp->unique_id : dev->gpu->adev->unique_id); sysfs_show_32bit_prop(buffer, offs, "num_xcc", diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile index 7329b8cc2576..8e949fe77312 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/Makefile @@ -39,7 +39,8 @@ AMDGPUDM = \ amdgpu_dm_psr.o \ amdgpu_dm_replay.o \ amdgpu_dm_quirks.o \ - amdgpu_dm_wb.o + amdgpu_dm_wb.o \ + amdgpu_dm_colorop.o ifdef CONFIG_DRM_AMD_DC_FP AMDGPUDM += dc_fpu.o diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 0d03e324d5b9..740711ac1037 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -551,13 +551,13 @@ static void schedule_dc_vmin_vmax(struct amdgpu_device *adev, struct dc_stream_state *stream, struct dc_crtc_timing_adjust *adjust) { - struct vupdate_offload_work *offload_work = kzalloc(sizeof(*offload_work), GFP_KERNEL); + struct vupdate_offload_work *offload_work = kzalloc(sizeof(*offload_work), GFP_NOWAIT); if (!offload_work) { drm_dbg_driver(adev_to_drm(adev), "Failed to allocate vupdate_offload_work\n"); return; } - struct dc_crtc_timing_adjust *adjust_copy = kzalloc(sizeof(*adjust_copy), GFP_KERNEL); + struct dc_crtc_timing_adjust *adjust_copy = kzalloc(sizeof(*adjust_copy), GFP_NOWAIT); if (!adjust_copy) { drm_dbg_driver(adev_to_drm(adev), "Failed to allocate adjust_copy\n"); kfree(offload_work); @@ -2085,8 +2085,6 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) dc_hardware_init(adev->dm.dc); - adev->dm.restore_backlight = true; - adev->dm.hpd_rx_offload_wq = hpd_rx_irq_create_workqueue(adev); if (!adev->dm.hpd_rx_offload_wq) { drm_err(adev_to_drm(adev), "failed to create hpd rx offload workqueue.\n"); @@ -3394,6 +3392,67 @@ static void apply_delay_after_dpcd_poweroff(struct amdgpu_device *adev, } } +/** + * amdgpu_dm_dump_links_and_sinks - Debug dump of all DC links and their sinks + * @adev: amdgpu device pointer + * + * Iterates through all DC links and dumps information about local and remote + * (MST) sinks. Should be called after connector detection is complete to see + * the final state of all links. + */ +static void amdgpu_dm_dump_links_and_sinks(struct amdgpu_device *adev) +{ + struct dc *dc = adev->dm.dc; + struct drm_device *dev = adev_to_drm(adev); + int li; + + if (!dc) + return; + + for (li = 0; li < dc->link_count; li++) { + struct dc_link *l = dc->links[li]; + const char *name = NULL; + int rs; + + if (!l) + continue; + if (l->local_sink && l->local_sink->edid_caps.display_name[0]) + name = l->local_sink->edid_caps.display_name; + else + name = "n/a"; + + drm_dbg_kms(dev, + "LINK_DUMP[%d]: local_sink=%p type=%d sink_signal=%d sink_count=%u edid_name=%s mst_capable=%d mst_alloc_streams=%d\n", + li, + l->local_sink, + l->type, + l->local_sink ? l->local_sink->sink_signal : SIGNAL_TYPE_NONE, + l->sink_count, + name, + l->dpcd_caps.is_mst_capable, + l->mst_stream_alloc_table.stream_count); + + /* Dump remote (MST) sinks if any */ + for (rs = 0; rs < l->sink_count; rs++) { + struct dc_sink *rsink = l->remote_sinks[rs]; + const char *rname = NULL; + + if (!rsink) + continue; + if (rsink->edid_caps.display_name[0]) + rname = rsink->edid_caps.display_name; + else + rname = "n/a"; + drm_dbg_kms(dev, + " REMOTE_SINK[%d:%d]: sink=%p signal=%d edid_name=%s\n", + li, rs, + rsink, + rsink->sink_signal, + rname); + } + } +} + static int dm_resume(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; @@ -3442,7 +3501,6 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) dc_set_power_state(dm->dc, DC_ACPI_CM_POWER_STATE_D0); dc_resume(dm->dc); - adev->dm.restore_backlight = true; amdgpu_dm_irq_resume_early(adev); @@ -3566,6 +3624,7 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) /* Do mst topology probing after resuming cached state*/ drm_connector_list_iter_begin(ddev, &iter); drm_for_each_connector_iter(connector, &iter) { + bool init = false; if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK) continue; @@ -3575,10 +3634,23 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) aconnector->mst_root) continue; - drm_dp_mst_topology_queue_probe(&aconnector->mst_mgr); + scoped_guard(mutex, &aconnector->mst_mgr.lock) { + init = !aconnector->mst_mgr.mst_primary; + } + if (init) + dm_helpers_dp_mst_start_top_mgr(aconnector->dc_link->ctx, + aconnector->dc_link, false); + else + drm_dp_mst_topology_queue_probe(&aconnector->mst_mgr); } drm_connector_list_iter_end(&iter); + /* Debug dump: list all DC links and their associated sinks after detection + * is complete for all connectors. This provides a comprehensive view of the + * final state without repeating the dump for each connector. + */ + amdgpu_dm_dump_links_and_sinks(adev); + amdgpu_dm_irq_resume_late(adev); amdgpu_dm_smu_write_watermarks_table(adev); @@ -3789,7 +3861,9 @@ void amdgpu_dm_update_connector_after_detect( drm_dbg_kms(dev, "DCHPD: connector_id=%d: Old sink=%p New sink=%p\n", aconnector->connector_id, aconnector->dc_sink, sink); - guard(mutex)(&dev->mode_config.mutex); + /* When polling, DRM has already locked the mutex for us. */ + if (!drm_kms_helper_is_poll_worker()) + mutex_lock(&dev->mode_config.mutex); /* * 1. Update status of the drm connector @@ -3852,6 +3926,101 @@ void amdgpu_dm_update_connector_after_detect( } update_subconnector_property(aconnector); + + /* When polling, the mutex will be unlocked for us by DRM. */ + if (!drm_kms_helper_is_poll_worker()) + mutex_unlock(&dev->mode_config.mutex); +} + +static bool are_sinks_equal(const struct dc_sink *sink1, const struct dc_sink *sink2) +{ + if (!sink1 || !sink2) + return false; + if (sink1->sink_signal != sink2->sink_signal) + return false; + + if (sink1->dc_edid.length != sink2->dc_edid.length) + return false; + + if (memcmp(sink1->dc_edid.raw_edid, sink2->dc_edid.raw_edid, + sink1->dc_edid.length) != 0) + return false; + return true; +} + + +/** + * DOC: hdmi_hpd_debounce_work + * + * HDMI HPD debounce delay in milliseconds. When an HDMI display toggles HPD + * (such as during power save transitions), this delay determines how long to + * wait before processing the HPD event. This allows distinguishing between a + * physical unplug (>hdmi_hpd_debounce_delay) + * and a spontaneous RX HPD toggle (<hdmi_hpd_debounce_delay). + * + * If the toggle is less than this delay, the driver compares sink capabilities + * and permits a hotplug event if they changed. + * + * The default value of 1500ms was chosen based on experimental testing with + * various monitors that exhibit spontaneous HPD toggling behavior. + */ +static void hdmi_hpd_debounce_work(struct work_struct *work) +{ + struct amdgpu_dm_connector *aconnector = + container_of(to_delayed_work(work), struct amdgpu_dm_connector, + hdmi_hpd_debounce_work); + struct drm_connector *connector = &aconnector->base; + struct drm_device *dev = connector->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + struct dc *dc = aconnector->dc_link->ctx->dc; + bool fake_reconnect = false; + bool reallow_idle = false; + bool ret = false; + guard(mutex)(&aconnector->hpd_lock); + + /* Re-detect the display */ + scoped_guard(mutex, &adev->dm.dc_lock) { + if (dc->caps.ips_support && dc->ctx->dmub_srv->idle_allowed) { + dc_allow_idle_optimizations(dc, false); + reallow_idle = true; + } + ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_HPD); + } + + if (ret) { + /* Apply workaround delay for certain panels */ + apply_delay_after_dpcd_poweroff(adev, aconnector->dc_sink); + /* Compare sinks to determine if this was a spontaneous HPD toggle */ + if (are_sinks_equal(aconnector->dc_link->local_sink, aconnector->hdmi_prev_sink)) { + /* + * Sinks match - this was a spontaneous HDMI HPD toggle. + */ + drm_dbg_kms(dev, "HDMI HPD: Sink unchanged after debounce, internal re-enable\n"); + fake_reconnect = true; + } + + /* Update connector state */ + amdgpu_dm_update_connector_after_detect(aconnector); + + drm_modeset_lock_all(dev); + dm_restore_drm_connector_state(dev, connector); + drm_modeset_unlock_all(dev); + + /* Only notify OS if sink actually changed */ + if (!fake_reconnect && aconnector->base.force == DRM_FORCE_UNSPECIFIED) + drm_kms_helper_hotplug_event(dev); + } + + /* Release the cached sink reference */ + if (aconnector->hdmi_prev_sink) { + dc_sink_release(aconnector->hdmi_prev_sink); + aconnector->hdmi_prev_sink = NULL; + } + + scoped_guard(mutex, &adev->dm.dc_lock) { + if (reallow_idle && dc->caps.ips_support) + dc_allow_idle_optimizations(dc, true); + } } static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector) @@ -3863,6 +4032,7 @@ static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector) struct dm_connector_state *dm_con_state = to_dm_connector_state(connector->state); struct dc *dc = aconnector->dc_link->ctx->dc; bool ret = false; + bool debounce_required = false; if (adev->dm.disable_hpd_irq) return; @@ -3885,6 +4055,14 @@ static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector) if (!dc_link_detect_connection_type(aconnector->dc_link, &new_connection_type)) drm_err(adev_to_drm(adev), "KMS: Failed to detect connector\n"); + /* + * Check for HDMI disconnect with debounce enabled. + */ + debounce_required = (aconnector->hdmi_hpd_debounce_delay_ms > 0 && + dc_is_hdmi_signal(aconnector->dc_link->connector_signal) && + new_connection_type == dc_connection_none && + aconnector->dc_link->local_sink != NULL); + if (aconnector->base.force && new_connection_type == dc_connection_none) { emulated_link_detect(aconnector->dc_link); @@ -3894,7 +4072,34 @@ static void handle_hpd_irq_helper(struct amdgpu_dm_connector *aconnector) if (aconnector->base.force == DRM_FORCE_UNSPECIFIED) drm_kms_helper_connector_hotplug_event(connector); + } else if (debounce_required) { + /* + * HDMI disconnect detected - schedule delayed work instead of + * processing immediately. This allows us to coalesce spurious + * HDMI signals from physical unplugs. + */ + drm_dbg_kms(dev, "HDMI HPD: Disconnect detected, scheduling debounce work (%u ms)\n", + aconnector->hdmi_hpd_debounce_delay_ms); + + /* Cache the current sink for later comparison */ + if (aconnector->hdmi_prev_sink) + dc_sink_release(aconnector->hdmi_prev_sink); + aconnector->hdmi_prev_sink = aconnector->dc_link->local_sink; + if (aconnector->hdmi_prev_sink) + dc_sink_retain(aconnector->hdmi_prev_sink); + + /* Schedule delayed detection. */ + if (mod_delayed_work(system_wq, + &aconnector->hdmi_hpd_debounce_work, + msecs_to_jiffies(aconnector->hdmi_hpd_debounce_delay_ms))) + drm_dbg_kms(dev, "HDMI HPD: Re-scheduled debounce work\n"); + } else { + + /* If the aconnector->hdmi_hpd_debounce_work is scheduled, exit early */ + if (delayed_work_pending(&aconnector->hdmi_hpd_debounce_work)) + return; + scoped_guard(mutex, &adev->dm.dc_lock) { dc_exit_ips_for_hw_access(dc); ret = dc_link_detect(aconnector->dc_link, DETECT_REASON_HPD); @@ -4920,6 +5125,21 @@ static void amdgpu_dm_backlight_set_level(struct amdgpu_display_manager *dm, struct dc_link *link; u32 brightness; bool rc, reallow_idle = false; + struct drm_connector *connector; + + list_for_each_entry(connector, &dm->ddev->mode_config.connector_list, head) { + struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector); + + if (aconnector->bl_idx != bl_idx) + continue; + + /* if connector is off, save the brightness for next time it's on */ + if (!aconnector->base.encoder) { + dm->brightness[bl_idx] = user_brightness; + dm->actual_brightness[bl_idx] = 0; + return; + } + } amdgpu_dm_update_backlight_caps(dm, bl_idx); caps = &dm->backlight_caps[bl_idx]; @@ -5136,6 +5356,7 @@ static int initialize_plane(struct amdgpu_display_manager *dm, static void setup_backlight_device(struct amdgpu_display_manager *dm, struct amdgpu_dm_connector *aconnector) { + struct amdgpu_dm_backlight_caps *caps; struct dc_link *link = aconnector->dc_link; int bl_idx = dm->num_of_edps; @@ -5155,6 +5376,13 @@ static void setup_backlight_device(struct amdgpu_display_manager *dm, dm->num_of_edps++; update_connector_ext_caps(aconnector); + caps = &dm->backlight_caps[aconnector->bl_idx]; + + /* Only offer ABM property when non-OLED and user didn't turn off by module parameter */ + if (!caps->ext_caps->bits.oled && amdgpu_dm_abm_level < 0) + drm_object_attach_property(&aconnector->base.base, + dm->adev->mode_info.abm_level_property, + ABM_SYSFS_CONTROL); } static void amdgpu_set_panel_orientation(struct drm_connector *connector); @@ -5410,6 +5638,12 @@ static int amdgpu_dm_initialize_drm_device(struct amdgpu_device *adev) amdgpu_set_panel_orientation(&aconnector->base); } + /* Debug dump: list all DC links and their associated sinks after detection + * is complete for all connectors. This provides a comprehensive view of the + * final state without repeating the dump for each connector. + */ + amdgpu_dm_dump_links_and_sinks(adev); + /* Software is initialized. Now we can register interrupt handlers. */ switch (adev->asic_type) { #if defined(CONFIG_DRM_AMD_DC_SI) @@ -5796,6 +6030,10 @@ fill_plane_color_attributes(const struct drm_plane_state *plane_state, *color_space = COLOR_SPACE_SRGB; + /* Ignore properties when DRM_CLIENT_CAP_PLANE_COLOR_PIPELINE is set */ + if (plane_state->state && plane_state->state->plane_color_pipeline) + return 0; + /* DRM color properties only affect non-RGB formats. */ if (format < SURFACE_PIXEL_FORMAT_VIDEO_BEGIN) return 0; @@ -7148,29 +7386,117 @@ finish: return stream; } +/** + * amdgpu_dm_connector_poll - Poll a connector to see if it's connected to a display + * @aconnector: DM connector to poll (owns @base drm_connector and @dc_link) + * @force: if true, force polling even when DAC load detection was used + * + * Used for connectors that don't support HPD (hotplug detection) to + * periodically check whether the connector is connected to a display. + * + * When connection was determined via DAC load detection, we avoid + * re-running it on normal polls to prevent visible glitches, unless + * @force is set. + * + * Return: The probed connector status (connected/disconnected/unknown). + */ static enum drm_connector_status -amdgpu_dm_connector_detect(struct drm_connector *connector, bool force) +amdgpu_dm_connector_poll(struct amdgpu_dm_connector *aconnector, bool force) { - bool connected; - struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector); + struct drm_connector *connector = &aconnector->base; + struct drm_device *dev = connector->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + struct dc_link *link = aconnector->dc_link; + enum dc_connection_type conn_type = dc_connection_none; + enum drm_connector_status status = connector_status_disconnected; - /* - * Notes: - * 1. This interface is NOT called in context of HPD irq. - * 2. This interface *is called* in context of user-mode ioctl. Which - * makes it a bad place for *any* MST-related activity. + /* When we determined the connection using DAC load detection, + * do NOT poll the connector do detect disconnect because + * that would run DAC load detection again which can cause + * visible visual glitches. + * + * Only allow to poll such a connector again when forcing. */ + if (!force && link->local_sink && link->type == dc_connection_dac_load) + return connector->status; - if (aconnector->base.force == DRM_FORCE_UNSPECIFIED && - !aconnector->fake_enable) - connected = (aconnector->dc_sink != NULL); - else - connected = (aconnector->base.force == DRM_FORCE_ON || - aconnector->base.force == DRM_FORCE_ON_DIGITAL); + mutex_lock(&aconnector->hpd_lock); + + if (dc_link_detect_connection_type(aconnector->dc_link, &conn_type) && + conn_type != dc_connection_none) { + mutex_lock(&adev->dm.dc_lock); + + /* Only call full link detection when a sink isn't created yet, + * ie. just when the display is plugged in, otherwise we risk flickering. + */ + if (link->local_sink || + dc_link_detect(link, DETECT_REASON_HPD)) + status = connector_status_connected; + + mutex_unlock(&adev->dm.dc_lock); + } + + if (connector->status != status) { + if (status == connector_status_disconnected) { + if (link->local_sink) + dc_sink_release(link->local_sink); + + link->local_sink = NULL; + link->dpcd_sink_count = 0; + link->type = dc_connection_none; + } + + amdgpu_dm_update_connector_after_detect(aconnector); + } + + mutex_unlock(&aconnector->hpd_lock); + return status; +} + +/** + * amdgpu_dm_connector_detect() - Detect whether a DRM connector is connected to a display + * + * A connector is considered connected when it has a sink that is not NULL. + * For connectors that support HPD (hotplug detection), the connection is + * handled in the HPD interrupt. + * For connectors that may not support HPD, such as analog connectors, + * DRM will call this function repeatedly to poll them. + * + * Notes: + * 1. This interface is NOT called in context of HPD irq. + * 2. This interface *is called* in context of user-mode ioctl. Which + * makes it a bad place for *any* MST-related activity. + * + * @connector: The DRM connector we are checking. We convert it to + * amdgpu_dm_connector so we can read the DC link and state. + * @force: If true, do a full detect again. This is used even when + * a lighter check would normally be used to avoid flicker. + * + * Return: The connector status (connected, disconnected, or unknown). + * + */ +static enum drm_connector_status +amdgpu_dm_connector_detect(struct drm_connector *connector, bool force) +{ + struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector); update_subconnector_property(aconnector); - return (connected ? connector_status_connected : + if (aconnector->base.force == DRM_FORCE_ON || + aconnector->base.force == DRM_FORCE_ON_DIGITAL) + return connector_status_connected; + else if (aconnector->base.force == DRM_FORCE_OFF) + return connector_status_disconnected; + + /* Poll analog connectors and only when either + * disconnected or connected to an analog display. + */ + if (drm_kms_helper_is_poll_worker() && + dc_connector_supports_analog(aconnector->dc_link->link_id.id) && + (!aconnector->dc_sink || aconnector->dc_sink->edid_caps.analog)) + return amdgpu_dm_connector_poll(aconnector, force); + + return (aconnector->dc_sink ? connector_status_connected : connector_status_disconnected); } @@ -7221,6 +7547,20 @@ int amdgpu_dm_connector_atomic_set_property(struct drm_connector *connector, } else if (property == adev->mode_info.underscan_property) { dm_new_state->underscan_enable = val; ret = 0; + } else if (property == adev->mode_info.abm_level_property) { + switch (val) { + case ABM_SYSFS_CONTROL: + dm_new_state->abm_sysfs_forbidden = false; + break; + case ABM_LEVEL_OFF: + dm_new_state->abm_sysfs_forbidden = true; + dm_new_state->abm_level = ABM_LEVEL_IMMEDIATE_DISABLE; + break; + default: + dm_new_state->abm_sysfs_forbidden = true; + dm_new_state->abm_level = val; + } + ret = 0; } return ret; @@ -7263,6 +7603,13 @@ int amdgpu_dm_connector_atomic_get_property(struct drm_connector *connector, } else if (property == adev->mode_info.underscan_property) { *val = dm_state->underscan_enable; ret = 0; + } else if (property == adev->mode_info.abm_level_property) { + if (!dm_state->abm_sysfs_forbidden) + *val = ABM_SYSFS_CONTROL; + else + *val = (dm_state->abm_level != ABM_LEVEL_IMMEDIATE_DISABLE) ? + dm_state->abm_level : 0; + ret = 0; } return ret; @@ -7315,10 +7662,16 @@ static ssize_t panel_power_savings_store(struct device *device, return -EINVAL; drm_modeset_lock(&dev->mode_config.connection_mutex, NULL); - to_dm_connector_state(connector->state)->abm_level = val ?: - ABM_LEVEL_IMMEDIATE_DISABLE; + if (to_dm_connector_state(connector->state)->abm_sysfs_forbidden) + ret = -EBUSY; + else + to_dm_connector_state(connector->state)->abm_level = val ?: + ABM_LEVEL_IMMEDIATE_DISABLE; drm_modeset_unlock(&dev->mode_config.connection_mutex); + if (ret) + return ret; + drm_kms_helper_hotplug_event(dev); return count; @@ -7383,6 +7736,13 @@ static void amdgpu_dm_connector_destroy(struct drm_connector *connector) if (aconnector->mst_mgr.dev) drm_dp_mst_topology_mgr_destroy(&aconnector->mst_mgr); + /* Cancel and flush any pending HDMI HPD debounce work */ + cancel_delayed_work_sync(&aconnector->hdmi_hpd_debounce_work); + if (aconnector->hdmi_prev_sink) { + dc_sink_release(aconnector->hdmi_prev_sink); + aconnector->hdmi_prev_sink = NULL; + } + if (aconnector->bl_idx != -1) { backlight_device_unregister(dm->backlight_dev[aconnector->bl_idx]); dm->backlight_dev[aconnector->bl_idx] = NULL; @@ -8033,7 +8393,7 @@ static int dm_encoder_helper_atomic_check(struct drm_encoder *encoder, "mode %dx%d@%dHz is not native, enabling scaling\n", adjusted_mode->hdisplay, adjusted_mode->vdisplay, drm_mode_vrefresh(adjusted_mode)); - dm_new_connector_state->scaling = RMX_FULL; + dm_new_connector_state->scaling = RMX_ASPECT; } return 0; } @@ -8158,7 +8518,7 @@ static int dm_update_mst_vcpi_slots_for_dsc(struct drm_atomic_state *state, return 0; } -static int to_drm_connector_type(enum signal_type st) +static int to_drm_connector_type(enum signal_type st, uint32_t connector_id) { switch (st) { case SIGNAL_TYPE_HDMI_TYPE_A: @@ -8174,6 +8534,10 @@ static int to_drm_connector_type(enum signal_type st) return DRM_MODE_CONNECTOR_DisplayPort; case SIGNAL_TYPE_DVI_DUAL_LINK: case SIGNAL_TYPE_DVI_SINGLE_LINK: + if (connector_id == CONNECTOR_ID_SINGLE_LINK_DVII || + connector_id == CONNECTOR_ID_DUAL_LINK_DVII) + return DRM_MODE_CONNECTOR_DVII; + return DRM_MODE_CONNECTOR_DVID; case SIGNAL_TYPE_VIRTUAL: return DRM_MODE_CONNECTOR_VIRTUAL; @@ -8225,7 +8589,7 @@ static void amdgpu_dm_get_native_mode(struct drm_connector *connector) static struct drm_display_mode * amdgpu_dm_create_common_mode(struct drm_encoder *encoder, - char *name, + const char *name, int hdisplay, int vdisplay) { struct drm_device *dev = encoder->dev; @@ -8247,6 +8611,24 @@ amdgpu_dm_create_common_mode(struct drm_encoder *encoder, } +static const struct amdgpu_dm_mode_size { + char name[DRM_DISPLAY_MODE_LEN]; + int w; + int h; +} common_modes[] = { + { "640x480", 640, 480}, + { "800x600", 800, 600}, + { "1024x768", 1024, 768}, + { "1280x720", 1280, 720}, + { "1280x800", 1280, 800}, + {"1280x1024", 1280, 1024}, + { "1440x900", 1440, 900}, + {"1680x1050", 1680, 1050}, + {"1600x1200", 1600, 1200}, + {"1920x1080", 1920, 1080}, + {"1920x1200", 1920, 1200} +}; + static void amdgpu_dm_connector_add_common_modes(struct drm_encoder *encoder, struct drm_connector *connector) { @@ -8257,23 +8639,6 @@ static void amdgpu_dm_connector_add_common_modes(struct drm_encoder *encoder, to_amdgpu_dm_connector(connector); int i; int n; - struct mode_size { - char name[DRM_DISPLAY_MODE_LEN]; - int w; - int h; - } common_modes[] = { - { "640x480", 640, 480}, - { "800x600", 800, 600}, - { "1024x768", 1024, 768}, - { "1280x720", 1280, 720}, - { "1280x800", 1280, 800}, - {"1280x1024", 1280, 1024}, - { "1440x900", 1440, 900}, - {"1680x1050", 1680, 1050}, - {"1600x1200", 1600, 1200}, - {"1920x1080", 1920, 1080}, - {"1920x1200", 1920, 1200} - }; if ((connector->connector_type != DRM_MODE_CONNECTOR_eDP) && (connector->connector_type != DRM_MODE_CONNECTOR_LVDS)) @@ -8474,6 +8839,16 @@ static void amdgpu_dm_connector_add_freesync_modes(struct drm_connector *connect if (!(amdgpu_freesync_vid_mode && drm_edid)) return; + if (!amdgpu_dm_connector->dc_sink || !amdgpu_dm_connector->dc_link) + return; + + if (!dc_supports_vrr(amdgpu_dm_connector->dc_sink->ctx->dce_version)) + return; + + if (dc_connector_supports_analog(amdgpu_dm_connector->dc_link->link_id.id) && + amdgpu_dm_connector->dc_sink->edid_caps.analog) + return; + if (amdgpu_dm_connector->max_vfreq - amdgpu_dm_connector->min_vfreq > 10) amdgpu_dm_connector->num_modes += add_fs_modes(amdgpu_dm_connector); @@ -8483,11 +8858,11 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector) { struct amdgpu_dm_connector *amdgpu_dm_connector = to_amdgpu_dm_connector(connector); + struct dc_link *dc_link = amdgpu_dm_connector->dc_link; struct drm_encoder *encoder; const struct drm_edid *drm_edid = amdgpu_dm_connector->drm_edid; - struct dc_link_settings *verified_link_cap = - &amdgpu_dm_connector->dc_link->verified_link_cap; - const struct dc *dc = amdgpu_dm_connector->dc_link->dc; + struct dc_link_settings *verified_link_cap = &dc_link->verified_link_cap; + const struct dc *dc = dc_link->dc; encoder = amdgpu_dm_connector_to_encoder(connector); @@ -8497,6 +8872,17 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector) if (dc->link_srv->dp_get_encoding_format(verified_link_cap) == DP_128b_132b_ENCODING) amdgpu_dm_connector->num_modes += drm_add_modes_noedid(connector, 1920, 1080); + + if (amdgpu_dm_connector->dc_sink && + amdgpu_dm_connector->dc_sink->edid_caps.analog && + dc_connector_supports_analog(dc_link->link_id.id)) { + /* Analog monitor connected by DAC load detection. + * Add common modes. It will be up to the user to select one that works. + */ + for (int i = 0; i < ARRAY_SIZE(common_modes); i++) + amdgpu_dm_connector->num_modes += drm_add_modes_noedid( + connector, common_modes[i].w, common_modes[i].h); + } } else { amdgpu_dm_connector_ddc_get_modes(connector, drm_edid); if (encoder) @@ -8544,6 +8930,10 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm, mutex_init(&aconnector->hpd_lock); mutex_init(&aconnector->handle_mst_msg_ready); + aconnector->hdmi_hpd_debounce_delay_ms = AMDGPU_DM_HDMI_HPD_DEBOUNCE_MS; + INIT_DELAYED_WORK(&aconnector->hdmi_hpd_debounce_work, hdmi_hpd_debounce_work); + aconnector->hdmi_prev_sink = NULL; + /* * configure support HPD hot plug connector_>polled default value is 0 * which means HPD hot plug not supported @@ -8565,6 +8955,11 @@ void amdgpu_dm_connector_init_helper(struct amdgpu_display_manager *dm, case DRM_MODE_CONNECTOR_DVID: aconnector->base.polled = DRM_CONNECTOR_POLL_HPD; break; + case DRM_MODE_CONNECTOR_DVII: + case DRM_MODE_CONNECTOR_VGA: + aconnector->base.polled = + DRM_CONNECTOR_POLL_CONNECT | DRM_CONNECTOR_POLL_DISCONNECT; + break; default: break; } @@ -8766,7 +9161,7 @@ static int amdgpu_dm_connector_init(struct amdgpu_display_manager *dm, goto out_free; } - connector_type = to_drm_connector_type(link->connector_signal); + connector_type = to_drm_connector_type(link->connector_signal, link->link_id.id); res = drm_connector_init_with_ddc( dm->ddev, @@ -9969,6 +10364,7 @@ static void amdgpu_dm_commit_streams(struct drm_atomic_state *state, bool mode_set_reset_required = false; u32 i; struct dc_commit_streams_params params = {dc_state->streams, dc_state->stream_count}; + bool set_backlight_level = false; /* Disable writeback */ for_each_old_connector_in_state(state, connector, old_con_state, i) { @@ -10088,6 +10484,7 @@ static void amdgpu_dm_commit_streams(struct drm_atomic_state *state, acrtc->hw_mode = new_crtc_state->mode; crtc->hwmode = new_crtc_state->mode; mode_set_reset_required = true; + set_backlight_level = true; } else if (modereset_required(new_crtc_state)) { drm_dbg_atomic(dev, "Atomic commit: RESET. crtc id %d:[%p]\n", @@ -10144,16 +10541,13 @@ static void amdgpu_dm_commit_streams(struct drm_atomic_state *state, * to fix a flicker issue. * It will cause the dm->actual_brightness is not the current panel brightness * level. (the dm->brightness is the correct panel level) - * So we set the backlight level with dm->brightness value after initial - * set mode. Use restore_backlight flag to avoid setting backlight level - * for every subsequent mode set. + * So we set the backlight level with dm->brightness value after set mode */ - if (dm->restore_backlight) { + if (set_backlight_level) { for (i = 0; i < dm->num_of_edps; i++) { if (dm->backlight_dev[i]) amdgpu_dm_backlight_set_level(dm, i, dm->brightness[i]); } - dm->restore_backlight = false; } } @@ -10523,7 +10917,7 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state) * Here we create an empty update on each plane. * To fix this, DC should permit updating only stream properties. */ - dummy_updates = kzalloc(sizeof(struct dc_surface_update) * MAX_SURFACES, GFP_ATOMIC); + dummy_updates = kzalloc(sizeof(struct dc_surface_update) * MAX_SURFACES, GFP_KERNEL); if (!dummy_updates) { drm_err(adev_to_drm(adev), "Failed to allocate memory for dummy_updates.\n"); continue; @@ -12450,7 +12844,7 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, int j = state->num_private_objs-1; dm_atomic_destroy_state(obj, - state->private_objs[i].state); + state->private_objs[i].state_to_destroy); /* If i is not at the end of the array then the * last element needs to be moved to where i was @@ -12461,7 +12855,7 @@ static int amdgpu_dm_atomic_check(struct drm_device *dev, state->private_objs[j]; state->private_objs[j].ptr = NULL; - state->private_objs[j].state = NULL; + state->private_objs[j].state_to_destroy = NULL; state->private_objs[j].old_state = NULL; state->private_objs[j].new_state = NULL; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index 009f206226f0..ef97cede9926 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -59,6 +59,7 @@ #define AMDGPU_HDR_MULT_DEFAULT (0x100000000LL) +#define AMDGPU_DM_HDMI_HPD_DEBOUNCE_MS 1500 /* #include "include/amdgpu_dal_power_if.h" #include "amdgpu_dm_irq.h" @@ -631,13 +632,6 @@ struct amdgpu_display_manager { u32 actual_brightness[AMDGPU_DM_MAX_NUM_EDP]; /** - * @restore_backlight: - * - * Flag to indicate whether to restore backlight after modeset. - */ - bool restore_backlight; - - /** * @aux_hpd_discon_quirk: * * quirk for hpd discon while aux is on-going. @@ -826,6 +820,11 @@ struct amdgpu_dm_connector { bool pack_sdp_v1_3; enum adaptive_sync_type as_type; struct amdgpu_hdmi_vsdb_info vsdb_info; + + /* HDMI HPD debounce support */ + unsigned int hdmi_hpd_debounce_delay_ms; + struct delayed_work hdmi_hpd_debounce_work; + struct dc_sink *hdmi_prev_sink; }; static inline void amdgpu_dm_set_mst_status(uint8_t *status, @@ -1000,6 +999,7 @@ struct dm_connector_state { bool underscan_enable; bool freesync_capable; bool update_hdcp; + bool abm_sysfs_forbidden; uint8_t abm_level; int vcpi_slots; uint64_t pbn; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c index a4ac6d442278..1dcc79b35225 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_color.c @@ -26,12 +26,39 @@ #include "amdgpu.h" #include "amdgpu_mode.h" #include "amdgpu_dm.h" +#include "amdgpu_dm_colorop.h" #include "dc.h" #include "modules/color/color_gamma.h" /** * DOC: overview * + * We have three types of color management in the AMD display driver. + * 1. the legacy &drm_crtc DEGAMMA, CTM, and GAMMA properties + * 2. AMD driver private color management on &drm_plane and &drm_crtc + * 3. AMD plane color pipeline + * + * The CRTC properties are the original color management. When they were + * implemented per-plane color management was not a thing yet. Because + * of that we could get away with plumbing the DEGAMMA and CTM + * properties to pre-blending HW functions. This is incompatible with + * per-plane color management, such as via the AMD private properties or + * the new drm_plane color pipeline. The only compatible CRTC property + * with per-plane color management is the GAMMA property as it is + * applied post-blending. + * + * The AMD driver private color management properties are only exposed + * when the kernel is built explicitly with -DAMD_PRIVATE_COLOR. They + * are temporary building blocks on the path to full-fledged &drm_plane + * and &drm_crtc color pipelines and lay the driver's groundwork for the + * color pipelines. + * + * The AMD plane color pipeline describes AMD's &drm_colorops via the + * &drm_plane's COLOR_PIPELINE property. + * + * drm_crtc Properties + * ------------------- + * * The DC interface to HW gives us the following color management blocks * per pipe (surface): * @@ -42,36 +69,93 @@ * - Surface regamma LUT (normalized) * - Output CSC (normalized) * - * But these aren't a direct mapping to DRM color properties. The current DRM - * interface exposes CRTC degamma, CRTC CTM and CRTC regamma while our hardware - * is essentially giving: + * But these aren't a direct mapping to DRM color properties. The + * current DRM interface exposes CRTC degamma, CRTC CTM and CRTC regamma + * while our hardware is essentially giving: * * Plane CTM -> Plane degamma -> Plane CTM -> Plane regamma -> Plane CTM * - * The input gamma LUT block isn't really applicable here since it operates - * on the actual input data itself rather than the HW fp representation. The - * input and output CSC blocks are technically available to use as part of - * the DC interface but are typically used internally by DC for conversions - * between color spaces. These could be blended together with user - * adjustments in the future but for now these should remain untouched. + * The input gamma LUT block isn't really applicable here since it + * operates on the actual input data itself rather than the HW fp + * representation. The input and output CSC blocks are technically + * available to use as part of the DC interface but are typically used + * internally by DC for conversions between color spaces. These could be + * blended together with user adjustments in the future but for now + * these should remain untouched. + * + * The pipe blending also happens after these blocks so we don't + * actually support any CRTC props with correct blending with multiple + * planes - but we can still support CRTC color management properties in + * DM in most single plane cases correctly with clever management of the + * DC interface in DM. + * + * As per DRM documentation, blocks should be in hardware bypass when + * their respective property is set to NULL. A linear DGM/RGM LUT should + * also considered as putting the respective block into bypass mode. + * + * This means that the following configuration is assumed to be the + * default: + * + * Plane DGM Bypass -> Plane CTM Bypass -> Plane RGM Bypass -> ... CRTC + * DGM Bypass -> CRTC CTM Bypass -> CRTC RGM Bypass + * + * AMD Private Color Management on drm_plane + * ----------------------------------------- + * + * The AMD private color management properties on a &drm_plane are: * - * The pipe blending also happens after these blocks so we don't actually - * support any CRTC props with correct blending with multiple planes - but we - * can still support CRTC color management properties in DM in most single - * plane cases correctly with clever management of the DC interface in DM. + * - AMD_PLANE_DEGAMMA_LUT + * - AMD_PLANE_DEGAMMA_LUT_SIZE + * - AMD_PLANE_DEGAMMA_TF + * - AMD_PLANE_HDR_MULT + * - AMD_PLANE_CTM + * - AMD_PLANE_SHAPER_LUT + * - AMD_PLANE_SHAPER_LUT_SIZE + * - AMD_PLANE_SHAPER_TF + * - AMD_PLANE_LUT3D + * - AMD_PLANE_LUT3D_SIZE + * - AMD_PLANE_BLEND_LUT + * - AMD_PLANE_BLEND_LUT_SIZE + * - AMD_PLANE_BLEND_TF * - * As per DRM documentation, blocks should be in hardware bypass when their - * respective property is set to NULL. A linear DGM/RGM LUT should also - * considered as putting the respective block into bypass mode. + * The AMD private color management property on a &drm_crtc is: * - * This means that the following - * configuration is assumed to be the default: + * - AMD_CRTC_REGAMMA_TF + * + * Use of these properties is discouraged. + * + * AMD plane color pipeline + * ------------------------ + * + * The AMD &drm_plane color pipeline is advertised for DCN generations + * 3.0 and newer. It exposes these elements in this order: + * + * 1. 1D curve colorop + * 2. Multiplier + * 3. 3x4 CTM + * 4. 1D curve colorop + * 5. 1D LUT + * 6. 3D LUT + * 7. 1D curve colorop + * 8. 1D LUT + * + * The multiplier (#2) is a simple multiplier that is applied to all + * channels. + * + * The 3x4 CTM (#3) is a simple 3x4 matrix. + * + * #1, and #7 are non-linear to linear curves. #4 is a linear to + * non-linear curve. They support sRGB, PQ, and BT.709/BT.2020 EOTFs or + * their inverse. + * + * The 1D LUTs (#5 and #8) are plain 4096 entry LUTs. + * + * The 3DLUT (#6) is a tetrahedrally interpolated 17 cube LUT. * - * Plane DGM Bypass -> Plane CTM Bypass -> Plane RGM Bypass -> ... - * CRTC DGM Bypass -> CRTC CTM Bypass -> CRTC RGM Bypass */ #define MAX_DRM_LUT_VALUE 0xFFFF +#define MAX_DRM_LUT32_VALUE 0xFFFFFFFF #define SDR_WHITE_LEVEL_INIT_VALUE 80 /** @@ -342,6 +426,21 @@ __extract_blob_lut(const struct drm_property_blob *blob, uint32_t *size) } /** + * __extract_blob_lut32 - Extracts the DRM lut and lut size from a blob. + * @blob: DRM color mgmt property blob + * @size: lut size + * + * Returns: + * DRM LUT or NULL + */ +static const struct drm_color_lut32 * +__extract_blob_lut32(const struct drm_property_blob *blob, uint32_t *size) +{ + *size = blob ? drm_color_lut32_size(blob) : 0; + return blob ? (struct drm_color_lut32 *)blob->data : NULL; +} + +/** * __is_lut_linear - check if the given lut is a linear mapping of values * @lut: given lut to check values * @size: lut size @@ -415,6 +514,24 @@ static void __drm_lut_to_dc_gamma(const struct drm_color_lut *lut, } /** + * __drm_lut32_to_dc_gamma - convert the drm_color_lut to dc_gamma. + * @lut: DRM lookup table for color conversion + * @gamma: DC gamma to set entries + * + * The conversion depends on the size of the lut - whether or not it's legacy. + */ +static void __drm_lut32_to_dc_gamma(const struct drm_color_lut32 *lut, struct dc_gamma *gamma) +{ + int i; + + for (i = 0; i < MAX_COLOR_LUT_ENTRIES; i++) { + gamma->entries.red[i] = dc_fixpt_from_fraction(lut[i].red, MAX_DRM_LUT32_VALUE); + gamma->entries.green[i] = dc_fixpt_from_fraction(lut[i].green, MAX_DRM_LUT32_VALUE); + gamma->entries.blue[i] = dc_fixpt_from_fraction(lut[i].blue, MAX_DRM_LUT32_VALUE); + } +} + +/** * __drm_ctm_to_dc_matrix - converts a DRM CTM to a DC CSC float matrix * @ctm: DRM color transformation matrix * @matrix: DC CSC float matrix @@ -566,6 +683,63 @@ static int __set_output_tf(struct dc_transfer_func *func, return res ? 0 : -ENOMEM; } +/** + * __set_output_tf_32 - calculates the output transfer function based on expected input space. + * @func: transfer function + * @lut: lookup table that defines the color space + * @lut_size: size of respective lut + * @has_rom: if ROM can be used for hardcoded curve + * + * Returns: + * 0 in case of success. -ENOMEM if fails. + */ +static int __set_output_tf_32(struct dc_transfer_func *func, + const struct drm_color_lut32 *lut, uint32_t lut_size, + bool has_rom) +{ + struct dc_gamma *gamma = NULL; + struct calculate_buffer cal_buffer = {0}; + bool res; + + cal_buffer.buffer_index = -1; + + if (lut_size) { + gamma = dc_create_gamma(); + if (!gamma) + return -ENOMEM; + + gamma->num_entries = lut_size; + __drm_lut32_to_dc_gamma(lut, gamma); + } + + if (func->tf == TRANSFER_FUNCTION_LINEAR) { + /* + * Color module doesn't like calculating regamma params + * on top of a linear input. But degamma params can be used + * instead to simulate this. + */ + if (gamma) + gamma->type = GAMMA_CUSTOM; + res = mod_color_calculate_degamma_params(NULL, func, + gamma, gamma != NULL); + } else { + /* + * Assume sRGB. The actual mapping will depend on whether the + * input was legacy or not. + */ + if (gamma) + gamma->type = GAMMA_CS_TFM_1D; + res = mod_color_calculate_regamma_params(func, gamma, gamma != NULL, + has_rom, NULL, &cal_buffer); + } + + if (gamma) + dc_gamma_release(&gamma); + + return res ? 0 : -ENOMEM; +} + + static int amdgpu_dm_set_atomic_regamma(struct dc_transfer_func *out_tf, const struct drm_color_lut *regamma_lut, uint32_t regamma_size, bool has_rom, @@ -638,6 +812,42 @@ static int __set_input_tf(struct dc_color_caps *caps, struct dc_transfer_func *f return res ? 0 : -ENOMEM; } +/** + * __set_input_tf_32 - calculates the input transfer function based on expected + * input space. + * @caps: dc color capabilities + * @func: transfer function + * @lut: lookup table that defines the color space + * @lut_size: size of respective lut. + * + * Returns: + * 0 in case of success. -ENOMEM if fails. + */ +static int __set_input_tf_32(struct dc_color_caps *caps, struct dc_transfer_func *func, + const struct drm_color_lut32 *lut, uint32_t lut_size) +{ + struct dc_gamma *gamma = NULL; + bool res; + + if (lut_size) { + gamma = dc_create_gamma(); + if (!gamma) + return -ENOMEM; + + gamma->type = GAMMA_CUSTOM; + gamma->num_entries = lut_size; + + __drm_lut32_to_dc_gamma(lut, gamma); + } + + res = mod_color_calculate_degamma_params(caps, func, gamma, gamma != NULL); + + if (gamma) + dc_gamma_release(&gamma); + + return res ? 0 : -ENOMEM; +} + static enum dc_transfer_func_predefined amdgpu_tf_to_dc_tf(enum amdgpu_transfer_function tf) { @@ -667,6 +877,27 @@ amdgpu_tf_to_dc_tf(enum amdgpu_transfer_function tf) } } +static enum dc_transfer_func_predefined +amdgpu_colorop_tf_to_dc_tf(enum drm_colorop_curve_1d_type tf) +{ + switch (tf) { + case DRM_COLOROP_1D_CURVE_SRGB_EOTF: + case DRM_COLOROP_1D_CURVE_SRGB_INV_EOTF: + return TRANSFER_FUNCTION_SRGB; + case DRM_COLOROP_1D_CURVE_PQ_125_EOTF: + case DRM_COLOROP_1D_CURVE_PQ_125_INV_EOTF: + return TRANSFER_FUNCTION_PQ; + case DRM_COLOROP_1D_CURVE_BT2020_INV_OETF: + case DRM_COLOROP_1D_CURVE_BT2020_OETF: + return TRANSFER_FUNCTION_BT709; + case DRM_COLOROP_1D_CURVE_GAMMA22: + case DRM_COLOROP_1D_CURVE_GAMMA22_INV: + return TRANSFER_FUNCTION_GAMMA22; + default: + return TRANSFER_FUNCTION_LINEAR; + } +} + static void __to_dc_lut3d_color(struct dc_rgb *rgb, const struct drm_color_lut lut, int bit_precision) @@ -720,6 +951,59 @@ static void __drm_3dlut_to_dc_3dlut(const struct drm_color_lut *lut, __to_dc_lut3d_color(&lut0[lut_i], lut[i], bit_depth); } +static void __to_dc_lut3d_32_color(struct dc_rgb *rgb, + const struct drm_color_lut32 lut, + int bit_precision) +{ + rgb->red = drm_color_lut32_extract(lut.red, bit_precision); + rgb->green = drm_color_lut32_extract(lut.green, bit_precision); + rgb->blue = drm_color_lut32_extract(lut.blue, bit_precision); +} + +static void __drm_3dlut32_to_dc_3dlut(const struct drm_color_lut32 *lut, + uint32_t lut3d_size, + struct tetrahedral_params *params, + bool use_tetrahedral_9, + int bit_depth) +{ + struct dc_rgb *lut0; + struct dc_rgb *lut1; + struct dc_rgb *lut2; + struct dc_rgb *lut3; + int lut_i, i; + + + if (use_tetrahedral_9) { + lut0 = params->tetrahedral_9.lut0; + lut1 = params->tetrahedral_9.lut1; + lut2 = params->tetrahedral_9.lut2; + lut3 = params->tetrahedral_9.lut3; + } else { + lut0 = params->tetrahedral_17.lut0; + lut1 = params->tetrahedral_17.lut1; + lut2 = params->tetrahedral_17.lut2; + lut3 = params->tetrahedral_17.lut3; + } + + for (lut_i = 0, i = 0; i < lut3d_size - 4; lut_i++, i += 4) { + /* + * We should consider the 3D LUT RGB values are distributed + * along four arrays lut0-3 where the first sizes 1229 and the + * other 1228. The bit depth supported for 3dlut channel is + * 12-bit, but DC also supports 10-bit. + * + * TODO: improve color pipeline API to enable the userspace set + * bit depth and 3D LUT size/stride, as specified by VA-API. + */ + __to_dc_lut3d_32_color(&lut0[lut_i], lut[i], bit_depth); + __to_dc_lut3d_32_color(&lut1[lut_i], lut[i + 1], bit_depth); + __to_dc_lut3d_32_color(&lut2[lut_i], lut[i + 2], bit_depth); + __to_dc_lut3d_32_color(&lut3[lut_i], lut[i + 3], bit_depth); + } + /* lut0 has 1229 points (lut_size/4 + 1) */ + __to_dc_lut3d_32_color(&lut0[lut_i], lut[i], bit_depth); +} + /* amdgpu_dm_atomic_lut3d - set DRM 3D LUT to DC stream * @drm_lut3d: user 3D LUT * @drm_lut3d_size: size of 3D LUT @@ -1178,6 +1462,360 @@ __set_dm_plane_degamma(struct drm_plane_state *plane_state, } static int +__set_colorop_in_tf_1d_curve(struct dc_plane_state *dc_plane_state, + struct drm_colorop_state *colorop_state) +{ + struct dc_transfer_func *tf = &dc_plane_state->in_transfer_func; + struct drm_colorop *colorop = colorop_state->colorop; + struct drm_device *drm = colorop->dev; + + if (colorop->type != DRM_COLOROP_1D_CURVE) + return -EINVAL; + + if (!(BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_degam_tfs)) + return -EINVAL; + + if (colorop_state->bypass) { + tf->type = TF_TYPE_BYPASS; + tf->tf = TRANSFER_FUNCTION_LINEAR; + return 0; + } + + drm_dbg(drm, "Degamma colorop with ID: %d\n", colorop->base.id); + + tf->type = TF_TYPE_PREDEFINED; + tf->tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type); + + return 0; +} + +static int +__set_dm_plane_colorop_degamma(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct drm_atomic_state *state = plane_state->state; + int i = 0; + + old_colorop = colorop; + + /* 1st op: 1d curve - degamma */ + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + (BIT(new_colorop_state->curve_1d_type) & amdgpu_dm_supported_degam_tfs)) { + colorop_state = new_colorop_state; + break; + } + } + + if (!colorop_state) + return -EINVAL; + + return __set_colorop_in_tf_1d_curve(dc_plane_state, colorop_state); +} + +static int +__set_dm_plane_colorop_3x4_matrix(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct drm_atomic_state *state = plane_state->state; + const struct drm_device *dev = colorop->dev; + const struct drm_property_blob *blob; + struct drm_color_ctm_3x4 *ctm = NULL; + int i = 0; + + /* 3x4 matrix */ + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + new_colorop_state->colorop->type == DRM_COLOROP_CTM_3X4) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_CTM_3X4) { + drm_dbg(dev, "3x4 matrix colorop with ID: %d\n", colorop->base.id); + blob = colorop_state->data; + if (blob->length == sizeof(struct drm_color_ctm_3x4)) { + ctm = (struct drm_color_ctm_3x4 *) blob->data; + __drm_ctm_3x4_to_dc_matrix(ctm, dc_plane_state->gamut_remap_matrix.matrix); + dc_plane_state->gamut_remap_matrix.enable_remap = true; + dc_plane_state->input_csc_color_matrix.enable_adjustment = false; + } else { + drm_warn(dev, "blob->length (%zu) isn't equal to drm_color_ctm_3x4 (%zu)\n", + blob->length, sizeof(struct drm_color_ctm_3x4)); + return -EINVAL; + } + } + + return 0; +} + +static int +__set_dm_plane_colorop_multiplier(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct drm_atomic_state *state = plane_state->state; + const struct drm_device *dev = colorop->dev; + int i = 0; + + /* Multiplier */ + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + new_colorop_state->colorop->type == DRM_COLOROP_MULTIPLIER) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_MULTIPLIER) { + drm_dbg(dev, "Multiplier colorop with ID: %d\n", colorop->base.id); + dc_plane_state->hdr_mult = amdgpu_dm_fixpt_from_s3132(colorop_state->multiplier); + } + + return 0; +} + +static int +__set_dm_plane_colorop_shaper(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct drm_atomic_state *state = plane_state->state; + enum dc_transfer_func_predefined default_tf = TRANSFER_FUNCTION_LINEAR; + struct dc_transfer_func *tf = &dc_plane_state->in_shaper_func; + const struct drm_color_lut32 *shaper_lut; + struct drm_device *dev = colorop->dev; + bool enabled = false; + u32 shaper_size; + int i = 0, ret = 0; + + /* 1D Curve - SHAPER TF */ + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + (BIT(new_colorop_state->curve_1d_type) & amdgpu_dm_supported_shaper_tfs)) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_CURVE) { + drm_dbg(dev, "Shaper TF colorop with ID: %d\n", colorop->base.id); + tf->type = TF_TYPE_DISTRIBUTED_POINTS; + tf->tf = default_tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type); + tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE; + ret = __set_output_tf(tf, 0, 0, false); + if (ret) + return ret; + enabled = true; + } + + /* 1D LUT - SHAPER LUT */ + colorop = old_colorop->next; + if (!colorop) { + drm_dbg(dev, "no Shaper LUT colorop found\n"); + return -EINVAL; + } + + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + new_colorop_state->colorop->type == DRM_COLOROP_1D_LUT) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_LUT) { + drm_dbg(dev, "Shaper LUT colorop with ID: %d\n", colorop->base.id); + tf->type = TF_TYPE_DISTRIBUTED_POINTS; + tf->tf = default_tf; + tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE; + shaper_lut = __extract_blob_lut32(colorop_state->data, &shaper_size); + shaper_size = shaper_lut != NULL ? shaper_size : 0; + + /* Custom LUT size must be the same as supported size */ + if (shaper_size == colorop->size) { + ret = __set_output_tf_32(tf, shaper_lut, shaper_size, false); + if (ret) + return ret; + enabled = true; + } + } + + if (!enabled) + tf->type = TF_TYPE_BYPASS; + + return 0; +} + +/* __set_colorop_3dlut - set DRM 3D LUT to DC stream + * @drm_lut3d: user 3D LUT + * @drm_lut3d_size: size of 3D LUT + * @lut3d: DC 3D LUT + * + * Map user 3D LUT data to DC 3D LUT and all necessary bits to program it + * on DCN accordingly. + * + * Returns: + * 0 on success. -EINVAL if drm_lut3d_size is zero. + */ +static int __set_colorop_3dlut(const struct drm_color_lut32 *drm_lut3d, + uint32_t drm_lut3d_size, + struct dc_3dlut *lut) +{ + if (!drm_lut3d_size) { + lut->state.bits.initialized = 0; + return -EINVAL; + } + + /* Only supports 17x17x17 3D LUT (12-bit) now */ + lut->lut_3d.use_12bits = true; + lut->lut_3d.use_tetrahedral_9 = false; + + lut->state.bits.initialized = 1; + __drm_3dlut32_to_dc_3dlut(drm_lut3d, drm_lut3d_size, &lut->lut_3d, + lut->lut_3d.use_tetrahedral_9, 12); + + return 0; +} + +static int +__set_dm_plane_colorop_3dlut(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct dc_transfer_func *tf = &dc_plane_state->in_shaper_func; + struct drm_atomic_state *state = plane_state->state; + const struct amdgpu_device *adev = drm_to_adev(colorop->dev); + const struct drm_device *dev = colorop->dev; + const struct drm_color_lut32 *lut3d; + uint32_t lut3d_size; + int i = 0, ret = 0; + + /* 3D LUT */ + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + new_colorop_state->colorop->type == DRM_COLOROP_3D_LUT) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_3D_LUT) { + if (!adev->dm.dc->caps.color.dpp.hw_3d_lut) { + drm_dbg(dev, "3D LUT is not supported by hardware\n"); + return -EINVAL; + } + + drm_dbg(dev, "3D LUT colorop with ID: %d\n", colorop->base.id); + lut3d = __extract_blob_lut32(colorop_state->data, &lut3d_size); + lut3d_size = lut3d != NULL ? lut3d_size : 0; + ret = __set_colorop_3dlut(lut3d, lut3d_size, &dc_plane_state->lut3d_func); + if (ret) { + drm_dbg(dev, "3D LUT colorop with ID: %d has LUT size = %d\n", + colorop->base.id, lut3d_size); + return ret; + } + + /* 3D LUT requires shaper. If shaper colorop is bypassed, enable shaper curve + * with TRANSFER_FUNCTION_LINEAR + */ + if (tf->type == TF_TYPE_BYPASS) { + tf->type = TF_TYPE_DISTRIBUTED_POINTS; + tf->tf = TRANSFER_FUNCTION_LINEAR; + tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE; + ret = __set_output_tf_32(tf, NULL, 0, false); + } + } + + return ret; +} + +static int +__set_dm_plane_colorop_blend(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state, + struct drm_colorop *colorop) +{ + struct drm_colorop *old_colorop; + struct drm_colorop_state *colorop_state = NULL, *new_colorop_state; + struct drm_atomic_state *state = plane_state->state; + enum dc_transfer_func_predefined default_tf = TRANSFER_FUNCTION_LINEAR; + struct dc_transfer_func *tf = &dc_plane_state->blend_tf; + const struct drm_color_lut32 *blend_lut = NULL; + struct drm_device *dev = colorop->dev; + uint32_t blend_size = 0; + int i = 0; + + /* 1D Curve - BLND TF */ + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + (BIT(new_colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_CURVE && + (BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) { + drm_dbg(dev, "Blend TF colorop with ID: %d\n", colorop->base.id); + tf->type = TF_TYPE_DISTRIBUTED_POINTS; + tf->tf = default_tf = amdgpu_colorop_tf_to_dc_tf(colorop_state->curve_1d_type); + tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE; + __set_input_tf_32(NULL, tf, blend_lut, blend_size); + } + + /* 1D Curve - BLND LUT */ + colorop = old_colorop->next; + if (!colorop) { + drm_dbg(dev, "no Blend LUT colorop found\n"); + return -EINVAL; + } + + old_colorop = colorop; + for_each_new_colorop_in_state(state, colorop, new_colorop_state, i) { + if (new_colorop_state->colorop == old_colorop && + new_colorop_state->colorop->type == DRM_COLOROP_1D_LUT) { + colorop_state = new_colorop_state; + break; + } + } + + if (colorop_state && !colorop_state->bypass && colorop->type == DRM_COLOROP_1D_LUT && + (BIT(colorop_state->curve_1d_type) & amdgpu_dm_supported_blnd_tfs)) { + drm_dbg(dev, "Blend LUT colorop with ID: %d\n", colorop->base.id); + tf->type = TF_TYPE_DISTRIBUTED_POINTS; + tf->tf = default_tf; + tf->sdr_ref_white_level = SDR_WHITE_LEVEL_INIT_VALUE; + blend_lut = __extract_blob_lut32(colorop_state->data, &blend_size); + blend_size = blend_lut != NULL ? blend_size : 0; + + /* Custom LUT size must be the same as supported size */ + if (blend_size == colorop->size) + __set_input_tf_32(NULL, tf, blend_lut, blend_size); + } + + return 0; +} + +static int amdgpu_dm_plane_set_color_properties(struct drm_plane_state *plane_state, struct dc_plane_state *dc_plane_state) { @@ -1227,6 +1865,93 @@ amdgpu_dm_plane_set_color_properties(struct drm_plane_state *plane_state, return 0; } +static int +amdgpu_dm_plane_set_colorop_properties(struct drm_plane_state *plane_state, + struct dc_plane_state *dc_plane_state) +{ + struct drm_colorop *colorop = plane_state->color_pipeline; + struct drm_device *dev = plane_state->plane->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + int ret; + + /* 1D Curve - DEGAM TF */ + if (!colorop) + return -EINVAL; + + ret = __set_dm_plane_colorop_degamma(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + + /* Multiplier */ + colorop = colorop->next; + if (!colorop) { + drm_dbg(dev, "no multiplier colorop found\n"); + return -EINVAL; + } + + ret = __set_dm_plane_colorop_multiplier(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + + /* 3x4 matrix */ + colorop = colorop->next; + if (!colorop) { + drm_dbg(dev, "no 3x4 matrix colorop found\n"); + return -EINVAL; + } + + ret = __set_dm_plane_colorop_3x4_matrix(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + + if (adev->dm.dc->caps.color.dpp.hw_3d_lut) { + /* 1D Curve & LUT - SHAPER TF & LUT */ + colorop = colorop->next; + if (!colorop) { + drm_dbg(dev, "no Shaper TF colorop found\n"); + return -EINVAL; + } + + ret = __set_dm_plane_colorop_shaper(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + + /* Shaper LUT colorop is already handled, just skip here */ + colorop = colorop->next; + if (!colorop) + return -EINVAL; + + /* 3D LUT */ + colorop = colorop->next; + if (!colorop) { + drm_dbg(dev, "no 3D LUT colorop found\n"); + return -EINVAL; + } + + ret = __set_dm_plane_colorop_3dlut(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + } + + /* 1D Curve & LUT - BLND TF & LUT */ + colorop = colorop->next; + if (!colorop) { + drm_dbg(dev, "no Blend TF colorop found\n"); + return -EINVAL; + } + + ret = __set_dm_plane_colorop_blend(plane_state, dc_plane_state, colorop); + if (ret) + return ret; + + /* BLND LUT colorop is already handled, just skip here */ + colorop = colorop->next; + if (!colorop) + return -EINVAL; + + return 0; +} + /** * amdgpu_dm_update_plane_color_mgmt: Maps DRM color management to DC plane. * @crtc: amdgpu_dm crtc state @@ -1323,5 +2048,8 @@ int amdgpu_dm_update_plane_color_mgmt(struct dm_crtc_state *crtc, dc_plane_state->input_csc_color_matrix.enable_adjustment = false; } + if (!amdgpu_dm_plane_set_colorop_properties(plane_state, dc_plane_state)) + return 0; + return amdgpu_dm_plane_set_color_properties(plane_state, dc_plane_state); } diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c new file mode 100644 index 000000000000..d585618b8064 --- /dev/null +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.c @@ -0,0 +1,209 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#include <drm/drm_print.h> +#include <drm/drm_plane.h> +#include <drm/drm_property.h> +#include <drm/drm_colorop.h> + +#include "amdgpu.h" +#include "amdgpu_dm_colorop.h" +#include "dc.h" + +const u64 amdgpu_dm_supported_degam_tfs = + BIT(DRM_COLOROP_1D_CURVE_SRGB_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_PQ_125_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_BT2020_INV_OETF) | + BIT(DRM_COLOROP_1D_CURVE_GAMMA22_INV); + +const u64 amdgpu_dm_supported_shaper_tfs = + BIT(DRM_COLOROP_1D_CURVE_SRGB_INV_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_PQ_125_INV_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_BT2020_OETF) | + BIT(DRM_COLOROP_1D_CURVE_GAMMA22); + +const u64 amdgpu_dm_supported_blnd_tfs = + BIT(DRM_COLOROP_1D_CURVE_SRGB_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_PQ_125_EOTF) | + BIT(DRM_COLOROP_1D_CURVE_BT2020_INV_OETF) | + BIT(DRM_COLOROP_1D_CURVE_GAMMA22_INV); + +#define MAX_COLOR_PIPELINE_OPS 10 + +#define LUT3D_SIZE 17 + +int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_prop_enum_list *list) +{ + struct drm_colorop *ops[MAX_COLOR_PIPELINE_OPS]; + struct drm_device *dev = plane->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + int ret; + int i = 0; + + memset(ops, 0, sizeof(ops)); + + /* 1D curve - DEGAM TF */ + ops[i] = kzalloc(sizeof(*ops[0]), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_curve_1d_init(dev, ops[i], plane, + amdgpu_dm_supported_degam_tfs, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + list->type = ops[i]->base.id; + list->name = kasprintf(GFP_KERNEL, "Color Pipeline %d", ops[i]->base.id); + + i++; + + /* Multiplier */ + ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_mult_init(dev, ops[i], plane, DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + + i++; + + /* 3x4 matrix */ + ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_ctm_3x4_init(dev, ops[i], plane, DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + + i++; + + if (adev->dm.dc->caps.color.dpp.hw_3d_lut) { + /* 1D curve - SHAPER TF */ + ops[i] = kzalloc(sizeof(*ops[0]), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_curve_1d_init(dev, ops[i], plane, + amdgpu_dm_supported_shaper_tfs, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + + i++; + + /* 1D LUT - SHAPER LUT */ + ops[i] = kzalloc(sizeof(*ops[0]), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_curve_1d_lut_init(dev, ops[i], plane, MAX_COLOR_LUT_ENTRIES, + DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + + i++; + + /* 3D LUT */ + ops[i] = kzalloc(sizeof(*ops[0]), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_3dlut_init(dev, ops[i], plane, LUT3D_SIZE, + DRM_COLOROP_LUT3D_INTERPOLATION_TETRAHEDRAL, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + + i++; + } + + /* 1D curve - BLND TF */ + ops[i] = kzalloc(sizeof(*ops[0]), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_curve_1d_init(dev, ops[i], plane, + amdgpu_dm_supported_blnd_tfs, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i - 1], ops[i]); + + i++; + + /* 1D LUT - BLND LUT */ + ops[i] = kzalloc(sizeof(struct drm_colorop), GFP_KERNEL); + if (!ops[i]) { + ret = -ENOMEM; + goto cleanup; + } + + ret = drm_plane_colorop_curve_1d_lut_init(dev, ops[i], plane, MAX_COLOR_LUT_ENTRIES, + DRM_COLOROP_LUT1D_INTERPOLATION_LINEAR, + DRM_COLOROP_FLAG_ALLOW_BYPASS); + if (ret) + goto cleanup; + + drm_colorop_set_next_property(ops[i-1], ops[i]); + return 0; + +cleanup: + if (ret == -ENOMEM) + drm_err(plane->dev, "KMS: Failed to allocate colorop\n"); + + drm_colorop_pipeline_destroy(dev); + + return ret; +} diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.h new file mode 100644 index 000000000000..2e1617ffc8ee --- /dev/null +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_colorop.h @@ -0,0 +1,36 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2023 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + * Authors: AMD + * + */ + +#ifndef __AMDGPU_DM_COLOROP_H__ +#define __AMDGPU_DM_COLOROP_H__ + +extern const u64 amdgpu_dm_supported_degam_tfs; +extern const u64 amdgpu_dm_supported_shaper_tfs; +extern const u64 amdgpu_dm_supported_blnd_tfs; + +int amdgpu_dm_initialize_default_pipeline(struct drm_plane *plane, struct drm_prop_enum_list *list); + +#endif /* __AMDGPU_DM_COLOROP_H__*/ diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index 1ec9d03ad747..697e232acebf 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -248,6 +248,8 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) struct vblank_control_work *vblank_work = container_of(work, struct vblank_control_work, work); struct amdgpu_display_manager *dm = vblank_work->dm; + struct amdgpu_device *adev = drm_to_adev(dm->ddev); + int r; mutex_lock(&dm->dc_lock); @@ -277,7 +279,16 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) if (dm->active_vblank_irq_count == 0) { dc_post_update_surfaces_to_stream(dm->dc); + + r = amdgpu_dpm_pause_power_profile(adev, true); + if (r) + dev_warn(adev->dev, "failed to set default power profile mode\n"); + dc_allow_idle_optimizations(dm->dc, true); + + r = amdgpu_dpm_pause_power_profile(adev, false); + if (r) + dev_warn(adev->dev, "failed to restore the power profile mode\n"); } mutex_unlock(&dm->dc_lock); @@ -297,8 +308,12 @@ static inline int amdgpu_dm_crtc_set_vblank(struct drm_crtc *crtc, bool enable) int irq_type; int rc = 0; - if (acrtc->otg_inst == -1) - goto skip; + if (enable && !acrtc->base.enabled) { + drm_dbg_vbl(crtc->dev, + "Reject vblank enable on unconfigured CRTC %d (enabled=%d)\n", + acrtc->crtc_id, acrtc->base.enabled); + return -EINVAL; + } irq_type = amdgpu_display_crtc_idx_to_irq_type(adev, acrtc->crtc_id); @@ -383,7 +398,7 @@ static inline int amdgpu_dm_crtc_set_vblank(struct drm_crtc *crtc, bool enable) return rc; } #endif -skip: + if (amdgpu_in_reset(adev)) return 0; @@ -721,7 +736,7 @@ int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, { struct amdgpu_crtc *acrtc = NULL; struct drm_plane *cursor_plane; - bool is_dcn; + bool has_degamma; int res = -ENOMEM; cursor_plane = kzalloc(sizeof(*cursor_plane), GFP_KERNEL); @@ -760,20 +775,18 @@ int amdgpu_dm_crtc_init(struct amdgpu_display_manager *dm, dm->adev->mode_info.crtcs[crtc_index] = acrtc; - /* Don't enable DRM CRTC degamma property for DCE since it doesn't - * support programmable degamma anywhere. - */ - is_dcn = dm->adev->dm.dc->caps.color.dpp.dcn_arch; - /* Dont't enable DRM CRTC degamma property for DCN401 since the - * pre-blending degamma LUT doesn't apply to cursor, and therefore - * can't work similar to a post-blending degamma LUT as in other hw - * versions. - * TODO: revisit it once KMS plane color API is merged. + /* Don't enable DRM CRTC degamma property for + * 1. Degamma is replaced by color pipeline. + * 2. DCE since it doesn't support programmable degamma anywhere. + * 3. DCN401 since pre-blending degamma LUT doesn't apply to cursor. */ - drm_crtc_enable_color_mgmt(&acrtc->base, - (is_dcn && - dm->adev->dm.dc->ctx->dce_version != DCN_VERSION_4_01) ? - MAX_COLOR_LUT_ENTRIES : 0, + if (plane->color_pipeline_property) + has_degamma = false; + else + has_degamma = dm->adev->dm.dc->caps.color.dpp.dcn_arch && + dm->adev->dm.dc->ctx->dce_version != DCN_VERSION_4_01; + + drm_crtc_enable_color_mgmt(&acrtc->base, has_degamma ? MAX_COLOR_LUT_ENTRIES : 0, true, MAX_COLOR_LUT_ENTRIES); drm_mode_crtc_set_gamma_size(&acrtc->base, MAX_COLOR_LEGACY_LUT_ENTRIES); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c index f263e1a4537e..a9839485f2a2 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c @@ -759,6 +759,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us int max_param_num = 11; enum dp_test_pattern test_pattern = DP_TEST_PATTERN_UNSUPPORTED; bool disable_hpd = false; + bool supports_hpd = link->irq_source_hpd != DC_IRQ_SOURCE_INVALID; bool valid_test_pattern = false; uint8_t param_nums = 0; /* init with default 80bit custom pattern */ @@ -850,7 +851,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us * because it might have been disabled after a test pattern was set. * AUX depends on HPD * sequence dependent, do not move! */ - if (!disable_hpd) + if (supports_hpd && !disable_hpd) dc_link_enable_hpd(link); prefer_link_settings.lane_count = link->verified_link_cap.lane_count; @@ -888,7 +889,7 @@ static ssize_t dp_phy_test_pattern_debugfs_write(struct file *f, const char __us * Need disable interrupt to avoid SW driver disable DP output. This is * done after the test pattern is set. */ - if (valid_test_pattern && disable_hpd) + if (valid_test_pattern && supports_hpd && disable_hpd) dc_link_disable_hpd(link); kfree(wr_buf); @@ -1302,7 +1303,8 @@ static int odm_combine_segments_show(struct seq_file *m, void *unused) if (connector->status != connector_status_connected) return -ENODEV; - if (pipe_ctx != NULL && pipe_ctx->stream_res.tg->funcs->get_odm_combine_segments) + if (pipe_ctx && pipe_ctx->stream_res.tg && + pipe_ctx->stream_res.tg->funcs->get_odm_combine_segments) pipe_ctx->stream_res.tg->funcs->get_odm_combine_segments(pipe_ctx->stream_res.tg, &segments); seq_printf(m, "%d\n", segments); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index 19038f336155..85ce558cefc5 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -201,6 +201,7 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, struct mod_hdcp_link_adjustment link_adjust; struct mod_hdcp_display_adjustment display_adjust; unsigned int conn_index = aconnector->base.index; + const struct dc *dc = aconnector->dc_link->dc; guard(mutex)(&hdcp_w->mutex); drm_connector_get(&aconnector->base); @@ -231,6 +232,9 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, link_adjust.hdcp1.disable = 1; link_adjust.hdcp2.force_type = MOD_HDCP_FORCE_TYPE_1; } + link_adjust.hdcp2.use_fw_locality_check = + (dc->caps.fused_io_supported || dc->debug.hdcp_lc_force_fw_enable); + link_adjust.hdcp2.use_sw_locality_fallback = dc->debug.hdcp_lc_enable_sw_fallback; schedule_delayed_work(&hdcp_w->property_validate_dwork, msecs_to_jiffies(DRM_HDCP_CHECK_PERIOD_MS)); @@ -534,6 +538,7 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index]; struct dc_sink *sink = NULL; bool link_is_hdcp14 = false; + const struct dc *dc = aconnector->dc_link->dc; if (config->dpms_off) { hdcp_remove_display(hdcp_work, link_index, aconnector); @@ -575,6 +580,8 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) link->adjust.auth_delay = 2; link->adjust.retry_limit = MAX_NUM_OF_ATTEMPTS; link->adjust.hdcp1.disable = 0; + link->adjust.hdcp2.use_fw_locality_check = (dc->caps.fused_io_supported || dc->debug.hdcp_lc_force_fw_enable); + link->adjust.hdcp2.use_sw_locality_fallback = dc->debug.hdcp_lc_enable_sw_fallback; hdcp_w->encryption_status[display->index] = MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF; DRM_DEBUG_DRIVER("[HDCP_DM] display %d, CP %d, type %d\n", aconnector->base.index, @@ -786,15 +793,8 @@ struct hdcp_workqueue *hdcp_create_workqueue(struct amdgpu_device *adev, ddc_funcs->read_i2c = lp_read_i2c; ddc_funcs->write_dpcd = lp_write_dpcd; ddc_funcs->read_dpcd = lp_read_dpcd; - - config->debug.lc_enable_sw_fallback = dc->debug.hdcp_lc_enable_sw_fallback; - if (dc->caps.fused_io_supported || dc->debug.hdcp_lc_force_fw_enable) { - ddc_funcs->atomic_write_poll_read_i2c = lp_atomic_write_poll_read_i2c; - ddc_funcs->atomic_write_poll_read_aux = lp_atomic_write_poll_read_aux; - } else { - ddc_funcs->atomic_write_poll_read_i2c = NULL; - ddc_funcs->atomic_write_poll_read_aux = NULL; - } + ddc_funcs->atomic_write_poll_read_i2c = lp_atomic_write_poll_read_i2c; + ddc_funcs->atomic_write_poll_read_aux = lp_atomic_write_poll_read_aux; memset(hdcp_work[i].aconnector, 0, sizeof(struct amdgpu_dm_connector *) * diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index fe100e4c9801..ac98c746c3de 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -83,6 +83,7 @@ static void apply_edid_quirks(struct drm_device *dev, struct edid *edid, struct edid_caps->panel_patch.remove_sink_ext_caps = true; break; case drm_edid_encode_panel_id('S', 'D', 'C', 0x4154): + case drm_edid_encode_panel_id('S', 'D', 'C', 0x4171): drm_dbg_driver(dev, "Disabling VSC on monitor with panel id %X\n", panel_id); edid_caps->panel_patch.disable_colorimetry = true; break; @@ -130,6 +131,7 @@ enum dc_edid_status dm_helpers_parse_edid_caps( edid_caps->serial_number = edid_buf->serial; edid_caps->manufacture_week = edid_buf->mfg_week; edid_caps->manufacture_year = edid_buf->mfg_year; + edid_caps->analog = !(edid_buf->input & DRM_EDID_INPUT_DIGITAL); drm_edid_get_monitor_name(edid_buf, edid_caps->display_name, @@ -996,8 +998,8 @@ enum dc_edid_status dm_helpers_read_local_edid( struct amdgpu_dm_connector *aconnector = link->priv; struct drm_connector *connector = &aconnector->base; struct i2c_adapter *ddc; - int retry = 3; - enum dc_edid_status edid_status; + int retry = 25; + enum dc_edid_status edid_status = EDID_NO_RESPONSE; const struct drm_edid *drm_edid; const struct edid *edid; @@ -1027,7 +1029,7 @@ enum dc_edid_status dm_helpers_read_local_edid( } if (!drm_edid) - return EDID_NO_RESPONSE; + continue; edid = drm_edid_raw(drm_edid); // FIXME: Get rid of drm_edid_raw() if (!edid || @@ -1045,7 +1047,7 @@ enum dc_edid_status dm_helpers_read_local_edid( &sink->dc_edid, &sink->edid_caps); - } while (edid_status == EDID_BAD_CHECKSUM && --retry > 0); + } while ((edid_status == EDID_BAD_CHECKSUM || edid_status == EDID_NO_RESPONSE) && --retry > 0); if (edid_status != EDID_OK) DRM_ERROR("EDID err: %d, on connector: %s", diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c index a1c722112c22..0a2a3f233a0e 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_irq.c @@ -476,6 +476,7 @@ void amdgpu_dm_irq_fini(struct amdgpu_device *adev) void amdgpu_dm_irq_suspend(struct amdgpu_device *adev) { + struct drm_device *dev = adev_to_drm(adev); int src; struct list_head *hnd_list_h; struct list_head *hnd_list_l; @@ -512,6 +513,9 @@ void amdgpu_dm_irq_suspend(struct amdgpu_device *adev) } DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_disable(dev); } void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev) @@ -537,6 +541,7 @@ void amdgpu_dm_irq_resume_early(struct amdgpu_device *adev) void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev) { + struct drm_device *dev = adev_to_drm(adev); int src; struct list_head *hnd_list_h, *hnd_list_l; unsigned long irq_table_flags; @@ -557,6 +562,9 @@ void amdgpu_dm_irq_resume_late(struct amdgpu_device *adev) } DM_IRQ_TABLE_UNLOCK(adev, irq_table_flags); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_enable(dev); } /* @@ -893,6 +901,7 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) struct drm_connector_list_iter iter; int irq_type; int i; + bool use_polling = false; /* First, clear all hpd and hpdrx interrupts */ for (i = DC_IRQ_SOURCE_HPD1; i <= DC_IRQ_SOURCE_HPD6RX; i++) { @@ -906,6 +915,8 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) struct amdgpu_dm_connector *amdgpu_dm_connector; const struct dc_link *dc_link; + use_polling |= connector->polled != DRM_CONNECTOR_POLL_HPD; + if (connector->connector_type == DRM_MODE_CONNECTOR_WRITEBACK) continue; @@ -947,6 +958,9 @@ void amdgpu_dm_hpd_init(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); + + if (use_polling) + drm_kms_helper_poll_init(dev); } /** @@ -997,4 +1011,7 @@ void amdgpu_dm_hpd_fini(struct amdgpu_device *adev) } } drm_connector_list_iter_end(&iter); + + if (dev->mode_config.poll_enabled) + drm_kms_helper_poll_fini(dev); } diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 5e92eaa67aa3..dbd1da4d85d3 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -884,26 +884,28 @@ struct dsc_mst_fairness_params { }; #if defined(CONFIG_DRM_AMD_DC_FP) -static uint16_t get_fec_overhead_multiplier(struct dc_link *dc_link) +static uint64_t kbps_to_pbn(int kbps, bool is_peak_pbn) { - u8 link_coding_cap; - uint16_t fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_8B_10B; + uint64_t effective_kbps = (uint64_t)kbps; - link_coding_cap = dc_link_dp_mst_decide_link_encoding_format(dc_link); - if (link_coding_cap == DP_128b_132b_ENCODING) - fec_overhead_multiplier_x1000 = PBN_FEC_OVERHEAD_MULTIPLIER_128B_132B; + if (is_peak_pbn) { // add 0.6% (1006/1000) overhead into effective kbps + effective_kbps *= 1006; + effective_kbps = div_u64(effective_kbps, 1000); + } - return fec_overhead_multiplier_x1000; + return (uint64_t) DIV64_U64_ROUND_UP(effective_kbps * 64, (54 * 8 * 1000)); } -static int kbps_to_peak_pbn(int kbps, uint16_t fec_overhead_multiplier_x1000) +static uint32_t pbn_to_kbps(unsigned int pbn, bool with_margin) { - u64 peak_kbps = kbps; + uint64_t pbn_effective = (uint64_t)pbn; + + if (with_margin) // deduct 0.6% (994/1000) overhead from effective pbn + pbn_effective *= (1000000 / PEAK_FACTOR_X1000); + else + pbn_effective *= 1000; - peak_kbps *= 1006; - peak_kbps *= fec_overhead_multiplier_x1000; - peak_kbps = div_u64(peak_kbps, 1000 * 1000); - return (int) DIV64_U64_ROUND_UP(peak_kbps * 64, (54 * 8 * 1000)); + return DIV_U64_ROUND_UP(pbn_effective * 8 * 54, 64); } static void set_dsc_configs_from_fairness_vars(struct dsc_mst_fairness_params *params, @@ -974,7 +976,7 @@ static int bpp_x16_from_pbn(struct dsc_mst_fairness_params param, int pbn) dc_dsc_get_default_config_option(param.sink->ctx->dc, &dsc_options); dsc_options.max_target_bpp_limit_override_x16 = drm_connector->display_info.max_dsc_bpp * 16; - kbps = div_u64((u64)pbn * 994 * 8 * 54, 64); + kbps = pbn_to_kbps(pbn, false); dc_dsc_compute_config( param.sink->ctx->dc->res_pool->dscs[0], ¶m.sink->dsc_caps.dsc_dec_caps, @@ -1003,12 +1005,11 @@ static int increase_dsc_bpp(struct drm_atomic_state *state, int link_timeslots_used; int fair_pbn_alloc; int ret = 0; - uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); for (i = 0; i < count; i++) { if (vars[i + k].dsc_enabled) { initial_slack[i] = - kbps_to_peak_pbn(params[i].bw_range.max_kbps, fec_overhead_multiplier_x1000) - vars[i + k].pbn; + kbps_to_pbn(params[i].bw_range.max_kbps, false) - vars[i + k].pbn; bpp_increased[i] = false; remaining_to_increase += 1; } else { @@ -1104,7 +1105,6 @@ static int try_disable_dsc(struct drm_atomic_state *state, int next_index; int remaining_to_try = 0; int ret; - uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); int var_pbn; for (i = 0; i < count; i++) { @@ -1137,7 +1137,7 @@ static int try_disable_dsc(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC index #%d, try no compression\n", next_index); var_pbn = vars[next_index].pbn; - vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.stream_kbps, fec_overhead_multiplier_x1000); + vars[next_index].pbn = kbps_to_pbn(params[next_index].bw_range.stream_kbps, true); ret = drm_dp_atomic_find_time_slots(state, params[next_index].port->mgr, params[next_index].port, @@ -1197,7 +1197,6 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, int count = 0; int i, k, ret; bool debugfs_overwrite = false; - uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); struct drm_connector_state *new_conn_state; memset(params, 0, sizeof(params)); @@ -1278,7 +1277,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC Try no compression\n"); for (i = 0; i < count; i++) { vars[i + k].aconnector = params[i].aconnector; - vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000); + vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.stream_kbps, false); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, params[i].port, @@ -1300,7 +1299,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, DRM_DEBUG_DRIVER("MST_DSC Try max compression\n"); for (i = 0; i < count; i++) { if (params[i].compression_possible && params[i].clock_force_enable != DSC_CLK_FORCE_DISABLE) { - vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.min_kbps, fec_overhead_multiplier_x1000); + vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.min_kbps, false); vars[i + k].dsc_enabled = true; vars[i + k].bpp_x16 = params[i].bw_range.min_target_bpp_x16; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, @@ -1308,7 +1307,7 @@ static int compute_mst_dsc_configs_for_link(struct drm_atomic_state *state, if (ret < 0) return ret; } else { - vars[i + k].pbn = kbps_to_peak_pbn(params[i].bw_range.stream_kbps, fec_overhead_multiplier_x1000); + vars[i + k].pbn = kbps_to_pbn(params[i].bw_range.stream_kbps, false); vars[i + k].dsc_enabled = false; vars[i + k].bpp_x16 = 0; ret = drm_dp_atomic_find_time_slots(state, params[i].port->mgr, @@ -1763,18 +1762,6 @@ clean_exit: return ret; } -static uint32_t kbps_from_pbn(unsigned int pbn) -{ - uint64_t kbps = (uint64_t)pbn; - - kbps *= (1000000 / PEAK_FACTOR_X1000); - kbps *= 8; - kbps *= 54; - kbps /= 64; - - return (uint32_t)kbps; -} - static bool is_dsc_common_config_possible(struct dc_stream_state *stream, struct dc_dsc_bw_range *bw_range) { @@ -1873,7 +1860,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( dc_link_get_highest_encoding_format(stream->link)); cur_link_settings = stream->link->verified_link_cap; root_link_bw_in_kbps = dc_link_bandwidth_kbps(aconnector->dc_link, &cur_link_settings); - virtual_channel_bw_in_kbps = kbps_from_pbn(aconnector->mst_output_port->full_pbn); + virtual_channel_bw_in_kbps = pbn_to_kbps(aconnector->mst_output_port->full_pbn, true); /* pick the end to end bw bottleneck */ end_to_end_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps); @@ -1926,7 +1913,7 @@ enum dc_status dm_dp_mst_is_port_support_mode( immediate_upstream_port = aconnector->mst_output_port->parent->port_parent; if (immediate_upstream_port) { - virtual_channel_bw_in_kbps = kbps_from_pbn(immediate_upstream_port->full_pbn); + virtual_channel_bw_in_kbps = pbn_to_kbps(immediate_upstream_port->full_pbn, true); virtual_channel_bw_in_kbps = min(root_link_bw_in_kbps, virtual_channel_bw_in_kbps); } else { /* For topology LCT 1 case - only one mstb*/ diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c index e027798ece03..2e3ee78999d9 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_plane.c @@ -37,6 +37,7 @@ #include "amdgpu_display.h" #include "amdgpu_dm_trace.h" #include "amdgpu_dm_plane.h" +#include "amdgpu_dm_colorop.h" #include "gc/gc_11_0_0_offset.h" #include "gc/gc_11_0_0_sh_mask.h" @@ -1782,6 +1783,39 @@ dm_atomic_plane_get_property(struct drm_plane *plane, return 0; } +#else + +#define MAX_COLOR_PIPELINES 5 + +static int +dm_plane_init_colorops(struct drm_plane *plane) +{ + struct drm_prop_enum_list pipelines[MAX_COLOR_PIPELINES]; + struct drm_device *dev = plane->dev; + struct amdgpu_device *adev = drm_to_adev(dev); + struct dc *dc = adev->dm.dc; + int len = 0; + int ret; + + if (plane->type == DRM_PLANE_TYPE_CURSOR) + return 0; + + /* initialize pipeline */ + if (dc->ctx->dce_version >= DCN_VERSION_3_0) { + ret = amdgpu_dm_initialize_default_pipeline(plane, &pipelines[len]); + if (ret) { + drm_err(plane->dev, "Failed to create color pipeline for plane %d: %d\n", + plane->base.id, ret); + return ret; + } + len++; + + /* Create COLOR_PIPELINE property and attach */ + drm_plane_create_color_pipeline_property(plane, pipelines, len); + } + + return 0; +} #endif static const struct drm_plane_funcs dm_plane_funcs = { @@ -1890,7 +1924,12 @@ int amdgpu_dm_plane_init(struct amdgpu_display_manager *dm, #ifdef AMD_PRIVATE_COLOR dm_atomic_plane_attach_color_mgmt_properties(dm, plane); +#else + res = dm_plane_init_colorops(plane); + if (res) + return res; #endif + /* Create (reset) the plane state */ if (plane->funcs->reset) plane->funcs->reset(plane); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c index 80704d709e44..da94e3544b65 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_replay.c @@ -162,7 +162,7 @@ bool amdgpu_dm_replay_enable(struct dc_stream_state *stream, bool wait) if (link) { link->dc->link_srv->edp_setup_replay(link, stream); - link->dc->link_srv->edp_set_coasting_vtotal(link, stream->timing.v_total); + link->dc->link_srv->edp_set_coasting_vtotal(link, stream->timing.v_total, 0); DRM_DEBUG_DRIVER("Enabling replay...\n"); link->dc->link_srv->edp_set_replay_allow_active(link, &replay_active, wait, false, NULL); return true; diff --git a/drivers/gpu/drm/amd/display/dc/Makefile b/drivers/gpu/drm/amd/display/dc/Makefile index dc943abd6dba..7277ed21552f 100644 --- a/drivers/gpu/drm/amd/display/dc/Makefile +++ b/drivers/gpu/drm/amd/display/dc/Makefile @@ -36,7 +36,7 @@ DC_LIBS += dcn30 DC_LIBS += dcn301 DC_LIBS += dcn31 DC_LIBS += dml -DC_LIBS += dml2 +DC_LIBS += dml2_0 DC_LIBS += soc_and_ip_translator endif diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c index 154fd2c18e88..d1471f34e419 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser.c @@ -67,7 +67,9 @@ static ATOM_HPD_INT_RECORD *get_hpd_record(struct bios_parser *bp, ATOM_OBJECT *object); static struct device_id device_type_from_device_id(uint16_t device_id); static uint32_t signal_to_ss_id(enum as_signal_type signal); -static uint32_t get_support_mask_for_device_id(struct device_id device_id); +static uint32_t get_support_mask_for_device_id( + enum dal_device_type device_type, + uint32_t enum_id); static ATOM_ENCODER_CAP_RECORD_V2 *get_encoder_cap_record( struct bios_parser *bp, ATOM_OBJECT *object); @@ -441,6 +443,7 @@ static enum bp_result get_firmware_info_v1_4( le32_to_cpu(firmware_info->ulMinPixelClockPLL_Output) * 10; info->pll_info.max_output_pxl_clk_pll_frequency = le32_to_cpu(firmware_info->ulMaxPixelClockPLL_Output) * 10; + info->max_pixel_clock = le16_to_cpu(firmware_info->usMaxPixelClock) * 10; if (firmware_info->usFirmwareCapability.sbfAccess.MemoryClockSS_Support) /* Since there is no information on the SS, report conservative @@ -497,6 +500,7 @@ static enum bp_result get_firmware_info_v2_1( info->external_clock_source_frequency_for_dp = le16_to_cpu(firmwareInfo->usUniphyDPModeExtClkFreq) * 10; info->min_allowed_bl_level = firmwareInfo->ucMinAllowedBL_Level; + info->max_pixel_clock = le16_to_cpu(firmwareInfo->usMaxPixelClock) * 10; /* There should be only one entry in the SS info table for Memory Clock */ @@ -736,18 +740,94 @@ static enum bp_result bios_parser_transmitter_control( return bp->cmd_tbl.transmitter_control(bp, cntl); } +static enum bp_result bios_parser_select_crtc_source( + struct dc_bios *dcb, + struct bp_crtc_source_select *bp_params) +{ + struct bios_parser *bp = BP_FROM_DCB(dcb); + + if (!bp->cmd_tbl.select_crtc_source) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.select_crtc_source(bp, bp_params); +} + static enum bp_result bios_parser_encoder_control( struct dc_bios *dcb, struct bp_encoder_control *cntl) { struct bios_parser *bp = BP_FROM_DCB(dcb); + if (cntl->engine_id == ENGINE_ID_DACA) { + if (!bp->cmd_tbl.dac1_encoder_control) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.dac1_encoder_control( + bp, cntl->action == ENCODER_CONTROL_ENABLE, + cntl->pixel_clock, ATOM_DAC1_PS2); + } else if (cntl->engine_id == ENGINE_ID_DACB) { + if (!bp->cmd_tbl.dac2_encoder_control) + return BP_RESULT_FAILURE; + + return bp->cmd_tbl.dac2_encoder_control( + bp, cntl->action == ENCODER_CONTROL_ENABLE, + cntl->pixel_clock, ATOM_DAC1_PS2); + } + if (!bp->cmd_tbl.dig_encoder_control) return BP_RESULT_FAILURE; return bp->cmd_tbl.dig_encoder_control(bp, cntl); } +static enum bp_result bios_parser_dac_load_detection( + struct dc_bios *dcb, + enum engine_id engine_id, + enum dal_device_type device_type, + uint32_t enum_id) +{ + struct bios_parser *bp = BP_FROM_DCB(dcb); + struct dc_context *ctx = dcb->ctx; + struct bp_load_detection_parameters bp_params = {0}; + enum bp_result bp_result; + uint32_t bios_0_scratch; + uint32_t device_id_mask = 0; + + bp_params.engine_id = engine_id; + bp_params.device_id = get_support_mask_for_device_id(device_type, enum_id); + + if (engine_id != ENGINE_ID_DACA && + engine_id != ENGINE_ID_DACB) + return BP_RESULT_UNSUPPORTED; + + if (!bp->cmd_tbl.dac_load_detection) + return BP_RESULT_UNSUPPORTED; + + if (bp_params.device_id == ATOM_DEVICE_CRT1_SUPPORT) + device_id_mask = ATOM_S0_CRT1_MASK; + else if (bp_params.device_id == ATOM_DEVICE_CRT2_SUPPORT) + device_id_mask = ATOM_S0_CRT2_MASK; + else + return BP_RESULT_UNSUPPORTED; + + /* BIOS will write the detected devices to BIOS_SCRATCH_0, clear corresponding bit */ + bios_0_scratch = dm_read_reg(ctx, bp->base.regs->BIOS_SCRATCH_0); + bios_0_scratch &= ~device_id_mask; + dm_write_reg(ctx, bp->base.regs->BIOS_SCRATCH_0, bios_0_scratch); + + bp_result = bp->cmd_tbl.dac_load_detection(bp, &bp_params); + + if (bp_result != BP_RESULT_OK) + return bp_result; + + bios_0_scratch = dm_read_reg(ctx, bp->base.regs->BIOS_SCRATCH_0); + + if (bios_0_scratch & device_id_mask) + return BP_RESULT_OK; + + return BP_RESULT_FAILURE; +} + static enum bp_result bios_parser_adjust_pixel_clock( struct dc_bios *dcb, struct bp_adjust_pixel_clock_parameters *bp_params) @@ -858,7 +938,7 @@ static bool bios_parser_is_device_id_supported( { struct bios_parser *bp = BP_FROM_DCB(dcb); - uint32_t mask = get_support_mask_for_device_id(id); + uint32_t mask = get_support_mask_for_device_id(id.device_type, id.enum_id); return (le16_to_cpu(bp->object_info_tbl.v1_1->usDeviceSupport) & mask) != 0; } @@ -2149,11 +2229,10 @@ static uint32_t signal_to_ss_id(enum as_signal_type signal) return clk_id_ss; } -static uint32_t get_support_mask_for_device_id(struct device_id device_id) +static uint32_t get_support_mask_for_device_id( + enum dal_device_type device_type, + uint32_t enum_id) { - enum dal_device_type device_type = device_id.device_type; - uint32_t enum_id = device_id.enum_id; - switch (device_type) { case DEVICE_TYPE_LCD: switch (enum_id) { @@ -2829,8 +2908,12 @@ static const struct dc_vbios_funcs vbios_funcs = { .is_device_id_supported = bios_parser_is_device_id_supported, /* COMMANDS */ + .select_crtc_source = bios_parser_select_crtc_source, + .encoder_control = bios_parser_encoder_control, + .dac_load_detection = bios_parser_dac_load_detection, + .transmitter_control = bios_parser_transmitter_control, .enable_crtc = bios_parser_enable_crtc, diff --git a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c index 04eb647acc4e..550a9f1d03f8 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/bios_parser2.c @@ -1480,10 +1480,10 @@ static enum bp_result get_embedded_panel_info_v2_1( /* not provided by VBIOS */ info->lcd_timing.misc_info.HORIZONTAL_CUT_OFF = 0; - info->lcd_timing.misc_info.H_SYNC_POLARITY = ~(uint32_t) (lvds->lcd_timing.miscinfo - & ATOM_HSYNC_POLARITY); - info->lcd_timing.misc_info.V_SYNC_POLARITY = ~(uint32_t) (lvds->lcd_timing.miscinfo - & ATOM_VSYNC_POLARITY); + info->lcd_timing.misc_info.H_SYNC_POLARITY = !(lvds->lcd_timing.miscinfo & + ATOM_HSYNC_POLARITY); + info->lcd_timing.misc_info.V_SYNC_POLARITY = !(lvds->lcd_timing.miscinfo & + ATOM_VSYNC_POLARITY); /* not provided by VBIOS */ info->lcd_timing.misc_info.VERTICAL_CUT_OFF = 0; diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.c b/drivers/gpu/drm/amd/display/dc/bios/command_table.c index 58e88778da7f..22457f417e65 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.c @@ -52,7 +52,9 @@ static void init_transmitter_control(struct bios_parser *bp); static void init_set_pixel_clock(struct bios_parser *bp); static void init_enable_spread_spectrum_on_ppll(struct bios_parser *bp); static void init_adjust_display_pll(struct bios_parser *bp); +static void init_select_crtc_source(struct bios_parser *bp); static void init_dac_encoder_control(struct bios_parser *bp); +static void init_dac_load_detection(struct bios_parser *bp); static void init_dac_output_control(struct bios_parser *bp); static void init_set_crtc_timing(struct bios_parser *bp); static void init_enable_crtc(struct bios_parser *bp); @@ -69,7 +71,9 @@ void dal_bios_parser_init_cmd_tbl(struct bios_parser *bp) init_set_pixel_clock(bp); init_enable_spread_spectrum_on_ppll(bp); init_adjust_display_pll(bp); + init_select_crtc_source(bp); init_dac_encoder_control(bp); + init_dac_load_detection(bp); init_dac_output_control(bp); init_set_crtc_timing(bp); init_enable_crtc(bp); @@ -1612,6 +1616,198 @@ static enum bp_result adjust_display_pll_v3( /******************************************************************************* ******************************************************************************** ** + ** SELECT CRTC SOURCE + ** + ******************************************************************************** + *******************************************************************************/ + +static enum bp_result select_crtc_source_v1( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); +static enum bp_result select_crtc_source_v2( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); +static enum bp_result select_crtc_source_v3( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); + +static void init_select_crtc_source(struct bios_parser *bp) +{ + switch (BIOS_CMD_TABLE_PARA_REVISION(SelectCRTC_Source)) { + case 1: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v1; + break; + case 2: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v2; + break; + case 3: + bp->cmd_tbl.select_crtc_source = select_crtc_source_v3; + break; + default: + bp->cmd_tbl.select_crtc_source = NULL; + break; + } +} + +static enum bp_result select_crtc_source_v1( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PS_ALLOCATION params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + switch (bp_params->engine_id) { + case ENGINE_ID_DACA: + params.ucDevice = ATOM_DEVICE_CRT1_INDEX; + break; + case ENGINE_ID_DACB: + params.ucDevice = ATOM_DEVICE_CRT2_INDEX; + break; + default: + return BP_RESULT_BADINPUT; + } + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + +static bool select_crtc_source_v2_encoder_id( + enum engine_id engine_id, uint8_t *out_encoder_id) +{ + uint8_t encoder_id = 0; + + switch (engine_id) { + case ENGINE_ID_DIGA: + encoder_id = ASIC_INT_DIG1_ENCODER_ID; + break; + case ENGINE_ID_DIGB: + encoder_id = ASIC_INT_DIG2_ENCODER_ID; + break; + case ENGINE_ID_DIGC: + encoder_id = ASIC_INT_DIG3_ENCODER_ID; + break; + case ENGINE_ID_DIGD: + encoder_id = ASIC_INT_DIG4_ENCODER_ID; + break; + case ENGINE_ID_DIGE: + encoder_id = ASIC_INT_DIG5_ENCODER_ID; + break; + case ENGINE_ID_DIGF: + encoder_id = ASIC_INT_DIG6_ENCODER_ID; + break; + case ENGINE_ID_DIGG: + encoder_id = ASIC_INT_DIG7_ENCODER_ID; + break; + case ENGINE_ID_DACA: + encoder_id = ASIC_INT_DAC1_ENCODER_ID; + break; + case ENGINE_ID_DACB: + encoder_id = ASIC_INT_DAC2_ENCODER_ID; + break; + default: + return false; + } + + *out_encoder_id = encoder_id; + return true; +} + +static bool select_crtc_source_v2_encoder_mode( + enum signal_type signal_type, uint8_t *out_encoder_mode) +{ + uint8_t encoder_mode = 0; + + switch (signal_type) { + case SIGNAL_TYPE_DVI_SINGLE_LINK: + case SIGNAL_TYPE_DVI_DUAL_LINK: + encoder_mode = ATOM_ENCODER_MODE_DVI; + break; + case SIGNAL_TYPE_HDMI_TYPE_A: + encoder_mode = ATOM_ENCODER_MODE_HDMI; + break; + case SIGNAL_TYPE_LVDS: + encoder_mode = ATOM_ENCODER_MODE_LVDS; + break; + case SIGNAL_TYPE_RGB: + encoder_mode = ATOM_ENCODER_MODE_CRT; + break; + case SIGNAL_TYPE_DISPLAY_PORT: + encoder_mode = ATOM_ENCODER_MODE_DP; + break; + case SIGNAL_TYPE_DISPLAY_PORT_MST: + encoder_mode = ATOM_ENCODER_MODE_DP_MST; + break; + case SIGNAL_TYPE_EDP: + encoder_mode = ATOM_ENCODER_MODE_DP; + break; + default: + return false; + } + + *out_encoder_mode = encoder_mode; + return true; +} + +static enum bp_result select_crtc_source_v2( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PARAMETERS_V3 params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + if (!select_crtc_source_v2_encoder_id( + bp_params->engine_id, + ¶ms.ucEncoderID)) + return BP_RESULT_BADINPUT; + if (!select_crtc_source_v2_encoder_mode( + bp_params->sink_signal, + ¶ms.ucEncodeMode)) + return BP_RESULT_BADINPUT; + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + +static enum bp_result select_crtc_source_v3( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + SELECT_CRTC_SOURCE_PARAMETERS_V3 params; + + if (!bp->cmd_helper->controller_id_to_atom(bp_params->controller_id, ¶ms.ucCRTC)) + return BP_RESULT_BADINPUT; + + if (!select_crtc_source_v2_encoder_id( + bp_params->engine_id, + ¶ms.ucEncoderID)) + return BP_RESULT_BADINPUT; + if (!select_crtc_source_v2_encoder_mode( + bp_params->sink_signal, + ¶ms.ucEncodeMode)) + return BP_RESULT_BADINPUT; + + params.ucDstBpc = bp_params->bit_depth; + + if (EXEC_BIOS_CMD_TABLE(SelectCRTC_Source, params)) + result = BP_RESULT_OK; + + return result; +} + +/******************************************************************************* + ******************************************************************************** + ** ** DAC ENCODER CONTROL ** ******************************************************************************** @@ -1711,6 +1907,96 @@ static enum bp_result dac2_encoder_control_v1( /******************************************************************************* ******************************************************************************** ** + ** DAC LOAD DETECTION + ** + ******************************************************************************** + *******************************************************************************/ + +static enum bp_result dac_load_detection_v1( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); + +static enum bp_result dac_load_detection_v3( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); + +static void init_dac_load_detection(struct bios_parser *bp) +{ + switch (BIOS_CMD_TABLE_PARA_REVISION(DAC_LoadDetection)) { + case 1: + case 2: + bp->cmd_tbl.dac_load_detection = dac_load_detection_v1; + break; + case 3: + default: + bp->cmd_tbl.dac_load_detection = dac_load_detection_v3; + break; + } +} + +static void dac_load_detect_prepare_params( + struct _DAC_LOAD_DETECTION_PS_ALLOCATION *params, + enum engine_id engine_id, + uint16_t device_id, + uint8_t misc) +{ + uint8_t dac_type = ENGINE_ID_DACA; + + if (engine_id == ENGINE_ID_DACB) + dac_type = ATOM_DAC_B; + + params->sDacload.usDeviceID = cpu_to_le16(device_id); + params->sDacload.ucDacType = dac_type; + params->sDacload.ucMisc = misc; +} + +static enum bp_result dac_load_detection_v1( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + DAC_LOAD_DETECTION_PS_ALLOCATION params; + + dac_load_detect_prepare_params( + ¶ms, + bp_params->engine_id, + bp_params->device_id, + 0); + + if (EXEC_BIOS_CMD_TABLE(DAC_LoadDetection, params)) + result = BP_RESULT_OK; + + return result; +} + +static enum bp_result dac_load_detection_v3( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params) +{ + enum bp_result result = BP_RESULT_FAILURE; + DAC_LOAD_DETECTION_PS_ALLOCATION params; + + uint8_t misc = 0; + + if (bp_params->device_id == ATOM_DEVICE_CV_SUPPORT || + bp_params->device_id == ATOM_DEVICE_TV1_SUPPORT) + misc = DAC_LOAD_MISC_YPrPb; + + dac_load_detect_prepare_params( + ¶ms, + bp_params->engine_id, + bp_params->device_id, + misc); + + if (EXEC_BIOS_CMD_TABLE(DAC_LoadDetection, params)) + result = BP_RESULT_OK; + + return result; +} + +/******************************************************************************* + ******************************************************************************** + ** ** DAC OUTPUT CONTROL ** ******************************************************************************** diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table.h b/drivers/gpu/drm/amd/display/dc/bios/command_table.h index ad533775e724..e89b1ba0048b 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table.h +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table.h @@ -52,6 +52,9 @@ struct cmd_tbl { enum bp_result (*adjust_display_pll)( struct bios_parser *bp, struct bp_adjust_pixel_clock_parameters *bp_params); + enum bp_result (*select_crtc_source)( + struct bios_parser *bp, + struct bp_crtc_source_select *bp_params); enum bp_result (*dac1_encoder_control)( struct bios_parser *bp, bool enable, @@ -68,6 +71,9 @@ struct cmd_tbl { enum bp_result (*dac2_output_control)( struct bios_parser *bp, bool enable); + enum bp_result (*dac_load_detection)( + struct bios_parser *bp, + struct bp_load_detection_parameters *bp_params); enum bp_result (*set_crtc_timing)( struct bios_parser *bp, struct bp_hw_crtc_timing_parameters *bp_params); diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c index 9e63fa72101c..db687a13174d 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn314/dcn314_clk_mgr.c @@ -509,16 +509,16 @@ void dcn314_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_b regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10; regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dppclk_bypass < 0 || regs_and_bypass->dppclk_bypass > 4) + if (regs_and_bypass->dppclk_bypass > 4) regs_and_bypass->dppclk_bypass = 0; regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dcfclk_bypass < 0 || regs_and_bypass->dcfclk_bypass > 4) + if (regs_and_bypass->dcfclk_bypass > 4) regs_and_bypass->dcfclk_bypass = 0; regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dispclk_bypass < 0 || regs_and_bypass->dispclk_bypass > 4) + if (regs_and_bypass->dispclk_bypass > 4) regs_and_bypass->dispclk_bypass = 0; regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dprefclk_bypass < 0 || regs_and_bypass->dprefclk_bypass > 4) + if (regs_and_bypass->dprefclk_bypass > 4) regs_and_bypass->dprefclk_bypass = 0; } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c index b315ed91e010..3a881451e9da 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c @@ -40,7 +40,7 @@ #include "dm_helpers.h" #include "dc_dmub_srv.h" - +#include "reg_helper.h" #include "logger_types.h" #undef DC_LOGGER #define DC_LOGGER \ @@ -48,9 +48,43 @@ #include "link_service.h" +#define MAX_INSTANCE 7 +#define MAX_SEGMENT 8 + +struct IP_BASE_INSTANCE { + unsigned int segment[MAX_SEGMENT]; +}; + +struct IP_BASE { + struct IP_BASE_INSTANCE instance[MAX_INSTANCE]; +}; + +static const struct IP_BASE CLK_BASE = { { { { 0x00016C00, 0x02401800, 0, 0, 0, 0, 0, 0 } }, + { { 0x00016E00, 0x02401C00, 0, 0, 0, 0, 0, 0 } }, + { { 0x00017000, 0x02402000, 0, 0, 0, 0, 0, 0 } }, + { { 0x00017200, 0x02402400, 0, 0, 0, 0, 0, 0 } }, + { { 0x0001B000, 0x0242D800, 0, 0, 0, 0, 0, 0 } }, + { { 0x0001B200, 0x0242DC00, 0, 0, 0, 0, 0, 0 } } } }; + +#define regCLK1_CLK0_CURRENT_CNT 0x0314 +#define regCLK1_CLK0_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK1_CURRENT_CNT 0x0315 +#define regCLK1_CLK1_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK2_CURRENT_CNT 0x0316 +#define regCLK1_CLK2_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK3_CURRENT_CNT 0x0317 +#define regCLK1_CLK3_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK4_CURRENT_CNT 0x0318 +#define regCLK1_CLK4_CURRENT_CNT_BASE_IDX 0 +#define regCLK1_CLK5_CURRENT_CNT 0x0319 +#define regCLK1_CLK5_CURRENT_CNT_BASE_IDX 0 + #define TO_CLK_MGR_DCN315(clk_mgr)\ container_of(clk_mgr, struct clk_mgr_dcn315, base) +#define REG(reg_name) \ + (CLK_BASE.instance[0].segment[reg ## reg_name ## _BASE_IDX] + reg ## reg_name) + #define UNSUPPORTED_DCFCLK 10000000 #define MIN_DPP_DISP_CLK 100000 @@ -245,9 +279,38 @@ static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base, dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } +static void dcn315_dump_clk_registers_internal(struct dcn35_clk_internal *internal, struct clk_mgr *clk_mgr_base) +{ + struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); + + // read dtbclk + internal->CLK1_CLK4_CURRENT_CNT = REG_READ(CLK1_CLK4_CURRENT_CNT); + + // read dcfclk + internal->CLK1_CLK3_CURRENT_CNT = REG_READ(CLK1_CLK3_CURRENT_CNT); + + // read dppclk + internal->CLK1_CLK1_CURRENT_CNT = REG_READ(CLK1_CLK1_CURRENT_CNT); + + // read dprefclk + internal->CLK1_CLK2_CURRENT_CNT = REG_READ(CLK1_CLK2_CURRENT_CNT); + + // read dispclk + internal->CLK1_CLK0_CURRENT_CNT = REG_READ(CLK1_CLK0_CURRENT_CNT); +} + static void dcn315_dump_clk_registers(struct clk_state_registers_and_bypass *regs_and_bypass, struct clk_mgr *clk_mgr_base, struct clk_log_info *log_info) { + struct dcn35_clk_internal internal = {0}; + + dcn315_dump_clk_registers_internal(&internal, clk_mgr_base); + + regs_and_bypass->dcfclk = internal.CLK1_CLK3_CURRENT_CNT / 10; + regs_and_bypass->dprefclk = internal.CLK1_CLK2_CURRENT_CNT / 10; + regs_and_bypass->dispclk = internal.CLK1_CLK0_CURRENT_CNT / 10; + regs_and_bypass->dppclk = internal.CLK1_CLK1_CURRENT_CNT / 10; + regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10; return; } @@ -594,13 +657,32 @@ static struct clk_mgr_funcs dcn315_funcs = { .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, .get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz, .update_clocks = dcn315_update_clocks, - .init_clocks = dcn31_init_clocks, + .init_clocks = dcn315_init_clocks, .enable_pme_wa = dcn315_enable_pme_wa, .are_clock_states_equal = dcn31_are_clock_states_equal, .notify_wm_ranges = dcn315_notify_wm_ranges }; extern struct clk_mgr_funcs dcn3_fpga_funcs; +void dcn315_init_clocks(struct clk_mgr *clk_mgr) +{ + struct clk_mgr_internal *clk_mgr_int = TO_CLK_MGR_INTERNAL(clk_mgr); + uint32_t ref_dtbclk = clk_mgr->clks.ref_dtbclk_khz; + struct clk_mgr_dcn315 *clk_mgr_dcn315 = TO_CLK_MGR_DCN315(clk_mgr_int); + struct clk_log_info log_info = {0}; + + memset(&(clk_mgr->clks), 0, sizeof(struct dc_clocks)); + // Assumption is that boot state always supports pstate + clk_mgr->clks.ref_dtbclk_khz = ref_dtbclk; // restore ref_dtbclk + clk_mgr->clks.p_state_change_support = true; + clk_mgr->clks.prev_p_state_change_support = true; + clk_mgr->clks.pwr_state = DCN_PWR_STATE_UNKNOWN; + clk_mgr->clks.zstate_support = DCN_ZSTATE_SUPPORT_UNKNOWN; + + dcn315_dump_clk_registers(&clk_mgr->boot_snapshot, &clk_mgr_dcn315->base.base, &log_info); + clk_mgr->clks.dispclk_khz = clk_mgr->boot_snapshot.dispclk * 1000; +} + void dcn315_clk_mgr_construct( struct dc_context *ctx, struct clk_mgr_dcn315 *clk_mgr, @@ -661,6 +743,7 @@ void dcn315_clk_mgr_construct( /* Saved clocks configured at boot for debug purposes */ dcn315_dump_clk_registers(&clk_mgr->base.base.boot_snapshot, &clk_mgr->base.base, &log_info); + clk_mgr->base.base.clks.dispclk_khz = clk_mgr->base.base.boot_snapshot.dispclk * 1000; clk_mgr->base.base.dprefclk_khz = 600000; clk_mgr->base.base.dprefclk_khz = dcn315_smu_get_dpref_clk(&clk_mgr->base); diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h index ac36ddf5dd1a..642ae3d4a790 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.h @@ -44,6 +44,7 @@ void dcn315_clk_mgr_construct(struct dc_context *ctx, struct pp_smu_funcs *pp_smu, struct dccg *dccg); +void dcn315_init_clocks(struct clk_mgr *clk_mgr); void dcn315_clk_mgr_destroy(struct clk_mgr_internal *clk_mgr_int); #endif //__DCN315_CLK_MGR_H__ diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c index b11383fba35f..dfd0c9505af0 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c @@ -394,6 +394,8 @@ void dcn35_update_clocks(struct clk_mgr *clk_mgr_base, display_count = dcn35_get_active_display_cnt_wa(dc, context, &all_active_disps); if (new_clocks->dtbclk_en && !new_clocks->ref_dtbclk_khz) new_clocks->ref_dtbclk_khz = 600000; + else if (!new_clocks->dtbclk_en && new_clocks->ref_dtbclk_khz > 590000) + new_clocks->ref_dtbclk_khz = 0; /* * if it is safe to lower, but we are already in the lower state, we don't have to do anything @@ -435,7 +437,7 @@ void dcn35_update_clocks(struct clk_mgr *clk_mgr_base, actual_dtbclk = REG_READ(CLK1_CLK4_CURRENT_CNT); - if (actual_dtbclk) { + if (actual_dtbclk > 590000) { clk_mgr_base->clks.ref_dtbclk_khz = new_clocks->ref_dtbclk_khz; clk_mgr_base->clks.dtbclk_en = new_clocks->dtbclk_en; } @@ -633,16 +635,16 @@ static void dcn35_save_clk_registers(struct clk_state_registers_and_bypass *regs regs_and_bypass->dtbclk = internal.CLK1_CLK4_CURRENT_CNT / 10; regs_and_bypass->dppclk_bypass = internal.CLK1_CLK1_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dppclk_bypass < 0 || regs_and_bypass->dppclk_bypass > 4) + if (regs_and_bypass->dppclk_bypass > 4) regs_and_bypass->dppclk_bypass = 0; regs_and_bypass->dcfclk_bypass = internal.CLK1_CLK3_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dcfclk_bypass < 0 || regs_and_bypass->dcfclk_bypass > 4) + if (regs_and_bypass->dcfclk_bypass > 4) regs_and_bypass->dcfclk_bypass = 0; regs_and_bypass->dispclk_bypass = internal.CLK1_CLK0_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dispclk_bypass < 0 || regs_and_bypass->dispclk_bypass > 4) + if (regs_and_bypass->dispclk_bypass > 4) regs_and_bypass->dispclk_bypass = 0; regs_and_bypass->dprefclk_bypass = internal.CLK1_CLK2_BYPASS_CNTL & 0x0007; - if (regs_and_bypass->dprefclk_bypass < 0 || regs_and_bypass->dprefclk_bypass > 4) + if (regs_and_bypass->dprefclk_bypass > 4) regs_and_bypass->dprefclk_bypass = 0; if (clk_mgr->base.base.ctx->dc->debug.pstate_enabled) { @@ -1293,6 +1295,35 @@ static void dcn35_update_clocks_fpga(struct clk_mgr *clk_mgr, dcn35_update_clocks_update_dtb_dto(clk_mgr_int, context, clk_mgr->clks.ref_dtbclk_khz); } +static unsigned int dcn35_get_max_clock_khz(struct clk_mgr *clk_mgr_base, enum clk_type clk_type) +{ + struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); + + unsigned int num_clk_levels; + + switch (clk_type) { + case CLK_TYPE_DISPCLK: + num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dispclk_levels; + return num_clk_levels ? + clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dispclk_mhz * 1000 : + clk_mgr->base.boot_snapshot.dispclk; + case CLK_TYPE_DPPCLK: + num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dppclk_levels; + return num_clk_levels ? + clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dppclk_mhz * 1000 : + clk_mgr->base.boot_snapshot.dppclk; + case CLK_TYPE_DSCCLK: + num_clk_levels = clk_mgr->base.bw_params->clk_table.num_entries_per_clk.num_dispclk_levels; + return num_clk_levels ? + clk_mgr->base.bw_params->clk_table.entries[num_clk_levels - 1].dispclk_mhz * 1000 / 3 : + clk_mgr->base.boot_snapshot.dispclk / 3; + default: + break; + } + + return 0; +} + static struct clk_mgr_funcs dcn35_funcs = { .get_dp_ref_clk_frequency = dce12_get_dp_ref_freq_khz, .get_dtb_ref_clk_frequency = dcn31_get_dtb_ref_freq_khz, @@ -1304,6 +1335,7 @@ static struct clk_mgr_funcs dcn35_funcs = { .set_low_power_state = dcn35_set_low_power_state, .exit_low_power_state = dcn35_exit_low_power_state, .is_ips_supported = dcn35_is_ips_supported, + .get_max_clock_khz = dcn35_get_max_clock_khz, }; struct clk_mgr_funcs dcn35_fpga_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 5f2d5638c819..8be9cbd43e18 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -83,7 +83,7 @@ #include "hw_sequencer_private.h" #if defined(CONFIG_DRM_AMD_DC_FP) -#include "dml2/dml2_internal_types.h" +#include "dml2_0/dml2_internal_types.h" #include "soc_and_ip_translator.h" #endif @@ -148,10 +148,16 @@ static const char DC_BUILD_ID[] = "production-build"; /* Private functions */ -static inline void elevate_update_type(enum surface_update_type *original, enum surface_update_type new) +static inline void elevate_update_type( + struct surface_update_descriptor *descriptor, + enum surface_update_type new_type, + enum dc_lock_descriptor new_locks +) { - if (new > *original) - *original = new; + if (new_type > descriptor->update_type) + descriptor->update_type = new_type; + + descriptor->lock_descriptor |= new_locks; } static void destroy_links(struct dc *dc) @@ -297,6 +303,7 @@ static bool create_links( link->link_id.id = CONNECTOR_ID_VIRTUAL; link->link_id.enum_id = ENUM_ID_1; link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED; + link->replay_settings.config.replay_version = DC_REPLAY_VERSION_UNSUPPORTED; link->link_enc = kzalloc(sizeof(*link->link_enc), GFP_KERNEL); if (!link->link_enc) { @@ -493,9 +500,14 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, 1, *adjust); stream->adjust.timing_adjust_pending = false; + + if (dc->hwss.notify_cursor_offload_drr_update) + dc->hwss.notify_cursor_offload_drr_update(dc, dc->current_state, stream); + return true; } } + return false; } @@ -1143,8 +1155,8 @@ static bool dc_construct(struct dc *dc, /* set i2c speed if not done by the respective dcnxxx__resource.c */ if (dc->caps.i2c_speed_in_khz_hdcp == 0) dc->caps.i2c_speed_in_khz_hdcp = dc->caps.i2c_speed_in_khz; - if (dc->caps.max_optimizable_video_width == 0) - dc->caps.max_optimizable_video_width = 5120; + if (dc->check_config.max_optimizable_video_width == 0) + dc->check_config.max_optimizable_video_width = 5120; dc->clk_mgr = dc_clk_mgr_create(dc->ctx, dc->res_pool->pp_smu, dc->res_pool->dccg); if (!dc->clk_mgr) goto fail; @@ -2135,6 +2147,14 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c if (!dcb->funcs->is_accelerated_mode(dcb)) { disable_vbios_mode_if_required(dc, context); dc->hwss.enable_accelerated_mode(dc, context); + } else if (get_seamless_boot_stream_count(dc->current_state) > 0) { + /* If the previous Stream still retains the apply seamless boot flag, + * it means the OS has not actually performed a flip yet. + * At this point, if we receive dc_commit_streams again, we should + * once more check whether the actual HW timing matches what the OS + * has provided + */ + disable_vbios_mode_if_required(dc, context); } if (dc->hwseq->funcs.wait_for_pipe_update_if_needed) { @@ -2158,8 +2178,8 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c */ if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, true, true, NULL, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, true); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, true); if (dc->hwss.update_dsc_pg) dc->hwss.update_dsc_pg(dc, context, false); @@ -2188,8 +2208,14 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc->hwss.wait_for_mpcc_disconnect(dc, dc->res_pool, pipe); } + for (i = 0; i < dc->current_state->stream_count; i++) + dc_dmub_srv_control_cursor_offload(dc, dc->current_state, dc->current_state->streams[i], false); + result = dc->hwss.apply_ctx_to_hw(dc, context); + for (i = 0; i < context->stream_count; i++) + dc_dmub_srv_control_cursor_offload(dc, context, context->streams[i], true); + if (result != DC_OK) { /* Application of dc_state to hardware stopped. */ dc->current_state->res_ctx.link_enc_cfg_ctx.mode = LINK_ENC_CFG_STEADY; @@ -2229,8 +2255,8 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c dc->hwss.commit_subvp_config(dc, context); if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, false, true, NULL, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, false); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, false); for (i = 0; i < context->stream_count; i++) { const struct dc_link *link = context->streams[i]->link; @@ -2645,47 +2671,50 @@ static bool is_surface_in_context( return false; } -static enum surface_update_type get_plane_info_update_type(const struct dc *dc, const struct dc_surface_update *u) +static struct surface_update_descriptor get_plane_info_update_type(const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; - enum surface_update_type update_type = UPDATE_TYPE_FAST; + struct surface_update_descriptor update_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (!u->plane_info) - return UPDATE_TYPE_FAST; + return update_type; + + // `plane_info` present means at least `STREAM` lock is required + elevate_update_type(&update_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); if (u->plane_info->color_space != u->surface->color_space) { update_flags->bits.color_space_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } if (u->plane_info->horizontal_mirror != u->surface->horizontal_mirror) { update_flags->bits.horizontal_mirror_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } if (u->plane_info->rotation != u->surface->rotation) { update_flags->bits.rotation_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (u->plane_info->format != u->surface->format) { update_flags->bits.pixel_format_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (u->plane_info->stereo_format != u->surface->stereo_format) { update_flags->bits.stereo_format_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (u->plane_info->per_pixel_alpha != u->surface->per_pixel_alpha) { update_flags->bits.per_pixel_alpha_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } if (u->plane_info->global_alpha_value != u->surface->global_alpha_value) { update_flags->bits.global_alpha_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } if (u->plane_info->dcc.enable != u->surface->dcc.enable @@ -2697,7 +2726,7 @@ static enum surface_update_type get_plane_info_update_type(const struct dc *dc, * recalculate stutter period. */ update_flags->bits.dcc_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (resource_pixel_format_to_bpp(u->plane_info->format) != @@ -2706,30 +2735,41 @@ static enum surface_update_type get_plane_info_update_type(const struct dc *dc, * and DML calculation */ update_flags->bits.bpp_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (u->plane_info->plane_size.surface_pitch != u->surface->plane_size.surface_pitch || u->plane_info->plane_size.chroma_pitch != u->surface->plane_size.chroma_pitch) { update_flags->bits.plane_size_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } + const struct dc_tiling_info *tiling = &u->plane_info->tiling_info; - if (memcmp(&u->plane_info->tiling_info, &u->surface->tiling_info, - sizeof(struct dc_tiling_info)) != 0) { + if (memcmp(tiling, &u->surface->tiling_info, sizeof(*tiling)) != 0) { update_flags->bits.swizzle_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_MED); - - /* todo: below are HW dependent, we should add a hook to - * DCE/N resource and validated there. - */ - if (!dc->debug.skip_full_updated_if_possible) { - /* swizzled mode requires RQ to be setup properly, - * thus need to run DML to calculate RQ settings - */ - update_flags->bits.bandwidth_change = 1; - elevate_update_type(&update_type, UPDATE_TYPE_FULL); + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); + + switch (tiling->gfxversion) { + case DcGfxVersion9: + case DcGfxVersion10: + case DcGfxVersion11: + if (tiling->gfx9.swizzle != DC_SW_LINEAR) { + update_flags->bits.bandwidth_change = 1; + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); + } + break; + case DcGfxAddr3: + if (tiling->gfx_addr3.swizzle != DC_ADDR3_SW_LINEAR) { + update_flags->bits.bandwidth_change = 1; + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); + } + break; + case DcGfxVersion7: + case DcGfxVersion8: + case DcGfxVersionUnknown: + default: + break; } } @@ -2737,14 +2777,18 @@ static enum surface_update_type get_plane_info_update_type(const struct dc *dc, return update_type; } -static enum surface_update_type get_scaling_info_update_type( - const struct dc *dc, +static struct surface_update_descriptor get_scaling_info_update_type( + const struct dc_check_config *check_config, const struct dc_surface_update *u) { union surface_update_flags *update_flags = &u->surface->update_flags; + struct surface_update_descriptor update_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (!u->scaling_info) - return UPDATE_TYPE_FAST; + return update_type; + + // `scaling_info` present means at least `STREAM` lock is required + elevate_update_type(&update_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); if (u->scaling_info->src_rect.width != u->surface->src_rect.width || u->scaling_info->src_rect.height != u->surface->src_rect.height @@ -2755,6 +2799,7 @@ static enum surface_update_type get_scaling_info_update_type( || u->scaling_info->scaling_quality.integer_scaling != u->surface->scaling_quality.integer_scaling) { update_flags->bits.scaling_change = 1; + elevate_update_type(&update_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); if (u->scaling_info->src_rect.width > u->surface->src_rect.width || u->scaling_info->src_rect.height > u->surface->src_rect.height) @@ -2768,7 +2813,7 @@ static enum surface_update_type get_scaling_info_update_type( /* Making dst rect smaller requires a bandwidth change */ update_flags->bits.bandwidth_change = 1; - if (u->scaling_info->src_rect.width > dc->caps.max_optimizable_video_width && + if (u->scaling_info->src_rect.width > check_config->max_optimizable_video_width && (u->scaling_info->clip_rect.width > u->surface->clip_rect.width || u->scaling_info->clip_rect.height > u->surface->clip_rect.height)) /* Changing clip size of a large surface may result in MPC slice count change */ @@ -2780,123 +2825,109 @@ static enum surface_update_type get_scaling_info_update_type( || u->scaling_info->clip_rect.x != u->surface->clip_rect.x || u->scaling_info->clip_rect.y != u->surface->clip_rect.y || u->scaling_info->dst_rect.x != u->surface->dst_rect.x - || u->scaling_info->dst_rect.y != u->surface->dst_rect.y) + || u->scaling_info->dst_rect.y != u->surface->dst_rect.y) { + elevate_update_type(&update_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); update_flags->bits.position_change = 1; + } - /* process every update flag before returning */ - if (update_flags->bits.clock_change - || update_flags->bits.bandwidth_change - || update_flags->bits.scaling_change) - return UPDATE_TYPE_FULL; - - if (update_flags->bits.position_change) - return UPDATE_TYPE_MED; - - return UPDATE_TYPE_FAST; + return update_type; } -static enum surface_update_type det_surface_update(const struct dc *dc, - const struct dc_surface_update *u) +static struct surface_update_descriptor det_surface_update( + const struct dc_check_config *check_config, + struct dc_surface_update *u) { - const struct dc_state *context = dc->current_state; - enum surface_update_type type; - enum surface_update_type overall_type = UPDATE_TYPE_FAST; + struct surface_update_descriptor overall_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; union surface_update_flags *update_flags = &u->surface->update_flags; - if (!is_surface_in_context(context, u->surface) || u->surface->force_full_update) { + if (u->surface->force_full_update) { update_flags->raw = 0xFFFFFFFF; - return UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); + return overall_type; } update_flags->raw = 0; // Reset all flags - type = get_plane_info_update_type(dc, u); - elevate_update_type(&overall_type, type); + struct surface_update_descriptor inner_type = get_plane_info_update_type(u); + + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); - type = get_scaling_info_update_type(dc, u); - elevate_update_type(&overall_type, type); + inner_type = get_scaling_info_update_type(check_config, u); + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); if (u->flip_addr) { update_flags->bits.addr_update = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + if (u->flip_addr->address.tmz_surface != u->surface->address.tmz_surface) { update_flags->bits.tmz_changed = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_FULL); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } } - if (u->in_transfer_func) + if (u->in_transfer_func) { update_flags->bits.in_transfer_func_change = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); + } - if (u->input_csc_color_matrix) + if (u->input_csc_color_matrix) { update_flags->bits.input_csc_change = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } - if (u->coeff_reduction_factor) + if (u->coeff_reduction_factor) { update_flags->bits.coeff_reduction_change = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } - if (u->gamut_remap_matrix) + if (u->gamut_remap_matrix) { update_flags->bits.gamut_remap_change = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } - if (u->blend_tf) + if (u->blend_tf || (u->gamma && dce_use_lut(u->plane_info ? u->plane_info->format : u->surface->format))) { update_flags->bits.gamma_change = 1; - - if (u->gamma) { - enum surface_pixel_format format = SURFACE_PIXEL_FORMAT_GRPH_BEGIN; - - if (u->plane_info) - format = u->plane_info->format; - else - format = u->surface->format; - - if (dce_use_lut(format)) - update_flags->bits.gamma_change = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); } - if (u->lut3d_func || u->func_shaper) + if (u->lut3d_func || u->func_shaper) { update_flags->bits.lut_3d = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } if (u->hdr_mult.value) if (u->hdr_mult.value != u->surface->hdr_mult.value) { + // TODO: Should be fast? update_flags->bits.hdr_mult = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_MED); + elevate_update_type(&overall_type, UPDATE_TYPE_MED, LOCK_DESCRIPTOR_STREAM); } if (u->sdr_white_level_nits) if (u->sdr_white_level_nits != u->surface->sdr_white_level_nits) { + // TODO: Should be fast? update_flags->bits.sdr_white_level_nits = 1; - elevate_update_type(&overall_type, UPDATE_TYPE_FULL); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (u->cm2_params) { - if ((u->cm2_params->component_settings.shaper_3dlut_setting - != u->surface->mcm_shaper_3dlut_setting) - || (u->cm2_params->component_settings.lut1d_enable - != u->surface->mcm_lut1d_enable)) - update_flags->bits.mcm_transfer_function_enable_change = 1; - if (u->cm2_params->cm2_luts.lut3d_data.lut3d_src - != u->surface->mcm_luts.lut3d_data.lut3d_src) + if (u->cm2_params->component_settings.shaper_3dlut_setting != u->surface->mcm_shaper_3dlut_setting + || u->cm2_params->component_settings.lut1d_enable != u->surface->mcm_lut1d_enable + || u->cm2_params->cm2_luts.lut3d_data.lut3d_src != u->surface->mcm_luts.lut3d_data.lut3d_src) { update_flags->bits.mcm_transfer_function_enable_change = 1; - } - if (update_flags->bits.in_transfer_func_change) { - type = UPDATE_TYPE_MED; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); + } } if (update_flags->bits.lut_3d && u->surface->mcm_luts.lut3d_data.lut3d_src != DC_CM2_TRANSFER_FUNC_SOURCE_VIDMEM) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); - } - if (update_flags->bits.mcm_transfer_function_enable_change) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } - if (dc->debug.enable_legacy_fast_update && + if (check_config->enable_legacy_fast_update && (update_flags->bits.gamma_change || update_flags->bits.gamut_remap_change || update_flags->bits.input_csc_change || update_flags->bits.coeff_reduction_change)) { - type = UPDATE_TYPE_FULL; - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } return overall_type; } @@ -2924,32 +2955,26 @@ static void force_immediate_gsl_plane_flip(struct dc *dc, struct dc_surface_upda } } -static enum surface_update_type check_update_surfaces_for_stream( - struct dc *dc, +static struct surface_update_descriptor check_update_surfaces_for_stream( + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status) + struct dc_stream_update *stream_update) { - int i; - enum surface_update_type overall_type = UPDATE_TYPE_FAST; - - if (dc->idle_optimizations_allowed || dc_can_clear_cursor_limit(dc)) - overall_type = UPDATE_TYPE_FULL; - - if (stream_status == NULL || stream_status->plane_count != surface_count) - overall_type = UPDATE_TYPE_FULL; + struct surface_update_descriptor overall_type = { UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_NONE }; if (stream_update && stream_update->pending_test_pattern) { - overall_type = UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } if (stream_update && stream_update->hw_cursor_req) { - overall_type = UPDATE_TYPE_FULL; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); } /* some stream updates require passive update */ if (stream_update) { + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + union stream_update_flags *su_flags = &stream_update->stream->update_flags; if ((stream_update->src.height != 0 && stream_update->src.width != 0) || @@ -2957,14 +2982,16 @@ static enum surface_update_type check_update_surfaces_for_stream( stream_update->integer_scaling_update) su_flags->bits.scaling = 1; - if (dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func) + if (check_config->enable_legacy_fast_update && stream_update->out_transfer_func) su_flags->bits.out_tf = 1; if (stream_update->abm_level) su_flags->bits.abm_level = 1; - if (stream_update->dpms_off) + if (stream_update->dpms_off) { su_flags->bits.dpms_off = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL | LOCK_DESCRIPTOR_LINK); + } if (stream_update->gamut_remap) su_flags->bits.gamut_remap = 1; @@ -2992,24 +3019,27 @@ static enum surface_update_type check_update_surfaces_for_stream( if (stream_update->output_color_space) su_flags->bits.out_csc = 1; - if (su_flags->raw != 0) - overall_type = UPDATE_TYPE_FULL; + // TODO: Make each elevation explicit, as to not override fast stream in crct_timing_adjust + if (su_flags->raw) + elevate_update_type(&overall_type, UPDATE_TYPE_FULL, LOCK_DESCRIPTOR_GLOBAL); - if (stream_update->output_csc_transform) + // Non-global cases + if (stream_update->output_csc_transform) { su_flags->bits.out_csc = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } - /* Output transfer function changes do not require bandwidth recalculation, - * so don't trigger a full update - */ - if (!dc->debug.enable_legacy_fast_update && stream_update->out_transfer_func) + if (!check_config->enable_legacy_fast_update && stream_update->out_transfer_func) { su_flags->bits.out_tf = 1; + elevate_update_type(&overall_type, UPDATE_TYPE_FAST, LOCK_DESCRIPTOR_STREAM); + } } - for (i = 0 ; i < surface_count; i++) { - enum surface_update_type type = - det_surface_update(dc, &updates[i]); + for (int i = 0 ; i < surface_count; i++) { + struct surface_update_descriptor inner_type = + det_surface_update(check_config, &updates[i]); - elevate_update_type(&overall_type, type); + elevate_update_type(&overall_type, inner_type.update_type, inner_type.lock_descriptor); } return overall_type; @@ -3020,44 +3050,18 @@ static enum surface_update_type check_update_surfaces_for_stream( * * See :c:type:`enum surface_update_type <surface_update_type>` for explanation of update types */ -enum surface_update_type dc_check_update_surfaces_for_stream( - struct dc *dc, +struct surface_update_descriptor dc_check_update_surfaces_for_stream( + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status) + struct dc_stream_update *stream_update) { - int i; - enum surface_update_type type; - if (stream_update) stream_update->stream->update_flags.raw = 0; - for (i = 0; i < surface_count; i++) + for (size_t i = 0; i < surface_count; i++) updates[i].surface->update_flags.raw = 0; - type = check_update_surfaces_for_stream(dc, updates, surface_count, stream_update, stream_status); - if (type == UPDATE_TYPE_FULL) { - if (stream_update) { - uint32_t dsc_changed = stream_update->stream->update_flags.bits.dsc_changed; - stream_update->stream->update_flags.raw = 0xFFFFFFFF; - stream_update->stream->update_flags.bits.dsc_changed = dsc_changed; - } - for (i = 0; i < surface_count; i++) - updates[i].surface->update_flags.raw = 0xFFFFFFFF; - } - - if (type == UPDATE_TYPE_FAST) { - // If there's an available clock comparator, we use that. - if (dc->clk_mgr->funcs->are_clock_states_equal) { - if (!dc->clk_mgr->funcs->are_clock_states_equal(&dc->clk_mgr->clks, &dc->current_state->bw_ctx.bw.dcn.clk)) - dc->optimized_required = true; - // Else we fallback to mem compare. - } else if (memcmp(&dc->current_state->bw_ctx.bw.dcn.clk, &dc->clk_mgr->clks, offsetof(struct dc_clocks, prev_p_state_change_support)) != 0) { - dc->optimized_required = true; - } - } - - return type; + return check_update_surfaces_for_stream(check_config, updates, surface_count, stream_update); } static struct dc_stream_status *stream_get_status( @@ -3388,7 +3392,11 @@ static void restore_planes_and_stream_state( for (i = 0; i < status->plane_count; i++) { dc_plane_copy_config(status->plane_states[i], &scratch->plane_states[i]); } + + // refcount is persistent + struct kref temp_refcount = stream->refcount; *stream = scratch->stream_state; + stream->refcount = temp_refcount; } /** @@ -3426,6 +3434,13 @@ static void update_seamless_boot_flags(struct dc *dc, } } +static bool full_update_required_weak( + const struct dc *dc, + const struct dc_surface_update *srf_updates, + int surface_count, + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream); + /** * update_planes_and_stream_state() - The function takes planes and stream * updates as inputs and determines the appropriate update type. If update type @@ -3472,7 +3487,10 @@ static bool update_planes_and_stream_state(struct dc *dc, context = dc->current_state; update_type = dc_check_update_surfaces_for_stream( - dc, srf_updates, surface_count, stream_update, stream_status); + &dc->check_config, srf_updates, surface_count, stream_update).update_type; + if (full_update_required_weak(dc, srf_updates, surface_count, stream_update, stream)) + update_type = UPDATE_TYPE_FULL; + /* It is possible to receive a flip for one plane while there are multiple flip_immediate planes in the same stream. * E.g. Desktop and MPO plane are flip_immediate but only the MPO plane received a flip * Force the other flip_immediate planes to flip so GSL doesn't wait for a flip that won't come. @@ -3504,6 +3522,16 @@ static bool update_planes_and_stream_state(struct dc *dc, } } + if (update_type == UPDATE_TYPE_FULL) { + if (stream_update) { + uint32_t dsc_changed = stream_update->stream->update_flags.bits.dsc_changed; + stream_update->stream->update_flags.raw = 0xFFFFFFFF; + stream_update->stream->update_flags.bits.dsc_changed = dsc_changed; + } + for (i = 0; i < surface_count; i++) + srf_updates[i].surface->update_flags.raw = 0xFFFFFFFF; + } + if (update_type >= update_surface_trace_level) update_surface_trace(dc, srf_updates, surface_count); @@ -4149,7 +4177,7 @@ static void commit_planes_for_stream(struct dc *dc, if ((update_type != UPDATE_TYPE_FAST) && stream->update_flags.bits.dsc_changed) if (top_pipe_to_program && top_pipe_to_program->stream_res.tg->funcs->lock_doublebuffer_enable) { - if (should_use_dmub_lock(stream->link)) { + if (should_use_dmub_inbox1_lock(dc, stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; @@ -4176,16 +4204,16 @@ static void commit_planes_for_stream(struct dc *dc, if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, true, should_lock_all_pipes, NULL, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, true); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, true); dc->hwss.interdependent_update_lock(dc, context, true); } else { if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, true, should_lock_all_pipes, top_pipe_to_program, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, true); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, true); /* Lock the top pipe while updating plane addrs, since freesync requires * plane addr update event triggers to be synchronized. @@ -4228,9 +4256,8 @@ static void commit_planes_for_stream(struct dc *dc, dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes, NULL, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, false); - + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, false); return; } @@ -4419,7 +4446,7 @@ static void commit_planes_for_stream(struct dc *dc, top_pipe_to_program->stream_res.tg, CRTC_STATE_VACTIVE); - if (should_use_dmub_lock(stream->link)) { + if (should_use_dmub_inbox1_lock(dc, stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; @@ -4467,13 +4494,13 @@ static void commit_planes_for_stream(struct dc *dc, if (should_lock_all_pipes && dc->hwss.interdependent_update_lock) { if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes, NULL, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, false); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, false); } else { if (dc->hwss.subvp_pipe_control_lock) dc->hwss.subvp_pipe_control_lock(dc, context, false, should_lock_all_pipes, top_pipe_to_program, subvp_prev_use); - if (dc->hwss.fams2_global_control_lock) - dc->hwss.fams2_global_control_lock(dc, context, false); + if (dc->hwss.dmub_hw_control_lock) + dc->hwss.dmub_hw_control_lock(dc, context, false); } // Fire manual trigger only when bottom plane is flipped @@ -4489,6 +4516,8 @@ static void commit_planes_for_stream(struct dc *dc, pipe_ctx->plane_state->skip_manual_trigger) continue; + if (dc->hwss.program_cursor_offload_now) + dc->hwss.program_cursor_offload_now(dc, pipe_ctx); if (pipe_ctx->stream_res.tg->funcs->program_manual_trigger) pipe_ctx->stream_res.tg->funcs->program_manual_trigger(pipe_ctx->stream_res.tg); } @@ -4994,7 +5023,7 @@ void populate_fast_updates(struct dc_fast_update *fast_update, } } -static bool fast_updates_exist(struct dc_fast_update *fast_update, int surface_count) +static bool fast_updates_exist(const struct dc_fast_update *fast_update, int surface_count) { int i; @@ -5035,18 +5064,44 @@ bool fast_nonaddr_updates_exist(struct dc_fast_update *fast_update, int surface_ return false; } -static bool full_update_required(struct dc *dc, - struct dc_surface_update *srf_updates, +static bool full_update_required_weak( + const struct dc *dc, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream) + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) { - - int i; - struct dc_stream_status *stream_status; const struct dc_state *context = dc->current_state; + if (srf_updates) + for (int i = 0; i < surface_count; i++) + if (!is_surface_in_context(context, srf_updates[i].surface)) + return true; - for (i = 0; i < surface_count; i++) { + if (stream) { + const struct dc_stream_status *stream_status = dc_stream_get_status_const(stream); + if (stream_status == NULL || stream_status->plane_count != surface_count) + return true; + } + if (dc->idle_optimizations_allowed) + return true; + + if (dc_can_clear_cursor_limit(dc)) + return true; + + return false; +} + +static bool full_update_required( + const struct dc *dc, + const struct dc_surface_update *srf_updates, + int surface_count, + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) +{ + if (full_update_required_weak(dc, srf_updates, surface_count, stream_update, stream)) + return true; + + for (int i = 0; i < surface_count; i++) { if (srf_updates && (srf_updates[i].plane_info || srf_updates[i].scaling_info || @@ -5062,8 +5117,7 @@ static bool full_update_required(struct dc *dc, srf_updates[i].flip_addr->address.tmz_surface != srf_updates[i].surface->address.tmz_surface) || (srf_updates[i].cm2_params && (srf_updates[i].cm2_params->component_settings.shaper_3dlut_setting != srf_updates[i].surface->mcm_shaper_3dlut_setting || - srf_updates[i].cm2_params->component_settings.lut1d_enable != srf_updates[i].surface->mcm_lut1d_enable)) || - !is_surface_in_context(context, srf_updates[i].surface))) + srf_updates[i].cm2_params->component_settings.lut1d_enable != srf_updates[i].surface->mcm_lut1d_enable)))) return true; } @@ -5099,26 +5153,16 @@ static bool full_update_required(struct dc *dc, stream_update->hw_cursor_req)) return true; - if (stream) { - stream_status = dc_stream_get_status(stream); - if (stream_status == NULL || stream_status->plane_count != surface_count) - return true; - } - if (dc->idle_optimizations_allowed) - return true; - - if (dc_can_clear_cursor_limit(dc)) - return true; - return false; } -static bool fast_update_only(struct dc *dc, - struct dc_fast_update *fast_update, - struct dc_surface_update *srf_updates, +static bool fast_update_only( + const struct dc *dc, + const struct dc_fast_update *fast_update, + const struct dc_surface_update *srf_updates, int surface_count, - struct dc_stream_update *stream_update, - struct dc_stream_state *stream) + const struct dc_stream_update *stream_update, + const struct dc_stream_state *stream) { return fast_updates_exist(fast_update, surface_count) && !full_update_required(dc, srf_updates, surface_count, stream_update, stream); @@ -5181,7 +5225,7 @@ static bool update_planes_and_stream_v2(struct dc *dc, commit_minimal_transition_state_in_dc_update(dc, context, stream, srf_updates, surface_count); - if (is_fast_update_only && !dc->debug.enable_legacy_fast_update) { + if (is_fast_update_only && !dc->check_config.enable_legacy_fast_update) { commit_planes_for_stream_fast(dc, srf_updates, surface_count, @@ -5224,7 +5268,7 @@ static void commit_planes_and_stream_update_on_current_context(struct dc *dc, stream_update); if (fast_update_only(dc, fast_update, srf_updates, surface_count, stream_update, stream) && - !dc->debug.enable_legacy_fast_update) + !dc->check_config.enable_legacy_fast_update) commit_planes_for_stream_fast(dc, srf_updates, surface_count, @@ -5350,7 +5394,8 @@ bool dc_update_planes_and_stream(struct dc *dc, * specially handle compatibility problems with transitions among those * features as they are now transparent to the new sequence. */ - if (dc->ctx->dce_version >= DCN_VERSION_4_01) + if (dc->ctx->dce_version >= DCN_VERSION_4_01 || dc->ctx->dce_version == DCN_VERSION_3_2 || + dc->ctx->dce_version == DCN_VERSION_3_21) ret = update_planes_and_stream_v3(dc, srf_updates, surface_count, stream, stream_update); else @@ -5935,6 +5980,101 @@ bool dc_process_dmub_aux_transfer_async(struct dc *dc, return true; } +bool dc_smart_power_oled_enable(const struct dc_link *link, bool enable, uint16_t peak_nits, + uint8_t debug_control, uint16_t fixed_CLL, uint32_t triggerline) +{ + bool status = false; + struct dc *dc = link->ctx->dc; + union dmub_rb_cmd cmd; + uint8_t otg_inst = 0; + unsigned int panel_inst = 0; + struct pipe_ctx *pipe_ctx = NULL; + struct resource_context *res_ctx = &link->ctx->dc->current_state->res_ctx; + int i = 0; + + // get panel_inst + if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) + return status; + + // get otg_inst + for (i = 0; i < MAX_PIPES; i++) { + if (res_ctx && + res_ctx->pipe_ctx[i].stream && + res_ctx->pipe_ctx[i].stream->link && + res_ctx->pipe_ctx[i].stream->link == link && + res_ctx->pipe_ctx[i].stream->link->connector_signal == SIGNAL_TYPE_EDP) { + pipe_ctx = &res_ctx->pipe_ctx[i]; + //TODO: refactor for multi edp support + break; + } + } + + if (pipe_ctx) + otg_inst = pipe_ctx->stream_res.tg->inst; + + // before enable smart power OLED, we need to call set pipe for DMUB to set ABM config + if (enable) { + if (dc->hwss.set_pipe && pipe_ctx) + dc->hwss.set_pipe(pipe_ctx); + } + + // fill in cmd + memset(&cmd, 0, sizeof(cmd)); + + cmd.smart_power_oled_enable.header.type = DMUB_CMD__SMART_POWER_OLED; + cmd.smart_power_oled_enable.header.sub_type = DMUB_CMD__SMART_POWER_OLED_ENABLE; + cmd.smart_power_oled_enable.header.payload_bytes = + sizeof(struct dmub_rb_cmd_smart_power_oled_enable_data) - sizeof(struct dmub_cmd_header); + cmd.smart_power_oled_enable.header.ret_status = 1; + cmd.smart_power_oled_enable.data.enable = enable; + cmd.smart_power_oled_enable.data.panel_inst = panel_inst; + cmd.smart_power_oled_enable.data.peak_nits = peak_nits; + cmd.smart_power_oled_enable.data.otg_inst = otg_inst; + cmd.smart_power_oled_enable.data.digfe_inst = link->link_enc->preferred_engine; + cmd.smart_power_oled_enable.data.digbe_inst = link->link_enc->transmitter; + + cmd.smart_power_oled_enable.data.debugcontrol = debug_control; + cmd.smart_power_oled_enable.data.triggerline = triggerline; + cmd.smart_power_oled_enable.data.fixed_max_cll = fixed_CLL; + + // send cmd + status = dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); + + return status; +} + +bool dc_smart_power_oled_get_max_cll(const struct dc_link *link, unsigned int *pCurrent_MaxCLL) +{ + struct dc *dc = link->ctx->dc; + union dmub_rb_cmd cmd; + bool status = false; + unsigned int panel_inst = 0; + + // get panel_inst + if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) + return status; + + // fill in cmd + memset(&cmd, 0, sizeof(cmd)); + + cmd.smart_power_oled_getmaxcll.header.type = DMUB_CMD__SMART_POWER_OLED; + cmd.smart_power_oled_getmaxcll.header.sub_type = DMUB_CMD__SMART_POWER_OLED_GETMAXCLL; + cmd.smart_power_oled_getmaxcll.header.payload_bytes = sizeof(cmd.smart_power_oled_getmaxcll.data); + cmd.smart_power_oled_getmaxcll.header.ret_status = 1; + + cmd.smart_power_oled_getmaxcll.data.input.panel_inst = panel_inst; + + // send cmd and wait for reply + status = dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT_WITH_REPLY); + + if (status) + *pCurrent_MaxCLL = cmd.smart_power_oled_getmaxcll.data.output.current_max_cll; + else + *pCurrent_MaxCLL = 0; + + return status; +} + uint8_t get_link_index_from_dpia_port_index(const struct dc *dc, uint8_t dpia_port_index) { @@ -6349,7 +6489,7 @@ bool dc_is_cursor_limit_pending(struct dc *dc) return false; } -bool dc_can_clear_cursor_limit(struct dc *dc) +bool dc_can_clear_cursor_limit(const struct dc *dc) { uint32_t i; @@ -6378,3 +6518,576 @@ void dc_get_underflow_debug_data_for_otg(struct dc *dc, int primary_otg_inst, if (dc->hwss.get_underflow_debug_data) dc->hwss.get_underflow_debug_data(dc, tg, out_data); } + +void dc_get_power_feature_status(struct dc *dc, int primary_otg_inst, + struct power_features *out_data) +{ + out_data->uclk_p_state = dc->current_state->clk_mgr->clks.p_state_change_support; + out_data->fams = dc->current_state->bw_ctx.bw.dcn.clk.fw_based_mclk_switching; +} + +bool dc_capture_register_software_state(struct dc *dc, struct dc_register_software_state *state) +{ + struct dc_state *context; + struct resource_context *res_ctx; + int i; + + if (!dc || !dc->current_state || !state) { + if (state) + state->state_valid = false; + return false; + } + + /* Initialize the state structure */ + memset(state, 0, sizeof(struct dc_register_software_state)); + + context = dc->current_state; + res_ctx = &context->res_ctx; + + /* Count active pipes and streams */ + state->active_pipe_count = 0; + state->active_stream_count = context->stream_count; + + for (i = 0; i < dc->res_pool->pipe_count; i++) { + if (res_ctx->pipe_ctx[i].stream) + state->active_pipe_count++; + } + + /* Capture HUBP programming state for each pipe */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + state->hubp[i].valid_stream = false; + if (!pipe_ctx->stream) + continue; + + state->hubp[i].valid_stream = true; + + /* HUBP register programming variables */ + if (pipe_ctx->stream_res.tg) + state->hubp[i].vtg_sel = pipe_ctx->stream_res.tg->inst; + + state->hubp[i].hubp_clock_enable = (pipe_ctx->plane_res.hubp != NULL) ? 1 : 0; + + state->hubp[i].valid_plane_state = false; + if (pipe_ctx->plane_state) { + state->hubp[i].valid_plane_state = true; + state->hubp[i].surface_pixel_format = pipe_ctx->plane_state->format; + state->hubp[i].rotation_angle = pipe_ctx->plane_state->rotation; + state->hubp[i].h_mirror_en = pipe_ctx->plane_state->horizontal_mirror ? 1 : 0; + + /* Surface size */ + if (pipe_ctx->plane_state->plane_size.surface_size.width > 0) { + state->hubp[i].surface_size_width = pipe_ctx->plane_state->plane_size.surface_size.width; + state->hubp[i].surface_size_height = pipe_ctx->plane_state->plane_size.surface_size.height; + } + + /* Viewport dimensions from scaler data */ + if (pipe_ctx->plane_state->src_rect.width > 0) { + state->hubp[i].pri_viewport_width = pipe_ctx->plane_state->src_rect.width; + state->hubp[i].pri_viewport_height = pipe_ctx->plane_state->src_rect.height; + state->hubp[i].pri_viewport_x_start = pipe_ctx->plane_state->src_rect.x; + state->hubp[i].pri_viewport_y_start = pipe_ctx->plane_state->src_rect.y; + } + + /* DCC settings */ + state->hubp[i].surface_dcc_en = (pipe_ctx->plane_state->dcc.enable) ? 1 : 0; + state->hubp[i].surface_dcc_ind_64b_blk = pipe_ctx->plane_state->dcc.independent_64b_blks; + state->hubp[i].surface_dcc_ind_128b_blk = pipe_ctx->plane_state->dcc.dcc_ind_blk; + + /* Surface pitch */ + state->hubp[i].surface_pitch = pipe_ctx->plane_state->plane_size.surface_pitch; + state->hubp[i].meta_pitch = pipe_ctx->plane_state->dcc.meta_pitch; + state->hubp[i].chroma_pitch = pipe_ctx->plane_state->plane_size.chroma_pitch; + state->hubp[i].meta_pitch_c = pipe_ctx->plane_state->dcc.meta_pitch_c; + + /* Surface addresses - primary */ + state->hubp[i].primary_surface_address_low = pipe_ctx->plane_state->address.grph.addr.low_part; + state->hubp[i].primary_surface_address_high = pipe_ctx->plane_state->address.grph.addr.high_part; + state->hubp[i].primary_meta_surface_address_low = pipe_ctx->plane_state->address.grph.meta_addr.low_part; + state->hubp[i].primary_meta_surface_address_high = pipe_ctx->plane_state->address.grph.meta_addr.high_part; + + /* TMZ settings */ + state->hubp[i].primary_surface_tmz = pipe_ctx->plane_state->address.tmz_surface; + state->hubp[i].primary_meta_surface_tmz = pipe_ctx->plane_state->address.tmz_surface; + + /* Tiling configuration */ + state->hubp[i].min_dc_gfx_version9 = false; + if (pipe_ctx->plane_state->tiling_info.gfxversion >= DcGfxVersion9) { + state->hubp[i].min_dc_gfx_version9 = true; + state->hubp[i].sw_mode = pipe_ctx->plane_state->tiling_info.gfx9.swizzle; + state->hubp[i].num_pipes = pipe_ctx->plane_state->tiling_info.gfx9.num_pipes; + state->hubp[i].num_banks = pipe_ctx->plane_state->tiling_info.gfx9.num_banks; + state->hubp[i].pipe_interleave = pipe_ctx->plane_state->tiling_info.gfx9.pipe_interleave; + state->hubp[i].num_shader_engines = pipe_ctx->plane_state->tiling_info.gfx9.num_shader_engines; + state->hubp[i].num_rb_per_se = pipe_ctx->plane_state->tiling_info.gfx9.num_rb_per_se; + state->hubp[i].num_pkrs = pipe_ctx->plane_state->tiling_info.gfx9.num_pkrs; + } + } + + /* DML Request Size Configuration */ + if (pipe_ctx->rq_regs.rq_regs_l.chunk_size > 0) { + state->hubp[i].rq_chunk_size = pipe_ctx->rq_regs.rq_regs_l.chunk_size; + state->hubp[i].rq_min_chunk_size = pipe_ctx->rq_regs.rq_regs_l.min_chunk_size; + state->hubp[i].rq_meta_chunk_size = pipe_ctx->rq_regs.rq_regs_l.meta_chunk_size; + state->hubp[i].rq_min_meta_chunk_size = pipe_ctx->rq_regs.rq_regs_l.min_meta_chunk_size; + state->hubp[i].rq_dpte_group_size = pipe_ctx->rq_regs.rq_regs_l.dpte_group_size; + state->hubp[i].rq_mpte_group_size = pipe_ctx->rq_regs.rq_regs_l.mpte_group_size; + state->hubp[i].rq_swath_height_l = pipe_ctx->rq_regs.rq_regs_l.swath_height; + state->hubp[i].rq_pte_row_height_l = pipe_ctx->rq_regs.rq_regs_l.pte_row_height_linear; + } + + /* Chroma request size configuration */ + if (pipe_ctx->rq_regs.rq_regs_c.chunk_size > 0) { + state->hubp[i].rq_chunk_size_c = pipe_ctx->rq_regs.rq_regs_c.chunk_size; + state->hubp[i].rq_min_chunk_size_c = pipe_ctx->rq_regs.rq_regs_c.min_chunk_size; + state->hubp[i].rq_meta_chunk_size_c = pipe_ctx->rq_regs.rq_regs_c.meta_chunk_size; + state->hubp[i].rq_min_meta_chunk_size_c = pipe_ctx->rq_regs.rq_regs_c.min_meta_chunk_size; + state->hubp[i].rq_dpte_group_size_c = pipe_ctx->rq_regs.rq_regs_c.dpte_group_size; + state->hubp[i].rq_mpte_group_size_c = pipe_ctx->rq_regs.rq_regs_c.mpte_group_size; + state->hubp[i].rq_swath_height_c = pipe_ctx->rq_regs.rq_regs_c.swath_height; + state->hubp[i].rq_pte_row_height_c = pipe_ctx->rq_regs.rq_regs_c.pte_row_height_linear; + } + + /* DML expansion modes */ + state->hubp[i].drq_expansion_mode = pipe_ctx->rq_regs.drq_expansion_mode; + state->hubp[i].prq_expansion_mode = pipe_ctx->rq_regs.prq_expansion_mode; + state->hubp[i].mrq_expansion_mode = pipe_ctx->rq_regs.mrq_expansion_mode; + state->hubp[i].crq_expansion_mode = pipe_ctx->rq_regs.crq_expansion_mode; + + /* DML DLG parameters - nominal */ + state->hubp[i].dst_y_per_vm_vblank = pipe_ctx->dlg_regs.dst_y_per_vm_vblank; + state->hubp[i].dst_y_per_row_vblank = pipe_ctx->dlg_regs.dst_y_per_row_vblank; + state->hubp[i].dst_y_per_vm_flip = pipe_ctx->dlg_regs.dst_y_per_vm_flip; + state->hubp[i].dst_y_per_row_flip = pipe_ctx->dlg_regs.dst_y_per_row_flip; + + /* DML prefetch settings */ + state->hubp[i].dst_y_prefetch = pipe_ctx->dlg_regs.dst_y_prefetch; + state->hubp[i].vratio_prefetch = pipe_ctx->dlg_regs.vratio_prefetch; + state->hubp[i].vratio_prefetch_c = pipe_ctx->dlg_regs.vratio_prefetch_c; + + /* TTU parameters */ + state->hubp[i].qos_level_low_wm = pipe_ctx->ttu_regs.qos_level_low_wm; + state->hubp[i].qos_level_high_wm = pipe_ctx->ttu_regs.qos_level_high_wm; + state->hubp[i].qos_level_flip = pipe_ctx->ttu_regs.qos_level_flip; + state->hubp[i].min_ttu_vblank = pipe_ctx->ttu_regs.min_ttu_vblank; + } + + /* Capture HUBBUB programming state */ + if (dc->res_pool->hubbub) { + /* Individual DET buffer sizes - software state variables that program DET registers */ + for (i = 0; i < 4 && i < dc->res_pool->pipe_count; i++) { + uint32_t det_size = res_ctx->pipe_ctx[i].det_buffer_size_kb; + switch (i) { + case 0: + state->hubbub.det0_size = det_size; + break; + case 1: + state->hubbub.det1_size = det_size; + break; + case 2: + state->hubbub.det2_size = det_size; + break; + case 3: + state->hubbub.det3_size = det_size; + break; + } + } + + /* Compression buffer configuration - software state that programs COMPBUF_SIZE register */ + // TODO: Handle logic for legacy DCN pre-DCN401 + state->hubbub.compbuf_size = context->bw_ctx.bw.dcn.arb_regs.compbuf_size; + } + + /* Capture DPP programming state for each pipe */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + if (!pipe_ctx->stream) + continue; + + state->dpp[i].dpp_clock_enable = (pipe_ctx->plane_res.dpp != NULL) ? 1 : 0; + + if (pipe_ctx->plane_state && pipe_ctx->plane_res.scl_data.recout.width > 0) { + /* Access dscl_prog_data directly - this contains the actual software state used for register programming */ + struct dscl_prog_data *dscl_data = &pipe_ctx->plane_res.scl_data.dscl_prog_data; + + /* Recout (Rectangle of Interest) configuration - software state that programs RECOUT registers */ + state->dpp[i].recout_start_x = dscl_data->recout.x; + state->dpp[i].recout_start_y = dscl_data->recout.y; + state->dpp[i].recout_width = dscl_data->recout.width; + state->dpp[i].recout_height = dscl_data->recout.height; + + /* MPC (Multiple Pipe/Plane Combiner) size - software state that programs MPC_SIZE registers */ + state->dpp[i].mpc_width = dscl_data->mpc_size.width; + state->dpp[i].mpc_height = dscl_data->mpc_size.height; + + /* DSCL mode - software state that programs SCL_MODE registers */ + state->dpp[i].dscl_mode = dscl_data->dscl_mode; + + /* Scaler ratios - software state that programs scale ratio registers (use actual programmed ratios) */ + state->dpp[i].horz_ratio_int = dscl_data->ratios.h_scale_ratio >> 19; // Extract integer part from programmed ratio + state->dpp[i].vert_ratio_int = dscl_data->ratios.v_scale_ratio >> 19; // Extract integer part from programmed ratio + + /* Basic scaler taps - software state that programs tap control registers (use actual programmed taps) */ + state->dpp[i].h_taps = dscl_data->taps.h_taps + 1; // dscl_prog_data.taps stores (taps - 1), so add 1 back + state->dpp[i].v_taps = dscl_data->taps.v_taps + 1; // dscl_prog_data.taps stores (taps - 1), so add 1 back + } + } + + /* Capture essential clock state for underflow analysis */ + if (dc->clk_mgr && dc->clk_mgr->clks.dispclk_khz > 0) { + /* Core display clocks affecting bandwidth and timing */ + state->dccg.dispclk_khz = dc->clk_mgr->clks.dispclk_khz; + + /* Per-pipe clock configuration - only capture what's essential */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + if (pipe_ctx->stream) { + /* Essential clocks that directly affect underflow risk */ + state->dccg.dppclk_khz[i] = dc->clk_mgr->clks.dppclk_khz; + state->dccg.pixclk_khz[i] = pipe_ctx->stream->timing.pix_clk_100hz / 10; + state->dccg.dppclk_enable[i] = 1; + + /* DP stream clock only for DP signals */ + if (pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT || + pipe_ctx->stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) { + state->dccg.dpstreamclk_enable[i] = 1; + } else { + state->dccg.dpstreamclk_enable[i] = 0; + } + } else { + /* Inactive pipe - no clocks */ + state->dccg.dppclk_khz[i] = 0; + state->dccg.pixclk_khz[i] = 0; + state->dccg.dppclk_enable[i] = 0; + if (i < 4) { + state->dccg.dpstreamclk_enable[i] = 0; + } + } + } + + /* DSC clock state - only when actually using DSC */ + for (i = 0; i < MAX_PIPES; i++) { + struct pipe_ctx *pipe_ctx = (i < dc->res_pool->pipe_count) ? &res_ctx->pipe_ctx[i] : NULL; + if (pipe_ctx && pipe_ctx->stream && pipe_ctx->stream->timing.dsc_cfg.num_slices_h > 0) { + state->dccg.dscclk_khz[i] = 400000; /* Typical DSC clock frequency */ + } else { + state->dccg.dscclk_khz[i] = 0; + } + } + + /* SYMCLK32 LE Control - only the essential HPO state for underflow analysis */ + for (i = 0; i < 2; i++) { + state->dccg.symclk32_le_enable[i] = 0; /* Default: disabled */ + } + + } + + /* Capture essential DSC configuration for underflow analysis */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + if (pipe_ctx->stream && pipe_ctx->stream->timing.dsc_cfg.num_slices_h > 0) { + /* DSC is enabled - capture essential configuration */ + state->dsc[i].dsc_clock_enable = 1; + + /* DSC configuration affecting bandwidth and timing */ + struct dc_dsc_config *dsc_cfg = &pipe_ctx->stream->timing.dsc_cfg; + state->dsc[i].dsc_num_slices_h = dsc_cfg->num_slices_h; + state->dsc[i].dsc_num_slices_v = dsc_cfg->num_slices_v; + state->dsc[i].dsc_bits_per_pixel = dsc_cfg->bits_per_pixel; + + /* OPP pipe source for DSC forwarding */ + if (pipe_ctx->stream_res.opp) { + state->dsc[i].dscrm_dsc_forward_enable = 1; + state->dsc[i].dscrm_dsc_opp_pipe_source = pipe_ctx->stream_res.opp->inst; + } else { + state->dsc[i].dscrm_dsc_forward_enable = 0; + state->dsc[i].dscrm_dsc_opp_pipe_source = 0; + } + } else { + /* DSC not enabled - clear all fields */ + memset(&state->dsc[i], 0, sizeof(state->dsc[i])); + } + } + + /* Capture MPC programming state - comprehensive register field coverage */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + if (pipe_ctx->plane_state && pipe_ctx->stream) { + struct dc_plane_state *plane_state = pipe_ctx->plane_state; + + /* MPCC blending tree and mode control - capture actual blend configuration */ + state->mpc.mpcc_mode[i] = (plane_state->blend_tf.type != TF_TYPE_BYPASS) ? 1 : 0; + state->mpc.mpcc_alpha_blend_mode[i] = plane_state->per_pixel_alpha ? 1 : 0; + state->mpc.mpcc_alpha_multiplied_mode[i] = plane_state->pre_multiplied_alpha ? 1 : 0; + state->mpc.mpcc_blnd_active_overlap_only[i] = 0; /* Default - no overlap restriction */ + state->mpc.mpcc_global_alpha[i] = plane_state->global_alpha_value; + state->mpc.mpcc_global_gain[i] = plane_state->global_alpha ? 255 : 0; + state->mpc.mpcc_bg_bpc[i] = 8; /* Standard 8-bit background */ + state->mpc.mpcc_bot_gain_mode[i] = 0; /* Standard gain mode */ + + /* MPCC blending tree connections - capture tree topology */ + if (pipe_ctx->bottom_pipe) { + state->mpc.mpcc_bot_sel[i] = pipe_ctx->bottom_pipe->pipe_idx; + } else { + state->mpc.mpcc_bot_sel[i] = 0xF; /* No bottom connection */ + } + state->mpc.mpcc_top_sel[i] = pipe_ctx->pipe_idx; /* This pipe's DPP ID */ + + /* MPCC output gamma control - capture gamma programming */ + if (plane_state->gamma_correction.type != GAMMA_CS_TFM_1D && plane_state->gamma_correction.num_entries > 0) { + state->mpc.mpcc_ogam_mode[i] = 1; /* Gamma enabled */ + state->mpc.mpcc_ogam_select[i] = 0; /* Bank A selection */ + state->mpc.mpcc_ogam_pwl_disable[i] = 0; /* PWL enabled */ + } else { + state->mpc.mpcc_ogam_mode[i] = 0; /* Bypass mode */ + state->mpc.mpcc_ogam_select[i] = 0; + state->mpc.mpcc_ogam_pwl_disable[i] = 1; /* PWL disabled */ + } + + /* MPCC pipe assignment and operational status */ + if (pipe_ctx->stream_res.opp) { + state->mpc.mpcc_opp_id[i] = pipe_ctx->stream_res.opp->inst; + } else { + state->mpc.mpcc_opp_id[i] = 0xF; /* No OPP assignment */ + } + + /* MPCC status indicators - active pipe state */ + state->mpc.mpcc_idle[i] = 0; /* Active pipe - not idle */ + state->mpc.mpcc_busy[i] = 1; /* Active pipe - busy processing */ + + } else { + /* Pipe not active - set disabled/idle state for all fields */ + state->mpc.mpcc_mode[i] = 0; + state->mpc.mpcc_alpha_blend_mode[i] = 0; + state->mpc.mpcc_alpha_multiplied_mode[i] = 0; + state->mpc.mpcc_blnd_active_overlap_only[i] = 0; + state->mpc.mpcc_global_alpha[i] = 0; + state->mpc.mpcc_global_gain[i] = 0; + state->mpc.mpcc_bg_bpc[i] = 0; + state->mpc.mpcc_bot_gain_mode[i] = 0; + state->mpc.mpcc_bot_sel[i] = 0xF; /* No bottom connection */ + state->mpc.mpcc_top_sel[i] = 0xF; /* No top connection */ + state->mpc.mpcc_ogam_mode[i] = 0; /* Bypass */ + state->mpc.mpcc_ogam_select[i] = 0; + state->mpc.mpcc_ogam_pwl_disable[i] = 1; /* PWL disabled */ + state->mpc.mpcc_opp_id[i] = 0xF; /* No OPP assignment */ + state->mpc.mpcc_idle[i] = 1; /* Idle */ + state->mpc.mpcc_busy[i] = 0; /* Not busy */ + } + } + + /* Capture OPP programming state for each pipe - comprehensive register field coverage */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + if (!pipe_ctx->stream) + continue; + + if (pipe_ctx->stream_res.opp) { + struct dc_crtc_timing *timing = &pipe_ctx->stream->timing; + + /* OPP Pipe Control */ + state->opp[i].opp_pipe_clock_enable = 1; /* Active pipe has clock enabled */ + + /* Display Pattern Generator (DPG) Control - 19 fields */ + if (pipe_ctx->stream->test_pattern.type != DP_TEST_PATTERN_VIDEO_MODE) { + state->opp[i].dpg_enable = 1; + } else { + /* Video mode - DPG disabled */ + state->opp[i].dpg_enable = 0; + } + + /* Format Control (FMT) - 18 fields */ + state->opp[i].fmt_pixel_encoding = timing->pixel_encoding; + + /* Chroma subsampling mode based on pixel encoding */ + if (timing->pixel_encoding == PIXEL_ENCODING_YCBCR420) { + state->opp[i].fmt_subsampling_mode = 1; /* 4:2:0 subsampling */ + } else if (timing->pixel_encoding == PIXEL_ENCODING_YCBCR422) { + state->opp[i].fmt_subsampling_mode = 2; /* 4:2:2 subsampling */ + } else { + state->opp[i].fmt_subsampling_mode = 0; /* No subsampling (4:4:4) */ + } + + state->opp[i].fmt_cbcr_bit_reduction_bypass = (timing->pixel_encoding == PIXEL_ENCODING_RGB) ? 1 : 0; + state->opp[i].fmt_stereosync_override = (timing->timing_3d_format != TIMING_3D_FORMAT_NONE) ? 1 : 0; + + /* Dithering control based on bit depth */ + if (timing->display_color_depth < COLOR_DEPTH_121212) { + state->opp[i].fmt_spatial_dither_frame_counter_max = 15; /* Typical frame counter max */ + state->opp[i].fmt_spatial_dither_frame_counter_bit_swap = 0; /* No bit swapping */ + state->opp[i].fmt_spatial_dither_enable = 1; + state->opp[i].fmt_spatial_dither_mode = 0; /* Spatial dithering mode */ + state->opp[i].fmt_spatial_dither_depth = timing->display_color_depth; + state->opp[i].fmt_temporal_dither_enable = 0; /* Spatial dithering preferred */ + } else { + state->opp[i].fmt_spatial_dither_frame_counter_max = 0; + state->opp[i].fmt_spatial_dither_frame_counter_bit_swap = 0; + state->opp[i].fmt_spatial_dither_enable = 0; + state->opp[i].fmt_spatial_dither_mode = 0; + state->opp[i].fmt_spatial_dither_depth = 0; + state->opp[i].fmt_temporal_dither_enable = 0; + } + + /* Truncation control for bit depth reduction */ + if (timing->display_color_depth < COLOR_DEPTH_121212) { + state->opp[i].fmt_truncate_enable = 1; + state->opp[i].fmt_truncate_depth = timing->display_color_depth; + state->opp[i].fmt_truncate_mode = 0; /* Round mode */ + } else { + state->opp[i].fmt_truncate_enable = 0; + state->opp[i].fmt_truncate_depth = 0; + state->opp[i].fmt_truncate_mode = 0; + } + + /* Data clamping control */ + state->opp[i].fmt_clamp_data_enable = 1; /* Clamping typically enabled */ + state->opp[i].fmt_clamp_color_format = timing->pixel_encoding; + + /* Dynamic expansion for limited range content */ + if (timing->pixel_encoding != PIXEL_ENCODING_RGB) { + state->opp[i].fmt_dynamic_exp_enable = 1; /* YCbCr typically needs expansion */ + state->opp[i].fmt_dynamic_exp_mode = 0; /* Standard expansion */ + } else { + state->opp[i].fmt_dynamic_exp_enable = 0; /* RGB typically full range */ + state->opp[i].fmt_dynamic_exp_mode = 0; + } + + /* Legacy field for compatibility */ + state->opp[i].fmt_bit_depth_control = timing->display_color_depth; + + /* Output Buffer (OPPBUF) Control - 6 fields */ + state->opp[i].oppbuf_active_width = timing->h_addressable; + state->opp[i].oppbuf_pixel_repetition = 0; /* No pixel repetition by default */ + + /* Multi-Stream Output (MSO) / ODM segmentation */ + if (pipe_ctx->next_odm_pipe) { + state->opp[i].oppbuf_display_segmentation = 1; /* Segmented display */ + state->opp[i].oppbuf_overlap_pixel_num = 0; /* ODM overlap pixels */ + } else { + state->opp[i].oppbuf_display_segmentation = 0; /* Single segment */ + state->opp[i].oppbuf_overlap_pixel_num = 0; + } + + /* 3D/Stereo control */ + if (timing->timing_3d_format != TIMING_3D_FORMAT_NONE) { + state->opp[i].oppbuf_3d_vact_space1_size = 30; /* Typical stereo blanking */ + state->opp[i].oppbuf_3d_vact_space2_size = 30; + } else { + state->opp[i].oppbuf_3d_vact_space1_size = 0; + state->opp[i].oppbuf_3d_vact_space2_size = 0; + } + + /* DSC Forward Config - 3 fields */ + if (timing->dsc_cfg.num_slices_h > 0) { + state->opp[i].dscrm_dsc_forward_enable = 1; + state->opp[i].dscrm_dsc_opp_pipe_source = pipe_ctx->stream_res.opp->inst; + state->opp[i].dscrm_dsc_forward_enable_status = 1; /* Status follows enable */ + } else { + state->opp[i].dscrm_dsc_forward_enable = 0; + state->opp[i].dscrm_dsc_opp_pipe_source = 0; + state->opp[i].dscrm_dsc_forward_enable_status = 0; + } + } else { + /* No OPP resource - set all fields to disabled state */ + memset(&state->opp[i], 0, sizeof(state->opp[i])); + } + } + + /* Capture OPTC programming state for each pipe - comprehensive register field coverage */ + for (i = 0; i < MAX_PIPES && i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &res_ctx->pipe_ctx[i]; + + if (!pipe_ctx->stream) + continue; + + if (pipe_ctx->stream_res.tg) { + struct dc_crtc_timing *timing = &pipe_ctx->stream->timing; + + state->optc[i].otg_master_inst = pipe_ctx->stream_res.tg->inst; + + /* OTG_CONTROL register - 5 fields */ + state->optc[i].otg_master_enable = 1; /* Active stream */ + state->optc[i].otg_disable_point_cntl = 0; /* Normal operation */ + state->optc[i].otg_start_point_cntl = 0; /* Normal start */ + state->optc[i].otg_field_number_cntl = (timing->flags.INTERLACE) ? 1 : 0; + state->optc[i].otg_out_mux = 0; /* Direct output */ + + /* OTG Horizontal Timing - 7 fields */ + state->optc[i].otg_h_total = timing->h_total; + state->optc[i].otg_h_blank_start = timing->h_addressable; + state->optc[i].otg_h_blank_end = timing->h_total - timing->h_front_porch; + state->optc[i].otg_h_sync_start = timing->h_addressable + timing->h_front_porch; + state->optc[i].otg_h_sync_end = timing->h_addressable + timing->h_front_porch + timing->h_sync_width; + state->optc[i].otg_h_sync_polarity = timing->flags.HSYNC_POSITIVE_POLARITY ? 0 : 1; + state->optc[i].otg_h_timing_div_mode = (pipe_ctx->next_odm_pipe) ? 1 : 0; /* ODM divide mode */ + + /* OTG Vertical Timing - 7 fields */ + state->optc[i].otg_v_total = timing->v_total; + state->optc[i].otg_v_blank_start = timing->v_addressable; + state->optc[i].otg_v_blank_end = timing->v_total - timing->v_front_porch; + state->optc[i].otg_v_sync_start = timing->v_addressable + timing->v_front_porch; + state->optc[i].otg_v_sync_end = timing->v_addressable + timing->v_front_porch + timing->v_sync_width; + state->optc[i].otg_v_sync_polarity = timing->flags.VSYNC_POSITIVE_POLARITY ? 0 : 1; + state->optc[i].otg_v_sync_mode = 0; /* Normal sync mode */ + + /* Initialize remaining core fields with appropriate defaults */ + // TODO: Update logic for accurate vtotal min/max + state->optc[i].otg_v_total_max = timing->v_total + 100; /* Typical DRR range */ + state->optc[i].otg_v_total_min = timing->v_total - 50; + state->optc[i].otg_v_total_mid = timing->v_total; + + /* ODM configuration */ + // TODO: Update logic to have complete ODM mappings (e.g. 3:1 and 4:1) stored in single pipe + if (pipe_ctx->next_odm_pipe) { + state->optc[i].optc_seg0_src_sel = pipe_ctx->stream_res.opp ? pipe_ctx->stream_res.opp->inst : 0; + state->optc[i].optc_seg1_src_sel = pipe_ctx->next_odm_pipe->stream_res.opp ? pipe_ctx->next_odm_pipe->stream_res.opp->inst : 0; + state->optc[i].optc_num_of_input_segment = 1; /* 2 segments - 1 */ + } else { + state->optc[i].optc_seg0_src_sel = pipe_ctx->stream_res.opp ? pipe_ctx->stream_res.opp->inst : 0; + state->optc[i].optc_seg1_src_sel = 0; + state->optc[i].optc_num_of_input_segment = 0; /* Single segment */ + } + + /* DSC configuration */ + if (timing->dsc_cfg.num_slices_h > 0) { + state->optc[i].optc_dsc_mode = 1; /* DSC enabled */ + state->optc[i].optc_dsc_bytes_per_pixel = timing->dsc_cfg.bits_per_pixel / 16; /* Convert to bytes */ + state->optc[i].optc_dsc_slice_width = timing->h_addressable / timing->dsc_cfg.num_slices_h; + } else { + state->optc[i].optc_dsc_mode = 0; + state->optc[i].optc_dsc_bytes_per_pixel = 0; + state->optc[i].optc_dsc_slice_width = 0; + } + + /* Essential control fields */ + state->optc[i].otg_stereo_enable = (timing->timing_3d_format != TIMING_3D_FORMAT_NONE) ? 1 : 0; + state->optc[i].otg_interlace_enable = timing->flags.INTERLACE ? 1 : 0; + state->optc[i].otg_clock_enable = 1; /* OTG clock enabled */ + state->optc[i].vtg0_enable = 1; /* VTG enabled for timing generation */ + + /* Initialize other key fields to defaults */ + state->optc[i].optc_input_pix_clk_en = 1; + state->optc[i].optc_segment_width = (pipe_ctx->next_odm_pipe) ? (timing->h_addressable / 2) : timing->h_addressable; + state->optc[i].otg_vready_offset = 1; + state->optc[i].otg_vstartup_start = timing->v_addressable + 10; + state->optc[i].otg_vupdate_offset = 0; + state->optc[i].otg_vupdate_width = 5; + } else { + /* No timing generator resource - initialize all fields to 0 */ + memset(&state->optc[i], 0, sizeof(state->optc[i])); + } + } + + state->state_valid = true; + return true; +} + +void dc_log_preos_dmcub_info(const struct dc *dc) +{ + dc_dmub_srv_log_preos_dmcub_info(dc->ctx->dmub_srv); +} diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index d82b1cb467f4..e2763b60482a 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -32,6 +32,13 @@ #include "resource.h" #include "dc_dmub_srv.h" #include "dc_state_priv.h" +#include "opp.h" +#include "dsc.h" +#include "dchubbub.h" +#include "dccg.h" +#include "abm.h" +#include "dcn10/dcn10_hubbub.h" +#include "dce/dmub_hw_lock_mgr.h" #define NUM_ELEMENTS(a) (sizeof(a) / sizeof((a)[0])) #define MAX_NUM_MCACHE 8 @@ -258,7 +265,7 @@ void color_space_to_black_color( black_color_format[BLACK_COLOR_FORMAT_RGB_LIMITED]; break; - /** + /* * Remove default and add case for all color space * so when we forget to add new color space * compiler will give a warning @@ -755,11 +762,13 @@ void hwss_build_fast_sequence(struct dc *dc, block_sequence[*num_steps].func = DMUB_SUBVP_PIPE_CONTROL_LOCK_FAST; (*num_steps)++; } - if (dc->hwss.fams2_global_control_lock_fast) { - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.dc = dc; - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.lock = true; - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.is_required = dc_state_is_fams2_in_use(dc, context); - block_sequence[*num_steps].func = DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST; + if (dc->hwss.dmub_hw_control_lock_fast) { + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.dc = dc; + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.lock = true; + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.is_required = + dc_state_is_fams2_in_use(dc, context) || + dmub_hw_lock_mgr_does_link_require_lock(dc, stream->link); + block_sequence[*num_steps].func = DMUB_HW_CONTROL_LOCK_FAST; (*num_steps)++; } if (dc->hwss.pipe_control_lock) { @@ -784,7 +793,7 @@ void hwss_build_fast_sequence(struct dc *dc, while (current_mpc_pipe) { if (current_mpc_pipe->plane_state) { if (dc->hwss.set_flip_control_gsl && current_mpc_pipe->plane_state->update_flags.raw) { - block_sequence[*num_steps].params.set_flip_control_gsl_params.pipe_ctx = current_mpc_pipe; + block_sequence[*num_steps].params.set_flip_control_gsl_params.hubp = current_mpc_pipe->plane_res.hubp; block_sequence[*num_steps].params.set_flip_control_gsl_params.flip_immediate = current_mpc_pipe->plane_state->flip_immediate; block_sequence[*num_steps].func = HUBP_SET_FLIP_CONTROL_GSL; (*num_steps)++; @@ -894,11 +903,11 @@ void hwss_build_fast_sequence(struct dc *dc, block_sequence[*num_steps].func = DMUB_SUBVP_PIPE_CONTROL_LOCK_FAST; (*num_steps)++; } - if (dc->hwss.fams2_global_control_lock_fast) { - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.dc = dc; - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.lock = false; - block_sequence[*num_steps].params.fams2_global_control_lock_fast_params.is_required = dc_state_is_fams2_in_use(dc, context); - block_sequence[*num_steps].func = DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST; + if (dc->hwss.dmub_hw_control_lock_fast) { + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.dc = dc; + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.lock = false; + block_sequence[*num_steps].params.dmub_hw_control_lock_fast_params.is_required = dc_state_is_fams2_in_use(dc, context); + block_sequence[*num_steps].func = DMUB_HW_CONTROL_LOCK_FAST; (*num_steps)++; } @@ -911,6 +920,13 @@ void hwss_build_fast_sequence(struct dc *dc, current_mpc_pipe->stream && current_mpc_pipe->plane_state && current_mpc_pipe->plane_state->update_flags.bits.addr_update && !current_mpc_pipe->plane_state->skip_manual_trigger) { + if (dc->hwss.program_cursor_offload_now) { + block_sequence[*num_steps].params.program_cursor_update_now_params.dc = dc; + block_sequence[*num_steps].params.program_cursor_update_now_params.pipe_ctx = current_mpc_pipe; + block_sequence[*num_steps].func = PROGRAM_CURSOR_UPDATE_NOW; + (*num_steps)++; + } + block_sequence[*num_steps].params.program_manual_trigger_params.pipe_ctx = current_mpc_pipe; block_sequence[*num_steps].func = OPTC_PROGRAM_MANUAL_TRIGGER; (*num_steps)++; @@ -942,8 +958,9 @@ void hwss_execute_sequence(struct dc *dc, params->pipe_control_lock_params.lock); break; case HUBP_SET_FLIP_CONTROL_GSL: - dc->hwss.set_flip_control_gsl(params->set_flip_control_gsl_params.pipe_ctx, - params->set_flip_control_gsl_params.flip_immediate); + params->set_flip_control_gsl_params.hubp->funcs->hubp_set_flip_control_surface_gsl( + params->set_flip_control_gsl_params.hubp, + params->set_flip_control_gsl_params.flip_immediate); break; case HUBP_PROGRAM_TRIPLEBUFFER: dc->hwss.program_triplebuffer(params->program_triplebuffer_params.dc, @@ -1001,8 +1018,301 @@ void hwss_execute_sequence(struct dc *dc, params->wait_for_dcc_meta_propagation_params.dc, params->wait_for_dcc_meta_propagation_params.top_pipe_to_program); break; - case DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST: - dc->hwss.fams2_global_control_lock_fast(params); + case DMUB_HW_CONTROL_LOCK_FAST: + dc->hwss.dmub_hw_control_lock_fast(params); + break; + case HUBP_PROGRAM_SURFACE_CONFIG: + hwss_program_surface_config(params); + break; + case HUBP_PROGRAM_MCACHE_ID: + hwss_program_mcache_id_and_split_coordinate(params); + break; + case PROGRAM_CURSOR_UPDATE_NOW: + dc->hwss.program_cursor_offload_now( + params->program_cursor_update_now_params.dc, + params->program_cursor_update_now_params.pipe_ctx); + break; + case HUBP_WAIT_PIPE_READ_START: + params->hubp_wait_pipe_read_start_params.hubp->funcs->hubp_wait_pipe_read_start( + params->hubp_wait_pipe_read_start_params.hubp); + break; + case HWS_APPLY_UPDATE_FLAGS_FOR_PHANTOM: + dc->hwss.apply_update_flags_for_phantom(params->apply_update_flags_for_phantom_params.pipe_ctx); + break; + case HWS_UPDATE_PHANTOM_VP_POSITION: + dc->hwss.update_phantom_vp_position(params->update_phantom_vp_position_params.dc, + params->update_phantom_vp_position_params.context, + params->update_phantom_vp_position_params.pipe_ctx); + break; + case OPTC_SET_ODM_COMBINE: + hwss_set_odm_combine(params); + break; + case OPTC_SET_ODM_BYPASS: + hwss_set_odm_bypass(params); + break; + case OPP_PIPE_CLOCK_CONTROL: + hwss_opp_pipe_clock_control(params); + break; + case OPP_PROGRAM_LEFT_EDGE_EXTRA_PIXEL: + hwss_opp_program_left_edge_extra_pixel(params); + break; + case DCCG_SET_DTO_DSCCLK: + hwss_dccg_set_dto_dscclk(params); + break; + case DSC_SET_CONFIG: + hwss_dsc_set_config(params); + break; + case DSC_ENABLE: + hwss_dsc_enable(params); + break; + case TG_SET_DSC_CONFIG: + hwss_tg_set_dsc_config(params); + break; + case DSC_DISCONNECT: + hwss_dsc_disconnect(params); + break; + case DSC_READ_STATE: + hwss_dsc_read_state(params); + break; + case DSC_CALCULATE_AND_SET_CONFIG: + hwss_dsc_calculate_and_set_config(params); + break; + case DSC_ENABLE_WITH_OPP: + hwss_dsc_enable_with_opp(params); + break; + case TG_PROGRAM_GLOBAL_SYNC: + hwss_tg_program_global_sync(params); + break; + case TG_WAIT_FOR_STATE: + hwss_tg_wait_for_state(params); + break; + case TG_SET_VTG_PARAMS: + hwss_tg_set_vtg_params(params); + break; + case TG_SETUP_VERTICAL_INTERRUPT2: + hwss_tg_setup_vertical_interrupt2(params); + break; + case DPP_SET_HDR_MULTIPLIER: + hwss_dpp_set_hdr_multiplier(params); + break; + case HUBP_PROGRAM_DET_SIZE: + hwss_program_det_size(params); + break; + case HUBP_PROGRAM_DET_SEGMENTS: + hwss_program_det_segments(params); + break; + case OPP_SET_DYN_EXPANSION: + hwss_opp_set_dyn_expansion(params); + break; + case OPP_PROGRAM_FMT: + hwss_opp_program_fmt(params); + break; + case OPP_PROGRAM_BIT_DEPTH_REDUCTION: + hwss_opp_program_bit_depth_reduction(params); + break; + case OPP_SET_DISP_PATTERN_GENERATOR: + hwss_opp_set_disp_pattern_generator(params); + break; + case ABM_SET_PIPE: + hwss_set_abm_pipe(params); + break; + case ABM_SET_LEVEL: + hwss_set_abm_level(params); + break; + case ABM_SET_IMMEDIATE_DISABLE: + hwss_set_abm_immediate_disable(params); + break; + case MPC_REMOVE_MPCC: + hwss_mpc_remove_mpcc(params); + break; + case OPP_SET_MPCC_DISCONNECT_PENDING: + hwss_opp_set_mpcc_disconnect_pending(params); + break; + case DC_SET_OPTIMIZED_REQUIRED: + hwss_dc_set_optimized_required(params); + break; + case HUBP_DISCONNECT: + hwss_hubp_disconnect(params); + break; + case HUBBUB_FORCE_PSTATE_CHANGE_CONTROL: + hwss_hubbub_force_pstate_change_control(params); + break; + case TG_ENABLE_CRTC: + hwss_tg_enable_crtc(params); + break; + case TG_SET_GSL: + hwss_tg_set_gsl(params); + break; + case TG_SET_GSL_SOURCE_SELECT: + hwss_tg_set_gsl_source_select(params); + break; + case HUBP_WAIT_FLIP_PENDING: + hwss_hubp_wait_flip_pending(params); + break; + case TG_WAIT_DOUBLE_BUFFER_PENDING: + hwss_tg_wait_double_buffer_pending(params); + break; + case UPDATE_FORCE_PSTATE: + hwss_update_force_pstate(params); + break; + case HUBBUB_APPLY_DEDCN21_147_WA: + hwss_hubbub_apply_dedcn21_147_wa(params); + break; + case HUBBUB_ALLOW_SELF_REFRESH_CONTROL: + hwss_hubbub_allow_self_refresh_control(params); + break; + case TG_GET_FRAME_COUNT: + hwss_tg_get_frame_count(params); + break; + case MPC_SET_DWB_MUX: + hwss_mpc_set_dwb_mux(params); + break; + case MPC_DISABLE_DWB_MUX: + hwss_mpc_disable_dwb_mux(params); + break; + case MCIF_WB_CONFIG_BUF: + hwss_mcif_wb_config_buf(params); + break; + case MCIF_WB_CONFIG_ARB: + hwss_mcif_wb_config_arb(params); + break; + case MCIF_WB_ENABLE: + hwss_mcif_wb_enable(params); + break; + case MCIF_WB_DISABLE: + hwss_mcif_wb_disable(params); + break; + case DWBC_ENABLE: + hwss_dwbc_enable(params); + break; + case DWBC_DISABLE: + hwss_dwbc_disable(params); + break; + case DWBC_UPDATE: + hwss_dwbc_update(params); + break; + case HUBP_UPDATE_MALL_SEL: + hwss_hubp_update_mall_sel(params); + break; + case HUBP_PREPARE_SUBVP_BUFFERING: + hwss_hubp_prepare_subvp_buffering(params); + break; + case HUBP_SET_BLANK_EN: + hwss_hubp_set_blank_en(params); + break; + case HUBP_DISABLE_CONTROL: + hwss_hubp_disable_control(params); + break; + case HUBBUB_SOFT_RESET: + hwss_hubbub_soft_reset(params); + break; + case HUBP_CLK_CNTL: + hwss_hubp_clk_cntl(params); + break; + case HUBP_INIT: + hwss_hubp_init(params); + break; + case HUBP_SET_VM_SYSTEM_APERTURE_SETTINGS: + hwss_hubp_set_vm_system_aperture_settings(params); + break; + case HUBP_SET_FLIP_INT: + hwss_hubp_set_flip_int(params); + break; + case DPP_DPPCLK_CONTROL: + hwss_dpp_dppclk_control(params); + break; + case DISABLE_PHANTOM_CRTC: + hwss_disable_phantom_crtc(params); + break; + case DSC_PG_STATUS: + hwss_dsc_pg_status(params); + break; + case DSC_WAIT_DISCONNECT_PENDING_CLEAR: + hwss_dsc_wait_disconnect_pending_clear(params); + break; + case DSC_DISABLE: + hwss_dsc_disable(params); + break; + case DCCG_SET_REF_DSCCLK: + hwss_dccg_set_ref_dscclk(params); + break; + case DPP_PG_CONTROL: + hwss_dpp_pg_control(params); + break; + case HUBP_PG_CONTROL: + hwss_hubp_pg_control(params); + break; + case HUBP_RESET: + hwss_hubp_reset(params); + break; + case DPP_RESET: + hwss_dpp_reset(params); + break; + case DPP_ROOT_CLOCK_CONTROL: + hwss_dpp_root_clock_control(params); + break; + case DC_IP_REQUEST_CNTL: + hwss_dc_ip_request_cntl(params); + break; + case DCCG_UPDATE_DPP_DTO: + hwss_dccg_update_dpp_dto(params); + break; + case HUBP_VTG_SEL: + hwss_hubp_vtg_sel(params); + break; + case HUBP_SETUP2: + hwss_hubp_setup2(params); + break; + case HUBP_SETUP: + hwss_hubp_setup(params); + break; + case HUBP_SET_UNBOUNDED_REQUESTING: + hwss_hubp_set_unbounded_requesting(params); + break; + case HUBP_SETUP_INTERDEPENDENT2: + hwss_hubp_setup_interdependent2(params); + break; + case HUBP_SETUP_INTERDEPENDENT: + hwss_hubp_setup_interdependent(params); + break; + case DPP_SET_CURSOR_MATRIX: + hwss_dpp_set_cursor_matrix(params); + break; + case MPC_UPDATE_BLENDING: + hwss_mpc_update_blending(params); + break; + case MPC_ASSERT_IDLE_MPCC: + hwss_mpc_assert_idle_mpcc(params); + break; + case MPC_INSERT_PLANE: + hwss_mpc_insert_plane(params); + break; + case DPP_SET_SCALER: + hwss_dpp_set_scaler(params); + break; + case HUBP_MEM_PROGRAM_VIEWPORT: + hwss_hubp_mem_program_viewport(params); + break; + case ABORT_CURSOR_OFFLOAD_UPDATE: + hwss_abort_cursor_offload_update(params); + break; + case SET_CURSOR_ATTRIBUTE: + hwss_set_cursor_attribute(params); + break; + case SET_CURSOR_POSITION: + hwss_set_cursor_position(params); + break; + case SET_CURSOR_SDR_WHITE_LEVEL: + hwss_set_cursor_sdr_white_level(params); + break; + case PROGRAM_OUTPUT_CSC: + hwss_program_output_csc(params); + break; + case HUBP_SET_BLANK: + hwss_hubp_set_blank(params); + break; + case PHANTOM_HUBP_POST_ENABLE: + hwss_phantom_hubp_post_enable(params); break; default: ASSERT(false); @@ -1011,6 +1321,338 @@ void hwss_execute_sequence(struct dc *dc, } } +/* + * Helper function to add OPTC pipe control lock to block sequence + */ +void hwss_add_optc_pipe_control_lock(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool lock) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.pipe_control_lock_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.pipe_control_lock_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.pipe_control_lock_params.lock = lock; + seq_state->steps[*seq_state->num_steps].func = OPTC_PIPE_CONTROL_LOCK; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP set flip control GSL to block sequence + */ +void hwss_add_hubp_set_flip_control_gsl(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool flip_immediate) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_flip_control_gsl_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.set_flip_control_gsl_params.flip_immediate = flip_immediate; + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_FLIP_CONTROL_GSL; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP program triplebuffer to block sequence + */ +void hwss_add_hubp_program_triplebuffer(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool enableTripleBuffer) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_triplebuffer_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.program_triplebuffer_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.program_triplebuffer_params.enableTripleBuffer = enableTripleBuffer; + seq_state->steps[*seq_state->num_steps].func = HUBP_PROGRAM_TRIPLEBUFFER; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP update plane address to block sequence + */ +void hwss_add_hubp_update_plane_addr(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.update_plane_addr_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.update_plane_addr_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = HUBP_UPDATE_PLANE_ADDR; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DPP set input transfer function to block sequence + */ +void hwss_add_dpp_set_input_transfer_func(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_plane_state *plane_state) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_input_transfer_func_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_input_transfer_func_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.set_input_transfer_func_params.plane_state = plane_state; + seq_state->steps[*seq_state->num_steps].func = DPP_SET_INPUT_TRANSFER_FUNC; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DPP program gamut remap to block sequence + */ +void hwss_add_dpp_program_gamut_remap(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_gamut_remap_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = DPP_PROGRAM_GAMUT_REMAP; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DPP program bias and scale to block sequence + */ +void hwss_add_dpp_program_bias_and_scale(struct block_sequence_state *seq_state, struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_bias_and_scale_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = DPP_PROGRAM_BIAS_AND_SCALE; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add OPTC program manual trigger to block sequence + */ +void hwss_add_optc_program_manual_trigger(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_manual_trigger_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = OPTC_PROGRAM_MANUAL_TRIGGER; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DPP set output transfer function to block sequence + */ +void hwss_add_dpp_set_output_transfer_func(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_stream_state *stream) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_output_transfer_func_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_output_transfer_func_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.set_output_transfer_func_params.stream = stream; + seq_state->steps[*seq_state->num_steps].func = DPP_SET_OUTPUT_TRANSFER_FUNC; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC update visual confirm to block sequence + */ +void hwss_add_mpc_update_visual_confirm(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + int mpcc_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.update_visual_confirm_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.update_visual_confirm_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.update_visual_confirm_params.mpcc_id = mpcc_id; + seq_state->steps[*seq_state->num_steps].func = MPC_UPDATE_VISUAL_CONFIRM; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC power on MPC mem PWR to block sequence + */ +void hwss_add_mpc_power_on_mpc_mem_pwr(struct block_sequence_state *seq_state, + struct mpc *mpc, + int mpcc_id, + bool power_on) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.power_on_mpc_mem_pwr_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.power_on_mpc_mem_pwr_params.mpcc_id = mpcc_id; + seq_state->steps[*seq_state->num_steps].params.power_on_mpc_mem_pwr_params.power_on = power_on; + seq_state->steps[*seq_state->num_steps].func = MPC_POWER_ON_MPC_MEM_PWR; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC set output CSC to block sequence + */ +void hwss_add_mpc_set_output_csc(struct block_sequence_state *seq_state, + struct mpc *mpc, + int opp_id, + const uint16_t *regval, + enum mpc_output_csc_mode ocsc_mode) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_output_csc_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.set_output_csc_params.opp_id = opp_id; + seq_state->steps[*seq_state->num_steps].params.set_output_csc_params.regval = regval; + seq_state->steps[*seq_state->num_steps].params.set_output_csc_params.ocsc_mode = ocsc_mode; + seq_state->steps[*seq_state->num_steps].func = MPC_SET_OUTPUT_CSC; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC set OCSC default to block sequence + */ +void hwss_add_mpc_set_ocsc_default(struct block_sequence_state *seq_state, + struct mpc *mpc, + int opp_id, + enum dc_color_space colorspace, + enum mpc_output_csc_mode ocsc_mode) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_ocsc_default_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.set_ocsc_default_params.opp_id = opp_id; + seq_state->steps[*seq_state->num_steps].params.set_ocsc_default_params.color_space = colorspace; + seq_state->steps[*seq_state->num_steps].params.set_ocsc_default_params.ocsc_mode = ocsc_mode; + seq_state->steps[*seq_state->num_steps].func = MPC_SET_OCSC_DEFAULT; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DMUB send DMCUB command to block sequence + */ +void hwss_add_dmub_send_dmcub_cmd(struct block_sequence_state *seq_state, + struct dc_context *ctx, + union dmub_rb_cmd *cmd, + enum dm_dmub_wait_type wait_type) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.send_dmcub_cmd_params.ctx = ctx; + seq_state->steps[*seq_state->num_steps].params.send_dmcub_cmd_params.cmd = cmd; + seq_state->steps[*seq_state->num_steps].params.send_dmcub_cmd_params.wait_type = wait_type; + seq_state->steps[*seq_state->num_steps].func = DMUB_SEND_DMCUB_CMD; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DMUB SubVP save surface address to block sequence + */ +void hwss_add_dmub_subvp_save_surf_addr(struct block_sequence_state *seq_state, + struct dc_dmub_srv *dc_dmub_srv, + struct dc_plane_address *addr, + uint8_t subvp_index) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.subvp_save_surf_addr.dc_dmub_srv = dc_dmub_srv; + seq_state->steps[*seq_state->num_steps].params.subvp_save_surf_addr.addr = addr; + seq_state->steps[*seq_state->num_steps].params.subvp_save_surf_addr.subvp_index = subvp_index; + seq_state->steps[*seq_state->num_steps].func = DMUB_SUBVP_SAVE_SURF_ADDR; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP wait for DCC meta propagation to block sequence + */ +void hwss_add_hubp_wait_for_dcc_meta_prop(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *top_pipe_to_program) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.wait_for_dcc_meta_propagation_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.wait_for_dcc_meta_propagation_params.top_pipe_to_program = top_pipe_to_program; + seq_state->steps[*seq_state->num_steps].func = HUBP_WAIT_FOR_DCC_META_PROP; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP wait pipe read start to block sequence + */ +void hwss_add_hubp_wait_pipe_read_start(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.hubp_wait_pipe_read_start_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].func = HUBP_WAIT_PIPE_READ_START; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HWS apply update flags for phantom to block sequence + */ +void hwss_add_hws_apply_update_flags_for_phantom(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.apply_update_flags_for_phantom_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = HWS_APPLY_UPDATE_FLAGS_FOR_PHANTOM; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HWS update phantom VP position to block sequence + */ +void hwss_add_hws_update_phantom_vp_position(struct block_sequence_state *seq_state, + struct dc *dc, + struct dc_state *context, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.update_phantom_vp_position_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.update_phantom_vp_position_params.context = context; + seq_state->steps[*seq_state->num_steps].params.update_phantom_vp_position_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = HWS_UPDATE_PHANTOM_VP_POSITION; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add OPTC set ODM combine to block sequence + */ +void hwss_add_optc_set_odm_combine(struct block_sequence_state *seq_state, + struct timing_generator *tg, int opp_inst[MAX_PIPES], int opp_head_count, + int odm_slice_width, int last_odm_slice_width) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_odm_combine_params.tg = tg; + memcpy(seq_state->steps[*seq_state->num_steps].params.set_odm_combine_params.opp_inst, opp_inst, sizeof(int) * MAX_PIPES); + seq_state->steps[*seq_state->num_steps].params.set_odm_combine_params.opp_head_count = opp_head_count; + seq_state->steps[*seq_state->num_steps].params.set_odm_combine_params.odm_slice_width = odm_slice_width; + seq_state->steps[*seq_state->num_steps].params.set_odm_combine_params.last_odm_slice_width = last_odm_slice_width; + seq_state->steps[*seq_state->num_steps].func = OPTC_SET_ODM_COMBINE; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add OPTC set ODM bypass to block sequence + */ +void hwss_add_optc_set_odm_bypass(struct block_sequence_state *seq_state, + struct timing_generator *tg, struct dc_crtc_timing *timing) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_odm_bypass_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.set_odm_bypass_params.timing = timing; + seq_state->steps[*seq_state->num_steps].func = OPTC_SET_ODM_BYPASS; + (*seq_state->num_steps)++; + } +} + void hwss_send_dmcub_cmd(union block_sequence_params *params) { struct dc_context *ctx = params->send_dmcub_cmd_params.ctx; @@ -1020,6 +1662,276 @@ void hwss_send_dmcub_cmd(union block_sequence_params *params) dc_wake_and_execute_dmub_cmd(ctx, cmd, wait_type); } +/* + * Helper function to add TG program global sync to block sequence + */ +void hwss_add_tg_program_global_sync(struct block_sequence_state *seq_state, + struct timing_generator *tg, + int vready_offset, + unsigned int vstartup_lines, + unsigned int vupdate_offset_pixels, + unsigned int vupdate_vupdate_width_pixels, + unsigned int pstate_keepout_start_lines) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.vready_offset = vready_offset; + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.vstartup_lines = vstartup_lines; + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.vupdate_offset_pixels = vupdate_offset_pixels; + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.vupdate_vupdate_width_pixels = vupdate_vupdate_width_pixels; + seq_state->steps[*seq_state->num_steps].params.tg_program_global_sync_params.pstate_keepout_start_lines = pstate_keepout_start_lines; + seq_state->steps[*seq_state->num_steps].func = TG_PROGRAM_GLOBAL_SYNC; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add TG wait for state to block sequence + */ +void hwss_add_tg_wait_for_state(struct block_sequence_state *seq_state, + struct timing_generator *tg, + enum crtc_state state) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_wait_for_state_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_wait_for_state_params.state = state; + seq_state->steps[*seq_state->num_steps].func = TG_WAIT_FOR_STATE; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add TG set VTG params to block sequence + */ +void hwss_add_tg_set_vtg_params(struct block_sequence_state *seq_state, + struct timing_generator *tg, + struct dc_crtc_timing *dc_crtc_timing, + bool program_fp2) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_set_vtg_params_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_set_vtg_params_params.timing = dc_crtc_timing; + seq_state->steps[*seq_state->num_steps].params.tg_set_vtg_params_params.program_fp2 = program_fp2; + seq_state->steps[*seq_state->num_steps].func = TG_SET_VTG_PARAMS; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add TG setup vertical interrupt2 to block sequence + */ +void hwss_add_tg_setup_vertical_interrupt2(struct block_sequence_state *seq_state, + struct timing_generator *tg, int start_line) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_setup_vertical_interrupt2_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_setup_vertical_interrupt2_params.start_line = start_line; + seq_state->steps[*seq_state->num_steps].func = TG_SETUP_VERTICAL_INTERRUPT2; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add DPP set HDR multiplier to block sequence + */ +void hwss_add_dpp_set_hdr_multiplier(struct block_sequence_state *seq_state, + struct dpp *dpp, uint32_t hw_mult) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.dpp_set_hdr_multiplier_params.dpp = dpp; + seq_state->steps[*seq_state->num_steps].params.dpp_set_hdr_multiplier_params.hw_mult = hw_mult; + seq_state->steps[*seq_state->num_steps].func = DPP_SET_HDR_MULTIPLIER; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP program DET size to block sequence + */ +void hwss_add_hubp_program_det_size(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + unsigned int hubp_inst, + unsigned int det_buffer_size_kb) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_det_size_params.hubbub = hubbub; + seq_state->steps[*seq_state->num_steps].params.program_det_size_params.hubp_inst = hubp_inst; + seq_state->steps[*seq_state->num_steps].params.program_det_size_params.det_buffer_size_kb = det_buffer_size_kb; + seq_state->steps[*seq_state->num_steps].func = HUBP_PROGRAM_DET_SIZE; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_program_mcache_id(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct dml2_hubp_pipe_mcache_regs *mcache_regs) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_mcache_id_and_split_coordinate.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.program_mcache_id_and_split_coordinate.mcache_regs = mcache_regs; + seq_state->steps[*seq_state->num_steps].func = HUBP_PROGRAM_MCACHE_ID; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubbub_force_pstate_change_control(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + bool enable, + bool wait) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.hubbub_force_pstate_change_control_params.hubbub = hubbub; + seq_state->steps[*seq_state->num_steps].params.hubbub_force_pstate_change_control_params.enable = enable; + seq_state->steps[*seq_state->num_steps].params.hubbub_force_pstate_change_control_params.wait = wait; + seq_state->steps[*seq_state->num_steps].func = HUBBUB_FORCE_PSTATE_CHANGE_CONTROL; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP program DET segments to block sequence + */ +void hwss_add_hubp_program_det_segments(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + unsigned int hubp_inst, + unsigned int det_size) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.program_det_segments_params.hubbub = hubbub; + seq_state->steps[*seq_state->num_steps].params.program_det_segments_params.hubp_inst = hubp_inst; + seq_state->steps[*seq_state->num_steps].params.program_det_segments_params.det_size = det_size; + seq_state->steps[*seq_state->num_steps].func = HUBP_PROGRAM_DET_SEGMENTS; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add OPP set dynamic expansion to block sequence + */ +void hwss_add_opp_set_dyn_expansion(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + enum dc_color_space color_space, + enum dc_color_depth color_depth, + enum signal_type signal) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.opp_set_dyn_expansion_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_set_dyn_expansion_params.color_space = color_space; + seq_state->steps[*seq_state->num_steps].params.opp_set_dyn_expansion_params.color_depth = color_depth; + seq_state->steps[*seq_state->num_steps].params.opp_set_dyn_expansion_params.signal = signal; + seq_state->steps[*seq_state->num_steps].func = OPP_SET_DYN_EXPANSION; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add OPP program FMT to block sequence + */ +void hwss_add_opp_program_fmt(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + struct bit_depth_reduction_params *fmt_bit_depth, + struct clamping_and_pixel_encoding_params *clamping) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.opp_program_fmt_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_program_fmt_params.fmt_bit_depth = fmt_bit_depth; + seq_state->steps[*seq_state->num_steps].params.opp_program_fmt_params.clamping = clamping; + seq_state->steps[*seq_state->num_steps].func = OPP_PROGRAM_FMT; + (*seq_state->num_steps)++; + } +} + +void hwss_add_opp_program_left_edge_extra_pixel(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + enum dc_pixel_encoding pixel_encoding, + bool is_otg_master) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = OPP_PROGRAM_LEFT_EDGE_EXTRA_PIXEL; + seq_state->steps[*seq_state->num_steps].params.opp_program_left_edge_extra_pixel_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_program_left_edge_extra_pixel_params.pixel_encoding = pixel_encoding; + seq_state->steps[*seq_state->num_steps].params.opp_program_left_edge_extra_pixel_params.is_otg_master = is_otg_master; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add ABM set pipe to block sequence + */ +void hwss_add_abm_set_pipe(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_abm_pipe_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_abm_pipe_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].func = ABM_SET_PIPE; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add ABM set level to block sequence + */ +void hwss_add_abm_set_level(struct block_sequence_state *seq_state, + struct abm *abm, + uint32_t abm_level) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.set_abm_level_params.abm = abm; + seq_state->steps[*seq_state->num_steps].params.set_abm_level_params.abm_level = abm_level; + seq_state->steps[*seq_state->num_steps].func = ABM_SET_LEVEL; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add TG enable CRTC to block sequence + */ +void hwss_add_tg_enable_crtc(struct block_sequence_state *seq_state, + struct timing_generator *tg) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_enable_crtc_params.tg = tg; + seq_state->steps[*seq_state->num_steps].func = TG_ENABLE_CRTC; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP wait flip pending to block sequence + */ +void hwss_add_hubp_wait_flip_pending(struct block_sequence_state *seq_state, + struct hubp *hubp, + unsigned int timeout_us, + unsigned int polling_interval_us) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.hubp_wait_flip_pending_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_wait_flip_pending_params.timeout_us = timeout_us; + seq_state->steps[*seq_state->num_steps].params.hubp_wait_flip_pending_params.polling_interval_us = polling_interval_us; + seq_state->steps[*seq_state->num_steps].func = HUBP_WAIT_FLIP_PENDING; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add TG wait double buffer pending to block sequence + */ +void hwss_add_tg_wait_double_buffer_pending(struct block_sequence_state *seq_state, + struct timing_generator *tg, + unsigned int timeout_us, + unsigned int polling_interval_us) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].params.tg_wait_double_buffer_pending_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_wait_double_buffer_pending_params.timeout_us = timeout_us; + seq_state->steps[*seq_state->num_steps].params.tg_wait_double_buffer_pending_params.polling_interval_us = polling_interval_us; + seq_state->steps[*seq_state->num_steps].func = TG_WAIT_DOUBLE_BUFFER_PENDING; + (*seq_state->num_steps)++; + } +} + void hwss_program_manual_trigger(union block_sequence_params *params) { struct pipe_ctx *pipe_ctx = params->program_manual_trigger_params.pipe_ctx; @@ -1046,12 +1958,6 @@ void hwss_setup_dpp(union block_sequence_params *params) plane_state->color_space, NULL); } - - if (dpp && dpp->funcs->set_cursor_matrix) { - dpp->funcs->set_cursor_matrix(dpp, - plane_state->color_space, - plane_state->cursor_csc_color_matrix); - } } void hwss_program_bias_and_scale(union block_sequence_params *params) @@ -1062,9 +1968,8 @@ void hwss_program_bias_and_scale(union block_sequence_params *params) struct dc_bias_and_scale bns_params = plane_state->bias_and_scale; //TODO :for CNVC set scale and bias registers if necessary - if (dpp->funcs->dpp_program_bias_and_scale) { + if (dpp->funcs->dpp_program_bias_and_scale) dpp->funcs->dpp_program_bias_and_scale(dpp, &bns_params); - } } void hwss_power_on_mpc_mem_pwr(union block_sequence_params *params) @@ -1114,6 +2019,39 @@ void hwss_subvp_save_surf_addr(union block_sequence_params *params) dc_dmub_srv_subvp_save_surf_addr(dc_dmub_srv, addr, subvp_index); } +void hwss_program_surface_config(union block_sequence_params *params) +{ + struct hubp *hubp = params->program_surface_config_params.hubp; + enum surface_pixel_format format = params->program_surface_config_params.format; + struct dc_tiling_info *tiling_info = params->program_surface_config_params.tiling_info; + struct plane_size size = params->program_surface_config_params.plane_size; + enum dc_rotation_angle rotation = params->program_surface_config_params.rotation; + struct dc_plane_dcc_param *dcc = params->program_surface_config_params.dcc; + bool horizontal_mirror = params->program_surface_config_params.horizontal_mirror; + int compat_level = params->program_surface_config_params.compat_level; + + hubp->funcs->hubp_program_surface_config( + hubp, + format, + tiling_info, + &size, + rotation, + dcc, + horizontal_mirror, + compat_level); + + hubp->power_gated = false; +} + +void hwss_program_mcache_id_and_split_coordinate(union block_sequence_params *params) +{ + struct hubp *hubp = params->program_mcache_id_and_split_coordinate.hubp; + struct dml2_hubp_pipe_mcache_regs *mcache_regs = params->program_mcache_id_and_split_coordinate.mcache_regs; + + hubp->funcs->hubp_program_mcache_id_and_split_coordinate(hubp, mcache_regs); + +} + void get_surface_tile_visual_confirm_color( struct pipe_ctx *pipe_ctx, struct tg_color *color) @@ -1188,6 +2126,7 @@ void hwss_wait_for_odm_update_pending_complete(struct dc *dc, struct dc_state *c void hwss_wait_for_no_pipes_pending(struct dc *dc, struct dc_state *context) { int i; + for (i = 0; i < MAX_PIPES; i++) { int count = 0; struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; @@ -1264,3 +2203,1869 @@ void hwss_process_outstanding_hw_updates(struct dc *dc, struct dc_state *dc_cont if (dc->hwss.program_outstanding_updates) dc->hwss.program_outstanding_updates(dc, dc_context); } + +void hwss_set_odm_combine(union block_sequence_params *params) +{ + struct timing_generator *tg = params->set_odm_combine_params.tg; + int *opp_inst = params->set_odm_combine_params.opp_inst; + int opp_head_count = params->set_odm_combine_params.opp_head_count; + int odm_slice_width = params->set_odm_combine_params.odm_slice_width; + int last_odm_slice_width = params->set_odm_combine_params.last_odm_slice_width; + + if (tg && tg->funcs->set_odm_combine) + tg->funcs->set_odm_combine(tg, opp_inst, opp_head_count, + odm_slice_width, last_odm_slice_width); +} + +void hwss_set_odm_bypass(union block_sequence_params *params) +{ + struct timing_generator *tg = params->set_odm_bypass_params.tg; + const struct dc_crtc_timing *timing = params->set_odm_bypass_params.timing; + + if (tg && tg->funcs->set_odm_bypass) + tg->funcs->set_odm_bypass(tg, timing); +} + +void hwss_opp_pipe_clock_control(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_pipe_clock_control_params.opp; + bool enable = params->opp_pipe_clock_control_params.enable; + + if (opp && opp->funcs->opp_pipe_clock_control) + opp->funcs->opp_pipe_clock_control(opp, enable); +} + +void hwss_opp_program_left_edge_extra_pixel(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_program_left_edge_extra_pixel_params.opp; + enum dc_pixel_encoding pixel_encoding = params->opp_program_left_edge_extra_pixel_params.pixel_encoding; + bool is_otg_master = params->opp_program_left_edge_extra_pixel_params.is_otg_master; + + if (opp && opp->funcs->opp_program_left_edge_extra_pixel) + opp->funcs->opp_program_left_edge_extra_pixel(opp, pixel_encoding, is_otg_master); +} + +void hwss_dccg_set_dto_dscclk(union block_sequence_params *params) +{ + struct dccg *dccg = params->dccg_set_dto_dscclk_params.dccg; + int inst = params->dccg_set_dto_dscclk_params.inst; + int num_slices_h = params->dccg_set_dto_dscclk_params.num_slices_h; + + if (dccg && dccg->funcs->set_dto_dscclk) + dccg->funcs->set_dto_dscclk(dccg, inst, num_slices_h); +} + +void hwss_dsc_set_config(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_set_config_params.dsc; + struct dsc_config *dsc_cfg = params->dsc_set_config_params.dsc_cfg; + struct dsc_optc_config *dsc_optc_cfg = params->dsc_set_config_params.dsc_optc_cfg; + + if (dsc && dsc->funcs->dsc_set_config) + dsc->funcs->dsc_set_config(dsc, dsc_cfg, dsc_optc_cfg); +} + +void hwss_dsc_enable(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_enable_params.dsc; + int opp_inst = params->dsc_enable_params.opp_inst; + + if (dsc && dsc->funcs->dsc_enable) + dsc->funcs->dsc_enable(dsc, opp_inst); +} + +void hwss_tg_set_dsc_config(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_set_dsc_config_params.tg; + enum optc_dsc_mode optc_dsc_mode = OPTC_DSC_DISABLED; + uint32_t bytes_per_pixel = 0; + uint32_t slice_width = 0; + + if (params->tg_set_dsc_config_params.enable) { + struct dsc_optc_config *dsc_optc_cfg = params->tg_set_dsc_config_params.dsc_optc_cfg; + + if (dsc_optc_cfg) { + bytes_per_pixel = dsc_optc_cfg->bytes_per_pixel; + slice_width = dsc_optc_cfg->slice_width; + optc_dsc_mode = dsc_optc_cfg->is_pixel_format_444 ? + OPTC_DSC_ENABLED_444 : OPTC_DSC_ENABLED_NATIVE_SUBSAMPLED; + } + } + + if (tg && tg->funcs->set_dsc_config) + tg->funcs->set_dsc_config(tg, optc_dsc_mode, bytes_per_pixel, slice_width); +} + +void hwss_dsc_disconnect(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_disconnect_params.dsc; + + if (dsc && dsc->funcs->dsc_disconnect) + dsc->funcs->dsc_disconnect(dsc); +} + +void hwss_dsc_read_state(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_read_state_params.dsc; + struct dcn_dsc_state *dsc_state = params->dsc_read_state_params.dsc_state; + + if (dsc && dsc->funcs->dsc_read_state) + dsc->funcs->dsc_read_state(dsc, dsc_state); +} + +void hwss_dsc_calculate_and_set_config(union block_sequence_params *params) +{ + struct pipe_ctx *pipe_ctx = params->dsc_calculate_and_set_config_params.pipe_ctx; + struct pipe_ctx *top_pipe = pipe_ctx; + bool enable = params->dsc_calculate_and_set_config_params.enable; + int opp_cnt = params->dsc_calculate_and_set_config_params.opp_cnt; + + struct display_stream_compressor *dsc = pipe_ctx->stream_res.dsc; + struct dc_stream_state *stream = pipe_ctx->stream; + + if (!dsc || !enable) + return; + + /* Calculate DSC configuration - extracted from dcn32_update_dsc_on_stream */ + struct dsc_config dsc_cfg; + + while (top_pipe->prev_odm_pipe) + top_pipe = top_pipe->prev_odm_pipe; + + dsc_cfg.pic_width = (stream->timing.h_addressable + top_pipe->dsc_padding_params.dsc_hactive_padding + + stream->timing.h_border_left + stream->timing.h_border_right) / opp_cnt; + dsc_cfg.pic_height = stream->timing.v_addressable + stream->timing.v_border_top + stream->timing.v_border_bottom; + dsc_cfg.pixel_encoding = stream->timing.pixel_encoding; + dsc_cfg.color_depth = stream->timing.display_color_depth; + dsc_cfg.is_odm = top_pipe->next_odm_pipe ? true : false; + dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; + dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = top_pipe->dsc_padding_params.dsc_hactive_padding; + + /* Set DSC configuration */ + if (dsc->funcs->dsc_set_config) + dsc->funcs->dsc_set_config(dsc, &dsc_cfg, + ¶ms->dsc_calculate_and_set_config_params.dsc_optc_cfg); +} + +void hwss_dsc_enable_with_opp(union block_sequence_params *params) +{ + struct pipe_ctx *pipe_ctx = params->dsc_enable_with_opp_params.pipe_ctx; + struct display_stream_compressor *dsc = pipe_ctx->stream_res.dsc; + + if (dsc && dsc->funcs->dsc_enable) + dsc->funcs->dsc_enable(dsc, pipe_ctx->stream_res.opp->inst); +} + +void hwss_tg_program_global_sync(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_program_global_sync_params.tg; + int vready_offset = params->tg_program_global_sync_params.vready_offset; + unsigned int vstartup_lines = params->tg_program_global_sync_params.vstartup_lines; + unsigned int vupdate_offset_pixels = params->tg_program_global_sync_params.vupdate_offset_pixels; + unsigned int vupdate_vupdate_width_pixels = params->tg_program_global_sync_params.vupdate_vupdate_width_pixels; + unsigned int pstate_keepout_start_lines = params->tg_program_global_sync_params.pstate_keepout_start_lines; + + if (tg->funcs->program_global_sync) { + tg->funcs->program_global_sync(tg, vready_offset, vstartup_lines, + vupdate_offset_pixels, vupdate_vupdate_width_pixels, pstate_keepout_start_lines); + } +} + +void hwss_tg_wait_for_state(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_wait_for_state_params.tg; + enum crtc_state state = params->tg_wait_for_state_params.state; + + if (tg->funcs->wait_for_state) + tg->funcs->wait_for_state(tg, state); +} + +void hwss_tg_set_vtg_params(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_set_vtg_params_params.tg; + struct dc_crtc_timing *timing = params->tg_set_vtg_params_params.timing; + bool program_fp2 = params->tg_set_vtg_params_params.program_fp2; + + if (tg->funcs->set_vtg_params) + tg->funcs->set_vtg_params(tg, timing, program_fp2); +} + +void hwss_tg_setup_vertical_interrupt2(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_setup_vertical_interrupt2_params.tg; + int start_line = params->tg_setup_vertical_interrupt2_params.start_line; + + if (tg->funcs->setup_vertical_interrupt2) + tg->funcs->setup_vertical_interrupt2(tg, start_line); +} + +void hwss_dpp_set_hdr_multiplier(union block_sequence_params *params) +{ + struct dpp *dpp = params->dpp_set_hdr_multiplier_params.dpp; + uint32_t hw_mult = params->dpp_set_hdr_multiplier_params.hw_mult; + + if (dpp->funcs->dpp_set_hdr_multiplier) + dpp->funcs->dpp_set_hdr_multiplier(dpp, hw_mult); +} + +void hwss_program_det_size(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->program_det_size_params.hubbub; + unsigned int hubp_inst = params->program_det_size_params.hubp_inst; + unsigned int det_buffer_size_kb = params->program_det_size_params.det_buffer_size_kb; + + if (hubbub->funcs->program_det_size) + hubbub->funcs->program_det_size(hubbub, hubp_inst, det_buffer_size_kb); +} + +void hwss_program_det_segments(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->program_det_segments_params.hubbub; + unsigned int hubp_inst = params->program_det_segments_params.hubp_inst; + unsigned int det_size = params->program_det_segments_params.det_size; + + if (hubbub->funcs->program_det_segments) + hubbub->funcs->program_det_segments(hubbub, hubp_inst, det_size); +} + +void hwss_opp_set_dyn_expansion(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_set_dyn_expansion_params.opp; + enum dc_color_space color_space = params->opp_set_dyn_expansion_params.color_space; + enum dc_color_depth color_depth = params->opp_set_dyn_expansion_params.color_depth; + enum signal_type signal = params->opp_set_dyn_expansion_params.signal; + + if (opp->funcs->opp_set_dyn_expansion) + opp->funcs->opp_set_dyn_expansion(opp, color_space, color_depth, signal); +} + +void hwss_opp_program_fmt(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_program_fmt_params.opp; + struct bit_depth_reduction_params *fmt_bit_depth = params->opp_program_fmt_params.fmt_bit_depth; + struct clamping_and_pixel_encoding_params *clamping = params->opp_program_fmt_params.clamping; + + if (opp->funcs->opp_program_fmt) + opp->funcs->opp_program_fmt(opp, fmt_bit_depth, clamping); +} + +void hwss_opp_program_bit_depth_reduction(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_program_bit_depth_reduction_params.opp; + bool use_default_params = params->opp_program_bit_depth_reduction_params.use_default_params; + struct pipe_ctx *pipe_ctx = params->opp_program_bit_depth_reduction_params.pipe_ctx; + struct bit_depth_reduction_params bit_depth_params; + + if (use_default_params) + memset(&bit_depth_params, 0, sizeof(bit_depth_params)); + else + resource_build_bit_depth_reduction_params(pipe_ctx->stream, &bit_depth_params); + + if (opp->funcs->opp_program_bit_depth_reduction) + opp->funcs->opp_program_bit_depth_reduction(opp, &bit_depth_params); +} + +void hwss_opp_set_disp_pattern_generator(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_set_disp_pattern_generator_params.opp; + enum controller_dp_test_pattern test_pattern = params->opp_set_disp_pattern_generator_params.test_pattern; + enum controller_dp_color_space color_space = params->opp_set_disp_pattern_generator_params.color_space; + enum dc_color_depth color_depth = params->opp_set_disp_pattern_generator_params.color_depth; + struct tg_color *solid_color = params->opp_set_disp_pattern_generator_params.use_solid_color ? + ¶ms->opp_set_disp_pattern_generator_params.solid_color : NULL; + int width = params->opp_set_disp_pattern_generator_params.width; + int height = params->opp_set_disp_pattern_generator_params.height; + int offset = params->opp_set_disp_pattern_generator_params.offset; + + if (opp && opp->funcs->opp_set_disp_pattern_generator) { + opp->funcs->opp_set_disp_pattern_generator(opp, test_pattern, color_space, + color_depth, solid_color, width, height, offset); + } +} + +void hwss_set_abm_pipe(union block_sequence_params *params) +{ + struct dc *dc = params->set_abm_pipe_params.dc; + struct pipe_ctx *pipe_ctx = params->set_abm_pipe_params.pipe_ctx; + + dc->hwss.set_pipe(pipe_ctx); +} + +void hwss_set_abm_level(union block_sequence_params *params) +{ + struct abm *abm = params->set_abm_level_params.abm; + unsigned int abm_level = params->set_abm_level_params.abm_level; + + if (abm->funcs->set_abm_level) + abm->funcs->set_abm_level(abm, abm_level); +} + +void hwss_set_abm_immediate_disable(union block_sequence_params *params) +{ + struct dc *dc = params->set_abm_immediate_disable_params.dc; + struct pipe_ctx *pipe_ctx = params->set_abm_immediate_disable_params.pipe_ctx; + + if (dc && dc->hwss.set_abm_immediate_disable) + dc->hwss.set_abm_immediate_disable(pipe_ctx); +} + +void hwss_mpc_remove_mpcc(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_remove_mpcc_params.mpc; + struct mpc_tree *mpc_tree_params = params->mpc_remove_mpcc_params.mpc_tree_params; + struct mpcc *mpcc_to_remove = params->mpc_remove_mpcc_params.mpcc_to_remove; + + mpc->funcs->remove_mpcc(mpc, mpc_tree_params, mpcc_to_remove); +} + +void hwss_opp_set_mpcc_disconnect_pending(union block_sequence_params *params) +{ + struct output_pixel_processor *opp = params->opp_set_mpcc_disconnect_pending_params.opp; + int mpcc_inst = params->opp_set_mpcc_disconnect_pending_params.mpcc_inst; + bool pending = params->opp_set_mpcc_disconnect_pending_params.pending; + + opp->mpcc_disconnect_pending[mpcc_inst] = pending; +} + +void hwss_dc_set_optimized_required(union block_sequence_params *params) +{ + struct dc *dc = params->dc_set_optimized_required_params.dc; + bool optimized_required = params->dc_set_optimized_required_params.optimized_required; + + dc->optimized_required = optimized_required; +} + +void hwss_hubp_disconnect(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_disconnect_params.hubp; + + if (hubp->funcs->hubp_disconnect) + hubp->funcs->hubp_disconnect(hubp); +} + +void hwss_hubbub_force_pstate_change_control(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->hubbub_force_pstate_change_control_params.hubbub; + bool enable = params->hubbub_force_pstate_change_control_params.enable; + bool wait = params->hubbub_force_pstate_change_control_params.wait; + + if (hubbub->funcs->force_pstate_change_control) { + hubbub->funcs->force_pstate_change_control(hubbub, enable, wait); + /* Add delay when enabling pstate change control */ + if (enable) + udelay(500); + } +} + +void hwss_tg_enable_crtc(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_enable_crtc_params.tg; + + if (tg->funcs->enable_crtc) + tg->funcs->enable_crtc(tg); +} + +void hwss_tg_set_gsl(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_set_gsl_params.tg; + struct gsl_params *gsl = ¶ms->tg_set_gsl_params.gsl; + + if (tg->funcs->set_gsl) + tg->funcs->set_gsl(tg, gsl); +} + +void hwss_tg_set_gsl_source_select(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_set_gsl_source_select_params.tg; + int group_idx = params->tg_set_gsl_source_select_params.group_idx; + uint32_t gsl_ready_signal = params->tg_set_gsl_source_select_params.gsl_ready_signal; + + if (tg->funcs->set_gsl_source_select) + tg->funcs->set_gsl_source_select(tg, group_idx, gsl_ready_signal); +} + +void hwss_hubp_wait_flip_pending(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_wait_flip_pending_params.hubp; + unsigned int timeout_us = params->hubp_wait_flip_pending_params.timeout_us; + unsigned int polling_interval_us = params->hubp_wait_flip_pending_params.polling_interval_us; + int j = 0; + + for (j = 0; j < timeout_us / polling_interval_us + && hubp->funcs->hubp_is_flip_pending(hubp); j++) + udelay(polling_interval_us); +} + +void hwss_tg_wait_double_buffer_pending(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_wait_double_buffer_pending_params.tg; + unsigned int timeout_us = params->tg_wait_double_buffer_pending_params.timeout_us; + unsigned int polling_interval_us = params->tg_wait_double_buffer_pending_params.polling_interval_us; + int j = 0; + + if (tg->funcs->get_optc_double_buffer_pending) { + for (j = 0; j < timeout_us / polling_interval_us + && tg->funcs->get_optc_double_buffer_pending(tg); j++) + udelay(polling_interval_us); + } +} + +void hwss_update_force_pstate(union block_sequence_params *params) +{ + struct dc *dc = params->update_force_pstate_params.dc; + struct dc_state *context = params->update_force_pstate_params.context; + struct dce_hwseq *hwseq = dc->hwseq; + + if (hwseq->funcs.update_force_pstate) + hwseq->funcs.update_force_pstate(dc, context); +} + +void hwss_hubbub_apply_dedcn21_147_wa(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->hubbub_apply_dedcn21_147_wa_params.hubbub; + + hubbub->funcs->apply_DEDCN21_147_wa(hubbub); +} + +void hwss_hubbub_allow_self_refresh_control(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->hubbub_allow_self_refresh_control_params.hubbub; + bool allow = params->hubbub_allow_self_refresh_control_params.allow; + + hubbub->funcs->allow_self_refresh_control(hubbub, allow); + + if (!allow && params->hubbub_allow_self_refresh_control_params.disallow_self_refresh_applied) + *params->hubbub_allow_self_refresh_control_params.disallow_self_refresh_applied = true; +} + +void hwss_tg_get_frame_count(union block_sequence_params *params) +{ + struct timing_generator *tg = params->tg_get_frame_count_params.tg; + unsigned int *frame_count = params->tg_get_frame_count_params.frame_count; + + *frame_count = tg->funcs->get_frame_count(tg); +} + +void hwss_mpc_set_dwb_mux(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_set_dwb_mux_params.mpc; + int dwb_id = params->mpc_set_dwb_mux_params.dwb_id; + int mpcc_id = params->mpc_set_dwb_mux_params.mpcc_id; + + if (mpc->funcs->set_dwb_mux) + mpc->funcs->set_dwb_mux(mpc, dwb_id, mpcc_id); +} + +void hwss_mpc_disable_dwb_mux(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_disable_dwb_mux_params.mpc; + unsigned int dwb_id = params->mpc_disable_dwb_mux_params.dwb_id; + + if (mpc->funcs->disable_dwb_mux) + mpc->funcs->disable_dwb_mux(mpc, dwb_id); +} + +void hwss_mcif_wb_config_buf(union block_sequence_params *params) +{ + struct mcif_wb *mcif_wb = params->mcif_wb_config_buf_params.mcif_wb; + struct mcif_buf_params *mcif_buf_params = params->mcif_wb_config_buf_params.mcif_buf_params; + unsigned int dest_height = params->mcif_wb_config_buf_params.dest_height; + + if (mcif_wb->funcs->config_mcif_buf) + mcif_wb->funcs->config_mcif_buf(mcif_wb, mcif_buf_params, dest_height); +} + +void hwss_mcif_wb_config_arb(union block_sequence_params *params) +{ + struct mcif_wb *mcif_wb = params->mcif_wb_config_arb_params.mcif_wb; + struct mcif_arb_params *mcif_arb_params = params->mcif_wb_config_arb_params.mcif_arb_params; + + if (mcif_wb->funcs->config_mcif_arb) + mcif_wb->funcs->config_mcif_arb(mcif_wb, mcif_arb_params); +} + +void hwss_mcif_wb_enable(union block_sequence_params *params) +{ + struct mcif_wb *mcif_wb = params->mcif_wb_enable_params.mcif_wb; + + if (mcif_wb->funcs->enable_mcif) + mcif_wb->funcs->enable_mcif(mcif_wb); +} + +void hwss_mcif_wb_disable(union block_sequence_params *params) +{ + struct mcif_wb *mcif_wb = params->mcif_wb_disable_params.mcif_wb; + + if (mcif_wb->funcs->disable_mcif) + mcif_wb->funcs->disable_mcif(mcif_wb); +} + +void hwss_dwbc_enable(union block_sequence_params *params) +{ + struct dwbc *dwb = params->dwbc_enable_params.dwb; + struct dc_dwb_params *dwb_params = params->dwbc_enable_params.dwb_params; + + if (dwb->funcs->enable) + dwb->funcs->enable(dwb, dwb_params); +} + +void hwss_dwbc_disable(union block_sequence_params *params) +{ + struct dwbc *dwb = params->dwbc_disable_params.dwb; + + if (dwb->funcs->disable) + dwb->funcs->disable(dwb); +} + +void hwss_dwbc_update(union block_sequence_params *params) +{ + struct dwbc *dwb = params->dwbc_update_params.dwb; + struct dc_dwb_params *dwb_params = params->dwbc_update_params.dwb_params; + + if (dwb->funcs->update) + dwb->funcs->update(dwb, dwb_params); +} + +void hwss_hubp_update_mall_sel(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_update_mall_sel_params.hubp; + uint32_t mall_sel = params->hubp_update_mall_sel_params.mall_sel; + bool cache_cursor = params->hubp_update_mall_sel_params.cache_cursor; + + if (hubp && hubp->funcs->hubp_update_mall_sel) + hubp->funcs->hubp_update_mall_sel(hubp, mall_sel, cache_cursor); +} + +void hwss_hubp_prepare_subvp_buffering(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_prepare_subvp_buffering_params.hubp; + bool enable = params->hubp_prepare_subvp_buffering_params.enable; + + if (hubp && hubp->funcs->hubp_prepare_subvp_buffering) + hubp->funcs->hubp_prepare_subvp_buffering(hubp, enable); +} + +void hwss_hubp_set_blank_en(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_set_blank_en_params.hubp; + bool enable = params->hubp_set_blank_en_params.enable; + + if (hubp && hubp->funcs->set_hubp_blank_en) + hubp->funcs->set_hubp_blank_en(hubp, enable); +} + +void hwss_hubp_disable_control(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_disable_control_params.hubp; + bool disable = params->hubp_disable_control_params.disable; + + if (hubp && hubp->funcs->hubp_disable_control) + hubp->funcs->hubp_disable_control(hubp, disable); +} + +void hwss_hubbub_soft_reset(union block_sequence_params *params) +{ + struct hubbub *hubbub = params->hubbub_soft_reset_params.hubbub; + bool reset = params->hubbub_soft_reset_params.reset; + + if (hubbub) + params->hubbub_soft_reset_params.hubbub_soft_reset(hubbub, reset); +} + +void hwss_hubp_clk_cntl(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_clk_cntl_params.hubp; + bool enable = params->hubp_clk_cntl_params.enable; + + if (hubp && hubp->funcs->hubp_clk_cntl) { + hubp->funcs->hubp_clk_cntl(hubp, enable); + hubp->power_gated = !enable; + } +} + +void hwss_hubp_init(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_init_params.hubp; + + if (hubp && hubp->funcs->hubp_init) + hubp->funcs->hubp_init(hubp); +} + +void hwss_hubp_set_vm_system_aperture_settings(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_set_vm_system_aperture_settings_params.hubp; + struct vm_system_aperture_param apt; + + apt.sys_default = params->hubp_set_vm_system_aperture_settings_params.sys_default; + apt.sys_high = params->hubp_set_vm_system_aperture_settings_params.sys_high; + apt.sys_low = params->hubp_set_vm_system_aperture_settings_params.sys_low; + + if (hubp && hubp->funcs->hubp_set_vm_system_aperture_settings) + hubp->funcs->hubp_set_vm_system_aperture_settings(hubp, &apt); +} + +void hwss_hubp_set_flip_int(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_set_flip_int_params.hubp; + + if (hubp && hubp->funcs->hubp_set_flip_int) + hubp->funcs->hubp_set_flip_int(hubp); +} + +void hwss_dpp_dppclk_control(union block_sequence_params *params) +{ + struct dpp *dpp = params->dpp_dppclk_control_params.dpp; + bool dppclk_div = params->dpp_dppclk_control_params.dppclk_div; + bool enable = params->dpp_dppclk_control_params.enable; + + if (dpp && dpp->funcs->dpp_dppclk_control) + dpp->funcs->dpp_dppclk_control(dpp, dppclk_div, enable); +} + +void hwss_disable_phantom_crtc(union block_sequence_params *params) +{ + struct timing_generator *tg = params->disable_phantom_crtc_params.tg; + + if (tg && tg->funcs->disable_phantom_crtc) + tg->funcs->disable_phantom_crtc(tg); +} + +void hwss_dsc_pg_status(union block_sequence_params *params) +{ + struct dce_hwseq *hws = params->dsc_pg_status_params.hws; + int dsc_inst = params->dsc_pg_status_params.dsc_inst; + + if (hws && hws->funcs.dsc_pg_status) + params->dsc_pg_status_params.is_ungated = hws->funcs.dsc_pg_status(hws, dsc_inst); +} + +void hwss_dsc_wait_disconnect_pending_clear(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_wait_disconnect_pending_clear_params.dsc; + + if (!params->dsc_wait_disconnect_pending_clear_params.is_ungated) + return; + if (*params->dsc_wait_disconnect_pending_clear_params.is_ungated == false) + return; + + if (dsc && dsc->funcs->dsc_wait_disconnect_pending_clear) + dsc->funcs->dsc_wait_disconnect_pending_clear(dsc); +} + +void hwss_dsc_disable(union block_sequence_params *params) +{ + struct display_stream_compressor *dsc = params->dsc_disable_params.dsc; + + if (!params->dsc_disable_params.is_ungated) + return; + if (*params->dsc_disable_params.is_ungated == false) + return; + + if (dsc && dsc->funcs->dsc_disable) + dsc->funcs->dsc_disable(dsc); +} + +void hwss_dccg_set_ref_dscclk(union block_sequence_params *params) +{ + struct dccg *dccg = params->dccg_set_ref_dscclk_params.dccg; + int dsc_inst = params->dccg_set_ref_dscclk_params.dsc_inst; + + if (!params->dccg_set_ref_dscclk_params.is_ungated) + return; + if (*params->dccg_set_ref_dscclk_params.is_ungated == false) + return; + + if (dccg && dccg->funcs->set_ref_dscclk) + dccg->funcs->set_ref_dscclk(dccg, dsc_inst); +} + +void hwss_dpp_pg_control(union block_sequence_params *params) +{ + struct dce_hwseq *hws = params->dpp_pg_control_params.hws; + unsigned int dpp_inst = params->dpp_pg_control_params.dpp_inst; + bool power_on = params->dpp_pg_control_params.power_on; + + if (hws->funcs.dpp_pg_control) + hws->funcs.dpp_pg_control(hws, dpp_inst, power_on); +} + +void hwss_hubp_pg_control(union block_sequence_params *params) +{ + struct dce_hwseq *hws = params->hubp_pg_control_params.hws; + unsigned int hubp_inst = params->hubp_pg_control_params.hubp_inst; + bool power_on = params->hubp_pg_control_params.power_on; + + if (hws->funcs.hubp_pg_control) + hws->funcs.hubp_pg_control(hws, hubp_inst, power_on); +} + +void hwss_hubp_reset(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_reset_params.hubp; + + if (hubp && hubp->funcs->hubp_reset) + hubp->funcs->hubp_reset(hubp); +} + +void hwss_dpp_reset(union block_sequence_params *params) +{ + struct dpp *dpp = params->dpp_reset_params.dpp; + + if (dpp && dpp->funcs->dpp_reset) + dpp->funcs->dpp_reset(dpp); +} + +void hwss_dpp_root_clock_control(union block_sequence_params *params) +{ + struct dce_hwseq *hws = params->dpp_root_clock_control_params.hws; + unsigned int dpp_inst = params->dpp_root_clock_control_params.dpp_inst; + bool clock_on = params->dpp_root_clock_control_params.clock_on; + + if (hws->funcs.dpp_root_clock_control) + hws->funcs.dpp_root_clock_control(hws, dpp_inst, clock_on); +} + +void hwss_dc_ip_request_cntl(union block_sequence_params *params) +{ + struct dc *dc = params->dc_ip_request_cntl_params.dc; + bool enable = params->dc_ip_request_cntl_params.enable; + struct dce_hwseq *hws = dc->hwseq; + + if (hws->funcs.dc_ip_request_cntl) + hws->funcs.dc_ip_request_cntl(dc, enable); +} + +void hwss_dccg_update_dpp_dto(union block_sequence_params *params) +{ + struct dccg *dccg = params->dccg_update_dpp_dto_params.dccg; + int dpp_inst = params->dccg_update_dpp_dto_params.dpp_inst; + int dppclk_khz = params->dccg_update_dpp_dto_params.dppclk_khz; + + if (dccg && dccg->funcs->update_dpp_dto) + dccg->funcs->update_dpp_dto(dccg, dpp_inst, dppclk_khz); +} + +void hwss_hubp_vtg_sel(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_vtg_sel_params.hubp; + uint32_t otg_inst = params->hubp_vtg_sel_params.otg_inst; + + if (hubp && hubp->funcs->hubp_vtg_sel) + hubp->funcs->hubp_vtg_sel(hubp, otg_inst); +} + +void hwss_hubp_setup2(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_setup2_params.hubp; + struct dml2_dchub_per_pipe_register_set *hubp_regs = params->hubp_setup2_params.hubp_regs; + union dml2_global_sync_programming *global_sync = params->hubp_setup2_params.global_sync; + struct dc_crtc_timing *timing = params->hubp_setup2_params.timing; + + if (hubp && hubp->funcs->hubp_setup2) + hubp->funcs->hubp_setup2(hubp, hubp_regs, global_sync, timing); +} + +void hwss_hubp_setup(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_setup_params.hubp; + struct _vcs_dpi_display_dlg_regs_st *dlg_regs = params->hubp_setup_params.dlg_regs; + struct _vcs_dpi_display_ttu_regs_st *ttu_regs = params->hubp_setup_params.ttu_regs; + struct _vcs_dpi_display_rq_regs_st *rq_regs = params->hubp_setup_params.rq_regs; + struct _vcs_dpi_display_pipe_dest_params_st *pipe_dest = params->hubp_setup_params.pipe_dest; + + if (hubp && hubp->funcs->hubp_setup) + hubp->funcs->hubp_setup(hubp, dlg_regs, ttu_regs, rq_regs, pipe_dest); +} + +void hwss_hubp_set_unbounded_requesting(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_set_unbounded_requesting_params.hubp; + bool unbounded_req = params->hubp_set_unbounded_requesting_params.unbounded_req; + + if (hubp && hubp->funcs->set_unbounded_requesting) + hubp->funcs->set_unbounded_requesting(hubp, unbounded_req); +} + +void hwss_hubp_setup_interdependent2(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_setup_interdependent2_params.hubp; + struct dml2_dchub_per_pipe_register_set *hubp_regs = params->hubp_setup_interdependent2_params.hubp_regs; + + if (hubp && hubp->funcs->hubp_setup_interdependent2) + hubp->funcs->hubp_setup_interdependent2(hubp, hubp_regs); +} + +void hwss_hubp_setup_interdependent(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_setup_interdependent_params.hubp; + struct _vcs_dpi_display_dlg_regs_st *dlg_regs = params->hubp_setup_interdependent_params.dlg_regs; + struct _vcs_dpi_display_ttu_regs_st *ttu_regs = params->hubp_setup_interdependent_params.ttu_regs; + + if (hubp && hubp->funcs->hubp_setup_interdependent) + hubp->funcs->hubp_setup_interdependent(hubp, dlg_regs, ttu_regs); +} + +void hwss_dpp_set_cursor_matrix(union block_sequence_params *params) +{ + struct dpp *dpp = params->dpp_set_cursor_matrix_params.dpp; + enum dc_color_space color_space = params->dpp_set_cursor_matrix_params.color_space; + struct dc_csc_transform *cursor_csc_color_matrix = params->dpp_set_cursor_matrix_params.cursor_csc_color_matrix; + + if (dpp && dpp->funcs->set_cursor_matrix) + dpp->funcs->set_cursor_matrix(dpp, color_space, *cursor_csc_color_matrix); +} + +void hwss_mpc_update_mpcc(union block_sequence_params *params) +{ + struct dc *dc = params->mpc_update_mpcc_params.dc; + struct pipe_ctx *pipe_ctx = params->mpc_update_mpcc_params.pipe_ctx; + struct dce_hwseq *hws = dc->hwseq; + + if (hws->funcs.update_mpcc) + hws->funcs.update_mpcc(dc, pipe_ctx); +} + +void hwss_mpc_update_blending(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_update_blending_params.mpc; + struct mpcc_blnd_cfg *blnd_cfg = ¶ms->mpc_update_blending_params.blnd_cfg; + int mpcc_id = params->mpc_update_blending_params.mpcc_id; + + if (mpc && mpc->funcs->update_blending) + mpc->funcs->update_blending(mpc, blnd_cfg, mpcc_id); +} + +void hwss_mpc_assert_idle_mpcc(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_assert_idle_mpcc_params.mpc; + int mpcc_id = params->mpc_assert_idle_mpcc_params.mpcc_id; + + if (mpc && mpc->funcs->wait_for_idle) + mpc->funcs->wait_for_idle(mpc, mpcc_id); +} + +void hwss_mpc_insert_plane(union block_sequence_params *params) +{ + struct mpc *mpc = params->mpc_insert_plane_params.mpc; + struct mpc_tree *tree = params->mpc_insert_plane_params.mpc_tree_params; + struct mpcc_blnd_cfg *blnd_cfg = ¶ms->mpc_insert_plane_params.blnd_cfg; + struct mpcc_sm_cfg *sm_cfg = params->mpc_insert_plane_params.sm_cfg; + struct mpcc *insert_above_mpcc = params->mpc_insert_plane_params.insert_above_mpcc; + int mpcc_id = params->mpc_insert_plane_params.mpcc_id; + int dpp_id = params->mpc_insert_plane_params.dpp_id; + + if (mpc && mpc->funcs->insert_plane) + mpc->funcs->insert_plane(mpc, tree, blnd_cfg, sm_cfg, insert_above_mpcc, + dpp_id, mpcc_id); +} + +void hwss_dpp_set_scaler(union block_sequence_params *params) +{ + struct dpp *dpp = params->dpp_set_scaler_params.dpp; + const struct scaler_data *scl_data = params->dpp_set_scaler_params.scl_data; + + if (dpp && dpp->funcs->dpp_set_scaler) + dpp->funcs->dpp_set_scaler(dpp, scl_data); +} + +void hwss_hubp_mem_program_viewport(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_mem_program_viewport_params.hubp; + const struct rect *viewport = params->hubp_mem_program_viewport_params.viewport; + const struct rect *viewport_c = params->hubp_mem_program_viewport_params.viewport_c; + + if (hubp && hubp->funcs->mem_program_viewport) + hubp->funcs->mem_program_viewport(hubp, viewport, viewport_c); +} + +void hwss_abort_cursor_offload_update(union block_sequence_params *params) +{ + struct dc *dc = params->abort_cursor_offload_update_params.dc; + struct pipe_ctx *pipe_ctx = params->abort_cursor_offload_update_params.pipe_ctx; + + if (dc && dc->hwss.abort_cursor_offload_update) + dc->hwss.abort_cursor_offload_update(dc, pipe_ctx); +} + +void hwss_set_cursor_attribute(union block_sequence_params *params) +{ + struct dc *dc = params->set_cursor_attribute_params.dc; + struct pipe_ctx *pipe_ctx = params->set_cursor_attribute_params.pipe_ctx; + + if (dc && dc->hwss.set_cursor_attribute) + dc->hwss.set_cursor_attribute(pipe_ctx); +} + +void hwss_set_cursor_position(union block_sequence_params *params) +{ + struct dc *dc = params->set_cursor_position_params.dc; + struct pipe_ctx *pipe_ctx = params->set_cursor_position_params.pipe_ctx; + + if (dc && dc->hwss.set_cursor_position) + dc->hwss.set_cursor_position(pipe_ctx); +} + +void hwss_set_cursor_sdr_white_level(union block_sequence_params *params) +{ + struct dc *dc = params->set_cursor_sdr_white_level_params.dc; + struct pipe_ctx *pipe_ctx = params->set_cursor_sdr_white_level_params.pipe_ctx; + + if (dc && dc->hwss.set_cursor_sdr_white_level) + dc->hwss.set_cursor_sdr_white_level(pipe_ctx); +} + +void hwss_program_output_csc(union block_sequence_params *params) +{ + struct dc *dc = params->program_output_csc_params.dc; + struct pipe_ctx *pipe_ctx = params->program_output_csc_params.pipe_ctx; + enum dc_color_space colorspace = params->program_output_csc_params.colorspace; + uint16_t *matrix = params->program_output_csc_params.matrix; + int opp_id = params->program_output_csc_params.opp_id; + + if (dc && dc->hwss.program_output_csc) + dc->hwss.program_output_csc(dc, pipe_ctx, colorspace, matrix, opp_id); +} + +void hwss_hubp_set_blank(union block_sequence_params *params) +{ + struct hubp *hubp = params->hubp_set_blank_params.hubp; + bool blank = params->hubp_set_blank_params.blank; + + if (hubp && hubp->funcs->set_blank) + hubp->funcs->set_blank(hubp, blank); +} + +void hwss_phantom_hubp_post_enable(union block_sequence_params *params) +{ + struct hubp *hubp = params->phantom_hubp_post_enable_params.hubp; + + if (hubp && hubp->funcs->phantom_hubp_post_enable) + hubp->funcs->phantom_hubp_post_enable(hubp); +} + +void hwss_add_dccg_set_dto_dscclk(struct block_sequence_state *seq_state, + struct dccg *dccg, int inst, int num_slices_h) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DCCG_SET_DTO_DSCCLK; + seq_state->steps[*seq_state->num_steps].params.dccg_set_dto_dscclk_params.dccg = dccg; + seq_state->steps[*seq_state->num_steps].params.dccg_set_dto_dscclk_params.inst = inst; + seq_state->steps[*seq_state->num_steps].params.dccg_set_dto_dscclk_params.num_slices_h = num_slices_h; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_calculate_and_set_config(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx, bool enable, int opp_cnt) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_CALCULATE_AND_SET_CONFIG; + seq_state->steps[*seq_state->num_steps].params.dsc_calculate_and_set_config_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.dsc_calculate_and_set_config_params.enable = enable; + seq_state->steps[*seq_state->num_steps].params.dsc_calculate_and_set_config_params.opp_cnt = opp_cnt; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mpc_remove_mpcc(struct block_sequence_state *seq_state, + struct mpc *mpc, struct mpc_tree *mpc_tree_params, struct mpcc *mpcc_to_remove) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_REMOVE_MPCC; + seq_state->steps[*seq_state->num_steps].params.mpc_remove_mpcc_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_remove_mpcc_params.mpc_tree_params = mpc_tree_params; + seq_state->steps[*seq_state->num_steps].params.mpc_remove_mpcc_params.mpcc_to_remove = mpcc_to_remove; + (*seq_state->num_steps)++; + } +} + +void hwss_add_opp_set_mpcc_disconnect_pending(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, int mpcc_inst, bool pending) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = OPP_SET_MPCC_DISCONNECT_PENDING; + seq_state->steps[*seq_state->num_steps].params.opp_set_mpcc_disconnect_pending_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_set_mpcc_disconnect_pending_params.mpcc_inst = mpcc_inst; + seq_state->steps[*seq_state->num_steps].params.opp_set_mpcc_disconnect_pending_params.pending = pending; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_disconnect(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_DISCONNECT; + seq_state->steps[*seq_state->num_steps].params.hubp_disconnect_params.hubp = hubp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_enable_with_opp(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_ENABLE_WITH_OPP; + seq_state->steps[*seq_state->num_steps].params.dsc_enable_with_opp_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_tg_set_dsc_config(struct block_sequence_state *seq_state, + struct timing_generator *tg, struct dsc_optc_config *dsc_optc_cfg, bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = TG_SET_DSC_CONFIG; + seq_state->steps[*seq_state->num_steps].params.tg_set_dsc_config_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_set_dsc_config_params.dsc_optc_cfg = dsc_optc_cfg; + seq_state->steps[*seq_state->num_steps].params.tg_set_dsc_config_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_disconnect(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_DISCONNECT; + seq_state->steps[*seq_state->num_steps].params.dsc_disconnect_params.dsc = dsc; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dc_set_optimized_required(struct block_sequence_state *seq_state, + struct dc *dc, bool optimized_required) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DC_SET_OPTIMIZED_REQUIRED; + seq_state->steps[*seq_state->num_steps].params.dc_set_optimized_required_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.dc_set_optimized_required_params.optimized_required = optimized_required; + (*seq_state->num_steps)++; + } +} + +void hwss_add_abm_set_immediate_disable(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = ABM_SET_IMMEDIATE_DISABLE; + seq_state->steps[*seq_state->num_steps].params.set_abm_immediate_disable_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_abm_immediate_disable_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_opp_set_disp_pattern_generator(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + enum controller_dp_test_pattern test_pattern, + enum controller_dp_color_space color_space, + enum dc_color_depth color_depth, + struct tg_color solid_color, + bool use_solid_color, + int width, + int height, + int offset) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = OPP_SET_DISP_PATTERN_GENERATOR; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.test_pattern = test_pattern; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.color_space = color_space; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.color_depth = color_depth; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.solid_color = solid_color; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.use_solid_color = use_solid_color; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.width = width; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.height = height; + seq_state->steps[*seq_state->num_steps].params.opp_set_disp_pattern_generator_params.offset = offset; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC update blending to block sequence + */ +void hwss_add_mpc_update_blending(struct block_sequence_state *seq_state, + struct mpc *mpc, + struct mpcc_blnd_cfg blnd_cfg, + int mpcc_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_UPDATE_BLENDING; + seq_state->steps[*seq_state->num_steps].params.mpc_update_blending_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_update_blending_params.blnd_cfg = blnd_cfg; + seq_state->steps[*seq_state->num_steps].params.mpc_update_blending_params.mpcc_id = mpcc_id; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC insert plane to block sequence + */ +void hwss_add_mpc_insert_plane(struct block_sequence_state *seq_state, + struct mpc *mpc, + struct mpc_tree *mpc_tree_params, + struct mpcc_blnd_cfg blnd_cfg, + struct mpcc_sm_cfg *sm_cfg, + struct mpcc *insert_above_mpcc, + int dpp_id, + int mpcc_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_INSERT_PLANE; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.mpc_tree_params = mpc_tree_params; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.blnd_cfg = blnd_cfg; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.sm_cfg = sm_cfg; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.insert_above_mpcc = insert_above_mpcc; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.dpp_id = dpp_id; + seq_state->steps[*seq_state->num_steps].params.mpc_insert_plane_params.mpcc_id = mpcc_id; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add MPC assert idle MPCC to block sequence + */ +void hwss_add_mpc_assert_idle_mpcc(struct block_sequence_state *seq_state, + struct mpc *mpc, + int mpcc_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_ASSERT_IDLE_MPCC; + seq_state->steps[*seq_state->num_steps].params.mpc_assert_idle_mpcc_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_assert_idle_mpcc_params.mpcc_id = mpcc_id; + (*seq_state->num_steps)++; + } +} + +/* + * Helper function to add HUBP set blank to block sequence + */ +void hwss_add_hubp_set_blank(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool blank) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_BLANK; + seq_state->steps[*seq_state->num_steps].params.hubp_set_blank_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_set_blank_params.blank = blank; + (*seq_state->num_steps)++; + } +} + +void hwss_add_opp_program_bit_depth_reduction(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + bool use_default_params, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = OPP_PROGRAM_BIT_DEPTH_REDUCTION; + seq_state->steps[*seq_state->num_steps].params.opp_program_bit_depth_reduction_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_program_bit_depth_reduction_params.use_default_params = use_default_params; + seq_state->steps[*seq_state->num_steps].params.opp_program_bit_depth_reduction_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dc_ip_request_cntl(struct block_sequence_state *seq_state, + struct dc *dc, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DC_IP_REQUEST_CNTL; + seq_state->steps[*seq_state->num_steps].params.dc_ip_request_cntl_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.dc_ip_request_cntl_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dwbc_update(struct block_sequence_state *seq_state, + struct dwbc *dwb, + struct dc_dwb_params *dwb_params) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DWBC_UPDATE; + seq_state->steps[*seq_state->num_steps].params.dwbc_update_params.dwb = dwb; + seq_state->steps[*seq_state->num_steps].params.dwbc_update_params.dwb_params = dwb_params; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mcif_wb_config_buf(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb, + struct mcif_buf_params *mcif_buf_params, + unsigned int dest_height) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MCIF_WB_CONFIG_BUF; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_config_buf_params.mcif_wb = mcif_wb; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_config_buf_params.mcif_buf_params = mcif_buf_params; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_config_buf_params.dest_height = dest_height; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mcif_wb_config_arb(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb, + struct mcif_arb_params *mcif_arb_params) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MCIF_WB_CONFIG_ARB; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_config_arb_params.mcif_wb = mcif_wb; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_config_arb_params.mcif_arb_params = mcif_arb_params; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mcif_wb_enable(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MCIF_WB_ENABLE; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_enable_params.mcif_wb = mcif_wb; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mcif_wb_disable(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MCIF_WB_DISABLE; + seq_state->steps[*seq_state->num_steps].params.mcif_wb_disable_params.mcif_wb = mcif_wb; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mpc_set_dwb_mux(struct block_sequence_state *seq_state, + struct mpc *mpc, + int dwb_id, + int mpcc_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_SET_DWB_MUX; + seq_state->steps[*seq_state->num_steps].params.mpc_set_dwb_mux_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_set_dwb_mux_params.dwb_id = dwb_id; + seq_state->steps[*seq_state->num_steps].params.mpc_set_dwb_mux_params.mpcc_id = mpcc_id; + (*seq_state->num_steps)++; + } +} + +void hwss_add_mpc_disable_dwb_mux(struct block_sequence_state *seq_state, + struct mpc *mpc, + unsigned int dwb_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = MPC_DISABLE_DWB_MUX; + seq_state->steps[*seq_state->num_steps].params.mpc_disable_dwb_mux_params.mpc = mpc; + seq_state->steps[*seq_state->num_steps].params.mpc_disable_dwb_mux_params.dwb_id = dwb_id; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dwbc_enable(struct block_sequence_state *seq_state, + struct dwbc *dwb, + struct dc_dwb_params *dwb_params) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DWBC_ENABLE; + seq_state->steps[*seq_state->num_steps].params.dwbc_enable_params.dwb = dwb; + seq_state->steps[*seq_state->num_steps].params.dwbc_enable_params.dwb_params = dwb_params; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dwbc_disable(struct block_sequence_state *seq_state, + struct dwbc *dwb) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DWBC_DISABLE; + seq_state->steps[*seq_state->num_steps].params.dwbc_disable_params.dwb = dwb; + (*seq_state->num_steps)++; + } +} + +void hwss_add_tg_set_gsl(struct block_sequence_state *seq_state, + struct timing_generator *tg, + struct gsl_params gsl) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = TG_SET_GSL; + seq_state->steps[*seq_state->num_steps].params.tg_set_gsl_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_set_gsl_params.gsl = gsl; + (*seq_state->num_steps)++; + } +} + +void hwss_add_tg_set_gsl_source_select(struct block_sequence_state *seq_state, + struct timing_generator *tg, + int group_idx, + uint32_t gsl_ready_signal) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = TG_SET_GSL_SOURCE_SELECT; + seq_state->steps[*seq_state->num_steps].params.tg_set_gsl_source_select_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_set_gsl_source_select_params.group_idx = group_idx; + seq_state->steps[*seq_state->num_steps].params.tg_set_gsl_source_select_params.gsl_ready_signal = gsl_ready_signal; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_update_mall_sel(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint32_t mall_sel, + bool cache_cursor) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_UPDATE_MALL_SEL; + seq_state->steps[*seq_state->num_steps].params.hubp_update_mall_sel_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_update_mall_sel_params.mall_sel = mall_sel; + seq_state->steps[*seq_state->num_steps].params.hubp_update_mall_sel_params.cache_cursor = cache_cursor; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_prepare_subvp_buffering(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_PREPARE_SUBVP_BUFFERING; + seq_state->steps[*seq_state->num_steps].params.hubp_prepare_subvp_buffering_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_prepare_subvp_buffering_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_set_blank_en(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_BLANK_EN; + seq_state->steps[*seq_state->num_steps].params.hubp_set_blank_en_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_set_blank_en_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_disable_control(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool disable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_DISABLE_CONTROL; + seq_state->steps[*seq_state->num_steps].params.hubp_disable_control_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_disable_control_params.disable = disable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubbub_soft_reset(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + void (*hubbub_soft_reset)(struct hubbub *hubbub, bool reset), + bool reset) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBBUB_SOFT_RESET; + seq_state->steps[*seq_state->num_steps].params.hubbub_soft_reset_params.hubbub = hubbub; + seq_state->steps[*seq_state->num_steps].params.hubbub_soft_reset_params.hubbub_soft_reset = hubbub_soft_reset; + seq_state->steps[*seq_state->num_steps].params.hubbub_soft_reset_params.reset = reset; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_clk_cntl(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_CLK_CNTL; + seq_state->steps[*seq_state->num_steps].params.hubp_clk_cntl_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_clk_cntl_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_dppclk_control(struct block_sequence_state *seq_state, + struct dpp *dpp, + bool dppclk_div, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_DPPCLK_CONTROL; + seq_state->steps[*seq_state->num_steps].params.dpp_dppclk_control_params.dpp = dpp; + seq_state->steps[*seq_state->num_steps].params.dpp_dppclk_control_params.dppclk_div = dppclk_div; + seq_state->steps[*seq_state->num_steps].params.dpp_dppclk_control_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_disable_phantom_crtc(struct block_sequence_state *seq_state, + struct timing_generator *tg) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DISABLE_PHANTOM_CRTC; + seq_state->steps[*seq_state->num_steps].params.disable_phantom_crtc_params.tg = tg; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_pg_status(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + int dsc_inst, + bool is_ungated) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_PG_STATUS; + seq_state->steps[*seq_state->num_steps].params.dsc_pg_status_params.hws = hws; + seq_state->steps[*seq_state->num_steps].params.dsc_pg_status_params.dsc_inst = dsc_inst; + seq_state->steps[*seq_state->num_steps].params.dsc_pg_status_params.is_ungated = is_ungated; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_wait_disconnect_pending_clear(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc, + bool *is_ungated) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_WAIT_DISCONNECT_PENDING_CLEAR; + seq_state->steps[*seq_state->num_steps].params.dsc_wait_disconnect_pending_clear_params.dsc = dsc; + seq_state->steps[*seq_state->num_steps].params.dsc_wait_disconnect_pending_clear_params.is_ungated = is_ungated; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dsc_disable(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc, + bool *is_ungated) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DSC_DISABLE; + seq_state->steps[*seq_state->num_steps].params.dsc_disable_params.dsc = dsc; + seq_state->steps[*seq_state->num_steps].params.dsc_disable_params.is_ungated = is_ungated; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dccg_set_ref_dscclk(struct block_sequence_state *seq_state, + struct dccg *dccg, + int dsc_inst, + bool *is_ungated) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DCCG_SET_REF_DSCCLK; + seq_state->steps[*seq_state->num_steps].params.dccg_set_ref_dscclk_params.dccg = dccg; + seq_state->steps[*seq_state->num_steps].params.dccg_set_ref_dscclk_params.dsc_inst = dsc_inst; + seq_state->steps[*seq_state->num_steps].params.dccg_set_ref_dscclk_params.is_ungated = is_ungated; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_root_clock_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int dpp_inst, + bool clock_on) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_ROOT_CLOCK_CONTROL; + seq_state->steps[*seq_state->num_steps].params.dpp_root_clock_control_params.hws = hws; + seq_state->steps[*seq_state->num_steps].params.dpp_root_clock_control_params.dpp_inst = dpp_inst; + seq_state->steps[*seq_state->num_steps].params.dpp_root_clock_control_params.clock_on = clock_on; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_pg_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int dpp_inst, + bool power_on) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_PG_CONTROL; + seq_state->steps[*seq_state->num_steps].params.dpp_pg_control_params.hws = hws; + seq_state->steps[*seq_state->num_steps].params.dpp_pg_control_params.dpp_inst = dpp_inst; + seq_state->steps[*seq_state->num_steps].params.dpp_pg_control_params.power_on = power_on; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_pg_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int hubp_inst, + bool power_on) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_PG_CONTROL; + seq_state->steps[*seq_state->num_steps].params.hubp_pg_control_params.hws = hws; + seq_state->steps[*seq_state->num_steps].params.hubp_pg_control_params.hubp_inst = hubp_inst; + seq_state->steps[*seq_state->num_steps].params.hubp_pg_control_params.power_on = power_on; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_init(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_INIT; + seq_state->steps[*seq_state->num_steps].params.hubp_init_params.hubp = hubp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_reset(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_RESET; + seq_state->steps[*seq_state->num_steps].params.hubp_reset_params.hubp = hubp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_reset(struct block_sequence_state *seq_state, + struct dpp *dpp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_RESET; + seq_state->steps[*seq_state->num_steps].params.dpp_reset_params.dpp = dpp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_opp_pipe_clock_control(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + bool enable) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = OPP_PIPE_CLOCK_CONTROL; + seq_state->steps[*seq_state->num_steps].params.opp_pipe_clock_control_params.opp = opp; + seq_state->steps[*seq_state->num_steps].params.opp_pipe_clock_control_params.enable = enable; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_set_vm_system_aperture_settings(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint64_t sys_default, + uint64_t sys_low, + uint64_t sys_high) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_VM_SYSTEM_APERTURE_SETTINGS; + seq_state->steps[*seq_state->num_steps].params.hubp_set_vm_system_aperture_settings_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_set_vm_system_aperture_settings_params.sys_default.quad_part = sys_default; + seq_state->steps[*seq_state->num_steps].params.hubp_set_vm_system_aperture_settings_params.sys_low.quad_part = sys_low; + seq_state->steps[*seq_state->num_steps].params.hubp_set_vm_system_aperture_settings_params.sys_high.quad_part = sys_high; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_set_flip_int(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_FLIP_INT; + seq_state->steps[*seq_state->num_steps].params.hubp_set_flip_int_params.hubp = hubp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dccg_update_dpp_dto(struct block_sequence_state *seq_state, + struct dccg *dccg, + int dpp_inst, + int dppclk_khz) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DCCG_UPDATE_DPP_DTO; + seq_state->steps[*seq_state->num_steps].params.dccg_update_dpp_dto_params.dccg = dccg; + seq_state->steps[*seq_state->num_steps].params.dccg_update_dpp_dto_params.dpp_inst = dpp_inst; + seq_state->steps[*seq_state->num_steps].params.dccg_update_dpp_dto_params.dppclk_khz = dppclk_khz; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_vtg_sel(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint32_t otg_inst) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_VTG_SEL; + seq_state->steps[*seq_state->num_steps].params.hubp_vtg_sel_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_vtg_sel_params.otg_inst = otg_inst; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_setup2(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct dml2_dchub_per_pipe_register_set *hubp_regs, + union dml2_global_sync_programming *global_sync, + struct dc_crtc_timing *timing) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SETUP2; + seq_state->steps[*seq_state->num_steps].params.hubp_setup2_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_setup2_params.hubp_regs = hubp_regs; + seq_state->steps[*seq_state->num_steps].params.hubp_setup2_params.global_sync = global_sync; + seq_state->steps[*seq_state->num_steps].params.hubp_setup2_params.timing = timing; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_setup(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct _vcs_dpi_display_dlg_regs_st *dlg_regs, + struct _vcs_dpi_display_ttu_regs_st *ttu_regs, + struct _vcs_dpi_display_rq_regs_st *rq_regs, + struct _vcs_dpi_display_pipe_dest_params_st *pipe_dest) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SETUP; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_params.dlg_regs = dlg_regs; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_params.ttu_regs = ttu_regs; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_params.rq_regs = rq_regs; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_params.pipe_dest = pipe_dest; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_set_unbounded_requesting(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool unbounded_req) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SET_UNBOUNDED_REQUESTING; + seq_state->steps[*seq_state->num_steps].params.hubp_set_unbounded_requesting_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_set_unbounded_requesting_params.unbounded_req = unbounded_req; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_setup_interdependent2(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct dml2_dchub_per_pipe_register_set *hubp_regs) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SETUP_INTERDEPENDENT2; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_interdependent2_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_interdependent2_params.hubp_regs = hubp_regs; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_setup_interdependent(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct _vcs_dpi_display_dlg_regs_st *dlg_regs, + struct _vcs_dpi_display_ttu_regs_st *ttu_regs) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_SETUP_INTERDEPENDENT; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_interdependent_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_interdependent_params.dlg_regs = dlg_regs; + seq_state->steps[*seq_state->num_steps].params.hubp_setup_interdependent_params.ttu_regs = ttu_regs; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_program_surface_config(struct block_sequence_state *seq_state, + struct hubp *hubp, + enum surface_pixel_format format, + struct dc_tiling_info *tiling_info, + struct plane_size plane_size, + enum dc_rotation_angle rotation, + struct dc_plane_dcc_param *dcc, + bool horizontal_mirror, + int compat_level) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_PROGRAM_SURFACE_CONFIG; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.format = format; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.tiling_info = tiling_info; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.plane_size = plane_size; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.rotation = rotation; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.dcc = dcc; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.horizontal_mirror = horizontal_mirror; + seq_state->steps[*seq_state->num_steps].params.program_surface_config_params.compat_level = compat_level; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_setup_dpp(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_SETUP_DPP; + seq_state->steps[*seq_state->num_steps].params.setup_dpp_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_set_cursor_matrix(struct block_sequence_state *seq_state, + struct dpp *dpp, + enum dc_color_space color_space, + struct dc_csc_transform *cursor_csc_color_matrix) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_SET_CURSOR_MATRIX; + seq_state->steps[*seq_state->num_steps].params.dpp_set_cursor_matrix_params.dpp = dpp; + seq_state->steps[*seq_state->num_steps].params.dpp_set_cursor_matrix_params.color_space = color_space; + seq_state->steps[*seq_state->num_steps].params.dpp_set_cursor_matrix_params.cursor_csc_color_matrix = cursor_csc_color_matrix; + (*seq_state->num_steps)++; + } +} + +void hwss_add_dpp_set_scaler(struct block_sequence_state *seq_state, + struct dpp *dpp, + const struct scaler_data *scl_data) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = DPP_SET_SCALER; + seq_state->steps[*seq_state->num_steps].params.dpp_set_scaler_params.dpp = dpp; + seq_state->steps[*seq_state->num_steps].params.dpp_set_scaler_params.scl_data = scl_data; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubp_mem_program_viewport(struct block_sequence_state *seq_state, + struct hubp *hubp, + const struct rect *viewport, + const struct rect *viewport_c) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBP_MEM_PROGRAM_VIEWPORT; + seq_state->steps[*seq_state->num_steps].params.hubp_mem_program_viewport_params.hubp = hubp; + seq_state->steps[*seq_state->num_steps].params.hubp_mem_program_viewport_params.viewport = viewport; + seq_state->steps[*seq_state->num_steps].params.hubp_mem_program_viewport_params.viewport_c = viewport_c; + (*seq_state->num_steps)++; + } +} + +void hwss_add_abort_cursor_offload_update(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = ABORT_CURSOR_OFFLOAD_UPDATE; + seq_state->steps[*seq_state->num_steps].params.abort_cursor_offload_update_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.abort_cursor_offload_update_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_set_cursor_attribute(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = SET_CURSOR_ATTRIBUTE; + seq_state->steps[*seq_state->num_steps].params.set_cursor_attribute_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_cursor_attribute_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_set_cursor_position(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = SET_CURSOR_POSITION; + seq_state->steps[*seq_state->num_steps].params.set_cursor_position_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_cursor_position_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_set_cursor_sdr_white_level(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = SET_CURSOR_SDR_WHITE_LEVEL; + seq_state->steps[*seq_state->num_steps].params.set_cursor_sdr_white_level_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.set_cursor_sdr_white_level_params.pipe_ctx = pipe_ctx; + (*seq_state->num_steps)++; + } +} + +void hwss_add_program_output_csc(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + enum dc_color_space colorspace, + uint16_t *matrix, + int opp_id) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = PROGRAM_OUTPUT_CSC; + seq_state->steps[*seq_state->num_steps].params.program_output_csc_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.program_output_csc_params.pipe_ctx = pipe_ctx; + seq_state->steps[*seq_state->num_steps].params.program_output_csc_params.colorspace = colorspace; + seq_state->steps[*seq_state->num_steps].params.program_output_csc_params.matrix = matrix; + seq_state->steps[*seq_state->num_steps].params.program_output_csc_params.opp_id = opp_id; + (*seq_state->num_steps)++; + } +} + +void hwss_add_phantom_hubp_post_enable(struct block_sequence_state *seq_state, + struct hubp *hubp) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = PHANTOM_HUBP_POST_ENABLE; + seq_state->steps[*seq_state->num_steps].params.phantom_hubp_post_enable_params.hubp = hubp; + (*seq_state->num_steps)++; + } +} + +void hwss_add_update_force_pstate(struct block_sequence_state *seq_state, + struct dc *dc, + struct dc_state *context) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = UPDATE_FORCE_PSTATE; + seq_state->steps[*seq_state->num_steps].params.update_force_pstate_params.dc = dc; + seq_state->steps[*seq_state->num_steps].params.update_force_pstate_params.context = context; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubbub_apply_dedcn21_147_wa(struct block_sequence_state *seq_state, + struct hubbub *hubbub) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBBUB_APPLY_DEDCN21_147_WA; + seq_state->steps[*seq_state->num_steps].params.hubbub_apply_dedcn21_147_wa_params.hubbub = hubbub; + (*seq_state->num_steps)++; + } +} + +void hwss_add_hubbub_allow_self_refresh_control(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + bool allow, + bool *disallow_self_refresh_applied) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = HUBBUB_ALLOW_SELF_REFRESH_CONTROL; + seq_state->steps[*seq_state->num_steps].params.hubbub_allow_self_refresh_control_params.hubbub = hubbub; + seq_state->steps[*seq_state->num_steps].params.hubbub_allow_self_refresh_control_params.allow = allow; + seq_state->steps[*seq_state->num_steps].params.hubbub_allow_self_refresh_control_params.disallow_self_refresh_applied = disallow_self_refresh_applied; + (*seq_state->num_steps)++; + } +} + +void hwss_add_tg_get_frame_count(struct block_sequence_state *seq_state, + struct timing_generator *tg, + unsigned int *frame_count) +{ + if (*seq_state->num_steps < MAX_HWSS_BLOCK_SEQUENCE_SIZE) { + seq_state->steps[*seq_state->num_steps].func = TG_GET_FRAME_COUNT; + seq_state->steps[*seq_state->num_steps].params.tg_get_frame_count_params.tg = tg; + seq_state->steps[*seq_state->num_steps].params.tg_get_frame_count_params.frame_count = frame_count; + (*seq_state->num_steps)++; + } +} diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c index a180f68f711c..deb23d20bca6 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_link_enc_cfg.c @@ -522,10 +522,10 @@ struct link_encoder *link_enc_cfg_get_link_enc_used_by_link( struct link_encoder *link_enc_cfg_get_next_avail_link_enc(struct dc *dc) { struct link_encoder *link_enc = NULL; - enum engine_id encs_assigned[MAX_DIG_LINK_ENCODERS]; + enum engine_id encs_assigned[MAX_LINK_ENCODERS]; int i; - for (i = 0; i < MAX_DIG_LINK_ENCODERS; i++) + for (i = 0; i < MAX_LINK_ENCODERS; i++) encs_assigned[i] = ENGINE_ID_UNKNOWN; /* Add assigned encoders to list. */ diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index bc5dedf5f60c..848c267ef11e 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -95,10 +95,44 @@ #define DC_LOGGER \ dc->ctx->logger #define DC_LOGGER_INIT(logger) -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define UNABLE_TO_SPLIT -1 +static void capture_pipe_topology_data(struct dc *dc, int plane_idx, int slice_idx, int stream_idx, + int dpp_inst, int opp_inst, int tg_inst, bool is_phantom_pipe) +{ + struct pipe_topology_snapshot *current_snapshot = &dc->debug_data.topology_history.snapshots[dc->debug_data.topology_history.current_snapshot_index]; + + if (current_snapshot->line_count >= MAX_PIPES) + return; + + current_snapshot->pipe_log_lines[current_snapshot->line_count].is_phantom_pipe = is_phantom_pipe; + current_snapshot->pipe_log_lines[current_snapshot->line_count].plane_idx = plane_idx; + current_snapshot->pipe_log_lines[current_snapshot->line_count].slice_idx = slice_idx; + current_snapshot->pipe_log_lines[current_snapshot->line_count].stream_idx = stream_idx; + current_snapshot->pipe_log_lines[current_snapshot->line_count].dpp_inst = dpp_inst; + current_snapshot->pipe_log_lines[current_snapshot->line_count].opp_inst = opp_inst; + current_snapshot->pipe_log_lines[current_snapshot->line_count].tg_inst = tg_inst; + + current_snapshot->line_count++; +} + +static void start_new_topology_snapshot(struct dc *dc, struct dc_state *state) +{ + // Move to next snapshot slot (circular buffer) + dc->debug_data.topology_history.current_snapshot_index = (dc->debug_data.topology_history.current_snapshot_index + 1) % MAX_TOPOLOGY_SNAPSHOTS; + + // Clear the new snapshot + struct pipe_topology_snapshot *current_snapshot = &dc->debug_data.topology_history.snapshots[dc->debug_data.topology_history.current_snapshot_index]; + memset(current_snapshot, 0, sizeof(*current_snapshot)); + + // Set metadata + current_snapshot->timestamp_us = dm_get_timestamp(dc->ctx); + current_snapshot->stream_count = state->stream_count; + current_snapshot->phantom_stream_count = state->phantom_stream_count; +} + enum dce_version resource_parse_asic_id(struct hw_asic_id asic_id) { enum dce_version dc_version = DCE_VERSION_UNKNOWN; @@ -446,6 +480,14 @@ bool resource_construct( DC_ERR("DC: failed to create stream_encoder!\n"); pool->stream_enc_count++; } + + for (i = 0; i < caps->num_analog_stream_encoder; i++) { + pool->stream_enc[caps->num_stream_encoder + i] = + create_funcs->create_stream_encoder(ENGINE_ID_DACA + i, ctx); + if (pool->stream_enc[caps->num_stream_encoder + i] == NULL) + DC_ERR("DC: failed to create analog stream_encoder %d!\n", i); + pool->stream_enc_count++; + } } pool->hpo_dp_stream_enc_count = 0; @@ -2303,10 +2345,11 @@ bool resource_is_odm_topology_changed(const struct pipe_ctx *otg_master_a, static void resource_log_pipe(struct dc *dc, struct pipe_ctx *pipe, int stream_idx, int slice_idx, int plane_idx, int slice_count, - bool is_primary) + bool is_primary, bool is_phantom_pipe) { DC_LOGGER_INIT(dc->ctx->logger); + // new format for logging: bit storing code if (slice_idx == 0 && plane_idx == 0 && is_primary) { /* case 0 (OTG master pipe with plane) */ DC_LOG_DC(" | plane%d slice%d stream%d|", @@ -2315,6 +2358,10 @@ static void resource_log_pipe(struct dc *dc, struct pipe_ctx *pipe, pipe->plane_res.dpp->inst, pipe->stream_res.opp->inst, pipe->stream_res.tg->inst); + capture_pipe_topology_data(dc, plane_idx, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } else if (slice_idx == 0 && plane_idx == -1) { /* case 1 (OTG master pipe without plane) */ DC_LOG_DC(" | slice%d stream%d|", @@ -2323,6 +2370,10 @@ static void resource_log_pipe(struct dc *dc, struct pipe_ctx *pipe, pipe->stream_res.opp->inst, pipe->stream_res.opp->inst, pipe->stream_res.tg->inst); + capture_pipe_topology_data(dc, 0xF, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } else if (slice_idx != 0 && plane_idx == 0 && is_primary) { /* case 2 (OPP head pipe with plane) */ DC_LOG_DC(" | plane%d slice%d | |", @@ -2330,27 +2381,43 @@ static void resource_log_pipe(struct dc *dc, struct pipe_ctx *pipe, DC_LOG_DC(" |DPP%d----OPP%d----| |", pipe->plane_res.dpp->inst, pipe->stream_res.opp->inst); + capture_pipe_topology_data(dc, plane_idx, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } else if (slice_idx != 0 && plane_idx == -1) { /* case 3 (OPP head pipe without plane) */ DC_LOG_DC(" | slice%d | |", slice_idx); DC_LOG_DC(" |DPG%d----OPP%d----| |", pipe->plane_res.dpp->inst, pipe->stream_res.opp->inst); + capture_pipe_topology_data(dc, 0xF, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } else if (slice_idx == slice_count - 1) { /* case 4 (DPP pipe in last slice) */ DC_LOG_DC(" | plane%d | |", plane_idx); DC_LOG_DC(" |DPP%d----| |", pipe->plane_res.dpp->inst); + capture_pipe_topology_data(dc, plane_idx, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } else { /* case 5 (DPP pipe not in last slice) */ DC_LOG_DC(" | plane%d | | |", plane_idx); DC_LOG_DC(" |DPP%d----| | |", pipe->plane_res.dpp->inst); + capture_pipe_topology_data(dc, plane_idx, slice_idx, stream_idx, + pipe->plane_res.dpp->inst, + pipe->stream_res.opp->inst, + pipe->stream_res.tg->inst, is_phantom_pipe); } } static void resource_log_pipe_for_stream(struct dc *dc, struct dc_state *state, - struct pipe_ctx *otg_master, int stream_idx) + struct pipe_ctx *otg_master, int stream_idx, bool is_phantom_pipe) { struct pipe_ctx *opp_heads[MAX_PIPES]; struct pipe_ctx *dpp_pipes[MAX_PIPES]; @@ -2376,12 +2443,12 @@ static void resource_log_pipe_for_stream(struct dc *dc, struct dc_state *state, resource_log_pipe(dc, dpp_pipes[dpp_idx], stream_idx, slice_idx, plane_idx, slice_count, - is_primary); + is_primary, is_phantom_pipe); } } else { resource_log_pipe(dc, opp_heads[slice_idx], stream_idx, slice_idx, plane_idx, - slice_count, true); + slice_count, true, is_phantom_pipe); } } @@ -2412,6 +2479,10 @@ void resource_log_pipe_topology_update(struct dc *dc, struct dc_state *state) struct pipe_ctx *otg_master; int stream_idx, phantom_stream_idx; DC_LOGGER_INIT(dc->ctx->logger); + bool is_phantom_pipe = false; + + // Start a new snapshot for this topology update + start_new_topology_snapshot(dc, state); DC_LOG_DC(" pipe topology update"); DC_LOG_DC(" ________________________"); @@ -2425,9 +2496,10 @@ void resource_log_pipe_topology_update(struct dc *dc, struct dc_state *state) if (!otg_master) continue; - resource_log_pipe_for_stream(dc, state, otg_master, stream_idx); + resource_log_pipe_for_stream(dc, state, otg_master, stream_idx, is_phantom_pipe); } if (state->phantom_stream_count > 0) { + is_phantom_pipe = true; DC_LOG_DC(" | (phantom pipes) |"); for (stream_idx = 0; stream_idx < state->stream_count; stream_idx++) { if (state->stream_status[stream_idx].mall_stream_config.type != SUBVP_MAIN) @@ -2440,7 +2512,7 @@ void resource_log_pipe_topology_update(struct dc *dc, struct dc_state *state) if (!otg_master) continue; - resource_log_pipe_for_stream(dc, state, otg_master, stream_idx); + resource_log_pipe_for_stream(dc, state, otg_master, stream_idx, is_phantom_pipe); } } DC_LOG_DC(" |________________________|\n"); @@ -2690,17 +2762,40 @@ static inline int find_fixed_dio_link_enc(const struct dc_link *link) } static inline int find_free_dio_link_enc(const struct resource_context *res_ctx, - const struct dc_link *link, const struct resource_pool *pool) + const struct dc_link *link, const struct resource_pool *pool, struct dc_stream_state *stream) { - int i; + int i, j = -1; + int stream_enc_inst = -1; int enc_count = pool->dig_link_enc_count; - /* for dpia, check preferred encoder first and then the next one */ - for (i = 0; i < enc_count; i++) - if (res_ctx->dio_link_enc_ref_cnts[(link->dpia_preferred_eng_id + i) % enc_count] == 0) - break; + /* Find stream encoder instance for the stream */ + if (stream) { + for (i = 0; i < pool->pipe_count; i++) { + if ((res_ctx->pipe_ctx[i].stream == stream) && + (res_ctx->pipe_ctx[i].stream_res.stream_enc != NULL)) { + stream_enc_inst = res_ctx->pipe_ctx[i].stream_res.stream_enc->id; + break; + } + } + } + + /* Assign dpia preferred > stream enc instance > available */ + for (i = 0; i < enc_count; i++) { + if (res_ctx->dio_link_enc_ref_cnts[i] == 0) { + if (j == -1) + j = i; + + if (link->dpia_preferred_eng_id == i) { + j = i; + break; + } - return (i >= 0 && i < enc_count) ? (link->dpia_preferred_eng_id + i) % enc_count : -1; + if (stream_enc_inst == i) { + j = stream_enc_inst; + } + } + } + return j; } static inline void acquire_dio_link_enc( @@ -2781,7 +2876,7 @@ static bool add_dio_link_enc_to_ctx(const struct dc *dc, retain_dio_link_enc(res_ctx, enc_index); } else { if (stream->link->is_dig_mapping_flexible) - enc_index = find_free_dio_link_enc(res_ctx, stream->link, pool); + enc_index = find_free_dio_link_enc(res_ctx, stream->link, pool, stream); else { int link_index = 0; @@ -2791,7 +2886,7 @@ static bool add_dio_link_enc_to_ctx(const struct dc *dc, * one into the acquiring link. */ if (enc_index >= 0 && is_dio_enc_acquired_by_other_link(stream->link, enc_index, &link_index)) { - int new_enc_index = find_free_dio_link_enc(res_ctx, dc->links[link_index], pool); + int new_enc_index = find_free_dio_link_enc(res_ctx, dc->links[link_index], pool, stream); if (new_enc_index >= 0) swap_dio_link_enc_to_muxable_ctx(context, pool, new_enc_index, enc_index); @@ -5201,7 +5296,7 @@ struct link_encoder *get_temp_dio_link_enc( enc_index = link->eng_id; if (enc_index < 0) - enc_index = find_free_dio_link_enc(res_ctx, link, pool); + enc_index = find_free_dio_link_enc(res_ctx, link, pool, NULL); if (enc_index >= 0) link_enc = pool->link_encoders[enc_index]; diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_state.c b/drivers/gpu/drm/amd/display/dc/core/dc_state.c index c61300a7cb1c..2de8ef4a58ec 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_state.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_state.c @@ -35,8 +35,8 @@ #include "link_enc_cfg.h" #if defined(CONFIG_DRM_AMD_DC_FP) -#include "dml2/dml2_wrapper.h" -#include "dml2/dml2_internal_types.h" +#include "dml2_0/dml2_wrapper.h" +#include "dml2_0/dml2_internal_types.h" #endif #define DC_LOGGER \ diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c index 9ac2d41f8fca..129cd5f84983 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_stream.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_stream.c @@ -224,6 +224,14 @@ struct dc_stream_status *dc_stream_get_status( return dc_state_get_stream_status(dc->current_state, stream); } +const struct dc_stream_status *dc_stream_get_status_const( + const struct dc_stream_state *stream) +{ + struct dc *dc = stream->ctx->dc; + + return dc_state_get_stream_status(dc->current_state, stream); +} + void program_cursor_attributes( struct dc *dc, struct dc_stream_state *stream) @@ -231,6 +239,7 @@ void program_cursor_attributes( int i; struct resource_context *res_ctx; struct pipe_ctx *pipe_to_program = NULL; + bool enable_cursor_offload = dc_dmub_srv_is_cursor_offload_enabled(dc); if (!stream) return; @@ -245,9 +254,14 @@ void program_cursor_attributes( if (!pipe_to_program) { pipe_to_program = pipe_ctx; - dc->hwss.cursor_lock(dc, pipe_to_program, true); - if (pipe_to_program->next_odm_pipe) - dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, true); + + if (enable_cursor_offload && dc->hwss.begin_cursor_offload_update) { + dc->hwss.begin_cursor_offload_update(dc, pipe_ctx); + } else { + dc->hwss.cursor_lock(dc, pipe_to_program, true); + if (pipe_to_program->next_odm_pipe) + dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, true); + } } dc->hwss.set_cursor_attribute(pipe_ctx); @@ -255,12 +269,18 @@ void program_cursor_attributes( dc_send_update_cursor_info_to_dmu(pipe_ctx, i); if (dc->hwss.set_cursor_sdr_white_level) dc->hwss.set_cursor_sdr_white_level(pipe_ctx); + if (enable_cursor_offload && dc->hwss.update_cursor_offload_pipe) + dc->hwss.update_cursor_offload_pipe(dc, pipe_ctx); } if (pipe_to_program) { - dc->hwss.cursor_lock(dc, pipe_to_program, false); - if (pipe_to_program->next_odm_pipe) - dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, false); + if (enable_cursor_offload && dc->hwss.commit_cursor_offload_update) { + dc->hwss.commit_cursor_offload_update(dc, pipe_to_program); + } else { + dc->hwss.cursor_lock(dc, pipe_to_program, false); + if (pipe_to_program->next_odm_pipe) + dc->hwss.cursor_lock(dc, pipe_to_program->next_odm_pipe, false); + } } } @@ -366,6 +386,7 @@ void program_cursor_position( int i; struct resource_context *res_ctx; struct pipe_ctx *pipe_to_program = NULL; + bool enable_cursor_offload = dc_dmub_srv_is_cursor_offload_enabled(dc); if (!stream) return; @@ -384,16 +405,27 @@ void program_cursor_position( if (!pipe_to_program) { pipe_to_program = pipe_ctx; - dc->hwss.cursor_lock(dc, pipe_to_program, true); + + if (enable_cursor_offload && dc->hwss.begin_cursor_offload_update) + dc->hwss.begin_cursor_offload_update(dc, pipe_ctx); + else + dc->hwss.cursor_lock(dc, pipe_to_program, true); } dc->hwss.set_cursor_position(pipe_ctx); + if (enable_cursor_offload && dc->hwss.update_cursor_offload_pipe) + dc->hwss.update_cursor_offload_pipe(dc, pipe_ctx); + if (dc->ctx->dmub_srv) dc_send_update_cursor_info_to_dmu(pipe_ctx, i); } - if (pipe_to_program) - dc->hwss.cursor_lock(dc, pipe_to_program, false); + if (pipe_to_program) { + if (enable_cursor_offload && dc->hwss.commit_cursor_offload_update) + dc->hwss.commit_cursor_offload_update(dc, pipe_to_program); + else + dc->hwss.cursor_lock(dc, pipe_to_program, false); + } } bool dc_stream_set_cursor_position( @@ -705,9 +737,14 @@ bool dc_stream_get_scanoutpos(const struct dc_stream_state *stream, { uint8_t i; bool ret = false; - struct dc *dc = stream->ctx->dc; - struct resource_context *res_ctx = - &dc->current_state->res_ctx; + struct dc *dc; + struct resource_context *res_ctx; + + if (!stream->ctx) + return false; + + dc = stream->ctx->dc; + res_ctx = &dc->current_state->res_ctx; dc_exit_ips_for_hw_access(dc); @@ -855,9 +892,11 @@ void dc_stream_log(const struct dc *dc, const struct dc_stream_state *stream) stream->sink->sink_signal != SIGNAL_TYPE_NONE) { DC_LOG_DC( - "\tdispname: %s signal: %x\n", + "\tsignal: %x dispname: %s manufacturer_id: 0x%x product_id: 0x%x\n", + stream->signal, stream->sink->edid_caps.display_name, - stream->signal); + stream->sink->edid_caps.manufacturer_id, + stream->sink->edid_caps.product_id); } } } diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 98f0b6b3c213..29edfa51ea2c 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -42,7 +42,7 @@ #include "inc/hw/dmcu.h" #include "dml/display_mode_lib.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dmub/inc/dmub_cmd.h" @@ -54,8 +54,16 @@ struct abm_save_restore; struct aux_payload; struct set_config_cmd_payload; struct dmub_notification; +struct dcn_hubbub_reg_state; +struct dcn_hubp_reg_state; +struct dcn_dpp_reg_state; +struct dcn_mpc_reg_state; +struct dcn_opp_reg_state; +struct dcn_dsc_reg_state; +struct dcn_optc_reg_state; +struct dcn_dccg_reg_state; -#define DC_VER "3.2.351" +#define DC_VER "3.2.359" /** * MAX_SURFACES - representative of the upper bound of surfaces that can be piped to a single CRTC @@ -278,6 +286,15 @@ struct dc_scl_caps { bool sharpener_support; }; +struct dc_check_config { + /** + * max video plane width that can be safely assumed to be always + * supported by single DPP pipe. + */ + unsigned int max_optimizable_video_width; + bool enable_legacy_fast_update; +}; + struct dc_caps { uint32_t max_streams; uint32_t max_links; @@ -293,11 +310,6 @@ struct dc_caps { unsigned int max_cursor_size; unsigned int max_buffered_cursor_size; unsigned int max_video_width; - /* - * max video plane width that can be safely assumed to be always - * supported by single DPP pipe. - */ - unsigned int max_optimizable_video_width; unsigned int min_horizontal_blanking_period; int linear_pitch_alignment; bool dcc_const_color; @@ -455,6 +467,18 @@ enum surface_update_type { UPDATE_TYPE_FULL, /* may need to shuffle resources */ }; +enum dc_lock_descriptor { + LOCK_DESCRIPTOR_NONE = 0x0, + LOCK_DESCRIPTOR_STREAM = 0x1, + LOCK_DESCRIPTOR_LINK = 0x2, + LOCK_DESCRIPTOR_GLOBAL = 0x4, +}; + +struct surface_update_descriptor { + enum surface_update_type update_type; + enum dc_lock_descriptor lock_descriptor; +}; + /* Forward declaration*/ struct dc; struct dc_plane_state; @@ -530,6 +554,7 @@ struct dc_config { bool set_pipe_unlock_order; bool enable_dpia_pre_training; bool unify_link_enc_assignment; + bool enable_cursor_offload; struct spl_sharpness_range dcn_sharpness_range; struct spl_sharpness_range dcn_override_sharpness_range; }; @@ -849,8 +874,7 @@ union dpia_debug_options { uint32_t enable_force_tbt3_work_around:1; /* bit 4 */ uint32_t disable_usb4_pm_support:1; /* bit 5 */ uint32_t enable_usb4_bw_zero_alloc_patch:1; /* bit 6 */ - uint32_t enable_bw_allocation_mode:1; /* bit 7 */ - uint32_t reserved:24; + uint32_t reserved:25; } bits; uint32_t raw; }; @@ -875,6 +899,7 @@ struct dc_debug_data { uint32_t ltFailCount; uint32_t i2cErrorCount; uint32_t auxErrorCount; + struct pipe_topology_history topology_history; }; struct dc_phy_addr_space_config { @@ -1120,7 +1145,6 @@ struct dc_debug_options { uint32_t fpo_vactive_min_active_margin_us; uint32_t fpo_vactive_max_blank_us; bool enable_hpo_pg_support; - bool enable_legacy_fast_update; bool disable_dc_mode_overwrite; bool replay_skip_crtc_disabled; bool ignore_pg;/*do nothing, let pmfw control it*/ @@ -1152,7 +1176,6 @@ struct dc_debug_options { bool enable_ips_visual_confirm; unsigned int sharpen_policy; unsigned int scale_to_sharpness_policy; - bool skip_full_updated_if_possible; unsigned int enable_oled_edp_power_up_opt; bool enable_hblank_borrow; bool force_subvp_df_throttle; @@ -1164,6 +1187,7 @@ struct dc_debug_options { unsigned int auxless_alpm_lfps_t1t2_us; short auxless_alpm_lfps_t1t2_offset_us; bool disable_stutter_for_wm_program; + bool enable_block_sequence_programming; }; @@ -1702,6 +1726,7 @@ struct dc { struct dc_debug_options debug; struct dc_versions versions; struct dc_caps caps; + struct dc_check_config check_config; struct dc_cap_funcs cap_funcs; struct dc_config config; struct dc_bounding_box_overrides bb_overrides; @@ -1830,20 +1855,26 @@ struct dc_surface_update { }; struct dc_underflow_debug_data { - uint32_t otg_inst; - uint32_t otg_underflow; - uint32_t h_position; - uint32_t v_position; - uint32_t otg_frame_count; - struct dc_underflow_per_hubp_debug_data { - uint32_t hubp_underflow; - uint32_t hubp_in_blank; - uint32_t hubp_readline; - uint32_t det_config_error; - } hubps[MAX_PIPES]; - uint32_t curr_det_sizes[MAX_PIPES]; - uint32_t target_det_sizes[MAX_PIPES]; - uint32_t compbuf_config_error; + struct dcn_hubbub_reg_state *hubbub_reg_state; + struct dcn_hubp_reg_state *hubp_reg_state[MAX_PIPES]; + struct dcn_dpp_reg_state *dpp_reg_state[MAX_PIPES]; + struct dcn_mpc_reg_state *mpc_reg_state[MAX_PIPES]; + struct dcn_opp_reg_state *opp_reg_state[MAX_PIPES]; + struct dcn_dsc_reg_state *dsc_reg_state[MAX_PIPES]; + struct dcn_optc_reg_state *optc_reg_state[MAX_PIPES]; + struct dcn_dccg_reg_state *dccg_reg_state[MAX_PIPES]; +}; + +struct power_features { + bool ips; + bool rcg; + bool replay; + bool dds; + bool sprs; + bool psr; + bool fams; + bool mpo; + bool uclk_p_state; }; /* @@ -2688,6 +2719,13 @@ bool dc_process_dmub_aux_transfer_async(struct dc *dc, uint32_t link_index, struct aux_payload *payload); +/* + * smart power OLED Interfaces + */ +bool dc_smart_power_oled_enable(const struct dc_link *link, bool enable, uint16_t peak_nits, + uint8_t debug_control, uint16_t fixed_CLL, uint32_t triggerline); +bool dc_smart_power_oled_get_max_cll(const struct dc_link *link, unsigned int *pCurrent_MaxCLL); + /* Get dc link index from dpia port index */ uint8_t get_link_index_from_dpia_port_index(const struct dc *dc, uint8_t dpia_port_index); @@ -2721,6 +2759,8 @@ unsigned int dc_get_det_buffer_size_from_state(const struct dc_state *context); bool dc_get_host_router_index(const struct dc_link *link, unsigned int *host_router_index); +void dc_log_preos_dmcub_info(const struct dc *dc); + /* DSC Interfaces */ #include "dc_dsc.h" @@ -2736,7 +2776,7 @@ bool dc_is_timing_changed(struct dc_stream_state *cur_stream, struct dc_stream_state *new_stream); bool dc_is_cursor_limit_pending(struct dc *dc); -bool dc_can_clear_cursor_limit(struct dc *dc); +bool dc_can_clear_cursor_limit(const struct dc *dc); /** * dc_get_underflow_debug_data_for_otg() - Retrieve underflow debug data. @@ -2751,4 +2791,493 @@ bool dc_can_clear_cursor_limit(struct dc *dc); */ void dc_get_underflow_debug_data_for_otg(struct dc *dc, int primary_otg_inst, struct dc_underflow_debug_data *out_data); +void dc_get_power_feature_status(struct dc *dc, int primary_otg_inst, struct power_features *out_data); + +/** + * Software state variables used to program register fields across the display pipeline + */ +struct dc_register_software_state { + /* HUBP register programming variables for each pipe */ + struct { + bool valid_plane_state; + bool valid_stream; + bool min_dc_gfx_version9; + uint32_t vtg_sel; /* DCHUBP_CNTL->HUBP_VTG_SEL from pipe_ctx->stream_res.tg->inst */ + uint32_t hubp_clock_enable; /* HUBP_CLK_CNTL->HUBP_CLOCK_ENABLE from power management */ + uint32_t surface_pixel_format; /* DCSURF_SURFACE_CONFIG->SURFACE_PIXEL_FORMAT from plane_state->format */ + uint32_t rotation_angle; /* DCSURF_SURFACE_CONFIG->ROTATION_ANGLE from plane_state->rotation */ + uint32_t h_mirror_en; /* DCSURF_SURFACE_CONFIG->H_MIRROR_EN from plane_state->horizontal_mirror */ + uint32_t surface_dcc_en; /* DCSURF_SURFACE_CONTROL->PRIMARY_SURFACE_DCC_EN from dcc->enable */ + uint32_t surface_size_width; /* HUBP_SIZE->SURFACE_SIZE_WIDTH from plane_size.surface_size.width */ + uint32_t surface_size_height; /* HUBP_SIZE->SURFACE_SIZE_HEIGHT from plane_size.surface_size.height */ + uint32_t pri_viewport_width; /* DCSURF_PRI_VIEWPORT_DIMENSION->PRI_VIEWPORT_WIDTH from scaler_data.viewport.width */ + uint32_t pri_viewport_height; /* DCSURF_PRI_VIEWPORT_DIMENSION->PRI_VIEWPORT_HEIGHT from scaler_data.viewport.height */ + uint32_t pri_viewport_x_start; /* DCSURF_PRI_VIEWPORT_START->PRI_VIEWPORT_X_START from scaler_data.viewport.x */ + uint32_t pri_viewport_y_start; /* DCSURF_PRI_VIEWPORT_START->PRI_VIEWPORT_Y_START from scaler_data.viewport.y */ + uint32_t cursor_enable; /* CURSOR_CONTROL->CURSOR_ENABLE from cursor_attributes.enable */ + uint32_t cursor_width; /* CURSOR_SETTINGS->CURSOR_WIDTH from cursor_position.width */ + uint32_t cursor_height; /* CURSOR_SETTINGS->CURSOR_HEIGHT from cursor_position.height */ + + /* Additional DCC configuration */ + uint32_t surface_dcc_ind_64b_blk; /* DCSURF_SURFACE_CONTROL->PRIMARY_SURFACE_DCC_IND_64B_BLK from dcc.independent_64b_blks */ + uint32_t surface_dcc_ind_128b_blk; /* DCSURF_SURFACE_CONTROL->PRIMARY_SURFACE_DCC_IND_128B_BLK from dcc.independent_128b_blks */ + + /* Surface pitch configuration */ + uint32_t surface_pitch; /* DCSURF_SURFACE_PITCH->PITCH from plane_size.surface_pitch */ + uint32_t meta_pitch; /* DCSURF_SURFACE_PITCH->META_PITCH from dcc.meta_pitch */ + uint32_t chroma_pitch; /* DCSURF_SURFACE_PITCH_C->PITCH_C from plane_size.chroma_pitch */ + uint32_t meta_pitch_c; /* DCSURF_SURFACE_PITCH_C->META_PITCH_C from dcc.meta_pitch_c */ + + /* Surface addresses */ + uint32_t primary_surface_address_low; /* DCSURF_PRIMARY_SURFACE_ADDRESS->PRIMARY_SURFACE_ADDRESS from address.grph.addr.low_part */ + uint32_t primary_surface_address_high; /* DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH->PRIMARY_SURFACE_ADDRESS_HIGH from address.grph.addr.high_part */ + uint32_t primary_meta_surface_address_low; /* DCSURF_PRIMARY_META_SURFACE_ADDRESS->PRIMARY_META_SURFACE_ADDRESS from address.grph.meta_addr.low_part */ + uint32_t primary_meta_surface_address_high; /* DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH->PRIMARY_META_SURFACE_ADDRESS_HIGH from address.grph.meta_addr.high_part */ + + /* TMZ configuration */ + uint32_t primary_surface_tmz; /* DCSURF_SURFACE_CONTROL->PRIMARY_SURFACE_TMZ from address.tmz_surface */ + uint32_t primary_meta_surface_tmz; /* DCSURF_SURFACE_CONTROL->PRIMARY_META_SURFACE_TMZ from address.tmz_surface */ + + /* Tiling configuration */ + uint32_t sw_mode; /* DCSURF_TILING_CONFIG->SW_MODE from tiling_info.gfx9.swizzle */ + uint32_t num_pipes; /* DCSURF_ADDR_CONFIG->NUM_PIPES from tiling_info.gfx9.num_pipes */ + uint32_t num_banks; /* DCSURF_ADDR_CONFIG->NUM_BANKS from tiling_info.gfx9.num_banks */ + uint32_t pipe_interleave; /* DCSURF_ADDR_CONFIG->PIPE_INTERLEAVE from tiling_info.gfx9.pipe_interleave */ + uint32_t num_shader_engines; /* DCSURF_ADDR_CONFIG->NUM_SE from tiling_info.gfx9.num_shader_engines */ + uint32_t num_rb_per_se; /* DCSURF_ADDR_CONFIG->NUM_RB_PER_SE from tiling_info.gfx9.num_rb_per_se */ + uint32_t num_pkrs; /* DCSURF_ADDR_CONFIG->NUM_PKRS from tiling_info.gfx9.num_pkrs */ + + /* DML Request Size Configuration - Luma */ + uint32_t rq_chunk_size; /* DCHUBP_REQ_SIZE_CONFIG->CHUNK_SIZE from rq_regs.rq_regs_l.chunk_size */ + uint32_t rq_min_chunk_size; /* DCHUBP_REQ_SIZE_CONFIG->MIN_CHUNK_SIZE from rq_regs.rq_regs_l.min_chunk_size */ + uint32_t rq_meta_chunk_size; /* DCHUBP_REQ_SIZE_CONFIG->META_CHUNK_SIZE from rq_regs.rq_regs_l.meta_chunk_size */ + uint32_t rq_min_meta_chunk_size; /* DCHUBP_REQ_SIZE_CONFIG->MIN_META_CHUNK_SIZE from rq_regs.rq_regs_l.min_meta_chunk_size */ + uint32_t rq_dpte_group_size; /* DCHUBP_REQ_SIZE_CONFIG->DPTE_GROUP_SIZE from rq_regs.rq_regs_l.dpte_group_size */ + uint32_t rq_mpte_group_size; /* DCHUBP_REQ_SIZE_CONFIG->MPTE_GROUP_SIZE from rq_regs.rq_regs_l.mpte_group_size */ + uint32_t rq_swath_height_l; /* DCHUBP_REQ_SIZE_CONFIG->SWATH_HEIGHT_L from rq_regs.rq_regs_l.swath_height */ + uint32_t rq_pte_row_height_l; /* DCHUBP_REQ_SIZE_CONFIG->PTE_ROW_HEIGHT_L from rq_regs.rq_regs_l.pte_row_height */ + + /* DML Request Size Configuration - Chroma */ + uint32_t rq_chunk_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->CHUNK_SIZE_C from rq_regs.rq_regs_c.chunk_size */ + uint32_t rq_min_chunk_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->MIN_CHUNK_SIZE_C from rq_regs.rq_regs_c.min_chunk_size */ + uint32_t rq_meta_chunk_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->META_CHUNK_SIZE_C from rq_regs.rq_regs_c.meta_chunk_size */ + uint32_t rq_min_meta_chunk_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->MIN_META_CHUNK_SIZE_C from rq_regs.rq_regs_c.min_meta_chunk_size */ + uint32_t rq_dpte_group_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->DPTE_GROUP_SIZE_C from rq_regs.rq_regs_c.dpte_group_size */ + uint32_t rq_mpte_group_size_c; /* DCHUBP_REQ_SIZE_CONFIG_C->MPTE_GROUP_SIZE_C from rq_regs.rq_regs_c.mpte_group_size */ + uint32_t rq_swath_height_c; /* DCHUBP_REQ_SIZE_CONFIG_C->SWATH_HEIGHT_C from rq_regs.rq_regs_c.swath_height */ + uint32_t rq_pte_row_height_c; /* DCHUBP_REQ_SIZE_CONFIG_C->PTE_ROW_HEIGHT_C from rq_regs.rq_regs_c.pte_row_height */ + + /* DML Expansion Modes */ + uint32_t drq_expansion_mode; /* DCN_EXPANSION_MODE->DRQ_EXPANSION_MODE from rq_regs.drq_expansion_mode */ + uint32_t prq_expansion_mode; /* DCN_EXPANSION_MODE->PRQ_EXPANSION_MODE from rq_regs.prq_expansion_mode */ + uint32_t mrq_expansion_mode; /* DCN_EXPANSION_MODE->MRQ_EXPANSION_MODE from rq_regs.mrq_expansion_mode */ + uint32_t crq_expansion_mode; /* DCN_EXPANSION_MODE->CRQ_EXPANSION_MODE from rq_regs.crq_expansion_mode */ + + /* DML DLG parameters - nominal */ + uint32_t dst_y_per_vm_vblank; /* NOM_PARAMETERS_0->DST_Y_PER_VM_VBLANK from dlg_regs.dst_y_per_vm_vblank */ + uint32_t dst_y_per_row_vblank; /* NOM_PARAMETERS_0->DST_Y_PER_ROW_VBLANK from dlg_regs.dst_y_per_row_vblank */ + uint32_t dst_y_per_vm_flip; /* NOM_PARAMETERS_1->DST_Y_PER_VM_FLIP from dlg_regs.dst_y_per_vm_flip */ + uint32_t dst_y_per_row_flip; /* NOM_PARAMETERS_1->DST_Y_PER_ROW_FLIP from dlg_regs.dst_y_per_row_flip */ + + /* DML prefetch settings */ + uint32_t dst_y_prefetch; /* PREFETCH_SETTINS->DST_Y_PREFETCH from dlg_regs.dst_y_prefetch */ + uint32_t vratio_prefetch; /* PREFETCH_SETTINS->VRATIO_PREFETCH from dlg_regs.vratio_prefetch */ + uint32_t vratio_prefetch_c; /* PREFETCH_SETTINS_C->VRATIO_PREFETCH_C from dlg_regs.vratio_prefetch_c */ + + /* TTU parameters */ + uint32_t qos_level_low_wm; /* TTU_CNTL1->QoSLevelLowWaterMark from ttu_regs.qos_level_low_wm */ + uint32_t qos_level_high_wm; /* TTU_CNTL1->QoSLevelHighWaterMark from ttu_regs.qos_level_high_wm */ + uint32_t qos_level_flip; /* TTU_CNTL2->QoS_LEVEL_FLIP_L from ttu_regs.qos_level_flip */ + uint32_t min_ttu_vblank; /* DCN_GLOBAL_TTU_CNTL->MIN_TTU_VBLANK from ttu_regs.min_ttu_vblank */ + } hubp[MAX_PIPES]; + + /* HUBBUB register programming variables */ + struct { + /* Individual DET buffer control per pipe - software state that programs DET registers */ + uint32_t det0_size; /* DCHUBBUB_DET0_CTRL->DET0_SIZE from hubbub->funcs->program_det_size(hubbub, 0, det_buffer_size_kb) */ + uint32_t det1_size; /* DCHUBBUB_DET1_CTRL->DET1_SIZE from hubbub->funcs->program_det_size(hubbub, 1, det_buffer_size_kb) */ + uint32_t det2_size; /* DCHUBBUB_DET2_CTRL->DET2_SIZE from hubbub->funcs->program_det_size(hubbub, 2, det_buffer_size_kb) */ + uint32_t det3_size; /* DCHUBBUB_DET3_CTRL->DET3_SIZE from hubbub->funcs->program_det_size(hubbub, 3, det_buffer_size_kb) */ + + /* Compression buffer control - software state that programs COMPBUF registers */ + uint32_t compbuf_size; /* DCHUBBUB_COMPBUF_CTRL->COMPBUF_SIZE from hubbub->funcs->program_compbuf_size(hubbub, compbuf_size_kb, safe_to_increase) */ + uint32_t compbuf_reserved_space_64b; /* COMPBUF_RESERVED_SPACE->COMPBUF_RESERVED_SPACE_64B from hubbub2->pixel_chunk_size / 32 */ + uint32_t compbuf_reserved_space_zs; /* COMPBUF_RESERVED_SPACE->COMPBUF_RESERVED_SPACE_ZS from hubbub2->pixel_chunk_size / 128 */ + } hubbub; + + /* DPP register programming variables for each pipe (simplified for available fields) */ + struct { + uint32_t dpp_clock_enable; /* DPP_CONTROL->DPP_CLOCK_ENABLE from dppclk_enable */ + + /* Recout (Rectangle of Interest) configuration */ + uint32_t recout_start_x; /* RECOUT_START->RECOUT_START_X from pipe_ctx->plane_res.scl_data.recout.x */ + uint32_t recout_start_y; /* RECOUT_START->RECOUT_START_Y from pipe_ctx->plane_res.scl_data.recout.y */ + uint32_t recout_width; /* RECOUT_SIZE->RECOUT_WIDTH from pipe_ctx->plane_res.scl_data.recout.width */ + uint32_t recout_height; /* RECOUT_SIZE->RECOUT_HEIGHT from pipe_ctx->plane_res.scl_data.recout.height */ + + /* MPC (Multiple Pipe/Plane Combiner) size configuration */ + uint32_t mpc_width; /* MPC_SIZE->MPC_WIDTH from pipe_ctx->plane_res.scl_data.h_active */ + uint32_t mpc_height; /* MPC_SIZE->MPC_HEIGHT from pipe_ctx->plane_res.scl_data.v_active */ + + /* DSCL mode configuration */ + uint32_t dscl_mode; /* SCL_MODE->DSCL_MODE from pipe_ctx->plane_res.scl_data.dscl_prog_data.dscl_mode */ + + /* Scaler ratios (simplified to integer parts) */ + uint32_t horz_ratio_int; /* SCL_HORZ_FILTER_SCALE_RATIO->SCL_H_SCALE_RATIO integer part from ratios.horz */ + uint32_t vert_ratio_int; /* SCL_VERT_FILTER_SCALE_RATIO->SCL_V_SCALE_RATIO integer part from ratios.vert */ + + /* Basic scaler taps */ + uint32_t h_taps; /* SCL_TAP_CONTROL->SCL_H_NUM_TAPS from taps.h_taps */ + uint32_t v_taps; /* SCL_TAP_CONTROL->SCL_V_NUM_TAPS from taps.v_taps */ + } dpp[MAX_PIPES]; + + /* DCCG register programming variables */ + struct { + /* Core Display Clock Control */ + uint32_t dispclk_khz; /* DENTIST_DISPCLK_CNTL->DENTIST_DISPCLK_WDIVIDER from clk_mgr.dispclk_khz */ + uint32_t dc_mem_global_pwr_req_dis; /* DC_MEM_GLOBAL_PWR_REQ_CNTL->DC_MEM_GLOBAL_PWR_REQ_DIS from memory power management settings */ + + /* DPP Clock Control - 4 fields per pipe */ + uint32_t dppclk_khz[MAX_PIPES]; /* DPPCLK_CTRL->DPPCLK_R_GATE_DISABLE from dpp_clocks[pipe] */ + uint32_t dppclk_enable[MAX_PIPES]; /* DPPCLK_CTRL->DPPCLK0_EN,DPPCLK1_EN,DPPCLK2_EN,DPPCLK3_EN from dccg31_update_dpp_dto() */ + uint32_t dppclk_dto_enable[MAX_PIPES]; /* DPPCLK_DTO_CTRL->DPPCLK_DTO_ENABLE from dccg->dpp_clock_gated[dpp_inst] state */ + uint32_t dppclk_dto_phase[MAX_PIPES]; /* DPPCLK0_DTO_PARAM->DPPCLK0_DTO_PHASE from phase calculation req_dppclk/ref_dppclk */ + uint32_t dppclk_dto_modulo[MAX_PIPES]; /* DPPCLK0_DTO_PARAM->DPPCLK0_DTO_MODULO from modulo = 0xff */ + + /* DSC Clock Control - 4 fields per DSC resource */ + uint32_t dscclk_khz[MAX_PIPES]; /* DSCCLK_DTO_CTRL->DSCCLK_DTO_ENABLE from dsc_clocks */ + uint32_t dscclk_dto_enable[MAX_PIPES]; /* DSCCLK_DTO_CTRL->DSCCLK0_DTO_ENABLE,DSCCLK1_DTO_ENABLE,DSCCLK2_DTO_ENABLE,DSCCLK3_DTO_ENABLE */ + uint32_t dscclk_dto_phase[MAX_PIPES]; /* DSCCLK0_DTO_PARAM->DSCCLK0_DTO_PHASE from dccg31_enable_dscclk() */ + uint32_t dscclk_dto_modulo[MAX_PIPES]; /* DSCCLK0_DTO_PARAM->DSCCLK0_DTO_MODULO from dccg31_enable_dscclk() */ + + /* Pixel Clock Control - per pipe */ + uint32_t pixclk_khz[MAX_PIPES]; /* PIXCLK_RESYNC_CNTL->PIXCLK_RESYNC_ENABLE from stream.timing.pix_clk_100hz */ + uint32_t otg_pixel_rate_div[MAX_PIPES]; /* OTG_PIXEL_RATE_DIV->OTG_PIXEL_RATE_DIV from OTG pixel rate divider control */ + uint32_t dtbclk_dto_enable[MAX_PIPES]; /* OTG0_PIXEL_RATE_CNTL->DTBCLK_DTO_ENABLE from dccg31_set_dtbclk_dto() */ + uint32_t pipe_dto_src_sel[MAX_PIPES]; /* OTG0_PIXEL_RATE_CNTL->PIPE_DTO_SRC_SEL from dccg31_set_dtbclk_dto() source selection */ + uint32_t dtbclk_dto_div[MAX_PIPES]; /* OTG0_PIXEL_RATE_CNTL->DTBCLK_DTO_DIV from dtbdto_div calculation */ + uint32_t otg_add_pixel[MAX_PIPES]; /* OTG0_PIXEL_RATE_CNTL->OTG_ADD_PIXEL from dccg31_otg_add_pixel() */ + uint32_t otg_drop_pixel[MAX_PIPES]; /* OTG0_PIXEL_RATE_CNTL->OTG_DROP_PIXEL from dccg31_otg_drop_pixel() */ + + /* DTBCLK DTO Control - 4 DTOs */ + uint32_t dtbclk_dto_modulo[4]; /* DTBCLK_DTO0_MODULO->DTBCLK_DTO0_MODULO from dccg31_set_dtbclk_dto() modulo calculation */ + uint32_t dtbclk_dto_phase[4]; /* DTBCLK_DTO0_PHASE->DTBCLK_DTO0_PHASE from phase calculation pixclk_khz/ref_dtbclk_khz */ + uint32_t dtbclk_dto_dbuf_en; /* DTBCLK_DTO_DBUF_EN->DTBCLK DTO data buffer enable */ + + /* DP Stream Clock Control - 4 pipes */ + uint32_t dpstreamclk_enable[MAX_PIPES]; /* DPSTREAMCLK_CNTL->DPSTREAMCLK_PIPE0_EN,DPSTREAMCLK_PIPE1_EN,DPSTREAMCLK_PIPE2_EN,DPSTREAMCLK_PIPE3_EN */ + uint32_t dp_dto_modulo[4]; /* DP_DTO0_MODULO->DP_DTO0_MODULO from DP stream DTO programming */ + uint32_t dp_dto_phase[4]; /* DP_DTO0_PHASE->DP_DTO0_PHASE from DP stream DTO programming */ + uint32_t dp_dto_dbuf_en; /* DP_DTO_DBUF_EN->DP DTO data buffer enable */ + + /* PHY Symbol Clock Control - 5 PHYs (A,B,C,D,E) */ + uint32_t phy_symclk_force_en[5]; /* PHYASYMCLK_CLOCK_CNTL->PHYASYMCLK_FORCE_EN from dccg31_set_physymclk() force_enable */ + uint32_t phy_symclk_force_src_sel[5]; /* PHYASYMCLK_CLOCK_CNTL->PHYASYMCLK_FORCE_SRC_SEL from dccg31_set_physymclk() clk_src */ + uint32_t phy_symclk_gate_disable[5]; /* DCCG_GATE_DISABLE_CNTL2->PHYASYMCLK_GATE_DISABLE from debug.root_clock_optimization.bits.physymclk */ + + /* SYMCLK32 SE Control - 4 instances */ + uint32_t symclk32_se_src_sel[4]; /* SYMCLK32_SE_CNTL->SYMCLK32_SE0_SRC_SEL from dccg31_enable_symclk32_se() with get_phy_mux_symclk() mapping */ + uint32_t symclk32_se_enable[4]; /* SYMCLK32_SE_CNTL->SYMCLK32_SE0_EN from dccg31_enable_symclk32_se() enable */ + uint32_t symclk32_se_gate_disable[4]; /* DCCG_GATE_DISABLE_CNTL3->SYMCLK32_SE0_GATE_DISABLE from debug.root_clock_optimization.bits.symclk32_se */ + + /* SYMCLK32 LE Control - 2 instances */ + uint32_t symclk32_le_src_sel[2]; /* SYMCLK32_LE_CNTL->SYMCLK32_LE0_SRC_SEL from dccg31_enable_symclk32_le() phyd32clk source */ + uint32_t symclk32_le_enable[2]; /* SYMCLK32_LE_CNTL->SYMCLK32_LE0_EN from dccg31_enable_symclk32_le() enable */ + uint32_t symclk32_le_gate_disable[2]; /* DCCG_GATE_DISABLE_CNTL3->SYMCLK32_LE0_GATE_DISABLE from debug.root_clock_optimization.bits.symclk32_le */ + + /* DPIA Clock Control */ + uint32_t dpiaclk_540m_dto_modulo; /* DPIACLK_540M_DTO_MODULO->DPIA 540MHz DTO modulo */ + uint32_t dpiaclk_540m_dto_phase; /* DPIACLK_540M_DTO_PHASE->DPIA 540MHz DTO phase */ + uint32_t dpiaclk_810m_dto_modulo; /* DPIACLK_810M_DTO_MODULO->DPIA 810MHz DTO modulo */ + uint32_t dpiaclk_810m_dto_phase; /* DPIACLK_810M_DTO_PHASE->DPIA 810MHz DTO phase */ + uint32_t dpiaclk_dto_cntl; /* DPIACLK_DTO_CNTL->DPIA clock DTO control */ + uint32_t dpiasymclk_cntl; /* DPIASYMCLK_CNTL->DPIA symbol clock control */ + + /* Clock Gating Control */ + uint32_t dccg_gate_disable_cntl; /* DCCG_GATE_DISABLE_CNTL->Clock gate disable control from dccg31_init() */ + uint32_t dpstreamclk_gate_disable; /* DCCG_GATE_DISABLE_CNTL3->DPSTREAMCLK_GATE_DISABLE from debug.root_clock_optimization.bits.dpstream */ + uint32_t dpstreamclk_root_gate_disable; /* DCCG_GATE_DISABLE_CNTL3->DPSTREAMCLK_ROOT_GATE_DISABLE from debug.root_clock_optimization.bits.dpstream */ + + /* VSync Control */ + uint32_t vsync_cnt_ctrl; /* DCCG_VSYNC_CNT_CTRL->VSync counter control */ + uint32_t vsync_cnt_int_ctrl; /* DCCG_VSYNC_CNT_INT_CTRL->VSync counter interrupt control */ + uint32_t vsync_otg_latch_value[6]; /* DCCG_VSYNC_OTG0_LATCH_VALUE->OTG0 VSync latch value (for OTG0-5) */ + + /* Time Base Control */ + uint32_t microsecond_time_base_div; /* MICROSECOND_TIME_BASE_DIV->Microsecond time base divider */ + uint32_t millisecond_time_base_div; /* MILLISECOND_TIME_BASE_DIV->Millisecond time base divider */ + } dccg; + + /* DSC essential configuration for underflow analysis */ + struct { + /* DSC active state - critical for bandwidth analysis */ + uint32_t dsc_clock_enable; /* DSC enabled - affects bandwidth requirements */ + + /* DSC configuration affecting bandwidth and timing */ + uint32_t dsc_num_slices_h; /* Horizontal slice count - affects throughput */ + uint32_t dsc_num_slices_v; /* Vertical slice count - affects throughput */ + uint32_t dsc_bits_per_pixel; /* Compression ratio - affects bandwidth */ + + /* OPP integration - affects pipeline flow */ + uint32_t dscrm_dsc_forward_enable; /* DSC forwarding to OPP enabled */ + uint32_t dscrm_dsc_opp_pipe_source; /* Which OPP receives DSC output */ + } dsc[MAX_PIPES]; + + /* MPC register programming variables */ + struct { + /* MPCC blending tree and mode control */ + uint32_t mpcc_mode[MAX_PIPES]; /* MPCC_CONTROL->MPCC_MODE from blend_cfg.blend_mode */ + uint32_t mpcc_alpha_blend_mode[MAX_PIPES]; /* MPCC_CONTROL->MPCC_ALPHA_BLND_MODE from blend_cfg.alpha_mode */ + uint32_t mpcc_alpha_multiplied_mode[MAX_PIPES]; /* MPCC_CONTROL->MPCC_ALPHA_MULTIPLIED_MODE from blend_cfg.pre_multiplied_alpha */ + uint32_t mpcc_blnd_active_overlap_only[MAX_PIPES]; /* MPCC_CONTROL->MPCC_BLND_ACTIVE_OVERLAP_ONLY from blend_cfg.overlap_only */ + uint32_t mpcc_global_alpha[MAX_PIPES]; /* MPCC_CONTROL->MPCC_GLOBAL_ALPHA from blend_cfg.global_alpha */ + uint32_t mpcc_global_gain[MAX_PIPES]; /* MPCC_CONTROL->MPCC_GLOBAL_GAIN from blend_cfg.global_gain */ + uint32_t mpcc_bg_bpc[MAX_PIPES]; /* MPCC_CONTROL->MPCC_BG_BPC from background color depth */ + uint32_t mpcc_bot_gain_mode[MAX_PIPES]; /* MPCC_CONTROL->MPCC_BOT_GAIN_MODE from bottom layer gain control */ + + /* MPCC blending tree connections */ + uint32_t mpcc_bot_sel[MAX_PIPES]; /* MPCC_BOT_SEL->MPCC_BOT_SEL from mpcc_state->bot_sel */ + uint32_t mpcc_top_sel[MAX_PIPES]; /* MPCC_TOP_SEL->MPCC_TOP_SEL from mpcc_state->dpp_id */ + + /* MPCC output gamma control */ + uint32_t mpcc_ogam_mode[MAX_PIPES]; /* MPCC_OGAM_CONTROL->MPCC_OGAM_MODE from output gamma mode */ + uint32_t mpcc_ogam_select[MAX_PIPES]; /* MPCC_OGAM_CONTROL->MPCC_OGAM_SELECT from gamma LUT bank selection */ + uint32_t mpcc_ogam_pwl_disable[MAX_PIPES]; /* MPCC_OGAM_CONTROL->MPCC_OGAM_PWL_DISABLE from PWL control */ + + /* MPCC pipe assignment and status */ + uint32_t mpcc_opp_id[MAX_PIPES]; /* MPCC_OPP_ID->MPCC_OPP_ID from mpcc_state->opp_id */ + uint32_t mpcc_idle[MAX_PIPES]; /* MPCC_STATUS->MPCC_IDLE from mpcc idle status */ + uint32_t mpcc_busy[MAX_PIPES]; /* MPCC_STATUS->MPCC_BUSY from mpcc busy status */ + + /* MPC output processing */ + uint32_t mpc_out_csc_mode; /* MPC_OUT_CSC_COEF->MPC_OUT_CSC_MODE from output_csc */ + uint32_t mpc_out_gamma_mode; /* MPC_OUT_GAMMA_LUT->MPC_OUT_GAMMA_MODE from output_gamma */ + } mpc; + + /* OPP register programming variables for each pipe */ + struct { + /* Display Pattern Generator (DPG) Control - 19 fields from DPG_CONTROL register */ + uint32_t dpg_enable; /* DPG_CONTROL->DPG_EN from test_pattern parameter (enable/disable) */ + + /* Format Control (FMT) - 18 fields from FMT_CONTROL register */ + uint32_t fmt_pixel_encoding; /* FMT_CONTROL->FMT_PIXEL_ENCODING from clamping->pixel_encoding */ + uint32_t fmt_subsampling_mode; /* FMT_CONTROL->FMT_SUBSAMPLING_MODE from force_chroma_subsampling_1tap */ + uint32_t fmt_cbcr_bit_reduction_bypass; /* FMT_CONTROL->FMT_CBCR_BIT_REDUCTION_BYPASS from pixel_encoding bypass control */ + uint32_t fmt_stereosync_override; /* FMT_CONTROL->FMT_STEREOSYNC_OVERRIDE from stereo timing override */ + uint32_t fmt_spatial_dither_frame_counter_max; /* FMT_CONTROL->FMT_SPATIAL_DITHER_FRAME_COUNTER_MAX from fmt_bit_depth->flags */ + uint32_t fmt_spatial_dither_frame_counter_bit_swap; /* FMT_CONTROL->FMT_SPATIAL_DITHER_FRAME_COUNTER_BIT_SWAP from dither control */ + uint32_t fmt_truncate_enable; /* FMT_CONTROL->FMT_TRUNCATE_EN from fmt_bit_depth->flags.TRUNCATE_ENABLED */ + uint32_t fmt_truncate_depth; /* FMT_CONTROL->FMT_TRUNCATE_DEPTH from fmt_bit_depth->flags.TRUNCATE_DEPTH */ + uint32_t fmt_truncate_mode; /* FMT_CONTROL->FMT_TRUNCATE_MODE from fmt_bit_depth->flags.TRUNCATE_MODE */ + uint32_t fmt_spatial_dither_enable; /* FMT_CONTROL->FMT_SPATIAL_DITHER_EN from fmt_bit_depth->flags.SPATIAL_DITHER_ENABLED */ + uint32_t fmt_spatial_dither_mode; /* FMT_CONTROL->FMT_SPATIAL_DITHER_MODE from fmt_bit_depth->flags.SPATIAL_DITHER_MODE */ + uint32_t fmt_spatial_dither_depth; /* FMT_CONTROL->FMT_SPATIAL_DITHER_DEPTH from fmt_bit_depth->flags.SPATIAL_DITHER_DEPTH */ + uint32_t fmt_temporal_dither_enable; /* FMT_CONTROL->FMT_TEMPORAL_DITHER_EN from fmt_bit_depth->flags.TEMPORAL_DITHER_ENABLED */ + uint32_t fmt_clamp_data_enable; /* FMT_CONTROL->FMT_CLAMP_DATA_EN from clamping->clamping_range enable */ + uint32_t fmt_clamp_color_format; /* FMT_CONTROL->FMT_CLAMP_COLOR_FORMAT from clamping->color_format */ + uint32_t fmt_dynamic_exp_enable; /* FMT_CONTROL->FMT_DYNAMIC_EXP_EN from color_sp/color_dpth/signal */ + uint32_t fmt_dynamic_exp_mode; /* FMT_CONTROL->FMT_DYNAMIC_EXP_MODE from color space mode mapping */ + uint32_t fmt_bit_depth_control; /* Legacy field - kept for compatibility */ + + /* OPP Pipe Control - 1 field from OPP_PIPE_CONTROL register */ + uint32_t opp_pipe_clock_enable; /* OPP_PIPE_CONTROL->OPP_PIPE_CLOCK_EN from enable parameter (bool) */ + + /* OPP CRC Control - 3 fields from OPP_PIPE_CRC_CONTROL register */ + uint32_t opp_crc_enable; /* OPP_PIPE_CRC_CONTROL->CRC_EN from CRC enable control */ + uint32_t opp_crc_select_source; /* OPP_PIPE_CRC_CONTROL->CRC_SELECT_SOURCE from CRC source selection */ + uint32_t opp_crc_stereo_cont; /* OPP_PIPE_CRC_CONTROL->CRC_STEREO_CONT from stereo continuous CRC */ + + /* Output Buffer (OPPBUF) Control - 6 fields from OPPBUF_CONTROL register */ + uint32_t oppbuf_active_width; /* OPPBUF_CONTROL->OPPBUF_ACTIVE_WIDTH from oppbuf_params->active_width */ + uint32_t oppbuf_pixel_repetition; /* OPPBUF_CONTROL->OPPBUF_PIXEL_REPETITION from oppbuf_params->pixel_repetition */ + uint32_t oppbuf_display_segmentation; /* OPPBUF_CONTROL->OPPBUF_DISPLAY_SEGMENTATION from oppbuf_params->mso_segmentation */ + uint32_t oppbuf_overlap_pixel_num; /* OPPBUF_CONTROL->OPPBUF_OVERLAP_PIXEL_NUM from oppbuf_params->mso_overlap_pixel_num */ + uint32_t oppbuf_3d_vact_space1_size; /* OPPBUF_CONTROL->OPPBUF_3D_VACT_SPACE1_SIZE from 3D timing space1_size */ + uint32_t oppbuf_3d_vact_space2_size; /* OPPBUF_CONTROL->OPPBUF_3D_VACT_SPACE2_SIZE from 3D timing space2_size */ + + /* DSC Forward Config - 3 fields from DSCRM_DSC_FORWARD_CONFIG register */ + uint32_t dscrm_dsc_forward_enable; /* DSCRM_DSC_FORWARD_CONFIG->DSCRM_DSC_FORWARD_EN from DSC forward enable control */ + uint32_t dscrm_dsc_opp_pipe_source; /* DSCRM_DSC_FORWARD_CONFIG->DSCRM_DSC_OPP_PIPE_SOURCE from opp_pipe parameter */ + uint32_t dscrm_dsc_forward_enable_status; /* DSCRM_DSC_FORWARD_CONFIG->DSCRM_DSC_FORWARD_EN_STATUS from DSC forward status (read-only) */ + } opp[MAX_PIPES]; + + /* OPTC register programming variables for each pipe */ + struct { + uint32_t otg_master_inst; + + /* OTG_CONTROL register - 5 fields for OTG control */ + uint32_t otg_master_enable; /* OTG_CONTROL->OTG_MASTER_EN from timing enable/disable control */ + uint32_t otg_disable_point_cntl; /* OTG_CONTROL->OTG_DISABLE_POINT_CNTL from disable timing control */ + uint32_t otg_start_point_cntl; /* OTG_CONTROL->OTG_START_POINT_CNTL from start timing control */ + uint32_t otg_field_number_cntl; /* OTG_CONTROL->OTG_FIELD_NUMBER_CNTL from interlace field control */ + uint32_t otg_out_mux; /* OTG_CONTROL->OTG_OUT_MUX from output mux selection */ + + /* OTG Horizontal Timing - 7 fields */ + uint32_t otg_h_total; /* OTG_H_TOTAL->OTG_H_TOTAL from dc_crtc_timing->h_total */ + uint32_t otg_h_blank_start; /* OTG_H_BLANK_START_END->OTG_H_BLANK_START from dc_crtc_timing->h_front_porch */ + uint32_t otg_h_blank_end; /* OTG_H_BLANK_START_END->OTG_H_BLANK_END from dc_crtc_timing->h_addressable_video_pixel_width */ + uint32_t otg_h_sync_start; /* OTG_H_SYNC_A->OTG_H_SYNC_A_START from dc_crtc_timing->h_sync_width */ + uint32_t otg_h_sync_end; /* OTG_H_SYNC_A->OTG_H_SYNC_A_END from calculated sync end position */ + uint32_t otg_h_sync_polarity; /* OTG_H_SYNC_A_CNTL->OTG_H_SYNC_A_POL from dc_crtc_timing->flags.HSYNC_POSITIVE_POLARITY */ + uint32_t otg_h_timing_div_mode; /* OTG_H_TIMING_CNTL->OTG_H_TIMING_DIV_MODE from horizontal timing division mode */ + + /* OTG Vertical Timing - 7 fields */ + uint32_t otg_v_total; /* OTG_V_TOTAL->OTG_V_TOTAL from dc_crtc_timing->v_total */ + uint32_t otg_v_blank_start; /* OTG_V_BLANK_START_END->OTG_V_BLANK_START from dc_crtc_timing->v_front_porch */ + uint32_t otg_v_blank_end; /* OTG_V_BLANK_START_END->OTG_V_BLANK_END from dc_crtc_timing->v_addressable_video_line_width */ + uint32_t otg_v_sync_start; /* OTG_V_SYNC_A->OTG_V_SYNC_A_START from dc_crtc_timing->v_sync_width */ + uint32_t otg_v_sync_end; /* OTG_V_SYNC_A->OTG_V_SYNC_A_END from calculated sync end position */ + uint32_t otg_v_sync_polarity; /* OTG_V_SYNC_A_CNTL->OTG_V_SYNC_A_POL from dc_crtc_timing->flags.VSYNC_POSITIVE_POLARITY */ + uint32_t otg_v_sync_mode; /* OTG_V_SYNC_A_CNTL->OTG_V_SYNC_MODE from sync mode selection */ + + /* OTG DRR (Dynamic Refresh Rate) Control - 8 fields */ + uint32_t otg_v_total_max; /* OTG_V_TOTAL_MAX->OTG_V_TOTAL_MAX from drr_params->vertical_total_max */ + uint32_t otg_v_total_min; /* OTG_V_TOTAL_MIN->OTG_V_TOTAL_MIN from drr_params->vertical_total_min */ + uint32_t otg_v_total_mid; /* OTG_V_TOTAL_MID->OTG_V_TOTAL_MID from drr_params->vertical_total_mid */ + uint32_t otg_v_total_max_sel; /* OTG_V_TOTAL_CONTROL->OTG_V_TOTAL_MAX_SEL from DRR max selection enable */ + uint32_t otg_v_total_min_sel; /* OTG_V_TOTAL_CONTROL->OTG_V_TOTAL_MIN_SEL from DRR min selection enable */ + uint32_t otg_vtotal_mid_replacing_max_en; /* OTG_V_TOTAL_CONTROL->OTG_VTOTAL_MID_REPLACING_MAX_EN from DRR mid-frame enable */ + uint32_t otg_vtotal_mid_frame_num; /* OTG_V_TOTAL_CONTROL->OTG_VTOTAL_MID_FRAME_NUM from drr_params->vertical_total_mid_frame_num */ + uint32_t otg_set_v_total_min_mask; /* OTG_V_TOTAL_CONTROL->OTG_SET_V_TOTAL_MIN_MASK from DRR trigger mask */ + uint32_t otg_force_lock_on_event; /* OTG_V_TOTAL_CONTROL->OTG_FORCE_LOCK_ON_EVENT from DRR force lock control */ + + /* OPTC Data Source and ODM - 6 fields */ + uint32_t optc_seg0_src_sel; /* OPTC_DATA_SOURCE_SELECT->OPTC_SEG0_SRC_SEL from opp_id[0] ODM segment 0 source */ + uint32_t optc_seg1_src_sel; /* OPTC_DATA_SOURCE_SELECT->OPTC_SEG1_SRC_SEL from opp_id[1] ODM segment 1 source */ + uint32_t optc_seg2_src_sel; /* OPTC_DATA_SOURCE_SELECT->OPTC_SEG2_SRC_SEL from opp_id[2] ODM segment 2 source */ + uint32_t optc_seg3_src_sel; /* OPTC_DATA_SOURCE_SELECT->OPTC_SEG3_SRC_SEL from opp_id[3] ODM segment 3 source */ + uint32_t optc_num_of_input_segment; /* OPTC_DATA_SOURCE_SELECT->OPTC_NUM_OF_INPUT_SEGMENT from opp_cnt-1 number of input segments */ + uint32_t optc_mem_sel; /* OPTC_MEMORY_CONFIG->OPTC_MEM_SEL from memory_mask ODM memory selection */ + + /* OPTC Data Format and DSC - 4 fields */ + uint32_t optc_data_format; /* OPTC_DATA_FORMAT_CONTROL->OPTC_DATA_FORMAT from data format selection */ + uint32_t optc_dsc_mode; /* OPTC_DATA_FORMAT_CONTROL->OPTC_DSC_MODE from dsc_mode parameter */ + uint32_t optc_dsc_bytes_per_pixel; /* OPTC_BYTES_PER_PIXEL->OPTC_DSC_BYTES_PER_PIXEL from dsc_bytes_per_pixel parameter */ + uint32_t optc_segment_width; /* OPTC_WIDTH_CONTROL->OPTC_SEGMENT_WIDTH from segment_width parameter */ + uint32_t optc_dsc_slice_width; /* OPTC_WIDTH_CONTROL->OPTC_DSC_SLICE_WIDTH from dsc_slice_width parameter */ + + /* OPTC Clock and Underflow Control - 4 fields */ + uint32_t optc_input_pix_clk_en; /* OPTC_INPUT_CLOCK_CONTROL->OPTC_INPUT_PIX_CLK_EN from pixel clock enable */ + uint32_t optc_underflow_occurred_status; /* OPTC_INPUT_GLOBAL_CONTROL->OPTC_UNDERFLOW_OCCURRED_STATUS from underflow status (read-only) */ + uint32_t optc_underflow_clear; /* OPTC_INPUT_GLOBAL_CONTROL->OPTC_UNDERFLOW_CLEAR from underflow clear control */ + uint32_t otg_clock_enable; /* OTG_CLOCK_CONTROL->OTG_CLOCK_EN from OTG clock enable */ + uint32_t otg_clock_gate_dis; /* OTG_CLOCK_CONTROL->OTG_CLOCK_GATE_DIS from clock gate disable */ + + /* OTG Stereo and 3D Control - 6 fields */ + uint32_t otg_stereo_enable; /* OTG_STEREO_CONTROL->OTG_STEREO_EN from stereo enable control */ + uint32_t otg_stereo_sync_output_line_num; /* OTG_STEREO_CONTROL->OTG_STEREO_SYNC_OUTPUT_LINE_NUM from timing->stereo_3d_format line num */ + uint32_t otg_stereo_sync_output_polarity; /* OTG_STEREO_CONTROL->OTG_STEREO_SYNC_OUTPUT_POLARITY from stereo polarity control */ + uint32_t otg_3d_structure_en; /* OTG_3D_STRUCTURE_CONTROL->OTG_3D_STRUCTURE_EN from 3D structure enable */ + uint32_t otg_3d_structure_v_update_mode; /* OTG_3D_STRUCTURE_CONTROL->OTG_3D_STRUCTURE_V_UPDATE_MODE from 3D vertical update mode */ + uint32_t otg_3d_structure_stereo_sel_ovr; /* OTG_3D_STRUCTURE_CONTROL->OTG_3D_STRUCTURE_STEREO_SEL_OVR from 3D stereo selection override */ + uint32_t otg_interlace_enable; /* OTG_INTERLACE_CONTROL->OTG_INTERLACE_ENABLE from dc_crtc_timing->flags.INTERLACE */ + + /* OTG GSL (Global Sync Lock) Control - 5 fields */ + uint32_t otg_gsl0_en; /* OTG_GSL_CONTROL->OTG_GSL0_EN from GSL group 0 enable */ + uint32_t otg_gsl1_en; /* OTG_GSL_CONTROL->OTG_GSL1_EN from GSL group 1 enable */ + uint32_t otg_gsl2_en; /* OTG_GSL_CONTROL->OTG_GSL2_EN from GSL group 2 enable */ + uint32_t otg_gsl_master_en; /* OTG_GSL_CONTROL->OTG_GSL_MASTER_EN from GSL master enable */ + uint32_t otg_gsl_master_mode; /* OTG_GSL_CONTROL->OTG_GSL_MASTER_MODE from gsl_params->gsl_master mode */ + + /* OTG DRR Advanced Control - 4 fields */ + uint32_t otg_v_total_last_used_by_drr; /* OTG_DRR_CONTROL->OTG_V_TOTAL_LAST_USED_BY_DRR from last used DRR V_TOTAL (read-only) */ + uint32_t otg_drr_trigger_window_start_x; /* OTG_DRR_TRIGGER_WINDOW->OTG_DRR_TRIGGER_WINDOW_START_X from window_start parameter */ + uint32_t otg_drr_trigger_window_end_x; /* OTG_DRR_TRIGGER_WINDOW->OTG_DRR_TRIGGER_WINDOW_END_X from window_end parameter */ + uint32_t otg_drr_v_total_change_limit; /* OTG_DRR_V_TOTAL_CHANGE->OTG_DRR_V_TOTAL_CHANGE_LIMIT from limit parameter */ + + /* OTG DSC Position Control - 2 fields */ + uint32_t otg_dsc_start_position_x; /* OTG_DSC_START_POSITION->OTG_DSC_START_POSITION_X from DSC start X position */ + uint32_t otg_dsc_start_position_line_num; /* OTG_DSC_START_POSITION->OTG_DSC_START_POSITION_LINE_NUM from DSC start line number */ + + /* OTG Double Buffer Control - 2 fields */ + uint32_t otg_drr_timing_dbuf_update_mode; /* OTG_DOUBLE_BUFFER_CONTROL->OTG_DRR_TIMING_DBUF_UPDATE_MODE from DRR double buffer mode */ + uint32_t otg_blank_data_double_buffer_en; /* OTG_DOUBLE_BUFFER_CONTROL->OTG_BLANK_DATA_DOUBLE_BUFFER_EN from blank data double buffer enable */ + + /* OTG Vertical Interrupts - 6 fields */ + uint32_t otg_vertical_interrupt0_int_enable; /* OTG_VERTICAL_INTERRUPT0_CONTROL->OTG_VERTICAL_INTERRUPT0_INT_ENABLE from interrupt 0 enable */ + uint32_t otg_vertical_interrupt0_line_start; /* OTG_VERTICAL_INTERRUPT0_POSITION->OTG_VERTICAL_INTERRUPT0_LINE_START from start_line parameter */ + uint32_t otg_vertical_interrupt1_int_enable; /* OTG_VERTICAL_INTERRUPT1_CONTROL->OTG_VERTICAL_INTERRUPT1_INT_ENABLE from interrupt 1 enable */ + uint32_t otg_vertical_interrupt1_line_start; /* OTG_VERTICAL_INTERRUPT1_POSITION->OTG_VERTICAL_INTERRUPT1_LINE_START from start_line parameter */ + uint32_t otg_vertical_interrupt2_int_enable; /* OTG_VERTICAL_INTERRUPT2_CONTROL->OTG_VERTICAL_INTERRUPT2_INT_ENABLE from interrupt 2 enable */ + uint32_t otg_vertical_interrupt2_line_start; /* OTG_VERTICAL_INTERRUPT2_POSITION->OTG_VERTICAL_INTERRUPT2_LINE_START from start_line parameter */ + + /* OTG Global Sync Parameters - 6 fields */ + uint32_t otg_vready_offset; /* OTG_VREADY_PARAM->OTG_VREADY_OFFSET from vready_offset parameter */ + uint32_t otg_vstartup_start; /* OTG_VSTARTUP_PARAM->OTG_VSTARTUP_START from vstartup_start parameter */ + uint32_t otg_vupdate_offset; /* OTG_VUPDATE_PARAM->OTG_VUPDATE_OFFSET from vupdate_offset parameter */ + uint32_t otg_vupdate_width; /* OTG_VUPDATE_PARAM->OTG_VUPDATE_WIDTH from vupdate_width parameter */ + uint32_t master_update_lock_vupdate_keepout_start_offset; /* OTG_VUPDATE_KEEPOUT->MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_START_OFFSET from pstate_keepout start */ + uint32_t master_update_lock_vupdate_keepout_end_offset; /* OTG_VUPDATE_KEEPOUT->MASTER_UPDATE_LOCK_VUPDATE_KEEPOUT_END_OFFSET from pstate_keepout end */ + + /* OTG Manual Trigger Control - 11 fields */ + uint32_t otg_triga_source_select; /* OTG_TRIGA_CNTL->OTG_TRIGA_SOURCE_SELECT from trigger A source selection */ + uint32_t otg_triga_source_pipe_select; /* OTG_TRIGA_CNTL->OTG_TRIGA_SOURCE_PIPE_SELECT from trigger A pipe selection */ + uint32_t otg_triga_rising_edge_detect_cntl; /* OTG_TRIGA_CNTL->OTG_TRIGA_RISING_EDGE_DETECT_CNTL from trigger A rising edge detect */ + uint32_t otg_triga_falling_edge_detect_cntl; /* OTG_TRIGA_CNTL->OTG_TRIGA_FALLING_EDGE_DETECT_CNTL from trigger A falling edge detect */ + uint32_t otg_triga_polarity_select; /* OTG_TRIGA_CNTL->OTG_TRIGA_POLARITY_SELECT from trigger A polarity selection */ + uint32_t otg_triga_frequency_select; /* OTG_TRIGA_CNTL->OTG_TRIGA_FREQUENCY_SELECT from trigger A frequency selection */ + uint32_t otg_triga_delay; /* OTG_TRIGA_CNTL->OTG_TRIGA_DELAY from trigger A delay */ + uint32_t otg_triga_clear; /* OTG_TRIGA_CNTL->OTG_TRIGA_CLEAR from trigger A clear */ + uint32_t otg_triga_manual_trig; /* OTG_TRIGA_MANUAL_TRIG->OTG_TRIGA_MANUAL_TRIG from manual trigger A */ + uint32_t otg_trigb_source_select; /* OTG_TRIGB_CNTL->OTG_TRIGB_SOURCE_SELECT from trigger B source selection */ + uint32_t otg_trigb_polarity_select; /* OTG_TRIGB_CNTL->OTG_TRIGB_POLARITY_SELECT from trigger B polarity selection */ + uint32_t otg_trigb_manual_trig; /* OTG_TRIGB_MANUAL_TRIG->OTG_TRIGB_MANUAL_TRIG from manual trigger B */ + + /* OTG Static Screen and Update Control - 6 fields */ + uint32_t otg_static_screen_event_mask; /* OTG_STATIC_SCREEN_CONTROL->OTG_STATIC_SCREEN_EVENT_MASK from event_triggers parameter */ + uint32_t otg_static_screen_frame_count; /* OTG_STATIC_SCREEN_CONTROL->OTG_STATIC_SCREEN_FRAME_COUNT from num_frames parameter */ + uint32_t master_update_lock; /* OTG_MASTER_UPDATE_LOCK->MASTER_UPDATE_LOCK from update lock control */ + uint32_t master_update_mode; /* OTG_MASTER_UPDATE_MODE->MASTER_UPDATE_MODE from update mode selection */ + uint32_t otg_force_count_now_mode; /* OTG_FORCE_COUNT_NOW_CNTL->OTG_FORCE_COUNT_NOW_MODE from force count mode */ + uint32_t otg_force_count_now_clear; /* OTG_FORCE_COUNT_NOW_CNTL->OTG_FORCE_COUNT_NOW_CLEAR from force count clear */ + + /* VTG Control - 3 fields */ + uint32_t vtg0_enable; /* CONTROL->VTG0_ENABLE from VTG enable control */ + uint32_t vtg0_fp2; /* CONTROL->VTG0_FP2 from VTG front porch 2 */ + uint32_t vtg0_vcount_init; /* CONTROL->VTG0_VCOUNT_INIT from VTG vertical count init */ + + /* OTG Status (Read-Only) - 12 fields */ + uint32_t otg_v_blank; /* OTG_STATUS->OTG_V_BLANK from vertical blank status (read-only) */ + uint32_t otg_v_active_disp; /* OTG_STATUS->OTG_V_ACTIVE_DISP from vertical active display (read-only) */ + uint32_t otg_frame_count; /* OTG_STATUS_FRAME_COUNT->OTG_FRAME_COUNT from frame count (read-only) */ + uint32_t otg_horz_count; /* OTG_STATUS_POSITION->OTG_HORZ_COUNT from horizontal position (read-only) */ + uint32_t otg_vert_count; /* OTG_STATUS_POSITION->OTG_VERT_COUNT from vertical position (read-only) */ + uint32_t otg_horz_count_hv; /* OTG_STATUS_HV_COUNT->OTG_HORZ_COUNT from horizontal count (read-only) */ + uint32_t otg_vert_count_nom; /* OTG_STATUS_HV_COUNT->OTG_VERT_COUNT_NOM from vertical count nominal (read-only) */ + uint32_t otg_flip_pending; /* OTG_PIPE_UPDATE_STATUS->OTG_FLIP_PENDING from flip pending status (read-only) */ + uint32_t otg_dc_reg_update_pending; /* OTG_PIPE_UPDATE_STATUS->OTG_DC_REG_UPDATE_PENDING from DC register update pending (read-only) */ + uint32_t otg_cursor_update_pending; /* OTG_PIPE_UPDATE_STATUS->OTG_CURSOR_UPDATE_PENDING from cursor update pending (read-only) */ + uint32_t otg_vupdate_keepout_status; /* OTG_PIPE_UPDATE_STATUS->OTG_VUPDATE_KEEPOUT_STATUS from VUPDATE keepout status (read-only) */ + } optc[MAX_PIPES]; + + /* Metadata */ + uint32_t active_pipe_count; + uint32_t active_stream_count; + bool state_valid; +}; + +/** + * dc_capture_register_software_state() - Capture software state for register programming + * @dc: DC context containing current display configuration + * @state: Pointer to dc_register_software_state structure to populate + * + * Extracts all software state variables that are used to program hardware register + * fields across the display driver pipeline. This provides a complete snapshot + * of the software configuration that drives hardware register programming. + * + * The function traverses the DC context and extracts values from: + * - Stream configurations (timing, format, DSC settings) + * - Plane states (surface format, rotation, scaling, cursor) + * - Pipe contexts (resource allocation, blending, viewport) + * - Clock manager (display clocks, DPP clocks, pixel clocks) + * - Resource context (DET buffer allocation, ODM configuration) + * + * This is essential for underflow debugging as it captures the exact software + * state that determines how registers are programmed, allowing analysis of + * whether underflow is caused by incorrect register programming or timing issues. + * + * Return: true if state was successfully captured, false on error + */ +bool dc_capture_register_software_state(struct dc *dc, struct dc_register_software_state *state); + #endif /* DC_INTERFACE_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h index 5fa5e2b63fb7..40d7a7d83c40 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_bios_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_bios_types.h @@ -91,9 +91,17 @@ struct dc_vbios_funcs { struct device_id id); /* COMMANDS */ + enum bp_result (*select_crtc_source)( + struct dc_bios *bios, + struct bp_crtc_source_select *bp_params); enum bp_result (*encoder_control)( struct dc_bios *bios, struct bp_encoder_control *cntl); + enum bp_result (*dac_load_detection)( + struct dc_bios *bios, + enum engine_id engine_id, + enum dal_device_type device_type, + uint32_t enum_id); enum bp_result (*transmitter_control)( struct dc_bios *bios, struct bp_transmitter_control *cntl); @@ -165,6 +173,7 @@ struct dc_vbios_funcs { }; struct bios_registers { + uint32_t BIOS_SCRATCH_0; uint32_t BIOS_SCRATCH_3; uint32_t BIOS_SCRATCH_6; }; diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 53a088ebddef..7b09af1cb306 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -442,7 +442,6 @@ bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, stru int i = 0, k = 0; int ramp_up_num_steps = 1; // TODO: Ramp is currently disabled. Reenable it. uint8_t visual_confirm_enabled; - int pipe_idx = 0; struct dc_stream_status *stream_status = NULL; if (dc == NULL) @@ -457,7 +456,7 @@ bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, stru cmd.fw_assisted_mclk_switch.config_data.visual_confirm_enabled = visual_confirm_enabled; if (should_manage_pstate) { - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { + for (i = 0; i < dc->res_pool->pipe_count; i++) { struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; if (!pipe->stream) @@ -472,7 +471,6 @@ bool dc_dmub_srv_p_state_delegate(struct dc *dc, bool should_manage_pstate, stru cmd.fw_assisted_mclk_switch.config_data.vactive_stretch_margin_us = dc->debug.fpo_vactive_margin_us; break; } - pipe_idx++; } } @@ -872,7 +870,7 @@ void dc_dmub_setup_subvp_dmub_command(struct dc *dc, bool enable) { uint8_t cmd_pipe_index = 0; - uint32_t i, pipe_idx; + uint32_t i; uint8_t subvp_count = 0; union dmub_rb_cmd cmd; struct pipe_ctx *subvp_pipes[2]; @@ -899,7 +897,7 @@ void dc_dmub_setup_subvp_dmub_command(struct dc *dc, if (enable) { // For each pipe that is a "main" SUBVP pipe, fill in pipe data for DMUB SUBVP cmd - for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) { + for (i = 0; i < dc->res_pool->pipe_count; i++) { struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; pipe_mall_type = dc_state_get_pipe_subvp_type(context, pipe); @@ -922,7 +920,6 @@ void dc_dmub_setup_subvp_dmub_command(struct dc *dc, populate_subvp_cmd_vblank_pipe_info(dc, context, &cmd, pipe, cmd_pipe_index++); } - pipe_idx++; } if (subvp_count == 2) { update_subvp_prefetch_end_to_mall_start(dc, context, &cmd, subvp_pipes); @@ -1174,6 +1171,100 @@ void dc_dmub_srv_subvp_save_surf_addr(const struct dc_dmub_srv *dc_dmub_srv, con dmub_srv_subvp_save_surf_addr(dc_dmub_srv->dmub, addr, subvp_index); } +void dc_dmub_srv_cursor_offload_init(struct dc *dc) +{ + struct dmub_rb_cmd_cursor_offload_init *init; + struct dc_dmub_srv *dc_dmub_srv = dc->ctx->dmub_srv; + union dmub_rb_cmd cmd; + + if (!dc->config.enable_cursor_offload) + return; + + if (!dc_dmub_srv->dmub->meta_info.feature_bits.bits.cursor_offload_v1_support) + return; + + if (!dc_dmub_srv->dmub->cursor_offload_fb.gpu_addr || !dc_dmub_srv->dmub->cursor_offload_fb.cpu_addr) + return; + + if (!dc_dmub_srv->dmub->cursor_offload_v1) + return; + + if (!dc_dmub_srv->dmub->shared_state) + return; + + memset(&cmd, 0, sizeof(cmd)); + + init = &cmd.cursor_offload_init; + init->header.type = DMUB_CMD__CURSOR_OFFLOAD; + init->header.sub_type = DMUB_CMD__CURSOR_OFFLOAD_INIT; + init->header.payload_bytes = sizeof(init->init_data); + init->init_data.state_addr.quad_part = dc_dmub_srv->dmub->cursor_offload_fb.gpu_addr; + init->init_data.state_size = dc_dmub_srv->dmub->cursor_offload_fb.size; + + dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); + + dc_dmub_srv->cursor_offload_enabled = true; +} + +void dc_dmub_srv_control_cursor_offload(struct dc *dc, struct dc_state *context, + const struct dc_stream_state *stream, bool enable) +{ + struct pipe_ctx const *pipe_ctx; + struct dmub_rb_cmd_cursor_offload_stream_cntl *cntl; + union dmub_rb_cmd cmd; + + if (!dc_dmub_srv_is_cursor_offload_enabled(dc)) + return; + + if (!stream) + return; + + pipe_ctx = resource_get_otg_master_for_stream(&context->res_ctx, stream); + if (!pipe_ctx || !pipe_ctx->stream_res.tg || pipe_ctx->stream != stream) + return; + + memset(&cmd, 0, sizeof(cmd)); + + cntl = &cmd.cursor_offload_stream_ctnl; + cntl->header.type = DMUB_CMD__CURSOR_OFFLOAD; + cntl->header.sub_type = + enable ? DMUB_CMD__CURSOR_OFFLOAD_STREAM_ENABLE : DMUB_CMD__CURSOR_OFFLOAD_STREAM_DISABLE; + cntl->header.payload_bytes = sizeof(cntl->data); + + cntl->data.otg_inst = pipe_ctx->stream_res.tg->inst; + cntl->data.line_time_in_ns = 1u + (uint32_t)(div64_u64(stream->timing.h_total * 1000000ull, + stream->timing.pix_clk_100hz / 10)); + + cntl->data.v_total_max = stream->adjust.v_total_max > stream->timing.v_total ? + stream->adjust.v_total_max : + stream->timing.v_total; + + dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, + enable ? DM_DMUB_WAIT_TYPE_NO_WAIT : DM_DMUB_WAIT_TYPE_WAIT); +} + +void dc_dmub_srv_program_cursor_now(struct dc *dc, const struct pipe_ctx *pipe) +{ + struct dmub_rb_cmd_cursor_offload_stream_cntl *cntl; + union dmub_rb_cmd cmd; + + if (!dc_dmub_srv_is_cursor_offload_enabled(dc)) + return; + + if (!pipe || !pipe->stream || !pipe->stream_res.tg) + return; + + memset(&cmd, 0, sizeof(cmd)); + + cntl = &cmd.cursor_offload_stream_ctnl; + cntl->header.type = DMUB_CMD__CURSOR_OFFLOAD; + cntl->header.sub_type = DMUB_CMD__CURSOR_OFFLOAD_STREAM_PROGRAM; + cntl->header.payload_bytes = sizeof(cntl->data); + cntl->data.otg_inst = pipe->stream_res.tg->inst; + + dc_wake_and_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_NO_WAIT); +} + bool dc_dmub_srv_is_hw_pwr_up(struct dc_dmub_srv *dc_dmub_srv, bool wait) { struct dc_context *dc_ctx; @@ -1993,6 +2084,9 @@ bool dmub_lsdma_init(struct dc_dmub_srv *dc_dmub_srv) struct dmub_cmd_lsdma_data *lsdma_data = &cmd.lsdma.lsdma_data; bool result; + if (!dc_dmub_srv->dmub->feature_caps.lsdma_support_in_dmu) + return false; + memset(&cmd, 0, sizeof(cmd)); cmd.cmd_common.header.type = DMUB_CMD__LSDMA; @@ -2231,6 +2325,11 @@ bool dmub_lsdma_send_poll_reg_write_command(struct dc_dmub_srv *dc_dmub_srv, uin return result; } +bool dc_dmub_srv_is_cursor_offload_enabled(const struct dc *dc) +{ + return dc->ctx->dmub_srv && dc->ctx->dmub_srv->cursor_offload_enabled; +} + void dc_dmub_srv_release_hw(const struct dc *dc) { struct dc_dmub_srv *dc_dmub_srv = dc->ctx->dmub_srv; @@ -2248,3 +2347,24 @@ void dc_dmub_srv_release_hw(const struct dc *dc) dm_execute_dmub_cmd(dc->ctx, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } + +void dc_dmub_srv_log_preos_dmcub_info(struct dc_dmub_srv *dc_dmub_srv) +{ + struct dmub_srv *dmub; + + if (!dc_dmub_srv || !dc_dmub_srv->dmub) + return; + + dmub = dc_dmub_srv->dmub; + + if (dmub_srv_get_preos_info(dmub)) { + DC_LOG_DEBUG("%s: PreOS DMCUB Info", __func__); + DC_LOG_DEBUG("fw_version : 0x%08x", dmub->preos_info.fw_version); + DC_LOG_DEBUG("boot_options : 0x%08x", dmub->preos_info.boot_options); + DC_LOG_DEBUG("boot_status : 0x%08x", dmub->preos_info.boot_status); + DC_LOG_DEBUG("trace_buffer_phy_addr : 0x%016llx", dmub->preos_info.trace_buffer_phy_addr); + DC_LOG_DEBUG("trace_buffer_size_bytes : 0x%08x", dmub->preos_info.trace_buffer_size); + DC_LOG_DEBUG("fb_base : 0x%016llx", dmub->preos_info.fb_base); + DC_LOG_DEBUG("fb_offset : 0x%016llx", dmub->preos_info.fb_offset); + } +} diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h index 7ef93444ef3c..72e0a41f39f0 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.h @@ -56,6 +56,7 @@ struct dc_dmub_srv { union dmub_shared_state_ips_driver_signals driver_signals; bool idle_allowed; bool needs_idle_wake; + bool cursor_offload_enabled; }; bool dc_dmub_srv_wait_for_pending(struct dc_dmub_srv *dc_dmub_srv); @@ -326,9 +327,51 @@ bool dc_dmub_srv_ips_query_residency_info(const struct dc_context *ctx, uint8_t enum ips_residency_mode ips_mode); /** + * dc_dmub_srv_cursor_offload_init() - Enables or disables cursor offloading for a stream. + * + * @dc: pointer to DC object + */ +void dc_dmub_srv_cursor_offload_init(struct dc *dc); + +/** + * dc_dmub_srv_control_cursor_offload() - Enables or disables cursor offloading for a stream. + * + * @dc: pointer to DC object + * @context: the DC context to reference for pipe allocations + * @stream: the stream to control + * @enable: true to enable cursor offload, false to disable + */ +void dc_dmub_srv_control_cursor_offload(struct dc *dc, struct dc_state *context, + const struct dc_stream_state *stream, bool enable); + +/** + * dc_dmub_srv_program_cursor_now() - Requests immediate cursor programming for a given pipe. + * + * @dc: pointer to DC object + * @pipe: top-most pipe for a stream. + */ +void dc_dmub_srv_program_cursor_now(struct dc *dc, const struct pipe_ctx *pipe); + +/** + * dc_dmub_srv_is_cursor_offload_enabled() - Checks if cursor offload is supported. + * + * @dc: pointer to DC object + * + * Return: true if cursor offload is supported, false otherwise + */ +bool dc_dmub_srv_is_cursor_offload_enabled(const struct dc *dc); + +/** * dc_dmub_srv_release_hw() - Notifies DMUB service that HW access is no longer required. * * @dc - pointer to DC object */ void dc_dmub_srv_release_hw(const struct dc *dc); + +/** + * dc_dmub_srv_log_preos_dmcub_info() - Logs preos dmcub fw info. + * + * @dc - pointer to DC object + */ +void dc_dmub_srv_log_preos_dmcub_info(struct dc_dmub_srv *dc_dmub_srv); #endif /* _DMUB_DC_SRV_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index db669ccb1d58..79e1696def63 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -1157,6 +1157,16 @@ struct dprx_states { bool cable_id_written; }; +union dpcd_panel_replay_capability_supported { + struct { + unsigned char PANEL_REPLAY_SUPPORT :1; + unsigned char SELECTIVE_UPDATE_SUPPORT :1; + unsigned char EARLY_TRANSPORT_SUPPORT :1; + unsigned char RESERVED :5; + } bits; + unsigned char raw; +}; + enum dpcd_downstream_port_max_bpc { DOWN_STREAM_MAX_8BPC = 0, DOWN_STREAM_MAX_10BPC, @@ -1280,6 +1290,7 @@ struct dpcd_caps { struct edp_psr_info psr_info; struct replay_info pr_info; + union dpcd_panel_replay_capability_supported pr_caps_supported; uint16_t edp_oled_emission_rate; union dp_receive_port0_cap receive_port0_cap; /* Indicates the number of SST links supported by MSO (Multi-Stream Output) */ @@ -1346,6 +1357,31 @@ union dpcd_replay_configuration { unsigned char raw; }; +union panel_replay_enable_and_configuration_1 { + struct { + unsigned char PANEL_REPLAY_ENABLE :1; + unsigned char PANEL_REPLAY_CRC_ENABLE :1; + unsigned char IRQ_HPD_ASSDP_MISSING :1; + unsigned char IRQ_HPD_VSCSDP_UNCORRECTABLE_ERROR :1; + unsigned char IRQ_HPD_RFB_ERROR :1; + unsigned char IRQ_HPD_ACTIVE_FRAME_CRC_ERROR :1; + unsigned char PANEL_REPLAY_SELECTIVE_UPDATE_ENABLE :1; + unsigned char PANEL_REPLAY_EARLY_TRANSPORT_ENABLE :1; + } bits; + unsigned char raw; +}; + +union panel_replay_enable_and_configuration_2 { + struct { + unsigned char SINK_REFRESH_RATE_UNLOCK_GRANTED :1; + unsigned char RESERVED :1; + unsigned char SU_Y_GRANULARITY_EXT_VALUE_ENABLED :1; + unsigned char SU_Y_GRANULARITY_EXT_VALUE :4; + unsigned char SU_REGION_SCAN_LINE_CAPTURE_INDICATION :1; + } bits; + unsigned char raw; +}; + union dpcd_alpm_configuration { struct { unsigned char ENABLE : 1; diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c index 55704d4457ef..37d1a79e8241 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c +++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c @@ -147,6 +147,8 @@ void translate_SPL_in_params_from_pipe_ctx(struct pipe_ctx *pipe_ctx, struct spl spl_in->prefer_easf = false; else if (pipe_ctx->stream->ctx->dc->debug.force_easf == 2) spl_in->disable_easf = true; + else if (pipe_ctx->stream->ctx->dc->debug.force_easf == 3) + spl_in->override_easf = true; /* Translate adaptive sharpening preference */ unsigned int sharpness_setting = pipe_ctx->stream->ctx->dc->debug.force_sharpness; unsigned int force_sharpness_level = pipe_ctx->stream->ctx->dc->debug.force_sharpness_level; diff --git a/drivers/gpu/drm/amd/display/dc/dc_stream.h b/drivers/gpu/drm/amd/display/dc/dc_stream.h index 76cf9fdedab0..321cfe92d799 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_stream.h +++ b/drivers/gpu/drm/amd/display/dc/dc_stream.h @@ -473,12 +473,11 @@ void dc_enable_stereo( /* Triggers multi-stream synchronization. */ void dc_trigger_sync(struct dc *dc, struct dc_state *context); -enum surface_update_type dc_check_update_surfaces_for_stream( - struct dc *dc, +struct surface_update_descriptor dc_check_update_surfaces_for_stream( + const struct dc_check_config *check_config, struct dc_surface_update *updates, int surface_count, - struct dc_stream_update *stream_update, - const struct dc_stream_status *stream_status); + struct dc_stream_update *stream_update); /** * Create a new default stream for the requested sink @@ -492,8 +491,8 @@ void update_stream_signal(struct dc_stream_state *stream, struct dc_sink *sink); void dc_stream_retain(struct dc_stream_state *dc_stream); void dc_stream_release(struct dc_stream_state *dc_stream); -struct dc_stream_status *dc_stream_get_status( - struct dc_stream_state *dc_stream); +struct dc_stream_status *dc_stream_get_status(struct dc_stream_state *dc_stream); +const struct dc_stream_status *dc_stream_get_status_const(const struct dc_stream_state *dc_stream); /******************************************************************************* * Cursor interfaces - To manages the cursor within a stream diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index b5aa03a3e39c..f46039f64203 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -185,6 +185,10 @@ struct dc_panel_patch { unsigned int wait_after_dpcd_poweroff_ms; }; +/** + * struct dc_edid_caps - Capabilities read from EDID. + * @analog: Whether the monitor is analog. Used by DVI-I handling. + */ struct dc_edid_caps { /* sink identification */ uint16_t manufacturer_id; @@ -212,6 +216,8 @@ struct dc_edid_caps { bool edid_hdmi; bool hdr_supported; bool rr_capable; + bool scdc_present; + bool analog; struct dc_panel_patch panel_patch; }; @@ -347,7 +353,8 @@ enum dc_connection_type { dc_connection_none, dc_connection_single, dc_connection_mst_branch, - dc_connection_sst_branch + dc_connection_sst_branch, + dc_connection_dac_load }; struct dc_csc_adjustments { @@ -934,6 +941,12 @@ enum dc_psr_version { DC_PSR_VERSION_UNSUPPORTED = 0xFFFFFFFF, }; +enum dc_replay_version { + DC_FREESYNC_REPLAY = 0, + DC_VESA_PANEL_REPLAY = 1, + DC_REPLAY_VERSION_UNSUPPORTED = 0XFF, +}; + /* Possible values of display_endpoint_id.endpoint */ enum display_endpoint_type { DISPLAY_ENDPOINT_PHY = 0, /* Physical connector. */ @@ -1086,6 +1099,7 @@ enum replay_FW_Message_type { Replay_Set_Residency_Frameupdate_Timer, Replay_Set_Pseudo_VTotal, Replay_Disabled_Adaptive_Sync_SDP, + Replay_Set_Version, Replay_Set_General_Cmd, }; @@ -1121,6 +1135,8 @@ union replay_low_refresh_rate_enable_options { }; struct replay_config { + /* Replay version */ + enum dc_replay_version replay_version; /* Replay feature is supported */ bool replay_supported; /* Replay caps support DPCD & EDID caps*/ @@ -1177,6 +1193,10 @@ struct replay_settings { uint32_t coasting_vtotal_table[PR_COASTING_TYPE_NUM]; /* Defer Update Coasting vtotal table */ uint32_t defer_update_coasting_vtotal_table[PR_COASTING_TYPE_NUM]; + /* Skip frame number table */ + uint32_t frame_skip_number_table[PR_COASTING_TYPE_NUM]; + /* Defer skip frame number table */ + uint32_t defer_frame_skip_number_table[PR_COASTING_TYPE_NUM]; /* Maximum link off frame count */ uint32_t link_off_frame_count; /* Replay pseudo vtotal for low refresh rate*/ @@ -1185,6 +1205,8 @@ struct replay_settings { uint16_t last_pseudo_vtotal; /* Replay desync error */ uint32_t replay_desync_error_fail_count; + /* The frame skip number dal send to DMUB */ + uint16_t frame_skip_number; }; /* To split out "global" and "per-panel" config settings. diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c index 5999b2da3a01..33d8bd91cb01 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.c @@ -148,7 +148,7 @@ struct dccg *dccg2_create( const struct dccg_shift *dccg_shift, const struct dccg_mask *dccg_mask) { - struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_ATOMIC); + struct dcn_dccg *dccg_dcn = kzalloc(sizeof(*dccg_dcn), GFP_KERNEL); struct dccg *base; if (dccg_dcn == NULL) { diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h index a9b88f5e0c04..8bdffd9ff31b 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn20/dcn20_dccg.h @@ -425,7 +425,69 @@ struct dccg_mask { uint32_t SYMCLKD_CLOCK_ENABLE; \ uint32_t SYMCLKE_CLOCK_ENABLE; \ uint32_t DP_DTO_MODULO[MAX_PIPES]; \ - uint32_t DP_DTO_PHASE[MAX_PIPES] + uint32_t DP_DTO_PHASE[MAX_PIPES]; \ + uint32_t DC_MEM_GLOBAL_PWR_REQ_CNTL; \ + uint32_t DCCG_AUDIO_DTO0_MODULE; \ + uint32_t DCCG_AUDIO_DTO0_PHASE; \ + uint32_t DCCG_AUDIO_DTO1_MODULE; \ + uint32_t DCCG_AUDIO_DTO1_PHASE; \ + uint32_t DCCG_CAC_STATUS; \ + uint32_t DCCG_CAC_STATUS2; \ + uint32_t DCCG_DISP_CNTL_REG; \ + uint32_t DCCG_DS_CNTL; \ + uint32_t DCCG_DS_DTO_INCR; \ + uint32_t DCCG_DS_DTO_MODULO; \ + uint32_t DCCG_DS_HW_CAL_INTERVAL; \ + uint32_t DCCG_GTC_CNTL; \ + uint32_t DCCG_GTC_CURRENT; \ + uint32_t DCCG_GTC_DTO_INCR; \ + uint32_t DCCG_GTC_DTO_MODULO; \ + uint32_t DCCG_PERFMON_CNTL; \ + uint32_t DCCG_PERFMON_CNTL2; \ + uint32_t DCCG_SOFT_RESET; \ + uint32_t DCCG_TEST_CLK_SEL; \ + uint32_t DCCG_VSYNC_CNT_CTRL; \ + uint32_t DCCG_VSYNC_CNT_INT_CTRL; \ + uint32_t DCCG_VSYNC_OTG0_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG1_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG2_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG3_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG4_LATCH_VALUE; \ + uint32_t DCCG_VSYNC_OTG5_LATCH_VALUE; \ + uint32_t DISPCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DP_DTO_DBUF_EN; \ + uint32_t DPIACLK_540M_DTO_MODULO; \ + uint32_t DPIACLK_540M_DTO_PHASE; \ + uint32_t DPIACLK_810M_DTO_MODULO; \ + uint32_t DPIACLK_810M_DTO_PHASE; \ + uint32_t DPIACLK_DTO_CNTL; \ + uint32_t DPIASYMCLK_CNTL; \ + uint32_t DPPCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DPREFCLK_CGTT_BLK_CTRL_REG; \ + uint32_t DPREFCLK_CNTL; \ + uint32_t DTBCLK_DTO_DBUF_EN; \ + uint32_t FORCE_SYMCLK_DISABLE; \ + uint32_t HDMICHARCLK0_CLOCK_CNTL; \ + uint32_t MICROSECOND_TIME_BASE_DIV; \ + uint32_t MILLISECOND_TIME_BASE_DIV; \ + uint32_t OTG0_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG0_PIXEL_RATE_CNTL; \ + uint32_t OTG1_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG1_PIXEL_RATE_CNTL; \ + uint32_t OTG2_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG2_PIXEL_RATE_CNTL; \ + uint32_t OTG3_PHYPLL_PIXEL_RATE_CNTL; \ + uint32_t OTG3_PIXEL_RATE_CNTL; \ + uint32_t PHYPLLA_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLB_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLC_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLD_PIXCLK_RESYNC_CNTL; \ + uint32_t PHYPLLE_PIXCLK_RESYNC_CNTL; \ + uint32_t REFCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SOCCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SYMCLK_CGTT_BLK_CTRL_REG; \ + uint32_t SYMCLK_PSP_CNTL + struct dccg_registers { DCCG_REG_VARIABLE_LIST; }; diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c index 8664f0c4c9b7..97df04b7e39d 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.c @@ -709,6 +709,128 @@ void dccg31_otg_drop_pixel(struct dccg *dccg, OTG_DROP_PIXEL[otg_inst], 1); } +void dccg31_read_reg_state(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state) +{ + struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg); + + dccg_reg_state->dc_mem_global_pwr_req_cntl = REG_READ(DC_MEM_GLOBAL_PWR_REQ_CNTL); + dccg_reg_state->dccg_audio_dtbclk_dto_modulo = REG_READ(DCCG_AUDIO_DTBCLK_DTO_MODULO); + dccg_reg_state->dccg_audio_dtbclk_dto_phase = REG_READ(DCCG_AUDIO_DTBCLK_DTO_PHASE); + dccg_reg_state->dccg_audio_dto_source = REG_READ(DCCG_AUDIO_DTO_SOURCE); + dccg_reg_state->dccg_audio_dto0_module = REG_READ(DCCG_AUDIO_DTO0_MODULE); + dccg_reg_state->dccg_audio_dto0_phase = REG_READ(DCCG_AUDIO_DTO0_PHASE); + dccg_reg_state->dccg_audio_dto1_module = REG_READ(DCCG_AUDIO_DTO1_MODULE); + dccg_reg_state->dccg_audio_dto1_phase = REG_READ(DCCG_AUDIO_DTO1_PHASE); + dccg_reg_state->dccg_cac_status = REG_READ(DCCG_CAC_STATUS); + dccg_reg_state->dccg_cac_status2 = REG_READ(DCCG_CAC_STATUS2); + dccg_reg_state->dccg_disp_cntl_reg = REG_READ(DCCG_DISP_CNTL_REG); + dccg_reg_state->dccg_ds_cntl = REG_READ(DCCG_DS_CNTL); + dccg_reg_state->dccg_ds_dto_incr = REG_READ(DCCG_DS_DTO_INCR); + dccg_reg_state->dccg_ds_dto_modulo = REG_READ(DCCG_DS_DTO_MODULO); + dccg_reg_state->dccg_ds_hw_cal_interval = REG_READ(DCCG_DS_HW_CAL_INTERVAL); + dccg_reg_state->dccg_gate_disable_cntl = REG_READ(DCCG_GATE_DISABLE_CNTL); + dccg_reg_state->dccg_gate_disable_cntl2 = REG_READ(DCCG_GATE_DISABLE_CNTL2); + dccg_reg_state->dccg_gate_disable_cntl3 = REG_READ(DCCG_GATE_DISABLE_CNTL3); + dccg_reg_state->dccg_gate_disable_cntl4 = REG_READ(DCCG_GATE_DISABLE_CNTL4); + dccg_reg_state->dccg_gate_disable_cntl5 = REG_READ(DCCG_GATE_DISABLE_CNTL5); + dccg_reg_state->dccg_gate_disable_cntl6 = REG_READ(DCCG_GATE_DISABLE_CNTL6); + dccg_reg_state->dccg_global_fgcg_rep_cntl = REG_READ(DCCG_GLOBAL_FGCG_REP_CNTL); + dccg_reg_state->dccg_gtc_cntl = REG_READ(DCCG_GTC_CNTL); + dccg_reg_state->dccg_gtc_current = REG_READ(DCCG_GTC_CURRENT); + dccg_reg_state->dccg_gtc_dto_incr = REG_READ(DCCG_GTC_DTO_INCR); + dccg_reg_state->dccg_gtc_dto_modulo = REG_READ(DCCG_GTC_DTO_MODULO); + dccg_reg_state->dccg_perfmon_cntl = REG_READ(DCCG_PERFMON_CNTL); + dccg_reg_state->dccg_perfmon_cntl2 = REG_READ(DCCG_PERFMON_CNTL2); + dccg_reg_state->dccg_soft_reset = REG_READ(DCCG_SOFT_RESET); + dccg_reg_state->dccg_test_clk_sel = REG_READ(DCCG_TEST_CLK_SEL); + dccg_reg_state->dccg_vsync_cnt_ctrl = REG_READ(DCCG_VSYNC_CNT_CTRL); + dccg_reg_state->dccg_vsync_cnt_int_ctrl = REG_READ(DCCG_VSYNC_CNT_INT_CTRL); + dccg_reg_state->dccg_vsync_otg0_latch_value = REG_READ(DCCG_VSYNC_OTG0_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg1_latch_value = REG_READ(DCCG_VSYNC_OTG1_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg2_latch_value = REG_READ(DCCG_VSYNC_OTG2_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg3_latch_value = REG_READ(DCCG_VSYNC_OTG3_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg4_latch_value = REG_READ(DCCG_VSYNC_OTG4_LATCH_VALUE); + dccg_reg_state->dccg_vsync_otg5_latch_value = REG_READ(DCCG_VSYNC_OTG5_LATCH_VALUE); + dccg_reg_state->dispclk_cgtt_blk_ctrl_reg = REG_READ(DISPCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dispclk_freq_change_cntl = REG_READ(DISPCLK_FREQ_CHANGE_CNTL); + dccg_reg_state->dp_dto_dbuf_en = REG_READ(DP_DTO_DBUF_EN); + dccg_reg_state->dp_dto0_modulo = REG_READ(DP_DTO_MODULO[0]); + dccg_reg_state->dp_dto0_phase = REG_READ(DP_DTO_PHASE[0]); + dccg_reg_state->dp_dto1_modulo = REG_READ(DP_DTO_MODULO[1]); + dccg_reg_state->dp_dto1_phase = REG_READ(DP_DTO_PHASE[1]); + dccg_reg_state->dp_dto2_modulo = REG_READ(DP_DTO_MODULO[2]); + dccg_reg_state->dp_dto2_phase = REG_READ(DP_DTO_PHASE[2]); + dccg_reg_state->dp_dto3_modulo = REG_READ(DP_DTO_MODULO[3]); + dccg_reg_state->dp_dto3_phase = REG_READ(DP_DTO_PHASE[3]); + dccg_reg_state->dpiaclk_540m_dto_modulo = REG_READ(DPIACLK_540M_DTO_MODULO); + dccg_reg_state->dpiaclk_540m_dto_phase = REG_READ(DPIACLK_540M_DTO_PHASE); + dccg_reg_state->dpiaclk_810m_dto_modulo = REG_READ(DPIACLK_810M_DTO_MODULO); + dccg_reg_state->dpiaclk_810m_dto_phase = REG_READ(DPIACLK_810M_DTO_PHASE); + dccg_reg_state->dpiaclk_dto_cntl = REG_READ(DPIACLK_DTO_CNTL); + dccg_reg_state->dpiasymclk_cntl = REG_READ(DPIASYMCLK_CNTL); + dccg_reg_state->dppclk_cgtt_blk_ctrl_reg = REG_READ(DPPCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dppclk_ctrl = REG_READ(DPPCLK_CTRL); + dccg_reg_state->dppclk_dto_ctrl = REG_READ(DPPCLK_DTO_CTRL); + dccg_reg_state->dppclk0_dto_param = REG_READ(DPPCLK_DTO_PARAM[0]); + dccg_reg_state->dppclk1_dto_param = REG_READ(DPPCLK_DTO_PARAM[1]); + dccg_reg_state->dppclk2_dto_param = REG_READ(DPPCLK_DTO_PARAM[2]); + dccg_reg_state->dppclk3_dto_param = REG_READ(DPPCLK_DTO_PARAM[3]); + dccg_reg_state->dprefclk_cgtt_blk_ctrl_reg = REG_READ(DPREFCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->dprefclk_cntl = REG_READ(DPREFCLK_CNTL); + dccg_reg_state->dpstreamclk_cntl = REG_READ(DPSTREAMCLK_CNTL); + dccg_reg_state->dscclk_dto_ctrl = REG_READ(DSCCLK_DTO_CTRL); + dccg_reg_state->dscclk0_dto_param = REG_READ(DSCCLK0_DTO_PARAM); + dccg_reg_state->dscclk1_dto_param = REG_READ(DSCCLK1_DTO_PARAM); + dccg_reg_state->dscclk2_dto_param = REG_READ(DSCCLK2_DTO_PARAM); + dccg_reg_state->dscclk3_dto_param = REG_READ(DSCCLK3_DTO_PARAM); + dccg_reg_state->dtbclk_dto_dbuf_en = REG_READ(DTBCLK_DTO_DBUF_EN); + dccg_reg_state->dtbclk_dto0_modulo = REG_READ(DTBCLK_DTO_MODULO[0]); + dccg_reg_state->dtbclk_dto0_phase = REG_READ(DTBCLK_DTO_PHASE[0]); + dccg_reg_state->dtbclk_dto1_modulo = REG_READ(DTBCLK_DTO_MODULO[1]); + dccg_reg_state->dtbclk_dto1_phase = REG_READ(DTBCLK_DTO_PHASE[1]); + dccg_reg_state->dtbclk_dto2_modulo = REG_READ(DTBCLK_DTO_MODULO[2]); + dccg_reg_state->dtbclk_dto2_phase = REG_READ(DTBCLK_DTO_PHASE[2]); + dccg_reg_state->dtbclk_dto3_modulo = REG_READ(DTBCLK_DTO_MODULO[3]); + dccg_reg_state->dtbclk_dto3_phase = REG_READ(DTBCLK_DTO_PHASE[3]); + dccg_reg_state->dtbclk_p_cntl = REG_READ(DTBCLK_P_CNTL); + dccg_reg_state->force_symclk_disable = REG_READ(FORCE_SYMCLK_DISABLE); + dccg_reg_state->hdmicharclk0_clock_cntl = REG_READ(HDMICHARCLK0_CLOCK_CNTL); + dccg_reg_state->hdmistreamclk_cntl = REG_READ(HDMISTREAMCLK_CNTL); + dccg_reg_state->hdmistreamclk0_dto_param = REG_READ(HDMISTREAMCLK0_DTO_PARAM); + dccg_reg_state->microsecond_time_base_div = REG_READ(MICROSECOND_TIME_BASE_DIV); + dccg_reg_state->millisecond_time_base_div = REG_READ(MILLISECOND_TIME_BASE_DIV); + dccg_reg_state->otg_pixel_rate_div = REG_READ(OTG_PIXEL_RATE_DIV); + dccg_reg_state->otg0_phypll_pixel_rate_cntl = REG_READ(OTG0_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg0_pixel_rate_cntl = REG_READ(OTG0_PIXEL_RATE_CNTL); + dccg_reg_state->otg1_phypll_pixel_rate_cntl = REG_READ(OTG1_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg1_pixel_rate_cntl = REG_READ(OTG1_PIXEL_RATE_CNTL); + dccg_reg_state->otg2_phypll_pixel_rate_cntl = REG_READ(OTG2_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg2_pixel_rate_cntl = REG_READ(OTG2_PIXEL_RATE_CNTL); + dccg_reg_state->otg3_phypll_pixel_rate_cntl = REG_READ(OTG3_PHYPLL_PIXEL_RATE_CNTL); + dccg_reg_state->otg3_pixel_rate_cntl = REG_READ(OTG3_PIXEL_RATE_CNTL); + dccg_reg_state->phyasymclk_clock_cntl = REG_READ(PHYASYMCLK_CLOCK_CNTL); + dccg_reg_state->phybsymclk_clock_cntl = REG_READ(PHYBSYMCLK_CLOCK_CNTL); + dccg_reg_state->phycsymclk_clock_cntl = REG_READ(PHYCSYMCLK_CLOCK_CNTL); + dccg_reg_state->phydsymclk_clock_cntl = REG_READ(PHYDSYMCLK_CLOCK_CNTL); + dccg_reg_state->phyesymclk_clock_cntl = REG_READ(PHYESYMCLK_CLOCK_CNTL); + dccg_reg_state->phyplla_pixclk_resync_cntl = REG_READ(PHYPLLA_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phypllb_pixclk_resync_cntl = REG_READ(PHYPLLB_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phypllc_pixclk_resync_cntl = REG_READ(PHYPLLC_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phyplld_pixclk_resync_cntl = REG_READ(PHYPLLD_PIXCLK_RESYNC_CNTL); + dccg_reg_state->phyplle_pixclk_resync_cntl = REG_READ(PHYPLLE_PIXCLK_RESYNC_CNTL); + dccg_reg_state->refclk_cgtt_blk_ctrl_reg = REG_READ(REFCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->socclk_cgtt_blk_ctrl_reg = REG_READ(SOCCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->symclk_cgtt_blk_ctrl_reg = REG_READ(SYMCLK_CGTT_BLK_CTRL_REG); + dccg_reg_state->symclk_psp_cntl = REG_READ(SYMCLK_PSP_CNTL); + dccg_reg_state->symclk32_le_cntl = REG_READ(SYMCLK32_LE_CNTL); + dccg_reg_state->symclk32_se_cntl = REG_READ(SYMCLK32_SE_CNTL); + dccg_reg_state->symclka_clock_enable = REG_READ(SYMCLKA_CLOCK_ENABLE); + dccg_reg_state->symclkb_clock_enable = REG_READ(SYMCLKB_CLOCK_ENABLE); + dccg_reg_state->symclkc_clock_enable = REG_READ(SYMCLKC_CLOCK_ENABLE); + dccg_reg_state->symclkd_clock_enable = REG_READ(SYMCLKD_CLOCK_ENABLE); + dccg_reg_state->symclke_clock_enable = REG_READ(SYMCLKE_CLOCK_ENABLE); +} + static const struct dccg_funcs dccg31_funcs = { .update_dpp_dto = dccg31_update_dpp_dto, .get_dccg_ref_freq = dccg31_get_dccg_ref_freq, @@ -727,6 +849,7 @@ static const struct dccg_funcs dccg31_funcs = { .set_dispclk_change_mode = dccg31_set_dispclk_change_mode, .disable_dsc = dccg31_disable_dscclk, .enable_dsc = dccg31_enable_dscclk, + .dccg_read_reg_state = dccg31_read_reg_state, }; struct dccg *dccg31_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h index cd261051dc2c..bf659920d4cc 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn31/dcn31_dccg.h @@ -236,4 +236,6 @@ void dccg31_disable_dscclk(struct dccg *dccg, int inst); void dccg31_enable_dscclk(struct dccg *dccg, int inst); +void dccg31_read_reg_state(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state); + #endif //__DCN31_DCCG_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c index 8f6edd8e9beb..ef3db6beba25 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.c @@ -377,7 +377,8 @@ static const struct dccg_funcs dccg314_funcs = { .get_pixel_rate_div = dccg314_get_pixel_rate_div, .trigger_dio_fifo_resync = dccg314_trigger_dio_fifo_resync, .set_valid_pixel_rate = dccg314_set_valid_pixel_rate, - .set_dtbclk_p_src = dccg314_set_dtbclk_p_src + .set_dtbclk_p_src = dccg314_set_dtbclk_p_src, + .dccg_read_reg_state = dccg31_read_reg_state }; struct dccg *dccg314_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h index 60ea1d248deb..a609635f35db 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn314/dcn314_dccg.h @@ -74,8 +74,7 @@ SR(DCCG_GATE_DISABLE_CNTL3),\ SR(HDMISTREAMCLK0_DTO_PARAM),\ SR(OTG_PIXEL_RATE_DIV),\ - SR(DTBCLK_P_CNTL),\ - SR(DCCG_AUDIO_DTO_SOURCE) + SR(DTBCLK_P_CNTL) #define DCCG_MASK_SH_LIST_DCN314_COMMON(mask_sh) \ DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\ diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c index de6d62401362..bd2f528137b2 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.c @@ -1114,6 +1114,16 @@ static void dccg35_trigger_dio_fifo_resync(struct dccg *dccg) if (dispclk_rdivider_value != 0) REG_UPDATE(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_WDIVIDER, dispclk_rdivider_value); } +static void dccg35_wait_for_dentist_change_done( + struct dccg *dccg) +{ + struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg); + + uint32_t dentist_dispclk_value = REG_READ(DENTIST_DISPCLK_CNTL); + + REG_WRITE(DENTIST_DISPCLK_CNTL, dentist_dispclk_value); + REG_WAIT(DENTIST_DISPCLK_CNTL, DENTIST_DISPCLK_CHG_DONE, 1, 50, 2000); +} static void dcn35_set_dppclk_enable(struct dccg *dccg, uint32_t dpp_inst, uint32_t enable) @@ -1174,9 +1184,9 @@ static void dccg35_update_dpp_dto(struct dccg *dccg, int dpp_inst, dcn35_set_dppclk_enable(dccg, dpp_inst, true); } else { dcn35_set_dppclk_enable(dccg, dpp_inst, false); - /*we have this in hwss: disable_plane*/ - //dccg35_set_dppclk_rcg(dccg, dpp_inst, true); + dccg35_set_dppclk_rcg(dccg, dpp_inst, true); } + udelay(10); dccg->pipe_dppclk_khz[dpp_inst] = req_dppclk; } @@ -1300,6 +1310,8 @@ static void dccg35_set_pixel_rate_div( BREAK_TO_DEBUGGER(); return; } + if (otg_inst < 4) + dccg35_wait_for_dentist_change_done(dccg); } static void dccg35_set_dtbclk_p_src( @@ -1411,7 +1423,7 @@ static void dccg35_set_dtbclk_dto( __func__, params->otg_inst, params->pixclk_khz, params->ref_dtbclk_khz, req_dtbclk_khz, phase, modulo); - } else { + } else if (!params->ref_dtbclk_khz && !req_dtbclk_khz) { switch (params->otg_inst) { case 0: REG_UPDATE(DCCG_GATE_DISABLE_CNTL5, DTBCLK_P0_GATE_DISABLE, 0); @@ -1664,7 +1676,7 @@ static void dccg35_dpp_root_clock_control( { struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg); - if (dccg->dpp_clock_gated[dpp_inst] == clock_on) + if (dccg->dpp_clock_gated[dpp_inst] != clock_on) return; if (clock_on) { @@ -1682,9 +1694,12 @@ static void dccg35_dpp_root_clock_control( DPPCLK0_DTO_PHASE, 0, DPPCLK0_DTO_MODULO, 1); /*we have this in hwss: disable_plane*/ - //dccg35_set_dppclk_rcg(dccg, dpp_inst, true); + dccg35_set_dppclk_rcg(dccg, dpp_inst, true); } + // wait for clock to fully ramp + udelay(10); + dccg->dpp_clock_gated[dpp_inst] = !clock_on; DC_LOG_DEBUG("%s: dpp_inst(%d) clock_on = %d\n", __func__, dpp_inst, clock_on); } @@ -2438,6 +2453,7 @@ static const struct dccg_funcs dccg35_funcs = { .disable_symclk_se = dccg35_disable_symclk_se, .set_dtbclk_p_src = dccg35_set_dtbclk_p_src, .dccg_root_gate_disable_control = dccg35_root_gate_disable_control, + .dccg_read_reg_state = dccg31_read_reg_state, }; struct dccg *dccg35_create( diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h index 51f98c5c51c4..7b9c36456cd9 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn35/dcn35_dccg.h @@ -41,8 +41,9 @@ SR(SYMCLKA_CLOCK_ENABLE),\ SR(SYMCLKB_CLOCK_ENABLE),\ SR(SYMCLKC_CLOCK_ENABLE),\ - SR(SYMCLKD_CLOCK_ENABLE),\ - SR(SYMCLKE_CLOCK_ENABLE) + SR(SYMCLKD_CLOCK_ENABLE), \ + SR(SYMCLKE_CLOCK_ENABLE),\ + SR(SYMCLK_PSP_CNTL) #define DCCG_MASK_SH_LIST_DCN35(mask_sh) \ DCCG_SFI(DPPCLK_DTO_CTRL, DTO_DB_EN, DPPCLK, 0, mask_sh),\ @@ -231,6 +232,14 @@ DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK1_GATE_DISABLE, mask_sh),\ DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK2_GATE_DISABLE, mask_sh),\ DCCG_SF(DCCG_GATE_DISABLE_CNTL5, DPSTREAMCLK3_GATE_DISABLE, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_DELAY, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_STEP_SIZE, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_FREQ_RAMP_DONE, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_MAX_ERRDET_CYCLES, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_RESET, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_STATE, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DCCG_FIFO_ERRDET_OVR_EN, mask_sh),\ + DCCG_SF(DISPCLK_FREQ_CHANGE_CNTL, DISPCLK_CHG_FWD_CORR_DISABLE, mask_sh),\ struct dccg *dccg35_create( struct dc_context *ctx, diff --git a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c index 0b8ed9b94d3c..663a18ee5162 100644 --- a/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c +++ b/drivers/gpu/drm/amd/display/dc/dccg/dcn401/dcn401_dccg.c @@ -886,6 +886,7 @@ static const struct dccg_funcs dccg401_funcs = { .enable_symclk_se = dccg401_enable_symclk_se, .disable_symclk_se = dccg401_disable_symclk_se, .set_dtbclk_p_src = dccg401_set_dtbclk_p_src, + .dccg_read_reg_state = dccg31_read_reg_state }; struct dccg *dccg401_create( diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c index a6006776333d..2dcf394edf22 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_abm.c @@ -283,7 +283,7 @@ struct abm *dce_abm_create( const struct dce_abm_shift *abm_shift, const struct dce_abm_mask *abm_mask) { - struct dce_abm *abm_dce = kzalloc(sizeof(*abm_dce), GFP_ATOMIC); + struct dce_abm *abm_dce = kzalloc(sizeof(*abm_dce), GFP_KERNEL); if (abm_dce == NULL) { BREAK_TO_DEBUGGER(); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c index eeed840073fe..fcad61c618a1 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_audio.c @@ -1143,7 +1143,8 @@ void dce_aud_wall_dto_setup( REG_UPDATE(DCCG_AUDIO_DTO1_PHASE, DCCG_AUDIO_DTO1_PHASE, clock_info.audio_dto_phase); - REG_UPDATE(DCCG_AUDIO_DTO_SOURCE, + if (aud->masks->DCCG_AUDIO_DTO2_USE_512FBR_DTO) + REG_UPDATE(DCCG_AUDIO_DTO_SOURCE, DCCG_AUDIO_DTO2_USE_512FBR_DTO, 1); } diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c index a8e79104b684..5f8fba45d98d 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_dmcu.c @@ -1126,7 +1126,7 @@ struct dmcu *dcn10_dmcu_create( const struct dce_dmcu_shift *dmcu_shift, const struct dce_dmcu_mask *dmcu_mask) { - struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC); + struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_KERNEL); if (dmcu_dce == NULL) { BREAK_TO_DEBUGGER(); @@ -1147,7 +1147,7 @@ struct dmcu *dcn20_dmcu_create( const struct dce_dmcu_shift *dmcu_shift, const struct dce_dmcu_mask *dmcu_mask) { - struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC); + struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_KERNEL); if (dmcu_dce == NULL) { BREAK_TO_DEBUGGER(); @@ -1168,7 +1168,7 @@ struct dmcu *dcn21_dmcu_create( const struct dce_dmcu_shift *dmcu_shift, const struct dce_dmcu_mask *dmcu_mask) { - struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_ATOMIC); + struct dce_dmcu *dmcu_dce = kzalloc(sizeof(*dmcu_dce), GFP_KERNEL); if (dmcu_dce == NULL) { BREAK_TO_DEBUGGER(); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c index 0c50fe266c8a..87dbb8d7ed27 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.c @@ -302,6 +302,10 @@ static void setup_panel_mode( if (ctx->dc->caps.psp_setup_panel_mode) return; + /* The code below is only applicable to encoders with a digital transmitter. */ + if (enc110->base.transmitter == TRANSMITTER_UNKNOWN) + return; + ASSERT(REG(DP_DPHY_INTERNAL_CTRL)); value = REG_READ(DP_DPHY_INTERNAL_CTRL); @@ -804,6 +808,33 @@ bool dce110_link_encoder_validate_dp_output( return true; } +static bool dce110_link_encoder_validate_rgb_output( + const struct dce110_link_encoder *enc110, + const struct dc_crtc_timing *crtc_timing) +{ + /* When the VBIOS doesn't specify any limits, use 400 MHz. + * The value comes from amdgpu_atombios_get_clock_info. + */ + uint32_t max_pixel_clock_khz = 400000; + + if (enc110->base.ctx->dc_bios->fw_info_valid && + enc110->base.ctx->dc_bios->fw_info.max_pixel_clock) { + max_pixel_clock_khz = + enc110->base.ctx->dc_bios->fw_info.max_pixel_clock; + } + + if (crtc_timing->pix_clk_100hz > max_pixel_clock_khz * 10) + return false; + + if (crtc_timing->display_color_depth != COLOR_DEPTH_888) + return false; + + if (crtc_timing->pixel_encoding != PIXEL_ENCODING_RGB) + return false; + + return true; +} + void dce110_link_encoder_construct( struct dce110_link_encoder *enc110, const struct encoder_init_data *init_data, @@ -824,6 +855,7 @@ void dce110_link_encoder_construct( enc110->base.connector = init_data->connector; enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; + enc110->base.analog_engine = init_data->analog_engine; enc110->base.features = *enc_features; @@ -847,6 +879,11 @@ void dce110_link_encoder_construct( SIGNAL_TYPE_EDP | SIGNAL_TYPE_HDMI_TYPE_A; + if ((enc110->base.connector.id == CONNECTOR_ID_DUAL_LINK_DVII || + enc110->base.connector.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + enc110->base.analog_engine != ENGINE_ID_UNKNOWN) + enc110->base.output_signals |= SIGNAL_TYPE_RGB; + /* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE. * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY. * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer @@ -885,6 +922,13 @@ void dce110_link_encoder_construct( enc110->base.preferred_engine = ENGINE_ID_DIGG; break; default: + if (init_data->analog_engine != ENGINE_ID_UNKNOWN) { + /* The connector is analog-only, ie. VGA */ + enc110->base.preferred_engine = init_data->analog_engine; + enc110->base.output_signals = SIGNAL_TYPE_RGB; + enc110->base.transmitter = TRANSMITTER_UNKNOWN; + break; + } ASSERT_CRITICAL(false); enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; } @@ -939,6 +983,10 @@ bool dce110_link_encoder_validate_output_with_stream( is_valid = dce110_link_encoder_validate_dp_output( enc110, &stream->timing); break; + case SIGNAL_TYPE_RGB: + is_valid = dce110_link_encoder_validate_rgb_output( + enc110, &stream->timing); + break; case SIGNAL_TYPE_EDP: case SIGNAL_TYPE_LVDS: is_valid = stream->timing.pixel_encoding == PIXEL_ENCODING_RGB; @@ -969,6 +1017,10 @@ void dce110_link_encoder_hw_init( cntl.coherent = false; cntl.hpd_sel = enc110->base.hpd_source; + /* The code below is only applicable to encoders with a digital transmitter. */ + if (enc110->base.transmitter == TRANSMITTER_UNKNOWN) + return; + if (enc110->base.connector.id == CONNECTOR_ID_EDP) cntl.signal = SIGNAL_TYPE_EDP; @@ -1034,6 +1086,8 @@ void dce110_link_encoder_setup( /* DP MST */ REG_UPDATE(DIG_BE_CNTL, DIG_MODE, 5); break; + case SIGNAL_TYPE_RGB: + break; default: ASSERT_CRITICAL(false); /* invalid mode ! */ @@ -1282,6 +1336,24 @@ void dce110_link_encoder_disable_output( struct bp_transmitter_control cntl = { 0 }; enum bp_result result; + switch (enc->analog_engine) { + case ENGINE_ID_DACA: + REG_UPDATE(DAC_ENABLE, DAC_ENABLE, 0); + break; + case ENGINE_ID_DACB: + /* DACB doesn't seem to be present on DCE6+, + * although there are references to it in the register file. + */ + DC_LOG_ERROR("%s DACB is unsupported\n", __func__); + break; + default: + break; + } + + /* The code below only applies to connectors that support digital signals. */ + if (enc->transmitter == TRANSMITTER_UNKNOWN) + return; + if (!dce110_is_dig_enabled(enc)) { /* OF_SKIP_POWER_DOWN_INACTIVE_ENCODER */ return; @@ -1726,6 +1798,7 @@ void dce60_link_encoder_construct( enc110->base.connector = init_data->connector; enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; + enc110->base.analog_engine = init_data->analog_engine; enc110->base.features = *enc_features; @@ -1749,6 +1822,11 @@ void dce60_link_encoder_construct( SIGNAL_TYPE_EDP | SIGNAL_TYPE_HDMI_TYPE_A; + if ((enc110->base.connector.id == CONNECTOR_ID_DUAL_LINK_DVII || + enc110->base.connector.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + enc110->base.analog_engine != ENGINE_ID_UNKNOWN) + enc110->base.output_signals |= SIGNAL_TYPE_RGB; + /* For DCE 8.0 and 8.1, by design, UNIPHY is hardwired to DIG_BE. * SW always assign DIG_FE 1:1 mapped to DIG_FE for non-MST UNIPHY. * SW assign DIG_FE to non-MST UNIPHY first and MST last. So prefer @@ -1787,6 +1865,13 @@ void dce60_link_encoder_construct( enc110->base.preferred_engine = ENGINE_ID_DIGG; break; default: + if (init_data->analog_engine != ENGINE_ID_UNKNOWN) { + /* The connector is analog-only, ie. VGA */ + enc110->base.preferred_engine = init_data->analog_engine; + enc110->base.output_signals = SIGNAL_TYPE_RGB; + enc110->base.transmitter = TRANSMITTER_UNKNOWN; + break; + } ASSERT_CRITICAL(false); enc110->base.preferred_engine = ENGINE_ID_UNKNOWN; } diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h index 261c70e01e33..c58b69bc319b 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_link_encoder.h @@ -101,18 +101,21 @@ SRI(DP_SEC_CNTL, DP, id), \ SRI(DP_VID_STREAM_CNTL, DP, id), \ SRI(DP_DPHY_FAST_TRAINING, DP, id), \ - SRI(DP_SEC_CNTL1, DP, id) + SRI(DP_SEC_CNTL1, DP, id), \ + SR(DAC_ENABLE) #endif #define LE_DCE80_REG_LIST(id)\ SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \ - LE_COMMON_REG_LIST_BASE(id) + LE_COMMON_REG_LIST_BASE(id), \ + SR(DAC_ENABLE) #define LE_DCE100_REG_LIST(id)\ LE_COMMON_REG_LIST_BASE(id), \ SRI(DP_DPHY_BS_SR_SWAP_CNTL, DP, id), \ SRI(DP_DPHY_INTERNAL_CTRL, DP, id), \ - SR(DCI_MEM_PWR_STATUS) + SR(DCI_MEM_PWR_STATUS), \ + SR(DAC_ENABLE) #define LE_DCE110_REG_LIST(id)\ LE_COMMON_REG_LIST_BASE(id), \ @@ -181,6 +184,9 @@ struct dce110_link_enc_registers { uint32_t DP_DPHY_BS_SR_SWAP_CNTL; uint32_t DP_DPHY_HBR2_PATTERN_CONTROL; uint32_t DP_SEC_CNTL1; + + /* DAC registers */ + uint32_t DAC_ENABLE; }; struct dce110_link_encoder { @@ -215,10 +221,6 @@ bool dce110_link_encoder_validate_dvi_output( enum signal_type signal, const struct dc_crtc_timing *crtc_timing); -bool dce110_link_encoder_validate_rgb_output( - const struct dce110_link_encoder *enc110, - const struct dc_crtc_timing *crtc_timing); - bool dce110_link_encoder_validate_dp_output( const struct dce110_link_encoder *enc110, const struct dc_crtc_timing *crtc_timing); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c index 1130d7619b26..574618d5d4a4 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c @@ -1567,3 +1567,17 @@ void dce110_stream_encoder_construct( enc110->se_shift = se_shift; enc110->se_mask = se_mask; } + +static const struct stream_encoder_funcs dce110_an_str_enc_funcs = {}; + +void dce110_analog_stream_encoder_construct( + struct dce110_stream_encoder *enc110, + struct dc_context *ctx, + struct dc_bios *bp, + enum engine_id eng_id) +{ + enc110->base.funcs = &dce110_an_str_enc_funcs; + enc110->base.ctx = ctx; + enc110->base.id = eng_id; + enc110->base.bp = bp; +} diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h index cc5020a8e1e1..068de1392121 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.h @@ -708,6 +708,11 @@ void dce110_stream_encoder_construct( const struct dce_stream_encoder_shift *se_shift, const struct dce_stream_encoder_mask *se_mask); +void dce110_analog_stream_encoder_construct( + struct dce110_stream_encoder *enc110, + struct dc_context *ctx, + struct dc_bios *bp, + enum engine_id eng_id); void dce110_se_audio_mute_control( struct stream_encoder *enc, bool mute); diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c index d37ecfdde4f1..5bfa2b0d2afd 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.c @@ -61,10 +61,9 @@ void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv, dc_dmub_srv_wait_for_inbox0_ack(dmub_srv); } -bool should_use_dmub_lock(struct dc_link *link) +bool dmub_hw_lock_mgr_does_link_require_lock(const struct dc *dc, const struct dc_link *link) { - /* ASIC doesn't support DMUB */ - if (!link->ctx->dmub_srv) + if (!link) return false; if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) @@ -73,16 +72,38 @@ bool should_use_dmub_lock(struct dc_link *link) if (link->replay_settings.replay_feature_enabled) return true; - /* only use HW lock for PSR1 on single eDP */ if (link->psr_settings.psr_version == DC_PSR_VERSION_1) { struct dc_link *edp_links[MAX_NUM_EDP]; int edp_num; - dc_get_edp_links(link->dc, edp_links, &edp_num); - + dc_get_edp_links(dc, edp_links, &edp_num); if (edp_num == 1) return true; } + return false; +} +bool dmub_hw_lock_mgr_does_context_require_lock(const struct dc *dc, const struct dc_state *context) +{ + if (!context) + return false; + for (int i = 0; i < context->stream_count; i++) { + const struct dc_link *link = context->streams[i]->link; + + if (dmub_hw_lock_mgr_does_link_require_lock(dc, link)) + return true; + } return false; } + +bool should_use_dmub_inbox1_lock(const struct dc *dc, const struct dc_link *link) +{ + /* ASIC doesn't support DMUB */ + if (!dc->ctx->dmub_srv) + return false; + + if (dc->ctx->dce_version >= DCN_VERSION_4_01) + return false; + + return dmub_hw_lock_mgr_does_link_require_lock(dc, link); +} diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h index 5a72b168fb4a..4c80ca8484ad 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_hw_lock_mgr.h @@ -37,6 +37,16 @@ void dmub_hw_lock_mgr_cmd(struct dc_dmub_srv *dmub_srv, void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv, union dmub_inbox0_cmd_lock_hw hw_lock_cmd); -bool should_use_dmub_lock(struct dc_link *link); +/** + * should_use_dmub_inbox1_lock() - Checks if the DMCUB hardware lock via inbox1 should be used. + * + * @dc: pointer to DC object + * @link: optional pointer to the link object to check for enabled link features + * + * Return: true if the inbox1 lock should be used, false otherwise + */ +bool should_use_dmub_inbox1_lock(const struct dc *dc, const struct dc_link *link); +bool dmub_hw_lock_mgr_does_link_require_lock(const struct dc *dc, const struct dc_link *link); +bool dmub_hw_lock_mgr_does_context_require_lock(const struct dc *dc, const struct dc_state *context); #endif /*_DMUB_HW_LOCK_MGR_H_ */ diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c index f9542edff14b..cf1372aaff6c 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.c @@ -213,7 +213,8 @@ static bool dmub_replay_copy_settings(struct dmub_replay *dmub, */ static void dmub_replay_set_coasting_vtotal(struct dmub_replay *dmub, uint32_t coasting_vtotal, - uint8_t panel_inst) + uint8_t panel_inst, + uint16_t frame_skip_number) { union dmub_rb_cmd cmd; struct dc_context *dc = dmub->ctx; @@ -227,6 +228,7 @@ static void dmub_replay_set_coasting_vtotal(struct dmub_replay *dmub, pCmd->header.payload_bytes = sizeof(struct dmub_cmd_replay_set_coasting_vtotal_data); pCmd->replay_set_coasting_vtotal_data.coasting_vtotal = (coasting_vtotal & 0xFFFF); pCmd->replay_set_coasting_vtotal_data.coasting_vtotal_high = (coasting_vtotal & 0xFFFF0000) >> 16; + pCmd->replay_set_coasting_vtotal_data.frame_skip_number = frame_skip_number; dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } @@ -283,7 +285,7 @@ static void dmub_replay_residency(struct dmub_replay *dmub, uint8_t panel_inst, * Set REPLAY power optimization flags and coasting vtotal. */ static void dmub_replay_set_power_opt_and_coasting_vtotal(struct dmub_replay *dmub, - unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal) + unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal, uint16_t frame_skip_number) { union dmub_rb_cmd cmd; struct dc_context *dc = dmub->ctx; @@ -301,6 +303,7 @@ static void dmub_replay_set_power_opt_and_coasting_vtotal(struct dmub_replay *dm pCmd->replay_set_power_opt_data.panel_inst = panel_inst; pCmd->replay_set_coasting_vtotal_data.coasting_vtotal = (coasting_vtotal & 0xFFFF); pCmd->replay_set_coasting_vtotal_data.coasting_vtotal_high = (coasting_vtotal & 0xFFFF0000) >> 16; + pCmd->replay_set_coasting_vtotal_data.frame_skip_number = frame_skip_number; dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT); } @@ -384,6 +387,19 @@ static void dmub_replay_send_cmd(struct dmub_replay *dmub, cmd.replay_disabled_adaptive_sync_sdp.data.force_disabled = cmd_element->disabled_adaptive_sync_sdp_data.force_disabled; break; + case Replay_Set_Version: + //Header + cmd.replay_set_version.header.sub_type = + DMUB_CMD__REPLAY_SET_VERSION; + cmd.replay_set_version.header.payload_bytes = + sizeof(struct dmub_rb_cmd_replay_set_version) - + sizeof(struct dmub_cmd_header); + //Cmd Body + cmd.replay_set_version.replay_set_version_data.panel_inst = + cmd_element->version_data.panel_inst; + cmd.replay_set_version.replay_set_version_data.version = + cmd_element->version_data.version; + break; case Replay_Set_General_Cmd: //Header cmd.replay_set_general_cmd.header.sub_type = diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h index e6346c0ffc0e..07c79739a980 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_replay.h @@ -27,11 +27,12 @@ struct dmub_replay_funcs { void (*replay_send_cmd)(struct dmub_replay *dmub, enum replay_FW_Message_type msg, union dmub_replay_cmd_set *cmd_element); void (*replay_set_coasting_vtotal)(struct dmub_replay *dmub, uint32_t coasting_vtotal, - uint8_t panel_inst); + uint8_t panel_inst, uint16_t frame_skip_number); void (*replay_residency)(struct dmub_replay *dmub, uint8_t panel_inst, uint32_t *residency, const bool is_start, const enum pr_residency_mode mode); void (*replay_set_power_opt_and_coasting_vtotal)(struct dmub_replay *dmub, - unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal); + unsigned int power_opt, uint8_t panel_inst, uint32_t coasting_vtotal, + uint16_t frame_skip_number); }; struct dmub_replay *dmub_replay_create(struct dc_context *ctx); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/Makefile b/drivers/gpu/drm/amd/display/dc/dml2/Makefile deleted file mode 100644 index 4c21ce42054c..000000000000 --- a/drivers/gpu/drm/amd/display/dc/dml2/Makefile +++ /dev/null @@ -1,141 +0,0 @@ -# SPDX-License-Identifier: MIT */ -# -# Copyright 2023 Advanced Micro Devices, Inc. -# -# Permission is hereby granted, free of charge, to any person obtaining a -# copy of this software and associated documentation files (the "Software"), -# to deal in the Software without restriction, including without limitation -# the rights to use, copy, modify, merge, publish, distribute, sublicense, -# and/or sell copies of the Software, and to permit persons to whom the -# Software is furnished to do so, subject to the following conditions: -# -# The above copyright notice and this permission notice shall be included in -# all copies or substantial portions of the Software. -# -# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR -# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, -# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL -# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR -# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, -# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR -# OTHER DEALINGS IN THE SOFTWARE. -# -# Authors: AMD -# -# Makefile for dml2. - -dml2_ccflags := $(CC_FLAGS_FPU) -dml2_rcflags := $(CC_FLAGS_NO_FPU) - -ifneq ($(CONFIG_FRAME_WARN),0) - ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) - ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) - frame_warn_limit := 4096 - else - frame_warn_limit := 3072 - endif - else - frame_warn_limit := 2048 - endif - - ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) - frame_warn_flag := -Wframe-larger-than=$(frame_warn_limit) - endif -endif - -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2 -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_core -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_mcg/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_dpmm/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_pmo/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/dml2_standalone_libraries/ -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/src/inc -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/inc -subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2/dml21/ - -CFLAGS_$(AMDDALPATH)/dc/dml2/display_mode_core.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/display_mode_util.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_wrapper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_utils.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_policy.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_translation_helper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_mall_phantom.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml_display_rq_dlg_calc.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml2_dc_resource_mgmt.o := $(dml2_ccflags) - -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/display_mode_core.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/display_mode_util.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_wrapper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_utils.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_policy.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_translation_helper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_mall_phantom.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml_display_rq_dlg_calc.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml2_dc_resource_mgmt.o := $(dml2_rcflags) - -DML2 = display_mode_core.o display_mode_util.o dml2_wrapper.o \ - dml2_utils.o dml2_policy.o dml2_translation_helper.o dml2_dc_resource_mgmt.o dml2_mall_phantom.o \ - dml_display_rq_dlg_calc.o - -AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2/,$(DML2)) - -AMD_DISPLAY_FILES += $(AMD_DAL_DML2) - -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_ccflags) -CFLAGS_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_ccflags) - -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/src/dml21_wrapper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_translation_helper.o := $(dml2_rcflags) -CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2/dml21/dml21_utils.o := $(dml2_rcflags) - -DML21 := src/dml2_top/dml2_top_interfaces.o -DML21 += src/dml2_top/dml2_top_soc15.o -DML21 += src/dml2_core/dml2_core_dcn4.o -DML21 += src/dml2_core/dml2_core_utils.o -DML21 += src/dml2_core/dml2_core_factory.o -DML21 += src/dml2_core/dml2_core_dcn4_calcs.o -DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o -DML21 += src/dml2_dpmm/dml2_dpmm_factory.o -DML21 += src/dml2_mcg/dml2_mcg_dcn4.o -DML21 += src/dml2_mcg/dml2_mcg_factory.o -DML21 += src/dml2_pmo/dml2_pmo_dcn3.o -DML21 += src/dml2_pmo/dml2_pmo_factory.o -DML21 += src/dml2_pmo/dml2_pmo_dcn4_fams2.o -DML21 += src/dml2_standalone_libraries/lib_float_math.o -DML21 += dml21_translation_helper.o -DML21 += dml21_wrapper.o -DML21 += dml21_utils.o - -AMD_DAL_DML21 = $(addprefix $(AMDDALPATH)/dc/dml2/dml21/,$(DML21)) - -AMD_DISPLAY_FILES += $(AMD_DAL_DML21) - diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile new file mode 100644 index 000000000000..97e068b6bf6b --- /dev/null +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/Makefile @@ -0,0 +1,140 @@ +# SPDX-License-Identifier: MIT */ +# +# Copyright 2023 Advanced Micro Devices, Inc. +# +# Permission is hereby granted, free of charge, to any person obtaining a +# copy of this software and associated documentation files (the "Software"), +# to deal in the Software without restriction, including without limitation +# the rights to use, copy, modify, merge, publish, distribute, sublicense, +# and/or sell copies of the Software, and to permit persons to whom the +# Software is furnished to do so, subject to the following conditions: +# +# The above copyright notice and this permission notice shall be included in +# all copies or substantial portions of the Software. +# +# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR +# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, +# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR +# OTHER DEALINGS IN THE SOFTWARE. +# +# Authors: AMD +# +# Makefile for dml2. + +dml2_ccflags := $(CC_FLAGS_FPU) +dml2_rcflags := $(CC_FLAGS_NO_FPU) + +ifneq ($(CONFIG_FRAME_WARN),0) + ifeq ($(filter y,$(CONFIG_KASAN)$(CONFIG_KCSAN)),y) + ifeq ($(CONFIG_CC_IS_CLANG)$(CONFIG_COMPILE_TEST),yy) + frame_warn_limit := 4096 + else + frame_warn_limit := 3072 + endif + else + frame_warn_limit := 2056 + endif + + ifeq ($(call test-lt, $(CONFIG_FRAME_WARN), $(frame_warn_limit)),y) + frame_warn_flag := -Wframe-larger-than=$(frame_warn_limit) + endif +endif + +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0 +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_core +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_mcg/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_dpmm/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_pmo/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/ +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/src/inc +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/inc +subdir-ccflags-y += -I$(FULL_AMD_DISPLAY_PATH)/dc/dml2_0/dml21/ + +CFLAGS_$(AMDDALPATH)/dc/dml2_0/display_mode_core.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/display_mode_util.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_wrapper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_utils.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_policy.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_translation_helper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_mall_phantom.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml_display_rq_dlg_calc.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml2_dc_resource_mgmt.o := $(dml2_ccflags) + +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/display_mode_core.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/display_mode_util.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_wrapper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_policy.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_translation_helper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_mall_phantom.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml_display_rq_dlg_calc.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml2_dc_resource_mgmt.o := $(dml2_rcflags) + +DML2 = display_mode_core.o display_mode_util.o dml2_wrapper.o \ + dml2_utils.o dml2_policy.o dml2_translation_helper.o dml2_dc_resource_mgmt.o dml2_mall_phantom.o \ + dml_display_rq_dlg_calc.o + +AMD_DAL_DML2 = $(addprefix $(AMDDALPATH)/dc/dml2_0/,$(DML2)) + +AMD_DISPLAY_FILES += $(AMD_DAL_DML2) + +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_ccflags) $(frame_warn_flag) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml21_wrapper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_translation_helper.o := $(dml2_ccflags) +CFLAGS_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_utils.o := $(dml2_ccflags) + +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/src/dml21_wrapper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_translation_helper.o := $(dml2_rcflags) +CFLAGS_REMOVE_$(AMDDALPATH)/dc/dml2_0/dml21/dml21_utils.o := $(dml2_rcflags) + +DML21 := src/dml2_top/dml2_top_interfaces.o +DML21 += src/dml2_top/dml2_top_soc15.o +DML21 += src/dml2_core/dml2_core_dcn4.o +DML21 += src/dml2_core/dml2_core_utils.o +DML21 += src/dml2_core/dml2_core_factory.o +DML21 += src/dml2_core/dml2_core_dcn4_calcs.o +DML21 += src/dml2_dpmm/dml2_dpmm_dcn4.o +DML21 += src/dml2_dpmm/dml2_dpmm_factory.o +DML21 += src/dml2_mcg/dml2_mcg_dcn4.o +DML21 += src/dml2_mcg/dml2_mcg_factory.o +DML21 += src/dml2_pmo/dml2_pmo_dcn3.o +DML21 += src/dml2_pmo/dml2_pmo_factory.o +DML21 += src/dml2_pmo/dml2_pmo_dcn4_fams2.o +DML21 += src/dml2_standalone_libraries/lib_float_math.o +DML21 += dml21_translation_helper.o +DML21 += dml21_wrapper.o +DML21 += dml21_utils.o + +AMD_DAL_DML21 = $(addprefix $(AMDDALPATH)/dc/dml2_0/dml21/,$(DML21)) + +AMD_DISPLAY_FILES += $(AMD_DAL_DML21) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h b/drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h index e450445bc05d..b954c9648fbe 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/cmntypes.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/cmntypes.h @@ -53,17 +53,17 @@ typedef const void *const_pvoid; typedef const char *const_pchar; typedef struct rgba_struct { - uint8 a; - uint8 r; - uint8 g; - uint8 b; + uint8 a; + uint8 r; + uint8 g; + uint8 b; } rgba_t; typedef struct { - uint8 blue; - uint8 green; - uint8 red; - uint8 alpha; + uint8 blue; + uint8 green; + uint8 red; + uint8 alpha; } gen_color_t; typedef union { @@ -87,7 +87,7 @@ typedef union { } uintfloat64; #ifndef UNREFERENCED_PARAMETER -#define UNREFERENCED_PARAMETER(x) x = x +#define UNREFERENCED_PARAMETER(x) (x = x) #endif #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c index 4b9b2e84d381..c468f492b876 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.c @@ -10205,6 +10205,7 @@ dml_bool_t dml_get_is_phantom_pipe(struct display_mode_lib_st *mode_lib, dml_uin return (mode_lib->ms.cache_display_cfg.plane.UseMALLForPStateChange[plane_idx] == dml_use_mall_pstate_change_phantom_pipe); } + #define dml_get_per_surface_var_func(variable, type, interval_var) type dml_get_##variable(struct display_mode_lib_st *mode_lib, dml_uint_t surface_idx) \ { \ dml_uint_t plane_idx; \ @@ -10333,3 +10334,4 @@ dml_get_per_surface_var_func(bigk_fragment_size, dml_uint_t, mode_lib->mp.BIGK_F dml_get_per_surface_var_func(dpte_bytes_per_row, dml_uint_t, mode_lib->mp.PixelPTEBytesPerRow); dml_get_per_surface_var_func(meta_bytes_per_row, dml_uint_t, mode_lib->mp.MetaRowByte); dml_get_per_surface_var_func(det_buffer_size_kbytes, dml_uint_t, mode_lib->ms.DETBufferSizeInKByte); + diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h index a38ed89c47a9..a38ed89c47a9 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h index dbeb08466092..5b40dcdc4406 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_core_structs.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_core_structs.h @@ -274,7 +274,6 @@ enum dml_clk_cfg_policy { dml_use_state_freq = 2 }; - struct soc_state_bounding_box_st { dml_float_t socclk_mhz; dml_float_t dscclk_mhz; @@ -1894,7 +1893,7 @@ struct display_mode_lib_scratch_st { struct CalculatePrefetchSchedule_params_st CalculatePrefetchSchedule_params; }; -/// @brief Represent the overall soc/ip enviroment. It contains data structure represent the soc/ip characteristic and also structures that hold calculation output +/// @brief Represent the overall soc/ip environment. It contains data structure represent the soc/ip characteristic and also structures that hold calculation output struct display_mode_lib_st { dml_uint_t project; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_lib_defines.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_lib_defines.h index 14d389525296..e574c81edf5e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_lib_defines.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_lib_defines.h @@ -52,7 +52,7 @@ #define __DML_VBA_DEBUG__ #define __DML_VBA_ENABLE_INLINE_CHECK_ 0 #define __DML_VBA_MIN_VSTARTUP__ 9 //<brief At which vstartup the DML start to try if the mode can be supported -#define __DML_ARB_TO_RET_DELAY__ 7 + 95 //<brief Delay in DCFCLK from ARB to DET (1st num is ARB to SDPIF, 2nd number is SDPIF to DET) +#define __DML_ARB_TO_RET_DELAY__ (7 + 95) //<brief Delay in DCFCLK from ARB to DET (1st num is ARB to SDPIF, 2nd number is SDPIF to DET) #define __DML_MIN_DCFCLK_FACTOR__ 1.15 //<brief fudge factor for min dcfclk calclation #define __DML_MAX_VRATIO_PRE__ 4.0 //<brief Prefetch schedule max vratio #define __DML_MAX_VRATIO_PRE_OTO__ 4.0 //<brief Prefetch schedule max vratio for one to one scheduling calculation for prefetch diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.c b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.c index 89890c88fd66..89890c88fd66 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h index 113b0265e1d1..a82b49cf7fb0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/display_mode_util.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/display_mode_util.h @@ -30,7 +30,6 @@ #include "display_mode_core_structs.h" #include "cmntypes.h" - #include "dml_assert.h" #include "dml_logging.h" @@ -72,5 +71,4 @@ __DML_DLL_EXPORT__ dml_uint_t dml_get_plane_idx(const struct display_mode_lib_st __DML_DLL_EXPORT__ dml_uint_t dml_get_pipe_idx(const struct display_mode_lib_st *mode_lib, dml_uint_t plane_idx); __DML_DLL_EXPORT__ void dml_calc_pipe_plane_mapping(const struct dml_hw_resource_st *hw, dml_uint_t *pipe_plane); - #endif diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c index bf5e7f4e0416..bf5e7f4e0416 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h index 9880d3e0398e..9880d3e0398e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_translation_helper.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c index ee721606b883..ee721606b883 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h index 4bff52eaaef8..4bff52eaaef8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_utils.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c index 08f7f03b1023..798abb2b2e67 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.c @@ -224,9 +224,7 @@ static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_s dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); /* Populate stream, plane mappings and other fields in display config. */ - DC_FP_START(); result = dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); - DC_FP_END(); if (!result) return false; @@ -281,9 +279,7 @@ static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *co dml_ctx->config.svp_pstate.callbacks.release_phantom_streams_and_planes(in_dc, context); mode_support->dml2_instance = dml_init->dml2_instance; - DC_FP_START(); dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); - DC_FP_END(); dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params.programming = dml_ctx->v21.mode_programming.programming; DC_FP_START(); is_supported = dml2_check_mode_supported(mode_support); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h index 15f92029d2e5..15f92029d2e5 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/dml21_wrapper.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h index 793e1c038efd..16a4f97bca4e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/bounding_boxes/dcn4_soc_bb.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/bounding_boxes/dcn4_soc_bb.h @@ -2,7 +2,6 @@ // // Copyright 2024 Advanced Micro Devices, Inc. - #ifndef __DML_DML_DCN4_SOC_BB__ #define __DML_DML_DCN4_SOC_BB__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h index 281d7ad230d8..281d7ad230d8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml2_external_lib_deps.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml2_external_lib_deps.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h index a64ec4dcf11a..a64ec4dcf11a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h index 91955bbe24b8..bf57df42d1d9 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_dchub_registers.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h @@ -46,7 +46,6 @@ struct dml2_display_dlg_regs { uint32_t dst_y_delta_drq_limit; uint32_t refcyc_per_vm_dmdata; uint32_t dmdata_dl_delta; - uint32_t dst_y_svp_drq_limit; // MRQ uint32_t refcyc_per_meta_chunk_vblank_l; @@ -122,6 +121,8 @@ struct dml2_display_rq_regs { uint32_t crq_expansion_mode; uint32_t plane1_base_address; uint32_t unbounded_request_enabled; + bool pte_buffer_mode; + bool force_one_row_for_frame; // MRQ uint32_t mrq_expansion_mode; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h index e8dc6471c0be..35aa954248cd 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_display_cfg_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_display_cfg_types.h @@ -49,6 +49,11 @@ enum dml2_source_format_class { dml2_422_packed_12 = 18 }; +enum dml2_sample_positioning { + dml2_interstitial = 0, + dml2_cosited = 1 +}; + enum dml2_rotation_angle { dml2_rotation_0 = 0, dml2_rotation_90 = 1, @@ -82,6 +87,15 @@ enum dml2_output_link_dp_rate { dml2_dp_rate_uhbr20 = 6 }; +enum dml2_pstate_type { + dml2_pstate_type_uclk = 0, + dml2_pstate_type_fclk = 1, + dml2_pstate_type_ppt = 2, + dml2_pstate_type_temp_read = 3, + dml2_pstate_type_dummy_pstate = 4, + dml2_pstate_type_count = 5 +}; + enum dml2_uclk_pstate_change_strategy { dml2_uclk_pstate_change_strategy_auto = 0, dml2_uclk_pstate_change_strategy_force_vactive = 1, @@ -222,7 +236,11 @@ struct dml2_composition_cfg { struct { bool enabled; + bool easf_enabled; + bool isharp_enabled; bool upsp_enabled; + enum dml2_sample_positioning upsp_sample_positioning; + unsigned int upsp_vtaps; struct { double h_ratio; double v_ratio; @@ -384,7 +402,7 @@ struct dml2_plane_parameters { // reserved_vblank_time_ns is the minimum time to reserve in vblank for Twait // The actual reserved vblank time used for the corresponding stream in mode_programming would be at least as much as this per-plane override. long reserved_vblank_time_ns; - unsigned int max_vactive_det_fill_delay_us; // 0 = no reserved time, +ve = explicit max delay + unsigned int max_vactive_det_fill_delay_us[dml2_pstate_type_count]; // 0 = no reserved time, +ve = explicit max delay unsigned int gpuvm_min_page_size_kbytes; unsigned int hostvm_min_page_size_kbytes; @@ -413,7 +431,6 @@ struct dml2_stream_parameters { bool disable_dynamic_odm; bool disable_subvp; int minimum_vblank_idle_requirement_us; - bool minimize_active_latency_hiding; struct { struct { @@ -456,6 +473,7 @@ struct dml2_display_cfg { bool enable; bool value; } force_nom_det_size_kbytes; + bool mode_support_check_disable; bool mcache_admissibility_check_disable; bool surface_viewport_size_check_disable; @@ -478,7 +496,6 @@ struct dml2_display_cfg { bool synchronize_ddr_displays_for_uclk_pstate_change; bool max_outstanding_when_urgent_expected_disable; bool enable_subvp_implicit_pmo; //enables PMO to switch pipe uclk strategy to subvp, and generate phantom programming - unsigned int best_effort_min_active_latency_hiding_us; bool all_streams_blanked; } overrides; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h index 8f624a912e78..8f624a912e78 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_policy_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_policy_types.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h index 176f55947664..1fbc520c2540 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_soc_parameter_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_soc_parameter_types.h @@ -89,8 +89,8 @@ struct dml2_soc_qos_parameters { struct dml2_soc_power_management_parameters { double dram_clk_change_blackout_us; - double dram_clk_change_read_only_us; - double dram_clk_change_write_only_us; + double dram_clk_change_read_only_us; // deprecated + double dram_clk_change_write_only_us; // deprecated double fclk_change_blackout_us; double g7_ppt_blackout_us; double g7_temperature_read_blackout_us; @@ -145,6 +145,8 @@ struct dml2_soc_bb { struct dml2_soc_vmin_clock_limits vmin_limit; double lower_bound_bandwidth_dchub; + double fraction_of_urgent_bandwidth_nominal_target; + double fraction_of_urgent_bandwidth_flip_target; unsigned int dprefclk_mhz; unsigned int xtalclk_mhz; unsigned int pcie_refclk_mhz; @@ -170,6 +172,7 @@ struct dml2_soc_bb { struct dml2_ip_capabilities { unsigned int pipe_count; unsigned int otg_count; + unsigned int TDLUT_33cube_count; unsigned int num_dsc; unsigned int max_num_dp2p0_streams; unsigned int max_num_hdmi_frl_outputs; @@ -188,7 +191,9 @@ struct dml2_ip_capabilities { unsigned int subvp_prefetch_end_to_mall_start_us; unsigned int subvp_fw_processing_delay; unsigned int max_vactive_det_fill_delay_us; - + unsigned int ppt_max_allow_delay_us; + unsigned int temp_read_max_allow_delay_us; + unsigned int dummy_pstate_max_allow_delay_us; /* FAMS2 delays */ struct { unsigned int max_allow_delay_us; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h index 41adb1104d0f..452e4a2e72c0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h @@ -70,6 +70,8 @@ struct dml2_pmo_options { bool disable_dyn_odm; bool disable_dyn_odm_for_multi_stream; bool disable_dyn_odm_for_stream_with_svp; + struct dml2_pmo_pstate_strategy *override_strategy_lists[DML2_MAX_PLANES]; + unsigned int num_override_strategies_per_list[DML2_MAX_PLANES]; }; struct dml2_options { @@ -310,6 +312,7 @@ struct dml2_mode_support_info { bool NumberOfOTGSupport; bool NumberOfHDMIFRLSupport; bool NumberOfDP2p0Support; + bool NumberOfTDLUT33cubeSupport; bool WritebackScaleRatioAndTapsSupport; bool CursorSupport; bool PitchSupport; @@ -357,6 +360,8 @@ struct dml2_mode_support_info { unsigned int AlignedCPitch[DML2_MAX_PLANES]; bool g6_temp_read_support; bool temp_read_or_ppt_support; + bool qos_bandwidth_support; + bool dcfclk_support; }; // dml2_mode_support_info struct dml2_display_cfg_programming { @@ -671,6 +676,8 @@ struct dml2_display_cfg_programming { unsigned int PrefetchMode[DML2_MAX_PLANES]; // LEGACY_ONLY bool ROBUrgencyAvoidance; double LowestPrefetchMargin; + + unsigned int pstate_recout_reduction_lines[DML2_MAX_PLANES]; } misc; struct dml2_mode_support_info mode_support_info; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c index 6ee37386f672..eba948e187c1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.c @@ -28,6 +28,7 @@ struct dml2_core_ip_params core_dcn4_ip_caps_base = { .writeback_interface_buffer_size_kbytes = 90, //Number of pipes after DCN Pipe harvesting .max_num_dpp = 4, + .max_num_opp = 4, .max_num_otg = 4, .max_num_wb = 1, .max_dchub_pscl_bw_pix_per_clk = 4, diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h index a68bb001a346..a68bb001a346 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index bf62d42b3f78..a02e9fd6b5ca 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -1303,6 +1303,7 @@ static double TruncToValidBPP( MinDSCBPP = 8; MaxDSCBPP = 16; } else { + if (Output == dml2_hdmi || Output == dml2_hdmifrl) { NonDSCBPP0 = 24; NonDSCBPP1 = 24; @@ -1320,6 +1321,7 @@ static double TruncToValidBPP( MaxDSCBPP = 16; } } + if (Output == dml2_dp2p0) { MaxLinkBPP = LinkBitRate * Lanes / PixelClock * 128.0 / 132.0 * 383.0 / 384.0 * 65536.0 / 65540.0; } else if (DSCEnable && Output == dml2_dp) { @@ -4047,7 +4049,9 @@ static bool ValidateODMMode(enum dml2_odm_mode ODMMode, bool UseDSC, unsigned int NumberOfDSCSlices, unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, unsigned int MaxNumDPP, + unsigned int MaxNumOPP, double DISPCLKRequired, unsigned int NumberOfDPPRequired, unsigned int MaxHActiveForDSC, @@ -4063,7 +4067,7 @@ static bool ValidateODMMode(enum dml2_odm_mode ODMMode, if (DISPCLKRequired > MaxDispclk) return false; - if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP) + if ((TotalNumberOfActiveDPP + NumberOfDPPRequired) > MaxNumDPP || (TotalNumberOfActiveOPP + NumberOfDPPRequired) > MaxNumOPP) return false; if (are_odm_segments_symmetrical) { if (HActive % (NumberOfDPPRequired * pixels_per_clock_cycle)) @@ -4109,7 +4113,9 @@ static noinline_for_stack void CalculateODMMode( double MaxDispclk, bool DSCEnable, unsigned int TotalNumberOfActiveDPP, + unsigned int TotalNumberOfActiveOPP, unsigned int MaxNumDPP, + unsigned int MaxNumOPP, double PixelClock, unsigned int NumberOfDSCSlices, @@ -4179,7 +4185,9 @@ static noinline_for_stack void CalculateODMMode( UseDSC, NumberOfDSCSlices, TotalNumberOfActiveDPP, + TotalNumberOfActiveOPP, MaxNumDPP, + MaxNumOPP, DISPCLKRequired, NumberOfDPPRequired, MaxHActiveForDSC, @@ -6964,7 +6972,7 @@ static void calculate_bytes_to_fetch_required_to_hide_latency( stream_index = p->display_cfg->plane_descriptors[plane_index].stream_index; - dst_lines_to_hide = (unsigned int)math_ceil(p->latency_to_hide_us / + dst_lines_to_hide = (unsigned int)math_ceil(p->latency_to_hide_us[0] / ((double)p->display_cfg->stream_descriptors[stream_index].timing.h_total / (double)p->display_cfg->stream_descriptors[stream_index].timing.pixel_clock_khz * 1000.0)); @@ -7061,9 +7069,9 @@ static void calculate_excess_vactive_bandwidth_required( excess_vactive_fill_bw_l[plane_index] = 0.0; excess_vactive_fill_bw_c[plane_index] = 0.0; - if (display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us > 0) { - excess_vactive_fill_bw_l[plane_index] = (double)bytes_required_l[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; - excess_vactive_fill_bw_c[plane_index] = (double)bytes_required_c[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us; + if (display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us[dml2_pstate_type_uclk] > 0) { + excess_vactive_fill_bw_l[plane_index] = (double)bytes_required_l[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us[dml2_pstate_type_uclk]; + excess_vactive_fill_bw_c[plane_index] = (double)bytes_required_c[plane_index] / (double)display_cfg->plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us[dml2_pstate_type_uclk]; } } } @@ -8358,6 +8366,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out CalculateSwathAndDETConfiguration(&mode_lib->scratch, CalculateSwathAndDETConfiguration_params); mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; mode_lib->ms.support.TotalAvailablePipesSupport = true; for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { @@ -8393,7 +8402,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.max_dispclk_freq_mhz, false, // DSCEnable mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), mode_lib->ms.support.NumberOfDSCSlices[k], @@ -8412,7 +8423,9 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.max_dispclk_freq_mhz, true, // DSCEnable mode_lib->ms.TotalNumberOfActiveDPP, + mode_lib->ms.TotalNumberOfActiveOPP, mode_lib->ip.max_num_dpp, + mode_lib->ip.max_num_opp, ((double)display_cfg->stream_descriptors[display_cfg->plane_descriptors[k].stream_index].timing.pixel_clock_khz / 1000), mode_lib->ms.support.NumberOfDSCSlices[k], @@ -8516,20 +8529,23 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 1; + mode_lib->ms.NoOfOPP[k] = 1; if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_4to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 4; + mode_lib->ms.NoOfOPP[k] = 4; } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_3to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 3; + mode_lib->ms.NoOfOPP[k] = 3; } else if (mode_lib->ms.ODMMode[k] == dml2_odm_mode_combine_2to1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 2; + mode_lib->ms.NoOfOPP[k] = 2; } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 2) { mode_lib->ms.MPCCombine[k] = true; mode_lib->ms.NoOfDPP[k] = 2; - mode_lib->ms.TotalNumberOfActiveDPP++; } else if (display_cfg->plane_descriptors[k].overrides.mpcc_combine_factor == 1) { mode_lib->ms.MPCCombine[k] = false; mode_lib->ms.NoOfDPP[k] = 1; @@ -8540,7 +8556,6 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out if ((mode_lib->ms.MinDPPCLKUsingSingleDPP[k] > mode_lib->ms.max_dppclk_freq_mhz) || !mode_lib->ms.SingleDPPViewportSizeSupportPerSurface[k]) { mode_lib->ms.MPCCombine[k] = true; mode_lib->ms.NoOfDPP[k] = 2; - mode_lib->ms.TotalNumberOfActiveDPP++; } } #if defined(__DML_VBA_DEBUG__) @@ -8548,8 +8563,16 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out #endif } + mode_lib->ms.TotalNumberOfActiveDPP = 0; + mode_lib->ms.TotalNumberOfActiveOPP = 0; + for (k = 0; k < mode_lib->ms.num_active_planes; ++k) { + mode_lib->ms.TotalNumberOfActiveDPP += mode_lib->ms.NoOfDPP[k]; + mode_lib->ms.TotalNumberOfActiveOPP += mode_lib->ms.NoOfOPP[k]; + } if (mode_lib->ms.TotalNumberOfActiveDPP > (unsigned int)mode_lib->ip.max_num_dpp) mode_lib->ms.support.TotalAvailablePipesSupport = false; + if (mode_lib->ms.TotalNumberOfActiveOPP > (unsigned int)mode_lib->ip.max_num_opp) + mode_lib->ms.support.TotalAvailablePipesSupport = false; mode_lib->ms.TotalNumberOfSingleDPPSurfaces = 0; @@ -9028,11 +9051,11 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->ms.SwathWidthC; calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->ms.SwathHeightY; calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->ms.SwathHeightC; - calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us[0] = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; /* outputs */ - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l[dml2_pstate_type_uclk]; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c[dml2_pstate_type_uclk]; calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); @@ -9040,8 +9063,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out calculate_excess_vactive_bandwidth_required( display_cfg, mode_lib->ms.num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, + s->pstate_bytes_required_l[dml2_pstate_type_uclk], + s->pstate_bytes_required_c[dml2_pstate_type_uclk], /* outputs */ mode_lib->ms.excess_vactive_fill_bw_l, mode_lib->ms.excess_vactive_fill_bw_c); @@ -9483,8 +9506,8 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out calculate_vactive_det_fill_latency( display_cfg, mode_lib->ms.num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, + s->pstate_bytes_required_l[dml2_pstate_type_uclk], + s->pstate_bytes_required_c[dml2_pstate_type_uclk], mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p0, mode_lib->ms.dcc_dram_bw_nom_overhead_factor_p1, mode_lib->ms.vactive_sw_bw_l, @@ -9492,7 +9515,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out mode_lib->ms.surface_avg_vactive_required_bw, mode_lib->ms.surface_peak_required_bw, /* outputs */ - mode_lib->ms.dram_change_vactive_det_fill_delay_us); + mode_lib->ms.pstate_vactive_det_fill_delay_us[dml2_pstate_type_uclk]); #ifdef __DML_VBA_DEBUG__ DML_LOG_VERBOSE("DML::%s: max_urgent_latency_us = %f\n", __func__, s->mSOCParameters.max_urgent_latency_us); @@ -10986,11 +11009,11 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_bytes_to_fetch_required_to_hide_latency_params->swath_width_c = mode_lib->mp.SwathWidthC; calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_l = mode_lib->mp.SwathHeightY; calculate_bytes_to_fetch_required_to_hide_latency_params->swath_height_c = mode_lib->mp.SwathHeightC; - calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; + calculate_bytes_to_fetch_required_to_hide_latency_params->latency_to_hide_us[0] = mode_lib->soc.power_management_parameters.dram_clk_change_blackout_us; /* outputs */ - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l; - calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_l = s->pstate_bytes_required_l[dml2_pstate_type_uclk]; + calculate_bytes_to_fetch_required_to_hide_latency_params->bytes_required_c = s->pstate_bytes_required_c[dml2_pstate_type_uclk]; calculate_bytes_to_fetch_required_to_hide_latency(calculate_bytes_to_fetch_required_to_hide_latency_params); @@ -10998,8 +11021,8 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_excess_vactive_bandwidth_required( display_cfg, s->num_active_planes, - s->pstate_bytes_required_l, - s->pstate_bytes_required_c, + s->pstate_bytes_required_l[dml2_pstate_type_uclk], + s->pstate_bytes_required_c[dml2_pstate_type_uclk], /* outputs */ mode_lib->mp.excess_vactive_fill_bw_l, mode_lib->mp.excess_vactive_fill_bw_c); @@ -12756,7 +12779,7 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna { const struct dml2_plane_parameters *plane_descriptor = &display_cfg->display_config.plane_descriptors[plane_index]; const struct dml2_stream_parameters *stream_descriptor = &display_cfg->display_config.stream_descriptors[plane_descriptor->stream_index]; - const struct dml2_fams2_meta *stream_fams2_meta = &display_cfg->stage3.stream_fams2_meta[plane_descriptor->stream_index]; + const struct dml2_pstate_meta *stream_pstate_meta = &display_cfg->stage3.stream_pstate_meta[plane_descriptor->stream_index]; struct dmub_fams2_cmd_stream_static_base_state *base_programming = &fams2_base_programming->stream_v1.base; union dmub_fams2_cmd_stream_static_sub_state *sub_programming = &fams2_sub_programming->stream_v1.sub_state; @@ -12771,24 +12794,24 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna /* from display configuration */ base_programming->htotal = (uint16_t)stream_descriptor->timing.h_total; base_programming->vtotal = (uint16_t)stream_descriptor->timing.v_total; - base_programming->vblank_start = (uint16_t)(stream_fams2_meta->nom_vtotal - + base_programming->vblank_start = (uint16_t)(stream_pstate_meta->nom_vtotal - stream_descriptor->timing.v_front_porch); - base_programming->vblank_end = (uint16_t)(stream_fams2_meta->nom_vtotal - + base_programming->vblank_end = (uint16_t)(stream_pstate_meta->nom_vtotal - stream_descriptor->timing.v_front_porch - stream_descriptor->timing.v_active); base_programming->config.bits.is_drr = stream_descriptor->timing.drr_config.enabled; /* from meta */ base_programming->otg_vline_time_ns = - (unsigned int)(stream_fams2_meta->otg_vline_time_us * 1000.0); - base_programming->scheduling_delay_otg_vlines = (uint8_t)stream_fams2_meta->scheduling_delay_otg_vlines; - base_programming->contention_delay_otg_vlines = (uint8_t)stream_fams2_meta->contention_delay_otg_vlines; - base_programming->vline_int_ack_delay_otg_vlines = (uint8_t)stream_fams2_meta->vertical_interrupt_ack_delay_otg_vlines; - base_programming->drr_keepout_otg_vline = (uint16_t)(stream_fams2_meta->nom_vtotal - + (unsigned int)(stream_pstate_meta->otg_vline_time_us * 1000.0); + base_programming->scheduling_delay_otg_vlines = (uint8_t)stream_pstate_meta->scheduling_delay_otg_vlines; + base_programming->contention_delay_otg_vlines = (uint8_t)stream_pstate_meta->contention_delay_otg_vlines; + base_programming->vline_int_ack_delay_otg_vlines = (uint8_t)stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines; + base_programming->drr_keepout_otg_vline = (uint16_t)(stream_pstate_meta->nom_vtotal - stream_descriptor->timing.v_front_porch - - stream_fams2_meta->method_drr.programming_delay_otg_vlines); - base_programming->allow_to_target_delay_otg_vlines = (uint8_t)stream_fams2_meta->allow_to_target_delay_otg_vlines; - base_programming->max_vtotal = (uint16_t)stream_fams2_meta->max_vtotal; + stream_pstate_meta->method_drr.programming_delay_otg_vlines); + base_programming->allow_to_target_delay_otg_vlines = (uint8_t)stream_pstate_meta->allow_to_target_delay_otg_vlines; + base_programming->max_vtotal = (uint16_t)stream_pstate_meta->max_vtotal; /* from core */ base_programming->config.bits.min_ttu_vblank_usable = true; @@ -12807,11 +12830,11 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna /* legacy vactive */ base_programming->type = FAMS2_STREAM_TYPE_VACTIVE; sub_programming->legacy.vactive_det_fill_delay_otg_vlines = - (uint8_t)stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; + (uint8_t)stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; base_programming->allow_start_otg_vline = - (uint16_t)stream_fams2_meta->method_vactive.common.allow_start_otg_vline; + (uint16_t)stream_pstate_meta->method_vactive.common.allow_start_otg_vline; base_programming->allow_end_otg_vline = - (uint16_t)stream_fams2_meta->method_vactive.common.allow_end_otg_vline; + (uint16_t)stream_pstate_meta->method_vactive.common.allow_end_otg_vline; base_programming->config.bits.clamp_vtotal_min = true; break; case dml2_pstate_method_vblank: @@ -12819,22 +12842,22 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna /* legacy vblank */ base_programming->type = FAMS2_STREAM_TYPE_VBLANK; base_programming->allow_start_otg_vline = - (uint16_t)stream_fams2_meta->method_vblank.common.allow_start_otg_vline; + (uint16_t)stream_pstate_meta->method_vblank.common.allow_start_otg_vline; base_programming->allow_end_otg_vline = - (uint16_t)stream_fams2_meta->method_vblank.common.allow_end_otg_vline; + (uint16_t)stream_pstate_meta->method_vblank.common.allow_end_otg_vline; base_programming->config.bits.clamp_vtotal_min = true; break; case dml2_pstate_method_fw_drr: /* drr */ base_programming->type = FAMS2_STREAM_TYPE_DRR; sub_programming->drr.programming_delay_otg_vlines = - (uint8_t)stream_fams2_meta->method_drr.programming_delay_otg_vlines; + (uint8_t)stream_pstate_meta->method_drr.programming_delay_otg_vlines; sub_programming->drr.nom_stretched_vtotal = - (uint16_t)stream_fams2_meta->method_drr.stretched_vtotal; + (uint16_t)stream_pstate_meta->method_drr.stretched_vtotal; base_programming->allow_start_otg_vline = - (uint16_t)stream_fams2_meta->method_drr.common.allow_start_otg_vline; + (uint16_t)stream_pstate_meta->method_drr.common.allow_start_otg_vline; base_programming->allow_end_otg_vline = - (uint16_t)stream_fams2_meta->method_drr.common.allow_end_otg_vline; + (uint16_t)stream_pstate_meta->method_drr.common.allow_end_otg_vline; /* drr only clamps to vtotal min for single display */ base_programming->config.bits.clamp_vtotal_min = display_cfg->display_config.num_streams == 1; sub_programming->drr.only_stretch_if_required = true; @@ -12847,13 +12870,13 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna (uint16_t)(plane_descriptor->composition.scaler_info.plane0.v_ratio * 1000.0); sub_programming->subvp.vratio_denominator = 1000; sub_programming->subvp.programming_delay_otg_vlines = - (uint8_t)stream_fams2_meta->method_subvp.programming_delay_otg_vlines; + (uint8_t)stream_pstate_meta->method_subvp.programming_delay_otg_vlines; sub_programming->subvp.prefetch_to_mall_otg_vlines = - (uint8_t)stream_fams2_meta->method_subvp.prefetch_to_mall_delay_otg_vlines; + (uint8_t)stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines; sub_programming->subvp.phantom_vtotal = - (uint16_t)stream_fams2_meta->method_subvp.phantom_vtotal; + (uint16_t)stream_pstate_meta->method_subvp.phantom_vtotal; sub_programming->subvp.phantom_vactive = - (uint16_t)stream_fams2_meta->method_subvp.phantom_vactive; + (uint16_t)stream_pstate_meta->method_subvp.phantom_vactive; sub_programming->subvp.config.bits.is_multi_planar = plane_descriptor->surface.plane1.height > 0; sub_programming->subvp.config.bits.is_yuv420 = @@ -12862,9 +12885,9 @@ void dml2_core_calcs_get_stream_fams2_programming(const struct dml2_core_interna plane_descriptor->pixel_format == dml2_420_12; base_programming->allow_start_otg_vline = - (uint16_t)stream_fams2_meta->method_subvp.common.allow_start_otg_vline; + (uint16_t)stream_pstate_meta->method_subvp.common.allow_start_otg_vline; base_programming->allow_end_otg_vline = - (uint16_t)stream_fams2_meta->method_subvp.common.allow_end_otg_vline; + (uint16_t)stream_pstate_meta->method_subvp.common.allow_end_otg_vline; base_programming->config.bits.clamp_vtotal_min = true; break; case dml2_pstate_method_reserved_hw: @@ -12920,7 +12943,8 @@ void dml2_core_calcs_get_plane_support_info(const struct dml2_display_cfg *displ out->active_latency_hiding_us = (int)mode_lib->ms.VActiveLatencyHidingUs[plane_idx]; - out->dram_change_vactive_det_fill_delay_us = (unsigned int)math_ceil(mode_lib->ms.dram_change_vactive_det_fill_delay_us[plane_idx]); + out->vactive_det_fill_delay_us[dml2_pstate_type_uclk] = + (unsigned int)math_ceil(mode_lib->ms.pstate_vactive_det_fill_delay_us[dml2_pstate_type_uclk][plane_idx]); } void dml2_core_calcs_get_stream_support_info(const struct dml2_display_cfg *display_cfg, const struct dml2_core_internal_display_mode_lib *mode_lib, struct core_stream_support_info *out, int plane_index) @@ -13001,7 +13025,7 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.mode_support_info.InvalidCombinationOfMALLUseForPState = mode_lib->ms.support.InvalidCombinationOfMALLUseForPState; out->informative.mode_support_info.ExceededMALLSize = mode_lib->ms.support.ExceededMALLSize; out->informative.mode_support_info.EnoughWritebackUnits = mode_lib->ms.support.EnoughWritebackUnits; - out->informative.mode_support_info.temp_read_or_ppt_support = mode_lib->ms.support.temp_read_or_ppt_support; + out->informative.mode_support_info.temp_read_or_ppt_support = mode_lib->ms.support.global_temp_read_or_ppt_supported; out->informative.mode_support_info.g6_temp_read_support = mode_lib->ms.support.g6_temp_read_support; out->informative.mode_support_info.ExceededMultistreamSlots = mode_lib->ms.support.ExceededMultistreamSlots; @@ -13027,7 +13051,10 @@ void dml2_core_calcs_get_informative(const struct dml2_core_internal_display_mod out->informative.mode_support_info.VRatioInPrefetchSupported = mode_lib->ms.support.VRatioInPrefetchSupported; out->informative.mode_support_info.DISPCLK_DPPCLK_Support = mode_lib->ms.support.DISPCLK_DPPCLK_Support; out->informative.mode_support_info.TotalAvailablePipesSupport = mode_lib->ms.support.TotalAvailablePipesSupport; + out->informative.mode_support_info.NumberOfTDLUT33cubeSupport = mode_lib->ms.support.NumberOfTDLUT33cubeSupport; out->informative.mode_support_info.ViewportSizeSupport = mode_lib->ms.support.ViewportSizeSupport; + out->informative.mode_support_info.qos_bandwidth_support = mode_lib->ms.support.qos_bandwidth_support; + out->informative.mode_support_info.dcfclk_support = mode_lib->ms.support.dcfclk_support; for (k = 0; k < out->display_config.num_planes; k++) { diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h index 27ef0e096b25..27ef0e096b25 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_dcn4_calcs.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c index 640087e862f8..cc4f0663c6d6 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.c @@ -15,6 +15,8 @@ bool dml2_core_create(enum dml2_project_id project_id, struct dml2_core_instance memset(out, 0, sizeof(struct dml2_core_instance)); + out->project_id = project_id; + switch (project_id) { case dml2_project_dcn4x_stage1: result = false; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h index 411c514fe65c..411c514fe65c 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_factory.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_factory.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h index ffb8c09f37a5..1087a8c926ff 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_shared_types.h @@ -36,7 +36,9 @@ struct dml2_core_ip_params { unsigned int max_line_buffer_lines; unsigned int writeback_interface_buffer_size_kbytes; unsigned int max_num_dpp; + unsigned int max_num_opp; unsigned int max_num_otg; + unsigned int TDLUT_33cube_count; unsigned int max_num_wb; unsigned int max_dchub_pscl_bw_pix_per_clk; unsigned int max_pscl_lb_bw_pix_per_clk; @@ -46,6 +48,7 @@ struct dml2_core_ip_params { double max_vscl_ratio; unsigned int max_hscl_taps; unsigned int max_vscl_taps; + unsigned int odm_combine_support_mask; unsigned int num_dsc; unsigned int maximum_dsc_bits_per_component; unsigned int maximum_pixels_per_line_per_dsc_unit; @@ -82,7 +85,6 @@ struct dml2_core_ip_params { unsigned int subvp_swath_height_margin_lines; unsigned int subvp_fw_processing_delay_us; unsigned int subvp_pstate_allow_width_us; - // MRQ bool dcn_mrq_present; unsigned int zero_size_buffer_entries; @@ -103,6 +105,8 @@ struct dml2_core_internal_DmlPipe { unsigned int DPPPerSurface; bool ScalerEnabled; bool UPSPEnabled; + unsigned int UPSPVTaps; + enum dml2_sample_positioning UPSPSamplePositioning; enum dml2_rotation_angle RotationAngle; bool mirrored; unsigned int ViewportHeight; @@ -230,6 +234,7 @@ struct dml2_core_internal_mode_support_info { bool MSOOrODMSplitWithNonDPLink; bool NotEnoughLanesForMSO; bool NumberOfOTGSupport; + bool NumberOfTDLUT33cubeSupport; bool NumberOfHDMIFRLSupport; bool NumberOfDP2p0Support; bool WritebackScaleRatioAndTapsSupport; @@ -259,8 +264,11 @@ struct dml2_core_internal_mode_support_info { bool DCCMetaBufferSizeNotExceeded; enum dml2_pstate_change_support DRAMClockChangeSupport[DML2_MAX_PLANES]; enum dml2_pstate_change_support FCLKChangeSupport[DML2_MAX_PLANES]; + enum dml2_pstate_change_support temp_read_or_ppt_support[DML2_MAX_PLANES]; + bool global_dram_clock_change_support_required; bool global_dram_clock_change_supported; bool global_fclk_change_supported; + bool global_temp_read_or_ppt_supported; bool USRRetrainingSupport; bool AvgBandwidthSupport; bool UrgVactiveBandwidthSupport; @@ -331,7 +339,6 @@ struct dml2_core_internal_mode_support_info { bool incorrect_imall_usage; bool g6_temp_read_support; - bool temp_read_or_ppt_support; struct dml2_core_internal_watermarks watermarks; bool dcfclk_support; @@ -566,6 +573,7 @@ struct dml2_core_internal_mode_support { enum dml2_odm_mode ODMMode[DML2_MAX_PLANES]; unsigned int SurfaceSizeInMALL[DML2_MAX_PLANES]; unsigned int NoOfDPP[DML2_MAX_PLANES]; + unsigned int NoOfOPP[DML2_MAX_PLANES]; bool MPCCombine[DML2_MAX_PLANES]; double dcfclk_deepsleep; double MinDPPCLKUsingSingleDPP[DML2_MAX_PLANES]; @@ -576,6 +584,7 @@ struct dml2_core_internal_mode_support { bool PTEBufferSizeNotExceeded[DML2_MAX_PLANES]; bool DCCMetaBufferSizeNotExceeded[DML2_MAX_PLANES]; unsigned int TotalNumberOfActiveDPP; + unsigned int TotalNumberOfActiveOPP; unsigned int TotalNumberOfSingleDPPSurfaces; unsigned int TotalNumberOfDCCActiveDPP; unsigned int Total3dlutActive; @@ -584,7 +593,7 @@ struct dml2_core_internal_mode_support { double VActiveLatencyHidingMargin[DML2_MAX_PLANES]; double VActiveLatencyHidingUs[DML2_MAX_PLANES]; unsigned int MaxVStartupLines[DML2_MAX_PLANES]; - double dram_change_vactive_det_fill_delay_us[DML2_MAX_PLANES]; + double pstate_vactive_det_fill_delay_us[dml2_pstate_type_count][DML2_MAX_PLANES]; unsigned int num_mcaches_l[DML2_MAX_PLANES]; unsigned int mcache_row_bytes_l[DML2_MAX_PLANES]; @@ -614,8 +623,8 @@ struct dml2_core_internal_mode_support { unsigned int dpte_row_bytes_per_row_l[DML2_MAX_PLANES]; unsigned int dpte_row_bytes_per_row_c[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_l[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_c[DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_l[dml2_pstate_type_count][DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_c[dml2_pstate_type_count][DML2_MAX_PLANES]; unsigned int cursor_bytes_per_chunk[DML2_MAX_PLANES]; unsigned int cursor_bytes_per_line[DML2_MAX_PLANES]; @@ -639,7 +648,7 @@ struct dml2_core_internal_mode_support { unsigned int DSTYAfterScaler[DML2_MAX_PLANES]; unsigned int DSTXAfterScaler[DML2_MAX_PLANES]; - enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; + enum dml2_pstate_method uclk_pstate_switch_modes[DML2_MAX_PLANES]; }; /// @brief A mega structure that houses various info for model programming step. @@ -830,6 +839,7 @@ struct dml2_core_internal_mode_program { double max_urgent_latency_us; double df_response_time_us; + enum dml2_pstate_method uclk_pstate_switch_modes[DML2_MAX_PLANES]; // ------------------- // Output // ------------------- @@ -956,11 +966,12 @@ struct dml2_core_internal_mode_program { double MaxActiveFCLKChangeLatencySupported; bool USRRetrainingSupport; bool g6_temp_read_support; - bool temp_read_or_ppt_support; enum dml2_pstate_change_support FCLKChangeSupport[DML2_MAX_PLANES]; enum dml2_pstate_change_support DRAMClockChangeSupport[DML2_MAX_PLANES]; + enum dml2_pstate_change_support temp_read_or_ppt_support[DML2_MAX_PLANES]; bool global_dram_clock_change_supported; bool global_fclk_change_supported; + bool global_temp_read_or_ppt_supported; double MaxActiveDRAMClockChangeLatencySupported[DML2_MAX_PLANES]; double WritebackAllowFCLKChangeEndPosition[DML2_MAX_PLANES]; double WritebackAllowDRAMClockChangeEndPosition[DML2_MAX_PLANES]; @@ -1127,8 +1138,8 @@ struct dml2_core_calcs_mode_support_locals { unsigned int cursor_bytes[DML2_MAX_PLANES]; bool stream_visited[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_l[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_c[DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_l[dml2_pstate_type_count][DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_c[dml2_pstate_type_count][DML2_MAX_PLANES]; double prefetch_sw_bytes[DML2_MAX_PLANES]; double Tpre_rounded[DML2_MAX_PLANES]; @@ -1219,8 +1230,8 @@ struct dml2_core_calcs_mode_programming_locals { double Tr0_trips_flip_rounded[DML2_MAX_PLANES]; unsigned int per_pipe_flip_bytes[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_l[DML2_MAX_PLANES]; - unsigned int pstate_bytes_required_c[DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_l[dml2_pstate_type_count][DML2_MAX_PLANES]; + unsigned int pstate_bytes_required_c[dml2_pstate_type_count][DML2_MAX_PLANES]; double prefetch_sw_bytes[DML2_MAX_PLANES]; double Tpre_rounded[DML2_MAX_PLANES]; @@ -1306,7 +1317,7 @@ struct dml2_core_calcs_CalculateVMRowAndSwath_params { unsigned int HostVMMinPageSize; unsigned int DCCMetaBufferSizeBytes; bool mrq_present; - enum dml2_pstate_method pstate_switch_modes[DML2_MAX_PLANES]; + enum dml2_pstate_method *uclk_pstate_switch_modes; // Output bool *PTEBufferSizeNotExceeded; @@ -1733,10 +1744,12 @@ struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_param unsigned int max_request_size_bytes; unsigned int *meta_row_height_l; unsigned int *meta_row_height_c; + enum dml2_pstate_method *uclk_pstate_switch_modes; // Output struct dml2_core_internal_watermarks *Watermark; enum dml2_pstate_change_support *DRAMClockChangeSupport; + bool *global_dram_clock_change_support_required; bool *global_dram_clock_change_supported; double *MaxActiveDRAMClockChangeLatencySupported; unsigned int *SubViewportLinesNeededInMALL; @@ -1747,10 +1760,10 @@ struct dml2_core_calcs_CalculateWatermarksMALLUseAndDRAMSpeedChangeSupport_param double *VActiveLatencyHidingMargin; double *VActiveLatencyHidingUs; bool *g6_temp_read_support; - bool *temp_read_or_ppt_support; + enum dml2_pstate_change_support *temp_read_or_ppt_support; + bool *global_temp_read_or_ppt_supported; }; - struct dml2_core_calcs_CalculateSwathAndDETConfiguration_params { const struct dml2_display_cfg *display_cfg; unsigned int ConfigReturnBufferSizeInKByte; @@ -2240,7 +2253,7 @@ struct dml2_core_calcs_calculate_bytes_to_fetch_required_to_hide_latency_params unsigned int *swath_width_c; unsigned int *swath_height_l; unsigned int *swath_height_c; - double latency_to_hide_us; + double latency_to_hide_us[DML2_MAX_PLANES]; /* outputs */ unsigned int *bytes_required_l; @@ -2308,6 +2321,7 @@ struct dml2_core_calcs_mode_support_ex { const struct dml2_display_cfg *in_display_cfg; const struct dml2_mcg_min_clock_table *min_clk_table; int min_clk_index; + enum dml2_project_id project_id; //unsigned int in_state_index; struct dml2_core_internal_mode_support_info *out_evaluation_info; }; @@ -2320,6 +2334,7 @@ struct dml2_core_calcs_mode_programming_ex { const struct dml2_mcg_min_clock_table *min_clk_table; const struct core_display_cfg_support_info *cfg_support_info; int min_clk_index; + enum dml2_project_id project_id; struct dml2_display_cfg_programming *programming; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.c index 5f301befed16..b57d0f6ea6a1 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.c @@ -306,6 +306,8 @@ void dml2_core_utils_print_mode_support_info(const struct dml2_core_internal_mod DML_LOG_VERBOSE("DML: support: ExceededMALLSize = %d\n", support->ExceededMALLSize); if (!fail_only || support->g6_temp_read_support == 0) DML_LOG_VERBOSE("DML: support: g6_temp_read_support = %d\n", support->g6_temp_read_support); + if (!fail_only || (support->global_dram_clock_change_supported == 0 && support->global_dram_clock_change_support_required)) + DML_LOG_VERBOSE("DML: support: dram_clock_change_support = %d\n", support->global_dram_clock_change_supported); if (!fail_only || support->ImmediateFlipSupport == 0) DML_LOG_VERBOSE("DML: support: ImmediateFlipSupport = %d\n", support->ImmediateFlipSupport); if (!fail_only || support->LinkCapacitySupport == 0) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h index 95f0d017add4..95f0d017add4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_utils.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_core/dml2_core_utils.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c index 22969a533a7b..22969a533a7b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h index e7b58f2efda4..e7b58f2efda4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_dcn4.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c index dfd01440737d..dfd01440737d 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h index 20ba2e446f1d..20ba2e446f1d 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_dpmm/dml2_dpmm_factory.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_dpmm/dml2_dpmm_factory.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c index a265f254152c..a265f254152c 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h index 02da6f45cbf7..f54fde8fba90 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_dcn4.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_dcn4.h @@ -10,4 +10,4 @@ bool mcg_dcn4_build_min_clock_table(struct dml2_mcg_build_min_clock_table_params_in_out *in_out); bool mcg_dcn4_unit_test(void); -#endif +#endif
\ No newline at end of file diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c index c60b8fe90819..c60b8fe90819 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h index ad307deca3b0..ad307deca3b0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_mcg/dml2_mcg_factory.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_mcg/dml2_mcg_factory.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c index 1b9579a32ff2..1b9579a32ff2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h index f00bd9e72a86..f00bd9e72a86 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn3.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn3.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c index d88b3e0082dd..c26e100fcaf2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c @@ -642,6 +642,11 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) int i = 0; struct dml2_pmo_instance *pmo = in_out->instance; + unsigned int base_list_size = 0; + const struct dml2_pmo_pstate_strategy *base_list = NULL; + unsigned int *expanded_list_size = NULL; + struct dml2_pmo_pstate_strategy *expanded_list = NULL; + pmo->soc_bb = in_out->soc_bb; pmo->ip_caps = in_out->ip_caps; pmo->mpc_combine_limit = 2; @@ -656,53 +661,71 @@ bool pmo_dcn4_fams2_initialize(struct dml2_pmo_initialize_in_out *in_out) pmo->options = in_out->options; /* generate permutations of p-state configs from base strategy list */ - for (i = 1; i <= PMO_DCN4_MAX_DISPLAYS; i++) { - switch (i) { + for (i = 0; i < PMO_DCN4_MAX_DISPLAYS; i++) { + switch (i+1) { case 1: - DML_ASSERT(base_strategy_list_1_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); - - /* populate list */ - pmo_dcn4_fams2_expand_base_pstate_strategies( - base_strategy_list_1_display, - base_strategy_list_1_display_size, - i, - pmo->init_data.pmo_dcn4.expanded_strategy_list_1_display, - &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_1_display; + base_list_size = base_strategy_list_1_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_1_display; + break; case 2: - DML_ASSERT(base_strategy_list_2_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); - - /* populate list */ - pmo_dcn4_fams2_expand_base_pstate_strategies( - base_strategy_list_2_display, - base_strategy_list_2_display_size, - i, - pmo->init_data.pmo_dcn4.expanded_strategy_list_2_display, - &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_2_display; + base_list_size = base_strategy_list_2_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_2_display; + break; case 3: - DML_ASSERT(base_strategy_list_3_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); - - /* populate list */ - pmo_dcn4_fams2_expand_base_pstate_strategies( - base_strategy_list_3_display, - base_strategy_list_3_display_size, - i, - pmo->init_data.pmo_dcn4.expanded_strategy_list_3_display, - &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_3_display; + base_list_size = base_strategy_list_3_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_3_display; + break; case 4: - DML_ASSERT(base_strategy_list_4_display_size <= PMO_DCN4_MAX_BASE_STRATEGIES); - - /* populate list */ - pmo_dcn4_fams2_expand_base_pstate_strategies( - base_strategy_list_4_display, - base_strategy_list_4_display_size, - i, - pmo->init_data.pmo_dcn4.expanded_strategy_list_4_display, - &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i - 1]); + if (pmo->options->override_strategy_lists[i] && pmo->options->num_override_strategies_per_list[i]) { + base_list = pmo->options->override_strategy_lists[i]; + base_list_size = pmo->options->num_override_strategies_per_list[i]; + } else { + base_list = base_strategy_list_4_display; + base_list_size = base_strategy_list_4_display_size; + } + + expanded_list_size = &pmo->init_data.pmo_dcn4.num_expanded_strategies_per_list[i]; + expanded_list = pmo->init_data.pmo_dcn4.expanded_strategy_list_4_display; + break; } + + DML_ASSERT(base_list_size <= PMO_DCN4_MAX_BASE_STRATEGIES); + + /* populate list */ + pmo_dcn4_fams2_expand_base_pstate_strategies( + base_list, + base_list_size, + i + 1, + expanded_list, + expanded_list_size); } return true; @@ -1026,13 +1049,13 @@ static bool all_timings_support_vblank(const struct dml2_pmo_instance *pmo, return synchronizable; } -static unsigned int calc_svp_microschedule(const struct dml2_fams2_meta *fams2_meta) +static unsigned int calc_svp_microschedule(const struct dml2_pstate_meta *pstate_meta) { - return fams2_meta->contention_delay_otg_vlines + - fams2_meta->method_subvp.programming_delay_otg_vlines + - fams2_meta->method_subvp.phantom_vtotal + - fams2_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + - fams2_meta->dram_clk_change_blackout_otg_vlines; + return pstate_meta->contention_delay_otg_vlines + + pstate_meta->method_subvp.programming_delay_otg_vlines + + pstate_meta->method_subvp.phantom_vtotal + + pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + + pstate_meta->blackout_otg_vlines; } static bool all_timings_support_drr(const struct dml2_pmo_instance *pmo, @@ -1042,29 +1065,29 @@ static bool all_timings_support_drr(const struct dml2_pmo_instance *pmo, unsigned int i; for (i = 0; i < DML2_MAX_PLANES; i++) { const struct dml2_stream_parameters *stream_descriptor; - const struct dml2_fams2_meta *stream_fams2_meta; + const struct dml2_pstate_meta *stream_pstate_meta; if (is_bit_set_in_bitfield(mask, i)) { stream_descriptor = &display_config->display_config.stream_descriptors[i]; - stream_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[i]; + stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; if (!stream_descriptor->timing.drr_config.enabled) return false; /* cannot support required vtotal */ - if (stream_fams2_meta->method_drr.stretched_vtotal > stream_fams2_meta->max_vtotal) { + if (stream_pstate_meta->method_drr.stretched_vtotal > stream_pstate_meta->max_vtotal) { return false; } /* check rr is within bounds */ - if (stream_fams2_meta->nom_refresh_rate_hz < pmo->fams_params.v2.drr.refresh_rate_limit_min || - stream_fams2_meta->nom_refresh_rate_hz > pmo->fams_params.v2.drr.refresh_rate_limit_max) { + if (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.drr.refresh_rate_limit_min || + stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.drr.refresh_rate_limit_max) { return false; } /* check required stretch is allowed */ if (stream_descriptor->timing.drr_config.max_instant_vtotal_delta > 0 && - stream_fams2_meta->method_drr.stretched_vtotal - stream_fams2_meta->nom_vtotal > stream_descriptor->timing.drr_config.max_instant_vtotal_delta) { + stream_pstate_meta->method_drr.stretched_vtotal - stream_pstate_meta->nom_vtotal > (int)stream_descriptor->timing.drr_config.max_instant_vtotal_delta) { return false; } } @@ -1079,7 +1102,7 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, { const struct dml2_stream_parameters *stream_descriptor; const struct dml2_plane_parameters *plane_descriptor; - const struct dml2_fams2_meta *stream_fams2_meta; + const struct dml2_pstate_meta *stream_pstate_meta; unsigned int microschedule_vlines; unsigned int i; unsigned int mcaches_per_plane; @@ -1124,13 +1147,13 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, for (i = 0; i < DML2_MAX_PLANES; i++) { if (is_bit_set_in_bitfield(mask, i)) { stream_descriptor = &display_config->display_config.stream_descriptors[i]; - stream_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[i]; + stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[i]; if (stream_descriptor->overrides.disable_subvp) { return false; } - microschedule_vlines = calc_svp_microschedule(&pmo->scratch.pmo_dcn4.stream_fams2_meta[i]); + microschedule_vlines = calc_svp_microschedule(&pmo->scratch.pmo_dcn4.stream_pstate_meta[i]); /* block if using an interlaced timing */ if (stream_descriptor->timing.interlaced) { @@ -1141,8 +1164,8 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, * 2) refresh rate must be within the allowed bounds */ if (microschedule_vlines >= stream_descriptor->timing.v_active || - (stream_fams2_meta->nom_refresh_rate_hz < pmo->fams_params.v2.subvp.refresh_rate_limit_min || - stream_fams2_meta->nom_refresh_rate_hz > pmo->fams_params.v2.subvp.refresh_rate_limit_max)) { + (stream_pstate_meta->nom_refresh_rate_hz < pmo->fams_params.v2.subvp.refresh_rate_limit_min || + stream_pstate_meta->nom_refresh_rate_hz > pmo->fams_params.v2.subvp.refresh_rate_limit_max)) { return false; } } @@ -1232,43 +1255,43 @@ static bool all_planes_match_method(const struct display_configuation_with_meta } static void build_method_scheduling_params( - struct dml2_fams2_per_method_common_meta *stream_method_fams2_meta, - struct dml2_fams2_meta *stream_fams2_meta) + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta, + struct dml2_pstate_meta *stream_pstate_meta) { - stream_method_fams2_meta->allow_time_us = - (double)((int)stream_method_fams2_meta->allow_end_otg_vline - (int)stream_method_fams2_meta->allow_start_otg_vline) * - stream_fams2_meta->otg_vline_time_us; - if (stream_method_fams2_meta->allow_time_us >= stream_method_fams2_meta->period_us) { + stream_method_pstate_meta->allow_time_us = + (double)((int)stream_method_pstate_meta->allow_end_otg_vline - (int)stream_method_pstate_meta->allow_start_otg_vline) * + stream_pstate_meta->otg_vline_time_us; + if (stream_method_pstate_meta->allow_time_us >= stream_method_pstate_meta->period_us) { /* when allow wave overlaps an entire frame, it is always schedulable (DRR can do this)*/ - stream_method_fams2_meta->disallow_time_us = 0.0; + stream_method_pstate_meta->disallow_time_us = 0.0; } else { - stream_method_fams2_meta->disallow_time_us = - stream_method_fams2_meta->period_us - stream_method_fams2_meta->allow_time_us; + stream_method_pstate_meta->disallow_time_us = + stream_method_pstate_meta->period_us - stream_method_pstate_meta->allow_time_us; } } -static struct dml2_fams2_per_method_common_meta *get_per_method_common_meta( +static struct dml2_pstate_per_method_common_meta *get_per_method_common_meta( struct dml2_pmo_instance *pmo, enum dml2_pstate_method stream_pstate_method, int stream_idx) { - struct dml2_fams2_per_method_common_meta *stream_method_fams2_meta = NULL; + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta = NULL; switch (stream_pstate_method) { case dml2_pstate_method_vactive: case dml2_pstate_method_fw_vactive_drr: - stream_method_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_idx].method_vactive.common; + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vactive.common; break; case dml2_pstate_method_vblank: case dml2_pstate_method_fw_vblank_drr: - stream_method_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_idx].method_vblank.common; + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_vblank.common; break; case dml2_pstate_method_fw_svp: case dml2_pstate_method_fw_svp_drr: - stream_method_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_idx].method_subvp.common; + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_subvp.common; break; case dml2_pstate_method_fw_drr: - stream_method_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_idx].method_drr.common; + stream_method_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_idx].method_drr.common; break; case dml2_pstate_method_reserved_hw: case dml2_pstate_method_reserved_fw: @@ -1277,10 +1300,10 @@ static struct dml2_fams2_per_method_common_meta *get_per_method_common_meta( case dml2_pstate_method_count: case dml2_pstate_method_na: default: - stream_method_fams2_meta = NULL; + stream_method_pstate_meta = NULL; } - return stream_method_fams2_meta; + return stream_method_pstate_meta; } static bool is_timing_group_schedulable( @@ -1288,10 +1311,10 @@ static bool is_timing_group_schedulable( const struct display_configuation_with_meta *display_cfg, const struct dml2_pmo_pstate_strategy *pstate_strategy, const unsigned int timing_group_idx, - struct dml2_fams2_per_method_common_meta *group_fams2_meta) + struct dml2_pstate_per_method_common_meta *group_pstate_meta) { unsigned int i; - struct dml2_fams2_per_method_common_meta *stream_method_fams2_meta; + struct dml2_pstate_per_method_common_meta *stream_method_pstate_meta; unsigned int base_stream_idx = 0; struct dml2_pmo_scratch *s = &pmo->scratch; @@ -1305,31 +1328,31 @@ static bool is_timing_group_schedulable( } /* init allow start and end lines for timing group */ - stream_method_fams2_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[base_stream_idx], base_stream_idx); - if (!stream_method_fams2_meta) + stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[base_stream_idx], base_stream_idx); + if (!stream_method_pstate_meta) return false; - group_fams2_meta->allow_start_otg_vline = stream_method_fams2_meta->allow_start_otg_vline; - group_fams2_meta->allow_end_otg_vline = stream_method_fams2_meta->allow_end_otg_vline; - group_fams2_meta->period_us = stream_method_fams2_meta->period_us; + group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; + group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; + group_pstate_meta->period_us = stream_method_pstate_meta->period_us; for (i = base_stream_idx + 1; i < display_cfg->display_config.num_streams; i++) { if (is_bit_set_in_bitfield(pmo->scratch.pmo_dcn4.synchronized_timing_group_masks[timing_group_idx], i)) { - stream_method_fams2_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[i], i); - if (!stream_method_fams2_meta) + stream_method_pstate_meta = get_per_method_common_meta(pmo, pstate_strategy->per_stream_pstate_method[i], i); + if (!stream_method_pstate_meta) continue; - if (group_fams2_meta->allow_start_otg_vline < stream_method_fams2_meta->allow_start_otg_vline) { + if (group_pstate_meta->allow_start_otg_vline < stream_method_pstate_meta->allow_start_otg_vline) { /* set group allow start to larger otg vline */ - group_fams2_meta->allow_start_otg_vline = stream_method_fams2_meta->allow_start_otg_vline; + group_pstate_meta->allow_start_otg_vline = stream_method_pstate_meta->allow_start_otg_vline; } - if (group_fams2_meta->allow_end_otg_vline > stream_method_fams2_meta->allow_end_otg_vline) { + if (group_pstate_meta->allow_end_otg_vline > stream_method_pstate_meta->allow_end_otg_vline) { /* set group allow end to smaller otg vline */ - group_fams2_meta->allow_end_otg_vline = stream_method_fams2_meta->allow_end_otg_vline; + group_pstate_meta->allow_end_otg_vline = stream_method_pstate_meta->allow_end_otg_vline; } /* check waveform still has positive width */ - if (group_fams2_meta->allow_start_otg_vline >= group_fams2_meta->allow_end_otg_vline) { + if (group_pstate_meta->allow_start_otg_vline >= group_pstate_meta->allow_end_otg_vline) { /* timing group is not schedulable */ return false; } @@ -1337,10 +1360,10 @@ static bool is_timing_group_schedulable( } /* calculate the rest of the meta */ - build_method_scheduling_params(group_fams2_meta, &pmo->scratch.pmo_dcn4.stream_fams2_meta[base_stream_idx]); + build_method_scheduling_params(group_pstate_meta, &pmo->scratch.pmo_dcn4.stream_pstate_meta[base_stream_idx]); - return group_fams2_meta->allow_time_us > 0.0 && - group_fams2_meta->disallow_time_us < pmo->ip_caps->fams2.max_allow_delay_us; + return group_pstate_meta->allow_time_us > 0.0 && + group_pstate_meta->disallow_time_us < pmo->ip_caps->fams2.max_allow_delay_us; } static bool is_config_schedulable( @@ -1354,7 +1377,7 @@ static bool is_config_schedulable( double max_allow_delay_us = 0.0; - memset(s->pmo_dcn4.group_common_fams2_meta, 0, sizeof(s->pmo_dcn4.group_common_fams2_meta)); + memset(s->pmo_dcn4.group_common_pstate_meta, 0, sizeof(s->pmo_dcn4.group_common_pstate_meta)); memset(s->pmo_dcn4.sorted_group_gtl_disallow_index, 0, sizeof(unsigned int) * DML2_MAX_PLANES); /* search for a general solution to the schedule */ @@ -1369,12 +1392,12 @@ static bool is_config_schedulable( for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { s->pmo_dcn4.sorted_group_gtl_disallow_index[i] = i; s->pmo_dcn4.sorted_group_gtl_period_index[i] = i; - if (!is_timing_group_schedulable(pmo, display_cfg, pstate_strategy, i, &s->pmo_dcn4.group_common_fams2_meta[i])) { + if (!is_timing_group_schedulable(pmo, display_cfg, pstate_strategy, i, &s->pmo_dcn4.group_common_pstate_meta[i])) { /* synchronized timing group was not schedulable */ schedulable = false; break; } - max_allow_delay_us += s->pmo_dcn4.group_common_fams2_meta[i].disallow_time_us; + max_allow_delay_us += s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us; } if ((schedulable && s->pmo_dcn4.num_timing_groups <= 1) || !schedulable) { @@ -1391,8 +1414,8 @@ static bool is_config_schedulable( bool swapped = false; for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { - double j_disallow_us = s->pmo_dcn4.group_common_fams2_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j]].disallow_time_us; - double jp1_disallow_us = s->pmo_dcn4.group_common_fams2_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]].disallow_time_us; + double j_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j]].disallow_time_us; + double jp1_disallow_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_disallow_index[j + 1]].disallow_time_us; if (j_disallow_us < jp1_disallow_us) { /* swap as A < B */ swap(s->pmo_dcn4.sorted_group_gtl_disallow_index[j], @@ -1410,19 +1433,19 @@ static bool is_config_schedulable( * other display, or when >2 streams continue to halve the remaining allow time. */ for (i = 0; i < s->pmo_dcn4.num_timing_groups; i++) { - if (s->pmo_dcn4.group_common_fams2_meta[i].disallow_time_us <= 0.0) { + if (s->pmo_dcn4.group_common_pstate_meta[i].disallow_time_us <= 0.0) { /* this timing group always allows */ continue; } - double max_allow_time_us = s->pmo_dcn4.group_common_fams2_meta[i].allow_time_us; + double max_allow_time_us = s->pmo_dcn4.group_common_pstate_meta[i].allow_time_us; for (j = 0; j < s->pmo_dcn4.num_timing_groups; j++) { unsigned int sorted_j = s->pmo_dcn4.sorted_group_gtl_disallow_index[j]; /* stream can't overlap itself */ - if (i != sorted_j && s->pmo_dcn4.group_common_fams2_meta[sorted_j].disallow_time_us > 0.0) { + if (i != sorted_j && s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us > 0.0) { max_allow_time_us = math_min2( - s->pmo_dcn4.group_common_fams2_meta[sorted_j].allow_time_us, - (max_allow_time_us - s->pmo_dcn4.group_common_fams2_meta[sorted_j].disallow_time_us) / 2); + s->pmo_dcn4.group_common_pstate_meta[sorted_j].allow_time_us, + (max_allow_time_us - s->pmo_dcn4.group_common_pstate_meta[sorted_j].disallow_time_us) / 2); if (max_allow_time_us < 0.0) { /* failed exit early */ @@ -1450,8 +1473,8 @@ static bool is_config_schedulable( bool swapped = false; for (j = 0; j < s->pmo_dcn4.num_timing_groups - 1; j++) { - double j_period_us = s->pmo_dcn4.group_common_fams2_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j]].period_us; - double jp1_period_us = s->pmo_dcn4.group_common_fams2_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]].period_us; + double j_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j]].period_us; + double jp1_period_us = s->pmo_dcn4.group_common_pstate_meta[s->pmo_dcn4.sorted_group_gtl_period_index[j + 1]].period_us; if (j_period_us < jp1_period_us) { /* swap as A < B */ swap(s->pmo_dcn4.sorted_group_gtl_period_index[j], @@ -1470,7 +1493,7 @@ static bool is_config_schedulable( unsigned int sorted_i = s->pmo_dcn4.sorted_group_gtl_period_index[i]; unsigned int sorted_ip1 = s->pmo_dcn4.sorted_group_gtl_period_index[i + 1]; - if (s->pmo_dcn4.group_common_fams2_meta[sorted_i].allow_time_us < s->pmo_dcn4.group_common_fams2_meta[sorted_ip1].period_us || + if (s->pmo_dcn4.group_common_pstate_meta[sorted_i].allow_time_us < s->pmo_dcn4.group_common_pstate_meta[sorted_ip1].period_us || (s->pmo_dcn4.group_is_drr_enabled[sorted_ip1] && s->pmo_dcn4.group_is_drr_active[sorted_ip1])) { schedulable = false; break; @@ -1492,18 +1515,18 @@ static bool is_config_schedulable( /* default period_0 > period_1 */ unsigned int lrg_idx = 0; unsigned int sml_idx = 1; - if (s->pmo_dcn4.group_common_fams2_meta[0].period_us < s->pmo_dcn4.group_common_fams2_meta[1].period_us) { + if (s->pmo_dcn4.group_common_pstate_meta[0].period_us < s->pmo_dcn4.group_common_pstate_meta[1].period_us) { /* period_0 < period_1 */ lrg_idx = 1; sml_idx = 0; } - period_ratio = s->pmo_dcn4.group_common_fams2_meta[lrg_idx].period_us / s->pmo_dcn4.group_common_fams2_meta[sml_idx].period_us; - shift_per_period = s->pmo_dcn4.group_common_fams2_meta[sml_idx].period_us * (period_ratio - math_floor(period_ratio)); - max_shift_us = s->pmo_dcn4.group_common_fams2_meta[lrg_idx].disallow_time_us - s->pmo_dcn4.group_common_fams2_meta[sml_idx].allow_time_us; - max_allow_delay_us = max_shift_us / shift_per_period * s->pmo_dcn4.group_common_fams2_meta[lrg_idx].period_us; + period_ratio = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us / s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us; + shift_per_period = s->pmo_dcn4.group_common_pstate_meta[sml_idx].period_us * (period_ratio - math_floor(period_ratio)); + max_shift_us = s->pmo_dcn4.group_common_pstate_meta[lrg_idx].disallow_time_us - s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us; + max_allow_delay_us = max_shift_us / shift_per_period * s->pmo_dcn4.group_common_pstate_meta[lrg_idx].period_us; if (shift_per_period > 0.0 && - shift_per_period < s->pmo_dcn4.group_common_fams2_meta[lrg_idx].allow_time_us + s->pmo_dcn4.group_common_fams2_meta[sml_idx].allow_time_us && + shift_per_period < s->pmo_dcn4.group_common_pstate_meta[lrg_idx].allow_time_us + s->pmo_dcn4.group_common_pstate_meta[sml_idx].allow_time_us && max_allow_delay_us < pmo->ip_caps->fams2.max_allow_delay_us) { schedulable = true; } @@ -1646,22 +1669,22 @@ static int get_vactive_pstate_margin(const struct display_configuation_with_meta return min_vactive_margin_us; } -static unsigned int get_vactive_det_fill_latency_delay_us(const struct display_configuation_with_meta *display_cfg, int plane_mask) +static int get_vactive_det_fill_latency_delay_us(const struct display_configuation_with_meta *display_cfg, int plane_mask) { unsigned char i; - unsigned int max_vactive_fill_us = 0; + int max_vactive_fill_us = 0; for (i = 0; i < DML2_MAX_PLANES; i++) { if (is_bit_set_in_bitfield(plane_mask, i)) { - if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us > max_vactive_fill_us) - max_vactive_fill_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].dram_change_vactive_det_fill_delay_us; + if (display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].vactive_det_fill_delay_us[dml2_pstate_type_uclk] > max_vactive_fill_us) + max_vactive_fill_us = display_cfg->mode_support_result.cfg_support_info.plane_support_info[i].vactive_det_fill_delay_us[dml2_pstate_type_uclk]; } } return max_vactive_fill_us; } -static void build_fams2_meta_per_stream(struct dml2_pmo_instance *pmo, +static void build_pstate_meta_per_stream(struct dml2_pmo_instance *pmo, struct display_configuation_with_meta *display_config, int stream_index) { @@ -1669,7 +1692,7 @@ static void build_fams2_meta_per_stream(struct dml2_pmo_instance *pmo, const struct dml2_stream_parameters *stream_descriptor = &display_config->display_config.stream_descriptors[stream_index]; const struct core_stream_support_info *stream_info = &display_config->mode_support_result.cfg_support_info.stream_support_info[stream_index]; const struct dml2_timing_cfg *timing = &stream_descriptor->timing; - struct dml2_fams2_meta *stream_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_index]; + struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; /* worst case all other streams require some programming at the same time, 0 if only 1 stream */ unsigned int contention_delay_us = (ip_caps->fams2.vertical_interrupt_ack_delay_us + @@ -1677,142 +1700,142 @@ static void build_fams2_meta_per_stream(struct dml2_pmo_instance *pmo, (display_config->display_config.num_streams - 1); /* common */ - stream_fams2_meta->valid = true; - stream_fams2_meta->otg_vline_time_us = (double)timing->h_total / timing->pixel_clock_khz * 1000.0; - stream_fams2_meta->nom_vtotal = stream_descriptor->timing.vblank_nom + stream_descriptor->timing.v_active; - stream_fams2_meta->nom_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / - (stream_fams2_meta->nom_vtotal * timing->h_total); - stream_fams2_meta->nom_frame_time_us = - (double)stream_fams2_meta->nom_vtotal * stream_fams2_meta->otg_vline_time_us; - stream_fams2_meta->vblank_start = timing->v_blank_end + timing->v_active; + stream_pstate_meta->valid = true; + stream_pstate_meta->otg_vline_time_us = (double)timing->h_total / timing->pixel_clock_khz * 1000.0; + stream_pstate_meta->nom_vtotal = stream_descriptor->timing.vblank_nom + stream_descriptor->timing.v_active; + stream_pstate_meta->nom_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / + (stream_pstate_meta->nom_vtotal * timing->h_total); + stream_pstate_meta->nom_frame_time_us = + (double)stream_pstate_meta->nom_vtotal * stream_pstate_meta->otg_vline_time_us; + stream_pstate_meta->vblank_start = timing->v_blank_end + timing->v_active; if (stream_descriptor->timing.drr_config.enabled == true) { if (stream_descriptor->timing.drr_config.min_refresh_uhz != 0.0) { - stream_fams2_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / + stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / ((double)stream_descriptor->timing.drr_config.min_refresh_uhz * stream_descriptor->timing.h_total) * 1e9); } else { /* assume min of 48Hz */ - stream_fams2_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / + stream_pstate_meta->max_vtotal = (unsigned int)math_floor((double)stream_descriptor->timing.pixel_clock_khz / (48000000.0 * stream_descriptor->timing.h_total) * 1e9); } } else { - stream_fams2_meta->max_vtotal = stream_fams2_meta->nom_vtotal; - } - stream_fams2_meta->min_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / - (stream_fams2_meta->max_vtotal * timing->h_total); - stream_fams2_meta->max_frame_time_us = - (double)stream_fams2_meta->max_vtotal * stream_fams2_meta->otg_vline_time_us; - - stream_fams2_meta->scheduling_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.scheduling_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->vertical_interrupt_ack_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.vertical_interrupt_ack_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->contention_delay_otg_vlines = - (unsigned int)math_ceil(contention_delay_us / stream_fams2_meta->otg_vline_time_us); + stream_pstate_meta->max_vtotal = stream_pstate_meta->nom_vtotal; + } + stream_pstate_meta->min_refresh_rate_hz = timing->pixel_clock_khz * 1000.0 / + (stream_pstate_meta->max_vtotal * timing->h_total); + stream_pstate_meta->max_frame_time_us = + (double)stream_pstate_meta->max_vtotal * stream_pstate_meta->otg_vline_time_us; + + stream_pstate_meta->scheduling_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.scheduling_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->vertical_interrupt_ack_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.vertical_interrupt_ack_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->contention_delay_otg_vlines = + (unsigned int)math_ceil(contention_delay_us / stream_pstate_meta->otg_vline_time_us); /* worst case allow to target needs to account for all streams' allow events overlapping, and 1 line for error */ - stream_fams2_meta->allow_to_target_delay_otg_vlines = - (unsigned int)(math_ceil((ip_caps->fams2.vertical_interrupt_ack_delay_us + contention_delay_us + ip_caps->fams2.allow_programming_delay_us) / stream_fams2_meta->otg_vline_time_us)) + 1; - stream_fams2_meta->min_allow_width_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.min_allow_width_us / stream_fams2_meta->otg_vline_time_us); + stream_pstate_meta->allow_to_target_delay_otg_vlines = + (unsigned int)(math_ceil((ip_caps->fams2.vertical_interrupt_ack_delay_us + contention_delay_us + ip_caps->fams2.allow_programming_delay_us) / stream_pstate_meta->otg_vline_time_us)) + 1; + stream_pstate_meta->min_allow_width_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.min_allow_width_us / stream_pstate_meta->otg_vline_time_us); /* this value should account for urgent latency */ - stream_fams2_meta->dram_clk_change_blackout_otg_vlines = + stream_pstate_meta->blackout_otg_vlines = (unsigned int)math_ceil(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us / - stream_fams2_meta->otg_vline_time_us); + stream_pstate_meta->otg_vline_time_us); /* scheduling params should be built based on the worst case for allow_time:disallow_time */ /* vactive */ if (display_config->display_config.num_streams == 1) { /* for single stream, guarantee at least an instant of allow */ - stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = (unsigned int)math_floor( + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = (unsigned int)math_floor( math_max2(0.0, - timing->v_active - math_max2(1.0, stream_fams2_meta->min_allow_width_otg_vlines) - stream_fams2_meta->dram_clk_change_blackout_otg_vlines)); + timing->v_active - math_max2(1.0, stream_pstate_meta->min_allow_width_otg_vlines) - stream_pstate_meta->blackout_otg_vlines)); } else { /* for multi stream, bound to a max fill time defined by IP caps */ - stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = - (unsigned int)math_floor((double)ip_caps->max_vactive_det_fill_delay_us / stream_fams2_meta->otg_vline_time_us); + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines = + (unsigned int)math_floor((double)ip_caps->max_vactive_det_fill_delay_us / stream_pstate_meta->otg_vline_time_us); } - stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_us = stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines * stream_fams2_meta->otg_vline_time_us; + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us = stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines * stream_pstate_meta->otg_vline_time_us; - if (stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_us > 0.0) { - stream_fams2_meta->method_vactive.common.allow_start_otg_vline = - timing->v_blank_end + stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; - stream_fams2_meta->method_vactive.common.allow_end_otg_vline = - stream_fams2_meta->vblank_start - - stream_fams2_meta->dram_clk_change_blackout_otg_vlines; + if (stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us > 0.0) { + stream_pstate_meta->method_vactive.common.allow_start_otg_vline = + timing->v_blank_end + stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_otg_vlines; + stream_pstate_meta->method_vactive.common.allow_end_otg_vline = + stream_pstate_meta->vblank_start - + stream_pstate_meta->blackout_otg_vlines; } else { - stream_fams2_meta->method_vactive.common.allow_start_otg_vline = 0; - stream_fams2_meta->method_vactive.common.allow_end_otg_vline = 0; + stream_pstate_meta->method_vactive.common.allow_start_otg_vline = 0; + stream_pstate_meta->method_vactive.common.allow_end_otg_vline = 0; } - stream_fams2_meta->method_vactive.common.period_us = stream_fams2_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_fams2_meta->method_vactive.common, stream_fams2_meta); + stream_pstate_meta->method_vactive.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_vactive.common, stream_pstate_meta); /* vblank */ - stream_fams2_meta->method_vblank.common.allow_start_otg_vline = stream_fams2_meta->vblank_start; - stream_fams2_meta->method_vblank.common.allow_end_otg_vline = - stream_fams2_meta->method_vblank.common.allow_start_otg_vline + 1; - stream_fams2_meta->method_vblank.common.period_us = stream_fams2_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_fams2_meta->method_vblank.common, stream_fams2_meta); + stream_pstate_meta->method_vblank.common.allow_start_otg_vline = stream_pstate_meta->vblank_start; + stream_pstate_meta->method_vblank.common.allow_end_otg_vline = + stream_pstate_meta->method_vblank.common.allow_start_otg_vline + 1; + stream_pstate_meta->method_vblank.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_vblank.common, stream_pstate_meta); /* subvp */ - stream_fams2_meta->method_subvp.programming_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_programming_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->method_subvp.df_throttle_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_df_throttle_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->method_subvp.prefetch_to_mall_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.subvp_prefetch_to_mall_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->method_subvp.phantom_vactive = - stream_fams2_meta->allow_to_target_delay_otg_vlines + - stream_fams2_meta->min_allow_width_otg_vlines + + stream_pstate_meta->method_subvp.programming_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_programming_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_df_throttle_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.subvp_prefetch_to_mall_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_subvp.phantom_vactive = + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + stream_info->phantom_min_v_active; - stream_fams2_meta->method_subvp.phantom_vfp = - stream_fams2_meta->method_subvp.df_throttle_delay_otg_vlines; + stream_pstate_meta->method_subvp.phantom_vfp = + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines; /* phantom vtotal = v_bp(vstartup) + v_sync(1) + v_fp(throttle_delay) + v_active(allow_to_target + min_allow + min_vactive)*/ - stream_fams2_meta->method_subvp.phantom_vtotal = + stream_pstate_meta->method_subvp.phantom_vtotal = stream_info->phantom_v_startup + - stream_fams2_meta->method_subvp.phantom_vfp + + stream_pstate_meta->method_subvp.phantom_vfp + 1 + - stream_fams2_meta->method_subvp.df_throttle_delay_otg_vlines + - stream_fams2_meta->method_subvp.phantom_vactive; - stream_fams2_meta->method_subvp.common.allow_start_otg_vline = + stream_pstate_meta->method_subvp.df_throttle_delay_otg_vlines + + stream_pstate_meta->method_subvp.phantom_vactive; + stream_pstate_meta->method_subvp.common.allow_start_otg_vline = stream_descriptor->timing.v_blank_end + - stream_fams2_meta->contention_delay_otg_vlines + - stream_fams2_meta->method_subvp.programming_delay_otg_vlines + - stream_fams2_meta->method_subvp.phantom_vtotal + - stream_fams2_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + - stream_fams2_meta->allow_to_target_delay_otg_vlines; - stream_fams2_meta->method_subvp.common.allow_end_otg_vline = - stream_fams2_meta->vblank_start - - stream_fams2_meta->dram_clk_change_blackout_otg_vlines; - stream_fams2_meta->method_subvp.common.period_us = stream_fams2_meta->nom_frame_time_us; - build_method_scheduling_params(&stream_fams2_meta->method_subvp.common, stream_fams2_meta); + stream_pstate_meta->contention_delay_otg_vlines + + stream_pstate_meta->method_subvp.programming_delay_otg_vlines + + stream_pstate_meta->method_subvp.phantom_vtotal + + stream_pstate_meta->method_subvp.prefetch_to_mall_delay_otg_vlines + + stream_pstate_meta->allow_to_target_delay_otg_vlines; + stream_pstate_meta->method_subvp.common.allow_end_otg_vline = + stream_pstate_meta->vblank_start - + stream_pstate_meta->blackout_otg_vlines; + stream_pstate_meta->method_subvp.common.period_us = stream_pstate_meta->nom_frame_time_us; + build_method_scheduling_params(&stream_pstate_meta->method_subvp.common, stream_pstate_meta); /* drr */ - stream_fams2_meta->method_drr.programming_delay_otg_vlines = - (unsigned int)math_ceil(ip_caps->fams2.drr_programming_delay_us / stream_fams2_meta->otg_vline_time_us); - stream_fams2_meta->method_drr.common.allow_start_otg_vline = - stream_fams2_meta->vblank_start + - stream_fams2_meta->allow_to_target_delay_otg_vlines; - stream_fams2_meta->method_drr.common.period_us = stream_fams2_meta->nom_frame_time_us; + stream_pstate_meta->method_drr.programming_delay_otg_vlines = + (unsigned int)math_ceil(ip_caps->fams2.drr_programming_delay_us / stream_pstate_meta->otg_vline_time_us); + stream_pstate_meta->method_drr.common.allow_start_otg_vline = + stream_pstate_meta->vblank_start + + stream_pstate_meta->allow_to_target_delay_otg_vlines; + stream_pstate_meta->method_drr.common.period_us = stream_pstate_meta->nom_frame_time_us; if (display_config->display_config.num_streams <= 1) { /* only need to stretch vblank for blackout time */ - stream_fams2_meta->method_drr.stretched_vtotal = - stream_fams2_meta->nom_vtotal + - stream_fams2_meta->allow_to_target_delay_otg_vlines + - stream_fams2_meta->min_allow_width_otg_vlines + - stream_fams2_meta->dram_clk_change_blackout_otg_vlines; + stream_pstate_meta->method_drr.stretched_vtotal = + stream_pstate_meta->nom_vtotal + + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + + stream_pstate_meta->blackout_otg_vlines; } else { /* multi display needs to always be schedulable */ - stream_fams2_meta->method_drr.stretched_vtotal = - stream_fams2_meta->nom_vtotal * 2 + - stream_fams2_meta->allow_to_target_delay_otg_vlines + - stream_fams2_meta->min_allow_width_otg_vlines + - stream_fams2_meta->dram_clk_change_blackout_otg_vlines; - } - stream_fams2_meta->method_drr.common.allow_end_otg_vline = - stream_fams2_meta->method_drr.stretched_vtotal - - stream_fams2_meta->dram_clk_change_blackout_otg_vlines; - build_method_scheduling_params(&stream_fams2_meta->method_drr.common, stream_fams2_meta); + stream_pstate_meta->method_drr.stretched_vtotal = + stream_pstate_meta->nom_vtotal * 2 + + stream_pstate_meta->allow_to_target_delay_otg_vlines + + stream_pstate_meta->min_allow_width_otg_vlines + + stream_pstate_meta->blackout_otg_vlines; + } + stream_pstate_meta->method_drr.common.allow_end_otg_vline = + stream_pstate_meta->method_drr.stretched_vtotal - + stream_pstate_meta->blackout_otg_vlines; + build_method_scheduling_params(&stream_pstate_meta->method_drr.common, stream_pstate_meta); } static void build_subvp_meta_per_stream(struct dml2_pmo_instance *pmo, @@ -1820,14 +1843,14 @@ static void build_subvp_meta_per_stream(struct dml2_pmo_instance *pmo, int stream_index) { struct dml2_implicit_svp_meta *stream_svp_meta = &pmo->scratch.pmo_dcn4.stream_svp_meta[stream_index]; - struct dml2_fams2_meta *stream_fams2_meta = &pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_index]; + struct dml2_pstate_meta *stream_pstate_meta = &pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index]; stream_svp_meta->valid = true; /* PMO FAMS2 precaulcates these values */ - stream_svp_meta->v_active = stream_fams2_meta->method_subvp.phantom_vactive; - stream_svp_meta->v_front_porch = stream_fams2_meta->method_subvp.phantom_vfp; - stream_svp_meta->v_total = stream_fams2_meta->method_subvp.phantom_vtotal; + stream_svp_meta->v_active = stream_pstate_meta->method_subvp.phantom_vactive; + stream_svp_meta->v_front_porch = stream_pstate_meta->method_subvp.phantom_vfp; + stream_svp_meta->v_total = stream_pstate_meta->method_subvp.phantom_vtotal; } bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_support_in_out *in_out) @@ -1879,7 +1902,7 @@ bool pmo_dcn4_fams2_init_for_pstate_support(struct dml2_pmo_init_for_pstate_supp set_bit_in_bitfield(&s->pmo_dcn4.stream_vactive_capability_mask, stream_index); /* FAMS2 meta */ - build_fams2_meta_per_stream(pmo, display_config, stream_index); + build_pstate_meta_per_stream(pmo, display_config, stream_index); /* SVP meta */ build_subvp_meta_per_stream(pmo, display_config, stream_index); @@ -1939,9 +1962,6 @@ static void reset_display_configuration(struct display_configuation_with_meta *d for (stream_index = 0; stream_index < display_config->display_config.num_streams; stream_index++) { display_config->stage3.stream_svp_meta[stream_index].valid = false; - - display_config->display_config.stream_descriptors[stream_index].overrides.minimize_active_latency_hiding = false; - display_config->display_config.overrides.best_effort_min_active_latency_hiding_us = 0; } for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { @@ -1974,7 +1994,6 @@ static void setup_planes_for_drr_by_mask(struct display_configuation_with_meta * plane->overrides.uclk_pstate_change_strategy = dml2_uclk_pstate_change_strategy_force_drr; display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_drr; - } } } @@ -2040,7 +2059,6 @@ static void setup_planes_for_vblank_by_mask(struct display_configuation_with_met plane->overrides.reserved_vblank_time_ns); display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vblank; - } } } @@ -2055,6 +2073,7 @@ static void setup_planes_for_vblank_drr_by_mask(struct display_configuation_with for (plane_index = 0; plane_index < display_config->display_config.num_planes; plane_index++) { if (is_bit_set_in_bitfield(plane_mask, plane_index)) { plane = &display_config->display_config.plane_descriptors[plane_index]; + plane->overrides.reserved_vblank_time_ns = (long)(pmo->soc_bb->power_management_parameters.dram_clk_change_blackout_us * 1000); display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vblank_drr; @@ -2076,8 +2095,8 @@ static void setup_planes_for_vactive_by_mask(struct display_configuation_with_me display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_vactive; if (!pmo->options->disable_vactive_det_fill_bw_pad) { - display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = - (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); + display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us[dml2_pstate_type_uclk] = + (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); } } } @@ -2097,8 +2116,8 @@ static void setup_planes_for_vactive_drr_by_mask(struct display_configuation_wit display_config->stage3.pstate_switch_modes[plane_index] = dml2_pstate_method_fw_vactive_drr; if (!pmo->options->disable_vactive_det_fill_bw_pad) { - display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us = - (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_fams2_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); + display_config->display_config.plane_descriptors[plane_index].overrides.max_vactive_det_fill_delay_us[dml2_pstate_type_uclk] = + (unsigned int)math_floor(pmo->scratch.pmo_dcn4.stream_pstate_meta[stream_index].method_vactive.max_vactive_det_fill_delay_us); } } } @@ -2144,9 +2163,9 @@ static bool setup_display_config(struct display_configuation_with_meta *display_ /* copy FAMS2 meta */ if (success) { display_config->stage3.fams2_required = fams2_required; - memcpy(&display_config->stage3.stream_fams2_meta, - &scratch->pmo_dcn4.stream_fams2_meta, - sizeof(struct dml2_fams2_meta) * DML2_MAX_PLANES); + memcpy(&display_config->stage3.stream_pstate_meta, + &scratch->pmo_dcn4.stream_pstate_meta, + sizeof(struct dml2_pstate_meta) * DML2_MAX_PLANES); } return success; @@ -2188,12 +2207,12 @@ bool pmo_dcn4_fams2_test_for_pstate_support(struct dml2_pmo_test_for_pstate_supp return false; for (stream_index = 0; stream_index < in_out->base_display_config->display_config.num_streams; stream_index++) { - struct dml2_fams2_meta *stream_fams2_meta = &s->pmo_dcn4.stream_fams2_meta[stream_index]; + struct dml2_pstate_meta *stream_pstate_meta = &s->pmo_dcn4.stream_pstate_meta[stream_index]; if (s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_vactive || s->pmo_dcn4.pstate_strategy_candidates[s->pmo_dcn4.cur_pstate_candidate].per_stream_pstate_method[stream_index] == dml2_pstate_method_fw_vactive_drr) { if (get_vactive_pstate_margin(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) < (MIN_VACTIVE_MARGIN_PCT * in_out->instance->soc_bb->power_management_parameters.dram_clk_change_blackout_us) || - get_vactive_det_fill_latency_delay_us(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) > stream_fams2_meta->method_vactive.max_vactive_det_fill_delay_us) { + get_vactive_det_fill_latency_delay_us(in_out->base_display_config, s->pmo_dcn4.stream_plane_mask[stream_index]) > stream_pstate_meta->method_vactive.max_vactive_det_fill_delay_us) { p_state_supported = false; break; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h index 6baab7ad6ecc..6baab7ad6ecc 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c index 55d2464365d0..55d2464365d0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h index 7218de1824cc..b90f6263cd85 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_factory.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_pmo/dml2_pmo_factory.h @@ -10,4 +10,4 @@ bool dml2_pmo_create(enum dml2_project_id project_id, struct dml2_pmo_instance *out); -#endif +#endif
\ No newline at end of file diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c index e17b5ceba447..e17b5ceba447 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h index e13b0c5939b0..e13b0c5939b0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_standalone_libraries/lib_float_math.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_standalone_libraries/lib_float_math.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c index 5a33e2f357f4..5a33e2f357f4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_interfaces.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_interfaces.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c index 5e14d85821e2..5e14d85821e2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h index 14d0ae03dce6..14d0ae03dce6 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_legacy.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_legacy.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c index 4a7c4c62111e..4a7c4c62111e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h index 53bd8602f9ef..53bd8602f9ef 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/dml2_top/dml2_top_soc15.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h index 611c80f4f1bf..611c80f4f1bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_debug.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_debug.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h index d52aa82283b3..1a6c0727cd2a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/inc/dml2_internal_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/src/inc/dml2_internal_shared_types.h @@ -152,7 +152,7 @@ struct core_plane_support_info { int active_latency_hiding_us; int mall_svp_size_requirement_ways; int nominal_vblank_pstate_latency_hiding_us; - unsigned int dram_change_vactive_det_fill_delay_us; + int vactive_det_fill_delay_us[dml2_pstate_type_count]; }; struct core_stream_support_info { @@ -209,6 +209,7 @@ struct dml2_core_mode_support_result { unsigned int uclk_pstate_supported; unsigned int fclk_pstate_supported; + struct dml2_core_internal_watermarks watermarks; } global; struct { @@ -255,56 +256,70 @@ struct dml2_implicit_svp_meta { unsigned long v_front_porch; }; -struct dml2_fams2_per_method_common_meta { +struct dml2_pstate_per_method_common_meta { /* generic params */ - unsigned int allow_start_otg_vline; - unsigned int allow_end_otg_vline; + int allow_start_otg_vline; + int allow_end_otg_vline; /* scheduling params */ double allow_time_us; double disallow_time_us; double period_us; }; -struct dml2_fams2_meta { +struct dml2_pstate_meta { bool valid; double otg_vline_time_us; - unsigned int scheduling_delay_otg_vlines; - unsigned int vertical_interrupt_ack_delay_otg_vlines; - unsigned int allow_to_target_delay_otg_vlines; - unsigned int contention_delay_otg_vlines; - unsigned int min_allow_width_otg_vlines; - unsigned int nom_vtotal; - unsigned int vblank_start; + int scheduling_delay_otg_vlines; + int vertical_interrupt_ack_delay_otg_vlines; + int allow_to_target_delay_otg_vlines; + int contention_delay_otg_vlines; + int min_allow_width_otg_vlines; + int nom_vtotal; + int vblank_start; double nom_refresh_rate_hz; double nom_frame_time_us; - unsigned int max_vtotal; + int max_vtotal; double min_refresh_rate_hz; double max_frame_time_us; - unsigned int dram_clk_change_blackout_otg_vlines; + int blackout_otg_vlines; + int max_allow_delay_otg_vlines; + double nom_vblank_time_us; struct { double max_vactive_det_fill_delay_us; - unsigned int max_vactive_det_fill_delay_otg_vlines; - struct dml2_fams2_per_method_common_meta common; + double vactive_latency_hiding_us; + double reserved_vblank_required_us; + int max_vactive_det_fill_delay_otg_vlines; + int reserved_blank_required_vlines; + struct dml2_pstate_per_method_common_meta common; } method_vactive; struct { - struct dml2_fams2_per_method_common_meta common; + struct dml2_pstate_per_method_common_meta common; } method_vblank; struct { - unsigned int programming_delay_otg_vlines; - unsigned int df_throttle_delay_otg_vlines; - unsigned int prefetch_to_mall_delay_otg_vlines; + int programming_delay_otg_vlines; + int df_throttle_delay_otg_vlines; + int prefetch_to_mall_delay_otg_vlines; unsigned long phantom_vactive; unsigned long phantom_vfp; unsigned long phantom_vtotal; - struct dml2_fams2_per_method_common_meta common; + struct dml2_pstate_per_method_common_meta common; } method_subvp; struct { - unsigned int programming_delay_otg_vlines; - unsigned int stretched_vtotal; - struct dml2_fams2_per_method_common_meta common; + int programming_delay_otg_vlines; + int stretched_vtotal; + struct dml2_pstate_per_method_common_meta common; } method_drr; }; +/* mask of synchronized timings by stream index */ +struct dml2_pmo_synchronized_timing_groups { + unsigned int num_timing_groups; + unsigned int synchronized_timing_group_masks[DML2_MAX_PLANES]; + bool group_is_drr_enabled[DML2_MAX_PLANES]; + bool group_is_drr_active[DML2_MAX_PLANES]; + double group_line_time_us[DML2_MAX_PLANES]; +}; + struct dml2_optimization_stage3_state { bool performed; bool success; @@ -319,7 +334,7 @@ struct dml2_optimization_stage3_state { // Meta-data for FAMS2 bool fams2_required; - struct dml2_fams2_meta stream_fams2_meta[DML2_MAX_PLANES]; + struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; int min_clk_index_for_latency; }; @@ -472,6 +487,7 @@ struct dml2_core_scratch { }; struct dml2_core_instance { + enum dml2_project_id project_id; struct dml2_mcg_min_clock_table *minimum_clock_table; struct dml2_core_internal_state_inputs inputs; struct dml2_core_internal_state_intermediates intermediates; @@ -619,6 +635,12 @@ struct dml2_pmo_optimize_for_stutter_in_out { #define PMO_DCN4_MAX_NUM_VARIANTS 2 #define PMO_DCN4_MAX_BASE_STRATEGIES 10 +struct dml2_scheduling_check_locals { + struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; + unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; +}; + struct dml2_pmo_scratch { union { struct { @@ -648,7 +670,7 @@ struct dml2_pmo_scratch { // Stores all the implicit SVP meta information indexed by stream index of the display // configuration under inspection, built at optimization stage init struct dml2_implicit_svp_meta stream_svp_meta[DML2_MAX_PLANES]; - struct dml2_fams2_meta stream_fams2_meta[DML2_MAX_PLANES]; + struct dml2_pstate_meta stream_pstate_meta[DML2_MAX_PLANES]; unsigned int optimal_vblank_reserved_time_for_stutter_us[DML2_PMO_STUTTER_CANDIDATE_LIST_SIZE]; unsigned int num_stutter_candidates; @@ -663,7 +685,7 @@ struct dml2_pmo_scratch { double group_line_time_us[DML2_MAX_PLANES]; /* scheduling check locals */ - struct dml2_fams2_per_method_common_meta group_common_fams2_meta[DML2_MAX_PLANES]; + struct dml2_pstate_per_method_common_meta group_common_pstate_meta[DML2_MAX_PLANES]; unsigned int sorted_group_gtl_disallow_index[DML2_MAX_PLANES]; unsigned int sorted_group_gtl_period_index[DML2_MAX_PLANES]; double group_phase_offset[DML2_MAX_PLANES]; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c index 4cfe64aa8492..4cfe64aa8492 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h index 1538b708d8be..1538b708d8be 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_resource_mgmt.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_resource_mgmt.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h index 7ca7f2a743c2..7ca7f2a743c2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_dc_types.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h index 140ec01545db..55b3e3ca54f7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_internal_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_internal_types.h @@ -23,7 +23,7 @@ * Authors: AMD * */ - + #ifndef __DML2_INTERNAL_TYPES_H__ #define __DML2_INTERNAL_TYPES_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c index c59f825cfae9..66040c877d68 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.c @@ -24,6 +24,7 @@ * */ + #include "dml2_dc_types.h" #include "dml2_internal_types.h" #include "dml2_utils.h" diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h index 9d64851f54e7..9d64851f54e7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_mall_phantom.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_mall_phantom.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c index ef693f608d59..ef693f608d59 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h index e83e05248592..e83e05248592 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_policy.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c index 3b866e876bf4..d834cb595afa 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.c @@ -301,6 +301,7 @@ void dml2_init_socbb_params(struct dml2_context *dml2, const struct dc *in_dc, s out->pct_ideal_dram_bw_after_urgent_pixel_only = 65.0; break; + case dml_project_dcn401: out->pct_ideal_fabric_bw_after_urgent = 76; //67; out->max_avg_sdp_bw_use_normal_percent = 75; //80; @@ -424,6 +425,8 @@ void dml2_init_soc_states(struct dml2_context *dml2, const struct dc *in_dc, p->in_states->state_array[1].dcfclk_mhz = 1434.0; p->in_states->state_array[1].dram_speed_mts = 1000 * transactions_per_mem_clock; break; + + case dml_project_dcn401: p->in_states->num_states = 2; transactions_per_mem_clock = 16; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h index d764773938f4..d764773938f4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_translation_helper.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c index 9a33158b63bf..9a33158b63bf 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h index 04fcfe637119..04fcfe637119 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_utils.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_utils.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c index 9deb03a18ccc..9deb03a18ccc 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h index c384e141cebc..c384e141cebc 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml2_wrapper.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h index 17f0972b1af7..17f0972b1af7 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_assert.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_assert.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h index f7d30b47beff..d459f93cf40b 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_depedencies.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_depedencies.h @@ -31,3 +31,4 @@ */ #include "os_types.h" #include "cmntypes.h" + diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c index 00d22e542469..00d22e542469 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.c +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.c diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h index bf491cf0582d..bf491cf0582d 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_display_rq_dlg_calc.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_display_rq_dlg_calc.h diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h index 2a2f84e07ca8..7fadbe6d7af4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml_logging.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml_logging.h @@ -23,6 +23,7 @@ * Authors: AMD * */ + #ifndef __DML_LOGGING_H__ #define __DML_LOGGING_H__ diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.c index 01480a04f85e..ce91e5d28956 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.c @@ -199,6 +199,8 @@ void dpp_reset(struct dpp *dpp_base) memset(&dpp->scl_data, 0, sizeof(dpp->scl_data)); memset(&dpp->pwl_data, 0, sizeof(dpp->pwl_data)); + + dpp_base->cursor_offload = false; } @@ -484,10 +486,12 @@ void dpp1_set_cursor_position( cur_en = 0; /* not visible beyond top edge*/ if (dpp_base->pos.cur0_ctl.bits.cur0_enable != cur_en) { - REG_UPDATE(CURSOR0_CONTROL, CUR0_ENABLE, cur_en); - - dpp_base->pos.cur0_ctl.bits.cur0_enable = cur_en; + if (!dpp_base->cursor_offload) + REG_UPDATE(CURSOR0_CONTROL, CUR0_ENABLE, cur_en); } + + dpp_base->pos.cur0_ctl.bits.cur0_enable = cur_en; + dpp_base->att.cur0_ctl.bits.cur0_enable = cur_en; } void dpp1_cnv_set_optional_cursor_attributes( @@ -497,8 +501,13 @@ void dpp1_cnv_set_optional_cursor_attributes( struct dcn10_dpp *dpp = TO_DCN10_DPP(dpp_base); if (attr) { - REG_UPDATE(CURSOR0_FP_SCALE_BIAS, CUR0_FP_BIAS, attr->bias); - REG_UPDATE(CURSOR0_FP_SCALE_BIAS, CUR0_FP_SCALE, attr->scale); + if (!dpp_base->cursor_offload) { + REG_UPDATE(CURSOR0_FP_SCALE_BIAS, CUR0_FP_BIAS, attr->bias); + REG_UPDATE(CURSOR0_FP_SCALE_BIAS, CUR0_FP_SCALE, attr->scale); + } + + dpp_base->att.fp_scale_bias.bits.fp_bias = attr->bias; + dpp_base->att.fp_scale_bias.bits.fp_scale = attr->scale; } } diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h index f466182963f7..b12f34345a58 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn10/dcn10_dpp.h @@ -1348,7 +1348,8 @@ struct dcn_dpp_mask { uint32_t CURSOR0_COLOR1; \ uint32_t DPP_CONTROL; \ uint32_t CM_HDR_MULT_COEF; \ - uint32_t CURSOR0_FP_SCALE_BIAS; + uint32_t CURSOR0_FP_SCALE_BIAS; \ + uint32_t OBUF_CONTROL; struct dcn_dpp_registers { DPP_COMMON_REG_VARIABLE_LIST @@ -1450,7 +1451,6 @@ void dpp1_set_degamma( void dpp1_set_degamma_pwl(struct dpp *dpp_base, const struct pwl_params *params); - void dpp_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s); diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c index 09be2a90cc79..ef4a16117181 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c @@ -84,6 +84,22 @@ void dpp30_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s) } } +void dpp30_read_reg_state(struct dpp *dpp_base, struct dcn_dpp_reg_state *dpp_reg_state) +{ + struct dcn3_dpp *dpp = TO_DCN30_DPP(dpp_base); + + dpp_reg_state->recout_start = REG_READ(RECOUT_START); + dpp_reg_state->recout_size = REG_READ(RECOUT_SIZE); + dpp_reg_state->scl_horz_filter_scale_ratio = REG_READ(SCL_HORZ_FILTER_SCALE_RATIO); + dpp_reg_state->scl_vert_filter_scale_ratio = REG_READ(SCL_VERT_FILTER_SCALE_RATIO); + dpp_reg_state->scl_mode = REG_READ(SCL_MODE); + dpp_reg_state->cm_control = REG_READ(CM_CONTROL); + dpp_reg_state->dpp_control = REG_READ(DPP_CONTROL); + dpp_reg_state->dscl_control = REG_READ(DSCL_CONTROL); + dpp_reg_state->obuf_control = REG_READ(OBUF_CONTROL); + dpp_reg_state->mpc_size = REG_READ(MPC_SIZE); +} + /*program post scaler scs block in dpp CM*/ void dpp3_program_post_csc( struct dpp *dpp_base, @@ -396,17 +412,21 @@ void dpp3_set_cursor_attributes( } } - REG_UPDATE_3(CURSOR0_CONTROL, - CUR0_MODE, color_format, - CUR0_EXPANSION_MODE, 0, - CUR0_ROM_EN, cur_rom_en); + if (!dpp_base->cursor_offload) + REG_UPDATE_3(CURSOR0_CONTROL, + CUR0_MODE, color_format, + CUR0_EXPANSION_MODE, 0, + CUR0_ROM_EN, cur_rom_en); if (color_format == CURSOR_MODE_MONO) { /* todo: clarify what to program these to */ - REG_UPDATE(CURSOR0_COLOR0, - CUR0_COLOR0, 0x00000000); - REG_UPDATE(CURSOR0_COLOR1, - CUR0_COLOR1, 0xFFFFFFFF); + + if (!dpp_base->cursor_offload) { + REG_UPDATE(CURSOR0_COLOR0, + CUR0_COLOR0, 0x00000000); + REG_UPDATE(CURSOR0_COLOR1, + CUR0_COLOR1, 0xFFFFFFFF); + } } dpp_base->att.cur0_ctl.bits.expansion_mode = 0; @@ -578,9 +598,6 @@ static void dpp3_power_on_blnd_lut( dpp_base->ctx->dc->optimized_required = true; dpp_base->deferred_reg_writes.bits.disable_blnd_lut = true; } - } else { - REG_SET(CM_MEM_PWR_CTRL, 0, - BLNDGAM_MEM_PWR_FORCE, power_on == true ? 0 : 1); } } diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h index f236824126e9..d4a70b4379ea 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.h @@ -594,6 +594,8 @@ void dpp3_program_CM_dealpha( void dpp30_read_state(struct dpp *dpp_base, struct dcn_dpp_state *s); +void dpp30_read_reg_state(struct dpp *dpp_base, struct dcn_dpp_reg_state *dpp_reg_state); + bool dpp3_get_optimal_number_of_taps( struct dpp *dpp, struct scaler_data *scl_data, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c index fa67e54bf94e..8a5aa5e86850 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn32/dcn32_dpp.c @@ -134,6 +134,7 @@ static struct dpp_funcs dcn32_dpp_funcs = { .dpp_dppclk_control = dpp1_dppclk_control, .dpp_set_hdr_multiplier = dpp3_set_hdr_multiplier, .dpp_get_gamut_remap = dpp3_cm_get_gamut_remap, + .dpp_read_reg_state = dpp30_read_reg_state, }; diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c index f7a373a3d70a..977d83bf7741 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn35/dcn35_dpp.c @@ -95,6 +95,7 @@ void dpp35_program_bias_and_scale_fcnv( static struct dpp_funcs dcn35_dpp_funcs = { .dpp_program_gamcor_lut = dpp3_program_gamcor_lut, .dpp_read_state = dpp30_read_state, + .dpp_read_reg_state = dpp30_read_reg_state, .dpp_reset = dpp_reset, .dpp_set_scaler = dpp1_dscl_set_scaler_manual_scale, .dpp_get_optimal_number_of_taps = dpp3_get_optimal_number_of_taps, diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c index 36187f890d5d..96c2c853de42 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp.c @@ -248,6 +248,7 @@ static struct dpp_funcs dcn401_dpp_funcs = { .set_optional_cursor_attributes = dpp401_set_optional_cursor_attributes, .dpp_dppclk_control = dpp1_dppclk_control, .dpp_set_hdr_multiplier = dpp3_set_hdr_multiplier, + .dpp_read_reg_state = dpp30_read_reg_state, .set_cursor_matrix = dpp401_set_cursor_matrix, }; diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c index 7aab77b58869..62bf7cea21d8 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c @@ -103,17 +103,21 @@ void dpp401_set_cursor_attributes( } } - REG_UPDATE_3(CURSOR0_CONTROL, - CUR0_MODE, color_format, - CUR0_EXPANSION_MODE, 0, - CUR0_ROM_EN, cur_rom_en); + if (!dpp_base->cursor_offload) + REG_UPDATE_3(CURSOR0_CONTROL, + CUR0_MODE, color_format, + CUR0_EXPANSION_MODE, 0, + CUR0_ROM_EN, cur_rom_en); if (color_format == CURSOR_MODE_MONO) { /* todo: clarify what to program these to */ - REG_UPDATE(CURSOR0_COLOR0, - CUR0_COLOR0, 0x00000000); - REG_UPDATE(CURSOR0_COLOR1, - CUR0_COLOR1, 0xFFFFFFFF); + + if (!dpp_base->cursor_offload) { + REG_UPDATE(CURSOR0_COLOR0, + CUR0_COLOR0, 0x00000000); + REG_UPDATE(CURSOR0_COLOR1, + CUR0_COLOR1, 0xFFFFFFFF); + } } dpp_base->att.cur0_ctl.bits.expansion_mode = 0; @@ -132,10 +136,12 @@ void dpp401_set_cursor_position( uint32_t cur_en = pos->enable ? 1 : 0; if (dpp_base->pos.cur0_ctl.bits.cur0_enable != cur_en) { - REG_UPDATE(CURSOR0_CONTROL, CUR0_ENABLE, cur_en); - - dpp_base->pos.cur0_ctl.bits.cur0_enable = cur_en; + if (!dpp_base->cursor_offload) + REG_UPDATE(CURSOR0_CONTROL, CUR0_ENABLE, cur_en); } + + dpp_base->pos.cur0_ctl.bits.cur0_enable = cur_en; + dpp_base->att.cur0_ctl.bits.cur0_enable = cur_en; } void dpp401_set_optional_cursor_attributes( @@ -145,10 +151,17 @@ void dpp401_set_optional_cursor_attributes( struct dcn401_dpp *dpp = TO_DCN401_DPP(dpp_base); if (attr) { - REG_UPDATE(CURSOR0_FP_SCALE_BIAS_G_Y, CUR0_FP_BIAS_G_Y, attr->bias); - REG_UPDATE(CURSOR0_FP_SCALE_BIAS_G_Y, CUR0_FP_SCALE_G_Y, attr->scale); - REG_UPDATE(CURSOR0_FP_SCALE_BIAS_RB_CRCB, CUR0_FP_BIAS_RB_CRCB, attr->bias); - REG_UPDATE(CURSOR0_FP_SCALE_BIAS_RB_CRCB, CUR0_FP_SCALE_RB_CRCB, attr->scale); + if (!dpp_base->cursor_offload) { + REG_UPDATE(CURSOR0_FP_SCALE_BIAS_G_Y, CUR0_FP_BIAS_G_Y, attr->bias); + REG_UPDATE(CURSOR0_FP_SCALE_BIAS_G_Y, CUR0_FP_SCALE_G_Y, attr->scale); + REG_UPDATE(CURSOR0_FP_SCALE_BIAS_RB_CRCB, CUR0_FP_BIAS_RB_CRCB, attr->bias); + REG_UPDATE(CURSOR0_FP_SCALE_BIAS_RB_CRCB, CUR0_FP_SCALE_RB_CRCB, attr->scale); + } + + dpp_base->att.fp_scale_bias_g_y.bits.fp_bias_g_y = attr->bias; + dpp_base->att.fp_scale_bias_g_y.bits.fp_scale_g_y = attr->scale; + dpp_base->att.fp_scale_bias_rb_crcb.bits.fp_bias_rb_crcb = attr->bias; + dpp_base->att.fp_scale_bias_rb_crcb.bits.fp_scale_rb_crcb = attr->scale; } } diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c index 89f0d999bf35..242f1e6f0d8f 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.c @@ -35,6 +35,7 @@ static void dsc_write_to_registers(struct display_stream_compressor *dsc, const static const struct dsc_funcs dcn20_dsc_funcs = { .dsc_get_enc_caps = dsc2_get_enc_caps, .dsc_read_state = dsc2_read_state, + .dsc_read_reg_state = dsc2_read_reg_state, .dsc_validate_stream = dsc2_validate_stream, .dsc_set_config = dsc2_set_config, .dsc_get_packed_pps = dsc2_get_packed_pps, @@ -155,6 +156,13 @@ void dsc2_read_state(struct display_stream_compressor *dsc, struct dcn_dsc_state DSCRM_DSC_OPP_PIPE_SOURCE, &s->dsc_opp_source); } +void dsc2_read_reg_state(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state) +{ + struct dcn20_dsc *dsc20 = TO_DCN20_DSC(dsc); + + dccg_reg_state->dsc_top_control = REG_READ(DSC_TOP_CONTROL); + dccg_reg_state->dscc_interrupt_control_status = REG_READ(DSCC_INTERRUPT_CONTROL_STATUS); +} bool dsc2_validate_stream(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg) { @@ -407,7 +415,7 @@ bool dsc_prepare_config(const struct dsc_config *dsc_cfg, struct dsc_reg_values dsc_reg_vals->ich_reset_at_eol = (dsc_cfg->is_odm || dsc_reg_vals->num_slices_h > 1) ? 0xF : 0; // Need to find the ceiling value for the slice width - dsc_reg_vals->pps.slice_width = (dsc_cfg->pic_width + dsc_cfg->dc_dsc_cfg.num_slices_h - 1) / dsc_cfg->dc_dsc_cfg.num_slices_h; + dsc_reg_vals->pps.slice_width = (dsc_cfg->pic_width + dsc_cfg->dsc_padding + dsc_cfg->dc_dsc_cfg.num_slices_h - 1) / dsc_cfg->dc_dsc_cfg.num_slices_h; // TODO: in addition to validating slice height (pic height must be divisible by slice height), // see what happens when the same condition doesn't apply for slice_width/pic_width. dsc_reg_vals->pps.slice_height = dsc_cfg->pic_height / dsc_cfg->dc_dsc_cfg.num_slices_v; diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h index a9c04fc95bd1..2337c3a97235 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn20/dcn20_dsc.h @@ -606,6 +606,7 @@ bool dsc2_get_packed_pps(struct display_stream_compressor *dsc, uint8_t *dsc_packed_pps); void dsc2_read_state(struct display_stream_compressor *dsc, struct dcn_dsc_state *s); +void dsc2_read_reg_state(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state); bool dsc2_validate_stream(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg); void dsc2_set_config(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg, struct dsc_optc_config *dsc_optc_cfg); diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c index 6f4f5a3c4861..e712985f7abd 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn35/dcn35_dsc.c @@ -28,10 +28,11 @@ #include "reg_helper.h" static void dsc35_enable(struct display_stream_compressor *dsc, int opp_pipe); +static void dsc35_get_single_enc_caps(struct dsc_enc_caps *dsc_enc_caps, unsigned int max_dscclk_khz); static const struct dsc_funcs dcn35_dsc_funcs = { - .dsc_get_enc_caps = dsc2_get_enc_caps, .dsc_read_state = dsc2_read_state, + .dsc_read_reg_state = dsc2_read_reg_state, .dsc_validate_stream = dsc2_validate_stream, .dsc_set_config = dsc2_set_config, .dsc_get_packed_pps = dsc2_get_packed_pps, @@ -39,6 +40,7 @@ static const struct dsc_funcs dcn35_dsc_funcs = { .dsc_disable = dsc2_disable, .dsc_disconnect = dsc2_disconnect, .dsc_wait_disconnect_pending_clear = dsc2_wait_disconnect_pending_clear, + .dsc_get_single_enc_caps = dsc35_get_single_enc_caps, }; /* Macro definitios for REG_SET macros*/ @@ -110,3 +112,31 @@ void dsc35_set_fgcg(struct dcn20_dsc *dsc20, bool enable) { REG_UPDATE(DSC_TOP_CONTROL, DSC_FGCG_REP_DIS, !enable); } + +void dsc35_get_single_enc_caps(struct dsc_enc_caps *dsc_enc_caps, unsigned int max_dscclk_khz) +{ + dsc_enc_caps->dsc_version = 0x21; /* v1.2 - DP spec defined it in reverse order and we kept it */ + + dsc_enc_caps->slice_caps.bits.NUM_SLICES_1 = 1; + dsc_enc_caps->slice_caps.bits.NUM_SLICES_2 = 1; + dsc_enc_caps->slice_caps.bits.NUM_SLICES_3 = 1; + dsc_enc_caps->slice_caps.bits.NUM_SLICES_4 = 1; + + dsc_enc_caps->lb_bit_depth = 13; + dsc_enc_caps->is_block_pred_supported = true; + + dsc_enc_caps->color_formats.bits.RGB = 1; + dsc_enc_caps->color_formats.bits.YCBCR_444 = 1; + dsc_enc_caps->color_formats.bits.YCBCR_SIMPLE_422 = 1; + dsc_enc_caps->color_formats.bits.YCBCR_NATIVE_422 = 0; + dsc_enc_caps->color_formats.bits.YCBCR_NATIVE_420 = 1; + + dsc_enc_caps->color_depth.bits.COLOR_DEPTH_8_BPC = 1; + dsc_enc_caps->color_depth.bits.COLOR_DEPTH_10_BPC = 1; + dsc_enc_caps->color_depth.bits.COLOR_DEPTH_12_BPC = 1; + + dsc_enc_caps->max_total_throughput_mps = max_dscclk_khz * 3 / 1000; + + dsc_enc_caps->max_slice_width = 5184; /* (including 64 overlap pixels for eDP MSO mode) */ + dsc_enc_caps->bpp_increment_div = 16; /* 1/16th of a bit */ +} diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c index 7bd92ae8b13e..c1bdbb38c690 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c +++ b/drivers/gpu/drm/amd/display/dc/dsc/dcn401/dcn401_dsc.c @@ -26,6 +26,7 @@ static const struct dsc_funcs dcn401_dsc_funcs = { .dsc_disconnect = dsc401_disconnect, .dsc_wait_disconnect_pending_clear = dsc401_wait_disconnect_pending_clear, .dsc_get_single_enc_caps = dsc401_get_single_enc_caps, + .dsc_read_reg_state = dsc2_read_reg_state }; /* Macro definitios for REG_SET macros*/ diff --git a/drivers/gpu/drm/amd/display/dc/dsc/dsc.h b/drivers/gpu/drm/amd/display/dc/dsc/dsc.h index b0bd1f9425b5..81c83d5fe042 100644 --- a/drivers/gpu/drm/amd/display/dc/dsc/dsc.h +++ b/drivers/gpu/drm/amd/display/dc/dsc/dsc.h @@ -41,6 +41,7 @@ struct dsc_config { enum dc_color_depth color_depth; /* Bits per component */ bool is_odm; struct dc_dsc_config dc_dsc_cfg; + uint32_t dsc_padding; }; @@ -65,6 +66,10 @@ struct dcn_dsc_state { uint32_t dsc_opp_source; }; +struct dcn_dsc_reg_state { + uint32_t dsc_top_control; + uint32_t dscc_interrupt_control_status; +}; /* DSC encoder capabilities * They differ from the DPCD DSC caps because they are based on AMD DSC encoder caps. @@ -99,6 +104,7 @@ struct dsc_enc_caps { struct dsc_funcs { void (*dsc_get_enc_caps)(struct dsc_enc_caps *dsc_enc_caps, int pixel_clock_100Hz); void (*dsc_read_state)(struct display_stream_compressor *dsc, struct dcn_dsc_state *s); + void (*dsc_read_reg_state)(struct display_stream_compressor *dsc, struct dcn_dsc_reg_state *dccg_reg_state); bool (*dsc_validate_stream)(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg); void (*dsc_set_config)(struct display_stream_compressor *dsc, const struct dsc_config *dsc_cfg, struct dsc_optc_config *dsc_optc_cfg); diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c index e7e5f6d4778e..181a93dc46e6 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.c @@ -440,33 +440,15 @@ void hubbub3_init_watermarks(struct hubbub *hubbub) REG_WRITE(DCHUBBUB_ARB_ALLOW_DRAM_CLK_CHANGE_WATERMARK_D, reg); } -void hubbub3_get_det_sizes(struct hubbub *hubbub, uint32_t *curr_det_sizes, uint32_t *target_det_sizes) +void hubbub3_read_reg_state(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state) { struct dcn20_hubbub *hubbub1 = TO_DCN20_HUBBUB(hubbub); - REG_GET_2(DCHUBBUB_DET0_CTRL, DET0_SIZE_CURRENT, &curr_det_sizes[0], - DET0_SIZE, &target_det_sizes[0]); - - REG_GET_2(DCHUBBUB_DET1_CTRL, DET1_SIZE_CURRENT, &curr_det_sizes[1], - DET1_SIZE, &target_det_sizes[1]); - - REG_GET_2(DCHUBBUB_DET2_CTRL, DET2_SIZE_CURRENT, &curr_det_sizes[2], - DET2_SIZE, &target_det_sizes[2]); - - REG_GET_2(DCHUBBUB_DET3_CTRL, DET3_SIZE_CURRENT, &curr_det_sizes[3], - DET3_SIZE, &target_det_sizes[3]); - -} - -uint32_t hubbub3_compbuf_config_error(struct hubbub *hubbub) -{ - struct dcn20_hubbub *hubbub1 = TO_DCN20_HUBBUB(hubbub); - uint32_t compbuf_config_error = 0; - - REG_GET(DCHUBBUB_COMPBUF_CTRL, CONFIG_ERROR, - &compbuf_config_error); - - return compbuf_config_error; + hubbub_reg_state->det0_ctrl = REG_READ(DCHUBBUB_DET0_CTRL); + hubbub_reg_state->det1_ctrl = REG_READ(DCHUBBUB_DET1_CTRL); + hubbub_reg_state->det2_ctrl = REG_READ(DCHUBBUB_DET2_CTRL); + hubbub_reg_state->det3_ctrl = REG_READ(DCHUBBUB_DET3_CTRL); + hubbub_reg_state->compbuf_ctrl = REG_READ(DCHUBBUB_COMPBUF_CTRL); } static const struct hubbub_funcs hubbub30_funcs = { @@ -486,8 +468,7 @@ static const struct hubbub_funcs hubbub30_funcs = { .force_pstate_change_control = hubbub3_force_pstate_change_control, .init_watermarks = hubbub3_init_watermarks, .hubbub_read_state = hubbub2_read_state, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub3_construct(struct dcn20_hubbub *hubbub3, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h index 49a469969d36..9e14de3ccaee 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn30/dcn30_hubbub.h @@ -133,10 +133,6 @@ void hubbub3_force_pstate_change_control(struct hubbub *hubbub, void hubbub3_init_watermarks(struct hubbub *hubbub); -void hubbub3_get_det_sizes(struct hubbub *hubbub, - uint32_t *curr_det_sizes, - uint32_t *target_det_sizes); - -uint32_t hubbub3_compbuf_config_error(struct hubbub *hubbub); +void hubbub3_read_reg_state(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state); #endif diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c index cdb20251a154..5a03758e3de6 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn31/dcn31_hubbub.c @@ -933,8 +933,8 @@ int hubbub31_init_dchub_sys_ctx(struct hubbub *hubbub, dcn20_vmid_setup(&hubbub2->vmid[15], &phys_config); } - - dcn21_dchvm_init(hubbub); + if (hubbub->funcs->dchvm_init) + hubbub->funcs->dchvm_init(hubbub); return NUM_VMID; } @@ -1071,8 +1071,8 @@ static const struct hubbub_funcs hubbub31_funcs = { .program_compbuf_size = dcn31_program_compbuf_size, .init_crb = dcn31_init_crb, .hubbub_read_state = hubbub2_read_state, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state, + .dchvm_init = dcn21_dchvm_init }; void hubbub31_construct(struct dcn20_hubbub *hubbub31, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c index 4d4ca6d77bbd..237331b35378 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn32/dcn32_hubbub.c @@ -1037,8 +1037,7 @@ static const struct hubbub_funcs hubbub32_funcs = { .force_usr_retraining_allow = hubbub32_force_usr_retraining_allow, .set_request_limit = hubbub32_set_request_limit, .get_mall_en = hubbub32_get_mall_en, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub32_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c index a443722a8632..43ba399f4822 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.c @@ -549,6 +549,55 @@ void hubbub35_init(struct hubbub *hubbub) memset(&hubbub2->watermarks.a.cstate_pstate, 0, sizeof(hubbub2->watermarks.a.cstate_pstate)); } +void dcn35_dchvm_init(struct hubbub *hubbub) +{ + struct dcn20_hubbub *hubbub2 = TO_DCN20_HUBBUB(hubbub); + uint32_t riommu_active; + int i; + + //Init DCHVM block + REG_UPDATE(DCHVM_CTRL0, HOSTVM_INIT_REQ, 1); + + //Poll until RIOMMU_ACTIVE = 1 + for (i = 0; i < 100; i++) { + REG_GET(DCHVM_RIOMMU_STAT0, RIOMMU_ACTIVE, &riommu_active); + + if (riommu_active) + break; + else + udelay(5); + } + + if (riommu_active) { + // Disable gating and memory power requests + REG_UPDATE(DCHVM_MEM_CTRL, HVM_GPUVMRET_PWR_REQ_DIS, 1); + REG_UPDATE_4(DCHVM_CLK_CTRL, + HVM_DISPCLK_R_GATE_DIS, 1, + HVM_DISPCLK_G_GATE_DIS, 1, + HVM_DCFCLK_R_GATE_DIS, 1, + HVM_DCFCLK_G_GATE_DIS, 1); + + //Reflect the power status of DCHUBBUB + REG_UPDATE(DCHVM_RIOMMU_CTRL0, HOSTVM_POWERSTATUS, 1); + + //Start rIOMMU prefetching + REG_UPDATE(DCHVM_RIOMMU_CTRL0, HOSTVM_PREFETCH_REQ, 1); + + //Poll until HOSTVM_PREFETCH_DONE = 1 + REG_WAIT(DCHVM_RIOMMU_STAT0, HOSTVM_PREFETCH_DONE, 1, 5, 100); + + //Enable memory power requests + REG_UPDATE(DCHVM_MEM_CTRL, HVM_GPUVMRET_PWR_REQ_DIS, 0); + // Enable dynamic clock gating + REG_UPDATE_4(DCHVM_CLK_CTRL, + HVM_DISPCLK_R_GATE_DIS, 0, + HVM_DISPCLK_G_GATE_DIS, 0, + HVM_DCFCLK_R_GATE_DIS, 0, + HVM_DCFCLK_G_GATE_DIS, 0); + hubbub->riommu_active = true; + } +} + /*static void hubbub35_set_request_limit(struct hubbub *hubbub, int memory_channel_count, int words_per_channel) @@ -589,8 +638,8 @@ static const struct hubbub_funcs hubbub35_funcs = { .hubbub_read_state = hubbub2_read_state, .force_usr_retraining_allow = hubbub32_force_usr_retraining_allow, .dchubbub_init = hubbub35_init, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state, + .dchvm_init = dcn35_dchvm_init }; void hubbub35_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.h b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.h index 23fecf88556c..9f65fff1bd4d 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.h +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn35/dcn35_hubbub.h @@ -168,4 +168,5 @@ void dcn35_program_compbuf_size(struct hubbub *hubbub, unsigned int compbuf_size_kb, bool safe_to_increase); void dcn35_init_crb(struct hubbub *hubbub); void hubbub35_init(struct hubbub *hubbub); +void dcn35_dchvm_init(struct hubbub *hubbub); #endif diff --git a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c index a36273a52880..d11afd1ce72a 100644 --- a/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c +++ b/drivers/gpu/drm/amd/display/dc/hubbub/dcn401/dcn401_hubbub.c @@ -1247,8 +1247,7 @@ static const struct hubbub_funcs hubbub4_01_funcs = { .program_compbuf_segments = dcn401_program_compbuf_segments, .wait_for_det_update = dcn401_wait_for_det_update, .program_arbiter = dcn401_program_arbiter, - .get_det_sizes = hubbub3_get_det_sizes, - .compbuf_config_error = hubbub3_compbuf_config_error, + .hubbub_read_reg_state = hubbub3_read_reg_state }; void hubbub401_construct(struct dcn20_hubbub *hubbub2, diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.c index 9b026600b90e..6378e3fd7249 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.c @@ -550,6 +550,7 @@ void hubp_reset(struct hubp *hubp) { memset(&hubp->pos, 0, sizeof(hubp->pos)); memset(&hubp->att, 0, sizeof(hubp->att)); + hubp->cursor_offload = false; } void hubp1_program_surface_config( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h index cf2eb9793008..f2571076fc50 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn10/dcn10_hubp.h @@ -105,7 +105,9 @@ SRI(DCN_CUR0_TTU_CNTL0, HUBPREQ, id),\ SRI(DCN_CUR0_TTU_CNTL1, HUBPREQ, id),\ SRI(HUBP_CLK_CNTL, HUBP, id),\ - SRI(HUBPRET_READ_LINE_VALUE, HUBPRET, id) + SRI(HUBPRET_READ_LINE_VALUE, HUBPRET, id),\ + SRI(HUBP_MEASURE_WIN_CTRL_DCFCLK, HUBP, id),\ + SRI(HUBP_MEASURE_WIN_CTRL_DPPCLK, HUBP, id) /* Register address initialization macro for ASICs with VM */ #define HUBP_REG_LIST_DCN_VM(id)\ @@ -251,7 +253,19 @@ uint32_t CURSOR_HOT_SPOT; \ uint32_t CURSOR_DST_OFFSET; \ uint32_t HUBP_CLK_CNTL; \ - uint32_t HUBPRET_READ_LINE_VALUE + uint32_t HUBPRET_READ_LINE_VALUE; \ + uint32_t HUBP_MEASURE_WIN_CTRL_DCFCLK; \ + uint32_t HUBP_MEASURE_WIN_CTRL_DPPCLK; \ + uint32_t HUBPRET_INTERRUPT; \ + uint32_t HUBPRET_MEM_PWR_CTRL; \ + uint32_t HUBPRET_MEM_PWR_STATUS; \ + uint32_t HUBPRET_READ_LINE_CTRL0; \ + uint32_t HUBPRET_READ_LINE_CTRL1; \ + uint32_t HUBPRET_READ_LINE0; \ + uint32_t HUBPRET_READ_LINE1; \ + uint32_t HUBPREQ_MEM_PWR_CTRL; \ + uint32_t HUBPREQ_MEM_PWR_STATUS + #define HUBP_SF(reg_name, field_name, post_fix)\ .field_name = reg_name ## __ ## field_name ## post_fix @@ -688,6 +702,123 @@ struct dcn_fl_regs_st { uint32_t lut_fl_mode; uint32_t lut_fl_format; }; +struct dcn_hubp_reg_state { + uint32_t hubp_cntl; + uint32_t mall_config; + uint32_t mall_sub_vp; + uint32_t hubp_req_size_config; + uint32_t hubp_req_size_config_c; + uint32_t vmpg_config; + uint32_t addr_config; + uint32_t pri_viewport_dimension; + uint32_t pri_viewport_dimension_c; + uint32_t pri_viewport_start; + uint32_t pri_viewport_start_c; + uint32_t sec_viewport_dimension; + uint32_t sec_viewport_dimension_c; + uint32_t sec_viewport_start; + uint32_t sec_viewport_start_c; + uint32_t surface_config; + uint32_t tiling_config; + uint32_t clk_cntl; + uint32_t mall_status; + uint32_t measure_win_ctrl_dcfclk; + uint32_t measure_win_ctrl_dppclk; + + uint32_t blank_offset_0; + uint32_t blank_offset_1; + uint32_t cursor_settings; + uint32_t dcn_cur0_ttu_cntl0; + uint32_t dcn_cur0_ttu_cntl1; + uint32_t dcn_cur1_ttu_cntl0; + uint32_t dcn_cur1_ttu_cntl1; + uint32_t dcn_dmdat_vm_cntl; + uint32_t dcn_expansion_mode; + uint32_t dcn_global_ttu_cntl; + uint32_t dcn_surf0_ttu_cntl0; + uint32_t dcn_surf0_ttu_cntl1; + uint32_t dcn_surf1_ttu_cntl0; + uint32_t dcn_surf1_ttu_cntl1; + uint32_t dcn_ttu_qos_wm; + uint32_t dcn_vm_mx_l1_tlb_cntl; + uint32_t dcn_vm_system_aperture_high_addr; + uint32_t dcn_vm_system_aperture_low_addr; + uint32_t dcsurf_flip_control; + uint32_t dcsurf_flip_control2; + uint32_t dcsurf_primary_meta_surface_address; + uint32_t dcsurf_primary_meta_surface_address_c; + uint32_t dcsurf_primary_meta_surface_address_high; + uint32_t dcsurf_primary_meta_surface_address_high_c; + uint32_t dcsurf_primary_surface_address; + uint32_t dcsurf_primary_surface_address_c; + uint32_t dcsurf_primary_surface_address_high; + uint32_t dcsurf_primary_surface_address_high_c; + uint32_t dcsurf_secondary_meta_surface_address; + uint32_t dcsurf_secondary_meta_surface_address_c; + uint32_t dcsurf_secondary_meta_surface_address_high; + uint32_t dcsurf_secondary_meta_surface_address_high_c; + uint32_t dcsurf_secondary_surface_address; + uint32_t dcsurf_secondary_surface_address_c; + uint32_t dcsurf_secondary_surface_address_high; + uint32_t dcsurf_secondary_surface_address_high_c; + uint32_t dcsurf_surface_control; + uint32_t dcsurf_surface_earliest_inuse; + uint32_t dcsurf_surface_earliest_inuse_c; + uint32_t dcsurf_surface_earliest_inuse_high; + uint32_t dcsurf_surface_earliest_inuse_high_c; + uint32_t dcsurf_surface_flip_interrupt; + uint32_t dcsurf_surface_inuse; + uint32_t dcsurf_surface_inuse_c; + uint32_t dcsurf_surface_inuse_high; + uint32_t dcsurf_surface_inuse_high_c; + uint32_t dcsurf_surface_pitch; + uint32_t dcsurf_surface_pitch_c; + uint32_t dst_after_scaler; + uint32_t dst_dimensions; + uint32_t dst_y_delta_drq_limit; + uint32_t flip_parameters_0; + uint32_t flip_parameters_1; + uint32_t flip_parameters_2; + uint32_t flip_parameters_3; + uint32_t flip_parameters_4; + uint32_t flip_parameters_5; + uint32_t flip_parameters_6; + uint32_t hubpreq_mem_pwr_ctrl; + uint32_t hubpreq_mem_pwr_status; + uint32_t nom_parameters_0; + uint32_t nom_parameters_1; + uint32_t nom_parameters_2; + uint32_t nom_parameters_3; + uint32_t nom_parameters_4; + uint32_t nom_parameters_5; + uint32_t nom_parameters_6; + uint32_t nom_parameters_7; + uint32_t per_line_delivery; + uint32_t per_line_delivery_pre; + uint32_t prefetch_settings; + uint32_t prefetch_settings_c; + uint32_t ref_freq_to_pix_freq; + uint32_t uclk_pstate_force; + uint32_t vblank_parameters_0; + uint32_t vblank_parameters_1; + uint32_t vblank_parameters_2; + uint32_t vblank_parameters_3; + uint32_t vblank_parameters_4; + uint32_t vblank_parameters_5; + uint32_t vblank_parameters_6; + uint32_t vmid_settings_0; + + uint32_t hubpret_control; + uint32_t hubpret_interrupt; + uint32_t hubpret_mem_pwr_ctrl; + uint32_t hubpret_mem_pwr_status; + uint32_t hubpret_read_line_ctrl0; + uint32_t hubpret_read_line_ctrl1; + uint32_t hubpret_read_line_status; + uint32_t hubpret_read_line_value; + uint32_t hubpret_read_line0; + uint32_t hubpret_read_line1; +}; struct dcn_hubp_state { struct _vcs_dpi_display_dlg_regs_st dlg_attr; @@ -718,7 +849,6 @@ struct dcn_hubp_state { uint32_t hubp_cntl; uint32_t flip_control; }; - struct dcn10_hubp { struct hubp base; struct dcn_hubp_state state; diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.c index 91259b896e03..92288de4cc10 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.c @@ -613,26 +613,28 @@ void hubp2_cursor_set_attributes( hubp->curs_attr = *attr; - REG_UPDATE(CURSOR_SURFACE_ADDRESS_HIGH, - CURSOR_SURFACE_ADDRESS_HIGH, attr->address.high_part); - REG_UPDATE(CURSOR_SURFACE_ADDRESS, - CURSOR_SURFACE_ADDRESS, attr->address.low_part); - - REG_UPDATE_2(CURSOR_SIZE, - CURSOR_WIDTH, attr->width, - CURSOR_HEIGHT, attr->height); - - REG_UPDATE_4(CURSOR_CONTROL, - CURSOR_MODE, attr->color_format, - CURSOR_2X_MAGNIFY, attr->attribute_flags.bits.ENABLE_MAGNIFICATION, - CURSOR_PITCH, hw_pitch, - CURSOR_LINES_PER_CHUNK, lpc); - - REG_SET_2(CURSOR_SETTINGS, 0, - /* no shift of the cursor HDL schedule */ - CURSOR0_DST_Y_OFFSET, 0, - /* used to shift the cursor chunk request deadline */ - CURSOR0_CHUNK_HDL_ADJUST, 3); + if (!hubp->cursor_offload) { + REG_UPDATE(CURSOR_SURFACE_ADDRESS_HIGH, + CURSOR_SURFACE_ADDRESS_HIGH, attr->address.high_part); + REG_UPDATE(CURSOR_SURFACE_ADDRESS, + CURSOR_SURFACE_ADDRESS, attr->address.low_part); + + REG_UPDATE_2(CURSOR_SIZE, + CURSOR_WIDTH, attr->width, + CURSOR_HEIGHT, attr->height); + + REG_UPDATE_4(CURSOR_CONTROL, + CURSOR_MODE, attr->color_format, + CURSOR_2X_MAGNIFY, attr->attribute_flags.bits.ENABLE_MAGNIFICATION, + CURSOR_PITCH, hw_pitch, + CURSOR_LINES_PER_CHUNK, lpc); + + REG_SET_2(CURSOR_SETTINGS, 0, + /* no shift of the cursor HDL schedule */ + CURSOR0_DST_Y_OFFSET, 0, + /* used to shift the cursor chunk request deadline */ + CURSOR0_CHUNK_HDL_ADJUST, 3); + } hubp->att.SURFACE_ADDR_HIGH = attr->address.high_part; hubp->att.SURFACE_ADDR = attr->address.low_part; @@ -1059,23 +1061,28 @@ void hubp2_cursor_set_position( cur_en = 0; /* not visible beyond top edge*/ if (hubp->pos.cur_ctl.bits.cur_enable != cur_en) { - if (cur_en && REG_READ(CURSOR_SURFACE_ADDRESS) == 0) + bool cursor_not_programmed = hubp->att.SURFACE_ADDR == 0 && hubp->att.SURFACE_ADDR_HIGH == 0; + + if (cur_en && cursor_not_programmed) hubp->funcs->set_cursor_attributes(hubp, &hubp->curs_attr); - REG_UPDATE(CURSOR_CONTROL, - CURSOR_ENABLE, cur_en); + if (!hubp->cursor_offload) + REG_UPDATE(CURSOR_CONTROL, CURSOR_ENABLE, cur_en); } - REG_SET_2(CURSOR_POSITION, 0, - CURSOR_X_POSITION, pos->x, - CURSOR_Y_POSITION, pos->y); + if (!hubp->cursor_offload) { + REG_SET_2(CURSOR_POSITION, 0, + CURSOR_X_POSITION, pos->x, + CURSOR_Y_POSITION, pos->y); - REG_SET_2(CURSOR_HOT_SPOT, 0, - CURSOR_HOT_SPOT_X, pos->x_hotspot, - CURSOR_HOT_SPOT_Y, pos->y_hotspot); + REG_SET_2(CURSOR_HOT_SPOT, 0, + CURSOR_HOT_SPOT_X, pos->x_hotspot, + CURSOR_HOT_SPOT_Y, pos->y_hotspot); + + REG_SET(CURSOR_DST_OFFSET, 0, + CURSOR_DST_X_OFFSET, dst_x_offset); + } - REG_SET(CURSOR_DST_OFFSET, 0, - CURSOR_DST_X_OFFSET, dst_x_offset); /* TODO Handle surface pixel formats other than 4:4:4 */ /* Cursor Position Register Config */ hubp->pos.cur_ctl.bits.cur_enable = cur_en; diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h index f325db555102..7062e6653062 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn20/dcn20_hubp.h @@ -145,7 +145,8 @@ uint32_t FLIP_PARAMETERS_2;\ uint32_t DCN_CUR1_TTU_CNTL0;\ uint32_t DCN_CUR1_TTU_CNTL1;\ - uint32_t VMID_SETTINGS_0 + uint32_t VMID_SETTINGS_0;\ + uint32_t DST_Y_DELTA_DRQ_LIMIT /*shared with dcn3.x*/ #define DCN21_HUBP_REG_COMMON_VARIABLE_LIST \ @@ -176,7 +177,10 @@ uint32_t HUBP_3DLUT_CONTROL;\ uint32_t HUBP_3DLUT_DLG_PARAM;\ uint32_t DCSURF_VIEWPORT_MCACHE_SPLIT_COORDINATE;\ - uint32_t DCHUBP_MCACHEID_CONFIG + uint32_t DCHUBP_MCACHEID_CONFIG;\ + uint32_t DCHUBP_MALL_SUB_VP;\ + uint32_t DCHUBP_ADDR_CONFIG;\ + uint32_t HUBP_MALL_STATUS #define DCN2_HUBP_REG_FIELD_VARIABLE_LIST(type) \ DCN_HUBP_REG_FIELD_BASE_LIST(type); \ diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn21/dcn21_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn21/dcn21_hubp.c index e2740482e1cf..08ea0a1b9e7f 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn21/dcn21_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn21/dcn21_hubp.c @@ -73,8 +73,6 @@ * On any mode switch, if the new reg values are smaller than the current values, * then update the regs with the new values. * - * Link to the ticket: http://ontrack-internal.amd.com/browse/DEDCN21-142 - * */ void apply_DEDCN21_142_wa_for_hostvm_deadline( struct hubp *hubp, diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c index 556214b2227d..0cc6f4558989 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.c @@ -476,6 +476,126 @@ void hubp3_read_state(struct hubp *hubp) } +void hubp3_read_reg_state(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state) +{ + struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); + + reg_state->hubp_cntl = REG_READ(DCHUBP_CNTL); + reg_state->mall_config = REG_READ(DCHUBP_MALL_CONFIG); + reg_state->mall_sub_vp = REG_READ(DCHUBP_MALL_SUB_VP); + reg_state->hubp_req_size_config = REG_READ(DCHUBP_REQ_SIZE_CONFIG); + reg_state->hubp_req_size_config_c = REG_READ(DCHUBP_REQ_SIZE_CONFIG_C); + reg_state->vmpg_config = REG_READ(DCHUBP_VMPG_CONFIG); + reg_state->addr_config = REG_READ(DCSURF_ADDR_CONFIG); + reg_state->pri_viewport_dimension = REG_READ(DCSURF_PRI_VIEWPORT_DIMENSION); + reg_state->pri_viewport_dimension_c = REG_READ(DCSURF_PRI_VIEWPORT_DIMENSION_C); + reg_state->pri_viewport_start = REG_READ(DCSURF_PRI_VIEWPORT_START); + reg_state->pri_viewport_start_c = REG_READ(DCSURF_PRI_VIEWPORT_START_C); + reg_state->sec_viewport_dimension = REG_READ(DCSURF_SEC_VIEWPORT_DIMENSION); + reg_state->sec_viewport_dimension_c = REG_READ(DCSURF_SEC_VIEWPORT_DIMENSION_C); + reg_state->sec_viewport_start = REG_READ(DCSURF_SEC_VIEWPORT_START); + reg_state->sec_viewport_start_c = REG_READ(DCSURF_SEC_VIEWPORT_START_C); + reg_state->surface_config = REG_READ(DCSURF_SURFACE_CONFIG); + reg_state->tiling_config = REG_READ(DCSURF_TILING_CONFIG); + reg_state->clk_cntl = REG_READ(HUBP_CLK_CNTL); + reg_state->mall_status = REG_READ(HUBP_MALL_STATUS); + reg_state->measure_win_ctrl_dcfclk = REG_READ(HUBP_MEASURE_WIN_CTRL_DCFCLK); + reg_state->measure_win_ctrl_dppclk = REG_READ(HUBP_MEASURE_WIN_CTRL_DPPCLK); + + reg_state->blank_offset_0 = REG_READ(BLANK_OFFSET_0); + reg_state->blank_offset_1 = REG_READ(BLANK_OFFSET_1); + reg_state->cursor_settings = REG_READ(CURSOR_SETTINGS); + reg_state->dcn_cur0_ttu_cntl0 = REG_READ(DCN_CUR0_TTU_CNTL0); + reg_state->dcn_cur0_ttu_cntl1 = REG_READ(DCN_CUR0_TTU_CNTL1); + reg_state->dcn_cur1_ttu_cntl0 = REG_READ(DCN_CUR1_TTU_CNTL0); + reg_state->dcn_cur1_ttu_cntl1 = REG_READ(DCN_CUR1_TTU_CNTL1); + reg_state->dcn_dmdat_vm_cntl = REG_READ(DCN_DMDATA_VM_CNTL); + reg_state->dcn_expansion_mode = REG_READ(DCN_EXPANSION_MODE); + reg_state->dcn_global_ttu_cntl = REG_READ(DCN_GLOBAL_TTU_CNTL); + reg_state->dcn_surf0_ttu_cntl0 = REG_READ(DCN_SURF0_TTU_CNTL0); + reg_state->dcn_surf0_ttu_cntl1 = REG_READ(DCN_SURF0_TTU_CNTL1); + reg_state->dcn_surf1_ttu_cntl0 = REG_READ(DCN_SURF1_TTU_CNTL0); + reg_state->dcn_surf1_ttu_cntl1 = REG_READ(DCN_SURF1_TTU_CNTL1); + reg_state->dcn_ttu_qos_wm = REG_READ(DCN_TTU_QOS_WM); + reg_state->dcn_vm_mx_l1_tlb_cntl = REG_READ(DCN_VM_MX_L1_TLB_CNTL); + reg_state->dcn_vm_system_aperture_high_addr = REG_READ(DCN_VM_SYSTEM_APERTURE_HIGH_ADDR); + reg_state->dcn_vm_system_aperture_low_addr = REG_READ(DCN_VM_SYSTEM_APERTURE_LOW_ADDR); + reg_state->dcsurf_flip_control = REG_READ(DCSURF_FLIP_CONTROL); + reg_state->dcsurf_flip_control2 = REG_READ(DCSURF_FLIP_CONTROL2); + reg_state->dcsurf_primary_meta_surface_address = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS); + reg_state->dcsurf_primary_meta_surface_address_c = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_C); + reg_state->dcsurf_primary_meta_surface_address_high = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_primary_meta_surface_address_high_c = REG_READ(DCSURF_PRIMARY_META_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_primary_surface_address = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS); + reg_state->dcsurf_primary_surface_address_c = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_C); + reg_state->dcsurf_primary_surface_address_high = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_primary_surface_address_high_c = REG_READ(DCSURF_PRIMARY_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_secondary_meta_surface_address = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS); + reg_state->dcsurf_secondary_meta_surface_address_c = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_C); + reg_state->dcsurf_secondary_meta_surface_address_high = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_secondary_meta_surface_address_high_c = REG_READ(DCSURF_SECONDARY_META_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_secondary_surface_address = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS); + reg_state->dcsurf_secondary_surface_address_c = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_C); + reg_state->dcsurf_secondary_surface_address_high = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH); + reg_state->dcsurf_secondary_surface_address_high_c = REG_READ(DCSURF_SECONDARY_SURFACE_ADDRESS_HIGH_C); + reg_state->dcsurf_surface_control = REG_READ(DCSURF_SURFACE_CONTROL); + reg_state->dcsurf_surface_earliest_inuse = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE); + reg_state->dcsurf_surface_earliest_inuse_c = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_C); + reg_state->dcsurf_surface_earliest_inuse_high = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_HIGH); + reg_state->dcsurf_surface_earliest_inuse_high_c = REG_READ(DCSURF_SURFACE_EARLIEST_INUSE_HIGH_C); + reg_state->dcsurf_surface_flip_interrupt = REG_READ(DCSURF_SURFACE_FLIP_INTERRUPT); + reg_state->dcsurf_surface_inuse = REG_READ(DCSURF_SURFACE_INUSE); + reg_state->dcsurf_surface_inuse_c = REG_READ(DCSURF_SURFACE_INUSE_C); + reg_state->dcsurf_surface_inuse_high = REG_READ(DCSURF_SURFACE_INUSE_HIGH); + reg_state->dcsurf_surface_inuse_high_c = REG_READ(DCSURF_SURFACE_INUSE_HIGH_C); + reg_state->dcsurf_surface_pitch = REG_READ(DCSURF_SURFACE_PITCH); + reg_state->dcsurf_surface_pitch_c = REG_READ(DCSURF_SURFACE_PITCH_C); + reg_state->dst_after_scaler = REG_READ(DST_AFTER_SCALER); + reg_state->dst_dimensions = REG_READ(DST_DIMENSIONS); + reg_state->dst_y_delta_drq_limit = REG_READ(DST_Y_DELTA_DRQ_LIMIT); + reg_state->flip_parameters_0 = REG_READ(FLIP_PARAMETERS_0); + reg_state->flip_parameters_1 = REG_READ(FLIP_PARAMETERS_1); + reg_state->flip_parameters_2 = REG_READ(FLIP_PARAMETERS_2); + reg_state->flip_parameters_3 = REG_READ(FLIP_PARAMETERS_3); + reg_state->flip_parameters_4 = REG_READ(FLIP_PARAMETERS_4); + reg_state->flip_parameters_5 = REG_READ(FLIP_PARAMETERS_5); + reg_state->flip_parameters_6 = REG_READ(FLIP_PARAMETERS_6); + reg_state->hubpreq_mem_pwr_ctrl = REG_READ(HUBPREQ_MEM_PWR_CTRL); + reg_state->hubpreq_mem_pwr_status = REG_READ(HUBPREQ_MEM_PWR_STATUS); + reg_state->nom_parameters_0 = REG_READ(NOM_PARAMETERS_0); + reg_state->nom_parameters_1 = REG_READ(NOM_PARAMETERS_1); + reg_state->nom_parameters_2 = REG_READ(NOM_PARAMETERS_2); + reg_state->nom_parameters_3 = REG_READ(NOM_PARAMETERS_3); + reg_state->nom_parameters_4 = REG_READ(NOM_PARAMETERS_4); + reg_state->nom_parameters_5 = REG_READ(NOM_PARAMETERS_5); + reg_state->nom_parameters_6 = REG_READ(NOM_PARAMETERS_6); + reg_state->nom_parameters_7 = REG_READ(NOM_PARAMETERS_7); + reg_state->per_line_delivery = REG_READ(PER_LINE_DELIVERY); + reg_state->per_line_delivery_pre = REG_READ(PER_LINE_DELIVERY_PRE); + reg_state->prefetch_settings = REG_READ(PREFETCH_SETTINGS); + reg_state->prefetch_settings_c = REG_READ(PREFETCH_SETTINGS_C); + reg_state->ref_freq_to_pix_freq = REG_READ(REF_FREQ_TO_PIX_FREQ); + reg_state->uclk_pstate_force = REG_READ(UCLK_PSTATE_FORCE); + reg_state->vblank_parameters_0 = REG_READ(VBLANK_PARAMETERS_0); + reg_state->vblank_parameters_1 = REG_READ(VBLANK_PARAMETERS_1); + reg_state->vblank_parameters_2 = REG_READ(VBLANK_PARAMETERS_2); + reg_state->vblank_parameters_3 = REG_READ(VBLANK_PARAMETERS_3); + reg_state->vblank_parameters_4 = REG_READ(VBLANK_PARAMETERS_4); + reg_state->vblank_parameters_5 = REG_READ(VBLANK_PARAMETERS_5); + reg_state->vblank_parameters_6 = REG_READ(VBLANK_PARAMETERS_6); + reg_state->vmid_settings_0 = REG_READ(VMID_SETTINGS_0); + reg_state->hubpret_control = REG_READ(HUBPRET_CONTROL); + reg_state->hubpret_interrupt = REG_READ(HUBPRET_INTERRUPT); + reg_state->hubpret_mem_pwr_ctrl = REG_READ(HUBPRET_MEM_PWR_CTRL); + reg_state->hubpret_mem_pwr_status = REG_READ(HUBPRET_MEM_PWR_STATUS); + reg_state->hubpret_read_line_ctrl0 = REG_READ(HUBPRET_READ_LINE_CTRL0); + reg_state->hubpret_read_line_ctrl1 = REG_READ(HUBPRET_READ_LINE_CTRL1); + reg_state->hubpret_read_line_status = REG_READ(HUBPRET_READ_LINE_STATUS); + reg_state->hubpret_read_line_value = REG_READ(HUBPRET_READ_LINE_VALUE); + reg_state->hubpret_read_line0 = REG_READ(HUBPRET_READ_LINE0); + reg_state->hubpret_read_line1 = REG_READ(HUBPRET_READ_LINE1); +} + void hubp3_setup( struct hubp *hubp, struct _vcs_dpi_display_dlg_regs_st *dlg_attr, @@ -505,30 +625,6 @@ void hubp3_init(struct hubp *hubp) hubp_reset(hubp); } -uint32_t hubp3_get_current_read_line(struct hubp *hubp) -{ - uint32_t read_line = 0; - struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - - REG_GET(HUBPRET_READ_LINE_VALUE, - PIPE_READ_LINE, - &read_line); - - return read_line; -} - -unsigned int hubp3_get_underflow_status(struct hubp *hubp) -{ - uint32_t hubp_underflow = 0; - struct dcn20_hubp *hubp2 = TO_DCN20_HUBP(hubp); - - REG_GET(DCHUBP_CNTL, - HUBP_UNDERFLOW_STATUS, - &hubp_underflow); - - return hubp_underflow; -} - static struct hubp_funcs dcn30_hubp_funcs = { .hubp_enable_tripleBuffer = hubp2_enable_triplebuffer, .hubp_is_triplebuffer_enabled = hubp2_is_triplebuffer_enabled, @@ -558,8 +654,7 @@ static struct hubp_funcs dcn30_hubp_funcs = { .hubp_soft_reset = hubp1_soft_reset, .hubp_set_flip_int = hubp1_set_flip_int, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp3_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h index 842f4eb72cc8..c767e9f4f9b3 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn30/dcn30_hubp.h @@ -296,6 +296,8 @@ void hubp3_dmdata_set_attributes( void hubp3_read_state(struct hubp *hubp); +void hubp3_read_reg_state(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state); + void hubp3_init(struct hubp *hubp); void hubp3_clear_tiling(struct hubp *hubp); diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c index 47101847c2b7..189045f85039 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn31/dcn31_hubp.c @@ -110,9 +110,7 @@ static struct hubp_funcs dcn31_hubp_funcs = { .hubp_in_blank = hubp1_in_blank, .program_extended_blank = hubp31_program_extended_blank, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state, }; bool hubp31_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c index a5f23bb2a76a..a781085b046b 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn32/dcn32_hubp.c @@ -118,29 +118,7 @@ void hubp32_cursor_set_attributes( uint32_t cursor_width = ((attr->width + 63) / 64) * 64; uint32_t cursor_height = attr->height; uint32_t cursor_size = cursor_width * cursor_height; - - hubp->curs_attr = *attr; - - REG_UPDATE(CURSOR_SURFACE_ADDRESS_HIGH, - CURSOR_SURFACE_ADDRESS_HIGH, attr->address.high_part); - REG_UPDATE(CURSOR_SURFACE_ADDRESS, - CURSOR_SURFACE_ADDRESS, attr->address.low_part); - - REG_UPDATE_2(CURSOR_SIZE, - CURSOR_WIDTH, attr->width, - CURSOR_HEIGHT, attr->height); - - REG_UPDATE_4(CURSOR_CONTROL, - CURSOR_MODE, attr->color_format, - CURSOR_2X_MAGNIFY, attr->attribute_flags.bits.ENABLE_MAGNIFICATION, - CURSOR_PITCH, hw_pitch, - CURSOR_LINES_PER_CHUNK, lpc); - - REG_SET_2(CURSOR_SETTINGS, 0, - /* no shift of the cursor HDL schedule */ - CURSOR0_DST_Y_OFFSET, 0, - /* used to shift the cursor chunk request deadline */ - CURSOR0_CHUNK_HDL_ADJUST, 3); + bool use_mall_for_cursor; switch (attr->color_format) { case CURSOR_MODE_MONO: @@ -158,11 +136,49 @@ void hubp32_cursor_set_attributes( cursor_size *= 8; break; } + use_mall_for_cursor = cursor_size > 16384 ? 1 : 0; + + hubp->curs_attr = *attr; - if (cursor_size > 16384) - REG_UPDATE(DCHUBP_MALL_CONFIG, USE_MALL_FOR_CURSOR, true); - else - REG_UPDATE(DCHUBP_MALL_CONFIG, USE_MALL_FOR_CURSOR, false); + if (!hubp->cursor_offload) { + REG_UPDATE(CURSOR_SURFACE_ADDRESS_HIGH, + CURSOR_SURFACE_ADDRESS_HIGH, attr->address.high_part); + REG_UPDATE(CURSOR_SURFACE_ADDRESS, + CURSOR_SURFACE_ADDRESS, attr->address.low_part); + + REG_UPDATE_2(CURSOR_SIZE, + CURSOR_WIDTH, attr->width, + CURSOR_HEIGHT, attr->height); + + REG_UPDATE_4(CURSOR_CONTROL, + CURSOR_MODE, attr->color_format, + CURSOR_2X_MAGNIFY, attr->attribute_flags.bits.ENABLE_MAGNIFICATION, + CURSOR_PITCH, hw_pitch, + CURSOR_LINES_PER_CHUNK, lpc); + + REG_SET_2(CURSOR_SETTINGS, 0, + /* no shift of the cursor HDL schedule */ + CURSOR0_DST_Y_OFFSET, 0, + /* used to shift the cursor chunk request deadline */ + CURSOR0_CHUNK_HDL_ADJUST, 3); + + REG_UPDATE(DCHUBP_MALL_CONFIG, USE_MALL_FOR_CURSOR, use_mall_for_cursor); + } + hubp->att.SURFACE_ADDR_HIGH = attr->address.high_part; + hubp->att.SURFACE_ADDR = attr->address.low_part; + hubp->att.size.bits.width = attr->width; + hubp->att.size.bits.height = attr->height; + hubp->att.cur_ctl.bits.mode = attr->color_format; + + hubp->cur_rect.w = attr->width; + hubp->cur_rect.h = attr->height; + + hubp->att.cur_ctl.bits.pitch = hw_pitch; + hubp->att.cur_ctl.bits.line_per_chunk = lpc; + hubp->att.cur_ctl.bits.cur_2x_magnify = attr->attribute_flags.bits.ENABLE_MAGNIFICATION; + hubp->att.settings.bits.dst_y_offset = 0; + hubp->att.settings.bits.chunk_hdl_adjust = 3; + hubp->use_mall_for_cursor = use_mall_for_cursor; } void hubp32_init(struct hubp *hubp) { @@ -206,9 +222,7 @@ static struct hubp_funcs dcn32_hubp_funcs = { .hubp_update_mall_sel = hubp32_update_mall_sel, .hubp_prepare_subvp_buffering = hubp32_prepare_subvp_buffering, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp32_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c index b140808f21af..79c583e258c7 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn35/dcn35_hubp.c @@ -209,6 +209,7 @@ static struct hubp_funcs dcn35_hubp_funcs = { .dmdata_load = hubp2_dmdata_load, .dmdata_status_done = hubp2_dmdata_status_done, .hubp_read_state = hubp3_read_state, + .hubp_read_reg_state = hubp3_read_reg_state, .hubp_clear_underflow = hubp2_clear_underflow, .hubp_set_flip_control_surface_gsl = hubp2_set_flip_control_surface_gsl, .hubp_init = hubp35_init, @@ -218,9 +219,6 @@ static struct hubp_funcs dcn35_hubp_funcs = { .hubp_in_blank = hubp1_in_blank, .program_extended_blank = hubp31_program_extended_blank_value, .hubp_clear_tiling = hubp3_clear_tiling, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, }; bool hubp35_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c index 0fcbc6a35be6..f01eae50d02f 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.c @@ -783,21 +783,23 @@ void hubp401_cursor_set_position( if (cur_en && REG_READ(CURSOR_SURFACE_ADDRESS) == 0) hubp->funcs->set_cursor_attributes(hubp, &hubp->curs_attr); - REG_UPDATE(CURSOR_CONTROL, - CURSOR_ENABLE, cur_en); + if (!hubp->cursor_offload) + REG_UPDATE(CURSOR_CONTROL, + CURSOR_ENABLE, cur_en); } - REG_SET_2(CURSOR_POSITION, 0, - CURSOR_X_POSITION, x_pos, - CURSOR_Y_POSITION, y_pos); + if (!hubp->cursor_offload) { + REG_SET_2(CURSOR_POSITION, 0, + CURSOR_X_POSITION, x_pos, + CURSOR_Y_POSITION, y_pos); - REG_SET_2(CURSOR_HOT_SPOT, 0, - CURSOR_HOT_SPOT_X, pos->x_hotspot, - CURSOR_HOT_SPOT_Y, pos->y_hotspot); - - REG_SET(CURSOR_DST_OFFSET, 0, - CURSOR_DST_X_OFFSET, dst_x_offset); + REG_SET_2(CURSOR_HOT_SPOT, 0, + CURSOR_HOT_SPOT_X, pos->x_hotspot, + CURSOR_HOT_SPOT_Y, pos->y_hotspot); + REG_SET(CURSOR_DST_OFFSET, 0, + CURSOR_DST_X_OFFSET, dst_x_offset); + } /* Cursor Position Register Config */ hubp->pos.cur_ctl.bits.cur_enable = cur_en; hubp->pos.position.bits.x_pos = pos->x; @@ -1071,9 +1073,7 @@ static struct hubp_funcs dcn401_hubp_funcs = { .hubp_get_3dlut_fl_done = hubp401_get_3dlut_fl_done, .hubp_clear_tiling = hubp401_clear_tiling, .hubp_program_3dlut_fl_config = hubp401_program_3dlut_fl_config, - .hubp_get_underflow_status = hubp3_get_underflow_status, - .hubp_get_current_read_line = hubp3_get_current_read_line, - .hubp_get_det_config_error = hubp31_get_det_config_error, + .hubp_read_reg_state = hubp3_read_reg_state }; bool hubp401_construct( diff --git a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h index fdabbeec8ffa..4570b8016de5 100644 --- a/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h +++ b/drivers/gpu/drm/amd/display/dc/hubp/dcn401/dcn401_hubp.h @@ -31,7 +31,7 @@ #include "dcn30/dcn30_hubp.h" #include "dcn31/dcn31_hubp.h" #include "dcn32/dcn32_hubp.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" #define HUBP_3DLUT_FL_REG_LIST_DCN401(inst)\ SRI_ARR_US(_3DLUT_FL_CONFIG, HUBP, inst),\ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index 24184b4eb352..8fe399939220 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -659,6 +659,20 @@ void dce110_update_info_frame(struct pipe_ctx *pipe_ctx) } } +static void +dce110_dac_encoder_control(struct pipe_ctx *pipe_ctx, bool enable) +{ + struct dc_link *link = pipe_ctx->stream->link; + struct dc_bios *bios = link->ctx->dc_bios; + struct bp_encoder_control encoder_control = {0}; + + encoder_control.action = enable ? ENCODER_CONTROL_ENABLE : ENCODER_CONTROL_DISABLE; + encoder_control.engine_id = link->link_enc->analog_engine; + encoder_control.pixel_clock = pipe_ctx->stream->timing.pix_clk_100hz / 10; + + bios->funcs->encoder_control(bios, &encoder_control); +} + void dce110_enable_stream(struct pipe_ctx *pipe_ctx) { enum dc_lane_count lane_count = @@ -671,7 +685,6 @@ void dce110_enable_stream(struct pipe_ctx *pipe_ctx) uint32_t early_control = 0; struct timing_generator *tg = pipe_ctx->stream_res.tg; - link_hwss->setup_stream_attribute(pipe_ctx); link_hwss->setup_stream_encoder(pipe_ctx); dc->hwss.update_info_frame(pipe_ctx); @@ -689,6 +702,9 @@ void dce110_enable_stream(struct pipe_ctx *pipe_ctx) early_control = lane_count; tg->funcs->set_early_control(tg, early_control); + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + dce110_dac_encoder_control(pipe_ctx, true); } static enum bp_result link_transmitter_control( @@ -1086,6 +1102,9 @@ void dce110_enable_audio_stream(struct pipe_ctx *pipe_ctx) if (!pipe_ctx->stream) return; + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + return; + dc = pipe_ctx->stream->ctx->dc; clk_mgr = dc->clk_mgr; link_hwss = get_link_hwss(pipe_ctx->stream->link, &pipe_ctx->link_res); @@ -1122,6 +1141,9 @@ void dce110_disable_audio_stream(struct pipe_ctx *pipe_ctx) if (!pipe_ctx || !pipe_ctx->stream) return; + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + return; + dc = pipe_ctx->stream->ctx->dc; clk_mgr = dc->clk_mgr; link_hwss = get_link_hwss(pipe_ctx->stream->link, &pipe_ctx->link_res); @@ -1196,6 +1218,9 @@ void dce110_disable_stream(struct pipe_ctx *pipe_ctx) dccg->funcs->disable_symclk_se(dccg, stream_enc->stream_enc_inst, link_enc->transmitter - TRANSMITTER_UNIPHY_A); } + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + dce110_dac_encoder_control(pipe_ctx, false); } void dce110_unblank_stream(struct pipe_ctx *pipe_ctx, @@ -1581,6 +1606,51 @@ static enum dc_status dce110_enable_stream_timing( return DC_OK; } +static void +dce110_select_crtc_source(struct pipe_ctx *pipe_ctx) +{ + struct dc_link *link = pipe_ctx->stream->link; + struct dc_bios *bios = link->ctx->dc_bios; + struct bp_crtc_source_select crtc_source_select = {0}; + enum engine_id engine_id = link->link_enc->preferred_engine; + uint8_t bit_depth; + + if (dc_is_rgb_signal(pipe_ctx->stream->signal)) + engine_id = link->link_enc->analog_engine; + + switch (pipe_ctx->stream->timing.display_color_depth) { + case COLOR_DEPTH_UNDEFINED: + bit_depth = 0; + break; + case COLOR_DEPTH_666: + bit_depth = 6; + break; + default: + case COLOR_DEPTH_888: + bit_depth = 8; + break; + case COLOR_DEPTH_101010: + bit_depth = 10; + break; + case COLOR_DEPTH_121212: + bit_depth = 12; + break; + case COLOR_DEPTH_141414: + bit_depth = 14; + break; + case COLOR_DEPTH_161616: + bit_depth = 16; + break; + } + + crtc_source_select.controller_id = CONTROLLER_ID_D0 + pipe_ctx->stream_res.tg->inst; + crtc_source_select.bit_depth = bit_depth; + crtc_source_select.engine_id = engine_id; + crtc_source_select.sink_signal = pipe_ctx->stream->signal; + + bios->funcs->select_crtc_source(bios, &crtc_source_select); +} + enum dc_status dce110_apply_single_controller_ctx_to_hw( struct pipe_ctx *pipe_ctx, struct dc_state *context, @@ -1600,6 +1670,10 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( hws->funcs.disable_stream_gating(dc, pipe_ctx); } + if (pipe_ctx->stream->signal == SIGNAL_TYPE_RGB) { + dce110_select_crtc_source(pipe_ctx); + } + if (pipe_ctx->stream_res.audio != NULL) { struct audio_output audio_output = {0}; @@ -1679,7 +1753,8 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( pipe_ctx->stream_res.tg->funcs->set_static_screen_control( pipe_ctx->stream_res.tg, event_triggers, 2); - if (!dc_is_virtual_signal(pipe_ctx->stream->signal)) + if (!dc_is_virtual_signal(pipe_ctx->stream->signal) && + !dc_is_rgb_signal(pipe_ctx->stream->signal)) pipe_ctx->stream_res.stream_enc->funcs->dig_connect_to_otg( pipe_ctx->stream_res.stream_enc, pipe_ctx->stream_res.tg->inst); @@ -1913,6 +1988,7 @@ void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context) bool can_apply_edp_fast_boot = false; bool can_apply_seamless_boot = false; bool keep_edp_vdd_on = false; + bool should_clean_dsc_block = true; struct dc_bios *dcb = dc->ctx->dc_bios; DC_LOGGER_INIT(); @@ -2005,9 +2081,15 @@ void dce110_enable_accelerated_mode(struct dc *dc, struct dc_state *context) power_down_all_hw_blocks(dc); /* DSC could be enabled on eDP during VBIOS post. - * To clean up dsc blocks if eDP is in link but not active. + * To clean up dsc blocks if all eDP dpms_off is true. */ - if (edp_link_with_sink && (edp_stream_num == 0)) + for (i = 0; i < edp_stream_num; i++) { + if (!edp_streams[i]->dpms_off) { + should_clean_dsc_block = false; + } + } + + if (should_clean_dsc_block) clean_up_dsc_blocks(dc); disable_vga_and_power_gate_all_controllers(dc); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c index e9fe97f0c4ea..fa62e40a9858 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c @@ -2245,7 +2245,7 @@ void dcn10_cursor_lock(struct dc *dc, struct pipe_ctx *pipe, bool lock) if (lock) delay_cursor_until_vupdate(dc, pipe); - if (pipe->stream && should_use_dmub_lock(pipe->stream->link)) { + if (pipe->stream && should_use_dmub_inbox1_lock(dc, pipe->stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; @@ -3090,6 +3090,9 @@ static void dcn10_update_dchubp_dpp( } if (pipe_ctx->stream->cursor_attributes.address.quad_part != 0) { + if (dc->hwss.abort_cursor_offload_update) + dc->hwss.abort_cursor_offload_update(dc, pipe_ctx); + dc->hwss.set_cursor_attribute(pipe_ctx); dc->hwss.set_cursor_position(pipe_ctx); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index 9477c9f9e196..c8ff8ae85a03 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -614,6 +614,14 @@ void dcn20_dpp_pg_control( * DOMAIN11_PGFSM_PWR_STATUS, pwr_status, * 1, 1000); */ + + /* Force disable cursor on plane powerdown on DPP 5 using dpp_force_disable_cursor */ + if (!power_on) { + struct dpp *dpp5 = hws->ctx->dc->res_pool->dpps[dpp_inst]; + if (dpp5 && dpp5->funcs->dpp_force_disable_cursor) + dpp5->funcs->dpp_force_disable_cursor(dpp5); + } + break; default: BREAK_TO_DEBUGGER(); @@ -1449,7 +1457,7 @@ void dcn20_pipe_control_lock( !flip_immediate) dcn20_setup_gsl_group_as_lock(dc, pipe, false); - if (pipe->stream && should_use_dmub_lock(pipe->stream->link)) { + if (pipe->stream && should_use_dmub_inbox1_lock(dc, pipe->stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; @@ -1793,6 +1801,9 @@ void dcn20_update_dchubp_dpp( if ((pipe_ctx->update_flags.bits.enable || pipe_ctx->update_flags.bits.opp_changed || pipe_ctx->update_flags.bits.scaler || viewport_changed == true) && pipe_ctx->stream->cursor_attributes.address.quad_part != 0) { + if (dc->hwss.abort_cursor_offload_update) + dc->hwss.abort_cursor_offload_update(dc, pipe_ctx); + dc->hwss.set_cursor_attribute(pipe_ctx); dc->hwss.set_cursor_position(pipe_ctx); @@ -3052,8 +3063,6 @@ void dcn20_enable_stream(struct pipe_ctx *pipe_ctx) link_enc->transmitter - TRANSMITTER_UNIPHY_A); } - link_hwss->setup_stream_attribute(pipe_ctx); - if (dc->res_pool->dccg->funcs->set_pixel_rate_div) dc->res_pool->dccg->funcs->set_pixel_rate_div( dc->res_pool->dccg, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index e47ed5571dfd..81bcadf5e57e 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -53,7 +53,8 @@ #include "link_service.h" #include "dc_state_priv.h" - +#define TO_DCN_DCCG(dccg)\ + container_of(dccg, struct dcn_dccg, base) #define DC_LOGGER_INIT(logger) @@ -1235,44 +1236,47 @@ void dcn30_get_underflow_debug_data(const struct dc *dc, { struct hubbub *hubbub = dc->res_pool->hubbub; - if (tg) { - uint32_t v_blank_start = 0, v_blank_end = 0; - - out_data->otg_inst = tg->inst; - - tg->funcs->get_scanoutpos(tg, - &v_blank_start, - &v_blank_end, - &out_data->h_position, - &out_data->v_position); - - out_data->otg_frame_count = tg->funcs->get_frame_count(tg); - - out_data->otg_underflow = tg->funcs->is_optc_underflow_occurred(tg); + if (hubbub) { + if (hubbub->funcs->hubbub_read_reg_state) { + hubbub->funcs->hubbub_read_reg_state(hubbub, out_data->hubbub_reg_state); + } } for (int i = 0; i < MAX_PIPES; i++) { struct hubp *hubp = dc->res_pool->hubps[i]; - - if (hubp) { - if (hubp->funcs->hubp_get_underflow_status) - out_data->hubps[i].hubp_underflow = hubp->funcs->hubp_get_underflow_status(hubp); - - if (hubp->funcs->hubp_in_blank) - out_data->hubps[i].hubp_in_blank = hubp->funcs->hubp_in_blank(hubp); - - if (hubp->funcs->hubp_get_current_read_line) - out_data->hubps[i].hubp_readline = hubp->funcs->hubp_get_current_read_line(hubp); - - if (hubp->funcs->hubp_get_det_config_error) - out_data->hubps[i].det_config_error = hubp->funcs->hubp_get_det_config_error(hubp); - } + struct dpp *dpp = dc->res_pool->dpps[i]; + struct output_pixel_processor *opp = dc->res_pool->opps[i]; + struct display_stream_compressor *dsc = dc->res_pool->dscs[i]; + struct mpc *mpc = dc->res_pool->mpc; + struct timing_generator *optc = dc->res_pool->timing_generators[i]; + struct dccg *dccg = dc->res_pool->dccg; + + if (hubp) + if (hubp->funcs->hubp_read_reg_state) + hubp->funcs->hubp_read_reg_state(hubp, out_data->hubp_reg_state[i]); + + if (dpp) + if (dpp->funcs->dpp_read_reg_state) + dpp->funcs->dpp_read_reg_state(dpp, out_data->dpp_reg_state[i]); + + if (opp) + if (opp->funcs->opp_read_reg_state) + opp->funcs->opp_read_reg_state(opp, out_data->opp_reg_state[i]); + + if (dsc) + if (dsc->funcs->dsc_read_reg_state) + dsc->funcs->dsc_read_reg_state(dsc, out_data->dsc_reg_state[i]); + + if (mpc) + if (mpc->funcs->mpc_read_reg_state) + mpc->funcs->mpc_read_reg_state(mpc, i, out_data->mpc_reg_state[i]); + + if (optc) + if (optc->funcs->optc_read_reg_state) + optc->funcs->optc_read_reg_state(optc, out_data->optc_reg_state[i]); + + if (dccg) + if (dccg->funcs->dccg_read_reg_state) + dccg->funcs->dccg_read_reg_state(dccg, out_data->dccg_reg_state[i]); } - - if (hubbub->funcs->get_det_sizes) - hubbub->funcs->get_det_sizes(hubbub, out_data->curr_det_sizes, out_data->target_det_sizes); - - if (hubbub->funcs->compbuf_config_error) - out_data->compbuf_config_error = hubbub->funcs->compbuf_config_error(hubbub); - } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c index b822f2dffff0..d1ecdb92b072 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c @@ -710,7 +710,8 @@ bool dcn31_set_backlight_level(struct pipe_ctx *pipe_ctx, panel_cntl->inst, panel_cntl->pwrseq_inst); - dmub_abm_set_backlight(dc, backlight_level_params, panel_cntl->inst); + if (backlight_level_params->control_type != BACKLIGHT_CONTROL_AMD_AUX) + dmub_abm_set_backlight(dc, backlight_level_params, panel_cntl->inst); return true; } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c index f925f669f2a4..4ee6ed610de0 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn314/dcn314_hwseq.c @@ -108,6 +108,7 @@ static void update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; ASSERT(dsc_cfg.dc_dsc_cfg.num_slices_h % opp_cnt == 0); dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; dsc->funcs->dsc_set_config(dsc, &dsc_cfg, &dsc_optc_cfg); dsc->funcs->dsc_enable(dsc, pipe_ctx->stream_res.opp->inst); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c index f39292952702..bf19ba65d09a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_hwseq.c @@ -1061,6 +1061,7 @@ void dcn32_update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; ASSERT(dsc_cfg.dc_dsc_cfg.num_slices_h % opp_cnt == 0); dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; if (should_use_dto_dscclk) dccg->funcs->set_dto_dscclk(dccg, dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c index 05011061822c..7aa0f452e8f7 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c @@ -364,6 +364,7 @@ static void update_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; ASSERT(dsc_cfg.dc_dsc_cfg.num_slices_h % opp_cnt == 0); dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; dsc->funcs->dsc_set_config(dsc, &dsc_cfg, &dsc_optc_cfg); dsc->funcs->dsc_enable(dsc, pipe_ctx->stream_res.opp->inst); @@ -816,8 +817,6 @@ void dcn35_enable_plane(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context) { struct dpp *dpp = pipe_ctx->plane_res.dpp; - struct dccg *dccg = dc->res_pool->dccg; - /* enable DCFCLK current DCHUB */ pipe_ctx->plane_res.hubp->funcs->hubp_clk_cntl(pipe_ctx->plane_res.hubp, true); @@ -825,7 +824,6 @@ void dcn35_enable_plane(struct dc *dc, struct pipe_ctx *pipe_ctx, /* initialize HUBP on power up */ pipe_ctx->plane_res.hubp->funcs->hubp_init(pipe_ctx->plane_res.hubp); /*make sure DPPCLK is on*/ - dccg->funcs->dccg_root_gate_disable_control(dccg, dpp->inst, true); dpp->funcs->dpp_dppclk_control(dpp, false, true); /* make sure OPP_PIPE_CLOCK_EN = 1 */ pipe_ctx->stream_res.opp->funcs->opp_pipe_clock_control( @@ -859,7 +857,6 @@ void dcn35_plane_atomic_disable(struct dc *dc, struct pipe_ctx *pipe_ctx) { struct hubp *hubp = pipe_ctx->plane_res.hubp; struct dpp *dpp = pipe_ctx->plane_res.dpp; - struct dccg *dccg = dc->res_pool->dccg; dc->hwss.wait_for_mpcc_disconnect(dc, dc->res_pool, pipe_ctx); @@ -878,7 +875,6 @@ void dcn35_plane_atomic_disable(struct dc *dc, struct pipe_ctx *pipe_ctx) hubp->funcs->hubp_clk_cntl(hubp, false); dpp->funcs->dpp_dppclk_control(dpp, false, false); - dccg->funcs->dccg_root_gate_disable_control(dccg, dpp->inst, false); hubp->power_gated = true; @@ -1592,3 +1588,141 @@ void dcn35_hardware_release(struct dc *dc) if (dc->hwss.hw_block_power_up) dc->hwss.hw_block_power_up(dc, &pg_update_state); } + +void dcn35_abort_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe) +{ + if (!dc_dmub_srv_is_cursor_offload_enabled(dc)) + return; + + /* + * Insert a blank update to modify the write index and set pipe_mask to 0. + * + * While the DMU is interlocked with driver full pipe programming via + * the DMU HW lock, if the cursor update begins to execute after a full + * pipe programming occurs there are two possible issues: + * + * 1. Outdated cursor information is programmed, replacing the current update + * 2. The cursor update in firmware holds the cursor lock, preventing + * the current update from being latched atomically in the same frame + * as the rest of the update. + * + * This blank update, treated as a no-op, will allow the firmware to skip + * the programming. + */ + + if (dc->hwss.begin_cursor_offload_update) + dc->hwss.begin_cursor_offload_update(dc, pipe); + + if (dc->hwss.commit_cursor_offload_update) + dc->hwss.commit_cursor_offload_update(dc, pipe); +} + +void dcn35_begin_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe) +{ + volatile struct dmub_cursor_offload_v1 *cs = dc->ctx->dmub_srv->dmub->cursor_offload_v1; + const struct pipe_ctx *top_pipe = resource_get_otg_master(pipe); + uint32_t stream_idx, write_idx, payload_idx; + + if (!top_pipe) + return; + + stream_idx = top_pipe->pipe_idx; + write_idx = cs->offload_streams[stream_idx].write_idx + 1; /* new payload (+1) */ + payload_idx = write_idx % ARRAY_SIZE(cs->offload_streams[stream_idx].payloads); + + cs->offload_streams[stream_idx].payloads[payload_idx].write_idx_start = write_idx; + + if (pipe->plane_res.hubp) + pipe->plane_res.hubp->cursor_offload = true; + + if (pipe->plane_res.dpp) + pipe->plane_res.dpp->cursor_offload = true; +} + +void dcn35_commit_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe) +{ + volatile struct dmub_cursor_offload_v1 *cs = dc->ctx->dmub_srv->dmub->cursor_offload_v1; + volatile struct dmub_shared_state_cursor_offload_stream_v1 *shared_stream; + const struct pipe_ctx *top_pipe = resource_get_otg_master(pipe); + uint32_t stream_idx, write_idx, payload_idx; + + if (pipe->plane_res.hubp) + pipe->plane_res.hubp->cursor_offload = false; + + if (pipe->plane_res.dpp) + pipe->plane_res.dpp->cursor_offload = false; + + if (!top_pipe) + return; + + stream_idx = top_pipe->pipe_idx; + write_idx = cs->offload_streams[stream_idx].write_idx + 1; /* new payload (+1) */ + payload_idx = write_idx % ARRAY_SIZE(cs->offload_streams[stream_idx].payloads); + + shared_stream = &dc->ctx->dmub_srv->dmub->shared_state[DMUB_SHARED_STATE_FEATURE__CURSOR_OFFLOAD_V1] + .data.cursor_offload_v1.offload_streams[stream_idx]; + + shared_stream->last_write_idx = write_idx; + + cs->offload_streams[stream_idx].write_idx = write_idx; + cs->offload_streams[stream_idx].payloads[payload_idx].write_idx_finish = write_idx; +} + +void dcn35_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pipe) +{ + volatile struct dmub_cursor_offload_v1 *cs = dc->ctx->dmub_srv->dmub->cursor_offload_v1; + const struct pipe_ctx *top_pipe = resource_get_otg_master(pipe); + const struct hubp *hubp = pipe->plane_res.hubp; + const struct dpp *dpp = pipe->plane_res.dpp; + volatile struct dmub_cursor_offload_pipe_data_dcn30_v1 *p; + uint32_t stream_idx, write_idx, payload_idx; + + if (!top_pipe || !hubp || !dpp) + return; + + stream_idx = top_pipe->pipe_idx; + write_idx = cs->offload_streams[stream_idx].write_idx + 1; /* new payload (+1) */ + payload_idx = write_idx % ARRAY_SIZE(cs->offload_streams[stream_idx].payloads); + + p = &cs->offload_streams[stream_idx].payloads[payload_idx].pipe_data[pipe->pipe_idx].dcn30; + + p->CURSOR0_0_CURSOR_SURFACE_ADDRESS = hubp->att.SURFACE_ADDR; + p->CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH = hubp->att.SURFACE_ADDR_HIGH; + p->CURSOR0_0_CURSOR_SIZE__CURSOR_WIDTH = hubp->att.size.bits.width; + p->CURSOR0_0_CURSOR_SIZE__CURSOR_HEIGHT = hubp->att.size.bits.height; + p->CURSOR0_0_CURSOR_POSITION__CURSOR_X_POSITION = hubp->pos.position.bits.x_pos; + p->CURSOR0_0_CURSOR_POSITION__CURSOR_Y_POSITION = hubp->pos.position.bits.y_pos; + p->CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_X = hubp->pos.hot_spot.bits.x_hot; + p->CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_Y = hubp->pos.hot_spot.bits.y_hot; + p->CURSOR0_0_CURSOR_DST_OFFSET__CURSOR_DST_X_OFFSET = hubp->pos.dst_offset.bits.dst_x_offset; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_ENABLE = hubp->pos.cur_ctl.bits.cur_enable; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE = hubp->att.cur_ctl.bits.mode; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY = hubp->pos.cur_ctl.bits.cur_2x_magnify; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH = hubp->att.cur_ctl.bits.pitch; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->att.cur_ctl.bits.line_per_chunk; + + p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_ENABLE = dpp->att.cur0_ctl.bits.cur0_enable; + p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_MODE = dpp->att.cur0_ctl.bits.mode; + p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_EXPANSION_MODE = dpp->att.cur0_ctl.bits.expansion_mode; + p->CNVC_CUR0_CURSOR0_CONTROL__CUR0_ROM_EN = dpp->att.cur0_ctl.bits.cur0_rom_en; + p->CNVC_CUR0_CURSOR0_COLOR0__CUR0_COLOR0 = 0x000000; + p->CNVC_CUR0_CURSOR0_COLOR1__CUR0_COLOR1 = 0xFFFFFF; + p->CNVC_CUR0_CURSOR0_FP_SCALE_BIAS__CUR0_FP_BIAS = dpp->att.fp_scale_bias.bits.fp_bias; + p->CNVC_CUR0_CURSOR0_FP_SCALE_BIAS__CUR0_FP_SCALE = dpp->att.fp_scale_bias.bits.fp_scale; + + p->HUBPREQ0_CURSOR_SETTINGS__CURSOR0_DST_Y_OFFSET = hubp->att.settings.bits.dst_y_offset; + p->HUBPREQ0_CURSOR_SETTINGS__CURSOR0_CHUNK_HDL_ADJUST = hubp->att.settings.bits.chunk_hdl_adjust; + + cs->offload_streams[stream_idx].payloads[payload_idx].pipe_mask |= (1u << pipe->pipe_idx); +} + +void dcn35_notify_cursor_offload_drr_update(struct dc *dc, struct dc_state *context, + const struct dc_stream_state *stream) +{ + dc_dmub_srv_control_cursor_offload(dc, context, stream, true); +} + +void dcn35_program_cursor_offload_now(struct dc *dc, const struct pipe_ctx *pipe) +{ + dc_dmub_srv_program_cursor_now(dc, pipe); +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.h index 0b1d6f608edd..1ff41dba556c 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.h @@ -101,4 +101,12 @@ bool dcn35_is_dp_dig_pixel_rate_div_policy(struct pipe_ctx *pipe_ctx); void dcn35_hardware_release(struct dc *dc); +void dcn35_abort_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe); +void dcn35_begin_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe); +void dcn35_commit_cursor_offload_update(struct dc *dc, const struct pipe_ctx *pipe); +void dcn35_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pipe); +void dcn35_notify_cursor_offload_drr_update(struct dc *dc, struct dc_state *context, + const struct dc_stream_state *stream); +void dcn35_program_cursor_offload_now(struct dc *dc, const struct pipe_ctx *pipe); + #endif /* __DC_HWSS_DCN35_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_init.c index f2f16a0bdb4f..5a66c9db2670 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_init.c @@ -86,6 +86,12 @@ static const struct hw_sequencer_funcs dcn35_funcs = { .set_cursor_position = dcn10_set_cursor_position, .set_cursor_attribute = dcn10_set_cursor_attribute, .set_cursor_sdr_white_level = dcn10_set_cursor_sdr_white_level, + .abort_cursor_offload_update = dcn35_abort_cursor_offload_update, + .begin_cursor_offload_update = dcn35_begin_cursor_offload_update, + .commit_cursor_offload_update = dcn35_commit_cursor_offload_update, + .update_cursor_offload_pipe = dcn35_update_cursor_offload_pipe, + .notify_cursor_offload_drr_update = dcn35_notify_cursor_offload_drr_update, + .program_cursor_offload_now = dcn35_program_cursor_offload_now, .setup_periodic_interrupt = dcn10_setup_periodic_interrupt, .set_clock = dcn10_set_clock, .get_clock = dcn10_get_clock, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 7c276c319086..2fbc22afb89c 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -26,9 +26,11 @@ #include "clk_mgr.h" #include "dsc.h" #include "link_service.h" +#include "custom_float.h" #include "dce/dmub_hw_lock_mgr.h" #include "dcn10/dcn10_cm_common.h" +#include "dcn10/dcn10_hubbub.h" #include "dcn20/dcn20_optc.h" #include "dcn30/dcn30_cm_common.h" #include "dcn32/dcn32_hwseq.h" @@ -36,6 +38,7 @@ #include "dcn401/dcn401_resource.h" #include "dc_state_priv.h" #include "link_enc_cfg.h" +#include "../hw_sequencer.h" #define DC_LOGGER_INIT(logger) @@ -200,6 +203,9 @@ void dcn401_init_hw(struct dc *dc) */ struct dc_link *link = dc->links[i]; + if (link->ep_type != DISPLAY_ENDPOINT_PHY) + continue; + link->link_enc->funcs->hw_init(link->link_enc); /* Check for enabled DIG to identify enabled display */ @@ -968,8 +974,6 @@ void dcn401_enable_stream(struct pipe_ctx *pipe_ctx) } } - link_hwss->setup_stream_attribute(pipe_ctx); - if (dc->res_pool->dccg->funcs->set_pixel_rate_div) { dc->res_pool->dccg->funcs->set_pixel_rate_div( dc->res_pool->dccg, @@ -1404,9 +1408,9 @@ void dcn401_prepare_bandwidth(struct dc *dc, } if (dc->debug.fams2_config.bits.enable) { - dcn401_fams2_global_control_lock(dc, context, true); + dcn401_dmub_hw_control_lock(dc, context, true); dcn401_fams2_update_config(dc, context, false); - dcn401_fams2_global_control_lock(dc, context, false); + dcn401_dmub_hw_control_lock(dc, context, false); } if (p_state_change_support != context->bw_ctx.bw.dcn.clk.p_state_change_support) { @@ -1425,9 +1429,9 @@ void dcn401_optimize_bandwidth( /* enable fams2 if needed */ if (dc->debug.fams2_config.bits.enable) { - dcn401_fams2_global_control_lock(dc, context, true); + dcn401_dmub_hw_control_lock(dc, context, true); dcn401_fams2_update_config(dc, context, true); - dcn401_fams2_global_control_lock(dc, context, false); + dcn401_dmub_hw_control_lock(dc, context, false); } /* program dchubbub watermarks */ @@ -1466,14 +1470,17 @@ void dcn401_optimize_bandwidth( } } -void dcn401_fams2_global_control_lock(struct dc *dc, +void dcn401_dmub_hw_control_lock(struct dc *dc, struct dc_state *context, bool lock) { /* use always for now */ union dmub_inbox0_cmd_lock_hw hw_lock_cmd = { 0 }; - if (!dc->ctx || !dc->ctx->dmub_srv || !dc->debug.fams2_config.bits.enable) + if (!dc->ctx || !dc->ctx->dmub_srv) + return; + + if (!dc->debug.fams2_config.bits.enable && !dc_dmub_srv_is_cursor_offload_enabled(dc)) return; hw_lock_cmd.bits.command_code = DMUB_INBOX0_CMD__HW_LOCK; @@ -1483,12 +1490,12 @@ void dcn401_fams2_global_control_lock(struct dc *dc, dmub_hw_lock_mgr_inbox0_cmd(dc->ctx->dmub_srv, hw_lock_cmd); } -void dcn401_fams2_global_control_lock_fast(union block_sequence_params *params) +void dcn401_dmub_hw_control_lock_fast(union block_sequence_params *params) { - struct dc *dc = params->fams2_global_control_lock_fast_params.dc; - bool lock = params->fams2_global_control_lock_fast_params.lock; + struct dc *dc = params->dmub_hw_control_lock_fast_params.dc; + bool lock = params->dmub_hw_control_lock_fast_params.lock; - if (params->fams2_global_control_lock_fast_params.is_required) { + if (params->dmub_hw_control_lock_fast_params.is_required) { union dmub_inbox0_cmd_lock_hw hw_lock_cmd = { 0 }; hw_lock_cmd.bits.command_code = DMUB_INBOX0_CMD__HW_LOCK; @@ -1595,6 +1602,143 @@ void dcn401_update_odm(struct dc *dc, struct dc_state *context, dc->hwseq->funcs.blank_pixel_data(dc, otg_master, true); } +static void dcn401_add_dsc_sequence_for_odm_change(struct dc *dc, struct dc_state *context, + struct pipe_ctx *otg_master, struct block_sequence_state *seq_state) +{ + struct pipe_ctx *old_pipe; + struct pipe_ctx *new_pipe; + struct pipe_ctx *old_opp_heads[MAX_PIPES]; + struct pipe_ctx *old_otg_master; + int old_opp_head_count = 0; + int i; + + old_otg_master = &dc->current_state->res_ctx.pipe_ctx[otg_master->pipe_idx]; + + if (resource_is_pipe_type(old_otg_master, OTG_MASTER)) { + old_opp_head_count = resource_get_opp_heads_for_otg_master(old_otg_master, + &dc->current_state->res_ctx, + old_opp_heads); + } else { + old_otg_master = NULL; + } + + /* Process new DSC configuration if DSC is enabled */ + if (otg_master->stream_res.dsc && otg_master->stream->timing.flags.DSC) { + struct dc_stream_state *stream = otg_master->stream; + struct pipe_ctx *odm_pipe; + int opp_cnt = 1; + int last_dsc_calc = 0; + bool should_use_dto_dscclk = (dc->res_pool->dccg->funcs->set_dto_dscclk != NULL) && + stream->timing.pix_clk_100hz > 480000; + + /* Count ODM pipes */ + for (odm_pipe = otg_master->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) + opp_cnt++; + + int num_slices_h = stream->timing.dsc_cfg.num_slices_h / opp_cnt; + + /* Step 1: Set DTO DSCCLK for main DSC if needed */ + if (should_use_dto_dscclk) { + hwss_add_dccg_set_dto_dscclk(seq_state, dc->res_pool->dccg, + otg_master->stream_res.dsc->inst, num_slices_h); + } + + /* Step 2: Calculate and set DSC config for main DSC */ + last_dsc_calc = *seq_state->num_steps; + hwss_add_dsc_calculate_and_set_config(seq_state, otg_master, true, opp_cnt); + + /* Step 3: Enable main DSC block */ + hwss_add_dsc_enable_with_opp(seq_state, otg_master); + + /* Step 4: Configure and enable ODM DSC blocks */ + for (odm_pipe = otg_master->next_odm_pipe; odm_pipe; odm_pipe = odm_pipe->next_odm_pipe) { + if (!odm_pipe->stream_res.dsc) + continue; + + /* Set DTO DSCCLK for ODM DSC if needed */ + if (should_use_dto_dscclk) { + hwss_add_dccg_set_dto_dscclk(seq_state, dc->res_pool->dccg, + odm_pipe->stream_res.dsc->inst, num_slices_h); + } + + /* Calculate and set DSC config for ODM DSC */ + last_dsc_calc = *seq_state->num_steps; + hwss_add_dsc_calculate_and_set_config(seq_state, odm_pipe, true, opp_cnt); + + /* Enable ODM DSC block */ + hwss_add_dsc_enable_with_opp(seq_state, odm_pipe); + } + + /* Step 5: Configure DSC in timing generator */ + hwss_add_tg_set_dsc_config(seq_state, otg_master->stream_res.tg, + &seq_state->steps[last_dsc_calc].params.dsc_calculate_and_set_config_params.dsc_optc_cfg, true); + } else if (otg_master->stream_res.dsc && !otg_master->stream->timing.flags.DSC) { + /* Disable DSC in OPTC */ + hwss_add_tg_set_dsc_config(seq_state, otg_master->stream_res.tg, NULL, false); + + hwss_add_dsc_disconnect(seq_state, otg_master->stream_res.dsc); + } + + /* Disable DSC for old pipes that no longer need it */ + if (old_otg_master && old_otg_master->stream_res.dsc) { + for (i = 0; i < old_opp_head_count; i++) { + old_pipe = old_opp_heads[i]; + new_pipe = &context->res_ctx.pipe_ctx[old_pipe->pipe_idx]; + + /* If old pipe had DSC but new pipe doesn't, disable the old DSC */ + if (old_pipe->stream_res.dsc && !new_pipe->stream_res.dsc) { + /* Then disconnect DSC block */ + hwss_add_dsc_disconnect(seq_state, old_pipe->stream_res.dsc); + } + } + } +} + +void dcn401_update_odm_sequence(struct dc *dc, struct dc_state *context, + struct pipe_ctx *otg_master, struct block_sequence_state *seq_state) +{ + struct pipe_ctx *opp_heads[MAX_PIPES]; + int opp_inst[MAX_PIPES] = {0}; + int opp_head_count; + int odm_slice_width = resource_get_odm_slice_dst_width(otg_master, false); + int last_odm_slice_width = resource_get_odm_slice_dst_width(otg_master, true); + int i; + + opp_head_count = resource_get_opp_heads_for_otg_master( + otg_master, &context->res_ctx, opp_heads); + + for (i = 0; i < opp_head_count; i++) + opp_inst[i] = opp_heads[i]->stream_res.opp->inst; + + /* Add ODM combine/bypass operation to sequence */ + if (opp_head_count > 1) { + hwss_add_optc_set_odm_combine(seq_state, otg_master->stream_res.tg, opp_inst, + opp_head_count, odm_slice_width, last_odm_slice_width); + } else { + hwss_add_optc_set_odm_bypass(seq_state, otg_master->stream_res.tg, &otg_master->stream->timing); + } + + /* Add OPP operations to sequence */ + for (i = 0; i < opp_head_count; i++) { + /* Add OPP pipe clock control operation */ + hwss_add_opp_pipe_clock_control(seq_state, opp_heads[i]->stream_res.opp, true); + + /* Add OPP program left edge extra pixel operation */ + hwss_add_opp_program_left_edge_extra_pixel(seq_state, opp_heads[i]->stream_res.opp, + opp_heads[i]->stream->timing.pixel_encoding, resource_is_pipe_type(opp_heads[i], OTG_MASTER)); + } + + /* Add DSC update operations to sequence */ + dcn401_add_dsc_sequence_for_odm_change(dc, context, otg_master, seq_state); + + /* Add blank pixel data operation if needed */ + if (!resource_is_pipe_type(otg_master, DPP_PIPE)) { + if (dc->hwseq->funcs.blank_pixel_data_sequence) + dc->hwseq->funcs.blank_pixel_data_sequence( + dc, otg_master, true, seq_state); + } +} + void dcn401_unblank_stream(struct pipe_ctx *pipe_ctx, struct dc_link_settings *link_settings) { @@ -2083,6 +2227,157 @@ void dcn401_program_pipe( } } +/* + * dcn401_program_pipe_sequence - Sequence-based version of dcn401_program_pipe + * + * This function creates a sequence-based version of the original dcn401_program_pipe + * function. Instead of directly calling hardware programming functions, it appends + * sequence steps to the provided block_sequence array that can later be executed + * as part of hwss_execute_sequence. + * + */ +void dcn401_program_pipe_sequence( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state) +{ + struct dce_hwseq *hws = dc->hwseq; + + /* Only need to unblank on top pipe */ + if (resource_is_pipe_type(pipe_ctx, OTG_MASTER)) { + if (pipe_ctx->update_flags.bits.enable || + pipe_ctx->update_flags.bits.odm || + pipe_ctx->stream->update_flags.bits.abm_level) { + if (dc->hwseq->funcs.blank_pixel_data_sequence) + dc->hwseq->funcs.blank_pixel_data_sequence(dc, pipe_ctx, + !pipe_ctx->plane_state || !pipe_ctx->plane_state->visible, + seq_state); + } + } + + /* Only update TG on top pipe */ + if (pipe_ctx->update_flags.bits.global_sync && !pipe_ctx->top_pipe + && !pipe_ctx->prev_odm_pipe) { + + /* Step 1: Program global sync */ + hwss_add_tg_program_global_sync(seq_state, pipe_ctx->stream_res.tg, + dcn401_calculate_vready_offset_for_group(pipe_ctx), + (unsigned int)pipe_ctx->global_sync.dcn4x.vstartup_lines, + (unsigned int)pipe_ctx->global_sync.dcn4x.vupdate_offset_pixels, + (unsigned int)pipe_ctx->global_sync.dcn4x.vupdate_vupdate_width_pixels, + (unsigned int)pipe_ctx->global_sync.dcn4x.pstate_keepout_start_lines); + + /* Step 2: Wait for VACTIVE state (if not phantom pipe) */ + if (dc_state_get_pipe_subvp_type(context, pipe_ctx) != SUBVP_PHANTOM) + hwss_add_tg_wait_for_state(seq_state, pipe_ctx->stream_res.tg, CRTC_STATE_VACTIVE); + + /* Step 3: Set VTG params */ + hwss_add_tg_set_vtg_params(seq_state, pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing, true); + + /* Step 4: Setup vupdate interrupt (if available) */ + if (hws->funcs.setup_vupdate_interrupt) + dcn401_setup_vupdate_interrupt_sequence(dc, pipe_ctx, seq_state); + } + + if (pipe_ctx->update_flags.bits.odm) { + if (hws->funcs.update_odm_sequence) + hws->funcs.update_odm_sequence(dc, context, pipe_ctx, seq_state); + } + + if (pipe_ctx->update_flags.bits.enable) { + if (dc->hwss.enable_plane_sequence) + dc->hwss.enable_plane_sequence(dc, pipe_ctx, context, seq_state); + } + + if (pipe_ctx->update_flags.bits.det_size) { + if (dc->res_pool->hubbub->funcs->program_det_size) { + hwss_add_hubp_program_det_size(seq_state, dc->res_pool->hubbub, + pipe_ctx->plane_res.hubp->inst, pipe_ctx->det_buffer_size_kb); + } + + if (dc->res_pool->hubbub->funcs->program_det_segments) { + hwss_add_hubp_program_det_segments(seq_state, dc->res_pool->hubbub, + pipe_ctx->plane_res.hubp->inst, pipe_ctx->hubp_regs.det_size); + } + } + + if (pipe_ctx->plane_state && (pipe_ctx->update_flags.raw || + pipe_ctx->plane_state->update_flags.raw || + pipe_ctx->stream->update_flags.raw)) { + + if (dc->hwss.update_dchubp_dpp_sequence) + dc->hwss.update_dchubp_dpp_sequence(dc, pipe_ctx, context, seq_state); + } + + if (pipe_ctx->plane_state && (pipe_ctx->update_flags.bits.enable || + pipe_ctx->plane_state->update_flags.bits.hdr_mult)) { + + hws->funcs.set_hdr_multiplier_sequence(pipe_ctx, seq_state); + } + + if (pipe_ctx->plane_state && + (pipe_ctx->plane_state->update_flags.bits.in_transfer_func_change || + pipe_ctx->plane_state->update_flags.bits.gamma_change || + pipe_ctx->plane_state->update_flags.bits.lut_3d || + pipe_ctx->update_flags.bits.enable)) { + + hwss_add_dpp_set_input_transfer_func(seq_state, dc, pipe_ctx, pipe_ctx->plane_state); + } + + /* dcn10_translate_regamma_to_hw_format takes 750us to finish + * only do gamma programming for powering on, internal memcmp to avoid + * updating on slave planes + */ + if (pipe_ctx->update_flags.bits.enable || + pipe_ctx->update_flags.bits.plane_changed || + pipe_ctx->stream->update_flags.bits.out_tf) { + hwss_add_dpp_set_output_transfer_func(seq_state, dc, pipe_ctx, pipe_ctx->stream); + } + + /* If the pipe has been enabled or has a different opp, we + * should reprogram the fmt. This deals with cases where + * interation between mpc and odm combine on different streams + * causes a different pipe to be chosen to odm combine with. + */ + if (pipe_ctx->update_flags.bits.enable + || pipe_ctx->update_flags.bits.opp_changed) { + + hwss_add_opp_set_dyn_expansion(seq_state, pipe_ctx->stream_res.opp, COLOR_SPACE_YCBCR601, + pipe_ctx->stream->timing.display_color_depth, pipe_ctx->stream->signal); + + hwss_add_opp_program_fmt(seq_state, pipe_ctx->stream_res.opp, + &pipe_ctx->stream->bit_depth_params, &pipe_ctx->stream->clamping); + } + + /* Set ABM pipe after other pipe configurations done */ + if ((pipe_ctx->plane_state && pipe_ctx->plane_state->visible)) { + if (pipe_ctx->stream_res.abm) { + hwss_add_abm_set_pipe(seq_state, dc, pipe_ctx); + + hwss_add_abm_set_level(seq_state, pipe_ctx->stream_res.abm, pipe_ctx->stream->abm_level); + } + } + + if (pipe_ctx->update_flags.bits.test_pattern_changed) { + struct output_pixel_processor *odm_opp = pipe_ctx->stream_res.opp; + + hwss_add_opp_program_bit_depth_reduction(seq_state, odm_opp, true, pipe_ctx); + + hwss_add_opp_set_disp_pattern_generator(seq_state, + odm_opp, + pipe_ctx->stream_res.test_pattern_params.test_pattern, + pipe_ctx->stream_res.test_pattern_params.color_space, + pipe_ctx->stream_res.test_pattern_params.color_depth, + (struct tg_color){0}, + false, + pipe_ctx->stream_res.test_pattern_params.width, + pipe_ctx->stream_res.test_pattern_params.height, + pipe_ctx->stream_res.test_pattern_params.offset); + } + +} + void dcn401_program_front_end_for_ctx( struct dc *dc, struct dc_state *context) @@ -2160,7 +2455,6 @@ void dcn401_program_front_end_for_ctx( && context->res_ctx.pipe_ctx[i].stream) hws->funcs.blank_pixel_data(dc, &context->res_ctx.pipe_ctx[i], true); - /* Disconnect mpcc */ for (i = 0; i < dc->res_pool->pipe_count; i++) if (context->res_ctx.pipe_ctx[i].update_flags.bits.disable @@ -2239,11 +2533,11 @@ void dcn401_program_front_end_for_ctx( /* Avoid underflow by check of pipe line read when adding 2nd plane. */ if (hws->wa.wait_hubpret_read_start_during_mpo_transition && - !pipe->top_pipe && - pipe->stream && - pipe->plane_res.hubp->funcs->hubp_wait_pipe_read_start && - dc->current_state->stream_status[0].plane_count == 1 && - context->stream_status[0].plane_count > 1) { + !pipe->top_pipe && + pipe->stream && + pipe->plane_res.hubp->funcs->hubp_wait_pipe_read_start && + dc->current_state->stream_status[0].plane_count == 1 && + context->stream_status[0].plane_count > 1) { pipe->plane_res.hubp->funcs->hubp_wait_pipe_read_start(pipe->plane_res.hubp); } } @@ -2355,7 +2649,6 @@ void dcn401_post_unlock_program_front_end( */ if (hwseq->funcs.update_force_pstate) dc->hwseq->funcs.update_force_pstate(dc, context); - /* Only program the MALL registers after all the main and phantom pipes * are done programming. */ @@ -2669,3 +2962,1084 @@ void dcn401_plane_atomic_power_down(struct dc *dc, if (hws->funcs.dpp_root_clock_control) hws->funcs.dpp_root_clock_control(hws, dpp->inst, false); } + +void dcn401_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pipe) +{ + volatile struct dmub_cursor_offload_v1 *cs = dc->ctx->dmub_srv->dmub->cursor_offload_v1; + const struct pipe_ctx *top_pipe = resource_get_otg_master(pipe); + const struct hubp *hubp = pipe->plane_res.hubp; + const struct dpp *dpp = pipe->plane_res.dpp; + volatile struct dmub_cursor_offload_pipe_data_dcn401_v1 *p; + uint32_t stream_idx, write_idx, payload_idx; + + if (!top_pipe || !hubp || !dpp) + return; + + stream_idx = top_pipe->pipe_idx; + write_idx = cs->offload_streams[stream_idx].write_idx + 1; /* new payload (+1) */ + payload_idx = write_idx % ARRAY_SIZE(cs->offload_streams[stream_idx].payloads); + + p = &cs->offload_streams[stream_idx].payloads[payload_idx].pipe_data[pipe->pipe_idx].dcn401; + + p->CURSOR0_0_CURSOR_SURFACE_ADDRESS = hubp->att.SURFACE_ADDR; + p->CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH = hubp->att.SURFACE_ADDR_HIGH; + p->CURSOR0_0_CURSOR_SIZE__CURSOR_WIDTH = hubp->att.size.bits.width; + p->CURSOR0_0_CURSOR_SIZE__CURSOR_HEIGHT = hubp->att.size.bits.height; + p->CURSOR0_0_CURSOR_POSITION__CURSOR_X_POSITION = hubp->pos.position.bits.x_pos; + p->CURSOR0_0_CURSOR_POSITION__CURSOR_Y_POSITION = hubp->pos.position.bits.y_pos; + p->CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_X = hubp->pos.hot_spot.bits.x_hot; + p->CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_Y = hubp->pos.hot_spot.bits.y_hot; + p->CURSOR0_0_CURSOR_DST_OFFSET__CURSOR_DST_X_OFFSET = hubp->pos.dst_offset.bits.dst_x_offset; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_ENABLE = hubp->pos.cur_ctl.bits.cur_enable; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE = hubp->att.cur_ctl.bits.mode; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY = hubp->pos.cur_ctl.bits.cur_2x_magnify; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH = hubp->att.cur_ctl.bits.pitch; + p->CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK = hubp->att.cur_ctl.bits.line_per_chunk; + + p->CM_CUR0_CURSOR0_CONTROL__CUR0_ENABLE = dpp->att.cur0_ctl.bits.cur0_enable; + p->CM_CUR0_CURSOR0_CONTROL__CUR0_MODE = dpp->att.cur0_ctl.bits.mode; + p->CM_CUR0_CURSOR0_CONTROL__CUR0_EXPANSION_MODE = dpp->att.cur0_ctl.bits.expansion_mode; + p->CM_CUR0_CURSOR0_CONTROL__CUR0_ROM_EN = dpp->att.cur0_ctl.bits.cur0_rom_en; + p->CM_CUR0_CURSOR0_COLOR0__CUR0_COLOR0 = 0x000000; + p->CM_CUR0_CURSOR0_COLOR1__CUR0_COLOR1 = 0xFFFFFF; + + p->CM_CUR0_CURSOR0_FP_SCALE_BIAS_G_Y__CUR0_FP_BIAS_G_Y = + dpp->att.fp_scale_bias_g_y.bits.fp_bias_g_y; + p->CM_CUR0_CURSOR0_FP_SCALE_BIAS_G_Y__CUR0_FP_SCALE_G_Y = + dpp->att.fp_scale_bias_g_y.bits.fp_scale_g_y; + p->CM_CUR0_CURSOR0_FP_SCALE_BIAS_RB_CRCB__CUR0_FP_BIAS_RB_CRCB = + dpp->att.fp_scale_bias_rb_crcb.bits.fp_bias_rb_crcb; + p->CM_CUR0_CURSOR0_FP_SCALE_BIAS_RB_CRCB__CUR0_FP_SCALE_RB_CRCB = + dpp->att.fp_scale_bias_rb_crcb.bits.fp_scale_rb_crcb; + + p->HUBPREQ0_CURSOR_SETTINGS__CURSOR0_DST_Y_OFFSET = hubp->att.settings.bits.dst_y_offset; + p->HUBPREQ0_CURSOR_SETTINGS__CURSOR0_CHUNK_HDL_ADJUST = hubp->att.settings.bits.chunk_hdl_adjust; + p->HUBP0_DCHUBP_MALL_CONFIG__USE_MALL_FOR_CURSOR = hubp->use_mall_for_cursor; + + cs->offload_streams[stream_idx].payloads[payload_idx].pipe_mask |= (1u << pipe->pipe_idx); +} + +void dcn401_plane_atomic_power_down_sequence(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp, + struct block_sequence_state *seq_state) +{ + struct dce_hwseq *hws = dc->hwseq; + uint32_t org_ip_request_cntl = 0; + + DC_LOGGER_INIT(dc->ctx->logger); + + /* Check and set DC_IP_REQUEST_CNTL if needed */ + if (REG(DC_IP_REQUEST_CNTL)) { + REG_GET(DC_IP_REQUEST_CNTL, IP_REQUEST_EN, &org_ip_request_cntl); + if (org_ip_request_cntl == 0) + hwss_add_dc_ip_request_cntl(seq_state, dc, true); + } + + /* DPP power gating control */ + hwss_add_dpp_pg_control(seq_state, hws, dpp->inst, false); + + /* HUBP power gating control */ + hwss_add_hubp_pg_control(seq_state, hws, hubp->inst, false); + + /* HUBP reset */ + hwss_add_hubp_reset(seq_state, hubp); + + /* DPP reset */ + hwss_add_dpp_reset(seq_state, dpp); + + /* Restore DC_IP_REQUEST_CNTL if it was originally 0 */ + if (org_ip_request_cntl == 0 && REG(DC_IP_REQUEST_CNTL)) + hwss_add_dc_ip_request_cntl(seq_state, dc, false); + + DC_LOG_DEBUG("Power gated front end %d\n", hubp->inst); + + /* DPP root clock control */ + hwss_add_dpp_root_clock_control(seq_state, hws, dpp->inst, false); +} + +/* trigger HW to start disconnect plane from stream on the next vsync using block sequence */ +void dcn401_plane_atomic_disconnect_sequence(struct dc *dc, + struct dc_state *state, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + struct hubp *hubp = pipe_ctx->plane_res.hubp; + int dpp_id = pipe_ctx->plane_res.dpp->inst; + struct mpc *mpc = dc->res_pool->mpc; + struct mpc_tree *mpc_tree_params; + struct mpcc *mpcc_to_remove = NULL; + struct output_pixel_processor *opp = pipe_ctx->stream_res.opp; + + mpc_tree_params = &(opp->mpc_tree_params); + mpcc_to_remove = mpc->funcs->get_mpcc_for_dpp(mpc_tree_params, dpp_id); + + /*Already reset*/ + if (mpcc_to_remove == NULL) + return; + + /* Step 1: Remove MPCC from MPC tree */ + hwss_add_mpc_remove_mpcc(seq_state, mpc, mpc_tree_params, mpcc_to_remove); + + // Phantom pipes have OTG disabled by default, so MPCC_STATUS will never assert idle, + // so don't wait for MPCC_IDLE in the programming sequence + if (dc_state_get_pipe_subvp_type(state, pipe_ctx) != SUBVP_PHANTOM) { + /* Step 2: Set MPCC disconnect pending flag */ + hwss_add_opp_set_mpcc_disconnect_pending(seq_state, opp, pipe_ctx->plane_res.mpcc_inst, true); + } + + /* Step 3: Set optimized required flag */ + hwss_add_dc_set_optimized_required(seq_state, dc, true); + + /* Step 4: Disconnect HUBP if function exists */ + if (hubp->funcs->hubp_disconnect) + hwss_add_hubp_disconnect(seq_state, hubp); + + /* Step 5: Verify pstate change high if debug sanity checks are enabled */ + if (dc->debug.sanity_checks) + dc->hwseq->funcs.verify_allow_pstate_change_high_sequence(dc, seq_state); +} + +void dcn401_blank_pixel_data_sequence( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool blank, + struct block_sequence_state *seq_state) +{ + struct tg_color black_color = {0}; + struct stream_resource *stream_res = &pipe_ctx->stream_res; + struct dc_stream_state *stream = pipe_ctx->stream; + enum dc_color_space color_space = stream->output_color_space; + enum controller_dp_test_pattern test_pattern = CONTROLLER_DP_TEST_PATTERN_SOLID_COLOR; + enum controller_dp_color_space test_pattern_color_space = CONTROLLER_DP_COLOR_SPACE_UDEFINED; + struct pipe_ctx *odm_pipe; + struct rect odm_slice_src; + + if (stream->link->test_pattern_enabled) + return; + + /* get opp dpg blank color */ + color_space_to_black_color(dc, color_space, &black_color); + + if (blank) { + /* Set ABM immediate disable */ + hwss_add_abm_set_immediate_disable(seq_state, dc, pipe_ctx); + + if (dc->debug.visual_confirm != VISUAL_CONFIRM_DISABLE) { + test_pattern = CONTROLLER_DP_TEST_PATTERN_COLORSQUARES; + test_pattern_color_space = CONTROLLER_DP_COLOR_SPACE_RGB; + } + } else { + test_pattern = CONTROLLER_DP_TEST_PATTERN_VIDEOMODE; + } + + odm_pipe = pipe_ctx; + + /* Set display pattern generator for all ODM pipes */ + while (odm_pipe->next_odm_pipe) { + odm_slice_src = resource_get_odm_slice_src_rect(odm_pipe); + + hwss_add_opp_set_disp_pattern_generator(seq_state, + odm_pipe->stream_res.opp, + test_pattern, + test_pattern_color_space, + stream->timing.display_color_depth, + black_color, + true, + odm_slice_src.width, + odm_slice_src.height, + odm_slice_src.x); + + odm_pipe = odm_pipe->next_odm_pipe; + } + + /* Set display pattern generator for final ODM pipe */ + odm_slice_src = resource_get_odm_slice_src_rect(odm_pipe); + + hwss_add_opp_set_disp_pattern_generator(seq_state, + odm_pipe->stream_res.opp, + test_pattern, + test_pattern_color_space, + stream->timing.display_color_depth, + black_color, + true, + odm_slice_src.width, + odm_slice_src.height, + odm_slice_src.x); + + /* Handle ABM level setting when not blanking */ + if (!blank) { + if (stream_res->abm) { + /* Set pipe for ABM */ + hwss_add_abm_set_pipe(seq_state, dc, pipe_ctx); + + /* Set ABM level */ + hwss_add_abm_set_level(seq_state, stream_res->abm, stream->abm_level); + } + } +} + +void dcn401_program_all_writeback_pipes_in_tree_sequence( + struct dc *dc, + const struct dc_stream_state *stream, + struct dc_state *context, + struct block_sequence_state *seq_state) +{ + struct dwbc *dwb; + int i_wb, i_pipe; + + if (!stream || stream->num_wb_info > dc->res_pool->res_cap->num_dwb) + return; + + /* For each writeback pipe */ + for (i_wb = 0; i_wb < stream->num_wb_info; i_wb++) { + /* Get direct pointer to writeback info */ + struct dc_writeback_info *wb_info = (struct dc_writeback_info *)&stream->writeback_info[i_wb]; + int mpcc_inst = -1; + + if (wb_info->wb_enabled) { + /* Get the MPCC instance for writeback_source_plane */ + for (i_pipe = 0; i_pipe < dc->res_pool->pipe_count; i_pipe++) { + struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i_pipe]; + + if (!pipe_ctx->plane_state) + continue; + + if (pipe_ctx->plane_state == wb_info->writeback_source_plane) { + mpcc_inst = pipe_ctx->plane_res.mpcc_inst; + break; + } + } + + if (mpcc_inst == -1) { + /* Disable writeback pipe and disconnect from MPCC + * if source plane has been removed + */ + dcn401_disable_writeback_sequence(dc, wb_info, seq_state); + continue; + } + + ASSERT(wb_info->dwb_pipe_inst < dc->res_pool->res_cap->num_dwb); + dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst]; + + if (dwb->funcs->is_enabled(dwb)) { + /* Writeback pipe already enabled, only need to update */ + dcn401_update_writeback_sequence(dc, wb_info, context, seq_state); + } else { + /* Enable writeback pipe and connect to MPCC */ + dcn401_enable_writeback_sequence(dc, wb_info, context, mpcc_inst, seq_state); + } + } else { + /* Disable writeback pipe and disconnect from MPCC */ + dcn401_disable_writeback_sequence(dc, wb_info, seq_state); + } + } +} + +void dcn401_enable_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct dc_state *context, + int mpcc_inst, + struct block_sequence_state *seq_state) +{ + struct dwbc *dwb; + struct mcif_wb *mcif_wb; + + if (!wb_info->wb_enabled || wb_info->dwb_pipe_inst >= dc->res_pool->res_cap->num_dwb) + return; + + dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst]; + mcif_wb = dc->res_pool->mcif_wb[wb_info->dwb_pipe_inst]; + + /* Update DWBC with new parameters */ + hwss_add_dwbc_update(seq_state, dwb, &wb_info->dwb_params); + + /* Configure MCIF_WB buffer settings */ + hwss_add_mcif_wb_config_buf(seq_state, mcif_wb, &wb_info->mcif_buf_params, wb_info->dwb_params.dest_height); + + /* Configure MCIF_WB arbitration */ + hwss_add_mcif_wb_config_arb(seq_state, mcif_wb, &context->bw_ctx.bw.dcn.bw_writeback.mcif_wb_arb[wb_info->dwb_pipe_inst]); + + /* Enable MCIF_WB */ + hwss_add_mcif_wb_enable(seq_state, mcif_wb); + + /* Set DWB MUX to connect writeback to MPCC */ + hwss_add_mpc_set_dwb_mux(seq_state, dc->res_pool->mpc, wb_info->dwb_pipe_inst, mpcc_inst); + + /* Enable DWBC */ + hwss_add_dwbc_enable(seq_state, dwb, &wb_info->dwb_params); +} + +void dcn401_disable_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct block_sequence_state *seq_state) +{ + struct dwbc *dwb; + struct mcif_wb *mcif_wb; + + if (wb_info->dwb_pipe_inst >= dc->res_pool->res_cap->num_dwb) + return; + + dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst]; + mcif_wb = dc->res_pool->mcif_wb[wb_info->dwb_pipe_inst]; + + /* Disable DWBC */ + hwss_add_dwbc_disable(seq_state, dwb); + + /* Disable DWB MUX */ + hwss_add_mpc_disable_dwb_mux(seq_state, dc->res_pool->mpc, wb_info->dwb_pipe_inst); + + /* Disable MCIF_WB */ + hwss_add_mcif_wb_disable(seq_state, mcif_wb); +} + +void dcn401_update_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct dc_state *context, + struct block_sequence_state *seq_state) +{ + struct dwbc *dwb; + struct mcif_wb *mcif_wb; + + if (!wb_info->wb_enabled || wb_info->dwb_pipe_inst >= dc->res_pool->res_cap->num_dwb) + return; + + dwb = dc->res_pool->dwbc[wb_info->dwb_pipe_inst]; + mcif_wb = dc->res_pool->mcif_wb[wb_info->dwb_pipe_inst]; + + /* Update writeback pipe */ + hwss_add_dwbc_update(seq_state, dwb, &wb_info->dwb_params); + + /* Update MCIF_WB buffer settings if needed */ + hwss_add_mcif_wb_config_buf(seq_state, mcif_wb, &wb_info->mcif_buf_params, wb_info->dwb_params.dest_height); +} + +static int find_free_gsl_group(const struct dc *dc) +{ + if (dc->res_pool->gsl_groups.gsl_0 == 0) + return 1; + if (dc->res_pool->gsl_groups.gsl_1 == 0) + return 2; + if (dc->res_pool->gsl_groups.gsl_2 == 0) + return 3; + + return 0; +} + +void dcn401_setup_gsl_group_as_lock_sequence( + const struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool enable, + struct block_sequence_state *seq_state) +{ + struct gsl_params gsl; + int group_idx; + + memset(&gsl, 0, sizeof(struct gsl_params)); + + if (enable) { + /* return if group already assigned since GSL was set up + * for vsync flip, we would unassign so it can't be "left over" + */ + if (pipe_ctx->stream_res.gsl_group > 0) + return; + + group_idx = find_free_gsl_group(dc); + ASSERT(group_idx != 0); + pipe_ctx->stream_res.gsl_group = group_idx; + + /* set gsl group reg field and mark resource used */ + switch (group_idx) { + case 1: + gsl.gsl0_en = 1; + dc->res_pool->gsl_groups.gsl_0 = 1; + break; + case 2: + gsl.gsl1_en = 1; + dc->res_pool->gsl_groups.gsl_1 = 1; + break; + case 3: + gsl.gsl2_en = 1; + dc->res_pool->gsl_groups.gsl_2 = 1; + break; + default: + BREAK_TO_DEBUGGER(); + return; // invalid case + } + gsl.gsl_master_en = 1; + } else { + group_idx = pipe_ctx->stream_res.gsl_group; + if (group_idx == 0) + return; // if not in use, just return + + pipe_ctx->stream_res.gsl_group = 0; + + /* unset gsl group reg field and mark resource free */ + switch (group_idx) { + case 1: + gsl.gsl0_en = 0; + dc->res_pool->gsl_groups.gsl_0 = 0; + break; + case 2: + gsl.gsl1_en = 0; + dc->res_pool->gsl_groups.gsl_1 = 0; + break; + case 3: + gsl.gsl2_en = 0; + dc->res_pool->gsl_groups.gsl_2 = 0; + break; + default: + BREAK_TO_DEBUGGER(); + return; + } + gsl.gsl_master_en = 0; + } + + hwss_add_tg_set_gsl(seq_state, pipe_ctx->stream_res.tg, gsl); + hwss_add_tg_set_gsl_source_select(seq_state, pipe_ctx->stream_res.tg, group_idx, enable ? 4 : 0); +} + +void dcn401_disable_plane_sequence( + struct dc *dc, + struct dc_state *state, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + bool is_phantom = dc_state_get_pipe_subvp_type(state, pipe_ctx) == SUBVP_PHANTOM; + struct timing_generator *tg = is_phantom ? pipe_ctx->stream_res.tg : NULL; + + if (!pipe_ctx->plane_res.hubp || pipe_ctx->plane_res.hubp->power_gated) + return; + + /* Wait for MPCC disconnect */ + if (dc->hwss.wait_for_mpcc_disconnect_sequence) + dc->hwss.wait_for_mpcc_disconnect_sequence(dc, dc->res_pool, pipe_ctx, seq_state); + + /* In flip immediate with pipe splitting case GSL is used for synchronization + * so we must disable it when the plane is disabled. + */ + if (pipe_ctx->stream_res.gsl_group != 0) + dcn401_setup_gsl_group_as_lock_sequence(dc, pipe_ctx, false, seq_state); + + /* Update HUBP mall sel */ + if (pipe_ctx->plane_res.hubp && pipe_ctx->plane_res.hubp->funcs->hubp_update_mall_sel) + hwss_add_hubp_update_mall_sel(seq_state, pipe_ctx->plane_res.hubp, 0, false); + + /* Set flip control GSL */ + hwss_add_hubp_set_flip_control_gsl(seq_state, pipe_ctx->plane_res.hubp, false); + + /* HUBP clock control */ + hwss_add_hubp_clk_cntl(seq_state, pipe_ctx->plane_res.hubp, false); + + /* DPP clock control */ + hwss_add_dpp_dppclk_control(seq_state, pipe_ctx->plane_res.dpp, false, false); + + /* Plane atomic power down */ + if (dc->hwseq->funcs.plane_atomic_power_down_sequence) + dc->hwseq->funcs.plane_atomic_power_down_sequence(dc, pipe_ctx->plane_res.dpp, + pipe_ctx->plane_res.hubp, seq_state); + + pipe_ctx->stream = NULL; + memset(&pipe_ctx->stream_res, 0, sizeof(pipe_ctx->stream_res)); + memset(&pipe_ctx->plane_res, 0, sizeof(pipe_ctx->plane_res)); + pipe_ctx->top_pipe = NULL; + pipe_ctx->bottom_pipe = NULL; + pipe_ctx->prev_odm_pipe = NULL; + pipe_ctx->next_odm_pipe = NULL; + pipe_ctx->plane_state = NULL; + + /* Turn back off the phantom OTG after the phantom plane is fully disabled */ + if (is_phantom && tg && tg->funcs->disable_phantom_crtc) + hwss_add_disable_phantom_crtc(seq_state, tg); +} + +void dcn401_post_unlock_reset_opp_sequence( + struct dc *dc, + struct pipe_ctx *opp_head, + struct block_sequence_state *seq_state) +{ + struct display_stream_compressor *dsc = opp_head->stream_res.dsc; + struct dccg *dccg = dc->res_pool->dccg; + + /* Wait for all DPP pipes in current mpc blending tree completes double + * buffered disconnection before resetting OPP + */ + if (dc->hwss.wait_for_mpcc_disconnect_sequence) + dc->hwss.wait_for_mpcc_disconnect_sequence(dc, dc->res_pool, opp_head, seq_state); + + if (dsc) { + bool *is_ungated = NULL; + /* Check DSC power gate status */ + if (dc->hwseq && dc->hwseq->funcs.dsc_pg_status) + hwss_add_dsc_pg_status(seq_state, dc->hwseq, dsc->inst, false); + + /* Seamless update specific where we will postpone non + * double buffered DSCCLK disable logic in post unlock + * sequence after DSC is disconnected from OPP but not + * yet power gated. + */ + + /* DSC wait disconnect pending clear */ + hwss_add_dsc_wait_disconnect_pending_clear(seq_state, dsc, is_ungated); + + /* DSC disable */ + hwss_add_dsc_disable(seq_state, dsc, is_ungated); + + /* Set reference DSCCLK */ + if (dccg && dccg->funcs->set_ref_dscclk) + hwss_add_dccg_set_ref_dscclk(seq_state, dccg, dsc->inst, 0); + } +} + +void dcn401_dc_ip_request_cntl(struct dc *dc, bool enable) +{ + struct dce_hwseq *hws = dc->hwseq; + + if (REG(DC_IP_REQUEST_CNTL)) + REG_SET(DC_IP_REQUEST_CNTL, 0, IP_REQUEST_EN, enable ? 1 : 0); +} + +void dcn401_enable_plane_sequence(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state) +{ + struct dce_hwseq *hws = dc->hwseq; + uint32_t org_ip_request_cntl = 0; + + if (!pipe_ctx->plane_res.dpp || !pipe_ctx->plane_res.hubp || !pipe_ctx->stream_res.opp) + return; + + if (REG(DC_IP_REQUEST_CNTL)) + REG_GET(DC_IP_REQUEST_CNTL, IP_REQUEST_EN, &org_ip_request_cntl); + + /* Step 1: DPP root clock control - enable clock */ + if (hws->funcs.dpp_root_clock_control) + hwss_add_dpp_root_clock_control(seq_state, hws, pipe_ctx->plane_res.dpp->inst, true); + + /* Step 2: Enable DC IP request (if needed) */ + if (hws->funcs.dc_ip_request_cntl) + hwss_add_dc_ip_request_cntl(seq_state, dc, true); + + /* Step 3: DPP power gating control - power on */ + if (REG(DC_IP_REQUEST_CNTL) && hws->funcs.dpp_pg_control) + hwss_add_dpp_pg_control(seq_state, hws, pipe_ctx->plane_res.dpp->inst, true); + + /* Step 4: HUBP power gating control - power on */ + if (REG(DC_IP_REQUEST_CNTL) && hws->funcs.hubp_pg_control) + hwss_add_hubp_pg_control(seq_state, hws, pipe_ctx->plane_res.hubp->inst, true); + + /* Step 5: Disable DC IP request (restore state) */ + if (org_ip_request_cntl == 0 && hws->funcs.dc_ip_request_cntl) + hwss_add_dc_ip_request_cntl(seq_state, dc, false); + + /* Step 6: HUBP clock control - enable DCFCLK */ + if (pipe_ctx->plane_res.hubp->funcs->hubp_clk_cntl) + hwss_add_hubp_clk_cntl(seq_state, pipe_ctx->plane_res.hubp, true); + + /* Step 7: HUBP initialization */ + if (pipe_ctx->plane_res.hubp->funcs->hubp_init) + hwss_add_hubp_init(seq_state, pipe_ctx->plane_res.hubp); + + /* Step 8: OPP pipe clock control - enable */ + if (pipe_ctx->stream_res.opp->funcs->opp_pipe_clock_control) + hwss_add_opp_pipe_clock_control(seq_state, pipe_ctx->stream_res.opp, true); + + /* Step 9: VM system aperture settings */ + if (dc->vm_pa_config.valid && pipe_ctx->plane_res.hubp->funcs->hubp_set_vm_system_aperture_settings) { + hwss_add_hubp_set_vm_system_aperture_settings(seq_state, pipe_ctx->plane_res.hubp, 0, + dc->vm_pa_config.system_aperture.start_addr, dc->vm_pa_config.system_aperture.end_addr); + } + + /* Step 10: Flip interrupt setup */ + if (!pipe_ctx->top_pipe + && pipe_ctx->plane_state + && pipe_ctx->plane_state->flip_int_enabled + && pipe_ctx->plane_res.hubp->funcs->hubp_set_flip_int) { + hwss_add_hubp_set_flip_int(seq_state, pipe_ctx->plane_res.hubp); + } +} + +void dcn401_update_dchubp_dpp_sequence(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state) +{ + struct dce_hwseq *hws = dc->hwseq; + struct hubp *hubp = pipe_ctx->plane_res.hubp; + struct dpp *dpp = pipe_ctx->plane_res.dpp; + struct dc_plane_state *plane_state = pipe_ctx->plane_state; + struct dccg *dccg = dc->res_pool->dccg; + bool viewport_changed = false; + enum mall_stream_type pipe_mall_type = dc_state_get_pipe_subvp_type(context, pipe_ctx); + + if (!hubp || !dpp || !plane_state) + return; + + /* Step 1: DPP DPPCLK control */ + if (pipe_ctx->update_flags.bits.dppclk) + hwss_add_dpp_dppclk_control(seq_state, dpp, false, true); + + /* Step 2: DCCG update DPP DTO */ + if (pipe_ctx->update_flags.bits.enable) + hwss_add_dccg_update_dpp_dto(seq_state, dccg, dpp->inst, pipe_ctx->plane_res.bw.dppclk_khz); + + /* Step 3: HUBP VTG selection */ + if (pipe_ctx->update_flags.bits.hubp_rq_dlg_ttu) { + hwss_add_hubp_vtg_sel(seq_state, hubp, pipe_ctx->stream_res.tg->inst); + + /* Step 4: HUBP setup (choose setup2 or setup) */ + if (hubp->funcs->hubp_setup2) { + hwss_add_hubp_setup2(seq_state, hubp, &pipe_ctx->hubp_regs, + &pipe_ctx->global_sync, &pipe_ctx->stream->timing); + } else if (hubp->funcs->hubp_setup) { + hwss_add_hubp_setup(seq_state, hubp, &pipe_ctx->dlg_regs, + &pipe_ctx->ttu_regs, &pipe_ctx->rq_regs, &pipe_ctx->pipe_dlg_param); + } + } + + /* Step 5: Set unbounded requesting */ + if (pipe_ctx->update_flags.bits.unbounded_req && hubp->funcs->set_unbounded_requesting) + hwss_add_hubp_set_unbounded_requesting(seq_state, hubp, pipe_ctx->unbounded_req); + + /* Step 6: HUBP interdependent setup */ + if (pipe_ctx->update_flags.bits.hubp_interdependent) { + if (hubp->funcs->hubp_setup_interdependent2) + hwss_add_hubp_setup_interdependent2(seq_state, hubp, &pipe_ctx->hubp_regs); + else if (hubp->funcs->hubp_setup_interdependent) + hwss_add_hubp_setup_interdependent(seq_state, hubp, &pipe_ctx->dlg_regs, &pipe_ctx->ttu_regs); + } + + /* Step 7: DPP setup - input CSC and format setup */ + if (pipe_ctx->update_flags.bits.enable || + pipe_ctx->update_flags.bits.plane_changed || + plane_state->update_flags.bits.bpp_change || + plane_state->update_flags.bits.input_csc_change || + plane_state->update_flags.bits.color_space_change || + plane_state->update_flags.bits.coeff_reduction_change) { + hwss_add_dpp_setup_dpp(seq_state, pipe_ctx); + + /* Step 8: DPP cursor matrix setup */ + if (dpp->funcs->set_cursor_matrix) { + hwss_add_dpp_set_cursor_matrix(seq_state, dpp, plane_state->color_space, + &plane_state->cursor_csc_color_matrix); + } + + /* Step 9: DPP program bias and scale */ + if (dpp->funcs->dpp_program_bias_and_scale) + hwss_add_dpp_program_bias_and_scale(seq_state, pipe_ctx); + } + + /* Step 10: MPCC updates */ + if (pipe_ctx->update_flags.bits.mpcc || + pipe_ctx->update_flags.bits.plane_changed || + plane_state->update_flags.bits.global_alpha_change || + plane_state->update_flags.bits.per_pixel_alpha_change) { + + /* Check if update_mpcc_sequence is implemented and prefer it over single MPC_UPDATE_MPCC step */ + if (hws->funcs.update_mpcc_sequence) + hws->funcs.update_mpcc_sequence(dc, pipe_ctx, seq_state); + } + + /* Step 11: DPP scaler setup */ + if (pipe_ctx->update_flags.bits.scaler || + plane_state->update_flags.bits.scaling_change || + plane_state->update_flags.bits.position_change || + plane_state->update_flags.bits.per_pixel_alpha_change || + pipe_ctx->stream->update_flags.bits.scaling) { + pipe_ctx->plane_res.scl_data.lb_params.alpha_en = pipe_ctx->plane_state->per_pixel_alpha; + ASSERT(pipe_ctx->plane_res.scl_data.lb_params.depth == LB_PIXEL_DEPTH_36BPP); + hwss_add_dpp_set_scaler(seq_state, pipe_ctx->plane_res.dpp, &pipe_ctx->plane_res.scl_data); + } + + /* Step 12: HUBP viewport programming */ + if (pipe_ctx->update_flags.bits.viewport || + (context == dc->current_state && plane_state->update_flags.bits.position_change) || + (context == dc->current_state && plane_state->update_flags.bits.scaling_change) || + (context == dc->current_state && pipe_ctx->stream->update_flags.bits.scaling)) { + hwss_add_hubp_mem_program_viewport(seq_state, hubp, + &pipe_ctx->plane_res.scl_data.viewport, &pipe_ctx->plane_res.scl_data.viewport_c); + viewport_changed = true; + } + + /* Step 13: HUBP program mcache if available */ + if (hubp->funcs->hubp_program_mcache_id_and_split_coordinate) + hwss_add_hubp_program_mcache_id(seq_state, hubp, &pipe_ctx->mcache_regs); + + /* Step 14: Cursor attribute setup */ + if ((pipe_ctx->update_flags.bits.enable || pipe_ctx->update_flags.bits.opp_changed || + pipe_ctx->update_flags.bits.scaler || viewport_changed == true) && + pipe_ctx->stream->cursor_attributes.address.quad_part != 0) { + + hwss_add_abort_cursor_offload_update(seq_state, dc, pipe_ctx); + + hwss_add_set_cursor_attribute(seq_state, dc, pipe_ctx); + + /* Step 15: Cursor position setup */ + hwss_add_set_cursor_position(seq_state, dc, pipe_ctx); + + /* Step 16: Cursor SDR white level */ + if (dc->hwss.set_cursor_sdr_white_level) + hwss_add_set_cursor_sdr_white_level(seq_state, dc, pipe_ctx); + } + + /* Step 17: Gamut remap and output CSC */ + if (pipe_ctx->update_flags.bits.enable || pipe_ctx->update_flags.bits.opp_changed || + pipe_ctx->update_flags.bits.plane_changed || + pipe_ctx->stream->update_flags.bits.gamut_remap || + plane_state->update_flags.bits.gamut_remap_change || + pipe_ctx->stream->update_flags.bits.out_csc) { + + /* Gamut remap */ + hwss_add_dpp_program_gamut_remap(seq_state, pipe_ctx); + + /* Output CSC */ + hwss_add_program_output_csc(seq_state, dc, pipe_ctx, pipe_ctx->stream->output_color_space, + pipe_ctx->stream->csc_color_matrix.matrix, hubp->opp_id); + } + + /* Step 18: HUBP surface configuration */ + if (pipe_ctx->update_flags.bits.enable || + pipe_ctx->update_flags.bits.plane_changed || + pipe_ctx->update_flags.bits.opp_changed || + plane_state->update_flags.bits.pixel_format_change || + plane_state->update_flags.bits.horizontal_mirror_change || + plane_state->update_flags.bits.rotation_change || + plane_state->update_flags.bits.swizzle_change || + plane_state->update_flags.bits.dcc_change || + plane_state->update_flags.bits.bpp_change || + plane_state->update_flags.bits.scaling_change || + plane_state->update_flags.bits.plane_size_change) { + struct plane_size size = plane_state->plane_size; + + size.surface_size = pipe_ctx->plane_res.scl_data.viewport; + hwss_add_hubp_program_surface_config(seq_state, hubp, + plane_state->format, &plane_state->tiling_info, size, + plane_state->rotation, &plane_state->dcc, + plane_state->horizontal_mirror, 0); + hubp->power_gated = false; + } + + /* Step 19: Update plane address (with SubVP support) */ + if (pipe_ctx->update_flags.bits.enable || + pipe_ctx->update_flags.bits.plane_changed || + plane_state->update_flags.bits.addr_update) { + + /* SubVP save surface address if needed */ + if (resource_is_pipe_type(pipe_ctx, OTG_MASTER) && pipe_mall_type == SUBVP_MAIN) { + hwss_add_dmub_subvp_save_surf_addr(seq_state, dc->ctx->dmub_srv, + &pipe_ctx->plane_state->address, pipe_ctx->subvp_index); + } + + /* Update plane address */ + hwss_add_hubp_update_plane_addr(seq_state, dc, pipe_ctx); + } + + /* Step 20: HUBP set blank - enable plane */ + if (pipe_ctx->update_flags.bits.enable) + hwss_add_hubp_set_blank(seq_state, hubp, false); + + /* Step 21: Phantom HUBP post enable */ + if (pipe_mall_type == SUBVP_PHANTOM && hubp->funcs->phantom_hubp_post_enable) + hwss_add_phantom_hubp_post_enable(seq_state, hubp); +} + +void dcn401_update_mpcc_sequence(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + struct hubp *hubp = pipe_ctx->plane_res.hubp; + struct mpcc_blnd_cfg blnd_cfg = {0}; + bool per_pixel_alpha; + int mpcc_id; + struct mpcc *new_mpcc; + struct mpc *mpc = dc->res_pool->mpc; + struct mpc_tree *mpc_tree_params = &(pipe_ctx->stream_res.opp->mpc_tree_params); + + if (!hubp || !pipe_ctx->plane_state) + return; + + per_pixel_alpha = pipe_ctx->plane_state->per_pixel_alpha; + + /* Initialize blend configuration */ + blnd_cfg.overlap_only = false; + blnd_cfg.global_gain = 0xff; + + if (per_pixel_alpha) { + blnd_cfg.pre_multiplied_alpha = pipe_ctx->plane_state->pre_multiplied_alpha; + if (pipe_ctx->plane_state->global_alpha) { + blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA_COMBINED_GLOBAL_GAIN; + blnd_cfg.global_gain = pipe_ctx->plane_state->global_alpha_value; + } else { + blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_PER_PIXEL_ALPHA; + } + } else { + blnd_cfg.pre_multiplied_alpha = false; + blnd_cfg.alpha_mode = MPCC_ALPHA_BLEND_MODE_GLOBAL_ALPHA; + } + + if (pipe_ctx->plane_state->global_alpha) + blnd_cfg.global_alpha = pipe_ctx->plane_state->global_alpha_value; + else + blnd_cfg.global_alpha = 0xff; + + blnd_cfg.background_color_bpc = 4; + blnd_cfg.bottom_gain_mode = 0; + blnd_cfg.top_gain = 0x1f000; + blnd_cfg.bottom_inside_gain = 0x1f000; + blnd_cfg.bottom_outside_gain = 0x1f000; + + if (pipe_ctx->plane_state->format == SURFACE_PIXEL_FORMAT_GRPH_RGBE_ALPHA) + blnd_cfg.pre_multiplied_alpha = false; + + /* MPCC instance is equal to HUBP instance */ + mpcc_id = hubp->inst; + + /* Step 1: Update blending if no full update needed */ + if (!pipe_ctx->plane_state->update_flags.bits.full_update && + !pipe_ctx->update_flags.bits.mpcc) { + + /* Update blending configuration */ + hwss_add_mpc_update_blending(seq_state, mpc, blnd_cfg, mpcc_id); + + /* Update visual confirm color */ + hwss_add_mpc_update_visual_confirm(seq_state, dc, pipe_ctx, mpcc_id); + return; + } + + /* Step 2: Get existing MPCC for DPP */ + new_mpcc = mpc->funcs->get_mpcc_for_dpp(mpc_tree_params, mpcc_id); + + /* Step 3: Remove MPCC if being used */ + if (new_mpcc != NULL) { + hwss_add_mpc_remove_mpcc(seq_state, mpc, mpc_tree_params, new_mpcc); + } else { + /* Step 4: Assert MPCC idle (debug only) */ + if (dc->debug.sanity_checks) + hwss_add_mpc_assert_idle_mpcc(seq_state, mpc, mpcc_id); + } + + /* Step 5: Insert new plane into MPC tree */ + hwss_add_mpc_insert_plane(seq_state, mpc, mpc_tree_params, blnd_cfg, NULL, NULL, hubp->inst, mpcc_id); + + /* Step 6: Update visual confirm color */ + hwss_add_mpc_update_visual_confirm(seq_state, dc, pipe_ctx, mpcc_id); + + /* Step 7: Set HUBP OPP and MPCC IDs */ + hubp->opp_id = pipe_ctx->stream_res.opp->inst; + hubp->mpcc_id = mpcc_id; +} + +static struct hubp *get_hubp_by_inst(struct resource_pool *res_pool, int mpcc_inst) +{ + int i; + + for (i = 0; i < res_pool->pipe_count; i++) { + if (res_pool->hubps[i]->inst == mpcc_inst) + return res_pool->hubps[i]; + } + ASSERT(false); + return NULL; +} + +void dcn401_wait_for_mpcc_disconnect_sequence( + struct dc *dc, + struct resource_pool *res_pool, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + int mpcc_inst; + + if (dc->debug.sanity_checks) + dc->hwseq->funcs.verify_allow_pstate_change_high_sequence(dc, seq_state); + + if (!pipe_ctx->stream_res.opp) + return; + + for (mpcc_inst = 0; mpcc_inst < MAX_PIPES; mpcc_inst++) { + if (pipe_ctx->stream_res.opp->mpcc_disconnect_pending[mpcc_inst]) { + struct hubp *hubp = get_hubp_by_inst(res_pool, mpcc_inst); + + if (pipe_ctx->stream_res.tg && + pipe_ctx->stream_res.tg->funcs->is_tg_enabled(pipe_ctx->stream_res.tg)) { + hwss_add_mpc_assert_idle_mpcc(seq_state, res_pool->mpc, mpcc_inst); + } + pipe_ctx->stream_res.opp->mpcc_disconnect_pending[mpcc_inst] = false; + if (hubp) + hwss_add_hubp_set_blank(seq_state, hubp, true); + } + } + + if (dc->debug.sanity_checks) + dc->hwseq->funcs.verify_allow_pstate_change_high_sequence(dc, seq_state); +} + +void dcn401_setup_vupdate_interrupt_sequence(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + struct timing_generator *tg = pipe_ctx->stream_res.tg; + int start_line = dc->hwss.get_vupdate_offset_from_vsync(pipe_ctx); + + if (start_line < 0) + start_line = 0; + + if (tg->funcs->setup_vertical_interrupt2) + hwss_add_tg_setup_vertical_interrupt2(seq_state, tg, start_line); +} + +void dcn401_set_hdr_multiplier_sequence(struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state) +{ + struct fixed31_32 multiplier = pipe_ctx->plane_state->hdr_mult; + uint32_t hw_mult = 0x1f000; // 1.0 default multiplier + struct custom_float_format fmt; + + fmt.exponenta_bits = 6; + fmt.mantissa_bits = 12; + fmt.sign = true; + + if (!dc_fixpt_eq(multiplier, dc_fixpt_from_int(0))) // check != 0 + convert_to_custom_float_format(multiplier, &fmt, &hw_mult); + + hwss_add_dpp_set_hdr_multiplier(seq_state, pipe_ctx->plane_res.dpp, hw_mult); +} + +void dcn401_program_mall_pipe_config_sequence(struct dc *dc, struct dc_state *context, + struct block_sequence_state *seq_state) +{ + int i; + unsigned int num_ways = dcn401_calculate_cab_allocation(dc, context); + bool cache_cursor = false; + + // Don't force p-state disallow -- can't block dummy p-state + + // Update MALL_SEL register for each pipe (break down update_mall_sel call) + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + struct hubp *hubp = pipe->plane_res.hubp; + + if (pipe->stream && pipe->plane_state && hubp && hubp->funcs->hubp_update_mall_sel) { + int cursor_size = hubp->curs_attr.pitch * hubp->curs_attr.height; + + switch (hubp->curs_attr.color_format) { + case CURSOR_MODE_MONO: + cursor_size /= 2; + break; + case CURSOR_MODE_COLOR_1BIT_AND: + case CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA: + case CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA: + cursor_size *= 4; + break; + + case CURSOR_MODE_COLOR_64BIT_FP_PRE_MULTIPLIED: + case CURSOR_MODE_COLOR_64BIT_FP_UN_PRE_MULTIPLIED: + default: + cursor_size *= 8; + break; + } + + if (cursor_size > 16384) + cache_cursor = true; + + if (dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + hwss_add_hubp_update_mall_sel(seq_state, hubp, 1, false); + } else { + // MALL not supported with Stereo3D + uint32_t mall_sel = (num_ways <= dc->caps.cache_num_ways && + pipe->stream->link->psr_settings.psr_version == DC_PSR_VERSION_UNSUPPORTED && + pipe->plane_state->address.type != PLN_ADDR_TYPE_GRPH_STEREO && + !pipe->plane_state->address.tmz_surface) ? 2 : 0; + hwss_add_hubp_update_mall_sel(seq_state, hubp, mall_sel, cache_cursor); + } + } + } + + // Program FORCE_ONE_ROW_FOR_FRAME and CURSOR_REQ_MODE for main subvp pipes + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i]; + struct hubp *hubp = pipe->plane_res.hubp; + + if (pipe->stream && hubp && hubp->funcs->hubp_prepare_subvp_buffering) { + if (dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_MAIN) + hwss_add_hubp_prepare_subvp_buffering(seq_state, hubp, true); + } + } +} + +void dcn401_verify_allow_pstate_change_high_sequence(struct dc *dc, + struct block_sequence_state *seq_state) +{ + struct hubbub *hubbub = dc->res_pool->hubbub; + + if (!hubbub->funcs->verify_allow_pstate_change_high) + return; + + if (!hubbub->funcs->verify_allow_pstate_change_high(hubbub)) { + /* Attempt hardware workaround force recovery */ + dcn401_hw_wa_force_recovery_sequence(dc, seq_state); + } +} + +bool dcn401_hw_wa_force_recovery_sequence(struct dc *dc, + struct block_sequence_state *seq_state) +{ + struct hubp *hubp; + unsigned int i; + + if (!dc->debug.recovery_enabled) + return false; + + /* Step 1: Set HUBP_BLANK_EN=1 for all active pipes */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + if (hubp != NULL && hubp->funcs->set_hubp_blank_en) + hwss_add_hubp_set_blank_en(seq_state, hubp, true); + } + } + + /* Step 2: DCHUBBUB_GLOBAL_SOFT_RESET=1 */ + hwss_add_hubbub_soft_reset(seq_state, dc->res_pool->hubbub, hubbub1_soft_reset, true); + + /* Step 3: Set HUBP_DISABLE=1 for all active pipes */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + if (hubp != NULL && hubp->funcs->hubp_disable_control) + hwss_add_hubp_disable_control(seq_state, hubp, true); + } + } + + /* Step 4: Set HUBP_DISABLE=0 for all active pipes */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + if (hubp != NULL && hubp->funcs->hubp_disable_control) + hwss_add_hubp_disable_control(seq_state, hubp, false); + } + } + + /* Step 5: DCHUBBUB_GLOBAL_SOFT_RESET=0 */ + hwss_add_hubbub_soft_reset(seq_state, dc->res_pool->hubbub, hubbub1_soft_reset, false); + + /* Step 6: Set HUBP_BLANK_EN=0 for all active pipes */ + for (i = 0; i < dc->res_pool->pipe_count; i++) { + struct pipe_ctx *pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; + + if (pipe_ctx != NULL) { + hubp = pipe_ctx->plane_res.hubp; + if (hubp != NULL && hubp->funcs->set_hubp_blank_en) + hwss_add_hubp_set_blank_en(seq_state, hubp, false); + } + } + + return true; +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h index 2621b7725267..f78162ab859b 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h @@ -9,6 +9,7 @@ #include "dc.h" #include "dc_stream.h" #include "hw_sequencer_private.h" +#include "hwss/hw_sequencer.h" #include "dcn401/dcn401_dccg.h" struct dc; @@ -73,15 +74,17 @@ void dcn401_optimize_bandwidth( struct dc *dc, struct dc_state *context); -void dcn401_fams2_global_control_lock(struct dc *dc, +void dcn401_dmub_hw_control_lock(struct dc *dc, struct dc_state *context, bool lock); void dcn401_fams2_update_config(struct dc *dc, struct dc_state *context, bool enable); -void dcn401_fams2_global_control_lock_fast(union block_sequence_params *params); +void dcn401_dmub_hw_control_lock_fast(union block_sequence_params *params); void dcn401_unblank_stream(struct pipe_ctx *pipe_ctx, struct dc_link_settings *link_settings); void dcn401_hardware_release(struct dc *dc); void dcn401_update_odm(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master); +void dcn401_update_odm_sequence(struct dc *dc, struct dc_state *context, + struct pipe_ctx *otg_master, struct block_sequence_state *seq_state); void adjust_hotspot_between_slices_for_2x_magnify(uint32_t cursor_width, struct dc_cursor_position *pos_cpy); void dcn401_wait_for_det_buffer_update_under_otg_master(struct dc *dc, struct dc_state *context, struct pipe_ctx *otg_master); void dcn401_interdependent_update_lock(struct dc *dc, struct dc_state *context, bool lock); @@ -97,6 +100,11 @@ void dcn401_program_pipe( struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context); +void dcn401_program_pipe_sequence( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); void dcn401_perform_3dlut_wa_unlock(struct pipe_ctx *pipe_ctx); void dcn401_program_front_end_for_ctx(struct dc *dc, struct dc_state *context); void dcn401_post_unlock_program_front_end(struct dc *dc, struct dc_state *context); @@ -109,5 +117,97 @@ void dcn401_detect_pipe_changes( void dcn401_plane_atomic_power_down(struct dc *dc, struct dpp *dpp, struct hubp *hubp); +void dcn401_plane_atomic_power_down_sequence(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp, + struct block_sequence_state *seq_state); +void dcn401_plane_atomic_disconnect_sequence(struct dc *dc, + struct dc_state *state, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); +void dcn401_blank_pixel_data_sequence( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool blank, + struct block_sequence_state *seq_state); void dcn401_initialize_min_clocks(struct dc *dc); +void dcn401_update_cursor_offload_pipe(struct dc *dc, const struct pipe_ctx *pipe); + +void dcn401_program_all_writeback_pipes_in_tree_sequence( + struct dc *dc, + const struct dc_stream_state *stream, + struct dc_state *context, + struct block_sequence_state *seq_state); + +void dcn401_enable_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct dc_state *context, + int mpcc_inst, + struct block_sequence_state *seq_state); + +void dcn401_disable_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct block_sequence_state *seq_state); + +void dcn401_update_writeback_sequence( + struct dc *dc, + struct dc_writeback_info *wb_info, + struct dc_state *context, + struct block_sequence_state *seq_state); + +void dcn401_setup_gsl_group_as_lock_sequence( + const struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool enable, + struct block_sequence_state *seq_state); + +void dcn401_disable_plane_sequence( + struct dc *dc, + struct dc_state *state, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); + +void dcn401_post_unlock_reset_opp_sequence( + struct dc *dc, + struct pipe_ctx *opp_head, + struct block_sequence_state *seq_state); + +void dcn401_dc_ip_request_cntl(struct dc *dc, bool enable); + +void dcn401_enable_plane_sequence(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); + +void dcn401_update_dchubp_dpp_sequence(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); + +void dcn401_update_mpcc_sequence(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); + +void dcn401_wait_for_mpcc_disconnect_sequence( + struct dc *dc, + struct resource_pool *res_pool, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); + +void dcn401_setup_vupdate_interrupt_sequence(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); + +void dcn401_set_hdr_multiplier_sequence(struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); + +void dcn401_program_mall_pipe_config_sequence(struct dc *dc, struct dc_state *context, + struct block_sequence_state *seq_state); + +void dcn401_verify_allow_pstate_change_high_sequence(struct dc *dc, + struct block_sequence_state *seq_state); + +bool dcn401_hw_wa_force_recovery_sequence(struct dc *dc, + struct block_sequence_state *seq_state); + #endif /* __DC_HWSS_DCN401_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c index d6e11b7e4fce..162096ce0bdf 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c @@ -9,6 +9,7 @@ #include "dcn30/dcn30_hwseq.h" #include "dcn31/dcn31_hwseq.h" #include "dcn32/dcn32_hwseq.h" +#include "dcn35/dcn35_hwseq.h" #include "dcn401/dcn401_hwseq.h" #include "dcn401_init.h" @@ -38,6 +39,7 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .enable_audio_stream = dce110_enable_audio_stream, .disable_audio_stream = dce110_disable_audio_stream, .disable_plane = dcn20_disable_plane, + .disable_plane_sequence = dcn401_disable_plane_sequence, .pipe_control_lock = dcn20_pipe_control_lock, .interdependent_update_lock = dcn401_interdependent_update_lock, .cursor_lock = dcn10_cursor_lock, @@ -53,6 +55,7 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .get_hw_state = dcn10_get_hw_state, .clear_status_bits = dcn10_clear_status_bits, .wait_for_mpcc_disconnect = dcn10_wait_for_mpcc_disconnect, + .wait_for_mpcc_disconnect_sequence = dcn401_wait_for_mpcc_disconnect_sequence, .edp_backlight_control = dce110_edp_backlight_control, .edp_power_control = dce110_edp_power_control, .edp_wait_for_hpd_ready = dce110_edp_wait_for_hpd_ready, @@ -60,6 +63,12 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .set_cursor_position = dcn401_set_cursor_position, .set_cursor_attribute = dcn10_set_cursor_attribute, .set_cursor_sdr_white_level = dcn10_set_cursor_sdr_white_level, + .abort_cursor_offload_update = dcn35_abort_cursor_offload_update, + .begin_cursor_offload_update = dcn35_begin_cursor_offload_update, + .commit_cursor_offload_update = dcn35_commit_cursor_offload_update, + .update_cursor_offload_pipe = dcn401_update_cursor_offload_pipe, + .notify_cursor_offload_drr_update = dcn35_notify_cursor_offload_drr_update, + .program_cursor_offload_now = dcn35_program_cursor_offload_now, .setup_periodic_interrupt = dcn10_setup_periodic_interrupt, .set_clock = dcn10_set_clock, .get_clock = dcn10_get_clock, @@ -95,55 +104,70 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .apply_update_flags_for_phantom = dcn32_apply_update_flags_for_phantom, .wait_for_dcc_meta_propagation = dcn401_wait_for_dcc_meta_propagation, .is_pipe_topology_transition_seamless = dcn32_is_pipe_topology_transition_seamless, - .fams2_global_control_lock = dcn401_fams2_global_control_lock, + .dmub_hw_control_lock = dcn401_dmub_hw_control_lock, .fams2_update_config = dcn401_fams2_update_config, - .fams2_global_control_lock_fast = dcn401_fams2_global_control_lock_fast, + .dmub_hw_control_lock_fast = dcn401_dmub_hw_control_lock_fast, .program_outstanding_updates = dcn401_program_outstanding_updates, .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, .detect_pipe_changes = dcn401_detect_pipe_changes, .enable_plane = dcn20_enable_plane, + .enable_plane_sequence = dcn401_enable_plane_sequence, .update_dchubp_dpp = dcn20_update_dchubp_dpp, + .update_dchubp_dpp_sequence = dcn401_update_dchubp_dpp_sequence, .post_unlock_reset_opp = dcn20_post_unlock_reset_opp, + .post_unlock_reset_opp_sequence = dcn401_post_unlock_reset_opp_sequence, .get_underflow_debug_data = dcn30_get_underflow_debug_data, }; static const struct hwseq_private_funcs dcn401_private_funcs = { .init_pipes = dcn10_init_pipes, .plane_atomic_disconnect = dcn10_plane_atomic_disconnect, + .plane_atomic_disconnect_sequence = dcn401_plane_atomic_disconnect_sequence, .update_mpcc = dcn20_update_mpcc, + .update_mpcc_sequence = dcn401_update_mpcc_sequence, .set_input_transfer_func = dcn32_set_input_transfer_func, .set_output_transfer_func = dcn401_set_output_transfer_func, .power_down = dce110_power_down, .enable_display_power_gating = dcn10_dummy_display_power_gating, .blank_pixel_data = dcn20_blank_pixel_data, + .blank_pixel_data_sequence = dcn401_blank_pixel_data_sequence, .reset_hw_ctx_wrap = dcn401_reset_hw_ctx_wrap, .enable_stream_timing = dcn401_enable_stream_timing, .edp_backlight_control = dce110_edp_backlight_control, .setup_vupdate_interrupt = dcn20_setup_vupdate_interrupt, + .setup_vupdate_interrupt_sequence = dcn401_setup_vupdate_interrupt_sequence, .did_underflow_occur = dcn10_did_underflow_occur, .init_blank = dcn32_init_blank, .disable_vga = dcn20_disable_vga, .bios_golden_init = dcn10_bios_golden_init, .plane_atomic_disable = dcn20_plane_atomic_disable, .plane_atomic_power_down = dcn401_plane_atomic_power_down, + .plane_atomic_power_down_sequence = dcn401_plane_atomic_power_down_sequence, .enable_power_gating_plane = dcn32_enable_power_gating_plane, .hubp_pg_control = dcn32_hubp_pg_control, .program_all_writeback_pipes_in_tree = dcn30_program_all_writeback_pipes_in_tree, + .program_all_writeback_pipes_in_tree_sequence = dcn401_program_all_writeback_pipes_in_tree_sequence, .update_odm = dcn401_update_odm, + .update_odm_sequence = dcn401_update_odm_sequence, .dsc_pg_control = dcn32_dsc_pg_control, .dsc_pg_status = dcn32_dsc_pg_status, .set_hdr_multiplier = dcn10_set_hdr_multiplier, + .set_hdr_multiplier_sequence = dcn401_set_hdr_multiplier_sequence, .verify_allow_pstate_change_high = dcn10_verify_allow_pstate_change_high, + .verify_allow_pstate_change_high_sequence = dcn401_verify_allow_pstate_change_high_sequence, .wait_for_blank_complete = dcn20_wait_for_blank_complete, .dccg_init = dcn20_dccg_init, .set_mcm_luts = dcn401_set_mcm_luts, .program_mall_pipe_config = dcn32_program_mall_pipe_config, + .program_mall_pipe_config_sequence = dcn401_program_mall_pipe_config_sequence, .update_mall_sel = dcn32_update_mall_sel, .calculate_dccg_k1_k2_values = NULL, .apply_single_controller_ctx_to_hw = dce110_apply_single_controller_ctx_to_hw, .reset_back_end_for_pipe = dcn401_reset_back_end_for_pipe, .populate_mcm_luts = NULL, .perform_3dlut_wa_unlock = dcn401_perform_3dlut_wa_unlock, + .program_pipe_sequence = dcn401_program_pipe_sequence, + .dc_ip_request_cntl = dcn401_dc_ip_request_cntl, }; void dcn401_hw_sequencer_init_functions(struct dc *dc) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index 1723bbcf2c46..8ed9eea40c56 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -31,6 +31,8 @@ #include "inc/hw/opp.h" #include "inc/hw/link_encoder.h" #include "inc/core_status.h" +#include "inc/hw/hw_shared.h" +#include "dsc/dsc.h" struct pipe_ctx; struct dc_state; @@ -48,6 +50,8 @@ struct dc_dmub_cmd; struct pg_block_update; struct drr_params; struct dc_underflow_debug_data; +struct dsc_optc_config; +struct vm_system_aperture_param; struct subvp_pipe_control_lock_fast_params { struct dc *dc; @@ -62,7 +66,7 @@ struct pipe_control_lock_params { }; struct set_flip_control_gsl_params { - struct pipe_ctx *pipe_ctx; + struct hubp *hubp; bool flip_immediate; }; @@ -148,12 +152,587 @@ struct wait_for_dcc_meta_propagation_params { const struct pipe_ctx *top_pipe_to_program; }; -struct fams2_global_control_lock_fast_params { +struct dmub_hw_control_lock_fast_params { struct dc *dc; bool is_required; bool lock; }; +struct program_surface_config_params { + struct hubp *hubp; + enum surface_pixel_format format; + struct dc_tiling_info *tiling_info; + struct plane_size plane_size; + enum dc_rotation_angle rotation; + struct dc_plane_dcc_param *dcc; + bool horizontal_mirror; + int compat_level; +}; + +struct program_mcache_id_and_split_coordinate { + struct hubp *hubp; + struct dml2_hubp_pipe_mcache_regs *mcache_regs; +}; + +struct program_cursor_update_now_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct hubp_wait_pipe_read_start_params { + struct hubp *hubp; +}; + +struct apply_update_flags_for_phantom_params { + struct pipe_ctx *pipe_ctx; +}; + +struct update_phantom_vp_position_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; + struct dc_state *context; +}; + +struct set_odm_combine_params { + struct timing_generator *tg; + int opp_inst[MAX_PIPES]; + int opp_head_count; + int odm_slice_width; + int last_odm_slice_width; +}; + +struct set_odm_bypass_params { + struct timing_generator *tg; + const struct dc_crtc_timing *timing; +}; + +struct opp_pipe_clock_control_params { + struct output_pixel_processor *opp; + bool enable; +}; + +struct opp_program_left_edge_extra_pixel_params { + struct output_pixel_processor *opp; + enum dc_pixel_encoding pixel_encoding; + bool is_otg_master; +}; + +struct dccg_set_dto_dscclk_params { + struct dccg *dccg; + int inst; + int num_slices_h; +}; + +struct dsc_set_config_params { + struct display_stream_compressor *dsc; + struct dsc_config *dsc_cfg; + struct dsc_optc_config *dsc_optc_cfg; +}; + +struct dsc_enable_params { + struct display_stream_compressor *dsc; + int opp_inst; +}; + +struct tg_set_dsc_config_params { + struct timing_generator *tg; + struct dsc_optc_config *dsc_optc_cfg; + bool enable; +}; + +struct dsc_disconnect_params { + struct display_stream_compressor *dsc; +}; + +struct dsc_read_state_params { + struct display_stream_compressor *dsc; + struct dcn_dsc_state *dsc_state; +}; + +struct dsc_calculate_and_set_config_params { + struct pipe_ctx *pipe_ctx; + struct dsc_optc_config dsc_optc_cfg; + bool enable; + int opp_cnt; +}; + +struct dsc_enable_with_opp_params { + struct pipe_ctx *pipe_ctx; +}; + +struct program_tg_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; + struct dc_state *context; +}; + +struct tg_program_global_sync_params { + struct timing_generator *tg; + int vready_offset; + unsigned int vstartup_lines; + unsigned int vupdate_offset_pixels; + unsigned int vupdate_vupdate_width_pixels; + unsigned int pstate_keepout_start_lines; +}; + +struct tg_wait_for_state_params { + struct timing_generator *tg; + enum crtc_state state; +}; + +struct tg_set_vtg_params_params { + struct timing_generator *tg; + struct dc_crtc_timing *timing; + bool program_fp2; +}; + +struct tg_set_gsl_params { + struct timing_generator *tg; + struct gsl_params gsl; +}; + +struct tg_set_gsl_source_select_params { + struct timing_generator *tg; + int group_idx; + uint32_t gsl_ready_signal; +}; + +struct setup_vupdate_interrupt_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct tg_setup_vertical_interrupt2_params { + struct timing_generator *tg; + int start_line; +}; + +struct dpp_set_hdr_multiplier_params { + struct dpp *dpp; + uint32_t hw_mult; +}; + +struct program_det_size_params { + struct hubbub *hubbub; + unsigned int hubp_inst; + unsigned int det_buffer_size_kb; +}; + +struct program_det_segments_params { + struct hubbub *hubbub; + unsigned int hubp_inst; + unsigned int det_size; +}; + +struct update_dchubp_dpp_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; + struct dc_state *context; +}; + +struct opp_set_dyn_expansion_params { + struct output_pixel_processor *opp; + enum dc_color_space color_space; + enum dc_color_depth color_depth; + enum signal_type signal; +}; + +struct opp_program_fmt_params { + struct output_pixel_processor *opp; + struct bit_depth_reduction_params *fmt_bit_depth; + struct clamping_and_pixel_encoding_params *clamping; +}; + +struct opp_program_bit_depth_reduction_params { + struct output_pixel_processor *opp; + bool use_default_params; + struct pipe_ctx *pipe_ctx; +}; + +struct opp_set_disp_pattern_generator_params { + struct output_pixel_processor *opp; + enum controller_dp_test_pattern test_pattern; + enum controller_dp_color_space color_space; + enum dc_color_depth color_depth; + struct tg_color solid_color; + bool use_solid_color; + int width; + int height; + int offset; +}; + +struct set_abm_pipe_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct set_abm_level_params { + struct abm *abm; + unsigned int abm_level; +}; + +struct set_abm_immediate_disable_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct set_disp_pattern_generator_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; + enum controller_dp_test_pattern test_pattern; + enum controller_dp_color_space color_space; + enum dc_color_depth color_depth; + const struct tg_color *solid_color; + int width; + int height; + int offset; +}; + +struct mpc_update_blending_params { + struct mpc *mpc; + struct mpcc_blnd_cfg blnd_cfg; + int mpcc_id; +}; + +struct mpc_assert_idle_mpcc_params { + struct mpc *mpc; + int mpcc_id; +}; + +struct mpc_insert_plane_params { + struct mpc *mpc; + struct mpc_tree *mpc_tree_params; + struct mpcc_blnd_cfg blnd_cfg; + struct mpcc_sm_cfg *sm_cfg; + struct mpcc *insert_above_mpcc; + int dpp_id; + int mpcc_id; +}; + +struct mpc_remove_mpcc_params { + struct mpc *mpc; + struct mpc_tree *mpc_tree_params; + struct mpcc *mpcc_to_remove; +}; + +struct opp_set_mpcc_disconnect_pending_params { + struct output_pixel_processor *opp; + int mpcc_inst; + bool pending; +}; + +struct dc_set_optimized_required_params { + struct dc *dc; + bool optimized_required; +}; + +struct hubp_disconnect_params { + struct hubp *hubp; +}; + +struct hubbub_force_pstate_change_control_params { + struct hubbub *hubbub; + bool enable; + bool wait; +}; + +struct tg_enable_crtc_params { + struct timing_generator *tg; +}; + +struct hubp_wait_flip_pending_params { + struct hubp *hubp; + unsigned int timeout_us; + unsigned int polling_interval_us; +}; + +struct tg_wait_double_buffer_pending_params { + struct timing_generator *tg; + unsigned int timeout_us; + unsigned int polling_interval_us; +}; + +struct update_force_pstate_params { + struct dc *dc; + struct dc_state *context; +}; + +struct hubbub_apply_dedcn21_147_wa_params { + struct hubbub *hubbub; +}; + +struct hubbub_allow_self_refresh_control_params { + struct hubbub *hubbub; + bool allow; + bool *disallow_self_refresh_applied; +}; + +struct tg_get_frame_count_params { + struct timing_generator *tg; + unsigned int *frame_count; +}; + +struct mpc_set_dwb_mux_params { + struct mpc *mpc; + int dwb_id; + int mpcc_id; +}; + +struct mpc_disable_dwb_mux_params { + struct mpc *mpc; + unsigned int dwb_id; +}; + +struct mcif_wb_config_buf_params { + struct mcif_wb *mcif_wb; + struct mcif_buf_params *mcif_buf_params; + unsigned int dest_height; +}; + +struct mcif_wb_config_arb_params { + struct mcif_wb *mcif_wb; + struct mcif_arb_params *mcif_arb_params; +}; + +struct mcif_wb_enable_params { + struct mcif_wb *mcif_wb; +}; + +struct mcif_wb_disable_params { + struct mcif_wb *mcif_wb; +}; + +struct dwbc_enable_params { + struct dwbc *dwb; + struct dc_dwb_params *dwb_params; +}; + +struct dwbc_disable_params { + struct dwbc *dwb; +}; + +struct dwbc_update_params { + struct dwbc *dwb; + struct dc_dwb_params *dwb_params; +}; + +struct hubp_update_mall_sel_params { + struct hubp *hubp; + uint32_t mall_sel; + bool cache_cursor; +}; + +struct hubp_prepare_subvp_buffering_params { + struct hubp *hubp; + bool enable; +}; + +struct hubp_set_blank_en_params { + struct hubp *hubp; + bool enable; +}; + +struct hubp_disable_control_params { + struct hubp *hubp; + bool disable; +}; + +struct hubbub_soft_reset_params { + struct hubbub *hubbub; + void (*hubbub_soft_reset)(struct hubbub *hubbub, bool reset); + bool reset; +}; + +struct hubp_clk_cntl_params { + struct hubp *hubp; + bool enable; +}; + +struct hubp_init_params { + struct hubp *hubp; +}; + +struct hubp_set_vm_system_aperture_settings_params { + struct hubp *hubp; + //struct vm_system_aperture_param apt; + PHYSICAL_ADDRESS_LOC sys_default; + PHYSICAL_ADDRESS_LOC sys_low; + PHYSICAL_ADDRESS_LOC sys_high; +}; + +struct hubp_set_flip_int_params { + struct hubp *hubp; +}; + +struct dpp_dppclk_control_params { + struct dpp *dpp; + bool dppclk_div; + bool enable; +}; + +struct disable_phantom_crtc_params { + struct timing_generator *tg; +}; + +struct dpp_pg_control_params { + struct dce_hwseq *hws; + unsigned int dpp_inst; + bool power_on; +}; + +struct hubp_pg_control_params { + struct dce_hwseq *hws; + unsigned int hubp_inst; + bool power_on; +}; + +struct hubp_reset_params { + struct hubp *hubp; +}; + +struct dpp_reset_params { + struct dpp *dpp; +}; + +struct dpp_root_clock_control_params { + struct dce_hwseq *hws; + unsigned int dpp_inst; + bool clock_on; +}; + +struct dc_ip_request_cntl_params { + struct dc *dc; + bool enable; +}; + +struct dsc_pg_status_params { + struct dce_hwseq *hws; + int dsc_inst; + bool is_ungated; +}; + +struct dsc_wait_disconnect_pending_clear_params { + struct display_stream_compressor *dsc; + bool *is_ungated; +}; + +struct dsc_disable_params { + struct display_stream_compressor *dsc; + bool *is_ungated; +}; + +struct dccg_set_ref_dscclk_params { + struct dccg *dccg; + int dsc_inst; + bool *is_ungated; +}; + +struct dccg_update_dpp_dto_params { + struct dccg *dccg; + int dpp_inst; + int dppclk_khz; +}; + +struct hubp_vtg_sel_params { + struct hubp *hubp; + uint32_t otg_inst; +}; + +struct hubp_setup2_params { + struct hubp *hubp; + struct dml2_dchub_per_pipe_register_set *hubp_regs; + union dml2_global_sync_programming *global_sync; + struct dc_crtc_timing *timing; +}; + +struct hubp_setup_params { + struct hubp *hubp; + struct _vcs_dpi_display_dlg_regs_st *dlg_regs; + struct _vcs_dpi_display_ttu_regs_st *ttu_regs; + struct _vcs_dpi_display_rq_regs_st *rq_regs; + struct _vcs_dpi_display_pipe_dest_params_st *pipe_dest; +}; + +struct hubp_set_unbounded_requesting_params { + struct hubp *hubp; + bool unbounded_req; +}; + +struct hubp_setup_interdependent2_params { + struct hubp *hubp; + struct dml2_dchub_per_pipe_register_set *hubp_regs; +}; + +struct hubp_setup_interdependent_params { + struct hubp *hubp; + struct _vcs_dpi_display_dlg_regs_st *dlg_regs; + struct _vcs_dpi_display_ttu_regs_st *ttu_regs; +}; + +struct dpp_set_cursor_matrix_params { + struct dpp *dpp; + enum dc_color_space color_space; + struct dc_csc_transform *cursor_csc_color_matrix; +}; + +struct mpc_update_mpcc_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct dpp_set_scaler_params { + struct dpp *dpp; + const struct scaler_data *scl_data; +}; + +struct hubp_mem_program_viewport_params { + struct hubp *hubp; + const struct rect *viewport; + const struct rect *viewport_c; +}; + +struct hubp_program_mcache_id_and_split_coordinate_params { + struct hubp *hubp; + struct mcache_regs_struct *mcache_regs; +}; + +struct abort_cursor_offload_update_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct set_cursor_attribute_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct set_cursor_position_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct set_cursor_sdr_white_level_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; +}; + +struct program_output_csc_params { + struct dc *dc; + struct pipe_ctx *pipe_ctx; + enum dc_color_space colorspace; + uint16_t *matrix; + int opp_id; +}; + +struct hubp_set_blank_params { + struct hubp *hubp; + bool blank; +}; + +struct phantom_hubp_post_enable_params { + struct hubp *hubp; +}; + union block_sequence_params { struct update_plane_addr_params update_plane_addr_params; struct subvp_pipe_control_lock_fast_params subvp_pipe_control_lock_fast_params; @@ -173,7 +752,108 @@ union block_sequence_params { struct set_ocsc_default_params set_ocsc_default_params; struct subvp_save_surf_addr subvp_save_surf_addr; struct wait_for_dcc_meta_propagation_params wait_for_dcc_meta_propagation_params; - struct fams2_global_control_lock_fast_params fams2_global_control_lock_fast_params; + struct dmub_hw_control_lock_fast_params dmub_hw_control_lock_fast_params; + struct program_surface_config_params program_surface_config_params; + struct program_mcache_id_and_split_coordinate program_mcache_id_and_split_coordinate; + struct program_cursor_update_now_params program_cursor_update_now_params; + struct hubp_wait_pipe_read_start_params hubp_wait_pipe_read_start_params; + struct apply_update_flags_for_phantom_params apply_update_flags_for_phantom_params; + struct update_phantom_vp_position_params update_phantom_vp_position_params; + struct set_odm_combine_params set_odm_combine_params; + struct set_odm_bypass_params set_odm_bypass_params; + struct opp_pipe_clock_control_params opp_pipe_clock_control_params; + struct opp_program_left_edge_extra_pixel_params opp_program_left_edge_extra_pixel_params; + struct dccg_set_dto_dscclk_params dccg_set_dto_dscclk_params; + struct dsc_set_config_params dsc_set_config_params; + struct dsc_enable_params dsc_enable_params; + struct tg_set_dsc_config_params tg_set_dsc_config_params; + struct dsc_disconnect_params dsc_disconnect_params; + struct dsc_read_state_params dsc_read_state_params; + struct dsc_calculate_and_set_config_params dsc_calculate_and_set_config_params; + struct dsc_enable_with_opp_params dsc_enable_with_opp_params; + struct program_tg_params program_tg_params; + struct tg_program_global_sync_params tg_program_global_sync_params; + struct tg_wait_for_state_params tg_wait_for_state_params; + struct tg_set_vtg_params_params tg_set_vtg_params_params; + struct tg_setup_vertical_interrupt2_params tg_setup_vertical_interrupt2_params; + struct dpp_set_hdr_multiplier_params dpp_set_hdr_multiplier_params; + struct tg_set_gsl_params tg_set_gsl_params; + struct tg_set_gsl_source_select_params tg_set_gsl_source_select_params; + struct setup_vupdate_interrupt_params setup_vupdate_interrupt_params; + struct program_det_size_params program_det_size_params; + struct program_det_segments_params program_det_segments_params; + struct update_dchubp_dpp_params update_dchubp_dpp_params; + struct opp_set_dyn_expansion_params opp_set_dyn_expansion_params; + struct opp_program_fmt_params opp_program_fmt_params; + struct opp_program_bit_depth_reduction_params opp_program_bit_depth_reduction_params; + struct opp_set_disp_pattern_generator_params opp_set_disp_pattern_generator_params; + struct set_abm_pipe_params set_abm_pipe_params; + struct set_abm_level_params set_abm_level_params; + struct set_abm_immediate_disable_params set_abm_immediate_disable_params; + struct set_disp_pattern_generator_params set_disp_pattern_generator_params; + struct mpc_remove_mpcc_params mpc_remove_mpcc_params; + struct opp_set_mpcc_disconnect_pending_params opp_set_mpcc_disconnect_pending_params; + struct dc_set_optimized_required_params dc_set_optimized_required_params; + struct hubp_disconnect_params hubp_disconnect_params; + struct hubbub_force_pstate_change_control_params hubbub_force_pstate_change_control_params; + struct tg_enable_crtc_params tg_enable_crtc_params; + struct hubp_wait_flip_pending_params hubp_wait_flip_pending_params; + struct tg_wait_double_buffer_pending_params tg_wait_double_buffer_pending_params; + struct update_force_pstate_params update_force_pstate_params; + struct hubbub_apply_dedcn21_147_wa_params hubbub_apply_dedcn21_147_wa_params; + struct hubbub_allow_self_refresh_control_params hubbub_allow_self_refresh_control_params; + struct tg_get_frame_count_params tg_get_frame_count_params; + struct mpc_set_dwb_mux_params mpc_set_dwb_mux_params; + struct mpc_disable_dwb_mux_params mpc_disable_dwb_mux_params; + struct mcif_wb_config_buf_params mcif_wb_config_buf_params; + struct mcif_wb_config_arb_params mcif_wb_config_arb_params; + struct mcif_wb_enable_params mcif_wb_enable_params; + struct mcif_wb_disable_params mcif_wb_disable_params; + struct dwbc_enable_params dwbc_enable_params; + struct dwbc_disable_params dwbc_disable_params; + struct dwbc_update_params dwbc_update_params; + struct hubp_update_mall_sel_params hubp_update_mall_sel_params; + struct hubp_prepare_subvp_buffering_params hubp_prepare_subvp_buffering_params; + struct hubp_set_blank_en_params hubp_set_blank_en_params; + struct hubp_disable_control_params hubp_disable_control_params; + struct hubbub_soft_reset_params hubbub_soft_reset_params; + struct hubp_clk_cntl_params hubp_clk_cntl_params; + struct hubp_init_params hubp_init_params; + struct hubp_set_vm_system_aperture_settings_params hubp_set_vm_system_aperture_settings_params; + struct hubp_set_flip_int_params hubp_set_flip_int_params; + struct dpp_dppclk_control_params dpp_dppclk_control_params; + struct disable_phantom_crtc_params disable_phantom_crtc_params; + struct dpp_pg_control_params dpp_pg_control_params; + struct hubp_pg_control_params hubp_pg_control_params; + struct hubp_reset_params hubp_reset_params; + struct dpp_reset_params dpp_reset_params; + struct dpp_root_clock_control_params dpp_root_clock_control_params; + struct dc_ip_request_cntl_params dc_ip_request_cntl_params; + struct dsc_pg_status_params dsc_pg_status_params; + struct dsc_wait_disconnect_pending_clear_params dsc_wait_disconnect_pending_clear_params; + struct dsc_disable_params dsc_disable_params; + struct dccg_set_ref_dscclk_params dccg_set_ref_dscclk_params; + struct dccg_update_dpp_dto_params dccg_update_dpp_dto_params; + struct hubp_vtg_sel_params hubp_vtg_sel_params; + struct hubp_setup2_params hubp_setup2_params; + struct hubp_setup_params hubp_setup_params; + struct hubp_set_unbounded_requesting_params hubp_set_unbounded_requesting_params; + struct hubp_setup_interdependent2_params hubp_setup_interdependent2_params; + struct hubp_setup_interdependent_params hubp_setup_interdependent_params; + struct dpp_set_cursor_matrix_params dpp_set_cursor_matrix_params; + struct mpc_update_mpcc_params mpc_update_mpcc_params; + struct mpc_update_blending_params mpc_update_blending_params; + struct mpc_assert_idle_mpcc_params mpc_assert_idle_mpcc_params; + struct mpc_insert_plane_params mpc_insert_plane_params; + struct dpp_set_scaler_params dpp_set_scaler_params; + struct hubp_mem_program_viewport_params hubp_mem_program_viewport_params; + struct abort_cursor_offload_update_params abort_cursor_offload_update_params; + struct set_cursor_attribute_params set_cursor_attribute_params; + struct set_cursor_position_params set_cursor_position_params; + struct set_cursor_sdr_white_level_params set_cursor_sdr_white_level_params; + struct program_output_csc_params program_output_csc_params; + struct hubp_set_blank_params hubp_set_blank_params; + struct phantom_hubp_post_enable_params phantom_hubp_post_enable_params; }; enum block_sequence_func { @@ -189,13 +869,111 @@ enum block_sequence_func { DPP_SETUP_DPP, DPP_PROGRAM_BIAS_AND_SCALE, DPP_SET_OUTPUT_TRANSFER_FUNC, + DPP_SET_HDR_MULTIPLIER, MPC_UPDATE_VISUAL_CONFIRM, MPC_POWER_ON_MPC_MEM_PWR, MPC_SET_OUTPUT_CSC, MPC_SET_OCSC_DEFAULT, DMUB_SUBVP_SAVE_SURF_ADDR, HUBP_WAIT_FOR_DCC_META_PROP, - DMUB_FAMS2_GLOBAL_CONTROL_LOCK_FAST, + DMUB_HW_CONTROL_LOCK_FAST, + HUBP_PROGRAM_SURFACE_CONFIG, + HUBP_PROGRAM_MCACHE_ID, + PROGRAM_CURSOR_UPDATE_NOW, + HUBP_WAIT_PIPE_READ_START, + HWS_APPLY_UPDATE_FLAGS_FOR_PHANTOM, + HWS_UPDATE_PHANTOM_VP_POSITION, + OPTC_SET_ODM_COMBINE, + OPTC_SET_ODM_BYPASS, + OPP_PIPE_CLOCK_CONTROL, + OPP_PROGRAM_LEFT_EDGE_EXTRA_PIXEL, + DCCG_SET_DTO_DSCCLK, + DSC_SET_CONFIG, + DSC_ENABLE, + TG_SET_DSC_CONFIG, + DSC_DISCONNECT, + DSC_READ_STATE, + DSC_CALCULATE_AND_SET_CONFIG, + DSC_ENABLE_WITH_OPP, + TG_PROGRAM_GLOBAL_SYNC, + TG_WAIT_FOR_STATE, + TG_SET_VTG_PARAMS, + TG_SETUP_VERTICAL_INTERRUPT2, + HUBP_PROGRAM_DET_SIZE, + HUBP_PROGRAM_DET_SEGMENTS, + OPP_SET_DYN_EXPANSION, + OPP_PROGRAM_FMT, + OPP_PROGRAM_BIT_DEPTH_REDUCTION, + OPP_SET_DISP_PATTERN_GENERATOR, + ABM_SET_PIPE, + ABM_SET_LEVEL, + ABM_SET_IMMEDIATE_DISABLE, + MPC_REMOVE_MPCC, + OPP_SET_MPCC_DISCONNECT_PENDING, + DC_SET_OPTIMIZED_REQUIRED, + HUBP_DISCONNECT, + HUBBUB_FORCE_PSTATE_CHANGE_CONTROL, + TG_ENABLE_CRTC, + TG_SET_GSL, + TG_SET_GSL_SOURCE_SELECT, + HUBP_WAIT_FLIP_PENDING, + TG_WAIT_DOUBLE_BUFFER_PENDING, + UPDATE_FORCE_PSTATE, + PROGRAM_MALL_PIPE_CONFIG, + HUBBUB_APPLY_DEDCN21_147_WA, + HUBBUB_ALLOW_SELF_REFRESH_CONTROL, + TG_GET_FRAME_COUNT, + MPC_SET_DWB_MUX, + MPC_DISABLE_DWB_MUX, + MCIF_WB_CONFIG_BUF, + MCIF_WB_CONFIG_ARB, + MCIF_WB_ENABLE, + MCIF_WB_DISABLE, + DWBC_ENABLE, + DWBC_DISABLE, + DWBC_UPDATE, + HUBP_UPDATE_MALL_SEL, + HUBP_PREPARE_SUBVP_BUFFERING, + HUBP_SET_BLANK_EN, + HUBP_DISABLE_CONTROL, + HUBBUB_SOFT_RESET, + HUBP_CLK_CNTL, + HUBP_INIT, + HUBP_SET_VM_SYSTEM_APERTURE_SETTINGS, + HUBP_SET_FLIP_INT, + DPP_DPPCLK_CONTROL, + DISABLE_PHANTOM_CRTC, + DSC_PG_STATUS, + DSC_WAIT_DISCONNECT_PENDING_CLEAR, + DSC_DISABLE, + DCCG_SET_REF_DSCCLK, + DPP_PG_CONTROL, + HUBP_PG_CONTROL, + HUBP_RESET, + DPP_RESET, + DPP_ROOT_CLOCK_CONTROL, + DC_IP_REQUEST_CNTL, + DCCG_UPDATE_DPP_DTO, + HUBP_VTG_SEL, + HUBP_SETUP2, + HUBP_SETUP, + HUBP_SET_UNBOUNDED_REQUESTING, + HUBP_SETUP_INTERDEPENDENT2, + HUBP_SETUP_INTERDEPENDENT, + DPP_SET_CURSOR_MATRIX, + MPC_UPDATE_BLENDING, + MPC_ASSERT_IDLE_MPCC, + MPC_INSERT_PLANE, + DPP_SET_SCALER, + HUBP_MEM_PROGRAM_VIEWPORT, + ABORT_CURSOR_OFFLOAD_UPDATE, + SET_CURSOR_ATTRIBUTE, + SET_CURSOR_POSITION, + SET_CURSOR_SDR_WHITE_LEVEL, + PROGRAM_OUTPUT_CSC, + HUBP_SET_LEGACY_TILING_COMPAT_LEVEL, + HUBP_SET_BLANK, + PHANTOM_HUBP_POST_ENABLE, /* This must be the last value in this enum, add new ones above */ HWSS_BLOCK_SEQUENCE_FUNC_COUNT }; @@ -205,6 +983,11 @@ struct block_sequence { enum block_sequence_func func; }; +struct block_sequence_state { + struct block_sequence *steps; + unsigned int *num_steps; +}; + #define MAX_HWSS_BLOCK_SEQUENCE_SIZE (HWSS_BLOCK_SEQUENCE_FUNC_COUNT * MAX_PIPES) struct hw_sequencer_funcs { @@ -222,6 +1005,8 @@ struct hw_sequencer_funcs { enum dc_status (*apply_ctx_to_hw)(struct dc *dc, struct dc_state *context); void (*disable_plane)(struct dc *dc, struct dc_state *state, struct pipe_ctx *pipe_ctx); + void (*disable_plane_sequence)(struct dc *dc, struct dc_state *state, struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); void (*disable_pixel_data)(struct dc *dc, struct pipe_ctx *pipe_ctx, bool blank); void (*apply_ctx_for_surface)(struct dc *dc, const struct dc_stream_state *stream, @@ -239,6 +1024,10 @@ struct hw_sequencer_funcs { void (*wait_for_mpcc_disconnect)(struct dc *dc, struct resource_pool *res_pool, struct pipe_ctx *pipe_ctx); + void (*wait_for_mpcc_disconnect_sequence)(struct dc *dc, + struct resource_pool *res_pool, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); void (*edp_backlight_control)( struct dc_link *link, bool enable); @@ -310,6 +1099,13 @@ struct hw_sequencer_funcs { void (*set_cursor_position)(struct pipe_ctx *pipe); void (*set_cursor_attribute)(struct pipe_ctx *pipe); void (*set_cursor_sdr_white_level)(struct pipe_ctx *pipe); + void (*abort_cursor_offload_update)(struct dc *dc, const struct pipe_ctx *pipe); + void (*begin_cursor_offload_update)(struct dc *dc, const struct pipe_ctx *pipe); + void (*commit_cursor_offload_update)(struct dc *dc, const struct pipe_ctx *pipe); + void (*update_cursor_offload_pipe)(struct dc *dc, const struct pipe_ctx *pipe); + void (*notify_cursor_offload_drr_update)(struct dc *dc, struct dc_state *context, + const struct dc_stream_state *stream); + void (*program_cursor_offload_now)(struct dc *dc, const struct pipe_ctx *pipe); /* Colour Related */ void (*program_gamut_remap)(struct pipe_ctx *pipe_ctx); @@ -452,13 +1248,13 @@ struct hw_sequencer_funcs { const struct dc_state *new_ctx); void (*wait_for_dcc_meta_propagation)(const struct dc *dc, const struct pipe_ctx *top_pipe_to_program); - void (*fams2_global_control_lock)(struct dc *dc, + void (*dmub_hw_control_lock)(struct dc *dc, struct dc_state *context, bool lock); void (*fams2_update_config)(struct dc *dc, struct dc_state *context, bool enable); - void (*fams2_global_control_lock_fast)(union block_sequence_params *params); + void (*dmub_hw_control_lock_fast)(union block_sequence_params *params); void (*set_long_vtotal)(struct pipe_ctx **pipe_ctx, int num_pipes, uint32_t v_total_min, uint32_t v_total_max); void (*program_outstanding_updates)(struct dc *dc, struct dc_state *context); @@ -471,11 +1267,23 @@ struct hw_sequencer_funcs { void (*enable_plane)(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context); + void (*enable_plane_sequence)(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); void (*update_dchubp_dpp)(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context); + void (*update_dchubp_dpp_sequence)(struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); void (*post_unlock_reset_opp)(struct dc *dc, struct pipe_ctx *opp_head); + void (*post_unlock_reset_opp_sequence)( + struct dc *dc, + struct pipe_ctx *opp_head, + struct block_sequence_state *seq_state); void (*get_underflow_debug_data)(const struct dc *dc, struct timing_generator *tg, struct dc_underflow_debug_data *out_data); @@ -588,4 +1396,630 @@ void hwss_set_ocsc_default(union block_sequence_params *params); void hwss_subvp_save_surf_addr(union block_sequence_params *params); +void hwss_program_surface_config(union block_sequence_params *params); + +void hwss_program_mcache_id_and_split_coordinate(union block_sequence_params *params); + +void hwss_set_odm_combine(union block_sequence_params *params); + +void hwss_set_odm_bypass(union block_sequence_params *params); + +void hwss_opp_pipe_clock_control(union block_sequence_params *params); + +void hwss_opp_program_left_edge_extra_pixel(union block_sequence_params *params); + +void hwss_blank_pixel_data(union block_sequence_params *params); + +void hwss_dccg_set_dto_dscclk(union block_sequence_params *params); + +void hwss_dsc_set_config(union block_sequence_params *params); + +void hwss_dsc_enable(union block_sequence_params *params); + +void hwss_tg_set_dsc_config(union block_sequence_params *params); + +void hwss_dsc_disconnect(union block_sequence_params *params); + +void hwss_dsc_read_state(union block_sequence_params *params); + +void hwss_dsc_calculate_and_set_config(union block_sequence_params *params); + +void hwss_dsc_enable_with_opp(union block_sequence_params *params); + +void hwss_program_tg(union block_sequence_params *params); + +void hwss_tg_program_global_sync(union block_sequence_params *params); + +void hwss_tg_wait_for_state(union block_sequence_params *params); + +void hwss_tg_set_vtg_params(union block_sequence_params *params); + +void hwss_tg_setup_vertical_interrupt2(union block_sequence_params *params); + +void hwss_dpp_set_hdr_multiplier(union block_sequence_params *params); + +void hwss_program_det_size(union block_sequence_params *params); + +void hwss_program_det_segments(union block_sequence_params *params); + +void hwss_opp_set_dyn_expansion(union block_sequence_params *params); + +void hwss_opp_program_fmt(union block_sequence_params *params); + +void hwss_opp_program_bit_depth_reduction(union block_sequence_params *params); + +void hwss_opp_set_disp_pattern_generator(union block_sequence_params *params); + +void hwss_set_abm_pipe(union block_sequence_params *params); + +void hwss_set_abm_level(union block_sequence_params *params); + +void hwss_set_abm_immediate_disable(union block_sequence_params *params); + +void hwss_mpc_remove_mpcc(union block_sequence_params *params); + +void hwss_opp_set_mpcc_disconnect_pending(union block_sequence_params *params); + +void hwss_dc_set_optimized_required(union block_sequence_params *params); + +void hwss_hubp_disconnect(union block_sequence_params *params); + +void hwss_hubbub_force_pstate_change_control(union block_sequence_params *params); + +void hwss_tg_enable_crtc(union block_sequence_params *params); + +void hwss_tg_set_gsl(union block_sequence_params *params); + +void hwss_tg_set_gsl_source_select(union block_sequence_params *params); + +void hwss_hubp_wait_flip_pending(union block_sequence_params *params); + +void hwss_tg_wait_double_buffer_pending(union block_sequence_params *params); + +void hwss_update_force_pstate(union block_sequence_params *params); + +void hwss_hubbub_apply_dedcn21_147_wa(union block_sequence_params *params); + +void hwss_hubbub_allow_self_refresh_control(union block_sequence_params *params); + +void hwss_tg_get_frame_count(union block_sequence_params *params); + +void hwss_mpc_set_dwb_mux(union block_sequence_params *params); + +void hwss_mpc_disable_dwb_mux(union block_sequence_params *params); + +void hwss_mcif_wb_config_buf(union block_sequence_params *params); + +void hwss_mcif_wb_config_arb(union block_sequence_params *params); + +void hwss_mcif_wb_enable(union block_sequence_params *params); + +void hwss_mcif_wb_disable(union block_sequence_params *params); + +void hwss_dwbc_enable(union block_sequence_params *params); + +void hwss_dwbc_disable(union block_sequence_params *params); + +void hwss_dwbc_update(union block_sequence_params *params); + +void hwss_hubp_update_mall_sel(union block_sequence_params *params); + +void hwss_hubp_prepare_subvp_buffering(union block_sequence_params *params); + +void hwss_hubp_set_blank_en(union block_sequence_params *params); + +void hwss_hubp_disable_control(union block_sequence_params *params); + +void hwss_hubbub_soft_reset(union block_sequence_params *params); + +void hwss_hubp_clk_cntl(union block_sequence_params *params); + +void hwss_hubp_init(union block_sequence_params *params); + +void hwss_hubp_set_vm_system_aperture_settings(union block_sequence_params *params); + +void hwss_hubp_set_flip_int(union block_sequence_params *params); + +void hwss_dpp_dppclk_control(union block_sequence_params *params); + +void hwss_disable_phantom_crtc(union block_sequence_params *params); + +void hwss_dsc_pg_status(union block_sequence_params *params); + +void hwss_dsc_wait_disconnect_pending_clear(union block_sequence_params *params); + +void hwss_dsc_disable(union block_sequence_params *params); + +void hwss_dccg_set_ref_dscclk(union block_sequence_params *params); + +void hwss_dpp_pg_control(union block_sequence_params *params); + +void hwss_hubp_pg_control(union block_sequence_params *params); + +void hwss_hubp_reset(union block_sequence_params *params); + +void hwss_dpp_reset(union block_sequence_params *params); + +void hwss_dpp_root_clock_control(union block_sequence_params *params); + +void hwss_dc_ip_request_cntl(union block_sequence_params *params); + +void hwss_dccg_update_dpp_dto(union block_sequence_params *params); + +void hwss_hubp_vtg_sel(union block_sequence_params *params); + +void hwss_hubp_setup2(union block_sequence_params *params); + +void hwss_hubp_setup(union block_sequence_params *params); + +void hwss_hubp_set_unbounded_requesting(union block_sequence_params *params); + +void hwss_hubp_setup_interdependent2(union block_sequence_params *params); + +void hwss_hubp_setup_interdependent(union block_sequence_params *params); + +void hwss_dpp_set_cursor_matrix(union block_sequence_params *params); + +void hwss_mpc_update_mpcc(union block_sequence_params *params); + +void hwss_mpc_update_blending(union block_sequence_params *params); + +void hwss_mpc_assert_idle_mpcc(union block_sequence_params *params); + +void hwss_mpc_insert_plane(union block_sequence_params *params); + +void hwss_dpp_set_scaler(union block_sequence_params *params); + +void hwss_hubp_mem_program_viewport(union block_sequence_params *params); + +void hwss_abort_cursor_offload_update(union block_sequence_params *params); + +void hwss_set_cursor_attribute(union block_sequence_params *params); + +void hwss_set_cursor_position(union block_sequence_params *params); + +void hwss_set_cursor_sdr_white_level(union block_sequence_params *params); + +void hwss_program_output_csc(union block_sequence_params *params); + +void hwss_hubp_set_legacy_tiling_compat_level(union block_sequence_params *params); + +void hwss_hubp_set_blank(union block_sequence_params *params); + +void hwss_phantom_hubp_post_enable(union block_sequence_params *params); + +void hwss_add_optc_pipe_control_lock(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx, bool lock); + +void hwss_add_hubp_set_flip_control_gsl(struct block_sequence_state *seq_state, + struct hubp *hubp, bool flip_immediate); + +void hwss_add_hubp_program_triplebuffer(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx, bool enableTripleBuffer); + +void hwss_add_hubp_update_plane_addr(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx); + +void hwss_add_dpp_set_input_transfer_func(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_plane_state *plane_state); + +void hwss_add_dpp_program_gamut_remap(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_dpp_program_bias_and_scale(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_optc_program_manual_trigger(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_dpp_set_output_transfer_func(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_stream_state *stream); + +void hwss_add_mpc_update_visual_confirm(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx, int mpcc_id); + +void hwss_add_mpc_power_on_mpc_mem_pwr(struct block_sequence_state *seq_state, + struct mpc *mpc, int mpcc_id, bool power_on); + +void hwss_add_mpc_set_output_csc(struct block_sequence_state *seq_state, + struct mpc *mpc, int opp_id, const uint16_t *regval, enum mpc_output_csc_mode ocsc_mode); + +void hwss_add_mpc_set_ocsc_default(struct block_sequence_state *seq_state, + struct mpc *mpc, int opp_id, enum dc_color_space colorspace, enum mpc_output_csc_mode ocsc_mode); + +void hwss_add_dmub_send_dmcub_cmd(struct block_sequence_state *seq_state, + struct dc_context *ctx, union dmub_rb_cmd *cmd, enum dm_dmub_wait_type wait_type); + +void hwss_add_dmub_subvp_save_surf_addr(struct block_sequence_state *seq_state, + struct dc_dmub_srv *dc_dmub_srv, struct dc_plane_address *addr, uint8_t subvp_index); + +void hwss_add_hubp_wait_for_dcc_meta_prop(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *top_pipe_to_program); + +void hwss_add_hubp_wait_pipe_read_start(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_hws_apply_update_flags_for_phantom(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_hws_update_phantom_vp_position(struct block_sequence_state *seq_state, + struct dc *dc, struct dc_state *context, struct pipe_ctx *pipe_ctx); + +void hwss_add_optc_set_odm_combine(struct block_sequence_state *seq_state, + struct timing_generator *tg, int opp_inst[MAX_PIPES], int opp_head_count, + int odm_slice_width, int last_odm_slice_width); + +void hwss_add_optc_set_odm_bypass(struct block_sequence_state *seq_state, + struct timing_generator *optc, struct dc_crtc_timing *timing); + +void hwss_add_tg_program_global_sync(struct block_sequence_state *seq_state, + struct timing_generator *tg, + int vready_offset, + unsigned int vstartup_lines, + unsigned int vupdate_offset_pixels, + unsigned int vupdate_vupdate_width_pixels, + unsigned int pstate_keepout_start_lines); + +void hwss_add_tg_wait_for_state(struct block_sequence_state *seq_state, + struct timing_generator *tg, enum crtc_state state); + +void hwss_add_tg_set_vtg_params(struct block_sequence_state *seq_state, + struct timing_generator *tg, struct dc_crtc_timing *dc_crtc_timing, bool program_fp2); + +void hwss_add_tg_setup_vertical_interrupt2(struct block_sequence_state *seq_state, + struct timing_generator *tg, int start_line); + +void hwss_add_dpp_set_hdr_multiplier(struct block_sequence_state *seq_state, + struct dpp *dpp, uint32_t hw_mult); + +void hwss_add_hubp_program_det_size(struct block_sequence_state *seq_state, + struct hubbub *hubbub, unsigned int hubp_inst, unsigned int det_buffer_size_kb); + +void hwss_add_hubp_program_mcache_id(struct block_sequence_state *seq_state, + struct hubp *hubp, struct dml2_hubp_pipe_mcache_regs *mcache_regs); + +void hwss_add_hubbub_force_pstate_change_control(struct block_sequence_state *seq_state, + struct hubbub *hubbub, bool enable, bool wait); + +void hwss_add_hubp_program_det_segments(struct block_sequence_state *seq_state, + struct hubbub *hubbub, unsigned int hubp_inst, unsigned int det_size); + +void hwss_add_opp_set_dyn_expansion(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, enum dc_color_space color_sp, + enum dc_color_depth color_dpth, enum signal_type signal); + +void hwss_add_opp_program_fmt(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, struct bit_depth_reduction_params *fmt_bit_depth, + struct clamping_and_pixel_encoding_params *clamping); + +void hwss_add_abm_set_pipe(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx); + +void hwss_add_abm_set_level(struct block_sequence_state *seq_state, + struct abm *abm, uint32_t abm_level); + +void hwss_add_tg_enable_crtc(struct block_sequence_state *seq_state, + struct timing_generator *tg); + +void hwss_add_hubp_wait_flip_pending(struct block_sequence_state *seq_state, + struct hubp *hubp, unsigned int timeout_us, unsigned int polling_interval_us); + +void hwss_add_tg_wait_double_buffer_pending(struct block_sequence_state *seq_state, + struct timing_generator *tg, unsigned int timeout_us, unsigned int polling_interval_us); + +void hwss_add_dccg_set_dto_dscclk(struct block_sequence_state *seq_state, + struct dccg *dccg, int inst, int num_slices_h); + +void hwss_add_dsc_calculate_and_set_config(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx, bool enable, int opp_cnt); + +void hwss_add_mpc_remove_mpcc(struct block_sequence_state *seq_state, + struct mpc *mpc, struct mpc_tree *mpc_tree_params, struct mpcc *mpcc_to_remove); + +void hwss_add_opp_set_mpcc_disconnect_pending(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, int mpcc_inst, bool pending); + +void hwss_add_hubp_disconnect(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_dsc_enable_with_opp(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_dsc_disconnect(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc); + +void hwss_add_dc_set_optimized_required(struct block_sequence_state *seq_state, + struct dc *dc, bool optimized_required); + +void hwss_add_abm_set_immediate_disable(struct block_sequence_state *seq_state, + struct dc *dc, struct pipe_ctx *pipe_ctx); + +void hwss_add_opp_set_disp_pattern_generator(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + enum controller_dp_test_pattern test_pattern, + enum controller_dp_color_space color_space, + enum dc_color_depth color_depth, + struct tg_color solid_color, + bool use_solid_color, + int width, + int height, + int offset); + +void hwss_add_opp_program_bit_depth_reduction(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + bool use_default_params, + struct pipe_ctx *pipe_ctx); + +void hwss_add_dc_ip_request_cntl(struct block_sequence_state *seq_state, + struct dc *dc, + bool enable); + +void hwss_add_dwbc_update(struct block_sequence_state *seq_state, + struct dwbc *dwb, + struct dc_dwb_params *dwb_params); + +void hwss_add_mcif_wb_config_buf(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb, + struct mcif_buf_params *mcif_buf_params, + unsigned int dest_height); + +void hwss_add_mcif_wb_config_arb(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb, + struct mcif_arb_params *mcif_arb_params); + +void hwss_add_mcif_wb_enable(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb); + +void hwss_add_mcif_wb_disable(struct block_sequence_state *seq_state, + struct mcif_wb *mcif_wb); + +void hwss_add_mpc_set_dwb_mux(struct block_sequence_state *seq_state, + struct mpc *mpc, + int dwb_id, + int mpcc_id); + +void hwss_add_mpc_disable_dwb_mux(struct block_sequence_state *seq_state, + struct mpc *mpc, + unsigned int dwb_id); + +void hwss_add_dwbc_enable(struct block_sequence_state *seq_state, + struct dwbc *dwb, + struct dc_dwb_params *dwb_params); + +void hwss_add_dwbc_disable(struct block_sequence_state *seq_state, + struct dwbc *dwb); + +void hwss_add_tg_set_gsl(struct block_sequence_state *seq_state, + struct timing_generator *tg, + struct gsl_params gsl); + +void hwss_add_tg_set_gsl_source_select(struct block_sequence_state *seq_state, + struct timing_generator *tg, + int group_idx, + uint32_t gsl_ready_signal); + +void hwss_add_hubp_update_mall_sel(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint32_t mall_sel, + bool cache_cursor); + +void hwss_add_hubp_prepare_subvp_buffering(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable); + +void hwss_add_hubp_set_blank_en(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable); + +void hwss_add_hubp_disable_control(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool disable); + +void hwss_add_hubbub_soft_reset(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + void (*hubbub_soft_reset)(struct hubbub *hubbub, bool reset), + bool reset); + +void hwss_add_hubp_clk_cntl(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool enable); + +void hwss_add_dpp_dppclk_control(struct block_sequence_state *seq_state, + struct dpp *dpp, + bool dppclk_div, + bool enable); + +void hwss_add_disable_phantom_crtc(struct block_sequence_state *seq_state, + struct timing_generator *tg); + +void hwss_add_dsc_pg_status(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + int dsc_inst, + bool is_ungated); + +void hwss_add_dsc_wait_disconnect_pending_clear(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc, + bool *is_ungated); + +void hwss_add_dsc_disable(struct block_sequence_state *seq_state, + struct display_stream_compressor *dsc, + bool *is_ungated); + +void hwss_add_dccg_set_ref_dscclk(struct block_sequence_state *seq_state, + struct dccg *dccg, + int dsc_inst, + bool *is_ungated); + +void hwss_add_dpp_root_clock_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int dpp_inst, + bool clock_on); + +void hwss_add_dpp_pg_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int dpp_inst, + bool power_on); + +void hwss_add_hubp_pg_control(struct block_sequence_state *seq_state, + struct dce_hwseq *hws, + unsigned int hubp_inst, + bool power_on); + +void hwss_add_hubp_set_blank(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool blank); + +void hwss_add_hubp_init(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_hubp_reset(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_dpp_reset(struct block_sequence_state *seq_state, + struct dpp *dpp); + +void hwss_add_opp_pipe_clock_control(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + bool enable); + +void hwss_add_hubp_set_vm_system_aperture_settings(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint64_t sys_default, + uint64_t sys_low, + uint64_t sys_high); + +void hwss_add_hubp_set_flip_int(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_dccg_update_dpp_dto(struct block_sequence_state *seq_state, + struct dccg *dccg, + int dpp_inst, + int dppclk_khz); + +void hwss_add_hubp_vtg_sel(struct block_sequence_state *seq_state, + struct hubp *hubp, + uint32_t otg_inst); + +void hwss_add_hubp_setup2(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct dml2_dchub_per_pipe_register_set *hubp_regs, + union dml2_global_sync_programming *global_sync, + struct dc_crtc_timing *timing); + +void hwss_add_hubp_setup(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct _vcs_dpi_display_dlg_regs_st *dlg_regs, + struct _vcs_dpi_display_ttu_regs_st *ttu_regs, + struct _vcs_dpi_display_rq_regs_st *rq_regs, + struct _vcs_dpi_display_pipe_dest_params_st *pipe_dest); + +void hwss_add_hubp_set_unbounded_requesting(struct block_sequence_state *seq_state, + struct hubp *hubp, + bool unbounded_req); + +void hwss_add_hubp_setup_interdependent2(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct dml2_dchub_per_pipe_register_set *hubp_regs); + +void hwss_add_hubp_setup_interdependent(struct block_sequence_state *seq_state, + struct hubp *hubp, + struct _vcs_dpi_display_dlg_regs_st *dlg_regs, + struct _vcs_dpi_display_ttu_regs_st *ttu_regs); +void hwss_add_hubp_program_surface_config(struct block_sequence_state *seq_state, + struct hubp *hubp, + enum surface_pixel_format format, + struct dc_tiling_info *tiling_info, + struct plane_size plane_size, + enum dc_rotation_angle rotation, + struct dc_plane_dcc_param *dcc, + bool horizontal_mirror, + int compat_level); + +void hwss_add_dpp_setup_dpp(struct block_sequence_state *seq_state, + struct pipe_ctx *pipe_ctx); + +void hwss_add_dpp_set_cursor_matrix(struct block_sequence_state *seq_state, + struct dpp *dpp, + enum dc_color_space color_space, + struct dc_csc_transform *cursor_csc_color_matrix); + +void hwss_add_mpc_update_blending(struct block_sequence_state *seq_state, + struct mpc *mpc, + struct mpcc_blnd_cfg blnd_cfg, + int mpcc_id); + +void hwss_add_mpc_assert_idle_mpcc(struct block_sequence_state *seq_state, + struct mpc *mpc, + int mpcc_id); + +void hwss_add_mpc_insert_plane(struct block_sequence_state *seq_state, + struct mpc *mpc, + struct mpc_tree *mpc_tree_params, + struct mpcc_blnd_cfg blnd_cfg, + struct mpcc_sm_cfg *sm_cfg, + struct mpcc *insert_above_mpcc, + int dpp_id, + int mpcc_id); + +void hwss_add_dpp_set_scaler(struct block_sequence_state *seq_state, + struct dpp *dpp, + const struct scaler_data *scl_data); + +void hwss_add_hubp_mem_program_viewport(struct block_sequence_state *seq_state, + struct hubp *hubp, + const struct rect *viewport, + const struct rect *viewport_c); + +void hwss_add_abort_cursor_offload_update(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx); + +void hwss_add_set_cursor_attribute(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx); + +void hwss_add_set_cursor_position(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx); + +void hwss_add_set_cursor_sdr_white_level(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx); + +void hwss_add_program_output_csc(struct block_sequence_state *seq_state, + struct dc *dc, + struct pipe_ctx *pipe_ctx, + enum dc_color_space colorspace, + uint16_t *matrix, + int opp_id); + +void hwss_add_phantom_hubp_post_enable(struct block_sequence_state *seq_state, + struct hubp *hubp); + +void hwss_add_update_force_pstate(struct block_sequence_state *seq_state, + struct dc *dc, + struct dc_state *context); + +void hwss_add_hubbub_apply_dedcn21_147_wa(struct block_sequence_state *seq_state, + struct hubbub *hubbub); + +void hwss_add_hubbub_allow_self_refresh_control(struct block_sequence_state *seq_state, + struct hubbub *hubbub, + bool allow, + bool *disallow_self_refresh_applied); + +void hwss_add_tg_get_frame_count(struct block_sequence_state *seq_state, + struct timing_generator *tg, + unsigned int *frame_count); + +void hwss_add_tg_set_dsc_config(struct block_sequence_state *seq_state, + struct timing_generator *tg, + struct dsc_optc_config *dsc_optc_cfg, + bool enable); + +void hwss_add_opp_program_left_edge_extra_pixel(struct block_sequence_state *seq_state, + struct output_pixel_processor *opp, + enum dc_pixel_encoding pixel_encoding, + bool is_otg_master); + #endif /* __DC_HW_SEQUENCER_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h index 1e2d247fbbac..406db231bc72 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h @@ -27,6 +27,7 @@ #define __DC_HW_SEQUENCER_PRIVATE_H__ #include "dc_types.h" +#include "hw_sequencer.h" enum pipe_gating_control { PIPE_GATING_CONTROL_DISABLE = 0, @@ -80,7 +81,13 @@ struct hwseq_private_funcs { void (*plane_atomic_disconnect)(struct dc *dc, struct dc_state *state, struct pipe_ctx *pipe_ctx); + void (*plane_atomic_disconnect_sequence)(struct dc *dc, + struct dc_state *state, + struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); void (*update_mpcc)(struct dc *dc, struct pipe_ctx *pipe_ctx); + void (*update_mpcc_sequence)(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); bool (*set_input_transfer_func)(struct dc *dc, struct pipe_ctx *pipe_ctx, const struct dc_plane_state *plane_state); @@ -97,6 +104,10 @@ struct hwseq_private_funcs { void (*blank_pixel_data)(struct dc *dc, struct pipe_ctx *pipe_ctx, bool blank); + void (*blank_pixel_data_sequence)(struct dc *dc, + struct pipe_ctx *pipe_ctx, + bool blank, + struct block_sequence_state *seq_state); enum dc_status (*enable_stream_timing)( struct pipe_ctx *pipe_ctx, struct dc_state *context, @@ -105,6 +116,8 @@ struct hwseq_private_funcs { bool enable); void (*setup_vupdate_interrupt)(struct dc *dc, struct pipe_ctx *pipe_ctx); + void (*setup_vupdate_interrupt_sequence)(struct dc *dc, struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); bool (*did_underflow_occur)(struct dc *dc, struct pipe_ctx *pipe_ctx); void (*init_blank)(struct dc *dc, struct timing_generator *tg); void (*disable_vga)(struct dce_hwseq *hws); @@ -112,6 +125,10 @@ struct hwseq_private_funcs { void (*plane_atomic_power_down)(struct dc *dc, struct dpp *dpp, struct hubp *hubp); + void (*plane_atomic_power_down_sequence)(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp, + struct block_sequence_state *seq_state); void (*plane_atomic_disable)(struct dc *dc, struct pipe_ctx *pipe_ctx); void (*enable_power_gating_plane)(struct dce_hwseq *hws, bool enable); @@ -140,15 +157,31 @@ struct hwseq_private_funcs { unsigned int dsc_inst); void (*update_odm)(struct dc *dc, struct dc_state *context, struct pipe_ctx *pipe_ctx); + void (*update_odm_sequence)(struct dc *dc, struct dc_state *context, + struct pipe_ctx *pipe_ctx, struct block_sequence_state *seq_state); void (*program_all_writeback_pipes_in_tree)(struct dc *dc, const struct dc_stream_state *stream, struct dc_state *context); + void (*program_all_writeback_pipes_in_tree_sequence)( + struct dc *dc, + const struct dc_stream_state *stream, + struct dc_state *context, + struct block_sequence_state *seq_state); bool (*s0i3_golden_init_wa)(struct dc *dc); void (*set_hdr_multiplier)(struct pipe_ctx *pipe_ctx); + void (*set_hdr_multiplier_sequence)(struct pipe_ctx *pipe_ctx, + struct block_sequence_state *seq_state); void (*verify_allow_pstate_change_high)(struct dc *dc); + void (*verify_allow_pstate_change_high_sequence)(struct dc *dc, + struct block_sequence_state *seq_state); void (*program_pipe)(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context); + void (*program_pipe_sequence)( + struct dc *dc, + struct pipe_ctx *pipe_ctx, + struct dc_state *context, + struct block_sequence_state *seq_state); bool (*wait_for_blank_complete)(struct output_pixel_processor *opp); void (*dccg_init)(struct dce_hwseq *hws); bool (*set_blend_lut)(struct pipe_ctx *pipe_ctx, @@ -163,6 +196,8 @@ struct hwseq_private_funcs { void (*enable_plane)(struct dc *dc, struct pipe_ctx *pipe_ctx, struct dc_state *context); void (*program_mall_pipe_config)(struct dc *dc, struct dc_state *context); + void (*program_mall_pipe_config_sequence)(struct dc *dc, struct dc_state *context, + struct block_sequence_state *seq_state); void (*update_force_pstate)(struct dc *dc, struct dc_state *context); void (*update_mall_sel)(struct dc *dc, struct dc_state *context); unsigned int (*calculate_dccg_k1_k2_values)(struct pipe_ctx *pipe_ctx, @@ -186,6 +221,7 @@ struct hwseq_private_funcs { void (*perform_3dlut_wa_unlock)(struct pipe_ctx *pipe_ctx); void (*wait_for_pipe_update_if_needed)(struct dc *dc, struct pipe_ctx *pipe_ctx, bool is_surface_update_only); void (*set_wait_for_update_needed_for_pipe)(struct dc *dc, struct pipe_ctx *pipe_ctx); + void (*dc_ip_request_cntl)(struct dc *dc, bool enable); }; struct dce_hwseq { diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index d11893f8c916..5ed2cd344804 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -58,8 +58,8 @@ #include "transform.h" #include "dpp.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" -#include "dml2/dml21/inc/dml_top_types.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_types.h" struct resource_pool; struct dc_state; @@ -274,7 +274,7 @@ struct resource_pool { /* An array for accessing the link encoder objects that have been created. * Index in array corresponds to engine ID - viz. 0: ENGINE_ID_DIGA */ - struct link_encoder *link_encoders[MAX_DIG_LINK_ENCODERS]; + struct link_encoder *link_encoders[MAX_LINK_ENCODERS]; /* Number of DIG link encoder objects created - i.e. number of valid * entries in link_encoders array. */ @@ -514,7 +514,7 @@ struct pipe_ctx { struct link_enc_cfg_context { enum link_enc_cfg_mode mode; struct link_enc_assignment link_enc_assignments[MAX_PIPES]; - enum engine_id link_enc_avail[MAX_DIG_LINK_ENCODERS]; + enum engine_id link_enc_avail[MAX_LINK_ENCODERS]; struct link_enc_assignment transient_assignments[MAX_PIPES]; }; @@ -526,8 +526,8 @@ struct resource_context { uint8_t dp_clock_source_ref_count; bool is_dsc_acquired[MAX_PIPES]; struct link_enc_cfg_context link_enc_cfg_ctx; - unsigned int dio_link_enc_to_link_idx[MAX_DIG_LINK_ENCODERS]; - int dio_link_enc_ref_cnts[MAX_DIG_LINK_ENCODERS]; + unsigned int dio_link_enc_to_link_idx[MAX_LINK_ENCODERS]; + int dio_link_enc_ref_cnts[MAX_LINK_ENCODERS]; bool is_hpo_dp_stream_enc_acquired[MAX_HPO_DP2_ENCODERS]; unsigned int hpo_dp_link_enc_to_link_idx[MAX_HPO_DP2_LINK_ENCODERS]; int hpo_dp_link_enc_ref_cnts[MAX_HPO_DP2_LINK_ENCODERS]; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/cursor_reg_cache.h b/drivers/gpu/drm/amd/display/dc/inc/hw/cursor_reg_cache.h index 45645f9fd86c..7ce2f417f86a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/cursor_reg_cache.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/cursor_reg_cache.h @@ -57,9 +57,9 @@ struct cursor_attribute_cache_hubp { } size; union reg_cursor_settings_cfg { struct { - uint32_t dst_y_offset: 8; - uint32_t chunk_hdl_adjust: 2; - uint32_t reserved: 22; + uint32_t dst_y_offset: 8; + uint32_t chunk_hdl_adjust: 2; + uint32_t reserved: 22; } bits; uint32_t raw; } settings; @@ -83,12 +83,34 @@ union reg_cur0_control_cfg { } bits; uint32_t raw; }; + struct cursor_position_cache_dpp { union reg_cur0_control_cfg cur0_ctl; }; struct cursor_attribute_cache_dpp { union reg_cur0_control_cfg cur0_ctl; + union reg_cur0_fp_scale_bias { + struct { + uint32_t fp_bias: 16; + uint32_t fp_scale: 16; + } bits; + uint32_t raw; + } fp_scale_bias; + union reg_cur0_fp_scale_bias_g_y { + struct { + uint32_t fp_bias_g_y: 16; + uint32_t fp_scale_g_y: 16; + } bits; + uint32_t raw; + } fp_scale_bias_g_y; + union reg_cur0_fp_scale_bias_rb_crcb { + struct { + uint32_t fp_bias_rb_crcb: 16; + uint32_t fp_scale_rb_crcb: 16; + } bits; + uint32_t raw; + } fp_scale_bias_rb_crcb; }; struct cursor_attributes_cfg { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h index 61c4d2a7db1c..500a601e99b5 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h @@ -71,6 +71,125 @@ enum pixel_rate_div { PIXEL_RATE_DIV_NA = 0xF }; +struct dcn_dccg_reg_state { + uint32_t dc_mem_global_pwr_req_cntl; + uint32_t dccg_audio_dtbclk_dto_modulo; + uint32_t dccg_audio_dtbclk_dto_phase; + uint32_t dccg_audio_dto_source; + uint32_t dccg_audio_dto0_module; + uint32_t dccg_audio_dto0_phase; + uint32_t dccg_audio_dto1_module; + uint32_t dccg_audio_dto1_phase; + uint32_t dccg_cac_status; + uint32_t dccg_cac_status2; + uint32_t dccg_disp_cntl_reg; + uint32_t dccg_ds_cntl; + uint32_t dccg_ds_dto_incr; + uint32_t dccg_ds_dto_modulo; + uint32_t dccg_ds_hw_cal_interval; + uint32_t dccg_gate_disable_cntl; + uint32_t dccg_gate_disable_cntl2; + uint32_t dccg_gate_disable_cntl3; + uint32_t dccg_gate_disable_cntl4; + uint32_t dccg_gate_disable_cntl5; + uint32_t dccg_gate_disable_cntl6; + uint32_t dccg_global_fgcg_rep_cntl; + uint32_t dccg_gtc_cntl; + uint32_t dccg_gtc_current; + uint32_t dccg_gtc_dto_incr; + uint32_t dccg_gtc_dto_modulo; + uint32_t dccg_perfmon_cntl; + uint32_t dccg_perfmon_cntl2; + uint32_t dccg_soft_reset; + uint32_t dccg_test_clk_sel; + uint32_t dccg_vsync_cnt_ctrl; + uint32_t dccg_vsync_cnt_int_ctrl; + uint32_t dccg_vsync_otg0_latch_value; + uint32_t dccg_vsync_otg1_latch_value; + uint32_t dccg_vsync_otg2_latch_value; + uint32_t dccg_vsync_otg3_latch_value; + uint32_t dccg_vsync_otg4_latch_value; + uint32_t dccg_vsync_otg5_latch_value; + uint32_t dispclk_cgtt_blk_ctrl_reg; + uint32_t dispclk_freq_change_cntl; + uint32_t dp_dto_dbuf_en; + uint32_t dp_dto0_modulo; + uint32_t dp_dto0_phase; + uint32_t dp_dto1_modulo; + uint32_t dp_dto1_phase; + uint32_t dp_dto2_modulo; + uint32_t dp_dto2_phase; + uint32_t dp_dto3_modulo; + uint32_t dp_dto3_phase; + uint32_t dpiaclk_540m_dto_modulo; + uint32_t dpiaclk_540m_dto_phase; + uint32_t dpiaclk_810m_dto_modulo; + uint32_t dpiaclk_810m_dto_phase; + uint32_t dpiaclk_dto_cntl; + uint32_t dpiasymclk_cntl; + uint32_t dppclk_cgtt_blk_ctrl_reg; + uint32_t dppclk_ctrl; + uint32_t dppclk_dto_ctrl; + uint32_t dppclk0_dto_param; + uint32_t dppclk1_dto_param; + uint32_t dppclk2_dto_param; + uint32_t dppclk3_dto_param; + uint32_t dprefclk_cgtt_blk_ctrl_reg; + uint32_t dprefclk_cntl; + uint32_t dpstreamclk_cntl; + uint32_t dscclk_dto_ctrl; + uint32_t dscclk0_dto_param; + uint32_t dscclk1_dto_param; + uint32_t dscclk2_dto_param; + uint32_t dscclk3_dto_param; + uint32_t dtbclk_dto_dbuf_en; + uint32_t dtbclk_dto0_modulo; + uint32_t dtbclk_dto0_phase; + uint32_t dtbclk_dto1_modulo; + uint32_t dtbclk_dto1_phase; + uint32_t dtbclk_dto2_modulo; + uint32_t dtbclk_dto2_phase; + uint32_t dtbclk_dto3_modulo; + uint32_t dtbclk_dto3_phase; + uint32_t dtbclk_p_cntl; + uint32_t force_symclk_disable; + uint32_t hdmicharclk0_clock_cntl; + uint32_t hdmistreamclk_cntl; + uint32_t hdmistreamclk0_dto_param; + uint32_t microsecond_time_base_div; + uint32_t millisecond_time_base_div; + uint32_t otg_pixel_rate_div; + uint32_t otg0_phypll_pixel_rate_cntl; + uint32_t otg0_pixel_rate_cntl; + uint32_t otg1_phypll_pixel_rate_cntl; + uint32_t otg1_pixel_rate_cntl; + uint32_t otg2_phypll_pixel_rate_cntl; + uint32_t otg2_pixel_rate_cntl; + uint32_t otg3_phypll_pixel_rate_cntl; + uint32_t otg3_pixel_rate_cntl; + uint32_t phyasymclk_clock_cntl; + uint32_t phybsymclk_clock_cntl; + uint32_t phycsymclk_clock_cntl; + uint32_t phydsymclk_clock_cntl; + uint32_t phyesymclk_clock_cntl; + uint32_t phyplla_pixclk_resync_cntl; + uint32_t phypllb_pixclk_resync_cntl; + uint32_t phypllc_pixclk_resync_cntl; + uint32_t phyplld_pixclk_resync_cntl; + uint32_t phyplle_pixclk_resync_cntl; + uint32_t refclk_cgtt_blk_ctrl_reg; + uint32_t socclk_cgtt_blk_ctrl_reg; + uint32_t symclk_cgtt_blk_ctrl_reg; + uint32_t symclk_psp_cntl; + uint32_t symclk32_le_cntl; + uint32_t symclk32_se_cntl; + uint32_t symclka_clock_enable; + uint32_t symclkb_clock_enable; + uint32_t symclkc_clock_enable; + uint32_t symclkd_clock_enable; + uint32_t symclke_clock_enable; +}; + struct dccg { struct dc_context *ctx; const struct dccg_funcs *funcs; @@ -81,7 +200,6 @@ struct dccg { //int audio_dtbclk_khz;/* TODO needs to be removed */ //int ref_dtbclk_khz;/* TODO needs to be removed */ }; - struct dtbclk_dto_params { const struct dc_crtc_timing *timing; int otg_inst; @@ -214,6 +332,7 @@ struct dccg_funcs { void (*set_dto_dscclk)(struct dccg *dccg, uint32_t dsc_inst, uint32_t num_slices_h); void (*set_ref_dscclk)(struct dccg *dccg, uint32_t dsc_inst); void (*dccg_root_gate_disable_control)(struct dccg *dccg, uint32_t pipe_idx, uint32_t disable_clock_gating); + void (*dccg_read_reg_state)(struct dccg *dccg, struct dcn_dccg_reg_state *dccg_reg_state); }; #endif //__DAL_DCCG_H__ diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h index 843a18287c83..1ddfa30411c8 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dchubbub.h @@ -137,6 +137,14 @@ struct dcn_hubbub_state { uint32_t dram_state_cntl; }; +struct dcn_hubbub_reg_state { + uint32_t det0_ctrl; + uint32_t det1_ctrl; + uint32_t det2_ctrl; + uint32_t det3_ctrl; + uint32_t compbuf_ctrl; +}; + struct hubbub_system_latencies { uint32_t max_latency_ns; uint32_t avg_latency_ns; @@ -216,6 +224,8 @@ struct hubbub_funcs { void (*init_watermarks)(struct hubbub *hubbub); + void (*hubbub_read_reg_state)(struct hubbub *hubbub, struct dcn_hubbub_reg_state *hubbub_reg_state); + /** * @program_det_size: * @@ -242,17 +252,39 @@ struct hubbub_funcs { void (*program_compbuf_segments)(struct hubbub *hubbub, unsigned compbuf_size_seg, bool safe_to_increase); void (*wait_for_det_update)(struct hubbub *hubbub, int hubp_inst); bool (*program_arbiter)(struct hubbub *hubbub, struct dml2_display_arb_regs *arb_regs, bool safe_to_lower); - void (*get_det_sizes)(struct hubbub *hubbub, uint32_t *curr_det_sizes, uint32_t *target_det_sizes); - uint32_t (*compbuf_config_error)(struct hubbub *hubbub); - struct hubbub_perfmon_funcs{ - void (*start_system_latency_measurement)(struct hubbub *hubbub); - void (*get_system_latency_result)(struct hubbub *hubbub, uint32_t refclk_mhz, struct hubbub_system_latencies *latencies); - void (*start_in_order_bandwidth_measurement)(struct hubbub *hubbub); - void (*get_in_order_bandwidth_result)(struct hubbub *hubbub, uint32_t refclk_mhz, uint32_t *bandwidth_mbps); - void (*start_urgent_ramp_latency_measurement)(struct hubbub *hubbub, const struct hubbub_urgent_latency_params *params); - void (*get_urgent_ramp_latency_result)(struct hubbub *hubbub, uint32_t refclk_mhz, uint32_t *latency_ns); + void (*dchvm_init)(struct hubbub *hubbub); + + struct hubbub_perfmon_funcs { void (*reset)(struct hubbub *hubbub); + void (*start_measuring_max_memory_latency_ns)( + struct hubbub *hubbub); + uint32_t (*get_max_memory_latency_ns)(struct hubbub *hubbub, + uint32_t refclk_mhz, uint32_t *sample_count); + void (*start_measuring_average_memory_latency_ns)( + struct hubbub *hubbub); + uint32_t (*get_average_memory_latency_ns)(struct hubbub *hubbub, + uint32_t refclk_mhz, uint32_t *sample_count); + void (*start_measuring_urgent_ramp_latency_ns)( + struct hubbub *hubbub, + const struct hubbub_urgent_latency_params *params); + uint32_t (*get_urgent_ramp_latency_ns)(struct hubbub *hubbub, + uint32_t refclk_mhz); + void (*start_measuring_unbounded_bandwidth_mbps)( + struct hubbub *hubbub); + uint32_t (*get_unbounded_bandwidth_mbps)(struct hubbub *hubbub, + uint32_t refclk_mhz, uint32_t *duration_ns); + void (*start_measuring_average_bandwidth_mbps)( + struct hubbub *hubbub); + uint32_t (*get_average_bandwidth_mbps)(struct hubbub *hubbub, + uint32_t refclk_mhz, uint32_t min_duration_ns, + uint32_t *duration_ns); } perfmon; + + struct hubbub_qos_funcs { + void (*force_display_nominal_profile)(struct hubbub *hubbub); + void (*force_display_urgent_profile)(struct hubbub *hubbub); + void (*reset_display_qos_profile)(struct hubbub *hubbub); + } qos; }; struct hubbub { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h index 1b7c085dc2cc..d88b57d4f512 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h @@ -65,7 +65,6 @@ union defer_reg_writes { } bits; uint32_t raw; }; - struct dpp { const struct dpp_funcs *funcs; struct dc_context *ctx; @@ -84,6 +83,7 @@ struct dpp { struct pwl_params shaper_params; bool cm_bypass_mode; + bool cursor_offload; struct cursor_position_cache_dpp pos; struct cursor_attribute_cache_dpp att; @@ -202,6 +202,19 @@ struct dcn_dpp_state { uint32_t gamcor_mode; }; +struct dcn_dpp_reg_state { + uint32_t recout_start; + uint32_t recout_size; + uint32_t scl_horz_filter_scale_ratio; + uint32_t scl_vert_filter_scale_ratio; + uint32_t scl_mode; + uint32_t cm_control; + uint32_t dpp_control; + uint32_t dscl_control; + uint32_t obuf_control; + uint32_t mpc_size; +}; + struct CM_bias_params { uint32_t cm_bias_cr_r; uint32_t cm_bias_y_g; @@ -225,6 +238,8 @@ struct dpp_funcs { void (*dpp_read_state)(struct dpp *dpp, struct dcn_dpp_state *s); + void (*dpp_read_reg_state)(struct dpp *dpp, struct dcn_dpp_reg_state *dpp_reg_state); + void (*dpp_reset)(struct dpp *dpp); void (*dpp_set_scaler)(struct dpp *dpp, diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h index 2b874d2cc61c..a79019365af8 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hubp.h @@ -41,8 +41,8 @@ #include "mem_input.h" #include "cursor_reg_cache.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" -#include "dml2/dml21/inc/dml_top_types.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_types.h" #define OPP_ID_INVALID 0xf #define MAX_TTU 0xffffff @@ -126,11 +126,13 @@ struct hubp { int mpcc_id; struct dc_cursor_attributes curs_attr; struct dc_cursor_position curs_pos; + bool cursor_offload; bool power_gated; struct cursor_position_cache_hubp pos; struct cursor_attribute_cache_hubp att; struct cursor_rect cur_rect; + bool use_mall_for_cursor; }; struct surface_flip_registers { @@ -236,6 +238,7 @@ struct hubp_funcs { void (*hubp_clk_cntl)(struct hubp *hubp, bool enable); void (*hubp_vtg_sel)(struct hubp *hubp, uint32_t otg_inst); void (*hubp_read_state)(struct hubp *hubp); + void (*hubp_read_reg_state)(struct hubp *hubp, struct dcn_hubp_reg_state *reg_state); void (*hubp_clear_underflow)(struct hubp *hubp); void (*hubp_disable_control)(struct hubp *hubp, bool disable_hubp); unsigned int (*hubp_get_underflow_status)(struct hubp *hubp); diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h index 41c76ba9ba56..a61d12ec61bc 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/hw_shared.h @@ -44,12 +44,67 @@ */ #define MAX_PIPES 6 #define MAX_PHANTOM_PIPES (MAX_PIPES / 2) -#define MAX_LINKS (MAX_PIPES * 2 +2) + +#define MAX_DPIA 6 +#define MAX_CONNECTOR 6 +#define MAX_VIRTUAL_LINKS 4 + +#define MAX_LINKS (MAX_DPIA + MAX_CONNECTOR + MAX_VIRTUAL_LINKS) + +/** + * define MAX_DIG_LINK_ENCODERS - maximum number of digital encoders + * + * Digital encoders are ENGINE_ID_DIGA...G, there are at most 7, + * although not every GPU may have that many. + */ #define MAX_DIG_LINK_ENCODERS 7 + +/** + * define MAX_DAC_LINK_ENCODERS - maximum number of analog link encoders + * + * Analog encoders are ENGINE_ID_DACA/B, there are at most 2, + * although not every GPU may have that many. Modern GPUs typically + * don't have analog encoders. + */ +#define MAX_DAC_LINK_ENCODERS 2 + +/** + * define MAX_LINK_ENCODERS - maximum number link encoders in total + * + * This includes both analog and digital encoders. + */ +#define MAX_LINK_ENCODERS (MAX_DIG_LINK_ENCODERS + MAX_DAC_LINK_ENCODERS) + #define MAX_DWB_PIPES 1 #define MAX_HPO_DP2_ENCODERS 4 #define MAX_HPO_DP2_LINK_ENCODERS 4 +/* Pipe topology snapshot structures */ +#define MAX_TOPOLOGY_SNAPSHOTS 4 + +struct pipe_topology_line { + bool is_phantom_pipe; + int plane_idx; + int slice_idx; + int stream_idx; + int dpp_inst; + int opp_inst; + int tg_inst; +}; + +struct pipe_topology_snapshot { + struct pipe_topology_line pipe_log_lines[MAX_PIPES]; + int line_count; + uint64_t timestamp_us; + int stream_count; + int phantom_stream_count; +}; + +struct pipe_topology_history { + struct pipe_topology_snapshot snapshots[MAX_TOPOLOGY_SNAPSHOTS]; + int current_snapshot_index; +}; + struct gamma_curve { uint32_t offset; uint32_t segments_num; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h index 08c16ba52a51..df512920a9fa 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/link_encoder.h @@ -47,6 +47,7 @@ struct encoder_init_data { enum hpd_source_id hpd_source; /* TODO: in DAL2, here was pointer to EventManagerInterface */ struct graphics_object_id encoder; + enum engine_id analog_engine; struct dc_context *ctx; enum transmitter transmitter; }; @@ -83,6 +84,7 @@ struct link_encoder { struct graphics_object_id connector; uint32_t output_signals; enum engine_id preferred_engine; + enum engine_id analog_engine; struct encoder_feature_support features; enum transmitter transmitter; enum hpd_source_id hpd_source; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h index 42fbc70f7056..d468bc85566a 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mem_input.h @@ -29,7 +29,7 @@ #include "include/grph_object_id.h" #include "dml/display_mode_structs.h" -#include "dml2/dml21/inc/dml_top_dchub_registers.h" +#include "dml2_0/dml21/inc/dml_top_dchub_registers.h" struct dchub_init_data; struct cstate_pstate_watermarks_st { diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h index 22960ee03dee..a8d1abe20f62 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/mpc.h @@ -350,6 +350,15 @@ struct mpcc_state { struct mpc_rmcm_regs rmcm_regs; }; +struct dcn_mpc_reg_state { + uint32_t mpcc_bot_sel; + uint32_t mpcc_control; + uint32_t mpcc_status; + uint32_t mpcc_top_sel; + uint32_t mpcc_opp_id; + uint32_t mpcc_ogam_control; +}; + /** * struct mpc_funcs - funcs */ @@ -373,6 +382,24 @@ struct mpc_funcs { struct mpc *mpc, int mpcc_inst, struct mpcc_state *s); + /** + * @mpc_read_reg_state: + * + * Read MPC register state for debugging underflow purposes. + * + * Parameters: + * + * - [in] mpc - MPC context + * - [out] reg_state - MPC register state structure + * + * Return: + * + * void + */ + void (*mpc_read_reg_state)( + struct mpc *mpc, + int mpcc_inst, + struct dcn_mpc_reg_state *mpc_reg_state); /** * @insert_plane: diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h index 747679cb4944..e1428a83ecbc 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/opp.h @@ -297,6 +297,16 @@ struct oppbuf_params { uint32_t num_segment_padded_pixels; }; +struct dcn_opp_reg_state { + uint32_t dpg_control; + uint32_t fmt_control; + uint32_t oppbuf_control; + uint32_t opp_pipe_control; + uint32_t opp_pipe_crc_control; + uint32_t opp_abm_control; + uint32_t dscrm_dsc_forward_config; +}; + struct opp_funcs { @@ -368,6 +378,9 @@ struct opp_funcs { struct output_pixel_processor *opp, enum dc_pixel_encoding pixel_encoding, bool is_primary); + + void (*opp_read_reg_state)( + struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index f2de2cf23859..da7bf59c4b9d 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -175,6 +175,135 @@ struct dcn_otg_state { uint32_t otg_double_buffer_control; }; +struct dcn_optc_reg_state { + uint32_t optc_bytes_per_pixel; + uint32_t optc_data_format_control; + uint32_t optc_data_source_select; + uint32_t optc_input_clock_control; + uint32_t optc_input_global_control; + uint32_t optc_input_spare_register; + uint32_t optc_memory_config; + uint32_t optc_rsmu_underflow; + uint32_t optc_underflow_threshold; + uint32_t optc_width_control; + + uint32_t otg_3d_structure_control; + uint32_t otg_clock_control; + uint32_t otg_control; + uint32_t otg_count_control; + uint32_t otg_count_reset; + uint32_t otg_crc_cntl; + uint32_t otg_crc_sig_blue_control_mask; + uint32_t otg_crc_sig_red_green_mask; + uint32_t otg_crc0_data_b; + uint32_t otg_crc0_data_rg; + uint32_t otg_crc0_windowa_x_control; + uint32_t otg_crc0_windowa_x_control_readback; + uint32_t otg_crc0_windowa_y_control; + uint32_t otg_crc0_windowa_y_control_readback; + uint32_t otg_crc0_windowb_x_control; + uint32_t otg_crc0_windowb_x_control_readback; + uint32_t otg_crc0_windowb_y_control; + uint32_t otg_crc0_windowb_y_control_readback; + uint32_t otg_crc1_data_b; + uint32_t otg_crc1_data_rg; + uint32_t otg_crc1_windowa_x_control; + uint32_t otg_crc1_windowa_x_control_readback; + uint32_t otg_crc1_windowa_y_control; + uint32_t otg_crc1_windowa_y_control_readback; + uint32_t otg_crc1_windowb_x_control; + uint32_t otg_crc1_windowb_x_control_readback; + uint32_t otg_crc1_windowb_y_control; + uint32_t otg_crc1_windowb_y_control_readback; + uint32_t otg_crc2_data_b; + uint32_t otg_crc2_data_rg; + uint32_t otg_crc3_data_b; + uint32_t otg_crc3_data_rg; + uint32_t otg_dlpc_control; + uint32_t otg_double_buffer_control; + uint32_t otg_drr_control2; + uint32_t otg_drr_control; + uint32_t otg_drr_timing_int_status; + uint32_t otg_drr_trigger_window; + uint32_t otg_drr_v_total_change; + uint32_t otg_drr_v_total_reach_range; + uint32_t otg_dsc_start_position; + uint32_t otg_force_count_now_cntl; + uint32_t otg_global_control0; + uint32_t otg_global_control1; + uint32_t otg_global_control2; + uint32_t otg_global_control3; + uint32_t otg_global_control4; + uint32_t otg_global_sync_status; + uint32_t otg_gsl_control; + uint32_t otg_gsl_vsync_gap; + uint32_t otg_gsl_window_x; + uint32_t otg_gsl_window_y; + uint32_t otg_h_blank_start_end; + uint32_t otg_h_sync_a; + uint32_t otg_h_sync_a_cntl; + uint32_t otg_h_timing_cntl; + uint32_t otg_h_total; + uint32_t otg_interlace_control; + uint32_t otg_interlace_status; + uint32_t otg_interrupt_control; + uint32_t otg_long_vblank_status; + uint32_t otg_m_const_dto0; + uint32_t otg_m_const_dto1; + uint32_t otg_manual_force_vsync_next_line; + uint32_t otg_master_en; + uint32_t otg_master_update_lock; + uint32_t otg_master_update_mode; + uint32_t otg_nom_vert_position; + uint32_t otg_pipe_update_status; + uint32_t otg_pixel_data_readback0; + uint32_t otg_pixel_data_readback1; + uint32_t otg_request_control; + uint32_t otg_snapshot_control; + uint32_t otg_snapshot_frame; + uint32_t otg_snapshot_position; + uint32_t otg_snapshot_status; + uint32_t otg_spare_register; + uint32_t otg_static_screen_control; + uint32_t otg_status; + uint32_t otg_status_frame_count; + uint32_t otg_status_hv_count; + uint32_t otg_status_position; + uint32_t otg_status_vf_count; + uint32_t otg_stereo_control; + uint32_t otg_stereo_force_next_eye; + uint32_t otg_stereo_status; + uint32_t otg_trig_manual_control; + uint32_t otg_triga_cntl; + uint32_t otg_triga_manual_trig; + uint32_t otg_trigb_cntl; + uint32_t otg_trigb_manual_trig; + uint32_t otg_update_lock; + uint32_t otg_v_blank_start_end; + uint32_t otg_v_count_stop_control; + uint32_t otg_v_count_stop_control2; + uint32_t otg_v_sync_a; + uint32_t otg_v_sync_a_cntl; + uint32_t otg_v_total; + uint32_t otg_v_total_control; + uint32_t otg_v_total_int_status; + uint32_t otg_v_total_max; + uint32_t otg_v_total_mid; + uint32_t otg_v_total_min; + uint32_t otg_vert_sync_control; + uint32_t otg_vertical_interrupt0_control; + uint32_t otg_vertical_interrupt0_position; + uint32_t otg_vertical_interrupt1_control; + uint32_t otg_vertical_interrupt1_position; + uint32_t otg_vertical_interrupt2_control; + uint32_t otg_vertical_interrupt2_position; + uint32_t otg_vready_param; + uint32_t otg_vstartup_param; + uint32_t otg_vsync_nom_int_status; + uint32_t otg_vupdate_keepout; + uint32_t otg_vupdate_param; +}; + /** * struct timing_generator - Entry point to Output Timing Generator feature. */ @@ -381,6 +510,7 @@ struct timing_generator_funcs { void (*set_vupdate_keepout)(struct timing_generator *tg, bool enable); bool (*wait_update_lock_status)(struct timing_generator *tg, bool locked); void (*read_otg_state)(struct timing_generator *tg, struct dcn_otg_state *s); + void (*optc_read_reg_state)(struct timing_generator *tg, struct dcn_optc_reg_state *optc_reg_state); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/inc/link_service.h b/drivers/gpu/drm/amd/display/dc/inc/link_service.h index 1e34e84160aa..6f94e48a24d1 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/link_service.h +++ b/drivers/gpu/drm/amd/display/dc/inc/link_service.h @@ -292,12 +292,12 @@ struct link_service { enum replay_FW_Message_type msg, union dmub_replay_cmd_set *cmd_data); bool (*edp_set_coasting_vtotal)( - struct dc_link *link, uint32_t coasting_vtotal); + struct dc_link *link, uint32_t coasting_vtotal, uint16_t frame_skip_number); bool (*edp_replay_residency)(const struct dc_link *link, unsigned int *residency, const bool is_start, const enum pr_residency_mode mode); bool (*edp_set_replay_power_opt_and_coasting_vtotal)(struct dc_link *link, - const unsigned int *power_opts, uint32_t coasting_vtotal); + const unsigned int *power_opts, uint32_t coasting_vtotal, uint16_t frame_skip_number); bool (*edp_wait_for_t12)(struct dc_link *link); bool (*edp_is_ilr_optimization_required)(struct dc_link *link, diff --git a/drivers/gpu/drm/amd/display/dc/inc/resource.h b/drivers/gpu/drm/amd/display/dc/inc/resource.h index 4e26a16a8743..79746d931471 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/resource.h +++ b/drivers/gpu/drm/amd/display/dc/inc/resource.h @@ -49,6 +49,7 @@ struct resource_caps { int num_video_plane; int num_audio; int num_stream_encoder; + int num_analog_stream_encoder; int num_pll; int num_dwb; int num_ddc; diff --git a/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c b/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c index 9e33bf937a69..1045c268672e 100644 --- a/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c +++ b/drivers/gpu/drm/amd/display/dc/link/accessories/link_dp_cts.c @@ -78,6 +78,7 @@ static void dp_retrain_link_dp_test(struct dc_link *link, struct audio_output audio_output[MAX_PIPES]; struct dc_stream_state *streams_on_link[MAX_PIPES]; int num_streams_on_link = 0; + struct dc *dc = (struct dc *)link->dc; needs_divider_update = (link->dc->link_srv->dp_get_encoding_format(link_setting) != link->dc->link_srv->dp_get_encoding_format((const struct dc_link_settings *) &link->cur_link_settings)); @@ -150,7 +151,7 @@ static void dp_retrain_link_dp_test(struct dc_link *link, if (streams_on_link[i] && streams_on_link[i]->link && streams_on_link[i]->link == link) { stream_update.stream = streams_on_link[i]; stream_update.dpms_off = &dpms_off; - dc_update_planes_and_stream(state->clk_mgr->ctx->dc, NULL, 0, streams_on_link[i], &stream_update); + dc_update_planes_and_stream(dc, NULL, 0, streams_on_link[i], &stream_update); } } } @@ -876,7 +877,7 @@ bool dp_set_test_pattern( return false; if (pipe_ctx->stream_res.tg->funcs->lock_doublebuffer_enable) { - if (should_use_dmub_lock(pipe_ctx->stream->link)) { + if (should_use_dmub_inbox1_lock(pipe_ctx->stream->link->dc, pipe_ctx->stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; @@ -924,7 +925,7 @@ bool dp_set_test_pattern( CRTC_STATE_VACTIVE); if (pipe_ctx->stream_res.tg->funcs->lock_doublebuffer_disable) { - if (should_use_dmub_lock(pipe_ctx->stream->link)) { + if (should_use_dmub_inbox1_lock(pipe_ctx->stream->link->dc, pipe_ctx->stream->link)) { union dmub_hw_lock_flags hw_locks = { 0 }; struct dmub_hw_lock_inst_flags inst_flags = { 0 }; diff --git a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c index 892907991f91..befa67b2b2ae 100644 --- a/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c +++ b/drivers/gpu/drm/amd/display/dc/link/hwss/link_hwss_dio.c @@ -58,8 +58,9 @@ void setup_dio_stream_encoder(struct pipe_ctx *pipe_ctx) return; } - link_enc->funcs->connect_dig_be_to_fe(link_enc, - pipe_ctx->stream_res.stream_enc->id, true); + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + link_enc->funcs->connect_dig_be_to_fe(link_enc, + pipe_ctx->stream_res.stream_enc->id, true); if (dc_is_dp_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->ctx->dc->link_srv->dp_trace_source_sequence(pipe_ctx->stream->link, DPCD_SOURCE_SEQ_AFTER_CONNECT_DIG_FE_BE); @@ -98,10 +99,13 @@ void reset_dio_stream_encoder(struct pipe_ctx *pipe_ctx) if (stream_enc->funcs->enable_stream) stream_enc->funcs->enable_stream(stream_enc, pipe_ctx->stream->signal, false); - link_enc->funcs->connect_dig_be_to_fe( - link_enc, - pipe_ctx->stream_res.stream_enc->id, - false); + + if (!dc_is_rgb_signal(pipe_ctx->stream->signal)) + link_enc->funcs->connect_dig_be_to_fe( + link_enc, + pipe_ctx->stream_res.stream_enc->id, + false); + if (dc_is_dp_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->ctx->dc->link_srv->dp_trace_source_sequence( pipe_ctx->stream->link, @@ -115,7 +119,8 @@ void setup_dio_stream_attribute(struct pipe_ctx *pipe_ctx) struct dc_stream_state *stream = pipe_ctx->stream; struct dc_link *link = stream->link; - if (!dc_is_virtual_signal(stream->signal)) + if (!dc_is_virtual_signal(stream->signal) && + !dc_is_rgb_signal(stream->signal)) stream_encoder->funcs->setup_stereo_sync( stream_encoder, pipe_ctx->stream_res.tg->inst, diff --git a/drivers/gpu/drm/amd/display/dc/link/link_detection.c b/drivers/gpu/drm/amd/display/dc/link/link_detection.c index 85303167a553..6d31f4967f1a 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -270,6 +270,10 @@ static void read_scdc_caps(struct ddc_service *ddc_service, uint8_t slave_address = HDMI_SCDC_ADDRESS; uint8_t offset = HDMI_SCDC_MANUFACTURER_OUI; + if (ddc_service->link->local_sink && + !ddc_service->link->local_sink->edid_caps.scdc_present) + return; + link_query_ddc_data(ddc_service, slave_address, &offset, sizeof(offset), sink->scdc_caps.manufacturer_OUI.byte, sizeof(sink->scdc_caps.manufacturer_OUI.byte)); @@ -858,6 +862,94 @@ static void verify_link_capability(struct dc_link *link, struct dc_sink *sink, verify_link_capability_non_destructive(link); } +/** + * link_detect_evaluate_edid_header() - Evaluate if an EDID header is acceptable. + * + * Evaluates an 8-byte EDID header to check if it's good enough + * for the purpose of determining whether a display is connected + * without reading the full EDID. + * + * @edid_header: The first 8 bytes of the EDID read from DDC. + * + * Return: true if the header looks valid (>= 6 of 8 bytes match the + * expected 00/FF pattern), false otherwise. + */ +static bool link_detect_evaluate_edid_header(uint8_t edid_header[8]) +{ + int edid_header_score = 0; + int i; + + for (i = 0; i < 8; ++i) + edid_header_score += edid_header[i] == ((i == 0 || i == 7) ? 0x00 : 0xff); + + return edid_header_score >= 6; +} + +/** + * link_detect_ddc_probe() - Probe the DDC to see if a display is connected. + * + * Detect whether a display is connected to DDC without reading full EDID. + * Reads only the EDID header (the first 8 bytes of EDID) from DDC and + * evaluates whether that matches. + * + * @link: DC link whose DDC/I2C is probed for the EDID header. + * + * Return: true if the EDID header was read and passes validation, + * false otherwise. + */ +static bool link_detect_ddc_probe(struct dc_link *link) +{ + if (!link->ddc) + return false; + + uint8_t edid_header[8] = {0}; + bool ddc_probed = i2c_read(link->ddc, 0x50, edid_header, sizeof(edid_header)); + + if (!ddc_probed) + return false; + + if (!link_detect_evaluate_edid_header(edid_header)) + return false; + + return true; +} + +/** + * link_detect_dac_load_detect() - Performs DAC load detection. + * + * Load detection can be used to detect the presence of an + * analog display when we can't read DDC. This causes a visible + * visual glitch so it should be used sparingly. + * + * @link: DC link to test using the DAC load-detect path. + * + * Return: true if the VBIOS load-detect call reports OK, false + * otherwise. + */ +static bool link_detect_dac_load_detect(struct dc_link *link) +{ + struct dc_bios *bios = link->ctx->dc_bios; + struct link_encoder *link_enc = link->link_enc; + enum engine_id engine_id = link_enc->preferred_engine; + enum dal_device_type device_type = DEVICE_TYPE_CRT; + enum bp_result bp_result; + uint32_t enum_id; + + switch (engine_id) { + case ENGINE_ID_DACB: + enum_id = 2; + break; + case ENGINE_ID_DACA: + default: + engine_id = ENGINE_ID_DACA; + enum_id = 1; + break; + } + + bp_result = bios->funcs->dac_load_detection(bios, engine_id, device_type, enum_id); + return bp_result == BP_RESULT_OK; +} + /* * detect_link_and_local_sink() - Detect if a sink is attached to a given link * @@ -942,6 +1034,12 @@ static bool detect_link_and_local_sink(struct dc_link *link, break; } + case SIGNAL_TYPE_RGB: { + sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C; + sink_caps.signal = SIGNAL_TYPE_RGB; + break; + } + case SIGNAL_TYPE_LVDS: { sink_caps.transaction_type = DDC_TRANSACTION_TYPE_I2C; sink_caps.signal = SIGNAL_TYPE_LVDS; @@ -1066,7 +1164,30 @@ static bool detect_link_and_local_sink(struct dc_link *link, DC_LOG_ERROR("Partial EDID valid, abandon invalid blocks.\n"); break; case EDID_NO_RESPONSE: + /* Analog connectors without EDID: + * - old monitor that actually doesn't have EDID + * - cheap DVI-A cable or adapter that doesn't connect DDC + */ + if (dc_connector_supports_analog(link->link_id.id)) { + /* If we didn't do DAC load detection yet, do it now + * to verify there really is a display connected. + */ + if (link->type != dc_connection_dac_load && + !link_detect_dac_load_detect(link)) { + if (prev_sink) + dc_sink_release(prev_sink); + link_disconnect_sink(link); + return false; + } + + DC_LOG_INFO("%s detected analog display without EDID\n", __func__); + link->type = dc_connection_dac_load; + sink->edid_caps.analog = true; + break; + } + DC_LOG_ERROR("No EDID read.\n"); + /* * Abort detection for non-DP connectors if we have * no EDID @@ -1133,14 +1254,23 @@ static bool detect_link_and_local_sink(struct dc_link *link, sink = prev_sink; prev_sink = NULL; } - query_hdcp_capability(sink->sink_signal, link); + + if (!sink->edid_caps.analog) + query_hdcp_capability(sink->sink_signal, link); } + /* DVI-I connector connected to analog display. */ + if ((link->link_id.id == CONNECTOR_ID_DUAL_LINK_DVII || + link->link_id.id == CONNECTOR_ID_SINGLE_LINK_DVII) && + sink->edid_caps.analog) + sink->sink_signal = SIGNAL_TYPE_RGB; + /* HDMI-DVI Dongle */ if (sink->sink_signal == SIGNAL_TYPE_HDMI_TYPE_A && !sink->edid_caps.edid_hdmi) sink->sink_signal = SIGNAL_TYPE_DVI_SINGLE_LINK; else if (dc_is_dvi_signal(sink->sink_signal) && + dc_is_dvi_signal(link->connector_signal) && aud_support->hdmi_audio_native && sink->edid_caps.edid_hdmi) sink->sink_signal = SIGNAL_TYPE_HDMI_TYPE_A; @@ -1232,6 +1362,36 @@ static bool detect_link_and_local_sink(struct dc_link *link, return true; } +/** + * link_detect_analog() - Determines if an analog sink is connected. + * + * @link: DC link to evaluate (must support analog signalling). + * @type: Updated with the detected connection type: + * dc_connection_single (analog via DDC), + * dc_connection_dac_load (via load-detect), + * or dc_connection_none. + * + * Return: true if detection completed. + */ +static bool link_detect_analog(struct dc_link *link, enum dc_connection_type *type) +{ + /* Don't care about connectors that don't support an analog signal. */ + ASSERT(dc_connector_supports_analog(link->link_id.id)); + + if (link_detect_ddc_probe(link)) { + *type = dc_connection_single; + return true; + } + + if (link_detect_dac_load_detect(link)) { + *type = dc_connection_dac_load; + return true; + } + + *type = dc_connection_none; + return true; +} + /* * link_detect_connection_type() - Determine if there is a sink connected * @@ -1248,6 +1408,17 @@ bool link_detect_connection_type(struct dc_link *link, enum dc_connection_type * return true; } + /* Ignore the HPD pin (if any) for analog connectors. + * Instead rely on DDC and DAC. + * + * - VGA connectors don't have any HPD at all. + * - Some DVI-A cables don't connect the HPD pin. + * - Some DVI-A cables pull up the HPD pin. + * (So it's high even when no display is connected.) + */ + if (dc_connector_supports_analog(link->link_id.id)) + return link_detect_analog(link, type); + if (link->connector_signal == SIGNAL_TYPE_EDP) { /*in case it is not on*/ if (!link->dc->config.edp_no_power_sequencing) diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c index 83419e1a9036..6ae134147617 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c @@ -841,6 +841,7 @@ void link_set_dsc_on_stream(struct pipe_ctx *pipe_ctx, bool enable) dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; ASSERT(dsc_cfg.dc_dsc_cfg.num_slices_h % opp_cnt == 0); dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; if (should_use_dto_dscclk) dccg->funcs->set_dto_dscclk(dccg, dsc->inst, dsc_cfg.dc_dsc_cfg.num_slices_h); @@ -970,6 +971,7 @@ bool link_set_dsc_pps_packet(struct pipe_ctx *pipe_ctx, bool enable, bool immedi dsc_cfg.color_depth = stream->timing.display_color_depth; dsc_cfg.is_odm = pipe_ctx->next_odm_pipe ? true : false; dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; dsc->funcs->dsc_get_packed_pps(dsc, &dsc_cfg, &dsc_packed_pps[0]); memcpy(&stream->dsc_packed_pps[0], &dsc_packed_pps[0], sizeof(stream->dsc_packed_pps)); @@ -2224,7 +2226,11 @@ static enum dc_status enable_link( { enum dc_status status = DC_ERROR_UNEXPECTED; struct dc_stream_state *stream = pipe_ctx->stream; - struct dc_link *link = stream->link; + struct dc_link *link = NULL; + + if (stream == NULL) + return DC_ERROR_UNEXPECTED; + link = stream->link; /* There's some scenarios where driver is unloaded with display * still enabled. When driver is reloaded, it may cause a display @@ -2256,6 +2262,9 @@ static enum dc_status enable_link( enable_link_lvds(pipe_ctx); status = DC_OK; break; + case SIGNAL_TYPE_RGB: + status = DC_OK; + break; case SIGNAL_TYPE_VIRTUAL: status = enable_link_virtual(pipe_ctx); break; @@ -2458,6 +2467,7 @@ void link_set_dpms_on( struct link_encoder *link_enc = pipe_ctx->link_res.dio_link_enc; enum otg_out_mux_dest otg_out_dest = OUT_MUX_DIO; struct vpg *vpg = pipe_ctx->stream_res.stream_enc->vpg; + const struct link_hwss *link_hwss = get_link_hwss(link, &pipe_ctx->link_res); bool apply_edp_fast_boot_optimization = pipe_ctx->stream->apply_edp_fast_boot_optimization; @@ -2502,6 +2512,8 @@ void link_set_dpms_on( pipe_ctx->stream_res.tg->funcs->set_out_mux(pipe_ctx->stream_res.tg, otg_out_dest); } + link_hwss->setup_stream_attribute(pipe_ctx); + pipe_ctx->stream->apply_edp_fast_boot_optimization = false; // Enable VPG before building infoframe diff --git a/drivers/gpu/drm/amd/display/dc/link/link_factory.c b/drivers/gpu/drm/amd/display/dc/link/link_factory.c index 31a73867cd4c..a6e2b0821969 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_factory.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_factory.c @@ -451,6 +451,46 @@ static enum channel_id get_ddc_line(struct dc_link *link) return channel; } +static enum engine_id find_analog_engine(struct dc_link *link) +{ + struct dc_bios *bp = link->ctx->dc_bios; + struct graphics_object_id encoder = {0}; + enum bp_result bp_result = BP_RESULT_OK; + int i; + + for (i = 0; i < 3; i++) { + bp_result = bp->funcs->get_src_obj(bp, link->link_id, i, &encoder); + + if (bp_result != BP_RESULT_OK) + return ENGINE_ID_UNKNOWN; + + switch (encoder.id) { + case ENCODER_ID_INTERNAL_DAC1: + case ENCODER_ID_INTERNAL_KLDSCP_DAC1: + return ENGINE_ID_DACA; + case ENCODER_ID_INTERNAL_DAC2: + case ENCODER_ID_INTERNAL_KLDSCP_DAC2: + return ENGINE_ID_DACB; + } + } + + return ENGINE_ID_UNKNOWN; +} + +static bool transmitter_supported(const enum transmitter transmitter) +{ + return transmitter != TRANSMITTER_UNKNOWN && + transmitter != TRANSMITTER_NUTMEG_CRT && + transmitter != TRANSMITTER_TRAVIS_CRT && + transmitter != TRANSMITTER_TRAVIS_LCD; +} + +static bool analog_engine_supported(const enum engine_id engine_id) +{ + return engine_id == ENGINE_ID_DACA || + engine_id == ENGINE_ID_DACB; +} + static bool construct_phy(struct dc_link *link, const struct link_init_data *init_params) { @@ -482,10 +522,23 @@ static bool construct_phy(struct dc_link *link, link->link_id = bios->funcs->get_connector_id(bios, init_params->connector_index); + /* Determine early if the link has any supported encoders, + * so that we avoid initializing DDC and HPD, etc. + */ + bp_funcs->get_src_obj(bios, link->link_id, 0, &enc_init_data.encoder); + enc_init_data.transmitter = translate_encoder_to_transmitter(enc_init_data.encoder); + enc_init_data.analog_engine = find_analog_engine(link); + link->ep_type = DISPLAY_ENDPOINT_PHY; DC_LOG_DC("BIOS object table - link_id: %d", link->link_id.id); + if (!transmitter_supported(enc_init_data.transmitter) && + !analog_engine_supported(enc_init_data.analog_engine)) { + DC_LOG_WARNING("link_id %d has unsupported encoder\n", link->link_id.id); + goto unsupported_fail; + } + if (bios->funcs->get_disp_connector_caps_info) { bios->funcs->get_disp_connector_caps_info(bios, link->link_id, &disp_connect_caps_info); link->is_internal_display = disp_connect_caps_info.INTERNAL_DISPLAY; @@ -530,6 +583,9 @@ static bool construct_phy(struct dc_link *link, case CONNECTOR_ID_DUAL_LINK_DVII: link->connector_signal = SIGNAL_TYPE_DVI_DUAL_LINK; break; + case CONNECTOR_ID_VGA: + link->connector_signal = SIGNAL_TYPE_RGB; + break; case CONNECTOR_ID_DISPLAY_PORT: case CONNECTOR_ID_MXM: case CONNECTOR_ID_USBC: @@ -611,16 +667,12 @@ static bool construct_phy(struct dc_link *link, dal_ddc_get_line(get_ddc_pin(link->ddc)); enc_init_data.ctx = dc_ctx; - bp_funcs->get_src_obj(dc_ctx->dc_bios, link->link_id, 0, - &enc_init_data.encoder); enc_init_data.connector = link->link_id; enc_init_data.channel = get_ddc_line(link); enc_init_data.hpd_source = get_hpd_line(link); link->hpd_src = enc_init_data.hpd_source; - enc_init_data.transmitter = - translate_encoder_to_transmitter(enc_init_data.encoder); link->link_enc = link->dc->res_pool->funcs->link_enc_create(dc_ctx, &enc_init_data); @@ -735,6 +787,7 @@ static bool construct_phy(struct dc_link *link, link->psr_settings.psr_vtotal_control_support = false; link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED; + link->replay_settings.config.replay_version = DC_REPLAY_VERSION_UNSUPPORTED; DC_LOG_DC("BIOS object table - %s finished successfully.\n", __func__); return true; @@ -753,6 +806,7 @@ create_fail: link->hpd_gpio = NULL; } +unsupported_fail: DC_LOG_DC("BIOS object table - %s failed.\n", __func__); return false; } @@ -816,9 +870,7 @@ static bool construct_dpia(struct dc_link *link, /* TODO: Create link encoder */ link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED; - - /* Some docks seem to NAK I2C writes to segment pointer with mot=0. */ - link->wa_flags.dp_mot_reset_segment = true; + link->replay_settings.config.replay_version = DC_REPLAY_VERSION_UNSUPPORTED; return true; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_ddc.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_ddc.c index 267180e7bc48..5d2bcce2f669 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_ddc.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_ddc.c @@ -549,7 +549,8 @@ void write_scdc_data(struct ddc_service *ddc_service, /*Lower than 340 Scramble bit from SCDC caps*/ if (ddc_service->link->local_sink && - ddc_service->link->local_sink->edid_caps.panel_patch.skip_scdc_overwrite) + (ddc_service->link->local_sink->edid_caps.panel_patch.skip_scdc_overwrite || + !ddc_service->link->local_sink->edid_caps.scdc_present)) return; link_query_ddc_data(ddc_service, slave_address, &offset, diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index b12c11bd6a14..ad90a0106938 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -357,7 +357,9 @@ bool dp_should_enable_fec(const struct dc_link *link) { bool force_disable = false; - if (link->fec_state == dc_link_fec_enabled) + if (link->dc->debug.disable_fec) + force_disable = true; + else if (link->fec_state == dc_link_fec_enabled) force_disable = false; else if (link->connector_signal != SIGNAL_TYPE_DISPLAY_PORT_MST && link->local_sink && @@ -424,6 +426,21 @@ static enum dc_link_rate get_link_rate_from_max_link_bw( return link_rate; } +static enum dc_lane_count get_lttpr_max_lane_count(struct dc_link *link) +{ + enum dc_lane_count lttpr_max_lane_count = LANE_COUNT_UNKNOWN; + + if (link->dpcd_caps.lttpr_caps.max_lane_count <= LANE_COUNT_DP_MAX) + lttpr_max_lane_count = link->dpcd_caps.lttpr_caps.max_lane_count; + + /* if bw_allocation is enabled and nrd_max_lane_count is set, use it */ + if (link->dpia_bw_alloc_config.bw_alloc_enabled && + link->dpia_bw_alloc_config.nrd_max_lane_count > 0) + lttpr_max_lane_count = link->dpia_bw_alloc_config.nrd_max_lane_count; + + return lttpr_max_lane_count; +} + static enum dc_link_rate get_lttpr_max_link_rate(struct dc_link *link) { @@ -438,6 +455,11 @@ static enum dc_link_rate get_lttpr_max_link_rate(struct dc_link *link) break; } + /* if bw_allocation is enabled and nrd_max_link_rate is set, use it */ + if (link->dpia_bw_alloc_config.bw_alloc_enabled && + link->dpia_bw_alloc_config.nrd_max_link_rate > 0) + lttpr_max_link_rate = link->dpia_bw_alloc_config.nrd_max_link_rate; + if (link->dpcd_caps.lttpr_caps.supported_128b_132b_rates.bits.UHBR20) lttpr_max_link_rate = LINK_RATE_UHBR20; else if (link->dpcd_caps.lttpr_caps.supported_128b_132b_rates.bits.UHBR13_5) @@ -1691,7 +1713,7 @@ static bool retrieve_link_cap(struct dc_link *link) union edp_configuration_cap edp_config_cap; union dp_downstream_port_present ds_port = { 0 }; enum dc_status status = DC_ERROR_UNEXPECTED; - uint32_t read_dpcd_retry_cnt = 3; + uint32_t read_dpcd_retry_cnt = 20; int i; struct dp_sink_hw_fw_revision dp_hw_fw_revision; const uint32_t post_oui_delay = 30; // 30ms @@ -1734,12 +1756,13 @@ static bool retrieve_link_cap(struct dc_link *link) } dpcd_set_source_specific_data(link); - /* Sink may need to configure internals based on vendor, so allow some - * time before proceeding with possibly vendor specific transactions - */ - msleep(post_oui_delay); for (i = 0; i < read_dpcd_retry_cnt; i++) { + /* + * Sink may need to configure internals based on vendor, so allow some + * time before proceeding with possibly vendor specific transactions + */ + msleep(post_oui_delay); status = core_link_read_dpcd( link, DP_DPCD_REV, @@ -1845,6 +1868,12 @@ static bool retrieve_link_cap(struct dc_link *link) link->dpcd_caps.is_mst_capable = read_is_mst_supported(link); DC_LOG_DC("%s: MST_Support: %s\n", __func__, str_yes_no(link->dpcd_caps.is_mst_capable)); + /* Some MST docks seem to NAK I2C writes to segment pointer with mot=0. */ + if (link->dpcd_caps.is_mst_capable) + link->wa_flags.dp_mot_reset_segment = true; + else + link->wa_flags.dp_mot_reset_segment = false; + get_active_converter_info(ds_port.byte, link); dp_wa_power_up_0010FA(link, dpcd_data, sizeof(dpcd_data)); @@ -2063,6 +2092,11 @@ static bool retrieve_link_cap(struct dc_link *link) link->dpcd_caps.max_uncompressed_pixel_rate_cap.raw, sizeof(link->dpcd_caps.max_uncompressed_pixel_rate_cap.raw)); + core_link_read_dpcd(link, + DP_PANEL_REPLAY_CAPABILITY_SUPPORT, + &link->dpcd_caps.pr_caps_supported.raw, + sizeof(link->dpcd_caps.pr_caps_supported.raw)); + /* Read DP tunneling information. */ status = dpcd_get_tunneling_device_data(link); if (status != DC_OK) @@ -2241,6 +2275,7 @@ const struct dc_link_settings *dp_get_verified_link_cap( struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) { struct dc_link_settings max_link_cap = {0}; + enum dc_lane_count lttpr_max_lane_count; enum dc_link_rate lttpr_max_link_rate; enum dc_link_rate cable_max_link_rate; struct resource_context *res_ctx = &link->dc->current_state->res_ctx; @@ -2305,8 +2340,11 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) /* Some LTTPR devices do not report valid DPCD revisions, if so, do not take it's link cap into consideration. */ if (link->dpcd_caps.lttpr_caps.revision.raw >= DPCD_REV_14) { - if (link->dpcd_caps.lttpr_caps.max_lane_count < max_link_cap.lane_count) - max_link_cap.lane_count = link->dpcd_caps.lttpr_caps.max_lane_count; + lttpr_max_lane_count = get_lttpr_max_lane_count(link); + + if (lttpr_max_lane_count < max_link_cap.lane_count) + max_link_cap.lane_count = lttpr_max_lane_count; + lttpr_max_link_rate = get_lttpr_max_link_rate(link); if (lttpr_max_link_rate < max_link_cap.link_rate) @@ -2412,6 +2450,11 @@ bool dp_verify_link_cap_with_retries( dp_trace_detect_lt_init(link); + DC_LOG_HW_LINK_TRAINING("%s: Link[%d] LinkRate=0x%x LaneCount=%d", + __func__, link->link_index, + known_limit_link_setting->link_rate, + known_limit_link_setting->lane_count); + if (link->link_enc && link->link_enc->features.flags.bits.DP_IS_USB_C && link->dc->debug.usbc_combo_phy_reset_wa) apply_usbc_combo_phy_reset_wa(link, known_limit_link_setting); @@ -2448,6 +2491,11 @@ bool dp_verify_link_cap_with_retries( dp_trace_lt_fail_count_update(link, fail_count, true); dp_trace_set_lt_end_timestamp(link, true); + DC_LOG_HW_LINK_TRAINING("%s: Link[%d] Exit. is_success=%d fail_count=%d", + __func__, link->link_index, + success, + fail_count); + return success; } diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c index 8a3c18ae97a7..c958d3f600c8 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_dpia_bw.c @@ -225,11 +225,6 @@ bool link_dpia_enable_usb4_dp_bw_alloc_mode(struct dc_link *link) bool ret = false; uint8_t val; - if (link->dc->debug.dpia_debug.bits.enable_bw_allocation_mode == false) { - DC_LOG_DEBUG("%s: link[%d] DPTX BW allocation mode disabled", __func__, link->link_index); - return false; - } - val = DPTX_BW_ALLOC_MODE_ENABLE | DPTX_BW_ALLOC_UNMASK_IRQ; if (core_link_write_dpcd(link, DPTX_BW_ALLOCATION_MODE_CONTROL, &val, sizeof(uint8_t)) == DC_OK) { @@ -273,17 +268,28 @@ bool link_dpia_enable_usb4_dp_bw_alloc_mode(struct dc_link *link) */ void link_dp_dpia_handle_bw_alloc_status(struct dc_link *link, uint8_t status) { - link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); - if (status & DP_TUNNELING_BW_REQUEST_SUCCEEDED) { DC_LOG_DEBUG("%s: BW Allocation request succeeded on link(%d)", __func__, link->link_index); - } else if (status & DP_TUNNELING_BW_REQUEST_FAILED) { + } + + if (status & DP_TUNNELING_BW_REQUEST_FAILED) { DC_LOG_DEBUG("%s: BW Allocation request failed on link(%d) allocated/estimated BW=%d", __func__, link->link_index, link->dpia_bw_alloc_config.estimated_bw); link_dpia_send_bw_alloc_request(link, link->dpia_bw_alloc_config.estimated_bw); - } else if (status & DP_TUNNELING_ESTIMATED_BW_CHANGED) { + } + + if (status & DP_TUNNELING_BW_ALLOC_CAP_CHANGED) { + link->dpia_bw_alloc_config.bw_granularity = get_bw_granularity(link); + + DC_LOG_DEBUG("%s: Granularity changed on link(%d) new granularity=%d", + __func__, link->link_index, link->dpia_bw_alloc_config.bw_granularity); + } + + if (status & DP_TUNNELING_ESTIMATED_BW_CHANGED) { + link->dpia_bw_alloc_config.estimated_bw = get_estimated_bw(link); + DC_LOG_DEBUG("%s: Estimated BW changed on link(%d) new estimated BW=%d", __func__, link->link_index, link->dpia_bw_alloc_config.estimated_bw); } diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_irq_handler.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_irq_handler.c index 693477413347..4b01ab0a5a7f 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_irq_handler.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_irq_handler.c @@ -398,10 +398,12 @@ bool dp_should_allow_hpd_rx_irq(const struct dc_link *link) * Don't handle RX IRQ unless one of following is met: * 1) The link is established (cur_link_settings != unknown) * 2) We know we're dealing with a branch device, SST or MST + * 3) The link is bw_alloc enabled. */ if ((link->cur_link_settings.lane_count != LANE_COUNT_UNKNOWN) || - is_dp_branch_device(link)) + is_dp_branch_device(link) || + link->dpia_bw_alloc_config.bw_alloc_enabled) return true; return false; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c index 5e806edbb9f6..c56e69eb27ef 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c @@ -949,7 +949,7 @@ bool edp_set_replay_allow_active(struct dc_link *link, const bool *allow_active, /* Set power optimization flag */ if (power_opts && link->replay_settings.replay_power_opt_active != *power_opts) { if (replay != NULL && link->replay_settings.replay_feature_enabled && - replay->funcs->replay_set_power_opt) { + replay->funcs->replay_set_power_opt) { replay->funcs->replay_set_power_opt(replay, *power_opts, panel_inst); link->replay_settings.replay_power_opt_active = *power_opts; } @@ -984,7 +984,117 @@ bool edp_get_replay_state(const struct dc_link *link, uint64_t *state) return true; } -bool edp_setup_replay(struct dc_link *link, const struct dc_stream_state *stream) +static bool edp_setup_panel_replay(struct dc_link *link, const struct dc_stream_state *stream) +{ + /* To-do: Setup Replay */ + struct dc *dc; + struct dmub_replay *replay; + int i; + unsigned int panel_inst; + struct replay_context replay_context = { 0 }; + unsigned int lineTimeInNs = 0; + + union panel_replay_enable_and_configuration_1 pr_config_1 = { 0 }; + union panel_replay_enable_and_configuration_2 pr_config_2 = { 0 }; + + union dpcd_alpm_configuration alpm_config; + + replay_context.controllerId = CONTROLLER_ID_UNDEFINED; + + if (!link) + return false; + + //Clear Panel Replay enable & config + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_1, + (uint8_t *)&(pr_config_1.raw), sizeof(uint8_t)); + + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_2, + (uint8_t *)&(pr_config_2.raw), sizeof(uint8_t)); + + if (!(link->replay_settings.config.replay_supported)) + return false; + + dc = link->ctx->dc; + + //not sure should keep or not + replay = dc->res_pool->replay; + + if (!replay) + return false; + + if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) + return false; + + replay_context.aux_inst = link->ddc->ddc_pin->hw_info.ddc_channel; + replay_context.digbe_inst = link->link_enc->transmitter; + replay_context.digfe_inst = link->link_enc->preferred_engine; + + for (i = 0; i < MAX_PIPES; i++) { + if (dc->current_state->res_ctx.pipe_ctx[i].stream + == stream) { + /* dmcu -1 for all controller id values, + * therefore +1 here + */ + replay_context.controllerId = + dc->current_state->res_ctx.pipe_ctx[i].stream_res.tg->inst + 1; + break; + } + } + + lineTimeInNs = + ((stream->timing.h_total * 1000000) / + (stream->timing.pix_clk_100hz / 10)) + 1; + + replay_context.line_time_in_ns = lineTimeInNs; + + link->replay_settings.replay_feature_enabled = + replay->funcs->replay_copy_settings(replay, link, &replay_context, panel_inst); + + if (link->replay_settings.replay_feature_enabled) { + pr_config_1.bits.PANEL_REPLAY_ENABLE = 1; + pr_config_1.bits.PANEL_REPLAY_CRC_ENABLE = 1; + pr_config_1.bits.IRQ_HPD_ASSDP_MISSING = 1; + pr_config_1.bits.IRQ_HPD_VSCSDP_UNCORRECTABLE_ERROR = 1; + pr_config_1.bits.IRQ_HPD_RFB_ERROR = 1; + pr_config_1.bits.IRQ_HPD_ACTIVE_FRAME_CRC_ERROR = 1; + pr_config_1.bits.PANEL_REPLAY_SELECTIVE_UPDATE_ENABLE = 1; + pr_config_1.bits.PANEL_REPLAY_EARLY_TRANSPORT_ENABLE = 1; + + pr_config_2.bits.SINK_REFRESH_RATE_UNLOCK_GRANTED = 0; + pr_config_2.bits.SU_Y_GRANULARITY_EXT_VALUE_ENABLED = 0; + pr_config_2.bits.SU_REGION_SCAN_LINE_CAPTURE_INDICATION = 0; + + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_1, + (uint8_t *)&(pr_config_1.raw), sizeof(uint8_t)); + + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_2, + (uint8_t *)&(pr_config_2.raw), sizeof(uint8_t)); + + //ALPM Setup + memset(&alpm_config, 0, sizeof(alpm_config)); + alpm_config.bits.ENABLE = link->replay_settings.config.alpm_mode != DC_ALPM_UNSUPPORTED ? 1 : 0; + + if (link->replay_settings.config.alpm_mode == DC_ALPM_AUXLESS) { + alpm_config.bits.ALPM_MODE_SEL = 1; + alpm_config.bits.ACDS_PERIOD_DURATION = 1; + } + + dm_helpers_dp_write_dpcd( + link->ctx, + link, + DP_RECEIVER_ALPM_CONFIG, + &alpm_config.raw, + sizeof(alpm_config.raw)); + } + + return true; +} + +static bool edp_setup_freesync_replay(struct dc_link *link, const struct dc_stream_state *stream) { /* To-do: Setup Replay */ struct dc *dc; @@ -1080,6 +1190,18 @@ bool edp_setup_replay(struct dc_link *link, const struct dc_stream_state *stream return true; } +bool edp_setup_replay(struct dc_link *link, const struct dc_stream_state *stream) +{ + if (!link) + return false; + if (link->replay_settings.config.replay_version == DC_VESA_PANEL_REPLAY) + return edp_setup_panel_replay(link, stream); + else if (link->replay_settings.config.replay_version == DC_FREESYNC_REPLAY) + return edp_setup_freesync_replay(link, stream); + else + return false; +} + /* * This is general Interface for Replay to set an 32 bit variable to dmub * replay_FW_Message_type: Indicates which instruction or variable pass to DMUB @@ -1110,7 +1232,7 @@ bool edp_send_replay_cmd(struct dc_link *link, return true; } -bool edp_set_coasting_vtotal(struct dc_link *link, uint32_t coasting_vtotal) +bool edp_set_coasting_vtotal(struct dc_link *link, uint32_t coasting_vtotal, uint16_t frame_skip_number) { struct dc *dc = link->ctx->dc; struct dmub_replay *replay = dc->res_pool->replay; @@ -1122,9 +1244,11 @@ bool edp_set_coasting_vtotal(struct dc_link *link, uint32_t coasting_vtotal) if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) return false; - if (coasting_vtotal && link->replay_settings.coasting_vtotal != coasting_vtotal) { - replay->funcs->replay_set_coasting_vtotal(replay, coasting_vtotal, panel_inst); + if (coasting_vtotal && (link->replay_settings.coasting_vtotal != coasting_vtotal || + link->replay_settings.frame_skip_number != frame_skip_number)) { + replay->funcs->replay_set_coasting_vtotal(replay, coasting_vtotal, panel_inst, frame_skip_number); link->replay_settings.coasting_vtotal = coasting_vtotal; + link->replay_settings.frame_skip_number = frame_skip_number; } return true; @@ -1152,7 +1276,7 @@ bool edp_replay_residency(const struct dc_link *link, } bool edp_set_replay_power_opt_and_coasting_vtotal(struct dc_link *link, - const unsigned int *power_opts, uint32_t coasting_vtotal) + const unsigned int *power_opts, uint32_t coasting_vtotal, uint16_t frame_skip_number) { struct dc *dc = link->ctx->dc; struct dmub_replay *replay = dc->res_pool->replay; @@ -1163,13 +1287,16 @@ bool edp_set_replay_power_opt_and_coasting_vtotal(struct dc_link *link, /* Only both power and coasting vtotal changed, this func could return true */ if (power_opts && link->replay_settings.replay_power_opt_active != *power_opts && - coasting_vtotal && link->replay_settings.coasting_vtotal != coasting_vtotal) { + (coasting_vtotal && + (link->replay_settings.coasting_vtotal != coasting_vtotal || + link->replay_settings.frame_skip_number != frame_skip_number))) { if (link->replay_settings.replay_feature_enabled && replay->funcs->replay_set_power_opt_and_coasting_vtotal) { replay->funcs->replay_set_power_opt_and_coasting_vtotal(replay, - *power_opts, panel_inst, coasting_vtotal); + *power_opts, panel_inst, coasting_vtotal, frame_skip_number); link->replay_settings.replay_power_opt_active = *power_opts; link->replay_settings.coasting_vtotal = coasting_vtotal; + link->replay_settings.frame_skip_number = frame_skip_number; } else return false; } else diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.h b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.h index 62a6344e613e..dd79c7cd2828 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.h +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.h @@ -59,12 +59,12 @@ bool edp_setup_replay(struct dc_link *link, bool edp_send_replay_cmd(struct dc_link *link, enum replay_FW_Message_type msg, union dmub_replay_cmd_set *cmd_data); -bool edp_set_coasting_vtotal(struct dc_link *link, uint32_t coasting_vtotal); +bool edp_set_coasting_vtotal(struct dc_link *link, uint32_t coasting_vtotal, uint16_t frame_skip_number); bool edp_replay_residency(const struct dc_link *link, unsigned int *residency, const bool is_start, const enum pr_residency_mode mode); bool edp_get_replay_state(const struct dc_link *link, uint64_t *state); bool edp_set_replay_power_opt_and_coasting_vtotal(struct dc_link *link, - const unsigned int *power_opts, uint32_t coasting_vtotal); + const unsigned int *power_opts, uint32_t coasting_vtotal, uint16_t frame_skip_number); bool edp_wait_for_t12(struct dc_link *link); bool edp_is_ilr_optimization_required(struct dc_link *link, struct dc_crtc_timing *crtc_timing); diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c index 85298b8a1b5e..6bfd2c1294e5 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.c @@ -1514,6 +1514,21 @@ static void mpc3_read_mpcc_state( MPCC_OGAM_SELECT_CURRENT, &s->rgam_lut); } +void mpc3_read_reg_state( + struct mpc *mpc, + int mpcc_inst, struct dcn_mpc_reg_state *mpc_reg_state) +{ + struct dcn30_mpc *mpc30 = TO_DCN30_MPC(mpc); + + mpc_reg_state->mpcc_bot_sel = REG_READ(MPCC_BOT_SEL[mpcc_inst]); + mpc_reg_state->mpcc_control = REG_READ(MPCC_CONTROL[mpcc_inst]); + mpc_reg_state->mpcc_ogam_control = REG_READ(MPCC_OGAM_CONTROL[mpcc_inst]); + mpc_reg_state->mpcc_opp_id = REG_READ(MPCC_OPP_ID[mpcc_inst]); + mpc_reg_state->mpcc_status = REG_READ(MPCC_STATUS[mpcc_inst]); + mpc_reg_state->mpcc_top_sel = REG_READ(MPCC_TOP_SEL[mpcc_inst]); + +} + static const struct mpc_funcs dcn30_mpc_funcs = { .read_mpcc_state = mpc3_read_mpcc_state, .insert_plane = mpc1_insert_plane, @@ -1544,6 +1559,7 @@ static const struct mpc_funcs dcn30_mpc_funcs = { .release_rmu = mpcc3_release_rmu, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, .set_mpc_mem_lp_mode = mpc3_set_mpc_mem_lp_mode, }; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h index 103f29900a2c..e2f147d17178 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn30/dcn30_mpc.h @@ -1096,6 +1096,11 @@ void mpc3_power_on_ogam_lut( struct mpc *mpc, int mpcc_id, bool power_on); +void mpc3_read_reg_state( + struct mpc *mpc, + int mpcc_inst, + struct dcn_mpc_reg_state *mpc_reg_state); + void mpc3_init_mpcc(struct mpcc *mpcc, int mpcc_inst); enum dc_lut_mode mpc3_get_ogam_current( diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c index 6f0e017a8ae2..83bbbf34bcac 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn32/dcn32_mpc.c @@ -1020,6 +1020,7 @@ static const struct mpc_funcs dcn32_mpc_funcs = { .release_rmu = NULL, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, }; diff --git a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c index e1a0308dee57..eeac13fdd6f5 100644 --- a/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c +++ b/drivers/gpu/drm/amd/display/dc/mpc/dcn401/dcn401_mpc.c @@ -598,6 +598,7 @@ static const struct mpc_funcs dcn401_mpc_funcs = { .release_rmu = NULL, .power_on_mpc_mem_pwr = mpc3_power_on_ogam_lut, .get_mpc_out_mux = mpc1_get_mpc_out_mux, + .mpc_read_reg_state = mpc3_read_reg_state, .set_bg_color = mpc1_set_bg_color, .set_movable_cm_location = mpc401_set_movable_cm_location, .update_3dlut_fast_load_select = mpc401_update_3dlut_fast_load_select, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c index 71e9288d60ed..45d418636d0c 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.c @@ -372,6 +372,17 @@ void opp1_pipe_clock_control(struct output_pixel_processor *opp, bool enable) REG_UPDATE(OPP_PIPE_CONTROL, OPP_PIPE_CLOCK_EN, regval); } + +void opp1_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn10_opp *oppn10 = TO_DCN10_OPP(opp); + + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); +} + /*****************************************/ /* Constructor, Destructor */ /*****************************************/ @@ -392,7 +403,8 @@ static const struct opp_funcs dcn10_opp_funcs = { .opp_program_dpg_dimensions = NULL, .dpg_is_blanked = NULL, .dpg_is_pending = NULL, - .opp_destroy = opp1_destroy + .opp_destroy = opp1_destroy, + .opp_read_reg_state = opp1_read_reg_state }; void dcn10_opp_construct(struct dcn10_opp *oppn10, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h index c87de68a509e..38d0d530a9b7 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn10/dcn10_opp.h @@ -63,7 +63,8 @@ uint32_t OPPBUF_CONTROL1; \ uint32_t OPPBUF_3D_PARAMETERS_0; \ uint32_t OPPBUF_3D_PARAMETERS_1; \ - uint32_t OPP_PIPE_CONTROL + uint32_t OPP_PIPE_CONTROL; \ + uint32_t OPP_PIPE_CRC_CONTROL #define OPP_MASK_SH_LIST_DCN(mask_sh) \ OPP_SF(FMT0_FMT_BIT_DEPTH_CONTROL, FMT_TRUNCATE_EN, mask_sh), \ @@ -153,7 +154,6 @@ struct dcn10_opp { const struct dcn10_opp_registers *regs; const struct dcn10_opp_shift *opp_shift; const struct dcn10_opp_mask *opp_mask; - bool is_write_to_ram_a_safe; }; @@ -188,4 +188,6 @@ void opp1_pipe_clock_control(struct output_pixel_processor *opp, bool enable); void opp1_destroy(struct output_pixel_processor **opp); +void opp1_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c index f5fe0cac7cb0..ce826a5be4c7 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.c @@ -377,6 +377,18 @@ uint32_t opp2_get_left_edge_extra_pixel_count(struct output_pixel_processor *opp return 0; } +void opp2_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn20_opp *oppn20 = TO_DCN20_OPP(opp); + + opp_reg_state->dpg_control = REG_READ(DPG_CONTROL); + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); + opp_reg_state->dscrm_dsc_forward_config = REG_READ(DSCRM_DSC_FORWARD_CONFIG); +} + /*****************************************/ /* Constructor, Destructor */ /*****************************************/ @@ -395,6 +407,7 @@ static struct opp_funcs dcn20_opp_funcs = { .opp_destroy = opp1_destroy, .opp_program_left_edge_extra_pixel = opp2_program_left_edge_extra_pixel, .opp_get_left_edge_extra_pixel_count = opp2_get_left_edge_extra_pixel_count, + .opp_read_reg_state = opp2_read_reg_state }; void dcn20_opp_construct(struct dcn20_opp *oppn20, diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h index 34936e6c49f3..fb0c047c1788 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn20/dcn20_opp.h @@ -59,7 +59,8 @@ uint32_t DPG_COLOUR_G_Y; \ uint32_t DPG_COLOUR_R_CR; \ uint32_t DPG_RAMP_CONTROL; \ - uint32_t DPG_STATUS + uint32_t DPG_STATUS; \ + uint32_t DSCRM_DSC_FORWARD_CONFIG #define OPP_DPG_MASK_SH_LIST(mask_sh) \ OPP_SF(DPG0_DPG_CONTROL, DPG_EN, mask_sh), \ @@ -171,4 +172,7 @@ void opp2_program_left_edge_extra_pixel ( uint32_t opp2_get_left_edge_extra_pixel_count(struct output_pixel_processor *opp, enum dc_pixel_encoding pixel_encoding, bool is_primary); + +void opp2_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); + #endif diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c index 3542b51c9aac..e11c4e16402f 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.c @@ -51,3 +51,16 @@ void dcn35_opp_set_fgcg(struct dcn20_opp *oppn20, bool enable) { REG_UPDATE(OPP_TOP_CLK_CONTROL, OPP_FGCG_REP_DIS, !enable); } + +void dcn35_opp_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state) +{ + struct dcn20_opp *oppn20 = TO_DCN20_OPP(opp); + + opp_reg_state->dpg_control = REG_READ(DPG_CONTROL); + opp_reg_state->fmt_control = REG_READ(FMT_CONTROL); + opp_reg_state->opp_abm_control = REG_READ(OPP_ABM_CONTROL); + opp_reg_state->opp_pipe_control = REG_READ(OPP_PIPE_CONTROL); + opp_reg_state->opp_pipe_crc_control = REG_READ(OPP_PIPE_CRC_CONTROL); + opp_reg_state->oppbuf_control = REG_READ(OPPBUF_CONTROL); + opp_reg_state->dscrm_dsc_forward_config = REG_READ(DSCRM_DSC_FORWARD_CONFIG); +} diff --git a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h index a9a413527801..c6cace90e8f2 100644 --- a/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h +++ b/drivers/gpu/drm/amd/display/dc/opp/dcn35/dcn35_opp.h @@ -31,7 +31,8 @@ #define OPP_REG_VARIABLE_LIST_DCN3_5 \ OPP_REG_VARIABLE_LIST_DCN2_0; \ - uint32_t OPP_TOP_CLK_CONTROL + uint32_t OPP_TOP_CLK_CONTROL; \ + uint32_t OPP_ABM_CONTROL #define OPP_MASK_SH_LIST_DCN35(mask_sh) \ OPP_MASK_SH_LIST_DCN20(mask_sh), \ @@ -64,4 +65,5 @@ void dcn35_opp_construct(struct dcn20_opp *oppn20, void dcn35_opp_set_fgcg(struct dcn20_opp *oppn20, bool enable); +void dcn35_opp_read_reg_state(struct output_pixel_processor *opp, struct dcn_opp_reg_state *opp_reg_state); #endif diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h index 8b2a8455eb56..803bcc25601c 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h @@ -209,7 +209,43 @@ uint32_t OPTC_WIDTH_CONTROL2; \ uint32_t OTG_PSTATE_REGISTER; \ uint32_t OTG_PIPE_UPDATE_STATUS; \ - uint32_t INTERRUPT_DEST + uint32_t INTERRUPT_DEST; \ + uint32_t OPTC_INPUT_SPARE_REGISTER; \ + uint32_t OPTC_RSMU_UNDERFLOW; \ + uint32_t OPTC_UNDERFLOW_THRESHOLD; \ + uint32_t OTG_COUNT_CONTROL; \ + uint32_t OTG_COUNT_RESET; \ + uint32_t OTG_CRC_SIG_BLUE_CONTROL_MASK; \ + uint32_t OTG_CRC_SIG_RED_GREEN_MASK; \ + uint32_t OTG_DLPC_CONTROL; \ + uint32_t OTG_DRR_CONTROL2; \ + uint32_t OTG_DRR_TIMING_INT_STATUS; \ + uint32_t OTG_GLOBAL_CONTROL3; \ + uint32_t OTG_GLOBAL_SYNC_STATUS; \ + uint32_t OTG_GSL_VSYNC_GAP; \ + uint32_t OTG_INTERLACE_STATUS; \ + uint32_t OTG_INTERRUPT_CONTROL; \ + uint32_t OTG_LONG_VBLANK_STATUS; \ + uint32_t OTG_MANUAL_FORCE_VSYNC_NEXT_LINE; \ + uint32_t OTG_MASTER_EN; \ + uint32_t OTG_PIXEL_DATA_READBACK0; \ + uint32_t OTG_PIXEL_DATA_READBACK1; \ + uint32_t OTG_REQUEST_CONTROL; \ + uint32_t OTG_SNAPSHOT_CONTROL; \ + uint32_t OTG_SNAPSHOT_FRAME; \ + uint32_t OTG_SNAPSHOT_POSITION; \ + uint32_t OTG_SNAPSHOT_STATUS; \ + uint32_t OTG_SPARE_REGISTER; \ + uint32_t OTG_STATUS_HV_COUNT; \ + uint32_t OTG_STATUS_VF_COUNT; \ + uint32_t OTG_STEREO_FORCE_NEXT_EYE; \ + uint32_t OTG_TRIG_MANUAL_CONTROL; \ + uint32_t OTG_TRIGB_CNTL; \ + uint32_t OTG_TRIGB_MANUAL_TRIG; \ + uint32_t OTG_UPDATE_LOCK; \ + uint32_t OTG_V_TOTAL_INT_STATUS; \ + uint32_t OTG_VSYNC_NOM_INT_STATUS + struct dcn_optc_registers { OPTC_REG_VARIABLE_LIST_DCN; diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c index 4f1830ba619f..c6417538090f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c @@ -315,6 +315,136 @@ void optc31_read_otg_state(struct timing_generator *optc, s->otg_double_buffer_control = REG_READ(OTG_DOUBLE_BUFFER_CONTROL); } +void optc31_read_reg_state(struct timing_generator *optc, struct dcn_optc_reg_state *optc_reg_state) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + + optc_reg_state->optc_bytes_per_pixel = REG_READ(OPTC_BYTES_PER_PIXEL); + optc_reg_state->optc_data_format_control = REG_READ(OPTC_DATA_FORMAT_CONTROL); + optc_reg_state->optc_data_source_select = REG_READ(OPTC_DATA_SOURCE_SELECT); + optc_reg_state->optc_input_clock_control = REG_READ(OPTC_INPUT_CLOCK_CONTROL); + optc_reg_state->optc_input_global_control = REG_READ(OPTC_INPUT_GLOBAL_CONTROL); + optc_reg_state->optc_input_spare_register = REG_READ(OPTC_INPUT_SPARE_REGISTER); + optc_reg_state->optc_memory_config = REG_READ(OPTC_MEMORY_CONFIG); + optc_reg_state->optc_rsmu_underflow = REG_READ(OPTC_RSMU_UNDERFLOW); + optc_reg_state->optc_underflow_threshold = REG_READ(OPTC_UNDERFLOW_THRESHOLD); + optc_reg_state->optc_width_control = REG_READ(OPTC_WIDTH_CONTROL); + optc_reg_state->otg_3d_structure_control = REG_READ(OTG_3D_STRUCTURE_CONTROL); + optc_reg_state->otg_clock_control = REG_READ(OTG_CLOCK_CONTROL); + optc_reg_state->otg_control = REG_READ(OTG_CONTROL); + optc_reg_state->otg_count_control = REG_READ(OTG_COUNT_CONTROL); + optc_reg_state->otg_count_reset = REG_READ(OTG_COUNT_RESET); + optc_reg_state->otg_crc_cntl = REG_READ(OTG_CRC_CNTL); + optc_reg_state->otg_crc_sig_blue_control_mask = REG_READ(OTG_CRC_SIG_BLUE_CONTROL_MASK); + optc_reg_state->otg_crc_sig_red_green_mask = REG_READ(OTG_CRC_SIG_RED_GREEN_MASK); + optc_reg_state->otg_crc0_data_b = REG_READ(OTG_CRC0_DATA_B); + optc_reg_state->otg_crc0_data_rg = REG_READ(OTG_CRC0_DATA_RG); + optc_reg_state->otg_crc0_windowa_x_control = REG_READ(OTG_CRC0_WINDOWA_X_CONTROL); + optc_reg_state->otg_crc0_windowa_x_control_readback = REG_READ(OTG_CRC0_WINDOWA_X_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowa_y_control = REG_READ(OTG_CRC0_WINDOWA_Y_CONTROL); + optc_reg_state->otg_crc0_windowa_y_control_readback = REG_READ(OTG_CRC0_WINDOWA_Y_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowb_x_control = REG_READ(OTG_CRC0_WINDOWB_X_CONTROL); + optc_reg_state->otg_crc0_windowb_x_control_readback = REG_READ(OTG_CRC0_WINDOWB_X_CONTROL_READBACK); + optc_reg_state->otg_crc0_windowb_y_control = REG_READ(OTG_CRC0_WINDOWB_Y_CONTROL); + optc_reg_state->otg_crc0_windowb_y_control_readback = REG_READ(OTG_CRC0_WINDOWB_Y_CONTROL_READBACK); + optc_reg_state->otg_crc1_data_b = REG_READ(OTG_CRC1_DATA_B); + optc_reg_state->otg_crc1_data_rg = REG_READ(OTG_CRC1_DATA_RG); + optc_reg_state->otg_crc1_windowa_x_control = REG_READ(OTG_CRC1_WINDOWA_X_CONTROL); + optc_reg_state->otg_crc1_windowa_x_control_readback = REG_READ(OTG_CRC1_WINDOWA_X_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowa_y_control = REG_READ(OTG_CRC1_WINDOWA_Y_CONTROL); + optc_reg_state->otg_crc1_windowa_y_control_readback = REG_READ(OTG_CRC1_WINDOWA_Y_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowb_x_control = REG_READ(OTG_CRC1_WINDOWB_X_CONTROL); + optc_reg_state->otg_crc1_windowb_x_control_readback = REG_READ(OTG_CRC1_WINDOWB_X_CONTROL_READBACK); + optc_reg_state->otg_crc1_windowb_y_control = REG_READ(OTG_CRC1_WINDOWB_Y_CONTROL); + optc_reg_state->otg_crc1_windowb_y_control_readback = REG_READ(OTG_CRC1_WINDOWB_Y_CONTROL_READBACK); + optc_reg_state->otg_crc2_data_b = REG_READ(OTG_CRC2_DATA_B); + optc_reg_state->otg_crc2_data_rg = REG_READ(OTG_CRC2_DATA_RG); + optc_reg_state->otg_crc3_data_b = REG_READ(OTG_CRC3_DATA_B); + optc_reg_state->otg_crc3_data_rg = REG_READ(OTG_CRC3_DATA_RG); + optc_reg_state->otg_dlpc_control = REG_READ(OTG_DLPC_CONTROL); + optc_reg_state->otg_double_buffer_control = REG_READ(OTG_DOUBLE_BUFFER_CONTROL); + optc_reg_state->otg_drr_control2 = REG_READ(OTG_DRR_CONTROL2); + optc_reg_state->otg_drr_control = REG_READ(OTG_DRR_CONTROL); + optc_reg_state->otg_drr_timing_int_status = REG_READ(OTG_DRR_TIMING_INT_STATUS); + optc_reg_state->otg_drr_trigger_window = REG_READ(OTG_DRR_TRIGGER_WINDOW); + optc_reg_state->otg_drr_v_total_change = REG_READ(OTG_DRR_V_TOTAL_CHANGE); + optc_reg_state->otg_dsc_start_position = REG_READ(OTG_DSC_START_POSITION); + optc_reg_state->otg_force_count_now_cntl = REG_READ(OTG_FORCE_COUNT_NOW_CNTL); + optc_reg_state->otg_global_control0 = REG_READ(OTG_GLOBAL_CONTROL0); + optc_reg_state->otg_global_control1 = REG_READ(OTG_GLOBAL_CONTROL1); + optc_reg_state->otg_global_control2 = REG_READ(OTG_GLOBAL_CONTROL2); + optc_reg_state->otg_global_control3 = REG_READ(OTG_GLOBAL_CONTROL3); + optc_reg_state->otg_global_control4 = REG_READ(OTG_GLOBAL_CONTROL4); + optc_reg_state->otg_global_sync_status = REG_READ(OTG_GLOBAL_SYNC_STATUS); + optc_reg_state->otg_gsl_control = REG_READ(OTG_GSL_CONTROL); + optc_reg_state->otg_gsl_vsync_gap = REG_READ(OTG_GSL_VSYNC_GAP); + optc_reg_state->otg_gsl_window_x = REG_READ(OTG_GSL_WINDOW_X); + optc_reg_state->otg_gsl_window_y = REG_READ(OTG_GSL_WINDOW_Y); + optc_reg_state->otg_h_blank_start_end = REG_READ(OTG_H_BLANK_START_END); + optc_reg_state->otg_h_sync_a = REG_READ(OTG_H_SYNC_A); + optc_reg_state->otg_h_sync_a_cntl = REG_READ(OTG_H_SYNC_A_CNTL); + optc_reg_state->otg_h_timing_cntl = REG_READ(OTG_H_TIMING_CNTL); + optc_reg_state->otg_h_total = REG_READ(OTG_H_TOTAL); + optc_reg_state->otg_interlace_control = REG_READ(OTG_INTERLACE_CONTROL); + optc_reg_state->otg_interlace_status = REG_READ(OTG_INTERLACE_STATUS); + optc_reg_state->otg_interrupt_control = REG_READ(OTG_INTERRUPT_CONTROL); + optc_reg_state->otg_long_vblank_status = REG_READ(OTG_LONG_VBLANK_STATUS); + optc_reg_state->otg_m_const_dto0 = REG_READ(OTG_M_CONST_DTO0); + optc_reg_state->otg_m_const_dto1 = REG_READ(OTG_M_CONST_DTO1); + optc_reg_state->otg_manual_force_vsync_next_line = REG_READ(OTG_MANUAL_FORCE_VSYNC_NEXT_LINE); + optc_reg_state->otg_master_en = REG_READ(OTG_MASTER_EN); + optc_reg_state->otg_master_update_lock = REG_READ(OTG_MASTER_UPDATE_LOCK); + optc_reg_state->otg_master_update_mode = REG_READ(OTG_MASTER_UPDATE_MODE); + optc_reg_state->otg_nom_vert_position = REG_READ(OTG_NOM_VERT_POSITION); + optc_reg_state->otg_pipe_update_status = REG_READ(OTG_PIPE_UPDATE_STATUS); + optc_reg_state->otg_pixel_data_readback0 = REG_READ(OTG_PIXEL_DATA_READBACK0); + optc_reg_state->otg_pixel_data_readback1 = REG_READ(OTG_PIXEL_DATA_READBACK1); + optc_reg_state->otg_request_control = REG_READ(OTG_REQUEST_CONTROL); + optc_reg_state->otg_snapshot_control = REG_READ(OTG_SNAPSHOT_CONTROL); + optc_reg_state->otg_snapshot_frame = REG_READ(OTG_SNAPSHOT_FRAME); + optc_reg_state->otg_snapshot_position = REG_READ(OTG_SNAPSHOT_POSITION); + optc_reg_state->otg_snapshot_status = REG_READ(OTG_SNAPSHOT_STATUS); + optc_reg_state->otg_spare_register = REG_READ(OTG_SPARE_REGISTER); + optc_reg_state->otg_static_screen_control = REG_READ(OTG_STATIC_SCREEN_CONTROL); + optc_reg_state->otg_status = REG_READ(OTG_STATUS); + optc_reg_state->otg_status_frame_count = REG_READ(OTG_STATUS_FRAME_COUNT); + optc_reg_state->otg_status_hv_count = REG_READ(OTG_STATUS_HV_COUNT); + optc_reg_state->otg_status_position = REG_READ(OTG_STATUS_POSITION); + optc_reg_state->otg_status_vf_count = REG_READ(OTG_STATUS_VF_COUNT); + optc_reg_state->otg_stereo_control = REG_READ(OTG_STEREO_CONTROL); + optc_reg_state->otg_stereo_force_next_eye = REG_READ(OTG_STEREO_FORCE_NEXT_EYE); + optc_reg_state->otg_stereo_status = REG_READ(OTG_STEREO_STATUS); + optc_reg_state->otg_trig_manual_control = REG_READ(OTG_TRIG_MANUAL_CONTROL); + optc_reg_state->otg_triga_cntl = REG_READ(OTG_TRIGA_CNTL); + optc_reg_state->otg_triga_manual_trig = REG_READ(OTG_TRIGA_MANUAL_TRIG); + optc_reg_state->otg_trigb_cntl = REG_READ(OTG_TRIGB_CNTL); + optc_reg_state->otg_trigb_manual_trig = REG_READ(OTG_TRIGB_MANUAL_TRIG); + optc_reg_state->otg_update_lock = REG_READ(OTG_UPDATE_LOCK); + optc_reg_state->otg_v_blank_start_end = REG_READ(OTG_V_BLANK_START_END); + optc_reg_state->otg_v_count_stop_control = REG_READ(OTG_V_COUNT_STOP_CONTROL); + optc_reg_state->otg_v_count_stop_control2 = REG_READ(OTG_V_COUNT_STOP_CONTROL2); + optc_reg_state->otg_v_sync_a = REG_READ(OTG_V_SYNC_A); + optc_reg_state->otg_v_sync_a_cntl = REG_READ(OTG_V_SYNC_A_CNTL); + optc_reg_state->otg_v_total = REG_READ(OTG_V_TOTAL); + optc_reg_state->otg_v_total_control = REG_READ(OTG_V_TOTAL_CONTROL); + optc_reg_state->otg_v_total_int_status = REG_READ(OTG_V_TOTAL_INT_STATUS); + optc_reg_state->otg_v_total_max = REG_READ(OTG_V_TOTAL_MAX); + optc_reg_state->otg_v_total_mid = REG_READ(OTG_V_TOTAL_MID); + optc_reg_state->otg_v_total_min = REG_READ(OTG_V_TOTAL_MIN); + optc_reg_state->otg_vert_sync_control = REG_READ(OTG_VERT_SYNC_CONTROL); + optc_reg_state->otg_vertical_interrupt0_control = REG_READ(OTG_VERTICAL_INTERRUPT0_CONTROL); + optc_reg_state->otg_vertical_interrupt0_position = REG_READ(OTG_VERTICAL_INTERRUPT0_POSITION); + optc_reg_state->otg_vertical_interrupt1_control = REG_READ(OTG_VERTICAL_INTERRUPT1_CONTROL); + optc_reg_state->otg_vertical_interrupt1_position = REG_READ(OTG_VERTICAL_INTERRUPT1_POSITION); + optc_reg_state->otg_vertical_interrupt2_control = REG_READ(OTG_VERTICAL_INTERRUPT2_CONTROL); + optc_reg_state->otg_vertical_interrupt2_position = REG_READ(OTG_VERTICAL_INTERRUPT2_POSITION); + optc_reg_state->otg_vready_param = REG_READ(OTG_VREADY_PARAM); + optc_reg_state->otg_vstartup_param = REG_READ(OTG_VSTARTUP_PARAM); + optc_reg_state->otg_vsync_nom_int_status = REG_READ(OTG_VSYNC_NOM_INT_STATUS); + optc_reg_state->otg_vupdate_keepout = REG_READ(OTG_VUPDATE_KEEPOUT); + optc_reg_state->otg_vupdate_param = REG_READ(OTG_VUPDATE_PARAM); +} + static const struct timing_generator_funcs dcn31_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, @@ -377,6 +507,7 @@ static const struct timing_generator_funcs dcn31_tg_funcs = { .init_odm = optc3_init_odm, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn31_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h index 0f72c274f40b..98f7d2e299c5 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h @@ -274,4 +274,6 @@ void optc3_init_odm(struct timing_generator *optc); void optc31_read_otg_state(struct timing_generator *optc, struct dcn_otg_state *s); +void optc31_read_reg_state(struct timing_generator *optc, struct dcn_optc_reg_state *optc_reg_state); + #endif /* __DC_OPTC_DCN31_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c index 4a2caca37255..43ff957288b2 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c @@ -256,6 +256,7 @@ static const struct timing_generator_funcs dcn314_tg_funcs = { .set_h_timing_div_manual_mode = optc314_set_h_timing_div_manual_mode, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn314_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c index b2b226bcd871..3dcb0d0c931c 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c @@ -365,6 +365,7 @@ static const struct timing_generator_funcs dcn32_tg_funcs = { .get_otg_double_buffer_pending = optc3_get_otg_update_pending, .get_pipe_update_pending = optc3_get_pipe_update_pending, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn32_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c index 52d5ea98c86b..f699e95059f3 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c @@ -511,6 +511,7 @@ static const struct timing_generator_funcs dcn35_tg_funcs = { .set_long_vtotal = optc35_set_long_vtotal, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn35_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index 5af13706e601..a8e978d1fae8 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -533,6 +533,7 @@ static const struct timing_generator_funcs dcn401_tg_funcs = { .set_vupdate_keepout = optc401_set_vupdate_keepout, .wait_update_lock_status = optc401_wait_update_lock_status, .read_otg_state = optc31_read_otg_state, + .optc_read_reg_state = optc31_read_reg_state, }; void dcn401_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c index c4b4dc3ad8c9..d40d91ec2035 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce100/dce100_resource.c @@ -78,6 +78,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -225,6 +226,7 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(4), link_regs(5), link_regs(6), + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -368,6 +370,7 @@ static const struct dce_abm_mask abm_mask = { #define DCFE_MEM_PWR_CTRL_REG_BASE 0x1b03 static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; @@ -375,6 +378,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -402,8 +406,10 @@ static const struct dc_plane_cap plane_cap = { } }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; #define CTX ctx @@ -484,6 +490,11 @@ static struct stream_encoder *dce100_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); return &enc110->base; @@ -624,7 +635,20 @@ static struct link_encoder *dce100_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = @@ -952,6 +976,10 @@ struct stream_encoder *dce100_find_first_free_match_stream_enc_for_link( int i; int j = -1; struct dc_link *link = stream->link; + enum engine_id preferred_engine = link->link_enc->preferred_engine; + + if (dc_is_rgb_signal(stream->signal)) + preferred_engine = link->link_enc->analog_engine; for (i = 0; i < pool->stream_enc_count; i++) { if (!res_ctx->is_stream_enc_acquired[i] && @@ -960,8 +988,7 @@ struct stream_encoder *dce100_find_first_free_match_stream_enc_for_link( * in daisy chain use case */ j = i; - if (pool->stream_enc[i]->id == - link->link_enc->preferred_engine) + if (pool->stream_enc[i]->id == preferred_engine) return pool->stream_enc[i]; } } @@ -1093,6 +1120,7 @@ static bool dce100_resource_construct( dc->caps.disable_dp_clk_share = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; for (i = 0; i < pool->base.pipe_count; i++) { pool->base.timing_generators[i] = diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c index cccde5a6f3cd..cd54382c0af3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce110/dce110_resource.c @@ -82,6 +82,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -377,6 +378,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; @@ -424,7 +426,9 @@ static const struct dc_plane_cap plane_cap = { 64 }; -static const struct dc_debug_options debug_defaults = { +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { .enable_legacy_fast_update = true, }; @@ -1376,6 +1380,7 @@ static bool dce110_resource_construct( dc->caps.is_apu = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c index 869a8e515fc0..3f0a6bc4dcc2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce112/dce112_resource.c @@ -76,6 +76,7 @@ #endif #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -385,6 +386,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; @@ -429,8 +431,10 @@ static const struct dc_plane_cap plane_cap = { 64 }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; #define CTX ctx @@ -1247,6 +1251,7 @@ static bool dce112_resource_construct( dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c index 540e04ec1e2d..b1570b6b1af3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce120/dce120_resource.c @@ -491,6 +491,7 @@ static struct dce_i2c_hw *dce120_i2c_hw_create( return dce_i2c_hw; } static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0 + NBIO_BASE(mmBIOS_SCRATCH_0_BASE_IDX), .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3 + NBIO_BASE(mmBIOS_SCRATCH_3_BASE_IDX), .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 + NBIO_BASE(mmBIOS_SCRATCH_6_BASE_IDX) }; @@ -526,8 +527,11 @@ static const struct dc_plane_cap plane_cap = { }; static const struct dc_debug_options debug_defaults = { - .disable_clock_gate = true, - .enable_legacy_fast_update = true, + .disable_clock_gate = true, +}; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; static struct clock_source *dce120_clock_source_create( @@ -1089,6 +1093,7 @@ static bool dce120_resource_construct( dc->caps.psp_setup_panel_mode = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c index b75be6ad64f6..f0152933bee2 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce60/dce60_resource.c @@ -80,6 +80,7 @@ #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -240,7 +241,9 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(2), link_regs(3), link_regs(4), - link_regs(5) + link_regs(5), + {0}, + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -366,6 +369,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; @@ -373,6 +377,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -382,6 +387,7 @@ static const struct resource_caps res_cap_61 = { .num_timing_generator = 4, .num_audio = 6, .num_stream_encoder = 6, + .num_analog_stream_encoder = 1, .num_pll = 3, .num_ddc = 6, }; @@ -389,6 +395,7 @@ static const struct resource_caps res_cap_61 = { static const struct resource_caps res_cap_64 = { .num_timing_generator = 2, .num_audio = 2, + .num_analog_stream_encoder = 1, .num_stream_encoder = 2, .num_pll = 3, .num_ddc = 2, @@ -599,6 +606,11 @@ static struct stream_encoder *dce60_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); @@ -718,7 +730,20 @@ static struct link_encoder *dce60_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = diff --git a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c index 5b7769745202..8687104cabb7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dce80/dce80_resource.c @@ -78,6 +78,7 @@ #ifndef mmBIOS_SCRATCH_2 + #define mmBIOS_SCRATCH_0 0x05C9 #define mmBIOS_SCRATCH_2 0x05CB #define mmBIOS_SCRATCH_3 0x05CC #define mmBIOS_SCRATCH_6 0x05CF @@ -241,6 +242,7 @@ static const struct dce110_link_enc_registers link_enc_regs[] = { link_regs(4), link_regs(5), link_regs(6), + { .DAC_ENABLE = mmDAC_ENABLE }, }; #define stream_enc_regs(id)\ @@ -368,6 +370,7 @@ static const struct dce110_clk_src_mask cs_mask = { }; static const struct bios_registers bios_regs = { + .BIOS_SCRATCH_0 = mmBIOS_SCRATCH_0, .BIOS_SCRATCH_3 = mmBIOS_SCRATCH_3, .BIOS_SCRATCH_6 = mmBIOS_SCRATCH_6 }; @@ -375,6 +378,7 @@ static const struct bios_registers bios_regs = { static const struct resource_caps res_cap = { .num_timing_generator = 6, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 3, .num_ddc = 6, @@ -383,6 +387,7 @@ static const struct resource_caps res_cap = { static const struct resource_caps res_cap_81 = { .num_timing_generator = 4, .num_audio = 7, + .num_analog_stream_encoder = 1, .num_stream_encoder = 7, .num_pll = 3, .num_ddc = 6, @@ -391,6 +396,7 @@ static const struct resource_caps res_cap_81 = { static const struct resource_caps res_cap_83 = { .num_timing_generator = 2, .num_audio = 6, + .num_analog_stream_encoder = 1, .num_stream_encoder = 6, .num_pll = 2, .num_ddc = 2, @@ -418,8 +424,10 @@ static const struct dc_plane_cap plane_cap = { } }; -static const struct dc_debug_options debug_defaults = { - .enable_legacy_fast_update = true, +static const struct dc_debug_options debug_defaults = { 0 }; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; static const struct dce_dmcu_registers dmcu_regs = { @@ -605,6 +613,11 @@ static struct stream_encoder *dce80_stream_encoder_create( if (!enc110) return NULL; + if (eng_id == ENGINE_ID_DACA || eng_id == ENGINE_ID_DACB) { + dce110_analog_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id); + return &enc110->base; + } + dce110_stream_encoder_construct(enc110, ctx, ctx->dc_bios, eng_id, &stream_enc_regs[eng_id], &se_shift, &se_mask); @@ -724,7 +737,20 @@ static struct link_encoder *dce80_link_encoder_create( kzalloc(sizeof(struct dce110_link_encoder), GFP_KERNEL); int link_regs_id; - if (!enc110 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) + if (!enc110) + return NULL; + + if (enc_init_data->connector.id == CONNECTOR_ID_VGA) { + dce110_link_encoder_construct(enc110, + enc_init_data, + &link_enc_feature, + &link_enc_regs[ENGINE_ID_DACA], + NULL, + NULL); + return &enc110->base; + } + + if (enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) return NULL; link_regs_id = @@ -919,6 +945,7 @@ static bool dce80_construct( dc->caps.dual_link_dvi = true; dc->caps.extended_aux_timeout_support = false; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * @@ -1320,6 +1347,7 @@ static bool dce83_construct( dc->caps.min_horizontal_blanking_period = 80; dc->caps.is_apu = true; dc->debug = debug_defaults; + dc->check_config = config_defaults; /************************************************* * Create resources * diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c index 652c05c35494..f12367adf145 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn10/dcn10_resource.c @@ -556,10 +556,13 @@ static const struct dc_debug_options debug_defaults_drv = { .recovery_enabled = false, /*enable this by default after testing.*/ .max_downscale_src_width = 3840, .underflow_assert_delay_us = 0xFFFFFFFF, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn10_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN10_DPP(*dpp)); @@ -1395,6 +1398,8 @@ static bool dcn10_resource_construct( dc->caps.color.mpc.ogam_rom_caps.pq = 0; dc->caps.color.mpc.ogam_rom_caps.hlg = 0; dc->caps.color.mpc.ocsc = 0; + dc->debug = debug_defaults_drv; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c index 84b38d2d6967..6679c1a14f2f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn20/dcn20_resource.c @@ -718,10 +718,13 @@ static const struct dc_debug_options debug_defaults_drv = { .scl_reset_length10 = true, .sanity_checks = false, .underflow_assert_delay_us = 0xFFFFFFFF, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + void dcn20_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN20_DPP(*dpp)); @@ -733,7 +736,7 @@ struct dpp *dcn20_dpp_create( uint32_t inst) { struct dcn20_dpp *dpp = - kzalloc(sizeof(struct dcn20_dpp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_dpp), GFP_KERNEL); if (!dpp) return NULL; @@ -751,7 +754,7 @@ struct input_pixel_processor *dcn20_ipp_create( struct dc_context *ctx, uint32_t inst) { struct dcn10_ipp *ipp = - kzalloc(sizeof(struct dcn10_ipp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn10_ipp), GFP_KERNEL); if (!ipp) { BREAK_TO_DEBUGGER(); @@ -768,7 +771,7 @@ struct output_pixel_processor *dcn20_opp_create( struct dc_context *ctx, uint32_t inst) { struct dcn20_opp *opp = - kzalloc(sizeof(struct dcn20_opp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_opp), GFP_KERNEL); if (!opp) { BREAK_TO_DEBUGGER(); @@ -785,7 +788,7 @@ struct dce_aux *dcn20_aux_engine_create( uint32_t inst) { struct aux_engine_dce110 *aux_engine = - kzalloc(sizeof(struct aux_engine_dce110), GFP_ATOMIC); + kzalloc(sizeof(struct aux_engine_dce110), GFP_KERNEL); if (!aux_engine) return NULL; @@ -823,7 +826,7 @@ struct dce_i2c_hw *dcn20_i2c_hw_create( uint32_t inst) { struct dce_i2c_hw *dce_i2c_hw = - kzalloc(sizeof(struct dce_i2c_hw), GFP_ATOMIC); + kzalloc(sizeof(struct dce_i2c_hw), GFP_KERNEL); if (!dce_i2c_hw) return NULL; @@ -835,8 +838,7 @@ struct dce_i2c_hw *dcn20_i2c_hw_create( } struct mpc *dcn20_mpc_create(struct dc_context *ctx) { - struct dcn20_mpc *mpc20 = kzalloc(sizeof(struct dcn20_mpc), - GFP_ATOMIC); + struct dcn20_mpc *mpc20 = kzalloc(sizeof(struct dcn20_mpc), GFP_KERNEL); if (!mpc20) return NULL; @@ -853,8 +855,7 @@ struct mpc *dcn20_mpc_create(struct dc_context *ctx) struct hubbub *dcn20_hubbub_create(struct dc_context *ctx) { int i; - struct dcn20_hubbub *hubbub = kzalloc(sizeof(struct dcn20_hubbub), - GFP_ATOMIC); + struct dcn20_hubbub *hubbub = kzalloc(sizeof(struct dcn20_hubbub), GFP_KERNEL); if (!hubbub) return NULL; @@ -882,7 +883,7 @@ struct timing_generator *dcn20_timing_generator_create( uint32_t instance) { struct optc *tgn10 = - kzalloc(sizeof(struct optc), GFP_ATOMIC); + kzalloc(sizeof(struct optc), GFP_KERNEL); if (!tgn10) return NULL; @@ -962,7 +963,7 @@ static struct clock_source *dcn20_clock_source_create( bool dp_clk_src) { struct dce110_clk_src *clk_src = - kzalloc(sizeof(struct dce110_clk_src), GFP_ATOMIC); + kzalloc(sizeof(struct dce110_clk_src), GFP_KERNEL); if (!clk_src) return NULL; @@ -1061,7 +1062,7 @@ struct display_stream_compressor *dcn20_dsc_create( struct dc_context *ctx, uint32_t inst) { struct dcn20_dsc *dsc = - kzalloc(sizeof(struct dcn20_dsc), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_dsc), GFP_KERNEL); if (!dsc) { BREAK_TO_DEBUGGER(); @@ -1198,7 +1199,7 @@ struct hubp *dcn20_hubp_create( uint32_t inst) { struct dcn20_hubp *hubp2 = - kzalloc(sizeof(struct dcn20_hubp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_hubp), GFP_KERNEL); if (!hubp2) return NULL; @@ -1668,6 +1669,7 @@ bool dcn20_validate_dsc(struct dc *dc, struct dc_state *new_ctx) dsc_cfg.is_odm = pipe_ctx->next_odm_pipe ? true : false; dsc_cfg.dc_dsc_cfg = stream->timing.dsc_cfg; dsc_cfg.dc_dsc_cfg.num_slices_h /= opp_cnt; + dsc_cfg.dsc_padding = pipe_ctx->dsc_padding_params.dsc_hactive_padding; if (!pipe_ctx->stream_res.dsc->funcs->dsc_validate_stream(pipe_ctx->stream_res.dsc, &dsc_cfg)) return false; @@ -2286,7 +2288,7 @@ bool dcn20_mmhubbub_create(struct dc_context *ctx, struct resource_pool *pool) static struct pp_smu_funcs *dcn20_pp_smu_create(struct dc_context *ctx) { - struct pp_smu_funcs *pp_smu = kzalloc(sizeof(*pp_smu), GFP_ATOMIC); + struct pp_smu_funcs *pp_smu = kzalloc(sizeof(*pp_smu), GFP_KERNEL); if (!pp_smu) return pp_smu; @@ -2472,6 +2474,7 @@ static bool dcn20_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->caps.dp_hdmi21_pcon_support = true; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; @@ -2765,7 +2768,7 @@ struct resource_pool *dcn20_create_resource_pool( struct dc *dc) { struct dcn20_resource_pool *pool = - kzalloc(sizeof(struct dcn20_resource_pool), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_resource_pool), GFP_KERNEL); if (!pool) return NULL; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c index e4a1338d21e0..055107843a70 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn201/dcn201_resource.c @@ -614,10 +614,13 @@ static const struct dc_debug_options debug_defaults_drv = { .sanity_checks = false, .underflow_assert_delay_us = 0xFFFFFFFF, .enable_tri_buf = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn201_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN201_DPP(*dpp)); @@ -629,7 +632,7 @@ static struct dpp *dcn201_dpp_create( uint32_t inst) { struct dcn201_dpp *dpp = - kzalloc(sizeof(struct dcn201_dpp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn201_dpp), GFP_KERNEL); if (!dpp) return NULL; @@ -646,7 +649,7 @@ static struct input_pixel_processor *dcn201_ipp_create( struct dc_context *ctx, uint32_t inst) { struct dcn10_ipp *ipp = - kzalloc(sizeof(struct dcn10_ipp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn10_ipp), GFP_KERNEL); if (!ipp) { return NULL; @@ -662,7 +665,7 @@ static struct output_pixel_processor *dcn201_opp_create( struct dc_context *ctx, uint32_t inst) { struct dcn201_opp *opp = - kzalloc(sizeof(struct dcn201_opp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn201_opp), GFP_KERNEL); if (!opp) { return NULL; @@ -677,7 +680,7 @@ static struct dce_aux *dcn201_aux_engine_create(struct dc_context *ctx, uint32_t inst) { struct aux_engine_dce110 *aux_engine = - kzalloc(sizeof(struct aux_engine_dce110), GFP_ATOMIC); + kzalloc(sizeof(struct aux_engine_dce110), GFP_KERNEL); if (!aux_engine) return NULL; @@ -710,7 +713,7 @@ static struct dce_i2c_hw *dcn201_i2c_hw_create(struct dc_context *ctx, uint32_t inst) { struct dce_i2c_hw *dce_i2c_hw = - kzalloc(sizeof(struct dce_i2c_hw), GFP_ATOMIC); + kzalloc(sizeof(struct dce_i2c_hw), GFP_KERNEL); if (!dce_i2c_hw) return NULL; @@ -723,8 +726,7 @@ static struct dce_i2c_hw *dcn201_i2c_hw_create(struct dc_context *ctx, static struct mpc *dcn201_mpc_create(struct dc_context *ctx, uint32_t num_mpcc) { - struct dcn201_mpc *mpc201 = kzalloc(sizeof(struct dcn201_mpc), - GFP_ATOMIC); + struct dcn201_mpc *mpc201 = kzalloc(sizeof(struct dcn201_mpc), GFP_KERNEL); if (!mpc201) return NULL; @@ -740,8 +742,7 @@ static struct mpc *dcn201_mpc_create(struct dc_context *ctx, uint32_t num_mpcc) static struct hubbub *dcn201_hubbub_create(struct dc_context *ctx) { - struct dcn20_hubbub *hubbub = kzalloc(sizeof(struct dcn20_hubbub), - GFP_ATOMIC); + struct dcn20_hubbub *hubbub = kzalloc(sizeof(struct dcn20_hubbub), GFP_KERNEL); if (!hubbub) return NULL; @@ -759,7 +760,7 @@ static struct timing_generator *dcn201_timing_generator_create( uint32_t instance) { struct optc *tgn10 = - kzalloc(sizeof(struct optc), GFP_ATOMIC); + kzalloc(sizeof(struct optc), GFP_KERNEL); if (!tgn10) return NULL; @@ -793,7 +794,7 @@ static struct link_encoder *dcn201_link_encoder_create( const struct encoder_init_data *enc_init_data) { struct dcn20_link_encoder *enc20 = - kzalloc(sizeof(struct dcn20_link_encoder), GFP_ATOMIC); + kzalloc(sizeof(struct dcn20_link_encoder), GFP_KERNEL); struct dcn10_link_encoder *enc10; if (!enc20 || enc_init_data->hpd_source >= ARRAY_SIZE(link_enc_hpd_regs)) @@ -821,7 +822,7 @@ static struct clock_source *dcn201_clock_source_create( bool dp_clk_src) { struct dce110_clk_src *clk_src = - kzalloc(sizeof(struct dce110_clk_src), GFP_ATOMIC); + kzalloc(sizeof(struct dce110_clk_src), GFP_KERNEL); if (!clk_src) return NULL; @@ -856,7 +857,7 @@ static struct stream_encoder *dcn201_stream_encoder_create( struct dc_context *ctx) { struct dcn10_stream_encoder *enc1 = - kzalloc(sizeof(struct dcn10_stream_encoder), GFP_ATOMIC); + kzalloc(sizeof(struct dcn10_stream_encoder), GFP_KERNEL); if (!enc1) return NULL; @@ -883,7 +884,7 @@ static const struct dce_hwseq_mask hwseq_mask = { static struct dce_hwseq *dcn201_hwseq_create( struct dc_context *ctx) { - struct dce_hwseq *hws = kzalloc(sizeof(struct dce_hwseq), GFP_ATOMIC); + struct dce_hwseq *hws = kzalloc(sizeof(struct dce_hwseq), GFP_KERNEL); if (hws) { hws->ctx = ctx; @@ -983,7 +984,7 @@ static struct hubp *dcn201_hubp_create( uint32_t inst) { struct dcn201_hubp *hubp201 = - kzalloc(sizeof(struct dcn201_hubp), GFP_ATOMIC); + kzalloc(sizeof(struct dcn201_hubp), GFP_KERNEL); if (!hubp201) return NULL; @@ -1153,6 +1154,7 @@ static bool dcn201_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->debug = debug_defaults_drv; + dc->check_config = config_defaults; /*a0 only, remove later*/ dc->work_arounds.no_connect_phy_config = true; @@ -1303,7 +1305,7 @@ struct resource_pool *dcn201_create_resource_pool( struct dc *dc) { struct dcn201_resource_pool *pool = - kzalloc(sizeof(struct dcn201_resource_pool), GFP_ATOMIC); + kzalloc(sizeof(struct dcn201_resource_pool), GFP_KERNEL); if (!pool) return NULL; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c index 918742a42ded..2060acd5ae09 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn21/dcn21_resource.c @@ -626,10 +626,13 @@ static const struct dc_debug_options debug_defaults_drv = { .usbc_combo_phy_reset_wa = true, .dmub_command_table = true, .use_max_lb = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1458,6 +1461,7 @@ static bool dcn21_resource_construct( dc->caps.color.mpc.ocsc = 1; dc->caps.dp_hdmi21_pcon_support = true; + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c index ff63f59ff928..d0ebb733e802 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn30/dcn30_resource.c @@ -727,10 +727,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -2374,6 +2377,7 @@ static bool dcn30_resource_construct( dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c index 82a205a7c25c..3ad6a3d4858e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn301/dcn301_resource.c @@ -701,10 +701,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = false, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static void dcn301_dpp_destroy(struct dpp **dpp) { kfree(TO_DCN20_DPP(*dpp)); @@ -1498,6 +1501,7 @@ static bool dcn301_resource_construct( bp_query_result = ctx->dc_bios->funcs->get_lttpr_interop(ctx->dc_bios, &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c index 61623cb518d9..c0d4a1dc94f8 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn302/dcn302_resource.c @@ -98,10 +98,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1290,6 +1293,7 @@ static bool dcn302_resource_construct( &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c index 02b9a84f2db3..75e09c2c283e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn303/dcn303_resource.c @@ -98,10 +98,13 @@ static const struct dc_debug_options debug_defaults_drv = { .dmub_command_table = true, .use_max_lb = true, .exit_idle_opt_for_cursor_updates = true, - .enable_legacy_fast_update = false, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1234,6 +1237,7 @@ static bool dcn303_resource_construct( bp_query_result = ctx->dc_bios->funcs->get_lttpr_interop(ctx->dc_bios, &is_vbios_interop_enabled); dc->caps.vbios_lttpr_aware = (bp_query_result == BP_RESULT_OK) && !!is_vbios_interop_enabled; } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c index 3ed7f50554e2..0d667b54ccf8 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn31/dcn31_resource.c @@ -888,12 +888,15 @@ static const struct dc_debug_options debug_defaults_drv = { } }, .disable_z10 = true, - .enable_legacy_fast_update = true, .enable_z9_disable_interface = true, /* Allow support for the PMFW interface for disable Z9*/ .dml_hostvm_override = DML_HOSTVM_OVERRIDE_FALSE, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1978,6 +1981,7 @@ static bool dcn31_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c index d4917a35b991..3ccde75a4ecb 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn314/dcn314_resource.c @@ -924,12 +924,15 @@ static const struct dc_debug_options debug_defaults_drv = { }, .seamless_boot_odm_combine = true, - .enable_legacy_fast_update = true, .using_dml2 = false, .disable_dsc_power_gate = true, .min_disp_clk_khz = 100000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1910,6 +1913,7 @@ static bool dcn314_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 82cc78c291d8..4e962f522f1b 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -887,9 +887,13 @@ static const struct dc_debug_options debug_defaults_drv = { .afmt = true, } }, - .enable_legacy_fast_update = true, .psr_power_use_phy_fsm = 0, .using_dml2 = false, + .min_disp_clk_khz = 100000, +}; + +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, }; static const struct dc_panel_config panel_config_defaults = { @@ -1939,6 +1943,7 @@ static bool dcn315_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c index 636110e48d01..5a95dd54cb42 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn316/dcn316_resource.c @@ -882,10 +882,13 @@ static const struct dc_debug_options debug_defaults_drv = { .afmt = true, } }, - .enable_legacy_fast_update = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1815,6 +1818,7 @@ static bool dcn316_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c index 3965a7f1b64b..b276fec3e479 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.c @@ -92,7 +92,7 @@ #include "dc_state_priv.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define DC_LOGGER_INIT(logger) @@ -738,10 +738,13 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_dp_plus_plus_wa = true, .fpo_vactive_min_active_margin_us = 200, .fpo_vactive_max_blank_us = 1000, - .enable_legacy_fast_update = false, .disable_stutter_for_wm_program = true }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn32_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -1844,7 +1847,7 @@ enum dc_status dcn32_validate_bandwidth(struct dc *dc, dc_state_set_stream_cursor_subvp_limit(stream, context, true); status = DC_FAIL_HW_CURSOR_SUPPORT; } - }; + } } if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_FAIL_HW_CURSOR_SUPPORT) { @@ -2197,7 +2200,8 @@ static bool dcn32_resource_construct( dc->caps.i2c_speed_in_khz_hdcp = 100; /*1.4 w/a applied by default*/ /* TODO: Bring max_cursor_size back to 256 after subvp cursor corruption is fixed*/ dc->caps.max_cursor_size = 64; - dc->caps.max_buffered_cursor_size = 64; // sqrt(16 * 1024 / 4) + /* floor(sqrt(buf_size_bytes / bpp ) * bpp, fixed_req_size) / bpp = max_width */ + dc->caps.max_buffered_cursor_size = 64; // floor(sqrt(16 * 1024 / 4) * 4, 256) / 4 = 64 dc->caps.min_horizontal_blanking_period = 80; dc->caps.dmdata_alloc_size = 2048; dc->caps.mall_size_per_mem_channel = 4; @@ -2294,6 +2298,7 @@ static bool dcn32_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c index ad214986f7ac..3466ca34c93f 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn321/dcn321_resource.c @@ -731,11 +731,14 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_subvp_high_refresh = false, .fpo_vactive_min_active_margin_us = 200, .fpo_vactive_max_blank_us = 1000, - .enable_legacy_fast_update = false, .disable_dc_mode_overwrite = true, .using_dml2 = false, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn321_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -1797,6 +1800,7 @@ static bool dcn321_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c index fff57f23f4f7..ef69898d2cc5 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn35/dcn35_resource.c @@ -33,7 +33,7 @@ #include "resource.h" #include "include/irq_service_interface.h" #include "dcn35_resource.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dcn20/dcn20_resource.h" #include "dcn30/dcn30_resource.h" @@ -767,7 +767,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -788,6 +787,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1946,6 +1949,7 @@ static bool dcn35_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index 0abd163b425e..f3c614c4490c 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -83,7 +83,7 @@ #include "vm_helper.h" #include "dcn20/dcn20_vmid.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "link_enc_cfg.h" #define DC_LOGGER_INIT(logger) @@ -747,7 +747,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -768,6 +767,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1917,6 +1920,7 @@ static bool dcn351_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c index ca125ee6c2fb..6469d5fe2e6d 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn36/dcn36_resource.c @@ -11,7 +11,7 @@ #include "resource.h" #include "include/irq_service_interface.h" #include "dcn36_resource.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #include "dcn20/dcn20_resource.h" #include "dcn30/dcn30_resource.h" @@ -748,7 +748,6 @@ static const struct dc_debug_options debug_defaults_drv = { .using_dml2 = true, .support_eDP1_5 = true, .enable_hpo_pg_support = false, - .enable_legacy_fast_update = true, .enable_single_display_2to1_odm_policy = true, .disable_idle_power_optimizations = false, .dmcub_emulation = false, @@ -769,6 +768,10 @@ static const struct dc_debug_options debug_defaults_drv = { .min_disp_clk_khz = 50000, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = true, +}; + static const struct dc_panel_config panel_config_defaults = { .psr = { .disable_psr = false, @@ -1918,6 +1921,7 @@ static bool dcn36_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 1d18807e4749..875ae97489d3 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -73,7 +73,7 @@ #include "dc_state_priv.h" -#include "dml2/dml2_wrapper.h" +#include "dml2_0/dml2_wrapper.h" #define DC_LOGGER_INIT(logger) @@ -721,7 +721,6 @@ static const struct dc_debug_options debug_defaults_drv = { .alloc_extra_way_for_cursor = true, .min_prefetch_in_strobe_ns = 60000, // 60us .disable_unbounded_requesting = false, - .enable_legacy_fast_update = false, .dcc_meta_propagation_delay_us = 10, .fams_version = { .minor = 1, @@ -737,6 +736,10 @@ static const struct dc_debug_options debug_defaults_drv = { .force_cositing = CHROMA_COSITING_NONE + 1, }; +static const struct dc_check_config config_defaults = { + .enable_legacy_fast_update = false, +}; + static struct dce_aux *dcn401_aux_engine_create( struct dc_context *ctx, uint32_t inst) @@ -1668,7 +1671,7 @@ enum dc_status dcn401_validate_bandwidth(struct dc *dc, dc_state_set_stream_cursor_subvp_limit(stream, context, true); status = DC_FAIL_HW_CURSOR_SUPPORT; } - }; + } } if (validate_mode == DC_VALIDATE_MODE_AND_PROGRAMMING && status == DC_FAIL_HW_CURSOR_SUPPORT) { @@ -1995,6 +1998,7 @@ static bool dcn401_resource_construct( dc->caps.vbios_lttpr_aware = true; } } + dc->check_config = config_defaults; if (dc->ctx->dce_environment == DCE_ENV_PRODUCTION_DRV) dc->debug = debug_defaults_drv; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h index 0fc66487d800..e1fa2e80a15a 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h @@ -227,7 +227,8 @@ void dcn401_prepare_mcache_programming(struct dc *dc, struct dc_state *context); #define LE_DCN401_REG_LIST_RI(id) \ LE_DCN3_REG_LIST_RI(id), \ SRI_ARR(DP_DPHY_INTERNAL_CTRL, DP, id), \ - SRI_ARR(DIG_BE_CLK_CNTL, DIG, id) + SRI_ARR(DIG_BE_CLK_CNTL, DIG, id),\ + SR_ARR(DIO_CLK_CNTL, id) /* DPP */ #define DPP_REG_LIST_DCN401_COMMON_RI(id) \ diff --git a/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h b/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h index 21d842857601..88c11b6be004 100644 --- a/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h +++ b/drivers/gpu/drm/amd/display/dc/soc_and_ip_translator/dcn401/dcn401_soc_and_ip_translator.h @@ -9,7 +9,7 @@ #include "dc.h" #include "clk_mgr.h" #include "soc_and_ip_translator.h" -#include "dml2/dml21/inc/dml_top_soc_parameter_types.h" +#include "dml2_0/dml21/inc/dml_top_soc_parameter_types.h" void dcn401_construct_soc_and_ip_translator(struct soc_and_ip_translator *soc_and_ip_translator); diff --git a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c index b1fb0f8a253a..7a839984dbc0 100644 --- a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c +++ b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl.c @@ -1018,6 +1018,21 @@ static bool spl_get_optimal_number_of_taps( spl_scratch->scl_data.taps.h_taps_c = 6; spl_scratch->scl_data.taps.v_taps_c = 6; } + + /* Override mode: keep EASF enabled but use input taps if valid */ + if (spl_in->override_easf) { + spl_scratch->scl_data.taps.h_taps = (in_taps->h_taps != 0) ? in_taps->h_taps : spl_scratch->scl_data.taps.h_taps; + spl_scratch->scl_data.taps.v_taps = (in_taps->v_taps != 0) ? in_taps->v_taps : spl_scratch->scl_data.taps.v_taps; + spl_scratch->scl_data.taps.h_taps_c = (in_taps->h_taps_c != 0) ? in_taps->h_taps_c : spl_scratch->scl_data.taps.h_taps_c; + spl_scratch->scl_data.taps.v_taps_c = (in_taps->v_taps_c != 0) ? in_taps->v_taps_c : spl_scratch->scl_data.taps.v_taps_c; + + if ((spl_scratch->scl_data.taps.h_taps > 6) || (spl_scratch->scl_data.taps.v_taps > 6)) + skip_easf = true; + if ((spl_scratch->scl_data.taps.h_taps > 1) && (spl_scratch->scl_data.taps.h_taps % 2)) + spl_scratch->scl_data.taps.h_taps--; + if ((spl_scratch->scl_data.taps.h_taps_c > 1) && (spl_scratch->scl_data.taps.h_taps_c % 2)) + spl_scratch->scl_data.taps.h_taps_c--; + } } /*Ensure we can support the requested number of vtaps*/ diff --git a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h index 23d254dea18f..20e4e52a77ac 100644 --- a/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h +++ b/drivers/gpu/drm/amd/display/dc/sspl/dc_spl_types.h @@ -545,6 +545,7 @@ struct spl_in { enum linear_light_scaling lls_pref; // Linear Light Scaling bool prefer_easf; bool disable_easf; + bool override_easf; /* If true, keep EASF enabled but use provided in_taps */ struct spl_debug debug; bool is_fullscreen; bool is_hdr_on; diff --git a/drivers/gpu/drm/amd/display/dc/virtual/virtual_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/virtual/virtual_stream_encoder.c index 6ffc74fc9dcd..ad088d70e189 100644 --- a/drivers/gpu/drm/amd/display/dc/virtual/virtual_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/virtual/virtual_stream_encoder.c @@ -44,11 +44,6 @@ static void virtual_stream_encoder_dvi_set_stream_attribute( struct dc_crtc_timing *crtc_timing, bool is_dual_link) {} -static void virtual_stream_encoder_lvds_set_stream_attribute( - struct stream_encoder *enc, - struct dc_crtc_timing *crtc_timing) -{} - static void virtual_stream_encoder_set_throttled_vcp_size( struct stream_encoder *enc, struct fixed31_32 avg_time_slots_per_mtp) @@ -120,8 +115,6 @@ static const struct stream_encoder_funcs virtual_str_enc_funcs = { virtual_stream_encoder_hdmi_set_stream_attribute, .dvi_set_stream_attribute = virtual_stream_encoder_dvi_set_stream_attribute, - .lvds_set_stream_attribute = - virtual_stream_encoder_lvds_set_stream_attribute, .set_throttled_vcp_size = virtual_stream_encoder_set_throttled_vcp_size, .update_hdmi_info_packets = diff --git a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h index 338fdc651f2c..9d0168986fe7 100644 --- a/drivers/gpu/drm/amd/display/dmub/dmub_srv.h +++ b/drivers/gpu/drm/amd/display/dmub/dmub_srv.h @@ -132,6 +132,7 @@ enum dmub_window_id { DMUB_WINDOW_IB_MEM, DMUB_WINDOW_SHARED_STATE, DMUB_WINDOW_LSDMA_BUFFER, + DMUB_WINDOW_CURSOR_OFFLOAD, DMUB_WINDOW_TOTAL, }; @@ -317,6 +318,7 @@ struct dmub_srv_hw_params { bool enable_non_transparent_setconfig; bool lower_hbr3_phy_ssc; bool override_hbr3_pll_vco; + bool disable_dpia_bw_allocation; }; /** @@ -361,6 +363,19 @@ struct dmub_diagnostic_data { uint8_t is_pwait : 1; }; +/** + * struct dmub_preos_info - preos fw info before loading post os fw. + */ +struct dmub_preos_info { + uint64_t fb_base; + uint64_t fb_offset; + uint64_t trace_buffer_phy_addr; + uint32_t trace_buffer_size; + uint32_t fw_version; + uint32_t boot_status; + uint32_t boot_options; +}; + struct dmub_srv_inbox { /* generic status */ uint64_t num_submitted; @@ -486,6 +501,7 @@ struct dmub_srv_hw_funcs { uint32_t (*get_current_time)(struct dmub_srv *dmub); void (*get_diagnostic_data)(struct dmub_srv *dmub); + bool (*get_preos_fw_info)(struct dmub_srv *dmub); bool (*should_detect)(struct dmub_srv *dmub); void (*init_reg_offsets)(struct dmub_srv *dmub, struct dc_context *ctx); @@ -535,7 +551,8 @@ struct dmub_srv_create_params { * @fw_version: the current firmware version, if any * @is_virtual: false if hardware support only * @shared_state: dmub shared state between firmware and driver - * @fw_state: dmub firmware state pointer + * @cursor_offload_v1: Cursor offload state + * @fw_state: dmub firmware state pointer (debug purpose only) */ struct dmub_srv { enum dmub_asic asic; @@ -544,7 +561,9 @@ struct dmub_srv { bool is_virtual; struct dmub_fb scratch_mem_fb; struct dmub_fb ib_mem_gart; + struct dmub_fb cursor_offload_fb; volatile struct dmub_shared_state_feature_block *shared_state; + volatile struct dmub_cursor_offload_v1 *cursor_offload_v1; volatile const struct dmub_fw_state *fw_state; /* private: internal use only */ @@ -583,6 +602,7 @@ struct dmub_srv { enum dmub_srv_power_state_type power_state; struct dmub_diagnostic_data debug; struct dmub_fb lsdma_rb_fb; + struct dmub_preos_info preos_info; }; /** @@ -1068,4 +1088,14 @@ enum dmub_status dmub_srv_wait_for_inbox_free(struct dmub_srv *dmub, */ enum dmub_status dmub_srv_update_inbox_status(struct dmub_srv *dmub); +/** + * dmub_srv_get_preos_info() - retrieves preos fw info + * @dmub: the dmub service + * + * Return: + * true - preos fw info retrieved successfully + * false - preos fw info not retrieved successfully + */ +bool dmub_srv_get_preos_info(struct dmub_srv *dmub); + #endif /* _DMUB_SRV_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h index 92248224b713..3f2a0ed02c59 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -485,7 +485,19 @@ union replay_debug_flags { */ uint32_t enable_visual_confirm_debug : 1; - uint32_t reserved : 18; + /** + * 0x4000 (bit 14) + * @debug_log_enabled: Debug Log Enabled + */ + uint32_t debug_log_enabled : 1; + + /** + * 0x8000 (bit 15) + * @enable_sub_feature_visual_confirm: Enable Sub Feature Visual Confirm + */ + uint32_t enable_sub_feature_visual_confirm : 1; + + uint32_t reserved : 16; } bitfields; uint32_t u32All; @@ -593,6 +605,104 @@ union replay_hw_flags { uint32_t u32All; }; +/** + * Flags that can be set by driver to change some Panel Replay behaviour. + */ +union pr_debug_flags { + struct { + /** + * 0x1 (bit 0) + * Enable visual confirm in FW. + */ + uint32_t visual_confirm : 1; + + /** + * 0x2 (bit 1) + * @skip_crc: Set if need to skip CRC. + */ + uint32_t skip_crc : 1; + + /** + * 0x4 (bit 2) + * @force_link_power_on: Force disable ALPM control + */ + uint32_t force_link_power_on : 1; + + /** + * 0x8 (bit 3) + * @force_phy_power_on: Force phy power on + */ + uint32_t force_phy_power_on : 1; + + /** + * 0x10 (bit 4) + * @skip_crtc_disabled: CRTC disable skipped + */ + uint32_t skip_crtc_disabled : 1; + + /* + * 0x20 (bit 5) + * @visual_confirm_rate_control: Enable Visual Confirm rate control detection + */ + uint32_t visual_confirm_rate_control : 1; + + uint32_t reserved : 26; + } bitfields; + + uint32_t u32All; +}; + +union pr_hw_flags { + struct { + /** + * @allow_alpm_fw_standby_mode: To indicate whether the + * ALPM FW standby mode is allowed + */ + uint32_t allow_alpm_fw_standby_mode : 1; + + /* + * @dsc_enable_status: DSC enable status in driver + */ + uint32_t dsc_enable_status : 1; + + /** + * @fec_enable_status: receive fec enable/disable status from driver + */ + uint32_t fec_enable_status : 1; + + /* + * @smu_optimizations_en: SMU power optimization. + * Only when active display is Replay capable and display enters Replay. + * Trigger interrupt to SMU to powerup/down. + */ + uint32_t smu_optimizations_en : 1; + + /** + * @phy_power_state: Indicates current phy power state + */ + uint32_t phy_power_state : 1; + + /** + * @link_power_state: Indicates current link power state + */ + uint32_t link_power_state : 1; + /** + * Use TPS3 signal when restore main link. + */ + uint32_t force_wakeup_by_tps3 : 1; + /** + * @is_alpm_initialized: Indicates whether ALPM is initialized + */ + uint32_t is_alpm_initialized : 1; + /** + * @alpm_mode: Indicates ALPM mode selected + */ + uint32_t alpm_mode : 2; + } bitfields; + + uint32_t u32All; +}; + union fw_assisted_mclk_switch_version { struct { uint8_t minor : 5; @@ -617,6 +727,7 @@ struct dmub_feature_caps { uint8_t replay_supported; uint8_t replay_reserved[3]; uint8_t abm_aux_backlight_support; + uint8_t lsdma_support_in_dmu; }; struct dmub_visual_confirm_color { @@ -629,6 +740,112 @@ struct dmub_visual_confirm_color { uint16_t panel_inst; }; +/** + * struct dmub_cursor_offload_pipe_data_dcn30_v1 - DCN30+ per pipe data. + */ +struct dmub_cursor_offload_pipe_data_dcn30_v1 { + uint32_t CURSOR0_0_CURSOR_SURFACE_ADDRESS; + uint32_t CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH; + uint32_t CURSOR0_0_CURSOR_SIZE__CURSOR_WIDTH : 16; + uint32_t CURSOR0_0_CURSOR_SIZE__CURSOR_HEIGHT : 16; + uint32_t CURSOR0_0_CURSOR_POSITION__CURSOR_X_POSITION : 16; + uint32_t CURSOR0_0_CURSOR_POSITION__CURSOR_Y_POSITION : 16; + uint32_t CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_X : 16; + uint32_t CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_Y : 16; + uint32_t CURSOR0_0_CURSOR_DST_OFFSET__CURSOR_DST_X_OFFSET : 13; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_ENABLE : 1; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE : 3; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY : 1; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH : 2; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK : 5; + uint32_t reserved0[4]; + uint32_t CNVC_CUR0_CURSOR0_CONTROL__CUR0_ENABLE : 1; + uint32_t CNVC_CUR0_CURSOR0_CONTROL__CUR0_MODE : 3; + uint32_t CNVC_CUR0_CURSOR0_CONTROL__CUR0_EXPANSION_MODE : 1; + uint32_t CNVC_CUR0_CURSOR0_CONTROL__CUR0_ROM_EN : 1; + uint32_t CNVC_CUR0_CURSOR0_COLOR0__CUR0_COLOR0 : 24; + uint32_t CNVC_CUR0_CURSOR0_COLOR1__CUR0_COLOR1 : 24; + uint32_t CNVC_CUR0_CURSOR0_FP_SCALE_BIAS__CUR0_FP_BIAS : 16; + uint32_t CNVC_CUR0_CURSOR0_FP_SCALE_BIAS__CUR0_FP_SCALE, : 16; + uint32_t reserved1[5]; + uint32_t HUBPREQ0_CURSOR_SETTINGS__CURSOR0_DST_Y_OFFSET : 8; + uint32_t HUBPREQ0_CURSOR_SETTINGS__CURSOR0_CHUNK_HDL_ADJUST : 8; + uint32_t reserved2[3]; +}; + +/** + * struct dmub_cursor_offload_pipe_data_dcn401_v1 - DCN401 per pipe data. + */ +struct dmub_cursor_offload_pipe_data_dcn401_v1 { + uint32_t CURSOR0_0_CURSOR_SURFACE_ADDRESS; + uint32_t CURSOR0_0_CURSOR_SURFACE_ADDRESS_HIGH; + uint32_t CURSOR0_0_CURSOR_SIZE__CURSOR_WIDTH : 16; + uint32_t CURSOR0_0_CURSOR_SIZE__CURSOR_HEIGHT : 16; + uint32_t CURSOR0_0_CURSOR_POSITION__CURSOR_X_POSITION : 16; + uint32_t CURSOR0_0_CURSOR_POSITION__CURSOR_Y_POSITION : 16; + uint32_t CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_X : 16; + uint32_t CURSOR0_0_CURSOR_HOT_SPOT__CURSOR_HOT_SPOT_Y : 16; + uint32_t CURSOR0_0_CURSOR_DST_OFFSET__CURSOR_DST_X_OFFSET : 13; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_ENABLE : 1; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_MODE : 3; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_2X_MAGNIFY : 1; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_PITCH : 2; + uint32_t CURSOR0_0_CURSOR_CONTROL__CURSOR_LINES_PER_CHUNK : 5; + uint32_t reserved0[4]; + uint32_t CM_CUR0_CURSOR0_CONTROL__CUR0_ENABLE : 1; + uint32_t CM_CUR0_CURSOR0_CONTROL__CUR0_MODE : 3; + uint32_t CM_CUR0_CURSOR0_CONTROL__CUR0_EXPANSION_MODE : 1; + uint32_t CM_CUR0_CURSOR0_CONTROL__CUR0_ROM_EN : 1; + uint32_t CM_CUR0_CURSOR0_COLOR0__CUR0_COLOR0 : 24; + uint32_t CM_CUR0_CURSOR0_COLOR1__CUR0_COLOR1 : 24; + uint32_t CM_CUR0_CURSOR0_FP_SCALE_BIAS_G_Y__CUR0_FP_BIAS_G_Y : 16; + uint32_t CM_CUR0_CURSOR0_FP_SCALE_BIAS_G_Y__CUR0_FP_SCALE_G_Y, : 16; + uint32_t CM_CUR0_CURSOR0_FP_SCALE_BIAS_RB_CRCB__CUR0_FP_BIAS_RB_CRCB : 16; + uint32_t CM_CUR0_CURSOR0_FP_SCALE_BIAS_RB_CRCB__CUR0_FP_SCALE_RB_CRCB : 16; + uint32_t reserved1[4]; + uint32_t HUBPREQ0_CURSOR_SETTINGS__CURSOR0_DST_Y_OFFSET : 8; + uint32_t HUBPREQ0_CURSOR_SETTINGS__CURSOR0_CHUNK_HDL_ADJUST : 8; + uint32_t HUBP0_DCHUBP_MALL_CONFIG__USE_MALL_FOR_CURSOR : 1; + uint32_t reserved2[3]; +}; + +/** + * struct dmub_cursor_offload_pipe_data_v1 - Per pipe data for cursor offload. + */ +struct dmub_cursor_offload_pipe_data_v1 { + union { + struct dmub_cursor_offload_pipe_data_dcn30_v1 dcn30; /**< DCN30 cursor data. */ + struct dmub_cursor_offload_pipe_data_dcn401_v1 dcn401; /**< DCN401 cursor data. */ + uint8_t payload[96]; /**< Guarantees the cursor pipe data size per-pipe. */ + }; +}; + +/** + * struct dmub_cursor_offload_payload_data_v1 - A payload of stream data. + */ +struct dmub_cursor_offload_payload_data_v1 { + uint32_t write_idx_start; /**< Write index, updated before pipe_data is written. */ + uint32_t write_idx_finish; /**< Write index, updated after pipe_data is written. */ + uint32_t pipe_mask; /**< Mask of pipes to update. */ + uint32_t reserved; /**< Reserved for future use. */ + struct dmub_cursor_offload_pipe_data_v1 pipe_data[6]; /**< Per-pipe cursor data. */ +}; + +/** + * struct dmub_cursor_offload_stream_v1 - Per-stream data for cursor offload. + */ +struct dmub_cursor_offload_stream_v1 { + struct dmub_cursor_offload_payload_data_v1 payloads[4]; /**< A small buffer of cursor payloads. */ + uint32_t write_idx; /**< The index of the last written payload. */ +}; + +/** + * struct dmub_cursor_offload_v1 - Cursor offload feature state. + */ +struct dmub_cursor_offload_v1 { + struct dmub_cursor_offload_stream_v1 offload_streams[6]; /**< Per-stream cursor offload data */ +}; + //============================================================================== //</DMUB_TYPES>================================================================= //============================================================================== @@ -648,7 +865,8 @@ struct dmub_visual_confirm_color { union dmub_fw_meta_feature_bits { struct { uint32_t shared_state_link_detection : 1; /**< 1 supports link detection via shared state */ - uint32_t reserved : 31; + uint32_t cursor_offload_v1_support: 1; /**< 1 supports cursor offload */ + uint32_t reserved : 30; } bits; /**< status bits */ uint32_t all; /**< 32-bit access to status bits */ }; @@ -814,6 +1032,28 @@ enum dmub_ips_comand_type { }; /** + * enum dmub_cursor_offload_comand_type - Cursor offload subcommands. + */ +enum dmub_cursor_offload_comand_type { + /** + * Initializes the cursor offload feature. + */ + DMUB_CMD__CURSOR_OFFLOAD_INIT = 0, + /** + * Enables cursor offloading for a stream and updates the timing parameters. + */ + DMUB_CMD__CURSOR_OFFLOAD_STREAM_ENABLE = 1, + /** + * Disables cursor offloading for a given stream. + */ + DMUB_CMD__CURSOR_OFFLOAD_STREAM_DISABLE = 2, + /** + * Programs the latest data for a given stream. + */ + DMUB_CMD__CURSOR_OFFLOAD_STREAM_PROGRAM = 3, +}; + +/** * union dmub_fw_boot_options - Boot option definitions for SCRATCH14 */ union dmub_fw_boot_options { @@ -844,7 +1084,8 @@ union dmub_fw_boot_options { uint32_t disable_sldo_opt: 1; /**< 1 to disable SLDO optimizations */ uint32_t lower_hbr3_phy_ssc: 1; /**< 1 to lower hbr3 phy ssc to 0.125 percent */ uint32_t override_hbr3_pll_vco: 1; /**< 1 to override the hbr3 pll vco to 0 */ - uint32_t reserved : 5; /**< reserved */ + uint32_t disable_dpia_bw_allocation: 1; /**< 1 to disable the USB4 DPIA BW allocation */ + uint32_t reserved : 4; /**< reserved */ } bits; /**< boot bits */ uint32_t all; /**< 32-bit access to bits */ }; @@ -877,6 +1118,7 @@ enum dmub_shared_state_feature_id { DMUB_SHARED_SHARE_FEATURE__IPS_FW = 1, DMUB_SHARED_SHARE_FEATURE__IPS_DRIVER = 2, DMUB_SHARED_SHARE_FEATURE__DEBUG_SETUP = 3, + DMUB_SHARED_STATE_FEATURE__CURSOR_OFFLOAD_V1 = 4, DMUB_SHARED_STATE_FEATURE__LAST, /* Total number of features. */ }; @@ -958,6 +1200,22 @@ struct dmub_shared_state_ips_driver { }; /* 248-bytes, fixed */ /** + * struct dmub_shared_state_cursor_offload_v1 - Header metadata for cursor offload. + */ +struct dmub_shared_state_cursor_offload_stream_v1 { + uint32_t last_write_idx; /**< Last write index */ + uint8_t reserved[28]; /**< Reserved bytes. */ +}; /* 32-bytes, fixed */ + +/** + * struct dmub_shared_state_cursor_offload_v1 - Header metadata for cursor offload. + */ +struct dmub_shared_state_cursor_offload_v1 { + struct dmub_shared_state_cursor_offload_stream_v1 offload_streams[6]; /**< stream state, 32-bytes each */ + uint8_t reserved[56]; /**< reserved for future use */ +}; /* 248-bytes, fixed */ + +/** * enum dmub_shared_state_feature_common - Generic payload. */ struct dmub_shared_state_feature_common { @@ -983,6 +1241,7 @@ struct dmub_shared_state_feature_block { struct dmub_shared_state_ips_fw ips_fw; /**< IPS firmware state */ struct dmub_shared_state_ips_driver ips_driver; /**< IPS driver state */ struct dmub_shared_state_debug_setup debug_setup; /**< Debug setup */ + struct dmub_shared_state_cursor_offload_v1 cursor_offload_v1; /**< Cursor offload */ } data; /**< Shared state data. */ }; /* 256-bytes, fixed */ @@ -1572,6 +1831,25 @@ enum dmub_cmd_type { */ DMUB_CMD__IPS = 91, + /** + * Command type use for Cursor offload. + */ + DMUB_CMD__CURSOR_OFFLOAD = 92, + + /** + * Command type used for all SMART_POWER_OLED commands. + */ + DMUB_CMD__SMART_POWER_OLED = 93, + + /** + * Command type use for all Panel Replay commands. + */ + DMUB_CMD__PR = 94, + + + /** + * Command type use for VBIOS shared commands. + */ DMUB_CMD__VBIOS = 128, }; @@ -2369,6 +2647,7 @@ struct dmub_cmd_fams2_global_config { union dmub_cmd_fams2_config { struct dmub_cmd_fams2_global_config global; +// coverity[cert_dcl37_c_violation:FALSE] errno.h, stddef.h, stdint.h not included in atombios.h struct dmub_fams2_stream_static_state stream; //v0 union { struct dmub_fams2_cmd_stream_static_base_state base; @@ -3981,6 +4260,33 @@ enum replay_state { }; /** + * Definition of a panel replay state + */ +enum pr_state { + PR_STATE_0 = 0x00, // State 0 steady state + // Pending SDP and Unlock before back to State 0 + PR_STATE_0_PENDING_SDP_AND_UNLOCK = 0x01, + PR_STATE_1 = 0x10, // State 1 + PR_STATE_2 = 0x20, // State 2 steady state + // Pending frame transmission before transition to State 2 + PR_STATE_2_PENDING_FRAME_TRANSMISSION = 0x30, + // Active and Powered Up + PR_STATE_2_POWERED = 0x31, + // Active and Powered Down, but need to blank HUBP after DPG_EN latch + PR_STATE_2_PENDING_HUBP_BLANK = 0x32, + // Active and Pending Power Up + PR_STATE_2_PENDING_POWER_UP = 0x33, + // Active and Powered Up, Pending DPG latch + PR_STATE_2_PENDING_LOCK_FOR_DPG_POWER_ON = 0x34, + // Active and Powered Up, Pending SDP and Unlock + PR_STATE_2_PENDING_SDP_AND_UNLOCK = 0x35, + // Pending transmission of AS SDP for timing sync, but no rfb update + PR_STATE_2_PENDING_AS_SDP = 0x36, + // Invalid + PR_STATE_INVALID = 0xFF, +}; + +/** * Replay command sub-types. */ enum dmub_cmd_replay_type { @@ -4030,6 +4336,25 @@ enum dmub_cmd_replay_type { DMUB_CMD__REPLAY_SET_GENERAL_CMD = 16, }; +/* + * Panel Replay sub-types + */ +enum dmub_cmd_panel_replay_type { + DMUB_CMD__PR_ENABLE = 0, + DMUB_CMD__PR_COPY_SETTINGS = 1, + DMUB_CMD__PR_UPDATE_STATE = 2, + DMUB_CMD__PR_GENERAL_CMD = 3, +}; + +enum dmub_cmd_panel_replay_state_update_subtype { + PR_STATE_UPDATE_COASTING_VTOTAL = 0x1, + PR_STATE_UPDATE_SYNC_MODE = 0x2, +}; + +enum dmub_cmd_panel_replay_general_subtype { + PR_GENERAL_CMD_DEBUG_OPTION = 0x1, +}; + /** * Replay general command sub-types. */ @@ -4045,6 +4370,7 @@ enum dmub_cmd_replay_general_subtype { REPLAY_GENERAL_CMD_DISABLED_DESYNC_ERROR_DETECTION, REPLAY_GENERAL_CMD_UPDATE_ERROR_STATUS, REPLAY_GENERAL_CMD_SET_LOW_RR_ACTIVATE, + REPLAY_GENERAL_CMD_VIDEO_CONFERENCING, }; struct dmub_alpm_auxless_data { @@ -4182,17 +4508,13 @@ struct dmub_cmd_replay_set_version_data { */ uint8_t panel_inst; /** - * PSR version that FW should implement. + * Replay version that FW should implement. */ enum replay_version version; /** - * PSR control version. - */ - uint8_t cmd_version; - /** * Explicit padding to 4 byte boundary. */ - uint8_t pad[2]; + uint8_t pad[3]; }; /** @@ -4238,6 +4560,45 @@ enum replay_enable { }; /** + * Data passed from driver to FW in a DMUB_CMD__SMART_POWER_OLED_ENABLE command. + */ +struct dmub_rb_cmd_smart_power_oled_enable_data { + /** + * SMART_POWER_OLED enable or disable. + */ + uint8_t enable; + /** + * Panel Instance. + * Panel isntance to identify which replay_state to use + * Currently the support is only for 0 or 1 + */ + uint8_t panel_inst; + + uint16_t peak_nits; + /** + * OTG HW instance. + */ + uint8_t otg_inst; + /** + * DIG FE HW instance. + */ + uint8_t digfe_inst; + /** + * DIG BE HW instance. + */ + uint8_t digbe_inst; + uint8_t debugcontrol; + /* + * vertical interrupt trigger line + */ + uint32_t triggerline; + + uint16_t fixed_max_cll; + + uint8_t pad[2]; +}; + +/** * Data passed from driver to FW in a DMUB_CMD__REPLAY_ENABLE command. */ struct dmub_rb_cmd_replay_enable_data { @@ -4408,9 +4769,9 @@ struct dmub_cmd_replay_set_coasting_vtotal_data { */ uint16_t coasting_vtotal_high; /** - * Explicit padding to 4 byte boundary. + * frame skip number. */ - uint8_t pad[2]; + uint16_t frame_skip_number; }; /** @@ -4571,6 +4932,58 @@ union dmub_replay_cmd_set { }; /** + * SMART POWER OLED command sub-types. + */ +enum dmub_cmd_smart_power_oled_type { + + /** + * Enable/Disable SMART_POWER_OLED. + */ + DMUB_CMD__SMART_POWER_OLED_ENABLE = 1, + /** + * Get current MaxCLL value if SMART POWER OLED is enabled. + */ + DMUB_CMD__SMART_POWER_OLED_GETMAXCLL = 2, +}; + +/** + * Definition of a DMUB_CMD__SMART_POWER_OLED command. + */ +struct dmub_rb_cmd_smart_power_oled_enable { + /** + * Command header. + */ + struct dmub_cmd_header header; + + struct dmub_rb_cmd_smart_power_oled_enable_data data; +}; + +struct dmub_cmd_smart_power_oled_getmaxcll_input { + uint8_t panel_inst; + uint8_t pad[3]; +}; + +struct dmub_cmd_smart_power_oled_getmaxcll_output { + uint16_t current_max_cll; + uint8_t pad[2]; +}; + +/** + * Definition of a DMUB_CMD__SMART_POWER_OLED command. + */ +struct dmub_rb_cmd_smart_power_oled_getmaxcll { + struct dmub_cmd_header header; /**< Command header */ + /** + * Data passed from driver to FW in a DMUB_CMD__SMART_POWER_OLED_GETMAXCLL command. + */ + union dmub_cmd_smart_power_oled_getmaxcll_data { + struct dmub_cmd_smart_power_oled_getmaxcll_input input; /**< Input */ + struct dmub_cmd_smart_power_oled_getmaxcll_output output; /**< Output */ + uint32_t output_raw; /**< Raw data output */ + } data; +}; + +/** * Set of HW components that can be locked. * * Note: If updating with more HW components, fields @@ -4652,6 +5065,7 @@ enum hw_lock_client { */ HW_LOCK_CLIENT_REPLAY = 4, HW_LOCK_CLIENT_FAMS2 = 5, + HW_LOCK_CLIENT_CURSOR_OFFLOAD = 6, /** * Invalid client. */ @@ -6064,6 +6478,257 @@ struct dmub_rb_cmd_ips_query_residency_info { }; /** + * struct dmub_cmd_cursor_offload_init_data - Payload for cursor offload init command. + */ +struct dmub_cmd_cursor_offload_init_data { + union dmub_addr state_addr; /**< State address for dmub_cursor_offload */ + uint32_t state_size; /**< State size for dmub_cursor_offload */ +}; + +/** + * struct dmub_rb_cmd_cursor_offload_init - Data for initializing cursor offload. + */ +struct dmub_rb_cmd_cursor_offload_init { + struct dmub_cmd_header header; + struct dmub_cmd_cursor_offload_init_data init_data; +}; + +/** + * struct dmub_cmd_cursor_offload_stream_data - Payload for cursor offload stream command. + */ +struct dmub_cmd_cursor_offload_stream_data { + uint32_t otg_inst: 4; /**< OTG instance to control */ + uint32_t reserved: 28; /**< Reserved for future use */ + uint32_t line_time_in_ns; /**< Line time in ns for the OTG */ + uint32_t v_total_max; /**< OTG v_total_max */ +}; + +/** + * struct dmub_rb_cmd_cursor_offload_stream_cntl - Controls a stream for cursor offload. + */ +struct dmub_rb_cmd_cursor_offload_stream_cntl { + struct dmub_cmd_header header; + struct dmub_cmd_cursor_offload_stream_data data; +}; + +/** + * Data passed from driver to FW in a DMUB_CMD__PR_ENABLE command. + */ +struct dmub_cmd_pr_enable_data { + /** + * Panel Replay enable or disable. + */ + uint8_t enable; + /** + * Panel Instance. + * Panel isntance to identify which replay_state to use + * Currently the support is only for 0 or 1 + */ + uint8_t panel_inst; + /** + * Phy state to enter. + * Values to use are defined in dmub_phy_fsm_state + */ + uint8_t phy_fsm_state; + /** + * Phy rate for DP - RBR/HBR/HBR2/HBR3. + * Set this using enum phy_link_rate. + * This does not support HDMI/DP2 for now. + */ + uint8_t phy_rate; + /** + * @hpo_stream_enc_inst: HPO stream encoder instance + */ + uint8_t hpo_stream_enc_inst; + /** + * @hpo_link_enc_inst: HPO link encoder instance + */ + uint8_t hpo_link_enc_inst; + /** + * @pad: Align structure to 4 byte boundary. + */ + uint8_t pad[2]; +}; + +/** + * Definition of a DMUB_CMD__PR_ENABLE command. + * Panel Replay enable/disable is controlled using action in data. + */ +struct dmub_rb_cmd_pr_enable { + /** + * Command header. + */ + struct dmub_cmd_header header; + + struct dmub_cmd_pr_enable_data data; +}; + +/** + * Data passed from driver to FW in a DMUB_CMD__PR_COPY_SETTINGS command. + */ +struct dmub_cmd_pr_copy_settings_data { + /** + * Flags that can be set by driver to change some replay behaviour. + */ + union pr_debug_flags debug; + + /** + * @flags: Flags used to determine feature functionality. + */ + union pr_hw_flags flags; + + /** + * DPP HW instance. + */ + uint8_t dpp_inst; + /** + * OTG HW instance. + */ + uint8_t otg_inst; + /** + * DIG FE HW instance. + */ + uint8_t digfe_inst; + /** + * DIG BE HW instance. + */ + uint8_t digbe_inst; + /** + * AUX HW instance. + */ + uint8_t aux_inst; + /** + * Panel Instance. + * Panel isntance to identify which psr_state to use + * Currently the support is only for 0 or 1 + */ + uint8_t panel_inst; + /** + * Length of each horizontal line in ns. + */ + uint32_t line_time_in_ns; + /** + * PHY instance. + */ + uint8_t dpphy_inst; + /** + * Determines if SMU optimzations are enabled/disabled. + */ + uint8_t smu_optimizations_en; + /* + * Use FSM state for Replay power up/down + */ + uint8_t use_phy_fsm; + /* + * Use FSFT afftet pixel clk + */ + uint32_t pix_clk_100hz; + /* + * Use Original pixel clock + */ + uint32_t sink_pix_clk_100hz; + /** + * Use for AUX-less ALPM LFPS wake operation + */ + struct dmub_alpm_auxless_data auxless_alpm_data; + /** + * @hpo_stream_enc_inst: HPO stream encoder instance + */ + uint8_t hpo_stream_enc_inst; + /** + * @hpo_link_enc_inst: HPO link encoder instance + */ + uint8_t hpo_link_enc_inst; + /** + * @pad: Align structure to 4 byte boundary. + */ + uint8_t pad[2]; +}; + +/** + * Definition of a DMUB_CMD__PR_COPY_SETTINGS command. + */ +struct dmub_rb_cmd_pr_copy_settings { + /** + * Command header. + */ + struct dmub_cmd_header header; + /** + * Data passed from driver to FW in a DMUB_CMD__PR_COPY_SETTINGS command. + */ + struct dmub_cmd_pr_copy_settings_data data; +}; + +struct dmub_cmd_pr_update_state_data { + /** + * Panel Instance. + * Panel isntance to identify which psr_state to use + * Currently the support is only for 0 or 1 + */ + uint8_t panel_inst; + + uint8_t pad[3]; // align to 4-byte boundary + /* + * Update flags to control the update behavior. + */ + uint32_t update_flag; + /** + * state/data to set. + */ + uint32_t coasting_vtotal; + uint32_t sync_mode; +}; + +struct dmub_cmd_pr_general_cmd_data { + /** + * Panel Instance. + * Panel isntance to identify which psr_state to use + * Currently the support is only for 0 or 1 + */ + uint8_t panel_inst; + /** + * subtype: PR general cmd sub type + */ + uint8_t subtype; + + uint8_t pad[2]; + /** + * config data by different subtypes + */ + union { + uint32_t u32All; + } data; +}; + +/** + * Definition of a DMUB_CMD__PR_UPDATE_STATE command. + */ +struct dmub_rb_cmd_pr_update_state { + /** + * Command header. + */ + struct dmub_cmd_header header; + /** + * Data passed from driver to FW in a DMUB_CMD__PR_UPDATE_STATE command. + */ + struct dmub_cmd_pr_update_state_data data; +}; + +/** + * Definition of a DMUB_CMD__PR_GENERAL_CMD command. + */ +struct dmub_rb_cmd_pr_general_cmd { + /** + * Command header. + */ + struct dmub_cmd_header header; + /** + * Data passed from driver to FW in a DMUB_CMD__PR_GENERAL_CMD command. + */ + struct dmub_cmd_pr_general_cmd_data data; +}; + +/** * union dmub_rb_cmd - DMUB inbox command. */ union dmub_rb_cmd { @@ -6392,6 +7057,38 @@ union dmub_rb_cmd { struct dmub_rb_cmd_ips_residency_cntl ips_residency_cntl; struct dmub_rb_cmd_ips_query_residency_info ips_query_residency_info; + /** + * Definition of a DMUB_CMD__CURSOR_OFFLOAD_INIT command. + */ + struct dmub_rb_cmd_cursor_offload_init cursor_offload_init; + /** + * Definition of a DMUB_CMD__CURSOR_OFFLOAD control commands. + * - DMUB_CMD__CURSOR_OFFLOAD_STREAM_ENABLE + * - DMUB_CMD__CURSOR_OFFLOAD_STREAM_DISABLE + * - DMUB_CMD__CURSOR_OFFLOAD_STREAM_PROGRAM + * - DMUB_CMD__CURSOR_OFFLOAD_STREAM_UPDATE_DRR + */ + struct dmub_rb_cmd_cursor_offload_stream_cntl cursor_offload_stream_ctnl; + /** + * Definition of a DMUB_CMD__SMART_POWER_OLED_ENABLE command. + */ + struct dmub_rb_cmd_smart_power_oled_enable smart_power_oled_enable; + /** + * Definition of a DMUB_CMD__DMUB_CMD__SMART_POWER_OLED_GETMAXCLL command. + */ + struct dmub_rb_cmd_smart_power_oled_getmaxcll smart_power_oled_getmaxcll; + /* + * Definition of a DMUB_CMD__REPLAY_COPY_SETTINGS command. + */ + struct dmub_rb_cmd_pr_copy_settings pr_copy_settings; + /** + * Definition of a DMUB_CMD__REPLAY_ENABLE command. + */ + struct dmub_rb_cmd_pr_enable pr_enable; + + struct dmub_rb_cmd_pr_update_state pr_update_state; + + struct dmub_rb_cmd_pr_general_cmd pr_general_cmd; }; /** diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c index 4777c7203b2c..cd04d7c756c3 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c @@ -380,6 +380,7 @@ void dmub_dcn31_enable_dmub_boot_options(struct dmub_srv *dmub, const struct dmu boot_options.bits.override_hbr3_pll_vco = params->override_hbr3_pll_vco; boot_options.bits.sel_mux_phy_c_d_phy_f_g = (dmub->asic == DMUB_ASIC_DCN31B) ? 1 : 0; + boot_options.bits.disable_dpia_bw_allocation = params->disable_dpia_bw_allocation; REG_WRITE(DMCUB_SCRATCH14, boot_options.all); } diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c index ce041f6239dc..7e9856289910 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn32.c @@ -89,50 +89,58 @@ static inline void dmub_dcn32_translate_addr(const union dmub_addr *addr_in, void dmub_dcn32_reset(struct dmub_srv *dmub) { union dmub_gpint_data_register cmd; - const uint32_t timeout = 100000; - uint32_t in_reset, is_enabled, scratch, i, pwait_mode; + const uint32_t timeout_us = 1 * 1000 * 1000; //1s + const uint32_t poll_delay_us = 1; //1us + uint32_t i = 0; + uint32_t enabled, in_reset, scratch, pwait_mode; - REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); - REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_enabled); + REG_GET(DMCUB_CNTL, + DMCUB_ENABLE, &enabled); + REG_GET(DMCUB_CNTL2, + DMCUB_SOFT_RESET, &in_reset); - if (in_reset == 0 && is_enabled != 0) { + if (enabled && in_reset == 0) { cmd.bits.status = 1; cmd.bits.command_code = DMUB_GPINT__STOP_FW; cmd.bits.param = 0; dmub->hw_funcs.set_gpint(dmub, cmd); - for (i = 0; i < timeout; ++i) { - if (dmub->hw_funcs.is_gpint_acked(dmub, cmd)) - break; - - udelay(1); - } - - for (i = 0; i < timeout; ++i) { + for (; i < timeout_us; i++) { scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; - udelay(1); + udelay(poll_delay_us); } - for (i = 0; i < timeout; ++i) { + for (; i < timeout_us; i++) { REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &pwait_mode); if (pwait_mode & (1 << 0)) break; - udelay(1); + udelay(poll_delay_us); } - /* Force reset in case we timed out, DMCUB is likely hung. */ } - if (is_enabled) { + if (enabled) { REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); udelay(1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); } + if (i >= timeout_us) { + /* timeout should never occur */ + BREAK_TO_DEBUGGER(); + } + + REG_UPDATE(DMCUB_REGION3_CW2_TOP_ADDRESS, DMCUB_REGION3_CW2_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW3_TOP_ADDRESS, DMCUB_REGION3_CW3_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW4_TOP_ADDRESS, DMCUB_REGION3_CW4_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW5_TOP_ADDRESS, DMCUB_REGION3_CW5_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW7_TOP_ADDRESS, DMCUB_REGION3_CW7_ENABLE, 0); + REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); @@ -141,7 +149,7 @@ void dmub_dcn32_reset(struct dmub_srv *dmub) REG_WRITE(DMCUB_OUTBOX0_WPTR, 0); REG_WRITE(DMCUB_SCRATCH0, 0); - /* Clear the GPINT command manually so we don't send anything during boot. */ + /* Clear the GPINT command manually so we don't reset again. */ cmd.all = 0; dmub->hw_funcs.set_gpint(dmub, cmd); } @@ -163,7 +171,9 @@ void dmub_dcn32_backdoor_load(struct dmub_srv *dmub, dmub_dcn32_get_fb_base_offset(dmub, &fb_base, &fb_offset); + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); dmub_dcn32_translate_addr(&cw0->offset, fb_base, fb_offset, &offset); @@ -193,7 +203,9 @@ void dmub_dcn32_backdoor_load_zfb_mode(struct dmub_srv *dmub, { union dmub_addr offset; + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); offset = cw0->offset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c index 834e5434ccb8..e13557ed97be 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c @@ -418,6 +418,7 @@ void dmub_dcn35_enable_dmub_boot_options(struct dmub_srv *dmub, const struct dmu boot_options.bits.disable_sldo_opt = params->disable_sldo_opt; boot_options.bits.enable_non_transparent_setconfig = params->enable_non_transparent_setconfig; boot_options.bits.lower_hbr3_phy_ssc = params->lower_hbr3_phy_ssc; + boot_options.bits.disable_dpia_bw_allocation = params->disable_dpia_bw_allocation; REG_WRITE(DMCUB_SCRATCH14, boot_options.all); } @@ -520,6 +521,45 @@ void dmub_dcn35_get_diagnostic_data(struct dmub_srv *dmub) dmub->debug.gpint_datain0 = REG_READ(DMCUB_GPINT_DATAIN0); } + +bool dmub_dcn35_get_preos_fw_info(struct dmub_srv *dmub) +{ + uint64_t region3_cw5_offset; + uint32_t top_addr, top_addr_enable, offset_low; + uint32_t offset_high, base_addr, fw_version; + bool is_vbios_fw = false; + + memset(&dmub->preos_info, 0, sizeof(dmub->preos_info)); + + fw_version = REG_READ(DMCUB_SCRATCH1); + is_vbios_fw = ((fw_version >> 6) & 0x01) ? true : false; + if (!is_vbios_fw) + return false; + + dmub->preos_info.boot_status = REG_READ(DMCUB_SCRATCH0); + dmub->preos_info.fw_version = REG_READ(DMCUB_SCRATCH1); + dmub->preos_info.boot_options = REG_READ(DMCUB_SCRATCH14); + REG_GET(DMCUB_REGION3_CW5_TOP_ADDRESS, + DMCUB_REGION3_CW5_ENABLE, &top_addr_enable); + if (top_addr_enable) { + dmub_dcn35_get_fb_base_offset(dmub, + &dmub->preos_info.fb_base, &dmub->preos_info.fb_offset); + offset_low = REG_READ(DMCUB_REGION3_CW5_OFFSET); + offset_high = REG_READ(DMCUB_REGION3_CW5_OFFSET_HIGH); + region3_cw5_offset = ((uint64_t)offset_high << 32) | offset_low; + dmub->preos_info.trace_buffer_phy_addr = region3_cw5_offset + - dmub->preos_info.fb_base + dmub->preos_info.fb_offset; + + REG_GET(DMCUB_REGION3_CW5_TOP_ADDRESS, + DMCUB_REGION3_CW5_TOP_ADDRESS, &top_addr); + base_addr = REG_READ(DMCUB_REGION3_CW5_BASE_ADDRESS) & 0x1FFFFFFF; + dmub->preos_info.trace_buffer_size = + (top_addr > base_addr) ? (top_addr - base_addr + 1) : 0; + } + + return true; +} + void dmub_dcn35_configure_dmub_in_system_memory(struct dmub_srv *dmub) { /* DMCUB_REGION3_TMR_AXI_SPACE values: diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h index 39fcb7275da5..92e6695a2c9b 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.h @@ -285,4 +285,6 @@ bool dmub_dcn35_is_hw_powered_up(struct dmub_srv *dmub); void dmub_srv_dcn35_regs_init(struct dmub_srv *dmub, struct dc_context *ctx); +bool dmub_dcn35_get_preos_fw_info(struct dmub_srv *dmub); + #endif /* _DMUB_DCN35_H_ */ diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c index b31adbd0d685..95542299e3b3 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c @@ -81,7 +81,7 @@ void dmub_dcn401_reset(struct dmub_srv *dmub) dmub->hw_funcs.set_gpint(dmub, cmd); for (; i < timeout_us; i++) { - scratch = dmub->hw_funcs.get_gpint_response(dmub); + scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; @@ -97,11 +97,24 @@ void dmub_dcn401_reset(struct dmub_srv *dmub) } } + if (enabled) { + REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); + udelay(1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); + } + if (i >= timeout_us) { /* timeout should never occur */ BREAK_TO_DEBUGGER(); } + REG_UPDATE(DMCUB_REGION3_CW2_TOP_ADDRESS, DMCUB_REGION3_CW2_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW3_TOP_ADDRESS, DMCUB_REGION3_CW3_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW4_TOP_ADDRESS, DMCUB_REGION3_CW4_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW5_TOP_ADDRESS, DMCUB_REGION3_CW5_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW6_TOP_ADDRESS, DMCUB_REGION3_CW6_ENABLE, 0); + REG_UPDATE(DMCUB_REGION3_CW7_TOP_ADDRESS, DMCUB_REGION3_CW7_ENABLE, 0); + REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); @@ -134,7 +147,6 @@ void dmub_dcn401_backdoor_load(struct dmub_srv *dmub, /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); dmub_dcn401_translate_addr(&cw0->offset, fb_base, fb_offset, &offset); @@ -168,7 +180,6 @@ void dmub_dcn401_backdoor_load_zfb_mode(struct dmub_srv *dmub, /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); offset = cw0->offset; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c index b17a19400c06..a6ae1d2e9685 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c @@ -66,7 +66,7 @@ #define DMUB_SCRATCH_MEM_SIZE (1024) /* Default indirect buffer size. */ -#define DMUB_IB_MEM_SIZE (1280) +#define DMUB_IB_MEM_SIZE (2560) /* Default LSDMA ring buffer size. */ #define DMUB_LSDMA_RB_SIZE (64 * 1024) @@ -359,6 +359,7 @@ static bool dmub_srv_hw_setup(struct dmub_srv *dmub, enum dmub_asic asic) funcs->get_current_time = dmub_dcn35_get_current_time; funcs->get_diagnostic_data = dmub_dcn35_get_diagnostic_data; + funcs->get_preos_fw_info = dmub_dcn35_get_preos_fw_info; funcs->init_reg_offsets = dmub_srv_dcn35_regs_init; if (asic == DMUB_ASIC_DCN351) @@ -564,10 +565,11 @@ enum dmub_status window_sizes[DMUB_WINDOW_4_MAILBOX] = DMUB_MAILBOX_SIZE; window_sizes[DMUB_WINDOW_5_TRACEBUFF] = trace_buffer_size; window_sizes[DMUB_WINDOW_6_FW_STATE] = fw_state_size; - window_sizes[DMUB_WINDOW_7_SCRATCH_MEM] = DMUB_SCRATCH_MEM_SIZE; + window_sizes[DMUB_WINDOW_7_SCRATCH_MEM] = dmub_align(DMUB_SCRATCH_MEM_SIZE, 64); window_sizes[DMUB_WINDOW_IB_MEM] = DMUB_IB_MEM_SIZE; window_sizes[DMUB_WINDOW_SHARED_STATE] = max(DMUB_FW_HEADER_SHARED_STATE_SIZE, shared_state_size); window_sizes[DMUB_WINDOW_LSDMA_BUFFER] = DMUB_LSDMA_RB_SIZE; + window_sizes[DMUB_WINDOW_CURSOR_OFFLOAD] = dmub_align(sizeof(struct dmub_cursor_offload_v1), 64); out->fb_size = dmub_srv_calc_regions_for_memory_type(params, out, window_sizes, DMUB_WINDOW_MEMORY_TYPE_FB); @@ -652,21 +654,22 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, struct dmub_fb *mail_fb = params->fb[DMUB_WINDOW_4_MAILBOX]; struct dmub_fb *tracebuff_fb = params->fb[DMUB_WINDOW_5_TRACEBUFF]; struct dmub_fb *fw_state_fb = params->fb[DMUB_WINDOW_6_FW_STATE]; - struct dmub_fb *scratch_mem_fb = params->fb[DMUB_WINDOW_7_SCRATCH_MEM]; - struct dmub_fb *ib_mem_gart = params->fb[DMUB_WINDOW_IB_MEM]; struct dmub_fb *shared_state_fb = params->fb[DMUB_WINDOW_SHARED_STATE]; struct dmub_rb_init_params rb_params, outbox0_rb_params; struct dmub_window cw0, cw1, cw2, cw3, cw4, cw5, cw6, region6; struct dmub_region inbox1, outbox1, outbox0; + uint32_t i; + if (!dmub->sw_init) return DMUB_STATUS_INVALID; - if (!inst_fb || !stack_fb || !data_fb || !bios_fb || !mail_fb || - !tracebuff_fb || !fw_state_fb || !scratch_mem_fb || !ib_mem_gart) { - ASSERT(0); - return DMUB_STATUS_INVALID; + for (i = 0; i < DMUB_WINDOW_TOTAL; ++i) { + if (!params->fb[i]) { + ASSERT(0); + return DMUB_STATUS_INVALID; + } } dmub->fb_base = params->fb_base; @@ -748,9 +751,11 @@ enum dmub_status dmub_srv_hw_init(struct dmub_srv *dmub, dmub->shared_state = shared_state_fb->cpu_addr; - dmub->scratch_mem_fb = *scratch_mem_fb; + dmub->scratch_mem_fb = *params->fb[DMUB_WINDOW_7_SCRATCH_MEM]; + dmub->ib_mem_gart = *params->fb[DMUB_WINDOW_IB_MEM]; - dmub->ib_mem_gart = *ib_mem_gart; + dmub->cursor_offload_fb = *params->fb[DMUB_WINDOW_CURSOR_OFFLOAD]; + dmub->cursor_offload_v1 = (struct dmub_cursor_offload_v1 *)dmub->cursor_offload_fb.cpu_addr; if (dmub->hw_funcs.setup_windows) dmub->hw_funcs.setup_windows(dmub, &cw2, &cw3, &cw4, &cw5, &cw6, ®ion6); @@ -1368,3 +1373,11 @@ enum dmub_status dmub_srv_update_inbox_status(struct dmub_srv *dmub) return DMUB_STATUS_OK; } + +bool dmub_srv_get_preos_info(struct dmub_srv *dmub) +{ + if (!dmub || !dmub->hw_funcs.get_preos_fw_info) + return false; + + return dmub->hw_funcs.get_preos_fw_info(dmub); +} diff --git a/drivers/gpu/drm/amd/display/include/bios_parser_types.h b/drivers/gpu/drm/amd/display/include/bios_parser_types.h index 812377d9e48f..973b6bdbac63 100644 --- a/drivers/gpu/drm/amd/display/include/bios_parser_types.h +++ b/drivers/gpu/drm/amd/display/include/bios_parser_types.h @@ -135,12 +135,8 @@ struct bp_external_encoder_control { struct bp_crtc_source_select { enum engine_id engine_id; enum controller_id controller_id; - /* from GPU Tx aka asic_signal */ - enum signal_type signal; - /* sink_signal may differ from asicSignal if Translator encoder */ enum signal_type sink_signal; - enum display_output_bit_depth display_output_bit_depth; - bool enable_dp_audio; + uint8_t bit_depth; }; struct bp_transmitter_control { @@ -166,6 +162,11 @@ struct bp_transmitter_control { bool single_pll_mode; }; +struct bp_load_detection_parameters { + enum engine_id engine_id; + uint16_t device_id; +}; + struct bp_hw_crtc_timing_parameters { enum controller_id controller_id; /* horizontal part */ diff --git a/drivers/gpu/drm/amd/display/include/dpcd_defs.h b/drivers/gpu/drm/amd/display/include/dpcd_defs.h index de8f3cfed6c8..07b937b92efc 100644 --- a/drivers/gpu/drm/amd/display/include/dpcd_defs.h +++ b/drivers/gpu/drm/amd/display/include/dpcd_defs.h @@ -30,6 +30,22 @@ #ifndef DP_SINK_HW_REVISION_START // can remove this once the define gets into linux drm_dp_helper.h #define DP_SINK_HW_REVISION_START 0x409 #endif +/* Panel Replay*/ +#ifndef DP_PANEL_REPLAY_CAPABILITY_SUPPORT // can remove this once the define gets into linux drm_dp_helper.h +#define DP_PANEL_REPLAY_CAPABILITY_SUPPORT 0x0b0 +#endif /* DP_PANEL_REPLAY_CAPABILITY_SUPPORT */ +#ifndef DP_PANEL_REPLAY_CAPABILITY // can remove this once the define gets into linux drm_dp_helper.h +#define DP_PANEL_REPLAY_CAPABILITY 0x0b1 +#endif /* DP_PANEL_REPLAY_CAPABILITY */ +#ifndef DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_1 // can remove this once the define gets into linux drm_dp_helper.h +#define DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_1 0x1b0 +#endif /* DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_1 */ +#ifndef DP_PANEL_REPLAY_ENABLE // can remove this once the define gets into linux drm_dp_helper.h +#define DP_PANEL_REPLAY_ENABLE (1 << 0) +#endif /* DP_PANEL_REPLAY_ENABLE */ +#ifndef DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_2 // can remove this once the define gets into linux drm_dp_helper.h +#define DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_2 0x1b1 +#endif /* DP_PANEL_REPLAY_ENABLE_AND_CONFIGURATION_2 */ enum dpcd_revision { DPCD_REV_10 = 0x10, diff --git a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h index cc467031651d..38a77fa9b4af 100644 --- a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h +++ b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h @@ -169,6 +169,7 @@ struct dc_firmware_info { uint32_t engine_clk_ss_percentage; } feature; + uint32_t max_pixel_clock; /* in KHz */ uint32_t default_display_engine_pll_frequency; /* in KHz */ uint32_t external_clock_source_frequency_for_dp; /* in KHz */ uint32_t smu_gpu_pll_output_freq; /* in KHz */ diff --git a/drivers/gpu/drm/amd/display/include/grph_object_id.h b/drivers/gpu/drm/amd/display/include/grph_object_id.h index 54e33062b3c0..1386fa124e85 100644 --- a/drivers/gpu/drm/amd/display/include/grph_object_id.h +++ b/drivers/gpu/drm/amd/display/include/grph_object_id.h @@ -310,4 +310,11 @@ static inline bool dal_graphics_object_id_equal( } return false; } + +static inline bool dc_connector_supports_analog(const enum connector_id conn) +{ + return conn == CONNECTOR_ID_VGA || + conn == CONNECTOR_ID_SINGLE_LINK_DVII || + conn == CONNECTOR_ID_DUAL_LINK_DVII; +} #endif diff --git a/drivers/gpu/drm/amd/display/include/signal_types.h b/drivers/gpu/drm/amd/display/include/signal_types.h index a10d6b988aab..3a2c2d2fb629 100644 --- a/drivers/gpu/drm/amd/display/include/signal_types.h +++ b/drivers/gpu/drm/amd/display/include/signal_types.h @@ -118,6 +118,18 @@ static inline bool dc_is_dvi_signal(enum signal_type signal) } } +/** + * dc_is_rgb_signal() - Whether the signal is analog RGB. + * + * Returns whether the given signal type is an analog RGB signal + * that is used with a DAC on VGA or DVI-I connectors. + * Not to be confused with other uses of "RGB", such as RGB color space. + */ +static inline bool dc_is_rgb_signal(enum signal_type signal) +{ + return (signal == SIGNAL_TYPE_RGB); +} + static inline bool dc_is_tmds_signal(enum signal_type signal) { switch (signal) { diff --git a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c index ce421bcddcb0..1aae46d703ba 100644 --- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c +++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c @@ -1260,6 +1260,17 @@ void mod_freesync_handle_v_update(struct mod_freesync *mod_freesync, update_v_total_for_static_ramp( core_freesync, stream, in_out_vrr); } + + /* + * If VRR is inactive, set vtotal min and max to nominal vtotal + */ + if (in_out_vrr->state == VRR_STATE_INACTIVE) { + in_out_vrr->adjust.v_total_min = + mod_freesync_calc_v_total_from_refresh(stream, + in_out_vrr->max_refresh_in_uhz); + in_out_vrr->adjust.v_total_max = in_out_vrr->adjust.v_total_min; + return; + } } unsigned long long mod_freesync_calc_nominal_field_rate( diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.c index c760216a6240..ca402ddcdacc 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.c @@ -354,7 +354,7 @@ enum mod_hdcp_status mod_hdcp_add_display(struct mod_hdcp *hdcp, /* reset retry counters */ reset_retry_counts(hdcp); - /* reset error trace */ + /* reset trace */ memset(&hdcp->connection.trace, 0, sizeof(hdcp->connection.trace)); /* add display to connection */ @@ -400,7 +400,7 @@ enum mod_hdcp_status mod_hdcp_remove_display(struct mod_hdcp *hdcp, /* clear retry counters */ reset_retry_counts(hdcp); - /* reset error trace */ + /* reset trace */ memset(&hdcp->connection.trace, 0, sizeof(hdcp->connection.trace)); /* remove display */ @@ -464,7 +464,7 @@ enum mod_hdcp_status mod_hdcp_update_display(struct mod_hdcp *hdcp, /* clear retry counters */ reset_retry_counts(hdcp); - /* reset error trace */ + /* reset trace */ memset(&hdcp->connection.trace, 0, sizeof(hdcp->connection.trace)); /* set new adjustment */ diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h index a37634942b07..26a351a184f3 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h @@ -88,6 +88,7 @@ struct mod_hdcp_transition_input_hdcp2 { uint8_t lc_init_write; uint8_t l_prime_available_poll; uint8_t l_prime_read; + uint8_t l_prime_combo_read; uint8_t l_prime_validation; uint8_t eks_prepare; uint8_t eks_write; @@ -508,7 +509,7 @@ static inline void set_auth_complete(struct mod_hdcp *hdcp, struct mod_hdcp_output *output) { output->auth_complete = 1; - mod_hdcp_log_ddc_trace(hdcp); + HDCP_AUTH_COMPLETE_TRACE(hdcp); } /* connection topology helpers */ diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c index 8bc377560787..1bbd728d4345 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp1_execution.c @@ -29,6 +29,7 @@ static inline enum mod_hdcp_status validate_bksv(struct mod_hdcp *hdcp) { uint64_t n = 0; uint8_t count = 0; + enum mod_hdcp_status status; u8 bksv[sizeof(n)] = { }; memcpy(bksv, hdcp->auth.msg.hdcp1.bksv, sizeof(hdcp->auth.msg.hdcp1.bksv)); @@ -38,8 +39,14 @@ static inline enum mod_hdcp_status validate_bksv(struct mod_hdcp *hdcp) count++; n &= (n - 1); } - return (count == 20) ? MOD_HDCP_STATUS_SUCCESS : - MOD_HDCP_STATUS_HDCP1_INVALID_BKSV; + + if (count == 20) { + hdcp->connection.trace.hdcp1.attempt_count++; + status = MOD_HDCP_STATUS_SUCCESS; + } else { + status = MOD_HDCP_STATUS_HDCP1_INVALID_BKSV; + } + return status; } static inline enum mod_hdcp_status check_ksv_ready(struct mod_hdcp *hdcp) @@ -135,6 +142,8 @@ static inline enum mod_hdcp_status check_device_count(struct mod_hdcp *hdcp) if (get_device_count(hdcp) == 0) return MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE; + hdcp->connection.trace.hdcp1.downstream_device_count = get_device_count(hdcp); + /* Some MST display may choose to report the internal panel as an HDCP RX. * To update this condition with 1(because the immediate repeater's internal * panel is possibly not included in DEVICE_COUNT) + get_device_count(hdcp). diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c index bb8ae80b37f8..27500abf9fee 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c @@ -48,6 +48,7 @@ static inline enum mod_hdcp_status check_receiver_id_list_ready(struct mod_hdcp static inline enum mod_hdcp_status check_hdcp2_capable(struct mod_hdcp *hdcp) { enum mod_hdcp_status status; + struct mod_hdcp_trace *trace = &hdcp->connection.trace; if (is_dp_hdcp(hdcp)) status = (hdcp->auth.msg.hdcp2.rxcaps_dp[0] == HDCP_2_2_RX_CAPS_VERSION_VAL) && @@ -55,9 +56,14 @@ static inline enum mod_hdcp_status check_hdcp2_capable(struct mod_hdcp *hdcp) MOD_HDCP_STATUS_SUCCESS : MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE; else - status = (hdcp->auth.msg.hdcp2.hdcp2version_hdmi & HDCP_2_2_HDMI_SUPPORT_MASK) ? - MOD_HDCP_STATUS_SUCCESS : - MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE; + status = (hdcp->auth.msg.hdcp2.hdcp2version_hdmi + & HDCP_2_2_HDMI_SUPPORT_MASK) + ? MOD_HDCP_STATUS_SUCCESS + : MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE; + + if (status == MOD_HDCP_STATUS_SUCCESS) + trace->hdcp2.attempt_count++; + return status; } @@ -201,10 +207,17 @@ static inline uint8_t get_device_count(struct mod_hdcp *hdcp) static enum mod_hdcp_status check_device_count(struct mod_hdcp *hdcp) { + struct mod_hdcp_trace *trace = &hdcp->connection.trace; + /* Avoid device count == 0 to do authentication */ if (get_device_count(hdcp) == 0) return MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE; + trace->hdcp2.downstream_device_count = get_device_count(hdcp); + trace->hdcp2.hdcp1_device_downstream = + HDCP_2_2_HDCP1_DEVICE_CONNECTED(hdcp->auth.msg.hdcp2.rx_id_list[2]); + trace->hdcp2.hdcp2_legacy_device_downstream = + HDCP_2_2_HDCP_2_0_REP_CONNECTED(hdcp->auth.msg.hdcp2.rx_id_list[2]); /* Some MST display may choose to report the internal panel as an HDCP RX. */ /* To update this condition with 1(because the immediate repeater's internal */ /* panel is possibly not included in DEVICE_COUNT) + get_device_count(hdcp). */ @@ -452,54 +465,11 @@ out: return status; } -static enum mod_hdcp_status locality_check_sw(struct mod_hdcp *hdcp, - struct mod_hdcp_event_context *event_ctx, - struct mod_hdcp_transition_input_hdcp2 *input) -{ - enum mod_hdcp_status status = MOD_HDCP_STATUS_SUCCESS; - - if (!mod_hdcp_execute_and_set(mod_hdcp_write_lc_init, - &input->lc_init_write, &status, - hdcp, "lc_init_write")) - goto out; - if (is_dp_hdcp(hdcp)) - msleep(16); - else - if (!mod_hdcp_execute_and_set(poll_l_prime_available, - &input->l_prime_available_poll, &status, - hdcp, "l_prime_available_poll")) - goto out; - if (!mod_hdcp_execute_and_set(mod_hdcp_read_l_prime, - &input->l_prime_read, &status, - hdcp, "l_prime_read")) - goto out; -out: - return status; -} - -static enum mod_hdcp_status locality_check_fw(struct mod_hdcp *hdcp, - struct mod_hdcp_event_context *event_ctx, - struct mod_hdcp_transition_input_hdcp2 *input) -{ - enum mod_hdcp_status status = MOD_HDCP_STATUS_SUCCESS; - - if (!mod_hdcp_execute_and_set(mod_hdcp_write_poll_read_lc_fw, - &input->l_prime_read, &status, - hdcp, "l_prime_read")) - goto out; - -out: - return status; -} - static enum mod_hdcp_status locality_check(struct mod_hdcp *hdcp, struct mod_hdcp_event_context *event_ctx, struct mod_hdcp_transition_input_hdcp2 *input) { enum mod_hdcp_status status = MOD_HDCP_STATUS_SUCCESS; - const bool use_fw = hdcp->config.ddc.funcs.atomic_write_poll_read_i2c - && hdcp->config.ddc.funcs.atomic_write_poll_read_aux - && !hdcp->connection.link.adjust.hdcp2.force_sw_locality_check; if (event_ctx->event != MOD_HDCP_EVENT_CALLBACK) { event_ctx->unexpected_event = 1; @@ -511,9 +481,28 @@ static enum mod_hdcp_status locality_check(struct mod_hdcp *hdcp, hdcp, "lc_init_prepare")) goto out; - status = (use_fw ? locality_check_fw : locality_check_sw)(hdcp, event_ctx, input); - if (status != MOD_HDCP_STATUS_SUCCESS) - goto out; + if (hdcp->connection.link.adjust.hdcp2.use_fw_locality_check) { + if (!mod_hdcp_execute_and_set(mod_hdcp_write_poll_read_lc_fw, + &input->l_prime_combo_read, &status, + hdcp, "l_prime_combo_read")) + goto out; + } else { + if (!mod_hdcp_execute_and_set(mod_hdcp_write_lc_init, + &input->lc_init_write, &status, + hdcp, "lc_init_write")) + goto out; + if (is_dp_hdcp(hdcp)) + msleep(16); + else + if (!mod_hdcp_execute_and_set(poll_l_prime_available, + &input->l_prime_available_poll, &status, + hdcp, "l_prime_available_poll")) + goto out; + if (!mod_hdcp_execute_and_set(mod_hdcp_read_l_prime, + &input->l_prime_read, &status, + hdcp, "l_prime_read")) + goto out; + } if (!mod_hdcp_execute_and_set(mod_hdcp_hdcp2_validate_l_prime, &input->l_prime_validation, &status, diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_transition.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_transition.c index 89ffb89e1932..9316312a4df5 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_transition.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_transition.c @@ -184,31 +184,33 @@ enum mod_hdcp_status mod_hdcp_hdcp2_transition(struct mod_hdcp *hdcp, callback_in_ms(0, output); set_state_id(hdcp, output, H2_A2_LOCALITY_CHECK); break; - case H2_A2_LOCALITY_CHECK: { - const bool use_fw = hdcp->config.ddc.funcs.atomic_write_poll_read_i2c - && !adjust->hdcp2.force_sw_locality_check; - - /* - * 1A-05: consider disconnection after LC init a failure - * 1A-13-1: consider invalid l' a failure - * 1A-13-2: consider l' timeout a failure - */ + case H2_A2_LOCALITY_CHECK: + /* 1A-05: consider disconnection after LC init a failure */ if (hdcp->state.stay_count > 10 || - input->lc_init_prepare != PASS || - (!use_fw && input->lc_init_write != PASS) || - (!use_fw && input->l_prime_available_poll != PASS)) { + input->lc_init_prepare != PASS) { fail_and_restart_in_ms(0, &status, output); break; - } else if (input->l_prime_read != PASS) { - if (use_fw && hdcp->config.debug.lc_enable_sw_fallback) { - adjust->hdcp2.force_sw_locality_check = true; + } else if (adjust->hdcp2.use_fw_locality_check && + input->l_prime_combo_read != PASS) { + /* 1A-13-2: consider l' timeout a failure */ + if (adjust->hdcp2.use_sw_locality_fallback) { + /* switch to software locality check */ + adjust->hdcp2.use_fw_locality_check = 0; callback_in_ms(0, output); + increment_stay_counter(hdcp); break; } - + fail_and_restart_in_ms(0, &status, output); + break; + } else if (!adjust->hdcp2.use_fw_locality_check && + (input->lc_init_write != PASS || + input->l_prime_available_poll != PASS || + input->l_prime_read != PASS)) { + /* 1A-13-2: consider l' timeout a failure */ fail_and_restart_in_ms(0, &status, output); break; } else if (input->l_prime_validation != PASS) { + /* 1A-13-1: consider invalid l' a failure */ callback_in_ms(0, output); increment_stay_counter(hdcp); break; @@ -216,7 +218,6 @@ enum mod_hdcp_status mod_hdcp_hdcp2_transition(struct mod_hdcp *hdcp, callback_in_ms(0, output); set_state_id(hdcp, output, H2_A3_EXCHANGE_KS_AND_TEST_FOR_REPEATER); break; - } case H2_A3_EXCHANGE_KS_AND_TEST_FOR_REPEATER: if (input->eks_prepare != PASS || input->eks_write != PASS) { @@ -510,26 +511,29 @@ enum mod_hdcp_status mod_hdcp_hdcp2_dp_transition(struct mod_hdcp *hdcp, callback_in_ms(0, output); set_state_id(hdcp, output, D2_A2_LOCALITY_CHECK); break; - case D2_A2_LOCALITY_CHECK: { - const bool use_fw = hdcp->config.ddc.funcs.atomic_write_poll_read_aux - && !adjust->hdcp2.force_sw_locality_check; - + case D2_A2_LOCALITY_CHECK: if (hdcp->state.stay_count > 10 || - input->lc_init_prepare != PASS || - (!use_fw && input->lc_init_write != PASS)) { - /* 1A-12: consider invalid l' a failure */ + input->lc_init_prepare != PASS) { fail_and_restart_in_ms(0, &status, output); break; - } else if (input->l_prime_read != PASS) { - if (use_fw && hdcp->config.debug.lc_enable_sw_fallback) { - adjust->hdcp2.force_sw_locality_check = true; + } else if (adjust->hdcp2.use_fw_locality_check && + input->l_prime_combo_read != PASS) { + if (adjust->hdcp2.use_sw_locality_fallback) { + /* switch to software locality check */ + adjust->hdcp2.use_fw_locality_check = 0; callback_in_ms(0, output); + increment_stay_counter(hdcp); break; } - + fail_and_restart_in_ms(0, &status, output); + break; + } else if (!adjust->hdcp2.use_fw_locality_check && + (input->lc_init_write != PASS || + input->l_prime_read != PASS)) { fail_and_restart_in_ms(0, &status, output); break; } else if (input->l_prime_validation != PASS) { + /* 1A-12: consider invalid l' a failure */ callback_in_ms(0, output); increment_stay_counter(hdcp); break; @@ -537,7 +541,6 @@ enum mod_hdcp_status mod_hdcp_hdcp2_dp_transition(struct mod_hdcp *hdcp, callback_in_ms(0, output); set_state_id(hdcp, output, D2_A34_EXCHANGE_KS_AND_TEST_FOR_REPEATER); break; - } case D2_A34_EXCHANGE_KS_AND_TEST_FOR_REPEATER: if (input->eks_prepare != PASS || input->eks_write != PASS) { diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_ddc.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_ddc.c index 2e6408579194..0ca39873f807 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_ddc.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_ddc.c @@ -758,6 +758,6 @@ enum mod_hdcp_status mod_hdcp_write_poll_read_lc_fw(struct mod_hdcp *hdcp) { const bool success = (is_dp_hdcp(hdcp) ? write_stall_read_lc_fw_aux : write_poll_read_lc_fw_i2c)(hdcp); - return success ? MOD_HDCP_STATUS_SUCCESS : MOD_HDCP_STATUS_DDC_FAILURE; + return success ? MOD_HDCP_STATUS_SUCCESS : MOD_HDCP_STATUS_HDCP2_LOCALITY_COMBO_READ_FAILURE; } diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c index 6b3b5f610907..5cb979c2cf8c 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c @@ -125,129 +125,11 @@ void mod_hdcp_log_ddc_trace(struct mod_hdcp *hdcp) } } +#define CASE_FORMAT(entry) case entry: return #entry; char *mod_hdcp_status_to_str(int32_t status) { switch (status) { - case MOD_HDCP_STATUS_SUCCESS: - return "MOD_HDCP_STATUS_SUCCESS"; - case MOD_HDCP_STATUS_FAILURE: - return "MOD_HDCP_STATUS_FAILURE"; - case MOD_HDCP_STATUS_RESET_NEEDED: - return "MOD_HDCP_STATUS_RESET_NEEDED"; - case MOD_HDCP_STATUS_DISPLAY_OUT_OF_BOUND: - return "MOD_HDCP_STATUS_DISPLAY_OUT_OF_BOUND"; - case MOD_HDCP_STATUS_DISPLAY_NOT_FOUND: - return "MOD_HDCP_STATUS_DISPLAY_NOT_FOUND"; - case MOD_HDCP_STATUS_INVALID_STATE: - return "MOD_HDCP_STATUS_INVALID_STATE"; - case MOD_HDCP_STATUS_NOT_IMPLEMENTED: - return "MOD_HDCP_STATUS_NOT_IMPLEMENTED"; - case MOD_HDCP_STATUS_INTERNAL_POLICY_FAILURE: - return "MOD_HDCP_STATUS_INTERNAL_POLICY_FAILURE"; - case MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE: - return "MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE"; - case MOD_HDCP_STATUS_CREATE_PSP_SERVICE_FAILURE: - return "MOD_HDCP_STATUS_CREATE_PSP_SERVICE_FAILURE"; - case MOD_HDCP_STATUS_DESTROY_PSP_SERVICE_FAILURE: - return "MOD_HDCP_STATUS_DESTROY_PSP_SERVICE_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_CREATE_SESSION_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_CREATE_SESSION_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_DESTROY_SESSION_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_DESTROY_SESSION_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_VALIDATE_ENCRYPTION_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_VALIDATE_ENCRYPTION_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_NOT_HDCP_REPEATER: - return "MOD_HDCP_STATUS_HDCP1_NOT_HDCP_REPEATER"; - case MOD_HDCP_STATUS_HDCP1_NOT_CAPABLE: - return "MOD_HDCP_STATUS_HDCP1_NOT_CAPABLE"; - case MOD_HDCP_STATUS_HDCP1_R0_PRIME_PENDING: - return "MOD_HDCP_STATUS_HDCP1_R0_PRIME_PENDING"; - case MOD_HDCP_STATUS_HDCP1_VALIDATE_RX_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_VALIDATE_RX_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_BKSV_REVOKED: - return "MOD_HDCP_STATUS_HDCP1_BKSV_REVOKED"; - case MOD_HDCP_STATUS_HDCP1_KSV_LIST_NOT_READY: - return "MOD_HDCP_STATUS_HDCP1_KSV_LIST_NOT_READY"; - case MOD_HDCP_STATUS_HDCP1_VALIDATE_KSV_LIST_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_VALIDATE_KSV_LIST_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_KSV_LIST_REVOKED: - return "MOD_HDCP_STATUS_HDCP1_KSV_LIST_REVOKED"; - case MOD_HDCP_STATUS_HDCP1_ENABLE_ENCRYPTION_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_ENABLE_ENCRYPTION_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_ENABLE_STREAM_ENCRYPTION_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_ENABLE_STREAM_ENCRYPTION_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_MAX_CASCADE_EXCEEDED_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_MAX_CASCADE_EXCEEDED_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_MAX_DEVS_EXCEEDED_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_MAX_DEVS_EXCEEDED_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_LINK_INTEGRITY_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_LINK_INTEGRITY_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_REAUTH_REQUEST_ISSUED: - return "MOD_HDCP_STATUS_HDCP1_REAUTH_REQUEST_ISSUED"; - case MOD_HDCP_STATUS_HDCP1_LINK_MAINTENANCE_FAILURE: - return "MOD_HDCP_STATUS_HDCP1_LINK_MAINTENANCE_FAILURE"; - case MOD_HDCP_STATUS_HDCP1_INVALID_BKSV: - return "MOD_HDCP_STATUS_HDCP1_INVALID_BKSV"; - case MOD_HDCP_STATUS_DDC_FAILURE: - return "MOD_HDCP_STATUS_DDC_FAILURE"; - case MOD_HDCP_STATUS_INVALID_OPERATION: - return "MOD_HDCP_STATUS_INVALID_OPERATION"; - case MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE: - return "MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE"; - case MOD_HDCP_STATUS_HDCP2_CREATE_SESSION_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_CREATE_SESSION_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_DESTROY_SESSION_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_DESTROY_SESSION_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_PREP_AKE_INIT_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_PREP_AKE_INIT_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_AKE_CERT_PENDING: - return "MOD_HDCP_STATUS_HDCP2_AKE_CERT_PENDING"; - case MOD_HDCP_STATUS_HDCP2_H_PRIME_PENDING: - return "MOD_HDCP_STATUS_HDCP2_H_PRIME_PENDING"; - case MOD_HDCP_STATUS_HDCP2_PAIRING_INFO_PENDING: - return "MOD_HDCP_STATUS_HDCP2_PAIRING_INFO_PENDING"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_AKE_CERT_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_AKE_CERT_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_AKE_CERT_REVOKED: - return "MOD_HDCP_STATUS_HDCP2_AKE_CERT_REVOKED"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_H_PRIME_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_H_PRIME_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_PAIRING_INFO_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_PAIRING_INFO_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_PREP_LC_INIT_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_PREP_LC_INIT_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_L_PRIME_PENDING: - return "MOD_HDCP_STATUS_HDCP2_L_PRIME_PENDING"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_L_PRIME_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_L_PRIME_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_PREP_EKS_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_PREP_EKS_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_ENABLE_ENCRYPTION_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_ENABLE_ENCRYPTION_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_RX_ID_LIST_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_RX_ID_LIST_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_REVOKED: - return "MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_REVOKED"; - case MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_NOT_READY: - return "MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_NOT_READY"; - case MOD_HDCP_STATUS_HDCP2_ENABLE_STREAM_ENCRYPTION_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_ENABLE_STREAM_ENCRYPTION_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_STREAM_READY_PENDING: - return "MOD_HDCP_STATUS_HDCP2_STREAM_READY_PENDING"; - case MOD_HDCP_STATUS_HDCP2_VALIDATE_STREAM_READY_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_VALIDATE_STREAM_READY_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_PREPARE_STREAM_MANAGEMENT_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_PREPARE_STREAM_MANAGEMENT_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_REAUTH_REQUEST: - return "MOD_HDCP_STATUS_HDCP2_REAUTH_REQUEST"; - case MOD_HDCP_STATUS_HDCP2_REAUTH_LINK_INTEGRITY_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_REAUTH_LINK_INTEGRITY_FAILURE"; - case MOD_HDCP_STATUS_HDCP2_DEVICE_COUNT_MISMATCH_FAILURE: - return "MOD_HDCP_STATUS_HDCP2_DEVICE_COUNT_MISMATCH_FAILURE"; - case MOD_HDCP_STATUS_UNSUPPORTED_PSP_VER_FAILURE: - return "MOD_HDCP_STATUS_UNSUPPORTED_PSP_VER_FAILURE"; + MOD_HDCP_STATUS_LIST(CASE_FORMAT) default: return "MOD_HDCP_STATUS_UNKNOWN"; } diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.h b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.h index 1d83c1b9da10..26553aa4c5ca 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.h +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.h @@ -31,6 +31,7 @@ #define HDCP_LOG_FSM(hdcp, ...) DRM_DEBUG_KMS(__VA_ARGS__) #define HDCP_LOG_TOP(hdcp, ...) pr_debug("[HDCP_TOP]:"__VA_ARGS__) #define HDCP_LOG_DDC(hdcp, ...) pr_debug("[HDCP_DDC]:"__VA_ARGS__) +#define HDCP_LOG_TRA(hdcp) do {} while (0) /* default logs */ #define HDCP_ERROR_TRACE(hdcp, status) \ @@ -131,4 +132,9 @@ HDCP_LOG_TOP(hdcp, "[Link %d] %s display %d", hdcp->config.index, __func__, i); \ } while (0) +#define HDCP_AUTH_COMPLETE_TRACE(hdcp) do { \ + mod_hdcp_log_ddc_trace(hdcp); \ + HDCP_LOG_TRA(hdcp); \ +} while (0) + #endif // MOD_HDCP_LOG_H_ diff --git a/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h b/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h index b51ddf2846df..835467225458 100644 --- a/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h +++ b/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h @@ -35,69 +35,74 @@ struct mod_hdcp; #define MAX_NUM_OF_DISPLAYS 6 #define MAX_NUM_OF_ATTEMPTS 4 #define MAX_NUM_OF_ERROR_TRACE 10 +#define MOD_HDCP_STATUS_LIST(FORMAT) \ + FORMAT(MOD_HDCP_STATUS_SUCCESS) \ + FORMAT(MOD_HDCP_STATUS_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_RESET_NEEDED) \ + FORMAT(MOD_HDCP_STATUS_DISPLAY_OUT_OF_BOUND) \ + FORMAT(MOD_HDCP_STATUS_DISPLAY_NOT_FOUND) \ + FORMAT(MOD_HDCP_STATUS_INVALID_STATE) \ + FORMAT(MOD_HDCP_STATUS_NOT_IMPLEMENTED) \ + FORMAT(MOD_HDCP_STATUS_INTERNAL_POLICY_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_CREATE_PSP_SERVICE_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_DESTROY_PSP_SERVICE_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_CREATE_SESSION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_DESTROY_SESSION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_VALIDATE_ENCRYPTION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_NOT_HDCP_REPEATER) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_NOT_CAPABLE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_R0_PRIME_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_VALIDATE_RX_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_BKSV_REVOKED) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_KSV_LIST_NOT_READY) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_VALIDATE_KSV_LIST_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_KSV_LIST_REVOKED) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_ENABLE_ENCRYPTION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_ENABLE_STREAM_ENCRYPTION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_MAX_CASCADE_EXCEEDED_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_MAX_DEVS_EXCEEDED_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_LINK_INTEGRITY_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_REAUTH_REQUEST_ISSUED) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_LINK_MAINTENANCE_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP1_INVALID_BKSV) \ + FORMAT(MOD_HDCP_STATUS_DDC_FAILURE) /* TODO: specific errors */ \ + FORMAT(MOD_HDCP_STATUS_INVALID_OPERATION) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_CREATE_SESSION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_DESTROY_SESSION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_PREP_AKE_INIT_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_AKE_CERT_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_H_PRIME_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_PAIRING_INFO_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_AKE_CERT_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_AKE_CERT_REVOKED) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_H_PRIME_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_PAIRING_INFO_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_PREP_LC_INIT_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_L_PRIME_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_L_PRIME_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_PREP_EKS_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_ENABLE_ENCRYPTION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_NOT_READY) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_RX_ID_LIST_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_REVOKED) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_ENABLE_STREAM_ENCRYPTION_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_STREAM_READY_PENDING) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_VALIDATE_STREAM_READY_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_PREPARE_STREAM_MANAGEMENT_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_REAUTH_REQUEST) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_REAUTH_LINK_INTEGRITY_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_DEVICE_COUNT_MISMATCH_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_UNSUPPORTED_PSP_VER_FAILURE) \ + FORMAT(MOD_HDCP_STATUS_HDCP2_LOCALITY_COMBO_READ_FAILURE) + +#define ENUM_FORMAT(entry) entry, /* detailed return status */ enum mod_hdcp_status { - MOD_HDCP_STATUS_SUCCESS = 0, - MOD_HDCP_STATUS_FAILURE, - MOD_HDCP_STATUS_RESET_NEEDED, - MOD_HDCP_STATUS_DISPLAY_OUT_OF_BOUND, - MOD_HDCP_STATUS_DISPLAY_NOT_FOUND, - MOD_HDCP_STATUS_INVALID_STATE, - MOD_HDCP_STATUS_NOT_IMPLEMENTED, - MOD_HDCP_STATUS_INTERNAL_POLICY_FAILURE, - MOD_HDCP_STATUS_UPDATE_TOPOLOGY_FAILURE, - MOD_HDCP_STATUS_CREATE_PSP_SERVICE_FAILURE, - MOD_HDCP_STATUS_DESTROY_PSP_SERVICE_FAILURE, - MOD_HDCP_STATUS_HDCP1_CREATE_SESSION_FAILURE, - MOD_HDCP_STATUS_HDCP1_DESTROY_SESSION_FAILURE, - MOD_HDCP_STATUS_HDCP1_VALIDATE_ENCRYPTION_FAILURE, - MOD_HDCP_STATUS_HDCP1_NOT_HDCP_REPEATER, - MOD_HDCP_STATUS_HDCP1_NOT_CAPABLE, - MOD_HDCP_STATUS_HDCP1_R0_PRIME_PENDING, - MOD_HDCP_STATUS_HDCP1_VALIDATE_RX_FAILURE, - MOD_HDCP_STATUS_HDCP1_BKSV_REVOKED, - MOD_HDCP_STATUS_HDCP1_KSV_LIST_NOT_READY, - MOD_HDCP_STATUS_HDCP1_VALIDATE_KSV_LIST_FAILURE, - MOD_HDCP_STATUS_HDCP1_KSV_LIST_REVOKED, - MOD_HDCP_STATUS_HDCP1_ENABLE_ENCRYPTION_FAILURE, - MOD_HDCP_STATUS_HDCP1_ENABLE_STREAM_ENCRYPTION_FAILURE, - MOD_HDCP_STATUS_HDCP1_MAX_CASCADE_EXCEEDED_FAILURE, - MOD_HDCP_STATUS_HDCP1_MAX_DEVS_EXCEEDED_FAILURE, - MOD_HDCP_STATUS_HDCP1_DEVICE_COUNT_MISMATCH_FAILURE, - MOD_HDCP_STATUS_HDCP1_LINK_INTEGRITY_FAILURE, - MOD_HDCP_STATUS_HDCP1_REAUTH_REQUEST_ISSUED, - MOD_HDCP_STATUS_HDCP1_LINK_MAINTENANCE_FAILURE, - MOD_HDCP_STATUS_HDCP1_INVALID_BKSV, - MOD_HDCP_STATUS_DDC_FAILURE, /* TODO: specific errors */ - MOD_HDCP_STATUS_INVALID_OPERATION, - MOD_HDCP_STATUS_HDCP2_NOT_CAPABLE, - MOD_HDCP_STATUS_HDCP2_CREATE_SESSION_FAILURE, - MOD_HDCP_STATUS_HDCP2_DESTROY_SESSION_FAILURE, - MOD_HDCP_STATUS_HDCP2_PREP_AKE_INIT_FAILURE, - MOD_HDCP_STATUS_HDCP2_AKE_CERT_PENDING, - MOD_HDCP_STATUS_HDCP2_H_PRIME_PENDING, - MOD_HDCP_STATUS_HDCP2_PAIRING_INFO_PENDING, - MOD_HDCP_STATUS_HDCP2_VALIDATE_AKE_CERT_FAILURE, - MOD_HDCP_STATUS_HDCP2_AKE_CERT_REVOKED, - MOD_HDCP_STATUS_HDCP2_VALIDATE_H_PRIME_FAILURE, - MOD_HDCP_STATUS_HDCP2_VALIDATE_PAIRING_INFO_FAILURE, - MOD_HDCP_STATUS_HDCP2_PREP_LC_INIT_FAILURE, - MOD_HDCP_STATUS_HDCP2_L_PRIME_PENDING, - MOD_HDCP_STATUS_HDCP2_VALIDATE_L_PRIME_FAILURE, - MOD_HDCP_STATUS_HDCP2_PREP_EKS_FAILURE, - MOD_HDCP_STATUS_HDCP2_ENABLE_ENCRYPTION_FAILURE, - MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_NOT_READY, - MOD_HDCP_STATUS_HDCP2_VALIDATE_RX_ID_LIST_FAILURE, - MOD_HDCP_STATUS_HDCP2_RX_ID_LIST_REVOKED, - MOD_HDCP_STATUS_HDCP2_ENABLE_STREAM_ENCRYPTION_FAILURE, - MOD_HDCP_STATUS_HDCP2_STREAM_READY_PENDING, - MOD_HDCP_STATUS_HDCP2_VALIDATE_STREAM_READY_FAILURE, - MOD_HDCP_STATUS_HDCP2_PREPARE_STREAM_MANAGEMENT_FAILURE, - MOD_HDCP_STATUS_HDCP2_REAUTH_REQUEST, - MOD_HDCP_STATUS_HDCP2_REAUTH_LINK_INTEGRITY_FAILURE, - MOD_HDCP_STATUS_HDCP2_DEVICE_COUNT_MISMATCH_FAILURE, - MOD_HDCP_STATUS_UNSUPPORTED_PSP_VER_FAILURE, + MOD_HDCP_STATUS_LIST(ENUM_FORMAT) }; struct mod_hdcp_displayport { @@ -214,8 +219,9 @@ struct mod_hdcp_link_adjustment_hdcp2 { uint8_t force_type : 2; uint8_t force_no_stored_km : 1; uint8_t increase_h_prime_timeout: 1; - uint8_t force_sw_locality_check : 1; - uint8_t reserved : 2; + uint8_t use_fw_locality_check : 1; + uint8_t use_sw_locality_fallback: 1; + uint8_t reserved : 1; }; struct mod_hdcp_link_adjustment { @@ -230,9 +236,23 @@ struct mod_hdcp_error { uint8_t state_id; }; +struct mod_hdcp1_trace { + uint8_t attempt_count; + uint8_t downstream_device_count; +}; + +struct mod_hdcp2_trace { + uint8_t attempt_count; + uint8_t downstream_device_count; + uint8_t hdcp1_device_downstream; + uint8_t hdcp2_legacy_device_downstream; +}; + struct mod_hdcp_trace { struct mod_hdcp_error errors[MAX_NUM_OF_ERROR_TRACE]; uint8_t error_count; + struct mod_hdcp1_trace hdcp1; + struct mod_hdcp2_trace hdcp2; }; enum mod_hdcp_encryption_status { @@ -303,10 +323,6 @@ struct mod_hdcp_display_query { struct mod_hdcp_config { struct mod_hdcp_psp psp; struct mod_hdcp_ddc ddc; - struct { - uint8_t lc_enable_sw_fallback : 1; - uint8_t reserved : 7; - } debug; uint8_t index; }; diff --git a/drivers/gpu/drm/amd/display/modules/power/power_helpers.c b/drivers/gpu/drm/amd/display/modules/power/power_helpers.c index 29ccd3532d13..fd139b219bf9 100644 --- a/drivers/gpu/drm/amd/display/modules/power/power_helpers.c +++ b/drivers/gpu/drm/amd/display/modules/power/power_helpers.c @@ -975,6 +975,34 @@ bool psr_su_set_dsc_slice_height(struct dc *dc, struct dc_link *link, return true; } +void set_replay_frame_skip_number(struct dc_link *link, + enum replay_coasting_vtotal_type type, + uint32_t coasting_vtotal_refresh_rate_mhz, + uint32_t flicker_free_refresh_rate_mhz, + bool is_defer) +{ + uint32_t *frame_skip_number_array = NULL; + uint32_t frame_skip_number = 0; + + if (link == NULL || flicker_free_refresh_rate_mhz == 0 || coasting_vtotal_refresh_rate_mhz == 0) + return; + + if (is_defer) + frame_skip_number_array = link->replay_settings.defer_frame_skip_number_table; + else + frame_skip_number_array = link->replay_settings.frame_skip_number_table; + + if (frame_skip_number_array == NULL) + return; + + frame_skip_number = coasting_vtotal_refresh_rate_mhz / flicker_free_refresh_rate_mhz; + + if (frame_skip_number >= 1) + frame_skip_number_array[type] = frame_skip_number - 1; + else + frame_skip_number_array[type] = 0; +} + void set_replay_defer_update_coasting_vtotal(struct dc_link *link, enum replay_coasting_vtotal_type type, uint32_t vtotal) @@ -987,6 +1015,8 @@ void update_replay_coasting_vtotal_from_defer(struct dc_link *link, { link->replay_settings.coasting_vtotal_table[type] = link->replay_settings.defer_update_coasting_vtotal_table[type]; + link->replay_settings.frame_skip_number_table[type] = + link->replay_settings.defer_frame_skip_number_table[type]; } void set_replay_coasting_vtotal(struct dc_link *link, @@ -1007,6 +1037,9 @@ void calculate_replay_link_off_frame_count(struct dc_link *link, uint8_t max_link_off_frame_count = 0; uint16_t max_deviation_line = 0, pixel_deviation_per_line = 0; + if (!link || link->replay_settings.config.replay_version != DC_FREESYNC_REPLAY) + return; + max_deviation_line = link->dpcd_caps.pr_info.max_deviation_line; pixel_deviation_per_line = link->dpcd_caps.pr_info.pixel_deviation_per_line; diff --git a/drivers/gpu/drm/amd/display/modules/power/power_helpers.h b/drivers/gpu/drm/amd/display/modules/power/power_helpers.h index 391209a3bf29..87d31d9dce5a 100644 --- a/drivers/gpu/drm/amd/display/modules/power/power_helpers.h +++ b/drivers/gpu/drm/amd/display/modules/power/power_helpers.h @@ -60,6 +60,11 @@ void set_replay_coasting_vtotal(struct dc_link *link, void set_replay_defer_update_coasting_vtotal(struct dc_link *link, enum replay_coasting_vtotal_type type, uint32_t vtotal); +void set_replay_frame_skip_number(struct dc_link *link, + enum replay_coasting_vtotal_type type, + uint32_t coasting_vtotal_refresh_rate_Mhz, + uint32_t flicker_free_refresh_rate_Mhz, + bool is_defer); void update_replay_coasting_vtotal_from_defer(struct dc_link *link, enum replay_coasting_vtotal_type type); void set_replay_low_rr_full_screen_video_src_vtotal(struct dc_link *link, uint16_t vtotal); diff --git a/drivers/gpu/drm/amd/include/amd_cper.h b/drivers/gpu/drm/amd/include/amd_cper.h index 086869264425..a252ee4c7874 100644 --- a/drivers/gpu/drm/amd/include/amd_cper.h +++ b/drivers/gpu/drm/amd/include/amd_cper.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2025 Advanced Micro Devices, Inc. * diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h index 75efda2969cf..17945094a138 100644 --- a/drivers/gpu/drm/amd/include/amd_shared.h +++ b/drivers/gpu/drm/amd/include/amd_shared.h @@ -109,6 +109,7 @@ enum amd_ip_block_type { AMD_IP_BLOCK_TYPE_VPE, AMD_IP_BLOCK_TYPE_UMSCH_MM, AMD_IP_BLOCK_TYPE_ISP, + AMD_IP_BLOCK_TYPE_RAS, AMD_IP_BLOCK_TYPE_NUM, }; diff --git a/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_d.h b/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_d.h index 2176548e9203..9778822dd2a0 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_d.h +++ b/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_d.h @@ -60,5 +60,10 @@ #define mmVCE_VCPU_CACHE_SIZE1 0x800C #define mmVCE_VCPU_CACHE_SIZE2 0x800E #define mmVCE_VCPU_CNTL 0x8005 +#define mmVCE_VCPU_SCRATCH7 0x8037 +#define mmVCE_FW_REG_STATUS 0x8384 +#define mmVCE_LMI_FW_PERIODIC_CTRL 0x8388 +#define mmVCE_LMI_FW_START_KEYSEL 0x8386 + #endif diff --git a/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_sh_mask.h index ea5b26b11cb1..1f82d6f5abde 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/vce/vce_1_0_sh_mask.h @@ -61,6 +61,8 @@ #define VCE_RB_WPTR__RB_WPTR__SHIFT 0x00000004 #define VCE_SOFT_RESET__ECPU_SOFT_RESET_MASK 0x00000001L #define VCE_SOFT_RESET__ECPU_SOFT_RESET__SHIFT 0x00000000 +#define VCE_SOFT_RESET__FME_SOFT_RESET_MASK 0x00000004L +#define VCE_SOFT_RESET__FME_SOFT_RESET__SHIFT 0x00000002 #define VCE_STATUS__JOB_BUSY_MASK 0x00000001L #define VCE_STATUS__JOB_BUSY__SHIFT 0x00000000 #define VCE_STATUS__UENC_BUSY_MASK 0x00000100L @@ -95,5 +97,13 @@ #define VCE_VCPU_CNTL__CLK_EN__SHIFT 0x00000000 #define VCE_VCPU_CNTL__RBBM_SOFT_RESET_MASK 0x00040000L #define VCE_VCPU_CNTL__RBBM_SOFT_RESET__SHIFT 0x00000012 +#define VCE_CLOCK_GATING_A__CGC_DYN_CLOCK_MODE_MASK 0x00010000 +#define VCE_CLOCK_GATING_A__CGC_DYN_CLOCK_MODE_SHIFT 0x00000010 +#define VCE_FW_REG_STATUS__BUSY_MASK 0x0000001 +#define VCE_FW_REG_STATUS__BUSY__SHIFT 0x0000001 +#define VCE_FW_REG_STATUS__PASS_MASK 0x0000008 +#define VCE_FW_REG_STATUS__PASS__SHIFT 0x0000003 +#define VCE_FW_REG_STATUS__DONE_MASK 0x0000800 +#define VCE_FW_REG_STATUS__DONE__SHIFT 0x000000b #endif diff --git a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h index 64b553e7de1a..e7fdcee22a71 100644 --- a/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h +++ b/drivers/gpu/drm/amd/include/ivsrcid/vcn/irqsrcs_vcn_5_0.h @@ -1,4 +1,4 @@ -/* SPDX-License-Identifier: GPL-2.0 */ +/* SPDX-License-Identifier: MIT */ /* * Copyright 2024 Advanced Micro Devices, Inc. All rights reserved. diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index 2b0cdb2a2775..2366e68262e6 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -454,7 +454,7 @@ struct amd_pm_funcs { bool gate, int inst); int (*set_clockgating_by_smu)(void *handle, uint32_t msg_id); - int (*set_power_limit)(void *handle, uint32_t n); + int (*set_power_limit)(void *handle, uint32_t limit_type, uint32_t n); int (*get_power_limit)(void *handle, uint32_t *limit, enum pp_power_limit_level pp_limit_level, enum pp_power_type power_type); @@ -532,6 +532,110 @@ struct metrics_table_header { uint8_t content_revision; }; +enum amdgpu_metrics_attr_id { + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_HOTSPOT, + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_MEM, + AMDGPU_METRICS_ATTR_ID_TEMPERATURE_VRSOC, + AMDGPU_METRICS_ATTR_ID_CURR_SOCKET_POWER, + AMDGPU_METRICS_ATTR_ID_AVERAGE_GFX_ACTIVITY, + AMDGPU_METRICS_ATTR_ID_AVERAGE_UMC_ACTIVITY, + AMDGPU_METRICS_ATTR_ID_MEM_MAX_BANDWIDTH, + AMDGPU_METRICS_ATTR_ID_ENERGY_ACCUMULATOR, + AMDGPU_METRICS_ATTR_ID_SYSTEM_CLOCK_COUNTER, + AMDGPU_METRICS_ATTR_ID_ACCUMULATION_COUNTER, + AMDGPU_METRICS_ATTR_ID_PROCHOT_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_PPT_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_SOCKET_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_VR_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_HBM_THM_RESIDENCY_ACC, + AMDGPU_METRICS_ATTR_ID_GFXCLK_LOCK_STATUS, + AMDGPU_METRICS_ATTR_ID_PCIE_LINK_WIDTH, + AMDGPU_METRICS_ATTR_ID_PCIE_LINK_SPEED, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_WIDTH, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_SPEED, + AMDGPU_METRICS_ATTR_ID_GFX_ACTIVITY_ACC, + AMDGPU_METRICS_ATTR_ID_MEM_ACTIVITY_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_BANDWIDTH_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_BANDWIDTH_INST, + AMDGPU_METRICS_ATTR_ID_PCIE_L0_TO_RECOV_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_REPLAY_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_REPLAY_ROVER_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_NAK_SENT_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_PCIE_NAK_RCVD_COUNT_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_READ_DATA_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_WRITE_DATA_ACC, + AMDGPU_METRICS_ATTR_ID_XGMI_LINK_STATUS, + AMDGPU_METRICS_ATTR_ID_FIRMWARE_TIMESTAMP, + AMDGPU_METRICS_ATTR_ID_CURRENT_GFXCLK, + AMDGPU_METRICS_ATTR_ID_CURRENT_SOCCLK, + AMDGPU_METRICS_ATTR_ID_CURRENT_VCLK0, + AMDGPU_METRICS_ATTR_ID_CURRENT_DCLK0, + AMDGPU_METRICS_ATTR_ID_CURRENT_UCLK, + AMDGPU_METRICS_ATTR_ID_NUM_PARTITION, + AMDGPU_METRICS_ATTR_ID_PCIE_LC_PERF_OTHER_END_RECOVERY, + AMDGPU_METRICS_ATTR_ID_GFX_BUSY_INST, + AMDGPU_METRICS_ATTR_ID_JPEG_BUSY, + AMDGPU_METRICS_ATTR_ID_VCN_BUSY, + AMDGPU_METRICS_ATTR_ID_GFX_BUSY_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_PPT_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_THM_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_LOW_UTILIZATION_ACC, + AMDGPU_METRICS_ATTR_ID_GFX_BELOW_HOST_LIMIT_TOTAL_ACC, + AMDGPU_METRICS_ATTR_ID_MAX, +}; + +enum amdgpu_metrics_attr_type { + AMDGPU_METRICS_TYPE_U8, + AMDGPU_METRICS_TYPE_S8, + AMDGPU_METRICS_TYPE_U16, + AMDGPU_METRICS_TYPE_S16, + AMDGPU_METRICS_TYPE_U32, + AMDGPU_METRICS_TYPE_S32, + AMDGPU_METRICS_TYPE_U64, + AMDGPU_METRICS_TYPE_S64, + AMDGPU_METRICS_TYPE_MAX, +}; + +enum amdgpu_metrics_attr_unit { + /* None */ + AMDGPU_METRICS_UNIT_NONE, + /* MHz*/ + AMDGPU_METRICS_UNIT_CLOCK_1, + /* Degree Celsius*/ + AMDGPU_METRICS_UNIT_TEMP_1, + /* Watts*/ + AMDGPU_METRICS_UNIT_POWER_1, + /* In nanoseconds*/ + AMDGPU_METRICS_UNIT_TIME_1, + /* In 10 nanoseconds*/ + AMDGPU_METRICS_UNIT_TIME_2, + /* Speed in GT/s */ + AMDGPU_METRICS_UNIT_SPEED_1, + /* Speed in 0.1 GT/s */ + AMDGPU_METRICS_UNIT_SPEED_2, + /* Bandwidth GB/s */ + AMDGPU_METRICS_UNIT_BW_1, + /* Data in KB */ + AMDGPU_METRICS_UNIT_DATA_1, + /* Percentage */ + AMDGPU_METRICS_UNIT_PERCENT, + AMDGPU_METRICS_UNIT_MAX, +}; + +#define AMDGPU_METRICS_ATTR_UNIT_MASK 0xFF000000 +#define AMDGPU_METRICS_ATTR_UNIT_SHIFT 24 +#define AMDGPU_METRICS_ATTR_TYPE_MASK 0x00F00000 +#define AMDGPU_METRICS_ATTR_TYPE_SHIFT 20 +#define AMDGPU_METRICS_ATTR_ID_MASK 0x000FFC00 +#define AMDGPU_METRICS_ATTR_ID_SHIFT 10 +#define AMDGPU_METRICS_ATTR_INST_MASK 0x000003FF +#define AMDGPU_METRICS_ATTR_INST_SHIFT 0 + +#define AMDGPU_METRICS_ENC_ATTR(unit, type, id, inst) \ + (((u64)(unit) << AMDGPU_METRICS_ATTR_UNIT_SHIFT) | \ + ((u64)(type) << AMDGPU_METRICS_ATTR_TYPE_SHIFT) | \ + ((u64)(id) << AMDGPU_METRICS_ATTR_ID_SHIFT) | (inst)) + /* * gpu_metrics_v1_0 is not recommended as it's not naturally aligned. * Use gpu_metrics_v1_1 or later instead. @@ -1221,6 +1325,19 @@ struct gpu_metrics_v1_8 { uint32_t pcie_lc_perf_other_end_recovery; }; +struct gpu_metrics_attr { + /* Field type encoded with AMDGPU_METRICS_ENC_ATTR */ + uint64_t attr_encoding; + /* Attribute value, depends on attr_encoding */ + void *attr_value; +}; + +struct gpu_metrics_v1_9 { + struct metrics_table_header common_header; + int attr_count; + struct gpu_metrics_attr metrics_attrs[]; +}; + /* * gpu_metrics_v2_0 is not recommended as it's not naturally aligned. * Use gpu_metrics_v2_1 or later instead. @@ -1703,4 +1820,10 @@ struct amdgpu_partition_metrics_v1_0 { uint64_t gfx_below_host_limit_total_acc[MAX_XCC]; }; +struct amdgpu_partition_metrics_v1_1 { + struct metrics_table_header common_header; + int attr_count; + struct gpu_metrics_attr metrics_attrs[]; +}; + #endif diff --git a/drivers/gpu/drm/amd/include/mes_v11_api_def.h b/drivers/gpu/drm/amd/include/mes_v11_api_def.h index ab1cfc92dbeb..f9629d42ada2 100644 --- a/drivers/gpu/drm/amd/include/mes_v11_api_def.h +++ b/drivers/gpu/drm/amd/include/mes_v11_api_def.h @@ -345,7 +345,8 @@ union MESAPI__REMOVE_QUEUE { uint32_t unmap_kiq_utility_queue : 1; uint32_t preempt_legacy_gfx_queue : 1; uint32_t unmap_legacy_queue : 1; - uint32_t reserved : 28; + uint32_t remove_queue_after_reset : 1; + uint32_t reserved : 27; }; struct MES_API_STATUS api_status; diff --git a/drivers/gpu/drm/amd/include/mes_v12_api_def.h b/drivers/gpu/drm/amd/include/mes_v12_api_def.h index 69611c7e30e3..2f12cba4eb66 100644 --- a/drivers/gpu/drm/amd/include/mes_v12_api_def.h +++ b/drivers/gpu/drm/amd/include/mes_v12_api_def.h @@ -399,7 +399,8 @@ union MESAPI__REMOVE_QUEUE { uint32_t unmap_kiq_utility_queue : 1; uint32_t preempt_legacy_gfx_queue : 1; uint32_t unmap_legacy_queue : 1; - uint32_t reserved : 28; + uint32_t remove_queue_after_reset : 1; + uint32_t reserved : 27; }; struct MES_API_STATUS api_status; diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c index 518d07afc7df..79b174e5326d 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c @@ -195,24 +195,6 @@ int amdgpu_dpm_set_mp1_state(struct amdgpu_device *adev, return ret; } -int amdgpu_dpm_notify_rlc_state(struct amdgpu_device *adev, bool en) -{ - int ret = 0; - const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; - - if (pp_funcs && pp_funcs->notify_rlc_state) { - mutex_lock(&adev->pm.mutex); - - ret = pp_funcs->notify_rlc_state( - adev->powerplay.pp_handle, - en); - - mutex_unlock(&adev->pm.mutex); - } - - return ret; -} - int amdgpu_dpm_is_baco_supported(struct amdgpu_device *adev) { const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; @@ -1205,8 +1187,11 @@ int amdgpu_dpm_get_pp_table(struct amdgpu_device *adev, char **table) const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; int ret = 0; - if (!pp_funcs->get_pp_table) - return 0; + if (!table) + return -EINVAL; + + if (amdgpu_sriov_vf(adev) || !pp_funcs->get_pp_table || adev->scpm_enabled) + return -EOPNOTSUPP; mutex_lock(&adev->pm.mutex); ret = pp_funcs->get_pp_table(adev->powerplay.pp_handle, @@ -1616,6 +1601,7 @@ int amdgpu_dpm_get_power_limit(struct amdgpu_device *adev, } int amdgpu_dpm_set_power_limit(struct amdgpu_device *adev, + uint32_t limit_type, uint32_t limit) { const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; @@ -1626,7 +1612,7 @@ int amdgpu_dpm_set_power_limit(struct amdgpu_device *adev, mutex_lock(&adev->pm.mutex); ret = pp_funcs->set_power_limit(adev->powerplay.pp_handle, - limit); + limit_type, limit); mutex_unlock(&adev->pm.mutex); return ret; @@ -1732,7 +1718,10 @@ int amdgpu_dpm_set_pp_table(struct amdgpu_device *adev, const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; int ret = 0; - if (!pp_funcs->set_pp_table) + if (!buf || !size) + return -EINVAL; + + if (amdgpu_sriov_vf(adev) || !pp_funcs->set_pp_table || adev->scpm_enabled) return -EOPNOTSUPP; mutex_lock(&adev->pm.mutex); @@ -2139,3 +2128,10 @@ ssize_t amdgpu_dpm_get_xcp_metrics(struct amdgpu_device *adev, int xcp_id, return ret; } + +const struct ras_smu_drv *amdgpu_dpm_get_ras_smu_driver(struct amdgpu_device *adev) +{ + void *pp_handle = adev->powerplay.pp_handle; + + return smu_get_ras_smu_driver(pp_handle); +} diff --git a/drivers/gpu/drm/amd/pm/amdgpu_pm.c b/drivers/gpu/drm/amd/pm/amdgpu_pm.c index b5fbb0fd1dc0..65296a819e6a 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c @@ -108,8 +108,9 @@ const char * const amdgpu_pp_profile_name[] = { static int amdgpu_pm_dev_state_check(struct amdgpu_device *adev, bool runpm) { bool runpm_check = runpm ? adev->in_runpm : false; + bool full_init = (adev->init_lvl->level == AMDGPU_INIT_LEVEL_DEFAULT); - if (amdgpu_in_reset(adev)) + if (amdgpu_in_reset(adev) || !full_init) return -EBUSY; if (adev->in_suspend && !runpm_check) @@ -173,7 +174,6 @@ static int amdgpu_pm_get_access_if_active(struct amdgpu_device *adev) */ static inline void amdgpu_pm_put_access(struct amdgpu_device *adev) { - pm_runtime_mark_last_busy(adev->dev); pm_runtime_put_autosuspend(adev->dev); } @@ -2506,7 +2506,7 @@ static struct amdgpu_device_attr amdgpu_device_attrs[] = { AMDGPU_DEVICE_ATTR_RO(pp_num_states, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF), AMDGPU_DEVICE_ATTR_RO(pp_cur_state, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF), AMDGPU_DEVICE_ATTR_RW(pp_force_state, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF), - AMDGPU_DEVICE_ATTR_RW(pp_table, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF), + AMDGPU_DEVICE_ATTR_RW(pp_table, ATTR_FLAG_BASIC), AMDGPU_DEVICE_ATTR_RW(pp_dpm_sclk, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF, .attr_update = pp_dpm_clk_default_attr_update), AMDGPU_DEVICE_ATTR_RW(pp_dpm_mclk, ATTR_FLAG_BASIC|ATTR_FLAG_ONEVF, @@ -2638,6 +2638,15 @@ static int default_attr_update(struct amdgpu_device *adev, struct amdgpu_device_ if (amdgpu_dpm_get_apu_thermal_limit(adev, &limit) == -EOPNOTSUPP) *states = ATTR_STATE_UNSUPPORTED; + } else if (DEVICE_ATTR_IS(pp_table)) { + int ret; + char *tmp = NULL; + + ret = amdgpu_dpm_get_pp_table(adev, &tmp); + if (ret == -EOPNOTSUPP || !tmp) + *states = ATTR_STATE_UNSUPPORTED; + else + *states = ATTR_STATE_SUPPORTED; } switch (gc_ver) { @@ -3372,7 +3381,9 @@ static ssize_t amdgpu_hwmon_show_power_label(struct device *dev, to_sensor_dev_attr(attr)->index == PP_PWR_TYPE_FAST ? "fastPPT" : "slowPPT"); else - return sysfs_emit(buf, "PPT\n"); + return sysfs_emit(buf, "%s\n", + to_sensor_dev_attr(attr)->index == PP_PWR_TYPE_FAST ? + "PPT1" : "PPT"); } static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev, @@ -3390,13 +3401,12 @@ static ssize_t amdgpu_hwmon_set_power_cap(struct device *dev, return err; value = value / 1000000; /* convert to Watt */ - value |= limit_type << 24; err = amdgpu_pm_get_access(adev); if (err < 0) return err; - err = amdgpu_dpm_set_power_limit(adev, value); + err = amdgpu_dpm_set_power_limit(adev, limit_type, value); amdgpu_pm_put_access(adev); @@ -3578,7 +3588,6 @@ static SENSOR_DEVICE_ATTR(power1_cap_min, S_IRUGO, amdgpu_hwmon_show_power_cap_m static SENSOR_DEVICE_ATTR(power1_cap, S_IRUGO | S_IWUSR, amdgpu_hwmon_show_power_cap, amdgpu_hwmon_set_power_cap, 0); static SENSOR_DEVICE_ATTR(power1_cap_default, S_IRUGO, amdgpu_hwmon_show_power_cap_default, NULL, 0); static SENSOR_DEVICE_ATTR(power1_label, S_IRUGO, amdgpu_hwmon_show_power_label, NULL, 0); -static SENSOR_DEVICE_ATTR(power2_average, S_IRUGO, amdgpu_hwmon_show_power_avg, NULL, 1); static SENSOR_DEVICE_ATTR(power2_cap_max, S_IRUGO, amdgpu_hwmon_show_power_cap_max, NULL, 1); static SENSOR_DEVICE_ATTR(power2_cap_min, S_IRUGO, amdgpu_hwmon_show_power_cap_min, NULL, 1); static SENSOR_DEVICE_ATTR(power2_cap, S_IRUGO | S_IWUSR, amdgpu_hwmon_show_power_cap, amdgpu_hwmon_set_power_cap, 1); @@ -3627,7 +3636,6 @@ static struct attribute *hwmon_attributes[] = { &sensor_dev_attr_power1_cap.dev_attr.attr, &sensor_dev_attr_power1_cap_default.dev_attr.attr, &sensor_dev_attr_power1_label.dev_attr.attr, - &sensor_dev_attr_power2_average.dev_attr.attr, &sensor_dev_attr_power2_cap_max.dev_attr.attr, &sensor_dev_attr_power2_cap_min.dev_attr.attr, &sensor_dev_attr_power2_cap.dev_attr.attr, @@ -3826,13 +3834,14 @@ static umode_t hwmon_attributes_visible(struct kobject *kobj, return 0; /* only Vangogh has fast PPT limit and power labels */ - if (!(gc_ver == IP_VERSION(10, 3, 1)) && - (attr == &sensor_dev_attr_power2_average.dev_attr.attr || - attr == &sensor_dev_attr_power2_cap_max.dev_attr.attr || + if ((attr == &sensor_dev_attr_power2_cap_max.dev_attr.attr || attr == &sensor_dev_attr_power2_cap_min.dev_attr.attr || attr == &sensor_dev_attr_power2_cap.dev_attr.attr || attr == &sensor_dev_attr_power2_cap_default.dev_attr.attr || - attr == &sensor_dev_attr_power2_label.dev_attr.attr)) + attr == &sensor_dev_attr_power2_label.dev_attr.attr) && + (amdgpu_dpm_get_power_limit(adev, &tmp, + PP_PWR_LIMIT_MAX, + PP_PWR_TYPE_FAST) == -EOPNOTSUPP)) return 0; return effective_mode; @@ -4724,14 +4733,14 @@ int amdgpu_pm_sysfs_init(struct amdgpu_device *adev) ret = devm_device_add_group(adev->dev, &amdgpu_pm_policy_attr_group); if (ret) - goto err_out0; + goto err_out1; } if (amdgpu_dpm_is_temp_metrics_supported(adev, SMU_TEMP_METRIC_GPUBOARD)) { ret = devm_device_add_group(adev->dev, &amdgpu_board_attr_group); if (ret) - goto err_out0; + goto err_out1; if (amdgpu_pm_get_sensor_generic(adev, AMDGPU_PP_SENSOR_MAXNODEPOWERLIMIT, (void *)&tmp) != -EOPNOTSUPP) { sysfs_add_file_to_group(&adev->dev->kobj, diff --git a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h index 65c1d98af26c..aa3f427819a0 100644 --- a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h +++ b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h @@ -424,8 +424,6 @@ int amdgpu_dpm_mode1_reset(struct amdgpu_device *adev); int amdgpu_dpm_set_mp1_state(struct amdgpu_device *adev, enum pp_mp1_state mp1_state); -int amdgpu_dpm_notify_rlc_state(struct amdgpu_device *adev, bool en); - int amdgpu_dpm_set_gfx_power_up_by_imu(struct amdgpu_device *adev); int amdgpu_dpm_baco_exit(struct amdgpu_device *adev); @@ -553,7 +551,7 @@ int amdgpu_dpm_get_power_limit(struct amdgpu_device *adev, enum pp_power_limit_level pp_limit_level, enum pp_power_type power_type); int amdgpu_dpm_set_power_limit(struct amdgpu_device *adev, - uint32_t limit); + uint32_t limit_type, uint32_t limit); int amdgpu_dpm_is_cclk_dpm_supported(struct amdgpu_device *adev); int amdgpu_dpm_debugfs_print_current_performance_level(struct amdgpu_device *adev, struct seq_file *m); @@ -614,5 +612,6 @@ int amdgpu_dpm_reset_vcn(struct amdgpu_device *adev, uint32_t inst_mask); bool amdgpu_dpm_reset_vcn_is_supported(struct amdgpu_device *adev); bool amdgpu_dpm_is_temp_metrics_supported(struct amdgpu_device *adev, enum smu_temp_metric_type type); +const struct ras_smu_drv *amdgpu_dpm_get_ras_smu_driver(struct amdgpu_device *adev); #endif diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c index cf9932e68055..1f539cc65f41 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.c @@ -2558,18 +2558,13 @@ static int si_enable_power_containment(struct amdgpu_device *adev, if (enable) { if (!si_should_disable_uvd_powertune(adev, amdgpu_new_state)) { smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_TDPClampingActive); - if (smc_result != PPSMC_Result_OK) { + if (smc_result != PPSMC_Result_OK) ret = -EINVAL; - ni_pi->pc_enabled = false; - } else { - ni_pi->pc_enabled = true; - } } } else { smc_result = amdgpu_si_send_msg_to_smc(adev, PPSMC_TDPClampingInactive); if (smc_result != PPSMC_Result_OK) ret = -EINVAL; - ni_pi->pc_enabled = false; } } @@ -3500,6 +3495,11 @@ static void si_apply_state_adjust_rules(struct amdgpu_device *adev, * for these GPUs to calculate bandwidth requirements. */ if (high_pixelclock_count) { + /* Work around flickering lines at the bottom edge + * of the screen when using a single 4K 60Hz monitor. + */ + disable_mclk_switching = true; + /* On Oland, we observe some flickering when two 4K 60Hz * displays are connected, possibly because voltage is too low. * Raise the voltage by requiring a higher SCLK. @@ -7046,13 +7046,20 @@ static void si_set_vce_clock(struct amdgpu_device *adev, if ((old_rps->evclk != new_rps->evclk) || (old_rps->ecclk != new_rps->ecclk)) { /* Turn the clocks on when encoding, off otherwise */ + dev_dbg(adev->dev, "set VCE clocks: %u, %u\n", new_rps->evclk, new_rps->ecclk); + if (new_rps->evclk || new_rps->ecclk) { - /* Place holder for future VCE1.0 porting to amdgpu - vce_v1_0_enable_mgcg(adev, false, false);*/ + amdgpu_asic_set_vce_clocks(adev, new_rps->evclk, new_rps->ecclk); + amdgpu_device_ip_set_clockgating_state( + adev, AMD_IP_BLOCK_TYPE_VCE, AMD_CG_STATE_UNGATE); + amdgpu_device_ip_set_powergating_state( + adev, AMD_IP_BLOCK_TYPE_VCE, AMD_PG_STATE_UNGATE); } else { - /* Place holder for future VCE1.0 porting to amdgpu - vce_v1_0_enable_mgcg(adev, true, false); - amdgpu_asic_set_vce_clocks(adev, new_rps->evclk, new_rps->ecclk);*/ + amdgpu_device_ip_set_powergating_state( + adev, AMD_IP_BLOCK_TYPE_VCE, AMD_PG_STATE_GATE); + amdgpu_device_ip_set_clockgating_state( + adev, AMD_IP_BLOCK_TYPE_VCE, AMD_CG_STATE_GATE); + amdgpu_asic_set_vce_clocks(adev, 0, 0); } } } @@ -7504,8 +7511,6 @@ static int si_dpm_init(struct amdgpu_device *adev) pi->pasi = CYPRESS_HASI_DFLT; pi->vrc = SISLANDS_VRC_DFLT; - pi->gfx_clock_gating = true; - eg_pi->sclk_deep_sleep = true; si_pi->sclk_deep_sleep_above_low = false; @@ -7516,7 +7521,6 @@ static int si_dpm_init(struct amdgpu_device *adev) eg_pi->dynamic_ac_timing = true; - eg_pi->light_sleep = true; #if defined(CONFIG_ACPI) eg_pi->pcie_performance_request = amdgpu_acpi_is_pcie_performance_request_supported(adev); @@ -7577,6 +7581,7 @@ static void si_dpm_debugfs_print_current_performance_level(void *handle, } else { pl = &ps->performance_levels[current_index]; seq_printf(m, "uvd vclk: %d dclk: %d\n", rps->vclk, rps->dclk); + seq_printf(m, "vce evclk: %d ecclk: %d\n", rps->evclk, rps->ecclk); seq_printf(m, "power level %d sclk: %u mclk: %u vddc: %u vddci: %u pcie gen: %u\n", current_index, pl->sclk, pl->mclk, pl->vddc, pl->vddci, pl->pcie_gen + 1); } diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.h b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.h index 11cb7874a6bb..3aed75fbf913 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.h +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/si_dpm.h @@ -38,11 +38,7 @@ #define MC_ARB_DRAM_TIMING2_2 0xa00 #define MC_ARB_DRAM_TIMING2_3 0xa01 -#define MAX_NO_OF_MVDD_VALUES 2 -#define MAX_NO_VREG_STEPS 32 #define NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 16 -#define SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE 32 -#define SMC_NISLANDS_MC_REGISTER_ARRAY_SET_COUNT 20 #define RV770_ASI_DFLT 1000 #define CYPRESS_HASI_DFLT 400000 #define PCIE_PERF_REQ_PECI_GEN1 2 @@ -51,11 +47,6 @@ #define RV770_DEFAULT_VCLK_FREQ 53300 /* 10 khz */ #define RV770_DEFAULT_DCLK_FREQ 40000 /* 10 khz */ -#define SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE 16 - -#define RV770_SMC_TABLE_ADDRESS 0xB000 -#define RV770_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 3 - #define SMC_STROBE_RATIO 0x0F #define SMC_STROBE_ENABLE 0x10 @@ -64,27 +55,6 @@ #define SMC_MC_RTT_ENABLE 0x04 #define SMC_MC_STUTTER_EN 0x08 -#define RV770_SMC_VOLTAGEMASK_VDDC 0 -#define RV770_SMC_VOLTAGEMASK_MVDD 1 -#define RV770_SMC_VOLTAGEMASK_VDDCI 2 -#define RV770_SMC_VOLTAGEMASK_MAX 4 - -#define NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE 16 -#define NISLANDS_SMC_STROBE_RATIO 0x0F -#define NISLANDS_SMC_STROBE_ENABLE 0x10 - -#define NISLANDS_SMC_MC_EDC_RD_FLAG 0x01 -#define NISLANDS_SMC_MC_EDC_WR_FLAG 0x02 -#define NISLANDS_SMC_MC_RTT_ENABLE 0x04 -#define NISLANDS_SMC_MC_STUTTER_EN 0x08 - -#define MAX_NO_VREG_STEPS 32 - -#define NISLANDS_SMC_VOLTAGEMASK_VDDC 0 -#define NISLANDS_SMC_VOLTAGEMASK_MVDD 1 -#define NISLANDS_SMC_VOLTAGEMASK_VDDCI 2 -#define NISLANDS_SMC_VOLTAGEMASK_MAX 4 - #define SISLANDS_MCREGISTERTABLE_INITIAL_SLOT 0 #define SISLANDS_MCREGISTERTABLE_ACPI_SLOT 1 #define SISLANDS_MCREGISTERTABLE_ULV_SLOT 2 @@ -219,32 +189,6 @@ enum si_cac_config_reg_type SISLANDS_CACCONFIG_MAX }; -enum si_power_level { - SI_POWER_LEVEL_LOW = 0, - SI_POWER_LEVEL_MEDIUM = 1, - SI_POWER_LEVEL_HIGH = 2, - SI_POWER_LEVEL_CTXSW = 3, -}; - -enum si_td { - SI_TD_AUTO, - SI_TD_UP, - SI_TD_DOWN, -}; - -enum si_display_watermark { - SI_DISPLAY_WATERMARK_LOW = 0, - SI_DISPLAY_WATERMARK_HIGH = 1, -}; - -enum si_display_gap -{ - SI_PM_DISPLAY_GAP_VBLANK_OR_WM = 0, - SI_PM_DISPLAY_GAP_VBLANK = 1, - SI_PM_DISPLAY_GAP_WATERMARK = 2, - SI_PM_DISPLAY_GAP_IGNORE = 3, -}; - extern const struct amdgpu_ip_block_version si_smu_ip_block; struct ni_leakage_coeffients @@ -258,56 +202,6 @@ struct ni_leakage_coeffients u32 t_ref; }; -struct SMC_Evergreen_MCRegisterAddress -{ - uint16_t s0; - uint16_t s1; -}; - -typedef struct SMC_Evergreen_MCRegisterAddress SMC_Evergreen_MCRegisterAddress; - -struct evergreen_mc_reg_entry { - u32 mclk_max; - u32 mc_data[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE]; -}; - -struct evergreen_mc_reg_table { - u8 last; - u8 num_entries; - u16 valid_flag; - struct evergreen_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES]; - SMC_Evergreen_MCRegisterAddress mc_reg_address[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE]; -}; - -struct SMC_Evergreen_MCRegisterSet -{ - uint32_t value[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE]; -}; - -typedef struct SMC_Evergreen_MCRegisterSet SMC_Evergreen_MCRegisterSet; - -struct SMC_Evergreen_MCRegisters -{ - uint8_t last; - uint8_t reserved[3]; - SMC_Evergreen_MCRegisterAddress address[SMC_EVERGREEN_MC_REGISTER_ARRAY_SIZE]; - SMC_Evergreen_MCRegisterSet data[5]; -}; - -typedef struct SMC_Evergreen_MCRegisters SMC_Evergreen_MCRegisters; - -struct SMC_NIslands_MCRegisterSet -{ - uint32_t value[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE]; -}; - -typedef struct SMC_NIslands_MCRegisterSet SMC_NIslands_MCRegisterSet; - -struct ni_mc_reg_entry { - u32 mclk_max; - u32 mc_data[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE]; -}; - struct SMC_NIslands_MCRegisterAddress { uint16_t s0; @@ -316,257 +210,20 @@ struct SMC_NIslands_MCRegisterAddress typedef struct SMC_NIslands_MCRegisterAddress SMC_NIslands_MCRegisterAddress; -struct SMC_NIslands_MCRegisters -{ - uint8_t last; - uint8_t reserved[3]; - SMC_NIslands_MCRegisterAddress address[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE]; - SMC_NIslands_MCRegisterSet data[SMC_NISLANDS_MC_REGISTER_ARRAY_SET_COUNT]; -}; - -typedef struct SMC_NIslands_MCRegisters SMC_NIslands_MCRegisters; - -struct evergreen_ulv_param { - bool supported; - struct rv7xx_pl *pl; -}; - -struct evergreen_arb_registers { - u32 mc_arb_dram_timing; - u32 mc_arb_dram_timing2; - u32 mc_arb_rfsh_rate; - u32 mc_arb_burst_time; -}; - -struct at { - u32 rlp; - u32 rmp; - u32 lhp; - u32 lmp; -}; - -struct ni_clock_registers { - u32 cg_spll_func_cntl; - u32 cg_spll_func_cntl_2; - u32 cg_spll_func_cntl_3; - u32 cg_spll_func_cntl_4; - u32 cg_spll_spread_spectrum; - u32 cg_spll_spread_spectrum_2; - u32 mclk_pwrmgt_cntl; - u32 dll_cntl; - u32 mpll_ad_func_cntl; - u32 mpll_ad_func_cntl_2; - u32 mpll_dq_func_cntl; - u32 mpll_dq_func_cntl_2; - u32 mpll_ss1; - u32 mpll_ss2; -}; - -struct RV770_SMC_SCLK_VALUE -{ - uint32_t vCG_SPLL_FUNC_CNTL; - uint32_t vCG_SPLL_FUNC_CNTL_2; - uint32_t vCG_SPLL_FUNC_CNTL_3; - uint32_t vCG_SPLL_SPREAD_SPECTRUM; - uint32_t vCG_SPLL_SPREAD_SPECTRUM_2; - uint32_t sclk_value; -}; - -typedef struct RV770_SMC_SCLK_VALUE RV770_SMC_SCLK_VALUE; - -struct RV770_SMC_MCLK_VALUE -{ - uint32_t vMPLL_AD_FUNC_CNTL; - uint32_t vMPLL_AD_FUNC_CNTL_2; - uint32_t vMPLL_DQ_FUNC_CNTL; - uint32_t vMPLL_DQ_FUNC_CNTL_2; - uint32_t vMCLK_PWRMGT_CNTL; - uint32_t vDLL_CNTL; - uint32_t vMPLL_SS; - uint32_t vMPLL_SS2; - uint32_t mclk_value; -}; - -typedef struct RV770_SMC_MCLK_VALUE RV770_SMC_MCLK_VALUE; - - -struct RV730_SMC_MCLK_VALUE -{ - uint32_t vMCLK_PWRMGT_CNTL; - uint32_t vDLL_CNTL; - uint32_t vMPLL_FUNC_CNTL; - uint32_t vMPLL_FUNC_CNTL2; - uint32_t vMPLL_FUNC_CNTL3; - uint32_t vMPLL_SS; - uint32_t vMPLL_SS2; - uint32_t mclk_value; -}; - -typedef struct RV730_SMC_MCLK_VALUE RV730_SMC_MCLK_VALUE; - -struct RV770_SMC_VOLTAGE_VALUE -{ - uint16_t value; - uint8_t index; - uint8_t padding; -}; - -typedef struct RV770_SMC_VOLTAGE_VALUE RV770_SMC_VOLTAGE_VALUE; - -union RV7XX_SMC_MCLK_VALUE -{ - RV770_SMC_MCLK_VALUE mclk770; - RV730_SMC_MCLK_VALUE mclk730; -}; - -typedef union RV7XX_SMC_MCLK_VALUE RV7XX_SMC_MCLK_VALUE, *LPRV7XX_SMC_MCLK_VALUE; - -struct RV770_SMC_HW_PERFORMANCE_LEVEL -{ - uint8_t arbValue; - union{ - uint8_t seqValue; - uint8_t ACIndex; - }; - uint8_t displayWatermark; - uint8_t gen2PCIE; - uint8_t gen2XSP; - uint8_t backbias; - uint8_t strobeMode; - uint8_t mcFlags; - uint32_t aT; - uint32_t bSP; - RV770_SMC_SCLK_VALUE sclk; - RV7XX_SMC_MCLK_VALUE mclk; - RV770_SMC_VOLTAGE_VALUE vddc; - RV770_SMC_VOLTAGE_VALUE mvdd; - RV770_SMC_VOLTAGE_VALUE vddci; - uint8_t reserved1; - uint8_t reserved2; - uint8_t stateFlags; - uint8_t padding; -}; - -typedef struct RV770_SMC_HW_PERFORMANCE_LEVEL RV770_SMC_HW_PERFORMANCE_LEVEL; - -struct RV770_SMC_SWSTATE -{ - uint8_t flags; - uint8_t padding1; - uint8_t padding2; - uint8_t padding3; - RV770_SMC_HW_PERFORMANCE_LEVEL levels[RV770_SMC_PERFORMANCE_LEVELS_PER_SWSTATE]; -}; - -typedef struct RV770_SMC_SWSTATE RV770_SMC_SWSTATE; - -struct RV770_SMC_VOLTAGEMASKTABLE -{ - uint8_t highMask[RV770_SMC_VOLTAGEMASK_MAX]; - uint32_t lowMask[RV770_SMC_VOLTAGEMASK_MAX]; -}; - -typedef struct RV770_SMC_VOLTAGEMASKTABLE RV770_SMC_VOLTAGEMASKTABLE; - -struct RV770_SMC_STATETABLE -{ - uint8_t thermalProtectType; - uint8_t systemFlags; - uint8_t maxVDDCIndexInPPTable; - uint8_t extraFlags; - uint8_t highSMIO[MAX_NO_VREG_STEPS]; - uint32_t lowSMIO[MAX_NO_VREG_STEPS]; - RV770_SMC_VOLTAGEMASKTABLE voltageMaskTable; - RV770_SMC_SWSTATE initialState; - RV770_SMC_SWSTATE ACPIState; - RV770_SMC_SWSTATE driverState; - RV770_SMC_SWSTATE ULVState; -}; - -typedef struct RV770_SMC_STATETABLE RV770_SMC_STATETABLE; - -struct vddc_table_entry { - u16 vddc; - u8 vddc_index; - u8 high_smio; - u32 low_smio; -}; - -struct rv770_clock_registers { - u32 cg_spll_func_cntl; - u32 cg_spll_func_cntl_2; - u32 cg_spll_func_cntl_3; - u32 cg_spll_spread_spectrum; - u32 cg_spll_spread_spectrum_2; - u32 mpll_ad_func_cntl; - u32 mpll_ad_func_cntl_2; - u32 mpll_dq_func_cntl; - u32 mpll_dq_func_cntl_2; - u32 mclk_pwrmgt_cntl; - u32 dll_cntl; - u32 mpll_ss1; - u32 mpll_ss2; -}; - -struct rv730_clock_registers { - u32 cg_spll_func_cntl; - u32 cg_spll_func_cntl_2; - u32 cg_spll_func_cntl_3; - u32 cg_spll_spread_spectrum; - u32 cg_spll_spread_spectrum_2; - u32 mclk_pwrmgt_cntl; - u32 dll_cntl; - u32 mpll_func_cntl; - u32 mpll_func_cntl2; - u32 mpll_func_cntl3; - u32 mpll_ss; - u32 mpll_ss2; -}; - -union r7xx_clock_registers { - struct rv770_clock_registers rv770; - struct rv730_clock_registers rv730; -}; - struct rv7xx_power_info { /* flags */ - bool mem_gddr5; - bool pcie_gen2; - bool dynamic_pcie_gen2; - bool acpi_pcie_gen2; - bool boot_in_gen2; bool voltage_control; /* vddc */ bool mvdd_control; bool sclk_ss; bool mclk_ss; bool dynamic_ss; - bool gfx_clock_gating; - bool mg_clock_gating; - bool mgcgtssm; - bool power_gating; bool thermal_protection; - bool display_gap; - bool dcodt; - bool ulps; - /* registers */ - union r7xx_clock_registers clk_regs; - u32 s0_vid_lower_smio_cntl; /* voltage */ - u32 vddc_mask_low; - u32 mvdd_mask_low; u32 mvdd_split_frequency; - u32 mvdd_low_smio[MAX_NO_OF_MVDD_VALUES]; u16 max_vddc; u16 max_vddc_in_table; u16 min_vddc_in_table; - struct vddc_table_entry vddc_table[MAX_NO_VREG_STEPS]; - u8 valid_vddc_entries; - /* dc odt */ - u32 mclk_odt_threshold; - u8 odt_value_0[2]; - u8 odt_value_1[2]; /* stored values */ - u32 boot_sclk; u16 acpi_vddc; u32 ref_div; u32 active_auto_throttle_sources; @@ -582,17 +239,6 @@ struct rv7xx_power_info { u32 asi; u32 pasi; u32 vrc; - u32 restricted_levels; - u32 rlp; - u32 rmp; - u32 lhp; - u32 lmp; - /* smc offsets */ - u16 state_table_start; - u16 soft_regs_start; - u16 sram_end; - /* scratch structs */ - RV770_SMC_STATETABLE smc_statetable; }; enum si_pcie_gen { @@ -611,44 +257,12 @@ struct rv7xx_pl { enum si_pcie_gen pcie_gen; /* si+ only */ }; -struct rv7xx_ps { - struct rv7xx_pl high; - struct rv7xx_pl medium; - struct rv7xx_pl low; - bool dc_compatible; -}; - struct si_ps { u16 performance_level_count; bool dc_compatible; struct rv7xx_pl performance_levels[NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE]; }; -struct ni_mc_reg_table { - u8 last; - u8 num_entries; - u16 valid_flag; - struct ni_mc_reg_entry mc_reg_table_entry[MAX_AC_TIMING_ENTRIES]; - SMC_NIslands_MCRegisterAddress mc_reg_address[SMC_NISLANDS_MC_REGISTER_ARRAY_SIZE]; -}; - -struct ni_cac_data -{ - struct ni_leakage_coeffients leakage_coefficients; - u32 i_leakage; - s32 leakage_minimum_temperature; - u32 pwr_const; - u32 dc_cac_value; - u32 bif_cac_value; - u32 lkge_pwr; - u8 mc_wr_weight; - u8 mc_rd_weight; - u8 allow_ovrflw; - u8 num_win_tdp; - u8 l2num_win_tdp; - u8 lts_truncate_n; -}; - struct evergreen_power_info { /* must be first! */ struct rv7xx_power_info rv7xx; @@ -657,203 +271,33 @@ struct evergreen_power_info { bool dynamic_ac_timing; bool abm; bool mcls; - bool light_sleep; - bool memory_transition; bool pcie_performance_request; - bool pcie_performance_request_registered; bool sclk_deep_sleep; - bool dll_default_on; - bool ls_clock_gating; bool smu_uvd_hs; bool uvd_enabled; /* stored values */ u16 acpi_vddci; - u8 mvdd_high_index; - u8 mvdd_low_index; u32 mclk_edc_wr_enable_threshold; - struct evergreen_mc_reg_table mc_reg_table; struct atom_voltage_table vddc_voltage_table; struct atom_voltage_table vddci_voltage_table; - struct evergreen_arb_registers bootup_arb_registers; - struct evergreen_ulv_param ulv; - struct at ats[2]; - /* smc offsets */ - u16 mc_reg_table_start; struct amdgpu_ps current_rps; - struct rv7xx_ps current_ps; struct amdgpu_ps requested_rps; - struct rv7xx_ps requested_ps; -}; - -struct PP_NIslands_Dpm2PerfLevel -{ - uint8_t MaxPS; - uint8_t TgtAct; - uint8_t MaxPS_StepInc; - uint8_t MaxPS_StepDec; - uint8_t PSST; - uint8_t NearTDPDec; - uint8_t AboveSafeInc; - uint8_t BelowSafeInc; - uint8_t PSDeltaLimit; - uint8_t PSDeltaWin; - uint8_t Reserved[6]; -}; - -typedef struct PP_NIslands_Dpm2PerfLevel PP_NIslands_Dpm2PerfLevel; - -struct PP_NIslands_DPM2Parameters -{ - uint32_t TDPLimit; - uint32_t NearTDPLimit; - uint32_t SafePowerLimit; - uint32_t PowerBoostLimit; -}; -typedef struct PP_NIslands_DPM2Parameters PP_NIslands_DPM2Parameters; - -struct NISLANDS_SMC_SCLK_VALUE -{ - uint32_t vCG_SPLL_FUNC_CNTL; - uint32_t vCG_SPLL_FUNC_CNTL_2; - uint32_t vCG_SPLL_FUNC_CNTL_3; - uint32_t vCG_SPLL_FUNC_CNTL_4; - uint32_t vCG_SPLL_SPREAD_SPECTRUM; - uint32_t vCG_SPLL_SPREAD_SPECTRUM_2; - uint32_t sclk_value; -}; - -typedef struct NISLANDS_SMC_SCLK_VALUE NISLANDS_SMC_SCLK_VALUE; - -struct NISLANDS_SMC_MCLK_VALUE -{ - uint32_t vMPLL_FUNC_CNTL; - uint32_t vMPLL_FUNC_CNTL_1; - uint32_t vMPLL_FUNC_CNTL_2; - uint32_t vMPLL_AD_FUNC_CNTL; - uint32_t vMPLL_AD_FUNC_CNTL_2; - uint32_t vMPLL_DQ_FUNC_CNTL; - uint32_t vMPLL_DQ_FUNC_CNTL_2; - uint32_t vMCLK_PWRMGT_CNTL; - uint32_t vDLL_CNTL; - uint32_t vMPLL_SS; - uint32_t vMPLL_SS2; - uint32_t mclk_value; -}; - -typedef struct NISLANDS_SMC_MCLK_VALUE NISLANDS_SMC_MCLK_VALUE; - -struct NISLANDS_SMC_VOLTAGE_VALUE -{ - uint16_t value; - uint8_t index; - uint8_t padding; -}; - -typedef struct NISLANDS_SMC_VOLTAGE_VALUE NISLANDS_SMC_VOLTAGE_VALUE; - -struct NISLANDS_SMC_HW_PERFORMANCE_LEVEL -{ - uint8_t arbValue; - uint8_t ACIndex; - uint8_t displayWatermark; - uint8_t gen2PCIE; - uint8_t reserved1; - uint8_t reserved2; - uint8_t strobeMode; - uint8_t mcFlags; - uint32_t aT; - uint32_t bSP; - NISLANDS_SMC_SCLK_VALUE sclk; - NISLANDS_SMC_MCLK_VALUE mclk; - NISLANDS_SMC_VOLTAGE_VALUE vddc; - NISLANDS_SMC_VOLTAGE_VALUE mvdd; - NISLANDS_SMC_VOLTAGE_VALUE vddci; - NISLANDS_SMC_VOLTAGE_VALUE std_vddc; - uint32_t powergate_en; - uint8_t hUp; - uint8_t hDown; - uint8_t stateFlags; - uint8_t arbRefreshState; - uint32_t SQPowerThrottle; - uint32_t SQPowerThrottle_2; - uint32_t reserved[2]; - PP_NIslands_Dpm2PerfLevel dpm2; -}; - -typedef struct NISLANDS_SMC_HW_PERFORMANCE_LEVEL NISLANDS_SMC_HW_PERFORMANCE_LEVEL; - -struct NISLANDS_SMC_SWSTATE -{ - uint8_t flags; - uint8_t levelCount; - uint8_t padding2; - uint8_t padding3; - NISLANDS_SMC_HW_PERFORMANCE_LEVEL levels[]; -}; - -typedef struct NISLANDS_SMC_SWSTATE NISLANDS_SMC_SWSTATE; - -struct NISLANDS_SMC_VOLTAGEMASKTABLE -{ - uint8_t highMask[NISLANDS_SMC_VOLTAGEMASK_MAX]; - uint32_t lowMask[NISLANDS_SMC_VOLTAGEMASK_MAX]; -}; - -typedef struct NISLANDS_SMC_VOLTAGEMASKTABLE NISLANDS_SMC_VOLTAGEMASKTABLE; - -#define NISLANDS_MAX_NO_VREG_STEPS 32 - -struct NISLANDS_SMC_STATETABLE -{ - uint8_t thermalProtectType; - uint8_t systemFlags; - uint8_t maxVDDCIndexInPPTable; - uint8_t extraFlags; - uint8_t highSMIO[NISLANDS_MAX_NO_VREG_STEPS]; - uint32_t lowSMIO[NISLANDS_MAX_NO_VREG_STEPS]; - NISLANDS_SMC_VOLTAGEMASKTABLE voltageMaskTable; - PP_NIslands_DPM2Parameters dpm2Params; - NISLANDS_SMC_SWSTATE initialState; - NISLANDS_SMC_SWSTATE ACPIState; - NISLANDS_SMC_SWSTATE ULVState; - NISLANDS_SMC_SWSTATE driverState; - NISLANDS_SMC_HW_PERFORMANCE_LEVEL dpmLevels[NISLANDS_MAX_SMC_PERFORMANCE_LEVELS_PER_SWSTATE - 1]; }; -typedef struct NISLANDS_SMC_STATETABLE NISLANDS_SMC_STATETABLE; - struct ni_power_info { /* must be first! */ struct evergreen_power_info eg; - struct ni_clock_registers clock_registers; - struct ni_mc_reg_table mc_reg_table; u32 mclk_rtt_mode_threshold; /* flags */ - bool use_power_boost_limit; bool support_cac_long_term_average; bool cac_enabled; bool cac_configuration_required; bool driver_calculate_cac_leakage; - bool pc_enabled; bool enable_power_containment; bool enable_cac; bool enable_sq_ramping; - /* smc offsets */ - u16 arb_table_start; - u16 fan_table_start; - u16 cac_table_start; - u16 spll_table_start; - /* CAC stuff */ - struct ni_cac_data cac_data; - u32 dc_cac_table[NISLANDS_DCCAC_MAX_LEVELS]; - const struct ni_cac_weights *cac_weights; - u8 lta_window_size; - u8 lts_truncate; struct si_ps current_ps; struct si_ps requested_ps; - /* scratch structs */ - SMC_NIslands_MCRegisters smc_mc_reg_table; - NISLANDS_SMC_STATETABLE smc_statetable; }; struct si_cac_config_reg @@ -952,7 +396,6 @@ struct si_leakage_voltage struct si_leakage_voltage_entry entries[SISLANDS_MAX_LEAKAGE_COUNT]; }; - struct si_ulv_param { bool supported; u32 cg_ulv_control; diff --git a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c index 554492dfa3c0..3aaf3dd71868 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c +++ b/drivers/gpu/drm/amd/pm/powerplay/amd_powerplay.c @@ -20,7 +20,6 @@ * OTHER DEALINGS IN THE SOFTWARE. * */ -#include "pp_debug.h" #include <linux/types.h> #include <linux/kernel.h> #include <linux/gfp.h> @@ -28,12 +27,10 @@ #include <linux/firmware.h> #include <linux/reboot.h> #include "amd_shared.h" -#include "amd_powerplay.h" #include "power_state.h" #include "amdgpu.h" #include "hwmgr.h" #include "amdgpu_dpm_internal.h" -#include "amdgpu_display.h" static const struct amd_pm_funcs pp_dpm_funcs; @@ -634,9 +631,12 @@ static int pp_dpm_get_pp_table(void *handle, char **table) { struct pp_hwmgr *hwmgr = handle; - if (!hwmgr || !hwmgr->pm_en || !hwmgr->soft_pp_table) + if (!hwmgr || !hwmgr->pm_en || !table) return -EINVAL; + if (!hwmgr->soft_pp_table) + return -EOPNOTSUPP; + *table = (char *)hwmgr->soft_pp_table; return hwmgr->soft_pp_table_size; } @@ -955,7 +955,7 @@ static int pp_dpm_switch_power_profile(void *handle, return 0; } -static int pp_set_power_limit(void *handle, uint32_t limit) +static int pp_set_power_limit(void *handle, uint32_t limit_type, uint32_t limit) { struct pp_hwmgr *hwmgr = handle; uint32_t max_power_limit; diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c index 8da882c51856..9b28c0728269 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu7_hwmgr.c @@ -5444,8 +5444,7 @@ static int smu7_get_thermal_temperature_range(struct pp_hwmgr *hwmgr, thermal_data->max = table_info->cac_dtp_table->usSoftwareShutdownTemp * PP_TEMPERATURE_UNITS_PER_CENTIGRADES; else if (hwmgr->pp_table_version == PP_TABLE_V0) - thermal_data->max = data->thermal_temp_setting.temperature_shutdown * - PP_TEMPERATURE_UNITS_PER_CENTIGRADES; + thermal_data->max = data->thermal_temp_setting.temperature_shutdown; thermal_data->sw_ctf_threshold = thermal_data->max; diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c index d2dbd90bb427..0a876c840c79 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/fiji_smumgr.c @@ -2024,7 +2024,7 @@ static int fiji_init_smc_table(struct pp_hwmgr *hwmgr) table->VoltageResponseTime = 0; table->PhaseResponseTime = 0; table->MemoryThermThrottleEnable = 1; - table->PCIeBootLinkLevel = 0; /* 0:Gen1 1:Gen2 2:Gen3*/ + table->PCIeBootLinkLevel = (uint8_t) (data->dpm_table.pcie_speed_table.count); table->PCIeGenInterval = 1; table->VRConfig = 0; diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c index 1f50f1e74c48..aa3ae9b115c4 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/iceland_smumgr.c @@ -2028,7 +2028,7 @@ static int iceland_init_smc_table(struct pp_hwmgr *hwmgr) table->VoltageResponseTime = 0; table->PhaseResponseTime = 0; table->MemoryThermThrottleEnable = 1; - table->PCIeBootLinkLevel = 0; + table->PCIeBootLinkLevel = (uint8_t) (data->dpm_table.pcie_speed_table.count); table->PCIeGenInterval = 1; result = iceland_populate_smc_svi2_config(hwmgr, table); diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu10_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu10_smumgr.c index ac9ec8257f82..38e19e5cad4d 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu10_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/smu10_smumgr.c @@ -139,7 +139,7 @@ static int smu10_copy_table_from_smc(struct pp_hwmgr *hwmgr, priv->smu_tables.entry[table_id].table_id, NULL); - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table, (uint8_t *)priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -164,7 +164,7 @@ static int smu10_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega10_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega10_smumgr.c index f9c0f117725d..0bf1bf5528c2 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega10_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega10_smumgr.c @@ -60,7 +60,7 @@ static int vega10_copy_table_from_smc(struct pp_hwmgr *hwmgr, priv->smu_tables.entry[table_id].table_id, NULL); - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -90,7 +90,7 @@ static int vega10_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega12_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega12_smumgr.c index d3ff6a831ed5..e2ba593faa5d 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega12_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega12_smumgr.c @@ -68,7 +68,7 @@ static int vega12_copy_table_from_smc(struct pp_hwmgr *hwmgr, "[CopyTableFromSMC] Attempt to Transfer Table From SMU Failed!", return -EINVAL); - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -98,7 +98,7 @@ static int vega12_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); PP_ASSERT_WITH_CODE(smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, diff --git a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega20_smumgr.c b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega20_smumgr.c index a5c95b180672..e3515156d26f 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega20_smumgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/smumgr/vega20_smumgr.c @@ -192,7 +192,7 @@ static int vega20_copy_table_from_smc(struct pp_hwmgr *hwmgr, "[CopyTableFromSMC] Attempt to Transfer Table From SMU Failed!", return ret); - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table, priv->smu_tables.entry[table_id].table, priv->smu_tables.entry[table_id].size); @@ -223,7 +223,7 @@ static int vega20_copy_table_to_smc(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[table_id].table, table, priv->smu_tables.entry[table_id].size); - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, @@ -256,7 +256,7 @@ int vega20_set_activity_monitor_coeff(struct pp_hwmgr *hwmgr, memcpy(priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].table, table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size); - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); PP_ASSERT_WITH_CODE((ret = smum_send_msg_to_smc_with_parameter(hwmgr, PPSMC_MSG_SetDriverDramAddrHigh, @@ -306,7 +306,7 @@ int vega20_get_activity_monitor_coeff(struct pp_hwmgr *hwmgr, "[GetActivityMonitor] Attempt to Transfer Table From SMU Failed!", return ret); - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].table, priv->smu_tables.entry[TABLE_ACTIVITY_MONITOR_COEFF].size); diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index fb8086859857..f51fa265230b 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -68,7 +68,7 @@ static int smu_handle_task(struct smu_context *smu, static int smu_reset(struct smu_context *smu); static int smu_set_fan_speed_pwm(void *handle, u32 speed); static int smu_set_fan_control_mode(void *handle, u32 value); -static int smu_set_power_limit(void *handle, uint32_t limit); +static int smu_set_power_limit(void *handle, uint32_t limit_type, uint32_t limit); static int smu_set_fan_speed_rpm(void *handle, uint32_t speed); static int smu_set_gfx_cgpg(struct smu_context *smu, bool enabled); static int smu_set_mp1_state(void *handle, enum pp_mp1_state mp1_state); @@ -508,11 +508,14 @@ static void smu_restore_dpm_user_profile(struct smu_context *smu) /* Enable restore flag */ smu->user_dpm_profile.flags |= SMU_DPM_USER_PROFILE_RESTORE; - /* set the user dpm power limit */ - if (smu->user_dpm_profile.power_limit) { - ret = smu_set_power_limit(smu, smu->user_dpm_profile.power_limit); + /* set the user dpm power limits */ + for (int i = SMU_DEFAULT_PPT_LIMIT; i < SMU_LIMIT_TYPE_COUNT; i++) { + if (!smu->user_dpm_profile.power_limits[i]) + continue; + ret = smu_set_power_limit(smu, i, + smu->user_dpm_profile.power_limits[i]); if (ret) - dev_err(smu->adev->dev, "Failed to set power limit value\n"); + dev_err(smu->adev->dev, "Failed to set %d power limit value\n", i); } /* set the user dpm clock configurations */ @@ -609,6 +612,17 @@ bool is_support_cclk_dpm(struct amdgpu_device *adev) return true; } +int amdgpu_smu_ras_send_msg(struct amdgpu_device *adev, enum smu_message_type msg, + uint32_t param, uint32_t *read_arg) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + int ret = -EOPNOTSUPP; + + if (smu->ppt_funcs && smu->ppt_funcs->ras_send_msg) + ret = smu->ppt_funcs->ras_send_msg(smu, msg, param, read_arg); + + return ret; +} static int smu_sys_get_pp_table(void *handle, char **table) @@ -620,7 +634,7 @@ static int smu_sys_get_pp_table(void *handle, return -EOPNOTSUPP; if (!smu_table->power_play_table && !smu_table->hardcode_pptable) - return -EINVAL; + return -EOPNOTSUPP; if (smu_table->hardcode_pptable) *table = smu_table->hardcode_pptable; @@ -1655,9 +1669,12 @@ static int smu_smc_hw_setup(struct smu_context *smu) if (adev->in_suspend && smu_is_dpm_running(smu)) { dev_info(adev->dev, "dpm has been enabled\n"); ret = smu_system_features_control(smu, true); - if (ret) + if (ret) { dev_err(adev->dev, "Failed system features control!\n"); - return ret; + return ret; + } + + return smu_enable_thermal_alert(smu); } break; default: @@ -2040,6 +2057,12 @@ static int smu_disable_dpms(struct smu_context *smu) smu->is_apu && (amdgpu_in_reset(adev) || adev->in_s0ix)) return 0; + /* vangogh s0ix */ + if ((amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 5, 0) || + amdgpu_ip_version(adev, MP1_HWIP, 0) == IP_VERSION(11, 5, 2)) && + adev->in_s0ix) + return 0; + /* * For gpu reset, runpm and hibernation through BACO, * BACO feature has to be kept enabled. @@ -2225,7 +2248,6 @@ static int smu_resume(struct amdgpu_ip_block *ip_block) int ret; struct amdgpu_device *adev = ip_block->adev; struct smu_context *smu = adev->powerplay.pp_handle; - struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); if (amdgpu_sriov_multi_vf_mode(adev)) return 0; @@ -2257,18 +2279,6 @@ static int smu_resume(struct amdgpu_ip_block *ip_block) adev->pm.dpm_enabled = true; - if (smu->current_power_limit) { - ret = smu_set_power_limit(smu, smu->current_power_limit); - if (ret && ret != -EOPNOTSUPP) - return ret; - } - - if (smu_dpm_ctx->dpm_level == AMD_DPM_FORCED_LEVEL_MANUAL && smu->od_enabled) { - ret = smu_od_edit_dpm_table(smu, PP_OD_COMMIT_DPM_TABLE, NULL, 0); - if (ret) - return ret; - } - dev_info(adev->dev, "SMU is resumed successfully!\n"); return 0; @@ -2796,6 +2806,17 @@ const struct amdgpu_ip_block_version smu_v14_0_ip_block = { .funcs = &smu_ip_funcs, }; +const struct ras_smu_drv *smu_get_ras_smu_driver(void *handle) +{ + struct smu_context *smu = (struct smu_context *)handle; + const struct ras_smu_drv *tmp = NULL; + int ret; + + ret = smu_get_ras_smu_drv(smu, &tmp); + + return ret ? NULL : tmp; +} + static int smu_load_microcode(void *handle) { struct smu_context *smu = handle; @@ -2889,6 +2910,9 @@ int smu_get_power_limit(void *handle, if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled) return -EOPNOTSUPP; + if (!limit) + return -EINVAL; + switch (pp_power_type) { case PP_PWR_TYPE_SUSTAINED: limit_type = SMU_DEFAULT_PPT_LIMIT; @@ -2920,6 +2944,8 @@ int smu_get_power_limit(void *handle, if (limit_type != SMU_DEFAULT_PPT_LIMIT) { if (smu->ppt_funcs->get_ppt_limit) ret = smu->ppt_funcs->get_ppt_limit(smu, limit, limit_type, limit_level); + else + return -EOPNOTSUPP; } else { switch (limit_level) { case SMU_PPT_LIMIT_CURRENT: @@ -2958,37 +2984,34 @@ int smu_get_power_limit(void *handle, return ret; } -static int smu_set_power_limit(void *handle, uint32_t limit) +static int smu_set_power_limit(void *handle, uint32_t limit_type, uint32_t limit) { struct smu_context *smu = handle; - uint32_t limit_type = limit >> 24; int ret = 0; if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled) return -EOPNOTSUPP; - limit &= (1<<24)-1; - if (limit_type != SMU_DEFAULT_PPT_LIMIT) - if (smu->ppt_funcs->set_power_limit) - return smu->ppt_funcs->set_power_limit(smu, limit_type, limit); - - if ((limit > smu->max_power_limit) || (limit < smu->min_power_limit)) { - dev_err(smu->adev->dev, - "New power limit (%d) is out of range [%d,%d]\n", - limit, smu->min_power_limit, smu->max_power_limit); - return -EINVAL; + if (limit_type == SMU_DEFAULT_PPT_LIMIT) { + if (!limit) + limit = smu->current_power_limit; + if ((limit > smu->max_power_limit) || (limit < smu->min_power_limit)) { + dev_err(smu->adev->dev, + "New power limit (%d) is out of range [%d,%d]\n", + limit, smu->min_power_limit, smu->max_power_limit); + return -EINVAL; + } } - if (!limit) - limit = smu->current_power_limit; - if (smu->ppt_funcs->set_power_limit) { ret = smu->ppt_funcs->set_power_limit(smu, limit_type, limit); - if (!ret && !(smu->user_dpm_profile.flags & SMU_DPM_USER_PROFILE_RESTORE)) - smu->user_dpm_profile.power_limit = limit; + if (ret) + return ret; + if (!(smu->user_dpm_profile.flags & SMU_DPM_USER_PROFILE_RESTORE)) + smu->user_dpm_profile.power_limits[limit_type] = limit; } - return ret; + return 0; } static int smu_print_smuclk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h index 582c186d8b62..8815fc70b63b 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h @@ -212,6 +212,7 @@ enum smu_power_src_type { enum smu_ppt_limit_type { SMU_DEFAULT_PPT_LIMIT = 0, SMU_FAST_PPT_LIMIT, + SMU_LIMIT_TYPE_COUNT, }; enum smu_ppt_limit_level { @@ -231,7 +232,7 @@ enum smu_memory_pool_size { struct smu_user_dpm_profile { uint32_t fan_mode; - uint32_t power_limit; + uint32_t power_limits[SMU_LIMIT_TYPE_COUNT]; uint32_t fan_speed_pwm; uint32_t fan_speed_rpm; uint32_t flags; @@ -1521,6 +1522,21 @@ struct pptable_funcs { */ ssize_t (*get_xcp_metrics)(struct smu_context *smu, int xcp_id, void *table); + /** + * @ras_send_msg: Send a message with a parameter from Ras + * &msg: Type of message. + * ¶m: Message parameter. + * &read_arg: SMU response (optional). + */ + int (*ras_send_msg)(struct smu_context *smu, + enum smu_message_type msg, uint32_t param, uint32_t *read_arg); + + + /** + * @get_ras_smu_drv: Get RAS smu driver interface + * Return: ras_smu_drv * + */ + int (*get_ras_smu_drv)(struct smu_context *smu, const struct ras_smu_drv **ras_smu_drv); }; typedef enum { @@ -1785,7 +1801,10 @@ int smu_set_pm_policy(struct smu_context *smu, enum pp_pm_policy p_type, int level); ssize_t smu_get_pm_policy_info(struct smu_context *smu, enum pp_pm_policy p_type, char *sysbuf); +const struct ras_smu_drv *smu_get_ras_smu_driver(void *handle); +int amdgpu_smu_ras_send_msg(struct amdgpu_device *adev, enum smu_message_type msg, + uint32_t param, uint32_t *readarg); #endif void smu_feature_cap_set(struct smu_context *smu, enum smu_feature_cap_id fea_id); diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h index bf6aa9620911..dd30d96e1ca2 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_pmfw.h @@ -87,7 +87,7 @@ typedef enum { /*37*/ FEATURE_DVO = 37, /*38*/ FEATURE_XVMINORPSM_CLKSTOP_DS = 38, /*39*/ FEATURE_GLOBAL_DPM = 39, -/*40*/ FEATURE_NODE_POWER_MANAGER = 40, +/*40*/ FEATURE_HROM_EN = 40, /*41*/ NUM_FEATURES = 41 } FEATURE_LIST_e; @@ -189,7 +189,7 @@ typedef enum { SVI_MAX_TEMP_ENTRIES, // 13 } SVI_TEMP_e; -#define SMU_METRICS_TABLE_VERSION 0x14 +#define SMU_METRICS_TABLE_VERSION 0x15 #define SMU_SYSTEM_METRICS_TABLE_VERSION 0x1 @@ -367,6 +367,11 @@ typedef struct { //Node Power Limit uint32_t MaxNodePowerLimit; + + // PPT1 Configuration + uint32_t PPT1Max; + uint32_t PPT1Min; + uint32_t PPT1Default; } StaticMetricsTable_t; #pragma pack(pop) diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_ppsmc.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_ppsmc.h index 4b066c42e0ec..d09b6ae9827e 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_ppsmc.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_12_ppsmc.h @@ -105,23 +105,21 @@ #define PPSMC_MSG_UpdatePccWaitDecMaxStr 0x4C #define PPSMC_MSG_ResetSDMA 0x4D #define PPSMC_MSG_GetRasTableVersion 0x4E -#define PPSMC_MSG_GetRmaStatus 0x4F -#define PPSMC_MSG_GetErrorCount 0x50 -#define PPSMC_MSG_GetBadPageCount 0x51 -#define PPSMC_MSG_GetBadPageInfo 0x52 -#define PPSMC_MSG_GetBadPagePaAddrLoHi 0x53 -#define PPSMC_MSG_SetTimestampLoHi 0x54 -#define PPSMC_MSG_GetTimestampLoHi 0x55 -#define PPSMC_MSG_GetRasPolicy 0x56 -#define PPSMC_MSG_DumpErrorRecord 0x57 +#define PPSMC_MSG_GetBadPageCount 0x50 +#define PPSMC_MSG_GetBadPageMcaAddress 0x51 +#define PPSMC_MSG_SetTimestamp 0x53 +#define PPSMC_MSG_SetTimestampHi 0x54 +#define PPSMC_MSG_GetTimestamp 0x55 +#define PPSMC_MSG_GetBadPageIpIdLoHi 0x57 #define PPSMC_MSG_EraseRasTable 0x58 #define PPSMC_MSG_GetStaticMetricsTable 0x59 #define PPSMC_MSG_ResetVfArbitersByIndex 0x5A -#define PPSMC_MSG_GetBadPageSeverity 0x5B #define PPSMC_MSG_GetSystemMetricsTable 0x5C #define PPSMC_MSG_GetSystemMetricsVersion 0x5D #define PPSMC_MSG_ResetVCN 0x5E -#define PPSMC_Message_Count 0x5F +#define PPSMC_MSG_SetFastPptLimit 0x5F +#define PPSMC_MSG_GetFastPptLimit 0x60 +#define PPSMC_Message_Count 0x61 //PPSMC Reset Types for driver msg argument #define PPSMC_RESET_TYPE_DRIVER_MODE_1_RESET 0x1 diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h index 2256c77da636..9b71a8afdd35 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/smu_types.h @@ -279,7 +279,16 @@ __SMU_DUMMY_MAP(ResetSDMA), \ __SMU_DUMMY_MAP(ResetVCN), \ __SMU_DUMMY_MAP(GetStaticMetricsTable), \ - __SMU_DUMMY_MAP(GetSystemMetricsTable), + __SMU_DUMMY_MAP(GetSystemMetricsTable), \ + __SMU_DUMMY_MAP(GetRASTableVersion), \ + __SMU_DUMMY_MAP(GetBadPageCount), \ + __SMU_DUMMY_MAP(GetBadPageMcaAddr), \ + __SMU_DUMMY_MAP(SetTimestamp), \ + __SMU_DUMMY_MAP(GetTimestamp), \ + __SMU_DUMMY_MAP(GetBadPageIpid), \ + __SMU_DUMMY_MAP(EraseRasTable), \ + __SMU_DUMMY_MAP(SetFastPptLimit), \ + __SMU_DUMMY_MAP(GetFastPptLimit), #undef __SMU_DUMMY_MAP #define __SMU_DUMMY_MAP(type) SMU_MSG_##type @@ -458,7 +467,8 @@ enum smu_clk_type { __SMU_DUMMY_MAP(GFX_EDC_XVMIN), \ __SMU_DUMMY_MAP(GFX_DIDT_XVMIN), \ __SMU_DUMMY_MAP(FAN_ABNORMAL), \ - __SMU_DUMMY_MAP(PIT), + __SMU_DUMMY_MAP(PIT), \ + __SMU_DUMMY_MAP(HROM_EN), #undef __SMU_DUMMY_MAP #define __SMU_DUMMY_MAP(feature) SMU_FEATURE_##feature##_BIT diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/cyan_skillfish_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/cyan_skillfish_ppt.c index 9548bd3c624b..55401e6b2b0b 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/cyan_skillfish_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/cyan_skillfish_ppt.c @@ -291,11 +291,12 @@ static int cyan_skillfish_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int ret = 0, size = 0; + int ret = 0, size = 0, start_offset = 0; uint32_t cur_value = 0; int i; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -353,7 +354,7 @@ static int cyan_skillfish_print_clk_levels(struct smu_context *smu, return ret; } - return size; + return size - start_offset; } static bool cyan_skillfish_is_dpm_running(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c index 0028f10ead42..7c9f77124ab2 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c @@ -1469,7 +1469,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { uint16_t *curve_settings; - int i, levels, size = 0, ret = 0; + int i, levels, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t freq_values[3] = {0}; uint32_t mark_index = 0; @@ -1484,6 +1484,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, uint32_t min_value, max_value; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_GFXCLK: @@ -1497,11 +1498,11 @@ static int navi10_print_clk_levels(struct smu_context *smu, case SMU_DCEFCLK: ret = navi10_get_current_clk_freq_by_table(smu, clk_type, &cur_value); if (ret) - return size; + return size - start_offset; ret = smu_v11_0_get_dpm_level_count(smu, clk_type, &count); if (ret) - return size; + return size - start_offset; ret = navi10_is_support_fine_grained_dpm(smu, clk_type); if (ret < 0) @@ -1511,7 +1512,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, for (i = 0; i < count; i++) { ret = smu_v11_0_get_dpm_freq_by_index(smu, clk_type, i, &value); if (ret) - return size; + return size - start_offset; size += sysfs_emit_at(buf, size, "%d: %uMhz %s\n", i, value, cur_value == value ? "*" : ""); @@ -1519,10 +1520,10 @@ static int navi10_print_clk_levels(struct smu_context *smu, } else { ret = smu_v11_0_get_dpm_freq_by_index(smu, clk_type, 0, &freq_values[0]); if (ret) - return size; + return size - start_offset; ret = smu_v11_0_get_dpm_freq_by_index(smu, clk_type, count - 1, &freq_values[2]); if (ret) - return size; + return size - start_offset; freq_values[1] = cur_value; mark_index = cur_value == freq_values[0] ? 0 : @@ -1653,7 +1654,7 @@ static int navi10_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int navi10_force_clk_levels(struct smu_context *smu, @@ -2888,7 +2889,7 @@ static int navi10_set_dummy_pstates_table_location(struct smu_context *smu) dummy_table += 0x1000; } - amdgpu_asic_flush_hdp(smu->adev, NULL); + amdgpu_hdp_flush(smu->adev, NULL); ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SET_DRIVER_DUMMY_TABLE_DRAM_ADDR_HIGH, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c index 31c2c0386b1f..774283ac7827 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c @@ -1281,7 +1281,7 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu, struct smu_11_0_7_overdrive_table *od_settings = smu->od_settings; OverDriveTable_t *od_table = (OverDriveTable_t *)table_context->overdrive_table; - int i, size = 0, ret = 0; + int i, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t freq_values[3] = {0}; uint32_t mark_index = 0; @@ -1289,6 +1289,7 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu, uint32_t min_value, max_value; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_GFXCLK: @@ -1434,7 +1435,7 @@ static int sienna_cichlid_print_clk_levels(struct smu_context *smu, } print_clk_out: - return size; + return size - start_offset; } static int sienna_cichlid_force_clk_levels(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c index 78e4186d06cc..b0d6487171d7 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/smu_v11_0.c @@ -1022,7 +1022,12 @@ int smu_v11_0_enable_thermal_alert(struct smu_context *smu) int smu_v11_0_disable_thermal_alert(struct smu_context *smu) { - return amdgpu_irq_put(smu->adev, &smu->irq_source, 0); + int ret = 0; + + if (smu->smu_table.thermal_controller_type) + ret = amdgpu_irq_put(smu->adev, &smu->irq_source, 0); + + return ret; } static uint16_t convert_to_vddc(uint8_t vid) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c index 2c9869feba61..9626da2dba58 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c @@ -565,7 +565,7 @@ static int vangogh_print_legacy_clk_levels(struct smu_context *smu, DpmClocks_t *clk_table = smu->smu_table.clocks_table; SmuMetrics_legacy_t metrics; struct smu_dpm_context *smu_dpm_ctx = &(smu->smu_dpm); - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; bool cur_value_match_level = false; @@ -576,6 +576,7 @@ static int vangogh_print_legacy_clk_levels(struct smu_context *smu, return ret; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -658,7 +659,7 @@ static int vangogh_print_legacy_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int vangogh_print_clk_levels(struct smu_context *smu, @@ -666,7 +667,7 @@ static int vangogh_print_clk_levels(struct smu_context *smu, { DpmClocks_t *clk_table = smu->smu_table.clocks_table; SmuMetrics_t metrics; - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; bool cur_value_match_level = false; uint32_t min, max; @@ -678,6 +679,7 @@ static int vangogh_print_clk_levels(struct smu_context *smu, return ret; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -779,7 +781,7 @@ static int vangogh_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int vangogh_common_print_clk_levels(struct smu_context *smu, @@ -2217,6 +2219,9 @@ static int vangogh_post_smu_init(struct smu_context *smu) uint32_t total_cu = adev->gfx.config.max_cu_per_sh * adev->gfx.config.max_sh_per_se * adev->gfx.config.max_shader_engines; + if (adev->in_s0ix) + return 0; + /* allow message will be sent after enable message on Vangogh*/ if (smu_cmn_feature_is_enabled(smu, SMU_FEATURE_DPM_GFXCLK_BIT) && (adev->pg_flags & AMD_PG_SUPPORT_GFX_PG)) { @@ -2308,8 +2313,7 @@ static int vangogh_get_power_limit(struct smu_context *smu, uint32_t *max_power_limit, uint32_t *min_power_limit) { - struct smu_11_5_power_context *power_context = - smu->smu_power.power_context; + struct smu_11_5_power_context *power_context = smu->smu_power.power_context; uint32_t ppt_limit; int ret = 0; @@ -2345,12 +2349,11 @@ static int vangogh_get_power_limit(struct smu_context *smu, } static int vangogh_get_ppt_limit(struct smu_context *smu, - uint32_t *ppt_limit, - enum smu_ppt_limit_type type, - enum smu_ppt_limit_level level) + uint32_t *ppt_limit, + enum smu_ppt_limit_type type, + enum smu_ppt_limit_level level) { - struct smu_11_5_power_context *power_context = - smu->smu_power.power_context; + struct smu_11_5_power_context *power_context = smu->smu_power.power_context; if (!power_context) return -EOPNOTSUPP; @@ -2399,7 +2402,6 @@ static int vangogh_set_power_limit(struct smu_context *smu, smu->current_power_limit = ppt_limit; break; case SMU_FAST_PPT_LIMIT: - ppt_limit &= ~(SMU_FAST_PPT_LIMIT << 24); if (ppt_limit > power_context->max_fast_ppt_limit) { dev_err(smu->adev->dev, "New power limit (%d) is over the max allowed %d\n", diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c index 3baf20f4c373..eaa9ea162f16 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu12/renoir_ppt.c @@ -494,7 +494,7 @@ static int renoir_set_fine_grain_gfx_freq_parameters(struct smu_context *smu) static int renoir_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0, min = 0, max = 0; SmuMetrics_t metrics; bool cur_value_match_level = false; @@ -506,6 +506,7 @@ static int renoir_print_clk_levels(struct smu_context *smu, return ret; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_RANGE: @@ -550,7 +551,7 @@ static int renoir_print_clk_levels(struct smu_context *smu, size += sysfs_emit_at(buf, size, "2: %uMhz %s\n", max, i == 2 ? "*" : ""); } - return size; + return size - start_offset; case SMU_SOCCLK: count = NUM_SOCCLK_DPM_LEVELS; cur_value = metrics.ClockFrequency[CLOCK_SOCCLK]; @@ -607,7 +608,7 @@ static int renoir_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static enum amd_pm_state_type renoir_get_current_power_state(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c index c1062e5f0393..677781060246 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c @@ -1195,15 +1195,16 @@ static int smu_v13_0_0_print_clk_levels(struct smu_context *smu, struct smu_13_0_dpm_table *single_dpm_table; struct smu_13_0_pcie_table *pcie_table; uint32_t gen_speed, lane_width; - int i, curr_freq, size = 0; + int i, curr_freq, size = 0, start_offset = 0; int32_t min_value, max_value; int ret = 0; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; if (amdgpu_ras_intr_triggered()) { size += sysfs_emit_at(buf, size, "unavailable\n"); - return size; + return size - start_offset; } switch (clk_type) { @@ -1534,7 +1535,7 @@ static int smu_v13_0_0_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c index cb3fea9e8cf3..9e635f733fbf 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_12_ppt.c @@ -34,6 +34,7 @@ #include "amdgpu_fru_eeprom.h" #include <linux/pci.h> #include "smu_cmn.h" +#include "amdgpu_ras.h" #undef MP1_Public #undef smnMP1_FIRMWARE_FLAGS @@ -58,7 +59,7 @@ #define NUM_JPEG_RINGS_FW 10 #define NUM_JPEG_RINGS_GPU_METRICS(gpu_metrics) \ - (ARRAY_SIZE(gpu_metrics->xcp_stats[0].jpeg_busy) / 4) + (ARRAY_SIZE(gpu_metrics->jpeg_busy) / 4) const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[SMU_FEATURE_COUNT] = { SMU_13_0_12_FEA_MAP(SMU_FEATURE_DATA_CALCULATIONS_BIT, FEATURE_DATA_CALCULATION), @@ -81,6 +82,7 @@ const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[SMU_FEATURE_COUNT] = SMU_13_0_12_FEA_MAP(SMU_FEATURE_DS_MPIOCLK_BIT, FEATURE_DS_MPIOCLK), SMU_13_0_12_FEA_MAP(SMU_FEATURE_DS_MP0CLK_BIT, FEATURE_DS_MP0CLK), SMU_13_0_12_FEA_MAP(SMU_FEATURE_PIT_BIT, FEATURE_PIT), + SMU_13_0_12_FEA_MAP(SMU_FEATURE_HROM_EN_BIT, FEATURE_HROM_EN), }; const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[SMU_MSG_MAX_COUNT] = { @@ -139,6 +141,15 @@ const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[SMU_MSG_MAX_COUNT] = MSG_MAP(ResetVCN, PPSMC_MSG_ResetVCN, 0), MSG_MAP(GetStaticMetricsTable, PPSMC_MSG_GetStaticMetricsTable, 1), MSG_MAP(GetSystemMetricsTable, PPSMC_MSG_GetSystemMetricsTable, 1), + MSG_MAP(GetRASTableVersion, PPSMC_MSG_GetRasTableVersion, 0), + MSG_MAP(GetBadPageCount, PPSMC_MSG_GetBadPageCount, 0), + MSG_MAP(GetBadPageMcaAddr, PPSMC_MSG_GetBadPageMcaAddress, 0), + MSG_MAP(SetTimestamp, PPSMC_MSG_SetTimestamp, 0), + MSG_MAP(GetTimestamp, PPSMC_MSG_GetTimestamp, 0), + MSG_MAP(GetBadPageIpid, PPSMC_MSG_GetBadPageIpIdLoHi, 0), + MSG_MAP(EraseRasTable, PPSMC_MSG_EraseRasTable, 0), + MSG_MAP(SetFastPptLimit, PPSMC_MSG_SetFastPptLimit, 1), + MSG_MAP(GetFastPptLimit, PPSMC_MSG_GetFastPptLimit, 1), }; int smu_v13_0_12_tables_init(struct smu_context *smu) @@ -345,6 +356,12 @@ int smu_v13_0_12_setup_driver_pptable(struct smu_context *smu) if (smu_v13_0_6_cap_supported(smu, SMU_CAP(NPM_METRICS))) pptable->MaxNodePowerLimit = SMUQ10_ROUND(static_metrics->MaxNodePowerLimit); + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(FAST_PPT)) && + static_metrics->PPT1Max) { + pptable->PPT1Max = static_metrics->PPT1Max; + pptable->PPT1Min = static_metrics->PPT1Min; + pptable->PPT1Default = static_metrics->PPT1Default; + } smu_v13_0_12_init_xgmi_data(smu, static_metrics); pptable->Init = true; } @@ -449,7 +466,7 @@ static int smu_v13_0_12_get_system_metrics_table(struct smu_context *smu) return ret; } - amdgpu_asic_invalidate_hdp(smu->adev, NULL); + amdgpu_hdp_invalidate(smu->adev, NULL); smu_table_cache_update_time(sys_table, jiffies); memcpy(sys_table->cache.buffer, table->cpu_addr, smu_v13_0_12_get_system_metrics_size()); @@ -719,15 +736,14 @@ static ssize_t smu_v13_0_12_get_temp_metrics(struct smu_context *smu, ssize_t smu_v13_0_12_get_xcp_metrics(struct smu_context *smu, struct amdgpu_xcp *xcp, void *table, void *smu_metrics) { const u8 num_jpeg_rings = NUM_JPEG_RINGS_FW; - struct amdgpu_partition_metrics_v1_0 *xcp_metrics; + struct smu_v13_0_6_partition_metrics *xcp_metrics; struct amdgpu_device *adev = smu->adev; MetricsTable_t *metrics; int inst, j, k, idx; u32 inst_mask; metrics = (MetricsTable_t *)smu_metrics; - xcp_metrics = (struct amdgpu_partition_metrics_v1_0 *) table; - smu_cmn_init_partition_metrics(xcp_metrics, 1, 0); + xcp_metrics = (struct smu_v13_0_6_partition_metrics *)table; amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask); idx = 0; for_each_inst(k, inst_mask) { @@ -772,22 +788,17 @@ ssize_t smu_v13_0_12_get_xcp_metrics(struct smu_context *smu, struct amdgpu_xcp return sizeof(*xcp_metrics); } -ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, void *smu_metrics) +void smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, + void *smu_metrics, + struct smu_v13_0_6_gpu_metrics *gpu_metrics) { - struct smu_table_context *smu_table = &smu->smu_table; - struct gpu_metrics_v1_8 *gpu_metrics = - (struct gpu_metrics_v1_8 *)smu_table->gpu_metrics_table; - int ret = 0, xcc_id, inst, i, j, k, idx; struct amdgpu_device *adev = smu->adev; + int ret = 0, xcc_id, inst, i, j; u8 num_jpeg_rings_gpu_metrics; MetricsTable_t *metrics; - struct amdgpu_xcp *xcp; - u32 inst_mask; metrics = (MetricsTable_t *)smu_metrics; - smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 8); - gpu_metrics->temperature_hotspot = SMUQ10_ROUND(metrics->MaxSocketTemperature); /* Individual HBM stack temperature is not reported */ @@ -877,60 +888,186 @@ ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, void gpu_metrics->xgmi_link_status[j] = ret; } - gpu_metrics->num_partition = adev->xcp_mgr->num_xcps; - num_jpeg_rings_gpu_metrics = NUM_JPEG_RINGS_GPU_METRICS(gpu_metrics); - for_each_xcp(adev->xcp_mgr, xcp, i) { - amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask); - idx = 0; - for_each_inst(k, inst_mask) { - /* Both JPEG and VCN has same instances */ - inst = GET_INST(VCN, k); - - for (j = 0; j < num_jpeg_rings_gpu_metrics; ++j) { - gpu_metrics->xcp_stats[i].jpeg_busy - [(idx * num_jpeg_rings_gpu_metrics) + j] = - SMUQ10_ROUND(metrics->JpegBusy - [(inst * NUM_JPEG_RINGS_FW) + j]); - } - gpu_metrics->xcp_stats[i].vcn_busy[idx] = - SMUQ10_ROUND(metrics->VcnBusy[inst]); - idx++; + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + inst = GET_INST(VCN, i); + + for (j = 0; j < num_jpeg_rings_gpu_metrics; ++j) { + gpu_metrics->jpeg_busy[(i * num_jpeg_rings_gpu_metrics) + + j] = + SMUQ10_ROUND( + metrics->JpegBusy[(inst * + NUM_JPEG_RINGS_FW) + + j]); } + gpu_metrics->vcn_busy[i] = SMUQ10_ROUND(metrics->VcnBusy[inst]); + } - amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask); - idx = 0; - for_each_inst(k, inst_mask) { - inst = GET_INST(GC, k); - gpu_metrics->xcp_stats[i].gfx_busy_inst[idx] = - SMUQ10_ROUND(metrics->GfxBusy[inst]); - gpu_metrics->xcp_stats[i].gfx_busy_acc[idx] = - SMUQ10_ROUND(metrics->GfxBusyAcc[inst]); - if (smu_v13_0_6_cap_supported(smu, SMU_CAP(HST_LIMIT_METRICS))) { - gpu_metrics->xcp_stats[i].gfx_below_host_limit_ppt_acc[idx] = - SMUQ10_ROUND(metrics->GfxclkBelowHostLimitPptAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_below_host_limit_thm_acc[idx] = - SMUQ10_ROUND(metrics->GfxclkBelowHostLimitThmAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_low_utilization_acc[idx] = - SMUQ10_ROUND(metrics->GfxclkLowUtilizationAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_below_host_limit_total_acc[idx] = - SMUQ10_ROUND(metrics->GfxclkBelowHostLimitTotalAcc[inst]); - } - idx++; - } + for (i = 0; i < NUM_XCC(adev->gfx.xcc_mask); ++i) { + inst = GET_INST(GC, i); + gpu_metrics->gfx_busy_inst[i] = + SMUQ10_ROUND(metrics->GfxBusy[inst]); + gpu_metrics->gfx_busy_acc[i] = + SMUQ10_ROUND(metrics->GfxBusyAcc[inst]); + if (smu_v13_0_6_cap_supported(smu, + SMU_CAP(HST_LIMIT_METRICS))) { + gpu_metrics + ->gfx_below_host_limit_ppt_acc[i] = SMUQ10_ROUND( + metrics->GfxclkBelowHostLimitPptAcc[inst]); + gpu_metrics + ->gfx_below_host_limit_thm_acc[i] = SMUQ10_ROUND( + metrics->GfxclkBelowHostLimitThmAcc[inst]); + gpu_metrics->gfx_low_utilization_acc[i] = SMUQ10_ROUND( + metrics->GfxclkLowUtilizationAcc[inst]); + gpu_metrics->gfx_below_host_limit_total_acc + [i] = SMUQ10_ROUND( + metrics->GfxclkBelowHostLimitTotalAcc[inst]); + }; } gpu_metrics->xgmi_link_width = metrics->XgmiWidth; gpu_metrics->xgmi_link_speed = metrics->XgmiBitrate; gpu_metrics->firmware_timestamp = metrics->Timestamp; - - *table = (void *)gpu_metrics; - - return sizeof(*gpu_metrics); } const struct smu_temp_funcs smu_v13_0_12_temp_funcs = { .temp_metrics_is_supported = smu_v13_0_12_is_temp_metrics_supported, .get_temp_metrics = smu_v13_0_12_get_temp_metrics, }; + +static int smu_v13_0_12_get_ras_table_version(struct amdgpu_device *adev, + uint32_t *table_version) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + + return smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetRASTableVersion, 0, table_version); +} + +static int smu_v13_0_12_get_badpage_count(struct amdgpu_device *adev, uint32_t *count, + uint32_t timeout) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + uint64_t end, now; + int ret = 0; + + now = (uint64_t)ktime_to_ms(ktime_get()); + end = now + timeout; + do { + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetBadPageCount, 0, count); + /* eeprom is not ready */ + if (ret != -EBUSY) + return ret; + mdelay(10); + now = (uint64_t)ktime_to_ms(ktime_get()); + } while (now < end); + + dev_err(adev->dev, + "smu get bad page count timeout!\n"); + return ret; +} + +static int smu_v13_0_12_set_timestamp(struct amdgpu_device *adev, uint64_t timestamp) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + + return smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_SetTimestamp, (uint32_t)timestamp, 0); +} + +static int smu_v13_0_12_get_timestamp(struct amdgpu_device *adev, + uint16_t index, uint64_t *timestamp) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + uint32_t temp; + int ret; + + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetTimestamp, index, &temp); + if (!ret) + *timestamp = temp; + + return ret; +} + +static int smu_v13_0_12_get_badpage_ipid(struct amdgpu_device *adev, + uint16_t index, uint64_t *ipid) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + uint32_t temp_arg, temp_ipid_lo, temp_ipid_high; + int ret; + + temp_arg = index | (1 << 16); + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetBadPageIpid, temp_arg, &temp_ipid_lo); + if (ret) + return ret; + + temp_arg = index | (2 << 16); + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetBadPageIpid, temp_arg, &temp_ipid_high); + if (!ret) + *ipid = (uint64_t)temp_ipid_high << 32 | temp_ipid_lo; + return ret; +} + +static int smu_v13_0_12_erase_ras_table(struct amdgpu_device *adev, + uint32_t *result) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + + return smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_EraseRasTable, 0, result); +} + +static int smu_v13_0_12_get_badpage_mca_addr(struct amdgpu_device *adev, + uint16_t index, uint64_t *mca_addr) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + uint32_t temp_arg, temp_addr_lo, temp_addr_high; + int ret; + + temp_arg = index | (1 << 16); + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetBadPageMcaAddr, temp_arg, &temp_addr_lo); + if (ret) + return ret; + + temp_arg = index | (2 << 16); + ret = smu_cmn_send_smc_msg_with_param(smu, + SMU_MSG_GetBadPageMcaAddr, temp_arg, &temp_addr_high); + if (!ret) + *mca_addr = (uint64_t)temp_addr_high << 32 | temp_addr_lo; + return ret; +} + +static const struct ras_eeprom_smu_funcs smu_v13_0_12_eeprom_smu_funcs = { + .get_ras_table_version = smu_v13_0_12_get_ras_table_version, + .get_badpage_count = smu_v13_0_12_get_badpage_count, + .get_badpage_mca_addr = smu_v13_0_12_get_badpage_mca_addr, + .set_timestamp = smu_v13_0_12_set_timestamp, + .get_timestamp = smu_v13_0_12_get_timestamp, + .get_badpage_ipid = smu_v13_0_12_get_badpage_ipid, + .erase_ras_table = smu_v13_0_12_erase_ras_table, +}; + +static void smu_v13_0_12_ras_smu_feature_flags(struct amdgpu_device *adev, uint64_t *flags) +{ + struct smu_context *smu = adev->powerplay.pp_handle; + + if (!flags) + return; + + *flags = 0ULL; + + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(RAS_EEPROM))) + *flags |= RAS_SMU_FEATURE_BIT__RAS_EEPROM; + +} + +const struct ras_smu_drv smu_v13_0_12_ras_smu_drv = { + .smu_eeprom_funcs = &smu_v13_0_12_eeprom_smu_funcs, + .ras_smu_feature_flags = smu_v13_0_12_ras_smu_feature_flags, +}; diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c index b081ae3e8f43..6908f9930f16 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_4_ppt.c @@ -497,11 +497,12 @@ static int smu_v13_0_4_get_dpm_level_count(struct smu_context *smu, static int smu_v13_0_4_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t min, max; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -565,7 +566,7 @@ static int smu_v13_0_4_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int smu_v13_0_4_read_sensor(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c index f5db181ef489..4576bf008b22 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c @@ -861,11 +861,12 @@ out: static int smu_v13_0_5_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t min = 0, max = 0; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -928,7 +929,7 @@ static int smu_v13_0_5_print_clk_levels(struct smu_context *smu, } print_clk_out: - return size; + return size - start_offset; } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index 285cf7979693..44e1cd821eec 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c @@ -356,6 +356,9 @@ static void smu_v13_0_12_init_caps(struct smu_context *smu) if (fw_ver > 0x04560900) smu_v13_0_6_cap_set(smu, SMU_CAP(VCN_RESET)); + if (fw_ver >= 0x04560D00) + smu_v13_0_6_cap_set(smu, SMU_CAP(FAST_PPT)); + if (fw_ver >= 0x04560700) { if (fw_ver >= 0x04560900) { smu_v13_0_6_cap_set(smu, SMU_CAP(TEMP_METRICS)); @@ -450,7 +453,8 @@ static void smu_v13_0_6_init_caps(struct smu_context *smu) ((pgm == 4) && (fw_ver >= 0x4557000))) smu_v13_0_6_cap_set(smu, SMU_CAP(SDMA_RESET)); - if ((pgm == 0) && (fw_ver >= 0x00558200)) + if ((pgm == 0 && fw_ver >= 0x00558200) || + (pgm == 7 && fw_ver >= 0x07551400)) smu_v13_0_6_cap_set(smu, SMU_CAP(VCN_RESET)); } @@ -548,7 +552,7 @@ static int smu_v13_0_6_tables_init(struct smu_context *smu) { struct smu_table_context *smu_table = &smu->smu_table; struct smu_table *tables = smu_table->tables; - void *gpu_metrics_table __free(kfree) = NULL; + struct smu_v13_0_6_gpu_metrics *gpu_metrics; void *driver_pptable __free(kfree) = NULL; void *metrics_table __free(kfree) = NULL; struct amdgpu_device *adev = smu->adev; @@ -578,24 +582,28 @@ static int smu_v13_0_6_tables_init(struct smu_context *smu) return -ENOMEM; smu_table->metrics_time = 0; - smu_table->gpu_metrics_table_size = sizeof(struct gpu_metrics_v1_8); - gpu_metrics_table = - kzalloc(smu_table->gpu_metrics_table_size, GFP_KERNEL); - if (!gpu_metrics_table) - return -ENOMEM; - driver_pptable = kzalloc(sizeof(struct PPTable_t), GFP_KERNEL); if (!driver_pptable) return -ENOMEM; + ret = smu_table_cache_init(smu, SMU_TABLE_SMU_METRICS, + sizeof(struct smu_v13_0_6_gpu_metrics), 1); + if (ret) + return ret; + + gpu_metrics = (struct smu_v13_0_6_gpu_metrics + *)(tables[SMU_TABLE_SMU_METRICS].cache.buffer); + + smu_v13_0_6_gpu_metrics_init(gpu_metrics, 1, 9); if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12)) { ret = smu_v13_0_12_tables_init(smu); - if (ret) + if (ret) { + smu_table_cache_fini(smu, SMU_TABLE_SMU_METRICS); return ret; + } } - smu_table->gpu_metrics_table = no_free_ptr(gpu_metrics_table); smu_table->metrics_table = no_free_ptr(metrics_table); smu_table->driver_pptable = no_free_ptr(driver_pptable); @@ -731,6 +739,7 @@ static int smu_v13_0_6_fini_smc_tables(struct smu_context *smu) { if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12)) smu_v13_0_12_tables_fini(smu); + smu_table_cache_fini(smu, SMU_TABLE_SMU_METRICS); return smu_v13_0_fini_smc_tables(smu); } @@ -765,7 +774,7 @@ int smu_v13_0_6_get_metrics_table(struct smu_context *smu, void *metrics_table, return ret; } - amdgpu_asic_invalidate_hdp(smu->adev, NULL); + amdgpu_hdp_invalidate(smu->adev, NULL); memcpy(smu_table->metrics_table, table->cpu_addr, table_size); smu_table->metrics_time = jiffies; @@ -844,12 +853,23 @@ int smu_v13_0_6_get_static_metrics_table(struct smu_context *smu) return ret; } - amdgpu_asic_invalidate_hdp(smu->adev, NULL); + amdgpu_hdp_invalidate(smu->adev, NULL); memcpy(smu_table->metrics_table, table->cpu_addr, table_size); return 0; } +static void smu_v13_0_6_update_caps(struct smu_context *smu) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct PPTable_t *pptable = + (struct PPTable_t *)smu_table->driver_pptable; + + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(FAST_PPT)) && + !pptable->PPT1Max) + smu_v13_0_6_cap_clear(smu, SMU_CAP(FAST_PPT)); +} + static int smu_v13_0_6_setup_driver_pptable(struct smu_context *smu) { struct smu_table_context *smu_table = &smu->smu_table; @@ -866,8 +886,12 @@ static int smu_v13_0_6_setup_driver_pptable(struct smu_context *smu) uint8_t max_width; if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12) && - smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) - return smu_v13_0_12_setup_driver_pptable(smu); + smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) { + ret = smu_v13_0_12_setup_driver_pptable(smu); + if (ret) + return ret; + goto out; + } /* Store one-time values in driver PPTable */ if (!pptable->Init) { @@ -947,7 +971,8 @@ static int smu_v13_0_6_setup_driver_pptable(struct smu_context *smu) smu_v13_0_6_fill_static_metrics_table(smu, static_metrics); } } - +out: + smu_v13_0_6_update_caps(smu); return 0; } @@ -1393,7 +1418,7 @@ static int smu_v13_0_6_print_clks(struct smu_context *smu, char *buf, int size, return -EINVAL; if (curr_clk < SMU_13_0_6_DSCLK_THRESHOLD) { - size = sysfs_emit_at(buf, size, "S: %uMhz *\n", curr_clk); + size += sysfs_emit_at(buf, size, "S: %uMhz *\n", curr_clk); for (i = 0; i < clocks.num_levels; i++) size += sysfs_emit_at(buf, size, "%d: %uMhz\n", i, clocks.data[i].clocks_in_khz / @@ -1428,7 +1453,7 @@ static int smu_v13_0_6_print_clks(struct smu_context *smu, char *buf, int size, static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, enum smu_clk_type type, char *buf) { - int now, size = 0; + int now, size = 0, start_offset = 0; int ret = 0; struct smu_umd_pstate_table *pstate_table = &smu->pstate_table; struct smu_13_0_dpm_table *single_dpm_table; @@ -1437,10 +1462,11 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, uint32_t min_clk, max_clk; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; if (amdgpu_ras_intr_triggered()) { size += sysfs_emit_at(buf, size, "unavailable\n"); - return size; + return size - start_offset; } dpm_context = smu_dpm->dpm_context; @@ -1512,9 +1538,13 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, single_dpm_table = &(dpm_context->dpm_tables.uclk_table); - return smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, - now, "mclk"); + ret = smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, + now, "mclk"); + if (ret < 0) + return ret; + size += ret; + break; case SMU_SOCCLK: ret = smu_v13_0_6_get_current_clk_freq_by_table(smu, SMU_SOCCLK, &now); @@ -1526,9 +1556,13 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, single_dpm_table = &(dpm_context->dpm_tables.soc_table); - return smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, - now, "socclk"); + ret = smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, + now, "socclk"); + if (ret < 0) + return ret; + size += ret; + break; case SMU_FCLK: ret = smu_v13_0_6_get_current_clk_freq_by_table(smu, SMU_FCLK, &now); @@ -1540,9 +1574,13 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, single_dpm_table = &(dpm_context->dpm_tables.fclk_table); - return smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, - now, "fclk"); + ret = smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, + now, "fclk"); + if (ret < 0) + return ret; + size += ret; + break; case SMU_VCLK: ret = smu_v13_0_6_get_current_clk_freq_by_table(smu, SMU_VCLK, &now); @@ -1554,9 +1592,13 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, single_dpm_table = &(dpm_context->dpm_tables.vclk_table); - return smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, - now, "vclk"); + ret = smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, + now, "vclk"); + if (ret < 0) + return ret; + size += ret; + break; case SMU_DCLK: ret = smu_v13_0_6_get_current_clk_freq_by_table(smu, SMU_DCLK, &now); @@ -1568,14 +1610,18 @@ static int smu_v13_0_6_print_clk_levels(struct smu_context *smu, single_dpm_table = &(dpm_context->dpm_tables.dclk_table); - return smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, - now, "dclk"); + ret = smu_v13_0_6_print_clks(smu, buf, size, single_dpm_table, + now, "dclk"); + if (ret < 0) + return ret; + size += ret; + break; default: break; } - return size; + return size - start_offset; } static int smu_v13_0_6_upload_dpm_level(struct smu_context *smu, bool max, @@ -1845,7 +1891,7 @@ static int smu_v13_0_6_get_power_limit(struct smu_context *smu, if (current_power_limit) *current_power_limit = power_limit; if (default_power_limit) - *default_power_limit = power_limit; + *default_power_limit = pptable->MaxSocketPowerLimit; if (max_power_limit) { *max_power_limit = pptable->MaxSocketPowerLimit; @@ -1860,9 +1906,66 @@ static int smu_v13_0_6_set_power_limit(struct smu_context *smu, enum smu_ppt_limit_type limit_type, uint32_t limit) { + struct smu_table_context *smu_table = &smu->smu_table; + struct PPTable_t *pptable = + (struct PPTable_t *)smu_table->driver_pptable; + int ret; + + if (limit_type == SMU_FAST_PPT_LIMIT) { + if (!smu_v13_0_6_cap_supported(smu, SMU_CAP(FAST_PPT))) + return -EOPNOTSUPP; + if (limit > pptable->PPT1Max || limit < pptable->PPT1Min) { + dev_err(smu->adev->dev, + "New power limit (%d) should be between min %d max %d\n", + limit, pptable->PPT1Min, pptable->PPT1Max); + return -EINVAL; + } + ret = smu_cmn_send_smc_msg_with_param(smu, SMU_MSG_SetFastPptLimit, + limit, NULL); + if (ret) + dev_err(smu->adev->dev, "Set fast PPT limit failed!\n"); + return ret; + } + return smu_v13_0_set_power_limit(smu, limit_type, limit); } +static int smu_v13_0_6_get_ppt_limit(struct smu_context *smu, + uint32_t *ppt_limit, + enum smu_ppt_limit_type type, + enum smu_ppt_limit_level level) +{ + struct smu_table_context *smu_table = &smu->smu_table; + struct PPTable_t *pptable = + (struct PPTable_t *)smu_table->driver_pptable; + int ret = 0; + + if (type == SMU_FAST_PPT_LIMIT) { + if (!smu_v13_0_6_cap_supported(smu, SMU_CAP(FAST_PPT))) + return -EOPNOTSUPP; + switch (level) { + case SMU_PPT_LIMIT_MAX: + *ppt_limit = pptable->PPT1Max; + break; + case SMU_PPT_LIMIT_CURRENT: + ret = smu_cmn_send_smc_msg(smu, SMU_MSG_GetFastPptLimit, ppt_limit); + if (ret) + dev_err(smu->adev->dev, "Get fast PPT limit failed!\n"); + break; + case SMU_PPT_LIMIT_DEFAULT: + *ppt_limit = pptable->PPT1Default; + break; + case SMU_PPT_LIMIT_MIN: + *ppt_limit = pptable->PPT1Min; + break; + default: + return -EOPNOTSUPP; + } + return ret; + } + return -EOPNOTSUPP; +} + static int smu_v13_0_6_irq_process(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -2383,7 +2486,7 @@ static int smu_v13_0_6_request_i2c_xfer(struct smu_context *smu, memcpy(table->cpu_addr, table_data, table_size); /* Flush hdp cache */ - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); ret = smu_cmn_send_smc_msg(smu, SMU_MSG_RequestI2cTransaction, NULL); @@ -2627,7 +2730,7 @@ static ssize_t smu_v13_0_6_get_xcp_metrics(struct smu_context *smu, int xcp_id, { const u8 num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS_4_0_3; int version = smu_v13_0_6_get_metrics_version(smu); - struct amdgpu_partition_metrics_v1_0 *xcp_metrics; + struct smu_v13_0_6_partition_metrics *xcp_metrics; MetricsTableV0_t *metrics_v0 __free(kfree) = NULL; struct amdgpu_device *adev = smu->adev; int ret, inst, i, j, k, idx; @@ -2647,8 +2750,8 @@ static ssize_t smu_v13_0_6_get_xcp_metrics(struct smu_context *smu, int xcp_id, if (i == adev->xcp_mgr->num_xcps) return -EINVAL; - xcp_metrics = (struct amdgpu_partition_metrics_v1_0 *)table; - smu_cmn_init_partition_metrics(xcp_metrics, 1, 0); + xcp_metrics = (struct smu_v13_0_6_partition_metrics *)table; + smu_v13_0_6_partition_metrics_init(xcp_metrics, 1, 1); metrics_v0 = kzalloc(METRICS_TABLE_SIZE, GFP_KERNEL); if (!metrics_v0) @@ -2740,18 +2843,16 @@ static ssize_t smu_v13_0_6_get_xcp_metrics(struct smu_context *smu, int xcp_id, static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table) { struct smu_table_context *smu_table = &smu->smu_table; - struct gpu_metrics_v1_8 *gpu_metrics = - (struct gpu_metrics_v1_8 *)smu_table->gpu_metrics_table; + struct smu_table *tables = smu_table->tables; + struct smu_v13_0_6_gpu_metrics *gpu_metrics; int version = smu_v13_0_6_get_metrics_version(smu); MetricsTableV0_t *metrics_v0 __free(kfree) = NULL; - int ret = 0, xcc_id, inst, i, j, k, idx; struct amdgpu_device *adev = smu->adev; + int ret = 0, xcc_id, inst, i, j; MetricsTableV1_t *metrics_v1; MetricsTableV2_t *metrics_v2; - struct amdgpu_xcp *xcp; u16 link_width_level; u8 num_jpeg_rings; - u32 inst_mask; bool per_inst; metrics_v0 = kzalloc(METRICS_TABLE_SIZE, GFP_KERNEL); @@ -2759,16 +2860,20 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table if (ret) return ret; - if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == - IP_VERSION(13, 0, 12) && - smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) - return smu_v13_0_12_get_gpu_metrics(smu, table, metrics_v0); + metrics_v2 = (MetricsTableV2_t *)metrics_v0; + gpu_metrics = (struct smu_v13_0_6_gpu_metrics + *)(tables[SMU_TABLE_SMU_METRICS].cache.buffer); + + if (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12) && + smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) { + smu_v13_0_12_get_gpu_metrics(smu, table, metrics_v0, + gpu_metrics); + goto fill; + } metrics_v1 = (MetricsTableV1_t *)metrics_v0; metrics_v2 = (MetricsTableV2_t *)metrics_v0; - smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 8); - gpu_metrics->temperature_hotspot = SMUQ10_ROUND(GET_METRIC_FIELD(MaxSocketTemperature, version)); /* Individual HBM stack temperature is not reported */ @@ -2889,55 +2994,49 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table gpu_metrics->xgmi_link_status[j] = ret; } - gpu_metrics->num_partition = adev->xcp_mgr->num_xcps; - per_inst = smu_v13_0_6_cap_supported(smu, SMU_CAP(PER_INST_METRICS)); num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS_4_0_3; - for_each_xcp(adev->xcp_mgr, xcp, i) { - amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_VCN, &inst_mask); - idx = 0; - for_each_inst(k, inst_mask) { - /* Both JPEG and VCN has same instances */ - inst = GET_INST(VCN, k); - - for (j = 0; j < num_jpeg_rings; ++j) { - gpu_metrics->xcp_stats[i].jpeg_busy - [(idx * num_jpeg_rings) + j] = - SMUQ10_ROUND(GET_METRIC_FIELD(JpegBusy, version) - [(inst * num_jpeg_rings) + j]); - } - gpu_metrics->xcp_stats[i].vcn_busy[idx] = - SMUQ10_ROUND(GET_METRIC_FIELD(VcnBusy, version)[inst]); - idx++; - - } + for (i = 0; i < adev->jpeg.num_jpeg_inst; ++i) { + inst = GET_INST(JPEG, i); + for (j = 0; j < num_jpeg_rings; ++j) + gpu_metrics->jpeg_busy[(i * num_jpeg_rings) + j] = + SMUQ10_ROUND(GET_METRIC_FIELD( + JpegBusy, + version)[(inst * num_jpeg_rings) + j]); + } + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + inst = GET_INST(VCN, i); + gpu_metrics->vcn_busy[i] = + SMUQ10_ROUND(GET_METRIC_FIELD(VcnBusy, version)[inst]); + } - if (per_inst) { - amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask); - idx = 0; - for_each_inst(k, inst_mask) { - inst = GET_INST(GC, k); - gpu_metrics->xcp_stats[i].gfx_busy_inst[idx] = - SMUQ10_ROUND(GET_GPU_METRIC_FIELD(GfxBusy, version)[inst]); - gpu_metrics->xcp_stats[i].gfx_busy_acc[idx] = - SMUQ10_ROUND(GET_GPU_METRIC_FIELD(GfxBusyAcc, - version)[inst]); - if (smu_v13_0_6_cap_supported(smu, SMU_CAP(HST_LIMIT_METRICS))) { - gpu_metrics->xcp_stats[i].gfx_below_host_limit_ppt_acc[idx] = - SMUQ10_ROUND - (metrics_v0->GfxclkBelowHostLimitPptAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_below_host_limit_thm_acc[idx] = - SMUQ10_ROUND - (metrics_v0->GfxclkBelowHostLimitThmAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_low_utilization_acc[idx] = - SMUQ10_ROUND - (metrics_v0->GfxclkLowUtilizationAcc[inst]); - gpu_metrics->xcp_stats[i].gfx_below_host_limit_total_acc[idx] = - SMUQ10_ROUND - (metrics_v0->GfxclkBelowHostLimitTotalAcc[inst]); - } - idx++; + if (per_inst) { + for (i = 0; i < NUM_XCC(adev->gfx.xcc_mask); ++i) { + inst = GET_INST(GC, i); + gpu_metrics->gfx_busy_inst[i] = SMUQ10_ROUND( + GET_GPU_METRIC_FIELD(GfxBusy, version)[inst]); + gpu_metrics->gfx_busy_acc[i] = SMUQ10_ROUND( + GET_GPU_METRIC_FIELD(GfxBusyAcc, + version)[inst]); + if (smu_v13_0_6_cap_supported( + smu, SMU_CAP(HST_LIMIT_METRICS))) { + gpu_metrics->gfx_below_host_limit_ppt_acc + [i] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitPptAcc + [inst]); + gpu_metrics->gfx_below_host_limit_thm_acc + [i] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitThmAcc + [inst]); + gpu_metrics->gfx_low_utilization_acc + [i] = SMUQ10_ROUND( + metrics_v0 + ->GfxclkLowUtilizationAcc[inst]); + gpu_metrics->gfx_below_host_limit_total_acc + [i] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitTotalAcc + [inst]); } } } @@ -2947,7 +3046,8 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table gpu_metrics->firmware_timestamp = GET_METRIC_FIELD(Timestamp, version); - *table = (void *)gpu_metrics; +fill: + *table = tables[SMU_TABLE_SMU_METRICS].cache.buffer; return sizeof(*gpu_metrics); } @@ -3226,6 +3326,24 @@ static int smu_v13_0_6_reset_vcn(struct smu_context *smu, uint32_t inst_mask) return ret; } +static int smu_v13_0_6_ras_send_msg(struct smu_context *smu, enum smu_message_type msg, uint32_t param, uint32_t *read_arg) +{ + int ret; + + switch (msg) { + case SMU_MSG_QueryValidMcaCount: + case SMU_MSG_QueryValidMcaCeCount: + case SMU_MSG_McaBankDumpDW: + case SMU_MSG_McaBankCeDumpDW: + case SMU_MSG_ClearMcaOnRead: + ret = smu_cmn_send_smc_msg_with_param(smu, msg, param, read_arg); + break; + default: + ret = -EPERM; + } + + return ret; +} static int smu_v13_0_6_post_init(struct smu_context *smu) { @@ -3863,6 +3981,29 @@ static void smu_v13_0_6_set_temp_funcs(struct smu_context *smu) == IP_VERSION(13, 0, 12)) ? &smu_v13_0_12_temp_funcs : NULL; } +static int smu_v13_0_6_get_ras_smu_drv(struct smu_context *smu, const struct ras_smu_drv **ras_smu_drv) +{ + if (!ras_smu_drv) + return -EINVAL; + + if (amdgpu_sriov_vf(smu->adev)) + return -EOPNOTSUPP; + + if (smu_cmn_feature_is_enabled(smu, SMU_FEATURE_HROM_EN_BIT)) + smu_v13_0_6_cap_set(smu, SMU_CAP(RAS_EEPROM)); + + switch (amdgpu_ip_version(smu->adev, MP1_HWIP, 0)) { + case IP_VERSION(13, 0, 12): + *ras_smu_drv = &smu_v13_0_12_ras_smu_drv; + break; + default: + *ras_smu_drv = NULL; + break; + } + + return 0; +} + static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { /* init dpm */ .get_allowed_feature_mask = smu_v13_0_6_get_allowed_feature_mask, @@ -3894,6 +4035,7 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { .get_enabled_mask = smu_v13_0_6_get_enabled_mask, .feature_is_enabled = smu_cmn_feature_is_enabled, .set_power_limit = smu_v13_0_6_set_power_limit, + .get_ppt_limit = smu_v13_0_6_get_ppt_limit, .set_xgmi_pstate = smu_v13_0_set_xgmi_pstate, .register_irq_handler = smu_v13_0_6_register_irq_handler, .enable_thermal_alert = smu_v13_0_enable_thermal_alert, @@ -3921,6 +4063,8 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { .reset_sdma = smu_v13_0_6_reset_sdma, .dpm_reset_vcn = smu_v13_0_6_reset_vcn, .post_init = smu_v13_0_6_post_init, + .ras_send_msg = smu_v13_0_6_ras_send_msg, + .get_ras_smu_drv = smu_v13_0_6_get_ras_smu_drv, }; void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h index 7ef5f3e66c27..6cbdd7c5ded9 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.h @@ -50,6 +50,9 @@ struct PPTable_t { uint32_t MinLclkDpmRange; uint64_t PublicSerialNumber_AID; uint32_t MaxNodePowerLimit; + uint32_t PPT1Max; + uint32_t PPT1Min; + uint32_t PPT1Default; bool Init; }; @@ -72,9 +75,18 @@ enum smu_v13_0_6_caps { SMU_CAP(PLDM_VERSION), SMU_CAP(TEMP_METRICS), SMU_CAP(NPM_METRICS), + SMU_CAP(RAS_EEPROM), + SMU_CAP(FAST_PPT), SMU_CAP(ALL), }; +#define SMU_13_0_6_NUM_XGMI_LINKS 8 +#define SMU_13_0_6_MAX_GFX_CLKS 8 +#define SMU_13_0_6_MAX_CLKS 4 +#define SMU_13_0_6_MAX_XCC 8 +#define SMU_13_0_6_MAX_VCN 4 +#define SMU_13_0_6_MAX_JPEG 40 + extern void smu_v13_0_6_set_ppt_funcs(struct smu_context *smu); bool smu_v13_0_6_cap_supported(struct smu_context *smu, enum smu_v13_0_6_caps cap); int smu_v13_0_6_get_static_metrics_table(struct smu_context *smu); @@ -87,7 +99,6 @@ size_t smu_v13_0_12_get_system_metrics_size(void); int smu_v13_0_12_setup_driver_pptable(struct smu_context *smu); int smu_v13_0_12_get_smu_metrics_data(struct smu_context *smu, MetricsMember_t member, uint32_t *value); -ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, void *smu_metrics); ssize_t smu_v13_0_12_get_xcp_metrics(struct smu_context *smu, struct amdgpu_xcp *xcp, void *table, void *smu_metrics); @@ -99,4 +110,156 @@ int smu_v13_0_12_get_npm_data(struct smu_context *smu, extern const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[]; extern const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[]; extern const struct smu_temp_funcs smu_v13_0_12_temp_funcs; +extern const struct ras_smu_drv smu_v13_0_12_ras_smu_drv; + +#if defined(SWSMU_CODE_LAYER_L2) +#include "smu_cmn.h" + +/* SMUv 13.0.6 GPU metrics*/ +#define SMU_13_0_6_METRICS_FIELDS(SMU_SCALAR, SMU_ARRAY) \ + SMU_SCALAR(SMU_MATTR(TEMPERATURE_HOTSPOT), SMU_MUNIT(TEMP_1), \ + SMU_MTYPE(U16), temperature_hotspot); \ + SMU_SCALAR(SMU_MATTR(TEMPERATURE_MEM), SMU_MUNIT(TEMP_1), \ + SMU_MTYPE(U16), temperature_mem); \ + SMU_SCALAR(SMU_MATTR(TEMPERATURE_VRSOC), SMU_MUNIT(TEMP_1), \ + SMU_MTYPE(U16), temperature_vrsoc); \ + SMU_SCALAR(SMU_MATTR(CURR_SOCKET_POWER), SMU_MUNIT(POWER_1), \ + SMU_MTYPE(U16), curr_socket_power); \ + SMU_SCALAR(SMU_MATTR(AVERAGE_GFX_ACTIVITY), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U16), average_gfx_activity); \ + SMU_SCALAR(SMU_MATTR(AVERAGE_UMC_ACTIVITY), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U16), average_umc_activity); \ + SMU_SCALAR(SMU_MATTR(MEM_MAX_BANDWIDTH), SMU_MUNIT(BW_1), \ + SMU_MTYPE(U64), mem_max_bandwidth); \ + SMU_SCALAR(SMU_MATTR(ENERGY_ACCUMULATOR), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), energy_accumulator); \ + SMU_SCALAR(SMU_MATTR(SYSTEM_CLOCK_COUNTER), SMU_MUNIT(TIME_1), \ + SMU_MTYPE(U64), system_clock_counter); \ + SMU_SCALAR(SMU_MATTR(ACCUMULATION_COUNTER), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), accumulation_counter); \ + SMU_SCALAR(SMU_MATTR(PROCHOT_RESIDENCY_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), prochot_residency_acc); \ + SMU_SCALAR(SMU_MATTR(PPT_RESIDENCY_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), ppt_residency_acc); \ + SMU_SCALAR(SMU_MATTR(SOCKET_THM_RESIDENCY_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), socket_thm_residency_acc); \ + SMU_SCALAR(SMU_MATTR(VR_THM_RESIDENCY_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), vr_thm_residency_acc); \ + SMU_SCALAR(SMU_MATTR(HBM_THM_RESIDENCY_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), hbm_thm_residency_acc); \ + SMU_SCALAR(SMU_MATTR(GFXCLK_LOCK_STATUS), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), gfxclk_lock_status); \ + SMU_SCALAR(SMU_MATTR(PCIE_LINK_WIDTH), SMU_MUNIT(NONE), \ + SMU_MTYPE(U16), pcie_link_width); \ + SMU_SCALAR(SMU_MATTR(PCIE_LINK_SPEED), SMU_MUNIT(SPEED_2), \ + SMU_MTYPE(U16), pcie_link_speed); \ + SMU_SCALAR(SMU_MATTR(XGMI_LINK_WIDTH), SMU_MUNIT(NONE), \ + SMU_MTYPE(U16), xgmi_link_width); \ + SMU_SCALAR(SMU_MATTR(XGMI_LINK_SPEED), SMU_MUNIT(SPEED_1), \ + SMU_MTYPE(U16), xgmi_link_speed); \ + SMU_SCALAR(SMU_MATTR(GFX_ACTIVITY_ACC), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U32), gfx_activity_acc); \ + SMU_SCALAR(SMU_MATTR(MEM_ACTIVITY_ACC), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U32), mem_activity_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_BANDWIDTH_ACC), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U64), pcie_bandwidth_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_BANDWIDTH_INST), SMU_MUNIT(BW_1), \ + SMU_MTYPE(U64), pcie_bandwidth_inst); \ + SMU_SCALAR(SMU_MATTR(PCIE_L0_TO_RECOV_COUNT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), pcie_l0_to_recov_count_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_REPLAY_COUNT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), pcie_replay_count_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_REPLAY_ROVER_COUNT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), pcie_replay_rover_count_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_NAK_SENT_COUNT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), pcie_nak_sent_count_acc); \ + SMU_SCALAR(SMU_MATTR(PCIE_NAK_RCVD_COUNT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U32), pcie_nak_rcvd_count_acc); \ + SMU_ARRAY(SMU_MATTR(XGMI_READ_DATA_ACC), SMU_MUNIT(DATA_1), \ + SMU_MTYPE(U64), xgmi_read_data_acc, \ + SMU_13_0_6_NUM_XGMI_LINKS); \ + SMU_ARRAY(SMU_MATTR(XGMI_WRITE_DATA_ACC), SMU_MUNIT(DATA_1), \ + SMU_MTYPE(U64), xgmi_write_data_acc, \ + SMU_13_0_6_NUM_XGMI_LINKS); \ + SMU_ARRAY(SMU_MATTR(XGMI_LINK_STATUS), SMU_MUNIT(NONE), \ + SMU_MTYPE(U16), xgmi_link_status, \ + SMU_13_0_6_NUM_XGMI_LINKS); \ + SMU_SCALAR(SMU_MATTR(FIRMWARE_TIMESTAMP), SMU_MUNIT(TIME_2), \ + SMU_MTYPE(U64), firmware_timestamp); \ + SMU_ARRAY(SMU_MATTR(CURRENT_GFXCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_gfxclk, SMU_13_0_6_MAX_GFX_CLKS); \ + SMU_ARRAY(SMU_MATTR(CURRENT_SOCCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_socclk, SMU_13_0_6_MAX_CLKS); \ + SMU_ARRAY(SMU_MATTR(CURRENT_VCLK0), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_vclk0, SMU_13_0_6_MAX_CLKS); \ + SMU_ARRAY(SMU_MATTR(CURRENT_DCLK0), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_dclk0, SMU_13_0_6_MAX_CLKS); \ + SMU_SCALAR(SMU_MATTR(CURRENT_UCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_uclk); \ + SMU_SCALAR(SMU_MATTR(PCIE_LC_PERF_OTHER_END_RECOVERY), \ + SMU_MUNIT(NONE), SMU_MTYPE(U32), \ + pcie_lc_perf_other_end_recovery); \ + SMU_ARRAY(SMU_MATTR(GFX_BUSY_INST), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U32), gfx_busy_inst, SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(JPEG_BUSY), SMU_MUNIT(PERCENT), SMU_MTYPE(U16), \ + jpeg_busy, SMU_13_0_6_MAX_JPEG); \ + SMU_ARRAY(SMU_MATTR(VCN_BUSY), SMU_MUNIT(PERCENT), SMU_MTYPE(U16), \ + vcn_busy, SMU_13_0_6_MAX_VCN); \ + SMU_ARRAY(SMU_MATTR(GFX_BUSY_ACC), SMU_MUNIT(PERCENT), SMU_MTYPE(U64), \ + gfx_busy_acc, SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_PPT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_ppt_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_THM_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_thm_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_LOW_UTILIZATION_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_low_utilization_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_TOTAL_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_total_acc, \ + SMU_13_0_6_MAX_XCC); + +DECLARE_SMU_METRICS_CLASS(smu_v13_0_6_gpu_metrics, SMU_13_0_6_METRICS_FIELDS); +void smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, + void *smu_metrics, + struct smu_v13_0_6_gpu_metrics *gpu_metrics); + +#define SMU_13_0_6_PARTITION_METRICS_FIELDS(SMU_SCALAR, SMU_ARRAY) \ + SMU_ARRAY(SMU_MATTR(CURRENT_GFXCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_gfxclk, SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(CURRENT_SOCCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_socclk, SMU_13_0_6_MAX_CLKS); \ + SMU_ARRAY(SMU_MATTR(CURRENT_VCLK0), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_vclk0, SMU_13_0_6_MAX_CLKS); \ + SMU_ARRAY(SMU_MATTR(CURRENT_DCLK0), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_dclk0, SMU_13_0_6_MAX_CLKS); \ + SMU_SCALAR(SMU_MATTR(CURRENT_UCLK), SMU_MUNIT(CLOCK_1), \ + SMU_MTYPE(U16), current_uclk); \ + SMU_ARRAY(SMU_MATTR(GFX_BUSY_INST), SMU_MUNIT(PERCENT), \ + SMU_MTYPE(U32), gfx_busy_inst, SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(JPEG_BUSY), SMU_MUNIT(PERCENT), SMU_MTYPE(U16), \ + jpeg_busy, SMU_13_0_6_MAX_JPEG); \ + SMU_ARRAY(SMU_MATTR(VCN_BUSY), SMU_MUNIT(PERCENT), SMU_MTYPE(U16), \ + vcn_busy, SMU_13_0_6_MAX_VCN); \ + SMU_ARRAY(SMU_MATTR(GFX_BUSY_ACC), SMU_MUNIT(PERCENT), SMU_MTYPE(U64), \ + gfx_busy_acc, SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_PPT_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_ppt_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_THM_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_thm_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_LOW_UTILIZATION_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_low_utilization_acc, \ + SMU_13_0_6_MAX_XCC); \ + SMU_ARRAY(SMU_MATTR(GFX_BELOW_HOST_LIMIT_TOTAL_ACC), SMU_MUNIT(NONE), \ + SMU_MTYPE(U64), gfx_below_host_limit_total_acc, \ + SMU_13_0_6_MAX_XCC); + +DECLARE_SMU_METRICS_CLASS(smu_v13_0_6_partition_metrics, + SMU_13_0_6_PARTITION_METRICS_FIELDS); + +#endif /* SWSMU_CODE_LAYER_L2 */ + #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c index c96fa5e49ed6..a3fc35b9011e 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_7_ppt.c @@ -1184,15 +1184,16 @@ static int smu_v13_0_7_print_clk_levels(struct smu_context *smu, struct smu_13_0_dpm_table *single_dpm_table; struct smu_13_0_pcie_table *pcie_table; uint32_t gen_speed, lane_width; - int i, curr_freq, size = 0; + int i, curr_freq, size = 0, start_offset = 0; int32_t min_value, max_value; int ret = 0; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; if (amdgpu_ras_intr_triggered()) { size += sysfs_emit_at(buf, size, "unavailable\n"); - return size; + return size - start_offset; } switch (clk_type) { @@ -1523,7 +1524,7 @@ static int smu_v13_0_7_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int smu_v13_0_7_od_restore_table_single(struct smu_context *smu, long input) diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c index 73b4506ef5a8..5d7e671fa3c3 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/yellow_carp_ppt.c @@ -1041,12 +1041,13 @@ static uint32_t yellow_carp_get_umd_pstate_clk_default(struct smu_context *smu, static int yellow_carp_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int i, idx, size = 0, ret = 0; + int i, idx, size = 0, ret = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t min, max; uint32_t clk_limit = 0; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -1111,7 +1112,7 @@ static int yellow_carp_print_clk_levels(struct smu_context *smu, } print_clk_out: - return size; + return size - start_offset; } static int yellow_carp_force_clk_levels(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c index fe00c84b1cc6..b1bd946d8e30 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_0_ppt.c @@ -1132,11 +1132,12 @@ static int smu_v14_0_common_get_dpm_level_count(struct smu_context *smu, static int smu_v14_0_0_print_clk_levels(struct smu_context *smu, enum smu_clk_type clk_type, char *buf) { - int i, idx, ret = 0, size = 0; + int i, idx, ret = 0, size = 0, start_offset = 0; uint32_t cur_value = 0, value = 0, count = 0; uint32_t min, max; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; switch (clk_type) { case SMU_OD_SCLK: @@ -1202,7 +1203,7 @@ static int smu_v14_0_0_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int smu_v14_0_0_set_soft_freq_limited_range(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c index 086501cc5213..2cea688c604f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu14/smu_v14_0_2_ppt.c @@ -1056,15 +1056,16 @@ static int smu_v14_0_2_print_clk_levels(struct smu_context *smu, struct smu_14_0_dpm_table *single_dpm_table; struct smu_14_0_pcie_table *pcie_table; uint32_t gen_speed, lane_width; - int i, curr_freq, size = 0; + int i, curr_freq, size = 0, start_offset = 0; int32_t min_value, max_value; int ret = 0; smu_cmn_get_sysfs_buf(&buf, &size); + start_offset = size; if (amdgpu_ras_intr_triggered()) { size += sysfs_emit_at(buf, size, "unavailable\n"); - return size; + return size - start_offset; } switch (clk_type) { @@ -1374,7 +1375,7 @@ static int smu_v14_0_2_print_clk_levels(struct smu_context *smu, break; } - return size; + return size - start_offset; } static int smu_v14_0_2_force_clk_levels(struct smu_context *smu, diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index f532f7c69259..4040ff926544 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -164,9 +164,13 @@ static void __smu_cmn_reg_print_error(struct smu_context *smu, msg_index, param, message); break; case SMU_RESP_BUSY_OTHER: - dev_err_ratelimited(adev->dev, - "SMU: I'm very busy for your command: index:%d param:0x%08X message:%s", - msg_index, param, message); + /* It is normal for SMU_MSG_GetBadPageCount to return busy + * so don't print error at this case. + */ + if (msg != SMU_MSG_GetBadPageCount) + dev_err_ratelimited(adev->dev, + "SMU: I'm very busy for your command: index:%d param:0x%08X message:%s", + msg_index, param, message); break; case SMU_RESP_DEBUG_END: dev_err_ratelimited(adev->dev, @@ -969,7 +973,7 @@ int smu_cmn_update_table(struct smu_context *smu, table_index); uint32_t table_size; int ret = 0; - if (!table_data || table_id >= SMU_TABLE_COUNT || table_id < 0) + if (!table_data || table_index >= SMU_TABLE_COUNT || table_id < 0) return -EINVAL; table_size = smu_table->tables[table_index].size; @@ -980,7 +984,7 @@ int smu_cmn_update_table(struct smu_context *smu, * Flush hdp cache: to guard the content seen by * GPU is consitent with CPU. */ - amdgpu_asic_flush_hdp(adev, NULL); + amdgpu_hdp_flush(adev, NULL); } ret = smu_cmn_send_smc_msg_with_param(smu, drv2smu ? @@ -992,7 +996,7 @@ int smu_cmn_update_table(struct smu_context *smu, return ret; if (!drv2smu) { - amdgpu_asic_invalidate_hdp(adev, NULL); + amdgpu_hdp_invalidate(adev, NULL); memcpy(table_data, table->cpu_addr, table_size); } diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h index 0ae91c8b6d72..8d7c4814c68f 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h @@ -202,5 +202,72 @@ void smu_cmn_get_backend_workload_mask(struct smu_context *smu, u32 workload_mask, u32 *backend_workload_mask); +/*SMU gpu metrics */ + +/* Attribute ID mapping */ +#define SMU_MATTR(X) AMDGPU_METRICS_ATTR_ID_##X +/* Type ID mapping */ +#define SMU_MTYPE(X) AMDGPU_METRICS_TYPE_##X +/* Unit ID mapping */ +#define SMU_MUNIT(X) AMDGPU_METRICS_UNIT_##X + +/* Map TYPEID to C type */ +#define SMU_CTYPE(TYPEID) SMU_CTYPE_##TYPEID + +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_U8 u8 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_S8 s8 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_U16 u16 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_S16 s16 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_U32 u32 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_S32 s32 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_U64 u64 +#define SMU_CTYPE_AMDGPU_METRICS_TYPE_S64 s64 + +/* struct members */ +#define SMU_METRICS_SCALAR(ID, UNIT, TYPEID, NAME) \ + u64 NAME##_ftype; \ + SMU_CTYPE(TYPEID) NAME + +#define SMU_METRICS_ARRAY(ID, UNIT, TYPEID, NAME, SIZE) \ + u64 NAME##_ftype; \ + SMU_CTYPE(TYPEID) NAME[SIZE] + +/* Init functions for scalar/array fields - init to 0xFFs */ +#define SMU_METRICS_INIT_SCALAR(ID, UNIT, TYPEID, NAME) \ + do { \ + obj->NAME##_ftype = \ + AMDGPU_METRICS_ENC_ATTR(UNIT, TYPEID, ID, 1); \ + obj->NAME = (SMU_CTYPE(TYPEID)) ~0; \ + count++; \ + } while (0) + +#define SMU_METRICS_INIT_ARRAY(ID, UNIT, TYPEID, NAME, SIZE) \ + do { \ + obj->NAME##_ftype = \ + AMDGPU_METRICS_ENC_ATTR(UNIT, TYPEID, ID, SIZE); \ + memset(obj->NAME, 0xFF, sizeof(obj->NAME)); \ + count++; \ + } while (0) + +/* Declare Metrics Class and Template object */ +#define DECLARE_SMU_METRICS_CLASS(CLASSNAME, SMU_METRICS_FIELD_LIST) \ + struct __packed CLASSNAME { \ + struct metrics_table_header header; \ + int attr_count; \ + SMU_METRICS_FIELD_LIST(SMU_METRICS_SCALAR, SMU_METRICS_ARRAY); \ + }; \ + static inline void CLASSNAME##_init(struct CLASSNAME *obj, \ + uint8_t frev, uint8_t crev) \ + { \ + int count = 0; \ + memset(obj, 0xFF, sizeof(*obj)); \ + obj->header.format_revision = frev; \ + obj->header.content_revision = crev; \ + obj->header.structure_size = sizeof(*obj); \ + SMU_METRICS_FIELD_LIST(SMU_METRICS_INIT_SCALAR, \ + SMU_METRICS_INIT_ARRAY) \ + obj->attr_count = count; \ + } + #endif #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_internal.h b/drivers/gpu/drm/amd/pm/swsmu/smu_internal.h index c09ecf1a68a0..34f6b4b1c3ba 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_internal.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_internal.h @@ -100,6 +100,7 @@ #define smu_is_asic_wbrf_supported(smu) smu_ppt_funcs(is_asic_wbrf_supported, false, smu) #define smu_enable_uclk_shadow(smu, enable) smu_ppt_funcs(enable_uclk_shadow, 0, smu, enable) #define smu_set_wbrf_exclusion_ranges(smu, freq_band_range) smu_ppt_funcs(set_wbrf_exclusion_ranges, -EOPNOTSUPP, smu, freq_band_range) +#define smu_get_ras_smu_drv(smu, ras_smu_drv) smu_ppt_funcs(get_ras_smu_drv, -EOPNOTSUPP, smu, ras_smu_drv) #endif #endif diff --git a/drivers/gpu/drm/amd/ras/Makefile b/drivers/gpu/drm/amd/ras/Makefile new file mode 100644 index 000000000000..bbdaba811d34 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/Makefile @@ -0,0 +1,34 @@ +# +# Copyright (c) 2025 Advanced Micro Devices, Inc. All rights reserved. +# +# Permission is hereby granted, free of charge, to any person obtaining a copy +# of this software and associated documentation files (the "Software"), to deal +# in the Software without restriction, including without limitation the rights +# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +# copies of the Software, and to permit persons to whom the Software is +# furnished to do so, subject to the following conditions: +# +# The above copyright notice and this permission notice shall be included in +# all copies or substantial portions of the Software. +# +# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +# THE SOFTWARE. + +ifeq ($(AMD_GPU_RAS_MGR),) + AMD_GPU_RAS_MGR := ras_mgr +endif + +subdir-ccflags-y += -I$(AMD_GPU_RAS_FULL_PATH)/rascore +subdir-ccflags-y += -I$(AMD_GPU_RAS_FULL_PATH)/$(AMD_GPU_RAS_MGR) + +RAS_LIBS = $(AMD_GPU_RAS_MGR) rascore + +AMD_RAS = $(addsuffix /Makefile, $(addprefix $(AMD_GPU_RAS_FULL_PATH)/,$(RAS_LIBS))) + +include $(AMD_RAS) + diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/Makefile b/drivers/gpu/drm/amd/ras/ras_mgr/Makefile new file mode 100644 index 000000000000..5e5a2cfa4068 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/Makefile @@ -0,0 +1,33 @@ +# Copyright 2025 Advanced Micro Devices, Inc. +# +# Permission is hereby granted, free of charge, to any person obtaining a +# copy of this software and associated documentation files (the "Software"), +# to deal in the Software without restriction, including without limitation +# the rights to use, copy, modify, merge, publish, distribute, sublicense, +# and/or sell copies of the Software, and to permit persons to whom the +# Software is furnished to do so, subject to the following conditions: +# +# The above copyright notice and this permission notice shall be included in +# all copies or substantial portions of the Software. +# +# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR +# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, +# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR +# OTHER DEALINGS IN THE SOFTWARE. + +RAS_MGR_FILES = amdgpu_ras_sys.o \ + amdgpu_ras_mgr.o \ + amdgpu_ras_eeprom_i2c.o \ + amdgpu_ras_mp1_v13_0.o \ + amdgpu_ras_cmd.o \ + amdgpu_ras_process.o \ + amdgpu_ras_nbio_v7_9.o + + +RAS_MGR = $(addprefix $(AMD_GPU_RAS_PATH)/ras_mgr/, $(RAS_MGR_FILES)) + +AMD_GPU_RAS_FILES += $(RAS_MGR) + diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.c new file mode 100644 index 000000000000..78419b7f7729 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.c @@ -0,0 +1,285 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include <linux/pci.h> +#include "amdgpu.h" +#include "amdgpu_ras.h" +#include "ras_sys.h" +#include "amdgpu_ras_cmd.h" +#include "amdgpu_ras_mgr.h" + +/* inject address is 52 bits */ +#define RAS_UMC_INJECT_ADDR_LIMIT (0x1ULL << 52) + +#define AMDGPU_RAS_TYPE_RASCORE 0x1 +#define AMDGPU_RAS_TYPE_AMDGPU 0x2 +#define AMDGPU_RAS_TYPE_VF 0x3 + +static int amdgpu_ras_trigger_error_prepare(struct ras_core_context *ras_core, + struct ras_cmd_inject_error_req *block_info) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int ret; + + if (block_info->block_id == TA_RAS_BLOCK__XGMI_WAFL) { + if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_DISALLOW)) + RAS_DEV_WARN(adev, "Failed to disallow df cstate"); + + ret = amdgpu_dpm_set_pm_policy(adev, PP_PM_POLICY_XGMI_PLPD, XGMI_PLPD_DISALLOW); + if (ret && (ret != -EOPNOTSUPP)) + RAS_DEV_WARN(adev, "Failed to disallow XGMI power down"); + } + + return 0; +} + +static int amdgpu_ras_trigger_error_end(struct ras_core_context *ras_core, + struct ras_cmd_inject_error_req *block_info) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int ret; + + if (block_info->block_id == TA_RAS_BLOCK__XGMI_WAFL) { + if (amdgpu_ras_intr_triggered()) + return 0; + + ret = amdgpu_dpm_set_pm_policy(adev, PP_PM_POLICY_XGMI_PLPD, XGMI_PLPD_DEFAULT); + if (ret && (ret != -EOPNOTSUPP)) + RAS_DEV_WARN(adev, "Failed to allow XGMI power down"); + + if (amdgpu_dpm_set_df_cstate(adev, DF_CSTATE_ALLOW)) + RAS_DEV_WARN(adev, "Failed to allow df cstate"); + } + + return 0; +} + +static uint64_t local_addr_to_xgmi_global_addr(struct ras_core_context *ras_core, + uint64_t addr) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct amdgpu_xgmi *xgmi = &adev->gmc.xgmi; + + return (addr + xgmi->physical_node_id * xgmi->node_segment_size); +} + +static int amdgpu_ras_inject_error(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct ras_cmd_inject_error_req *req = + (struct ras_cmd_inject_error_req *)cmd->input_buff_raw; + int ret = RAS_CMD__ERROR_GENERIC; + + if (req->block_id == RAS_BLOCK_ID__UMC) { + if (amdgpu_ras_mgr_check_retired_addr(adev, req->address)) { + RAS_DEV_WARN(ras_core->dev, + "RAS WARN: inject: 0x%llx has already been marked as bad!\n", + req->address); + return RAS_CMD__ERROR_ACCESS_DENIED; + } + + if ((req->address >= adev->gmc.mc_vram_size && + adev->gmc.mc_vram_size) || + (req->address >= RAS_UMC_INJECT_ADDR_LIMIT)) { + RAS_DEV_WARN(adev, "RAS WARN: input address 0x%llx is invalid.", + req->address); + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + } + + /* Calculate XGMI relative offset */ + if (adev->gmc.xgmi.num_physical_nodes > 1 && + req->block_id != RAS_BLOCK_ID__GFX) { + req->address = local_addr_to_xgmi_global_addr(ras_core, req->address); + } + } + + amdgpu_ras_trigger_error_prepare(ras_core, req); + ret = rascore_handle_cmd(ras_core, cmd, data); + amdgpu_ras_trigger_error_end(ras_core, req); + if (ret) { + RAS_DEV_ERR(adev, "ras inject block %u failed %d\n", req->block_id, ret); + ret = RAS_CMD__ERROR_ACCESS_DENIED; + } + + + return ret; +} + +static int amdgpu_ras_get_ras_safe_fb_addr_ranges(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct ras_cmd_dev_handle *input_data = + (struct ras_cmd_dev_handle *)cmd->input_buff_raw; + struct ras_cmd_ras_safe_fb_address_ranges_rsp *ranges = + (struct ras_cmd_ras_safe_fb_address_ranges_rsp *)cmd->output_buff_raw; + struct amdgpu_mem_partition_info *mem_ranges; + uint32_t i = 0; + + if (cmd->input_size != sizeof(*input_data)) + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + + mem_ranges = adev->gmc.mem_partitions; + for (i = 0; i < adev->gmc.num_mem_partitions; i++) { + ranges->range[i].start = mem_ranges[i].range.fpfn << AMDGPU_GPU_PAGE_SHIFT; + ranges->range[i].size = mem_ranges[i].size; + ranges->range[i].idx = i; + } + + ranges->num_ranges = adev->gmc.num_mem_partitions; + + ranges->version = 0; + cmd->output_size = sizeof(struct ras_cmd_ras_safe_fb_address_ranges_rsp); + + return RAS_CMD__SUCCESS; +} + +static int ras_translate_fb_address(struct ras_core_context *ras_core, + enum ras_fb_addr_type src_type, + enum ras_fb_addr_type dest_type, + union ras_translate_fb_address *src_addr, + union ras_translate_fb_address *dest_addr) +{ + uint64_t soc_phy_addr; + int ret = RAS_CMD__SUCCESS; + + /* Does not need to be queued as event as this is a SW translation */ + switch (src_type) { + case RAS_FB_ADDR_SOC_PHY: + soc_phy_addr = src_addr->soc_phy_addr; + break; + case RAS_FB_ADDR_BANK: + ret = ras_cmd_translate_bank_to_soc_pa(ras_core, + src_addr->bank_addr, &soc_phy_addr); + if (ret) + return RAS_CMD__ERROR_GENERIC; + break; + default: + return RAS_CMD__ERROR_INVALID_CMD; + } + + switch (dest_type) { + case RAS_FB_ADDR_SOC_PHY: + dest_addr->soc_phy_addr = soc_phy_addr; + break; + case RAS_FB_ADDR_BANK: + ret = ras_cmd_translate_soc_pa_to_bank(ras_core, + soc_phy_addr, &dest_addr->bank_addr); + if (ret) + return RAS_CMD__ERROR_GENERIC; + break; + default: + return RAS_CMD__ERROR_INVALID_CMD; + } + + return ret; +} + +static int amdgpu_ras_translate_fb_address(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_translate_fb_address_req *req_buff = + (struct ras_cmd_translate_fb_address_req *)cmd->input_buff_raw; + struct ras_cmd_translate_fb_address_rsp *rsp_buff = + (struct ras_cmd_translate_fb_address_rsp *)cmd->output_buff_raw; + int ret = RAS_CMD__ERROR_GENERIC; + + if (cmd->input_size != sizeof(struct ras_cmd_translate_fb_address_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + if ((req_buff->src_addr_type >= RAS_FB_ADDR_UNKNOWN) || + (req_buff->dest_addr_type >= RAS_FB_ADDR_UNKNOWN)) + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + + ret = ras_translate_fb_address(ras_core, req_buff->src_addr_type, + req_buff->dest_addr_type, &req_buff->trans_addr, &rsp_buff->trans_addr); + if (ret) + return RAS_CMD__ERROR_GENERIC; + + rsp_buff->version = 0; + cmd->output_size = sizeof(struct ras_cmd_translate_fb_address_rsp); + + return RAS_CMD__SUCCESS; +} + +static struct ras_cmd_func_map amdgpu_ras_cmd_maps[] = { + {RAS_CMD__INJECT_ERROR, amdgpu_ras_inject_error}, + {RAS_CMD__GET_SAFE_FB_ADDRESS_RANGES, amdgpu_ras_get_ras_safe_fb_addr_ranges}, + {RAS_CMD__TRANSLATE_FB_ADDRESS, amdgpu_ras_translate_fb_address}, +}; + +int amdgpu_ras_handle_cmd(struct ras_core_context *ras_core, struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_func_map *ras_cmd = NULL; + int i, res; + + for (i = 0; i < ARRAY_SIZE(amdgpu_ras_cmd_maps); i++) { + if (cmd->cmd_id == amdgpu_ras_cmd_maps[i].cmd_id) { + ras_cmd = &amdgpu_ras_cmd_maps[i]; + break; + } + } + + if (ras_cmd) + res = ras_cmd->func(ras_core, cmd, NULL); + else + res = RAS_CMD__ERROR_UKNOWN_CMD; + + return res; +} + +int amdgpu_ras_submit_cmd(struct ras_core_context *ras_core, struct ras_cmd_ctx *cmd) +{ + struct ras_core_context *cmd_core = ras_core; + int timeout = 60; + int res; + + cmd->cmd_res = RAS_CMD__ERROR_INVALID_CMD; + cmd->output_size = 0; + + if (!ras_core_is_enabled(cmd_core)) + return RAS_CMD__ERROR_ACCESS_DENIED; + + while (ras_core_gpu_in_reset(cmd_core)) { + msleep(1000); + if (!timeout--) + return RAS_CMD__ERROR_TIMEOUT; + } + + res = amdgpu_ras_handle_cmd(cmd_core, cmd, NULL); + if (res == RAS_CMD__ERROR_UKNOWN_CMD) + res = rascore_handle_cmd(cmd_core, cmd, NULL); + + cmd->cmd_res = res; + + if (cmd->output_size > cmd->output_buf_size) { + RAS_DEV_ERR(cmd_core->dev, + "Output size 0x%x exceeds output buffer size 0x%x!\n", + cmd->output_size, cmd->output_buf_size); + return RAS_CMD__SUCCESS_EXEED_BUFFER; + } + + return RAS_CMD__SUCCESS; +} diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.h new file mode 100644 index 000000000000..5973b156cc85 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_cmd.h @@ -0,0 +1,54 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __AMDGPU_RAS_CMD_H__ +#define __AMDGPU_RAS_CMD_H__ +#include "ras.h" + +enum amdgpu_ras_cmd_id { + RAS_CMD__AMDGPU_BEGIN = RAS_CMD_ID_AMDGPU_START, + RAS_CMD__TRANSLATE_MEMORY_FD, + RAS_CMD__AMDGPU_SUPPORTED_MAX = RAS_CMD_ID_AMDGPU_END, +}; + +struct ras_cmd_translate_memory_fd_req { + struct ras_cmd_dev_handle dev; + uint32_t type; + uint32_t fd; + uint64_t address; + uint32_t reserved[4]; +}; + +struct ras_cmd_translate_memory_fd_rsp { + uint32_t version; + uint32_t padding; + uint64_t start; + uint64_t size; + uint32_t reserved[2]; +}; + +int amdgpu_ras_handle_cmd(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data); +int amdgpu_ras_submit_cmd(struct ras_core_context *ras_core, struct ras_cmd_ctx *cmd); + +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.c new file mode 100644 index 000000000000..3ed3ff42b7e1 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.c @@ -0,0 +1,182 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright (c) 2025 Advanced Micro Devices, Inc. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "amdgpu.h" +#include "amdgpu_atomfirmware.h" +#include "amdgpu_ras_eeprom.h" +#include "amdgpu_ras_mgr.h" +#include "amdgpu_ras_eeprom_i2c.h" +#include "ras_eeprom.h" + +/* These are memory addresses as would be seen by one or more EEPROM + * chips strung on the I2C bus, usually by manipulating pins 1-3 of a + * set of EEPROM devices. They form a continuous memory space. + * + * The I2C device address includes the device type identifier, 1010b, + * which is a reserved value and indicates that this is an I2C EEPROM + * device. It also includes the top 3 bits of the 19 bit EEPROM memory + * address, namely bits 18, 17, and 16. This makes up the 7 bit + * address sent on the I2C bus with bit 0 being the direction bit, + * which is not represented here, and sent by the hardware directly. + * + * For instance, + * 50h = 1010000b => device type identifier 1010b, bits 18:16 = 000b, address 0. + * 54h = 1010100b => --"--, bits 18:16 = 100b, address 40000h. + * 56h = 1010110b => --"--, bits 18:16 = 110b, address 60000h. + * Depending on the size of the I2C EEPROM device(s), bits 18:16 may + * address memory in a device or a device on the I2C bus, depending on + * the status of pins 1-3. See top of amdgpu_eeprom.c. + * + * The RAS table lives either at address 0 or address 40000h of EEPROM. + */ +#define EEPROM_I2C_MADDR_0 0x0 +#define EEPROM_I2C_MADDR_4 0x40000 + +#define MAKE_I2C_ADDR(_aa) ((0xA << 3) | (((_aa) >> 16) & 0xF)) +#define to_amdgpu_ras(x) (container_of(x, struct amdgpu_ras, eeprom_control)) + +#define EEPROM_PAGE_BITS 8 +#define EEPROM_PAGE_SIZE (1U << EEPROM_PAGE_BITS) +#define EEPROM_PAGE_MASK (EEPROM_PAGE_SIZE - 1) + +#define EEPROM_OFFSET_SIZE 2 + +static int ras_eeprom_i2c_config(struct ras_core_context *ras_core) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + u8 i2c_addr; + + if (amdgpu_atomfirmware_ras_rom_addr(adev, &i2c_addr)) { + /* The address given by VBIOS is an 8-bit, wire-format + * address, i.e. the most significant byte. + * + * Normalize it to a 19-bit EEPROM address. Remove the + * device type identifier and make it a 7-bit address; + * then make it a 19-bit EEPROM address. See top of + * amdgpu_eeprom.c. + */ + i2c_addr = (i2c_addr & 0x0F) >> 1; + control->i2c_address = ((u32) i2c_addr) << 16; + return 0; + } + + switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { + case IP_VERSION(13, 0, 5): + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 10): + case IP_VERSION(13, 0, 12): + case IP_VERSION(13, 0, 14): + control->i2c_address = EEPROM_I2C_MADDR_4; + return 0; + default: + return -ENODATA; + } + return -ENODATA; +} + +static int ras_eeprom_i2c_xfer(struct ras_core_context *ras_core, u32 eeprom_addr, + u8 *eeprom_buf, u32 buf_size, bool read) +{ + struct i2c_adapter *i2c_adap = ras_core->ras_eeprom.i2c_adapter; + u8 eeprom_offset_buf[EEPROM_OFFSET_SIZE]; + struct i2c_msg msgs[] = { + { + .flags = 0, + .len = EEPROM_OFFSET_SIZE, + .buf = eeprom_offset_buf, + }, + { + .flags = read ? I2C_M_RD : 0, + }, + }; + const u8 *p = eeprom_buf; + int r; + u16 len; + + for (r = 0; buf_size > 0; + buf_size -= len, eeprom_addr += len, eeprom_buf += len) { + /* Set the EEPROM address we want to write to/read from. + */ + msgs[0].addr = MAKE_I2C_ADDR(eeprom_addr); + msgs[1].addr = msgs[0].addr; + msgs[0].buf[0] = (eeprom_addr >> 8) & 0xff; + msgs[0].buf[1] = eeprom_addr & 0xff; + + if (!read) { + /* Write the maximum amount of data, without + * crossing the device's page boundary, as per + * its spec. Partial page writes are allowed, + * starting at any location within the page, + * so long as the page boundary isn't crossed + * over (actually the page pointer rolls + * over). + * + * As per the AT24CM02 EEPROM spec, after + * writing into a page, the I2C driver should + * terminate the transfer, i.e. in + * "i2c_transfer()" below, with a STOP + * condition, so that the self-timed write + * cycle begins. This is implied for the + * "i2c_transfer()" abstraction. + */ + len = min(EEPROM_PAGE_SIZE - (eeprom_addr & EEPROM_PAGE_MASK), + buf_size); + } else { + /* Reading from the EEPROM has no limitation + * on the number of bytes read from the EEPROM + * device--they are simply sequenced out. + * Keep in mind that i2c_msg.len is u16 type. + */ + len = min(U16_MAX, buf_size); + } + msgs[1].len = len; + msgs[1].buf = eeprom_buf; + + + /* This constitutes a START-STOP transaction. + */ + r = i2c_transfer(i2c_adap, msgs, ARRAY_SIZE(msgs)); + if (r != ARRAY_SIZE(msgs)) + break; + + if (!read) { + /* According to EEPROM specs the length of the + * self-writing cycle, tWR (tW), is 10 ms. + * + * TODO: Use polling on ACK, aka Acknowledge + * Polling, to minimize waiting for the + * internal write cycle to complete, as it is + * usually smaller than tWR (tW). + */ + msleep(10); + } + } + + return r < 0 ? r : eeprom_buf - p; +} + +const struct ras_eeprom_sys_func amdgpu_ras_eeprom_i2c_sys_func = { + .eeprom_i2c_xfer = ras_eeprom_i2c_xfer, + .update_eeprom_i2c_config = ras_eeprom_i2c_config, +}; diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.h new file mode 100644 index 000000000000..3b5878605411 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.h @@ -0,0 +1,27 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright (C) 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included + * in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS + * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN + * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN + * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. + */ +#ifndef __AMDGPU_RAS_EEPROM_I2C_H__ +#define __AMDGPU_RAS_EEPROM_I2C_H__ +#include "ras.h" + +extern const struct ras_eeprom_sys_func amdgpu_ras_eeprom_i2c_sys_func; +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c new file mode 100644 index 000000000000..afe8135b6258 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c @@ -0,0 +1,648 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu.h" +#include "amdgpu_reset.h" +#include "amdgpu_xgmi.h" +#include "ras_sys.h" +#include "amdgpu_ras_mgr.h" +#include "amdgpu_ras_cmd.h" +#include "amdgpu_ras_process.h" +#include "amdgpu_ras_eeprom_i2c.h" +#include "amdgpu_ras_mp1_v13_0.h" +#include "amdgpu_ras_nbio_v7_9.h" + +#define MAX_SOCKET_NUM_PER_HIVE 8 +#define MAX_AID_NUM_PER_SOCKET 4 +#define MAX_XCD_NUM_PER_AID 2 + +/* typical ECC bad page rate is 1 bad page per 100MB VRAM */ +#define TYPICAL_ECC_BAD_PAGE_RATE (100ULL * SZ_1M) + +#define COUNT_BAD_PAGE_THRESHOLD(size) (((size) >> 21) << 4) + +/* Reserve 8 physical dram row for possible retirement. + * In worst cases, it will lose 8 * 2MB memory in vram domain + */ +#define RAS_RESERVED_VRAM_SIZE_DEFAULT (16ULL << 20) + + +static void ras_mgr_init_event_mgr(struct ras_event_manager *mgr) +{ + struct ras_event_state *event_state; + int i; + + memset(mgr, 0, sizeof(*mgr)); + atomic64_set(&mgr->seqno, 0); + + for (i = 0; i < ARRAY_SIZE(mgr->event_state); i++) { + event_state = &mgr->event_state[i]; + event_state->last_seqno = RAS_EVENT_INVALID_ID; + atomic64_set(&event_state->count, 0); + } +} + +static void amdgpu_ras_mgr_init_event_mgr(struct ras_core_context *ras_core) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + struct ras_event_manager *event_mgr; + struct amdgpu_hive_info *hive; + + hive = amdgpu_get_xgmi_hive(adev); + event_mgr = hive ? &hive->event_mgr : &ras_mgr->ras_event_mgr; + + /* init event manager with node 0 on xgmi system */ + if (!amdgpu_reset_in_recovery(adev)) { + if (!hive || adev->gmc.xgmi.node_id == 0) + ras_mgr_init_event_mgr(event_mgr); + } + + if (hive) + amdgpu_put_xgmi_hive(hive); +} + +static int amdgpu_ras_mgr_init_aca_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_aca_config *aca_cfg = &config->aca_cfg; + + aca_cfg->socket_num_per_hive = MAX_SOCKET_NUM_PER_HIVE; + aca_cfg->aid_num_per_socket = MAX_AID_NUM_PER_SOCKET; + aca_cfg->xcd_num_per_aid = MAX_XCD_NUM_PER_AID; + + return 0; +} + +static int amdgpu_ras_mgr_init_eeprom_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_eeprom_config *eeprom_cfg = &config->eeprom_cfg; + + eeprom_cfg->eeprom_sys_fn = &amdgpu_ras_eeprom_i2c_sys_func; + eeprom_cfg->eeprom_i2c_adapter = adev->pm.ras_eeprom_i2c_bus; + if (eeprom_cfg->eeprom_i2c_adapter) { + const struct i2c_adapter_quirks *quirks = + ((struct i2c_adapter *)eeprom_cfg->eeprom_i2c_adapter)->quirks; + + if (quirks) { + eeprom_cfg->max_i2c_read_len = quirks->max_read_len; + eeprom_cfg->max_i2c_write_len = quirks->max_write_len; + } + } + + /* + * amdgpu_bad_page_threshold is used to config + * the threshold for the number of bad pages. + * -1: Threshold is set to default value + * Driver will issue a warning message when threshold is reached + * and continue runtime services. + * 0: Disable bad page retirement + * Driver will not retire bad pages + * which is intended for debugging purpose. + * -2: Threshold is determined by a formula + * that assumes 1 bad page per 100M of local memory. + * Driver will continue runtime services when threhold is reached. + * 0 < threshold < max number of bad page records in EEPROM, + * A user-defined threshold is set + * Driver will halt runtime services when this custom threshold is reached. + */ + if (amdgpu_bad_page_threshold == NONSTOP_OVER_THRESHOLD) + eeprom_cfg->eeprom_record_threshold_count = + div64_u64(adev->gmc.mc_vram_size, TYPICAL_ECC_BAD_PAGE_RATE); + else if (amdgpu_bad_page_threshold == WARN_NONSTOP_OVER_THRESHOLD) + eeprom_cfg->eeprom_record_threshold_count = + COUNT_BAD_PAGE_THRESHOLD(RAS_RESERVED_VRAM_SIZE_DEFAULT); + else + eeprom_cfg->eeprom_record_threshold_count = amdgpu_bad_page_threshold; + + eeprom_cfg->eeprom_record_threshold_config = amdgpu_bad_page_threshold; + + return 0; +} + +static int amdgpu_ras_mgr_init_mp1_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_mp1_config *mp1_cfg = &config->mp1_cfg; + int ret = 0; + + switch (config->mp1_ip_version) { + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 14): + case IP_VERSION(13, 0, 12): + mp1_cfg->mp1_sys_fn = &amdgpu_ras_mp1_sys_func_v13_0; + break; + default: + RAS_DEV_ERR(adev, + "The mp1(0x%x) ras config is not right!\n", + config->mp1_ip_version); + ret = -EINVAL; + break; + } + + return ret; +} + +static int amdgpu_ras_mgr_init_nbio_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_nbio_config *nbio_cfg = &config->nbio_cfg; + int ret = 0; + + switch (config->nbio_ip_version) { + case IP_VERSION(7, 9, 0): + case IP_VERSION(7, 9, 1): + nbio_cfg->nbio_sys_fn = &amdgpu_ras_nbio_sys_func_v7_9; + break; + default: + RAS_DEV_ERR(adev, + "The nbio(0x%x) ras config is not right!\n", + config->nbio_ip_version); + ret = -EINVAL; + break; + } + + return ret; +} + +static int amdgpu_ras_mgr_get_ras_psp_system_status(struct ras_core_context *ras_core, + struct ras_psp_sys_status *status) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct ta_context *context = &adev->psp.ras_context.context; + + status->initialized = context->initialized; + status->session_id = context->session_id; + status->psp_cmd_mutex = &adev->psp.mutex; + + return 0; +} + +static int amdgpu_ras_mgr_get_ras_ta_init_param(struct ras_core_context *ras_core, + struct ras_ta_init_param *ras_ta_param) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + uint32_t nps_mode; + + if (amdgpu_ras_is_poison_mode_supported(adev)) + ras_ta_param->poison_mode_en = 1; + + if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) + ras_ta_param->dgpu_mode = 1; + + ras_ta_param->xcc_mask = adev->gfx.xcc_mask; + ras_ta_param->channel_dis_num = hweight32(adev->gmc.m_half_use) * 2; + + ras_ta_param->active_umc_mask = adev->umc.active_mask; + + if (!amdgpu_ras_mgr_get_curr_nps_mode(adev, &nps_mode)) + ras_ta_param->nps_mode = nps_mode; + + return 0; +} + +const struct ras_psp_sys_func amdgpu_ras_psp_sys_func = { + .get_ras_psp_system_status = amdgpu_ras_mgr_get_ras_psp_system_status, + .get_ras_ta_init_param = amdgpu_ras_mgr_get_ras_ta_init_param, +}; + +static int amdgpu_ras_mgr_init_psp_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_psp_config *psp_cfg = &config->psp_cfg; + + psp_cfg->psp_sys_fn = &amdgpu_ras_psp_sys_func; + + return 0; +} + +static int amdgpu_ras_mgr_init_umc_config(struct amdgpu_device *adev, + struct ras_core_config *config) +{ + struct ras_umc_config *umc_cfg = &config->umc_cfg; + + umc_cfg->umc_vram_type = adev->gmc.vram_type; + + return 0; +} + +static struct ras_core_context *amdgpu_ras_mgr_create_ras_core(struct amdgpu_device *adev) +{ + struct ras_core_config init_config; + + memset(&init_config, 0, sizeof(init_config)); + + init_config.umc_ip_version = amdgpu_ip_version(adev, UMC_HWIP, 0); + init_config.mp1_ip_version = amdgpu_ip_version(adev, MP1_HWIP, 0); + init_config.gfx_ip_version = amdgpu_ip_version(adev, GC_HWIP, 0); + init_config.nbio_ip_version = amdgpu_ip_version(adev, NBIO_HWIP, 0); + init_config.psp_ip_version = amdgpu_ip_version(adev, MP1_HWIP, 0); + + if (init_config.umc_ip_version == IP_VERSION(12, 0, 0) || + init_config.umc_ip_version == IP_VERSION(12, 5, 0)) + init_config.aca_ip_version = IP_VERSION(1, 0, 0); + + init_config.sys_fn = &amdgpu_ras_sys_fn; + init_config.ras_eeprom_supported = true; + init_config.poison_supported = + amdgpu_ras_is_poison_mode_supported(adev); + + amdgpu_ras_mgr_init_aca_config(adev, &init_config); + amdgpu_ras_mgr_init_eeprom_config(adev, &init_config); + amdgpu_ras_mgr_init_mp1_config(adev, &init_config); + amdgpu_ras_mgr_init_nbio_config(adev, &init_config); + amdgpu_ras_mgr_init_psp_config(adev, &init_config); + amdgpu_ras_mgr_init_umc_config(adev, &init_config); + + return ras_core_create(&init_config); +} + +static int amdgpu_ras_mgr_sw_init(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_ras_mgr *ras_mgr; + int ret = 0; + + /* Disabled by default */ + con->uniras_enabled = false; + + /* Enabled only in debug mode */ + if (adev->debug_enable_ras_aca) { + con->uniras_enabled = true; + RAS_DEV_INFO(adev, "Debug amdgpu uniras!"); + } + + if (!con->uniras_enabled) + return 0; + + ras_mgr = kzalloc(sizeof(*ras_mgr), GFP_KERNEL); + if (!ras_mgr) + return -EINVAL; + + con->ras_mgr = ras_mgr; + ras_mgr->adev = adev; + + ras_mgr->ras_core = amdgpu_ras_mgr_create_ras_core(adev); + if (!ras_mgr->ras_core) { + RAS_DEV_ERR(adev, "Failed to create ras core!\n"); + ret = -EINVAL; + goto err; + } + + ras_mgr->ras_core->dev = adev; + + amdgpu_ras_process_init(adev); + ras_core_sw_init(ras_mgr->ras_core); + amdgpu_ras_mgr_init_event_mgr(ras_mgr->ras_core); + return 0; + +err: + kfree(ras_mgr); + return ret; +} + +static int amdgpu_ras_mgr_sw_fini(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_ras_mgr *ras_mgr = (struct amdgpu_ras_mgr *)con->ras_mgr; + + if (!con->uniras_enabled) + return 0; + + if (!ras_mgr) + return 0; + + amdgpu_ras_process_fini(adev); + ras_core_sw_fini(ras_mgr->ras_core); + ras_core_destroy(ras_mgr->ras_core); + ras_mgr->ras_core = NULL; + + kfree(con->ras_mgr); + con->ras_mgr = NULL; + + return 0; +} + +static int amdgpu_ras_mgr_hw_init(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + int ret; + + if (!con->uniras_enabled) + return 0; + + if (!ras_mgr || !ras_mgr->ras_core) + return -EINVAL; + + ret = ras_core_hw_init(ras_mgr->ras_core); + if (ret) { + RAS_DEV_ERR(adev, "Failed to initialize ras core!\n"); + return ret; + } + + ras_mgr->ras_is_ready = true; + + amdgpu_enable_uniras(adev, true); + + RAS_DEV_INFO(adev, "AMDGPU RAS Is Ready.\n"); + return 0; +} + +static int amdgpu_ras_mgr_hw_fini(struct amdgpu_ip_block *ip_block) +{ + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!con->uniras_enabled) + return 0; + + if (!ras_mgr || !ras_mgr->ras_core) + return -EINVAL; + + ras_core_hw_fini(ras_mgr->ras_core); + + ras_mgr->ras_is_ready = false; + + return 0; +} + +struct amdgpu_ras_mgr *amdgpu_ras_mgr_get_context(struct amdgpu_device *adev) +{ + if (!adev || !adev->psp.ras_context.ras) + return NULL; + + return (struct amdgpu_ras_mgr *)adev->psp.ras_context.ras->ras_mgr; +} + +static const struct amd_ip_funcs __maybe_unused ras_v1_0_ip_funcs = { + .name = "ras_v1_0", + .sw_init = amdgpu_ras_mgr_sw_init, + .sw_fini = amdgpu_ras_mgr_sw_fini, + .hw_init = amdgpu_ras_mgr_hw_init, + .hw_fini = amdgpu_ras_mgr_hw_fini, +}; + +const struct amdgpu_ip_block_version ras_v1_0_ip_block = { + .type = AMD_IP_BLOCK_TYPE_RAS, + .major = 1, + .minor = 0, + .rev = 0, + .funcs = &ras_v1_0_ip_funcs, +}; + +int amdgpu_enable_uniras(struct amdgpu_device *adev, bool enable) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!ras_mgr || !ras_mgr->ras_core) + return -EPERM; + + if (amdgpu_sriov_vf(adev)) + return -EPERM; + + RAS_DEV_INFO(adev, "Enable amdgpu unified ras!"); + return ras_core_set_status(ras_mgr->ras_core, enable); +} + +bool amdgpu_uniras_enabled(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!ras_mgr || !ras_mgr->ras_core) + return false; + + if (amdgpu_sriov_vf(adev)) + return false; + + return ras_core_is_enabled(ras_mgr->ras_core); +} + +static bool amdgpu_ras_mgr_is_ready(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (ras_mgr && ras_mgr->ras_core && ras_mgr->ras_is_ready && + ras_core_is_ready(ras_mgr->ras_core)) + return true; + + return false; +} + +int amdgpu_ras_mgr_handle_fatal_interrupt(struct amdgpu_device *adev, void *data) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + return ras_core_handle_nbio_irq(ras_mgr->ras_core, data); +} + +uint64_t amdgpu_ras_mgr_gen_ras_event_seqno(struct amdgpu_device *adev, + enum ras_seqno_type seqno_type) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + int ret; + uint64_t seq_no; + + if (!amdgpu_ras_mgr_is_ready(adev) || + (seqno_type >= RAS_SEQNO_TYPE_COUNT_MAX)) + return 0; + + seq_no = ras_core_gen_seqno(ras_mgr->ras_core, seqno_type); + + if ((seqno_type == RAS_SEQNO_TYPE_DE) || + (seqno_type == RAS_SEQNO_TYPE_POISON_CONSUMPTION)) { + ret = ras_core_put_seqno(ras_mgr->ras_core, seqno_type, seq_no); + if (ret) + RAS_DEV_WARN(adev, "There are too many ras interrupts!"); + } + + return seq_no; +} + +int amdgpu_ras_mgr_handle_controller_interrupt(struct amdgpu_device *adev, void *data) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + struct ras_ih_info *ih_info = (struct ras_ih_info *)data; + uint64_t seq_no = 0; + int ret = 0; + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + if (ih_info && (ih_info->block == AMDGPU_RAS_BLOCK__UMC)) { + if (ras_mgr->ras_core->poison_supported) { + seq_no = amdgpu_ras_mgr_gen_ras_event_seqno(adev, RAS_SEQNO_TYPE_DE); + RAS_DEV_INFO(adev, + "{%llu} RAS poison is created, no user action is needed.\n", + seq_no); + } + + ret = amdgpu_ras_process_handle_umc_interrupt(adev, ih_info); + } else if (ras_mgr->ras_core->poison_supported) { + ret = amdgpu_ras_process_handle_unexpected_interrupt(adev, ih_info); + } else { + RAS_DEV_WARN(adev, + "No RAS interrupt handler for non-UMC block with poison disabled.\n"); + } + + return ret; +} + +int amdgpu_ras_mgr_handle_consumer_interrupt(struct amdgpu_device *adev, void *data) +{ + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + return amdgpu_ras_process_handle_consumption_interrupt(adev, data); +} + +int amdgpu_ras_mgr_update_ras_ecc(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + return ras_core_update_ecc_info(ras_mgr->ras_core); +} + +int amdgpu_ras_mgr_reset_gpu(struct amdgpu_device *adev, uint32_t flags) +{ + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + con->gpu_reset_flags |= flags; + return amdgpu_ras_reset_gpu(adev); +} + +bool amdgpu_ras_mgr_check_eeprom_safety_watermark(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!amdgpu_ras_mgr_is_ready(adev)) + return false; + + return ras_eeprom_check_safety_watermark(ras_mgr->ras_core); +} + +int amdgpu_ras_mgr_get_curr_nps_mode(struct amdgpu_device *adev, + uint32_t *nps_mode) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + uint32_t mode; + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EINVAL; + + mode = ras_core_get_curr_nps_mode(ras_mgr->ras_core); + if (!mode || mode > AMDGPU_NPS8_PARTITION_MODE) + return -EINVAL; + + *nps_mode = mode; + + return 0; +} + +bool amdgpu_ras_mgr_check_retired_addr(struct amdgpu_device *adev, + uint64_t addr) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!amdgpu_ras_mgr_is_ready(adev)) + return false; + + return ras_umc_check_retired_addr(ras_mgr->ras_core, addr); +} + +bool amdgpu_ras_mgr_is_rma(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!ras_mgr || !ras_mgr->ras_core || !ras_mgr->ras_is_ready) + return false; + + return ras_core_gpu_is_rma(ras_mgr->ras_core); +} + +int amdgpu_ras_mgr_handle_ras_cmd(struct amdgpu_device *adev, + uint32_t cmd_id, void *input, uint32_t input_size, + void *output, uint32_t out_size) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + struct ras_cmd_ctx *cmd_ctx; + uint32_t ctx_buf_size = PAGE_SIZE; + int ret; + + if (!amdgpu_ras_mgr_is_ready(adev)) + return -EPERM; + + cmd_ctx = kzalloc(ctx_buf_size, GFP_KERNEL); + if (!cmd_ctx) + return -ENOMEM; + + cmd_ctx->cmd_id = cmd_id; + + memcpy(cmd_ctx->input_buff_raw, input, input_size); + cmd_ctx->input_size = input_size; + cmd_ctx->output_buf_size = ctx_buf_size - sizeof(*cmd_ctx); + + ret = amdgpu_ras_submit_cmd(ras_mgr->ras_core, cmd_ctx); + if (!ret && !cmd_ctx->cmd_res && output && (out_size == cmd_ctx->output_size)) + memcpy(output, cmd_ctx->output_buff_raw, cmd_ctx->output_size); + + kfree(cmd_ctx); + + return ret; +} + +int amdgpu_ras_mgr_pre_reset(struct amdgpu_device *adev) +{ + if (!amdgpu_ras_mgr_is_ready(adev)) { + RAS_DEV_ERR(adev, "Invalid ras suspend!\n"); + return -EPERM; + } + + amdgpu_ras_process_pre_reset(adev); + return 0; +} + +int amdgpu_ras_mgr_post_reset(struct amdgpu_device *adev) +{ + if (!amdgpu_ras_mgr_is_ready(adev)) { + RAS_DEV_ERR(adev, "Invalid ras resume!\n"); + return -EPERM; + } + + amdgpu_ras_process_post_reset(adev); + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h new file mode 100644 index 000000000000..8fb7eb4b8f13 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h @@ -0,0 +1,83 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright (c) 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ +#ifndef __AMDGPU_RAS_MGR_H__ +#define __AMDGPU_RAS_MGR_H__ +#include "ras.h" +#include "amdgpu_ras_process.h" + +enum ras_ih_type { + RAS_IH_NONE, + RAS_IH_FROM_BLOCK_CONTROLLER, + RAS_IH_FROM_CONSUMER_CLIENT, + RAS_IH_FROM_FATAL_ERROR, +}; + +struct ras_ih_info { + uint32_t block; + union { + struct amdgpu_iv_entry iv_entry; + struct { + uint16_t pasid; + uint32_t reset; + pasid_notify pasid_fn; + void *data; + }; + }; +}; + +struct amdgpu_ras_mgr { + struct amdgpu_device *adev; + struct ras_core_context *ras_core; + struct delayed_work retire_page_dwork; + struct ras_event_manager ras_event_mgr; + uint64_t last_poison_consumption_seqno; + bool ras_is_ready; + + bool is_paused; + struct completion ras_event_done; +}; + +extern const struct amdgpu_ip_block_version ras_v1_0_ip_block; + +struct amdgpu_ras_mgr *amdgpu_ras_mgr_get_context( + struct amdgpu_device *adev); +int amdgpu_enable_uniras(struct amdgpu_device *adev, bool enable); +bool amdgpu_uniras_enabled(struct amdgpu_device *adev); +int amdgpu_ras_mgr_handle_fatal_interrupt(struct amdgpu_device *adev, void *data); +int amdgpu_ras_mgr_handle_controller_interrupt(struct amdgpu_device *adev, void *data); +int amdgpu_ras_mgr_handle_consumer_interrupt(struct amdgpu_device *adev, void *data); +int amdgpu_ras_mgr_update_ras_ecc(struct amdgpu_device *adev); +int amdgpu_ras_mgr_reset_gpu(struct amdgpu_device *adev, uint32_t flags); +uint64_t amdgpu_ras_mgr_gen_ras_event_seqno(struct amdgpu_device *adev, + enum ras_seqno_type seqno_type); +bool amdgpu_ras_mgr_check_eeprom_safety_watermark(struct amdgpu_device *adev); +int amdgpu_ras_mgr_get_curr_nps_mode(struct amdgpu_device *adev, uint32_t *nps_mode); +bool amdgpu_ras_mgr_check_retired_addr(struct amdgpu_device *adev, + uint64_t addr); +bool amdgpu_ras_mgr_is_rma(struct amdgpu_device *adev); +int amdgpu_ras_mgr_handle_ras_cmd(struct amdgpu_device *adev, + uint32_t cmd_id, void *input, uint32_t input_size, + void *output, uint32_t out_size); +int amdgpu_ras_mgr_pre_reset(struct amdgpu_device *adev); +int amdgpu_ras_mgr_post_reset(struct amdgpu_device *adev); +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.c new file mode 100644 index 000000000000..79a51b1603ac --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.c @@ -0,0 +1,94 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "amdgpu_smu.h" +#include "amdgpu_reset.h" +#include "amdgpu_ras_mp1_v13_0.h" + +#define RAS_MP1_MSG_QueryValidMcaCeCount 0x3A +#define RAS_MP1_MSG_McaBankCeDumpDW 0x3B + +static int mp1_v13_0_get_valid_bank_count(struct ras_core_context *ras_core, + u32 msg, u32 *count) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + u32 smu_msg; + int ret = 0; + + if (!count) + return -EINVAL; + + smu_msg = (msg == RAS_MP1_MSG_QueryValidMcaCeCount) ? + SMU_MSG_QueryValidMcaCeCount : SMU_MSG_QueryValidMcaCount; + + if (down_read_trylock(&adev->reset_domain->sem)) { + ret = amdgpu_smu_ras_send_msg(adev, smu_msg, 0, count); + up_read(&adev->reset_domain->sem); + } else { + ret = -RAS_CORE_GPU_IN_MODE1_RESET; + } + + if (ret) + *count = 0; + + return ret; +} + +static int mp1_v13_0_dump_valid_bank(struct ras_core_context *ras_core, + u32 msg, u32 idx, u32 reg_idx, u64 *val) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + uint32_t data[2] = {0, 0}; + uint32_t param; + int ret = 0; + int i, offset; + u32 smu_msg = (msg == RAS_MP1_MSG_McaBankCeDumpDW) ? + SMU_MSG_McaBankCeDumpDW : SMU_MSG_McaBankDumpDW; + + if (down_read_trylock(&adev->reset_domain->sem)) { + offset = reg_idx * 8; + for (i = 0; i < ARRAY_SIZE(data); i++) { + param = ((idx & 0xffff) << 16) | ((offset + (i << 2)) & 0xfffc); + ret = amdgpu_smu_ras_send_msg(adev, smu_msg, param, &data[i]); + if (ret) { + RAS_DEV_ERR(adev, "ACA failed to read register[%d], offset:0x%x\n", + reg_idx, offset); + break; + } + } + up_read(&adev->reset_domain->sem); + + if (!ret) + *val = (uint64_t)data[1] << 32 | data[0]; + } else { + ret = -RAS_CORE_GPU_IN_MODE1_RESET; + } + + return ret; +} + +const struct ras_mp1_sys_func amdgpu_ras_mp1_sys_func_v13_0 = { + .mp1_get_valid_bank_count = mp1_v13_0_get_valid_bank_count, + .mp1_dump_valid_bank = mp1_v13_0_dump_valid_bank, +}; + diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.h new file mode 100644 index 000000000000..71c614ae1ae4 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mp1_v13_0.h @@ -0,0 +1,30 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +#ifndef __AMDGPU_RAS_MP1_V13_0_H__ +#define __AMDGPU_RAS_MP1_V13_0_H__ +#include "ras.h" + +extern const struct ras_mp1_sys_func amdgpu_ras_mp1_sys_func_v13_0; + +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.c new file mode 100644 index 000000000000..2783f5875c7c --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.c @@ -0,0 +1,125 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "amdgpu_ras_mgr.h" +#include "amdgpu_ras_nbio_v7_9.h" +#include "nbio/nbio_7_9_0_offset.h" +#include "nbio/nbio_7_9_0_sh_mask.h" +#include "ivsrcid/nbio/irqsrcs_nbif_7_4.h" + +static int nbio_v7_9_set_ras_controller_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *src, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + /* Dummy function, there is no initialization operation in driver */ + + return 0; +} + +static int nbio_v7_9_process_ras_controller_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + /* By design, the ih cookie for ras_controller_irq should be written + * to BIFring instead of general iv ring. However, due to known bif ring + * hw bug, it has to be disabled. There is no chance the process function + * will be involked. Just left it as a dummy one. + */ + return 0; +} + +static int nbio_v7_9_set_ras_err_event_athub_irq_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *src, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + /* Dummy function, there is no initialization operation in driver */ + + return 0; +} + +static int nbio_v7_9_process_err_event_athub_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + /* By design, the ih cookie for err_event_athub_irq should be written + * to BIFring instead of general iv ring. However, due to known bif ring + * hw bug, it has to be disabled. There is no chance the process function + * will be involked. Just left it as a dummy one. + */ + return 0; +} + +static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_controller_irq_funcs = { + .set = nbio_v7_9_set_ras_controller_irq_state, + .process = nbio_v7_9_process_ras_controller_irq, +}; + +static const struct amdgpu_irq_src_funcs nbio_v7_9_ras_err_event_athub_irq_funcs = { + .set = nbio_v7_9_set_ras_err_event_athub_irq_state, + .process = nbio_v7_9_process_err_event_athub_irq, +}; + +static int nbio_v7_9_init_ras_controller_interrupt(struct ras_core_context *ras_core, bool state) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int r; + + /* init the irq funcs */ + adev->nbio.ras_controller_irq.funcs = + &nbio_v7_9_ras_controller_irq_funcs; + adev->nbio.ras_controller_irq.num_types = 1; + + /* register ras controller interrupt */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, + NBIF_7_4__SRCID__RAS_CONTROLLER_INTERRUPT, + &adev->nbio.ras_controller_irq); + + return r; +} + +static int nbio_v7_9_init_ras_err_event_athub_interrupt(struct ras_core_context *ras_core, + bool state) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int r; + + /* init the irq funcs */ + adev->nbio.ras_err_event_athub_irq.funcs = + &nbio_v7_9_ras_err_event_athub_irq_funcs; + adev->nbio.ras_err_event_athub_irq.num_types = 1; + + /* register ras err event athub interrupt */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_BIF, + NBIF_7_4__SRCID__ERREVENT_ATHUB_INTERRUPT, + &adev->nbio.ras_err_event_athub_irq); + + return r; +} + +const struct ras_nbio_sys_func amdgpu_ras_nbio_sys_func_v7_9 = { + .set_ras_controller_irq_state = nbio_v7_9_init_ras_controller_interrupt, + .set_ras_err_event_athub_irq_state = nbio_v7_9_init_ras_err_event_athub_interrupt, +}; diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.h new file mode 100644 index 000000000000..272259e9a0e7 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_nbio_v7_9.h @@ -0,0 +1,30 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __AMDGPU_RAS_NBIO_V7_9_H__ +#define __AMDGPU_RAS_NBIO_V7_9_H__ + +extern const struct ras_nbio_sys_func amdgpu_ras_nbio_sys_func_v7_9; + +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c new file mode 100644 index 000000000000..5782c007de71 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c @@ -0,0 +1,190 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright (c) 2025 Advanced Micro Devices, Inc. All rights reserved. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ + +#include "amdgpu.h" +#include "amdgpu_reset.h" +#include "amdgpu_xgmi.h" +#include "ras_sys.h" +#include "amdgpu_ras_mgr.h" +#include "amdgpu_ras_process.h" + +#define RAS_MGR_RETIRE_PAGE_INTERVAL 100 +#define RAS_EVENT_PROCESS_TIMEOUT 1200 + +static void ras_process_retire_page_dwork(struct work_struct *work) +{ + struct amdgpu_ras_mgr *ras_mgr = + container_of(work, struct amdgpu_ras_mgr, retire_page_dwork.work); + struct amdgpu_device *adev = ras_mgr->adev; + int ret; + + if (amdgpu_ras_is_rma(adev)) + return; + + /* If gpu reset is ongoing, delay retiring the bad pages */ + if (amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) { + schedule_delayed_work(&ras_mgr->retire_page_dwork, + msecs_to_jiffies(RAS_MGR_RETIRE_PAGE_INTERVAL * 3)); + return; + } + + ret = ras_umc_handle_bad_pages(ras_mgr->ras_core, NULL); + if (!ret) + schedule_delayed_work(&ras_mgr->retire_page_dwork, + msecs_to_jiffies(RAS_MGR_RETIRE_PAGE_INTERVAL)); +} + +int amdgpu_ras_process_init(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + ras_mgr->is_paused = false; + init_completion(&ras_mgr->ras_event_done); + + INIT_DELAYED_WORK(&ras_mgr->retire_page_dwork, ras_process_retire_page_dwork); + + return 0; +} + +int amdgpu_ras_process_fini(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + ras_mgr->is_paused = false; + /* Save all cached bad pages to eeprom */ + flush_delayed_work(&ras_mgr->retire_page_dwork); + cancel_delayed_work_sync(&ras_mgr->retire_page_dwork); + return 0; +} + +int amdgpu_ras_process_handle_umc_interrupt(struct amdgpu_device *adev, void *data) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!ras_mgr->ras_core) + return -EINVAL; + + return ras_process_add_interrupt_req(ras_mgr->ras_core, NULL, true); +} + +int amdgpu_ras_process_handle_unexpected_interrupt(struct amdgpu_device *adev, void *data) +{ + amdgpu_ras_set_fed(adev, true); + return amdgpu_ras_mgr_reset_gpu(adev, AMDGPU_RAS_GPU_RESET_MODE1_RESET); +} + +int amdgpu_ras_process_handle_consumption_interrupt(struct amdgpu_device *adev, void *data) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + struct ras_ih_info *ih_info = (struct ras_ih_info *)data; + struct ras_event_req req; + uint64_t seqno; + + if (!ih_info) + return -EINVAL; + + memset(&req, 0, sizeof(req)); + req.block = ih_info->block; + req.data = ih_info->data; + req.pasid = ih_info->pasid; + req.pasid_fn = ih_info->pasid_fn; + req.reset = ih_info->reset; + + seqno = ras_core_get_seqno(ras_mgr->ras_core, + RAS_SEQNO_TYPE_POISON_CONSUMPTION, false); + + /* When the ACA register cannot be read from FW, the poison + * consumption seqno in the fifo will not pop up, so it is + * necessary to check whether the seqno is the previous seqno. + */ + if (seqno == ras_mgr->last_poison_consumption_seqno) { + /* Pop and discard the previous seqno */ + ras_core_get_seqno(ras_mgr->ras_core, + RAS_SEQNO_TYPE_POISON_CONSUMPTION, true); + seqno = ras_core_get_seqno(ras_mgr->ras_core, + RAS_SEQNO_TYPE_POISON_CONSUMPTION, false); + } + ras_mgr->last_poison_consumption_seqno = seqno; + req.seqno = seqno; + + return ras_process_add_interrupt_req(ras_mgr->ras_core, &req, false); +} + +int amdgpu_ras_process_begin(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (ras_mgr->is_paused) + return -EAGAIN; + + reinit_completion(&ras_mgr->ras_event_done); + return 0; +} + +int amdgpu_ras_process_end(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + complete(&ras_mgr->ras_event_done); + return 0; +} + +int amdgpu_ras_process_pre_reset(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + long rc; + + if (!ras_mgr || !ras_mgr->ras_core) + return -EINVAL; + + if (!ras_mgr->ras_core->is_initialized) + return -EPERM; + + ras_mgr->is_paused = true; + + /* Wait for RAS event processing to complete */ + rc = wait_for_completion_interruptible_timeout(&ras_mgr->ras_event_done, + msecs_to_jiffies(RAS_EVENT_PROCESS_TIMEOUT)); + if (rc <= 0) + RAS_DEV_WARN(adev, "Waiting for ras process to complete %s\n", + rc ? "interrupted" : "timeout"); + + flush_delayed_work(&ras_mgr->retire_page_dwork); + return 0; +} + +int amdgpu_ras_process_post_reset(struct amdgpu_device *adev) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + + if (!ras_mgr || !ras_mgr->ras_core) + return -EINVAL; + + if (!ras_mgr->ras_core->is_initialized) + return -EPERM; + + ras_mgr->is_paused = false; + + schedule_delayed_work(&ras_mgr->retire_page_dwork, 0); + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.h b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.h new file mode 100644 index 000000000000..d55cdaeac441 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.h @@ -0,0 +1,41 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright (c) 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ +#ifndef __AMDGPU_RAS_PROCESS_H__ +#define __AMDGPU_RAS_PROCESS_H__ +#include "ras_process.h" +#include "amdgpu_ras_mgr.h" + +enum ras_ih_type; +int amdgpu_ras_process_init(struct amdgpu_device *adev); +int amdgpu_ras_process_fini(struct amdgpu_device *adev); +int amdgpu_ras_process_handle_umc_interrupt(struct amdgpu_device *adev, + void *data); +int amdgpu_ras_process_handle_unexpected_interrupt(struct amdgpu_device *adev, + void *data); +int amdgpu_ras_process_handle_consumption_interrupt(struct amdgpu_device *adev, + void *data); +int amdgpu_ras_process_begin(struct amdgpu_device *adev); +int amdgpu_ras_process_end(struct amdgpu_device *adev); +int amdgpu_ras_process_pre_reset(struct amdgpu_device *adev); +int amdgpu_ras_process_post_reset(struct amdgpu_device *adev); +#endif diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_sys.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_sys.c new file mode 100644 index 000000000000..45ed8c3b5563 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_sys.c @@ -0,0 +1,279 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras_sys.h" +#include "amdgpu_ras_mgr.h" +#include "amdgpu_ras.h" +#include "amdgpu_reset.h" + +static int amdgpu_ras_sys_detect_fatal_event(struct ras_core_context *ras_core, void *data) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int ret; + uint64_t seq_no; + + ret = amdgpu_ras_global_ras_isr(adev); + if (ret) + return ret; + + seq_no = amdgpu_ras_mgr_gen_ras_event_seqno(adev, RAS_SEQNO_TYPE_UE); + RAS_DEV_INFO(adev, + "{%llu} Uncorrectable hardware error(ERREVENT_ATHUB_INTERRUPT) detected!\n", + seq_no); + + return amdgpu_ras_process_handle_unexpected_interrupt(adev, data); +} + +static int amdgpu_ras_sys_poison_consumption_event(struct ras_core_context *ras_core, + void *data) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct ras_event_req *req = (struct ras_event_req *)data; + pasid_notify pasid_fn; + + if (!req) + return -EINVAL; + + if (req->pasid_fn) { + pasid_fn = (pasid_notify)req->pasid_fn; + pasid_fn(adev, req->pasid, req->data); + } + + return 0; +} + +static int amdgpu_ras_sys_gen_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, uint64_t *seqno) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(adev); + struct ras_event_manager *event_mgr; + struct ras_event_state *event_state; + struct amdgpu_hive_info *hive; + enum ras_event_type event_type; + uint64_t seq_no; + + if (!ras_mgr || !seqno || + (seqno_type >= RAS_SEQNO_TYPE_COUNT_MAX)) + return -EINVAL; + + switch (seqno_type) { + case RAS_SEQNO_TYPE_UE: + event_type = RAS_EVENT_TYPE_FATAL; + break; + case RAS_SEQNO_TYPE_CE: + case RAS_SEQNO_TYPE_DE: + event_type = RAS_EVENT_TYPE_POISON_CREATION; + break; + case RAS_SEQNO_TYPE_POISON_CONSUMPTION: + event_type = RAS_EVENT_TYPE_POISON_CONSUMPTION; + break; + default: + event_type = RAS_EVENT_TYPE_INVALID; + break; + } + + hive = amdgpu_get_xgmi_hive(adev); + event_mgr = hive ? &hive->event_mgr : &ras_mgr->ras_event_mgr; + event_state = &event_mgr->event_state[event_type]; + if ((event_type == RAS_EVENT_TYPE_FATAL) && amdgpu_ras_in_recovery(adev)) { + seq_no = event_state->last_seqno; + } else { + seq_no = atomic64_inc_return(&event_mgr->seqno); + event_state->last_seqno = seq_no; + atomic64_inc(&event_state->count); + } + amdgpu_put_xgmi_hive(hive); + + *seqno = seq_no; + return 0; + +} + +static int amdgpu_ras_sys_event_notifier(struct ras_core_context *ras_core, + enum ras_notify_event event_id, void *data) +{ + struct amdgpu_ras_mgr *ras_mgr = amdgpu_ras_mgr_get_context(ras_core->dev); + int ret = 0; + + switch (event_id) { + case RAS_EVENT_ID__BAD_PAGE_DETECTED: + schedule_delayed_work(&ras_mgr->retire_page_dwork, 0); + break; + case RAS_EVENT_ID__POISON_CONSUMPTION: + amdgpu_ras_sys_poison_consumption_event(ras_core, data); + break; + case RAS_EVENT_ID__RESERVE_BAD_PAGE: + ret = amdgpu_ras_reserve_page(ras_core->dev, *(uint64_t *)data); + break; + case RAS_EVENT_ID__FATAL_ERROR_DETECTED: + ret = amdgpu_ras_sys_detect_fatal_event(ras_core, data); + break; + case RAS_EVENT_ID__UPDATE_BAD_PAGE_NUM: + ret = amdgpu_dpm_send_hbm_bad_pages_num(ras_core->dev, *(uint32_t *)data); + break; + case RAS_EVENT_ID__UPDATE_BAD_CHANNEL_BITMAP: + ret = amdgpu_dpm_send_hbm_bad_channel_flag(ras_core->dev, *(uint32_t *)data); + break; + case RAS_EVENT_ID__DEVICE_RMA: + ras_log_ring_add_log_event(ras_core, RAS_LOG_EVENT_RMA, NULL, NULL); + ret = amdgpu_dpm_send_rma_reason(ras_core->dev); + break; + case RAS_EVENT_ID__RESET_GPU: + ret = amdgpu_ras_mgr_reset_gpu(ras_core->dev, *(uint32_t *)data); + break; + case RAS_EVENT_ID__RAS_EVENT_PROC_BEGIN: + ret = amdgpu_ras_process_begin(ras_core->dev); + break; + case RAS_EVENT_ID__RAS_EVENT_PROC_END: + ret = amdgpu_ras_process_end(ras_core->dev); + break; + default: + RAS_DEV_WARN(ras_core->dev, "Invalid ras notify event:%d\n", event_id); + break; + } + + return ret; +} + +static u64 amdgpu_ras_sys_get_utc_second_timestamp(struct ras_core_context *ras_core) +{ + return ktime_get_real_seconds(); +} + +static int amdgpu_ras_sys_check_gpu_status(struct ras_core_context *ras_core, + uint32_t *status) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + uint32_t gpu_status = 0; + + if (amdgpu_in_reset(adev) || amdgpu_ras_in_recovery(adev)) + gpu_status |= RAS_GPU_STATUS__IN_RESET; + + if (amdgpu_sriov_vf(adev)) + gpu_status |= RAS_GPU_STATUS__IS_VF; + + *status = gpu_status; + + return 0; +} + +static int amdgpu_ras_sys_get_device_system_info(struct ras_core_context *ras_core, + struct device_system_info *dev_info) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + + dev_info->device_id = adev->pdev->device; + dev_info->vendor_id = adev->pdev->vendor; + dev_info->socket_id = adev->smuio.funcs->get_socket_id(adev); + + return 0; +} + +static int amdgpu_ras_sys_gpu_reset_lock(struct ras_core_context *ras_core, + bool down, bool try) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + int ret = 0; + + if (down && try) + ret = down_read_trylock(&adev->reset_domain->sem); + else if (down) + down_read(&adev->reset_domain->sem); + else + up_read(&adev->reset_domain->sem); + + return ret; +} + +static bool amdgpu_ras_sys_detect_ras_interrupt(struct ras_core_context *ras_core) +{ + return !!atomic_read(&amdgpu_ras_in_intr); +} + +static int amdgpu_ras_sys_get_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem) +{ + struct amdgpu_device *adev = (struct amdgpu_device *)ras_core->dev; + struct psp_context *psp = &adev->psp; + struct psp_ring *psp_ring; + struct ta_mem_context *mem_ctx; + + if (mem_type == GPU_MEM_TYPE_RAS_PSP_RING) { + psp_ring = &psp->km_ring; + gpu_mem->mem_bo = adev->firmware.rbuf; + gpu_mem->mem_size = psp_ring->ring_size; + gpu_mem->mem_mc_addr = psp_ring->ring_mem_mc_addr; + gpu_mem->mem_cpu_addr = psp_ring->ring_mem; + } else if (mem_type == GPU_MEM_TYPE_RAS_PSP_CMD) { + gpu_mem->mem_bo = psp->cmd_buf_bo; + gpu_mem->mem_size = PSP_CMD_BUFFER_SIZE; + gpu_mem->mem_mc_addr = psp->cmd_buf_mc_addr; + gpu_mem->mem_cpu_addr = psp->cmd_buf_mem; + } else if (mem_type == GPU_MEM_TYPE_RAS_PSP_FENCE) { + gpu_mem->mem_bo = psp->fence_buf_bo; + gpu_mem->mem_size = PSP_FENCE_BUFFER_SIZE; + gpu_mem->mem_mc_addr = psp->fence_buf_mc_addr; + gpu_mem->mem_cpu_addr = psp->fence_buf; + } else if (mem_type == GPU_MEM_TYPE_RAS_TA_FW) { + gpu_mem->mem_bo = psp->fw_pri_bo; + gpu_mem->mem_size = PSP_1_MEG; + gpu_mem->mem_mc_addr = psp->fw_pri_mc_addr; + gpu_mem->mem_cpu_addr = psp->fw_pri_buf; + } else if (mem_type == GPU_MEM_TYPE_RAS_TA_CMD) { + mem_ctx = &psp->ras_context.context.mem_context; + gpu_mem->mem_bo = mem_ctx->shared_bo; + gpu_mem->mem_size = mem_ctx->shared_mem_size; + gpu_mem->mem_mc_addr = mem_ctx->shared_mc_addr; + gpu_mem->mem_cpu_addr = mem_ctx->shared_buf; + } else { + return -EINVAL; + } + + if (!gpu_mem->mem_bo || !gpu_mem->mem_size || + !gpu_mem->mem_mc_addr || !gpu_mem->mem_cpu_addr) { + RAS_DEV_ERR(ras_core->dev, "The ras psp gpu memory is invalid!\n"); + return -ENOMEM; + } + + return 0; +} + +static int amdgpu_ras_sys_put_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem) +{ + + return 0; +} + +const struct ras_sys_func amdgpu_ras_sys_fn = { + .ras_notifier = amdgpu_ras_sys_event_notifier, + .get_utc_second_timestamp = amdgpu_ras_sys_get_utc_second_timestamp, + .gen_seqno = amdgpu_ras_sys_gen_seqno, + .check_gpu_status = amdgpu_ras_sys_check_gpu_status, + .get_device_system_info = amdgpu_ras_sys_get_device_system_info, + .gpu_reset_lock = amdgpu_ras_sys_gpu_reset_lock, + .detect_ras_interrupt = amdgpu_ras_sys_detect_ras_interrupt, + .get_gpu_mem = amdgpu_ras_sys_get_gpu_mem, + .put_gpu_mem = amdgpu_ras_sys_put_gpu_mem, +}; diff --git a/drivers/gpu/drm/amd/ras/ras_mgr/ras_sys.h b/drivers/gpu/drm/amd/ras/ras_mgr/ras_sys.h new file mode 100644 index 000000000000..8156531a7b63 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/ras_mgr/ras_sys.h @@ -0,0 +1,110 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_SYS_H__ +#define __RAS_SYS_H__ +#include <linux/stdarg.h> +#include <linux/printk.h> +#include <linux/dev_printk.h> +#include <linux/mempool.h> +#include "amdgpu.h" + +#define RAS_DEV_ERR(device, fmt, ...) \ + do { \ + if (device) \ + dev_err(((struct amdgpu_device *)device)->dev, fmt, ##__VA_ARGS__); \ + else \ + printk(KERN_ERR fmt, ##__VA_ARGS__); \ + } while (0) + +#define RAS_DEV_WARN(device, fmt, ...) \ + do { \ + if (device) \ + dev_warn(((struct amdgpu_device *)device)->dev, fmt, ##__VA_ARGS__); \ + else \ + printk(KERN_WARNING fmt, ##__VA_ARGS__); \ + } while (0) + +#define RAS_DEV_INFO(device, fmt, ...) \ + do { \ + if (device) \ + dev_info(((struct amdgpu_device *)device)->dev, fmt, ##__VA_ARGS__); \ + else \ + printk(KERN_INFO fmt, ##__VA_ARGS__); \ + } while (0) + +#define RAS_DEV_DBG(device, fmt, ...) \ + do { \ + if (device) \ + dev_dbg(((struct amdgpu_device *)device)->dev, fmt, ##__VA_ARGS__); \ + else \ + printk(KERN_DEBUG fmt, ##__VA_ARGS__); \ + } while (0) + +#define RAS_INFO(fmt, ...) printk(KERN_INFO fmt, ##__VA_ARGS__) + +#define RAS_DEV_RREG32_SOC15(dev, ip, inst, reg) \ +({ \ + struct amdgpu_device *adev = (struct amdgpu_device *)dev; \ + __RREG32_SOC15_RLC__(adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg, \ + 0, ip##_HWIP, inst); \ +}) + +#define RAS_DEV_WREG32_SOC15(dev, ip, inst, reg, value) \ +({ \ + struct amdgpu_device *adev = (struct amdgpu_device *)dev; \ + __WREG32_SOC15_RLC__((adev->reg_offset[ip##_HWIP][inst][reg##_BASE_IDX] + reg), \ + value, 0, ip##_HWIP, inst); \ +}) + +/* GET_INST returns the physical instance corresponding to a logical instance */ +#define RAS_GET_INST(dev, ip, inst) \ +({ \ + struct amdgpu_device *adev = (struct amdgpu_device *)dev; \ + adev->ip_map.logical_to_dev_inst ? \ + adev->ip_map.logical_to_dev_inst(adev, ip##_HWIP, inst) : inst; \ +}) + +#define RAS_GET_MASK(dev, ip, mask) \ +({ \ + struct amdgpu_device *adev = (struct amdgpu_device *)dev; \ + (adev->ip_map.logical_to_dev_mask ? \ + adev->ip_map.logical_to_dev_mask(adev, ip##_HWIP, mask) : mask); \ +}) + +static inline void *ras_radix_tree_delete_iter(struct radix_tree_root *root, void *iter) +{ + return radix_tree_delete(root, ((struct radix_tree_iter *)iter)->index); +} + +static inline long ras_wait_event_interruptible_timeout(void *wq_head, + int (*condition)(void *param), void *param, unsigned int timeout) +{ + return wait_event_interruptible_timeout(*(wait_queue_head_t *)wq_head, + condition(param), timeout); +} + +extern const struct ras_sys_func amdgpu_ras_sys_fn; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/Makefile b/drivers/gpu/drm/amd/ras/rascore/Makefile index e69de29bb2d1..e826a1f86424 100644 --- a/drivers/gpu/drm/amd/ras/rascore/Makefile +++ b/drivers/gpu/drm/amd/ras/rascore/Makefile @@ -0,0 +1,44 @@ +# +# Copyright 2025 Advanced Micro Devices, Inc. +# +# Permission is hereby granted, free of charge, to any person obtaining a +# copy of this software and associated documentation files (the "Software"), +# to deal in the Software without restriction, including without limitation +# the rights to use, copy, modify, merge, publish, distribute, sublicense, +# and/or sell copies of the Software, and to permit persons to whom the +# Software is furnished to do so, subject to the following conditions: +# +# The above copyright notice and this permission notice shall be included in +# all copies or substantial portions of the Software. +# +# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL +# THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR +# OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, +# ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR +# OTHER DEALINGS IN THE SOFTWARE. +# +RAS_CORE_FILES = ras_core.o \ + ras_mp1.o \ + ras_mp1_v13_0.o \ + ras_aca.o \ + ras_aca_v1_0.o \ + ras_eeprom.o \ + ras_umc.o \ + ras_umc_v12_0.o \ + ras_cmd.o \ + ras_gfx.o \ + ras_gfx_v9_0.o \ + ras_process.o \ + ras_nbio.o \ + ras_nbio_v7_9.o \ + ras_log_ring.o \ + ras_cper.o \ + ras_psp.o \ + ras_psp_v13_0.o + + +RAS_CORE = $(addprefix $(AMD_GPU_RAS_PATH)/rascore/,$(RAS_CORE_FILES)) + +AMD_GPU_RAS_FILES += $(RAS_CORE) diff --git a/drivers/gpu/drm/amd/ras/rascore/ras.h b/drivers/gpu/drm/amd/ras/rascore/ras.h new file mode 100644 index 000000000000..3396b2e0949d --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras.h @@ -0,0 +1,370 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_H__ +#define __RAS_H__ +#include "ras_sys.h" +#include "ras_umc.h" +#include "ras_aca.h" +#include "ras_eeprom.h" +#include "ras_core_status.h" +#include "ras_process.h" +#include "ras_gfx.h" +#include "ras_cmd.h" +#include "ras_nbio.h" +#include "ras_mp1.h" +#include "ras_psp.h" +#include "ras_log_ring.h" + +#define RAS_HW_ERR "[Hardware Error]: " + +#define RAS_GPU_PAGE_SHIFT 12 +#define RAS_ADDR_TO_PFN(addr) ((addr) >> RAS_GPU_PAGE_SHIFT) +#define RAS_PFN_TO_ADDR(pfn) ((pfn) << RAS_GPU_PAGE_SHIFT) + +#define RAS_CORE_RESET_GPU 0x10000 + +#define GPU_RESET_CAUSE_POISON (RAS_CORE_RESET_GPU | 0x0001) +#define GPU_RESET_CAUSE_FATAL (RAS_CORE_RESET_GPU | 0x0002) +#define GPU_RESET_CAUSE_RMA (RAS_CORE_RESET_GPU | 0x0004) + +enum ras_block_id { + RAS_BLOCK_ID__UMC = 0, + RAS_BLOCK_ID__SDMA, + RAS_BLOCK_ID__GFX, + RAS_BLOCK_ID__MMHUB, + RAS_BLOCK_ID__ATHUB, + RAS_BLOCK_ID__PCIE_BIF, + RAS_BLOCK_ID__HDP, + RAS_BLOCK_ID__XGMI_WAFL, + RAS_BLOCK_ID__DF, + RAS_BLOCK_ID__SMN, + RAS_BLOCK_ID__SEM, + RAS_BLOCK_ID__MP0, + RAS_BLOCK_ID__MP1, + RAS_BLOCK_ID__FUSE, + RAS_BLOCK_ID__MCA, + RAS_BLOCK_ID__VCN, + RAS_BLOCK_ID__JPEG, + RAS_BLOCK_ID__IH, + RAS_BLOCK_ID__MPIO, + + RAS_BLOCK_ID__LAST +}; + +enum ras_ecc_err_type { + RAS_ECC_ERR__NONE = 0, + RAS_ECC_ERR__PARITY = 1, + RAS_ECC_ERR__SINGLE_CORRECTABLE = 2, + RAS_ECC_ERR__MULTI_UNCORRECTABLE = 4, + RAS_ECC_ERR__POISON = 8, +}; + +enum ras_err_type { + RAS_ERR_TYPE__UE = 0, + RAS_ERR_TYPE__CE, + RAS_ERR_TYPE__DE, + RAS_ERR_TYPE__LAST +}; + +enum ras_seqno_type { + RAS_SEQNO_TYPE_INVALID = 0, + RAS_SEQNO_TYPE_UE, + RAS_SEQNO_TYPE_CE, + RAS_SEQNO_TYPE_DE, + RAS_SEQNO_TYPE_POISON_CONSUMPTION, + RAS_SEQNO_TYPE_COUNT_MAX, +}; + +enum ras_seqno_fifo { + SEQNO_FIFO_INVALID = 0, + SEQNO_FIFO_POISON_CREATION, + SEQNO_FIFO_POISON_CONSUMPTION, + SEQNO_FIFO_COUNT_MAX +}; + +enum ras_notify_event { + RAS_EVENT_ID__NONE, + RAS_EVENT_ID__BAD_PAGE_DETECTED, + RAS_EVENT_ID__POISON_CONSUMPTION, + RAS_EVENT_ID__RESERVE_BAD_PAGE, + RAS_EVENT_ID__DEVICE_RMA, + RAS_EVENT_ID__UPDATE_BAD_PAGE_NUM, + RAS_EVENT_ID__UPDATE_BAD_CHANNEL_BITMAP, + RAS_EVENT_ID__FATAL_ERROR_DETECTED, + RAS_EVENT_ID__RESET_GPU, + RAS_EVENT_ID__RESET_VF, + RAS_EVENT_ID__RAS_EVENT_PROC_BEGIN, + RAS_EVENT_ID__RAS_EVENT_PROC_END, +}; + +enum ras_gpu_status { + RAS_GPU_STATUS__NOT_READY = 0, + RAS_GPU_STATUS__READY = 0x1, + RAS_GPU_STATUS__IN_RESET = 0x2, + RAS_GPU_STATUS__IS_RMA = 0x4, + RAS_GPU_STATUS__IS_VF = 0x8, +}; + +struct ras_core_context; +struct ras_bank_ecc; +struct ras_umc; +struct ras_aca; +struct ras_process; +struct ras_nbio; +struct ras_log_ring; +struct ras_psp; + +struct ras_mp1_sys_func { + int (*mp1_get_valid_bank_count)(struct ras_core_context *ras_core, + u32 msg, u32 *count); + int (*mp1_dump_valid_bank)(struct ras_core_context *ras_core, + u32 msg, u32 idx, u32 reg_idx, u64 *val); +}; + +struct ras_eeprom_sys_func { + int (*eeprom_i2c_xfer)(struct ras_core_context *ras_core, + u32 eeprom_addr, u8 *eeprom_buf, u32 buf_size, bool read); + int (*update_eeprom_i2c_config)(struct ras_core_context *ras_core); +}; + +struct ras_nbio_sys_func { + int (*set_ras_controller_irq_state)(struct ras_core_context *ras_core, + bool state); + int (*set_ras_err_event_athub_irq_state)(struct ras_core_context *ras_core, + bool state); +}; + +struct ras_time { + int tm_sec; + int tm_min; + int tm_hour; + int tm_mday; + int tm_mon; + long tm_year; +}; + +struct device_system_info { + uint32_t device_id; + uint32_t vendor_id; + uint32_t socket_id; +}; + +enum gpu_mem_type { + GPU_MEM_TYPE_DEFAULT, + GPU_MEM_TYPE_RAS_PSP_RING, + GPU_MEM_TYPE_RAS_PSP_CMD, + GPU_MEM_TYPE_RAS_PSP_FENCE, + GPU_MEM_TYPE_RAS_TA_FW, + GPU_MEM_TYPE_RAS_TA_CMD, +}; + +struct ras_psp_sys_func { + int (*get_ras_psp_system_status)(struct ras_core_context *ras_core, + struct ras_psp_sys_status *status); + int (*get_ras_ta_init_param)(struct ras_core_context *ras_core, + struct ras_ta_init_param *ras_ta_param); +}; + +struct ras_sys_func { + int (*gpu_reset_lock)(struct ras_core_context *ras_core, + bool down, bool try); + int (*check_gpu_status)(struct ras_core_context *ras_core, + uint32_t *status); + int (*gen_seqno)(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, uint64_t *seqno); + int (*async_handle_ras_event)(struct ras_core_context *ras_core, void *data); + int (*ras_notifier)(struct ras_core_context *ras_core, + enum ras_notify_event event_id, void *data); + u64 (*get_utc_second_timestamp)(struct ras_core_context *ras_core); + int (*get_device_system_info)(struct ras_core_context *ras_core, + struct device_system_info *dev_info); + bool (*detect_ras_interrupt)(struct ras_core_context *ras_core); + int (*get_gpu_mem)(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem); + int (*put_gpu_mem)(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem); +}; + +struct ras_ecc_count { + uint64_t new_ce_count; + uint64_t total_ce_count; + uint64_t new_ue_count; + uint64_t total_ue_count; + uint64_t new_de_count; + uint64_t total_de_count; +}; + +struct ras_bank_ecc { + uint32_t nps; + uint64_t seq_no; + uint64_t status; + uint64_t ipid; + uint64_t addr; +}; + +struct ras_bank_ecc_node { + struct list_head node; + struct ras_bank_ecc ecc; +}; + +struct ras_aca_config { + u32 socket_num_per_hive; + u32 aid_num_per_socket; + u32 xcd_num_per_aid; +}; + +struct ras_mp1_config { + const struct ras_mp1_sys_func *mp1_sys_fn; +}; + +struct ras_nbio_config { + const struct ras_nbio_sys_func *nbio_sys_fn; +}; + +struct ras_psp_config { + const struct ras_psp_sys_func *psp_sys_fn; +}; + +struct ras_umc_config { + uint32_t umc_vram_type; +}; + +struct ras_eeprom_config { + const struct ras_eeprom_sys_func *eeprom_sys_fn; + int eeprom_record_threshold_config; + uint32_t eeprom_record_threshold_count; + void *eeprom_i2c_adapter; + u32 eeprom_i2c_addr; + u32 eeprom_i2c_port; + u16 max_i2c_read_len; + u16 max_i2c_write_len; +}; + +struct ras_core_config { + u32 aca_ip_version; + u32 umc_ip_version; + u32 mp1_ip_version; + u32 gfx_ip_version; + u32 nbio_ip_version; + u32 psp_ip_version; + + bool poison_supported; + bool ras_eeprom_supported; + const struct ras_sys_func *sys_fn; + + struct ras_aca_config aca_cfg; + struct ras_mp1_config mp1_cfg; + struct ras_nbio_config nbio_cfg; + struct ras_psp_config psp_cfg; + struct ras_eeprom_config eeprom_cfg; + struct ras_umc_config umc_cfg; +}; + +struct ras_core_context { + void *dev; + struct ras_core_config *config; + u32 socket_num_per_hive; + u32 aid_num_per_socket; + u32 xcd_num_per_aid; + int max_ue_banks_per_query; + int max_ce_banks_per_query; + struct ras_aca ras_aca; + + bool ras_eeprom_supported; + struct ras_eeprom_control ras_eeprom; + + struct ras_psp ras_psp; + struct ras_umc ras_umc; + struct ras_nbio ras_nbio; + struct ras_gfx ras_gfx; + struct ras_mp1 ras_mp1; + struct ras_process ras_proc; + struct ras_cmd_mgr ras_cmd; + struct ras_log_ring ras_log_ring; + + const struct ras_sys_func *sys_fn; + + /* is poison mode supported */ + bool poison_supported; + + bool is_rma; + bool is_initialized; + + struct kfifo de_seqno_fifo; + struct kfifo consumption_seqno_fifo; + spinlock_t seqno_lock; + + bool ras_core_enabled; +}; + +struct ras_core_context *ras_core_create(struct ras_core_config *init_config); +void ras_core_destroy(struct ras_core_context *ras_core); +int ras_core_sw_init(struct ras_core_context *ras_core); +int ras_core_sw_fini(struct ras_core_context *ras_core); +int ras_core_hw_init(struct ras_core_context *ras_core); +int ras_core_hw_fini(struct ras_core_context *ras_core); +bool ras_core_is_ready(struct ras_core_context *ras_core); +uint64_t ras_core_gen_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type); +uint64_t ras_core_get_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, bool pop); + +int ras_core_put_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, uint64_t seqno); + +int ras_core_update_ecc_info(struct ras_core_context *ras_core); +int ras_core_query_block_ecc_data(struct ras_core_context *ras_core, + enum ras_block_id block, struct ras_ecc_count *ecc_count); + +bool ras_core_gpu_in_reset(struct ras_core_context *ras_core); +bool ras_core_gpu_is_rma(struct ras_core_context *ras_core); +bool ras_core_gpu_is_vf(struct ras_core_context *ras_core); +bool ras_core_handle_nbio_irq(struct ras_core_context *ras_core, void *data); +int ras_core_handle_fatal_error(struct ras_core_context *ras_core); + +uint32_t ras_core_get_curr_nps_mode(struct ras_core_context *ras_core); +const char *ras_core_get_ras_block_name(enum ras_block_id block_id); +int ras_core_convert_timestamp_to_time(struct ras_core_context *ras_core, + uint64_t timestamp, struct ras_time *tm); + +int ras_core_set_status(struct ras_core_context *ras_core, bool enable); +bool ras_core_is_enabled(struct ras_core_context *ras_core); +uint64_t ras_core_get_utc_second_timestamp(struct ras_core_context *ras_core); +int ras_core_translate_soc_pa_and_bank(struct ras_core_context *ras_core, + uint64_t *soc_pa, struct umc_bank_addr *bank_addr, bool bank_to_pa); +bool ras_core_ras_interrupt_detected(struct ras_core_context *ras_core); +int ras_core_get_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem); +int ras_core_put_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem); +bool ras_core_check_safety_watermark(struct ras_core_context *ras_core); +int ras_core_down_trylock_gpu_reset_lock(struct ras_core_context *ras_core); +void ras_core_down_gpu_reset_lock(struct ras_core_context *ras_core); +void ras_core_up_gpu_reset_lock(struct ras_core_context *ras_core); +int ras_core_event_notify(struct ras_core_context *ras_core, + enum ras_notify_event event_id, void *data); +int ras_core_get_device_system_info(struct ras_core_context *ras_core, + struct device_system_info *dev_info); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_aca.c b/drivers/gpu/drm/amd/ras/rascore/ras_aca.c new file mode 100644 index 000000000000..e433c70d2989 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_aca.c @@ -0,0 +1,672 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_aca.h" +#include "ras_aca_v1_0.h" +#include "ras_mp1_v13_0.h" + +#define ACA_MARK_FATAL_FLAG 0x100 +#define ACA_MARK_UE_READ_FLAG 0x1 + +#define blk_name(block_id) ras_core_get_ras_block_name(block_id) + +static struct aca_regs_dump { + const char *name; + int reg_idx; +} aca_regs[] = { + {"CONTROL", ACA_REG_IDX__CTL}, + {"STATUS", ACA_REG_IDX__STATUS}, + {"ADDR", ACA_REG_IDX__ADDR}, + {"MISC", ACA_REG_IDX__MISC0}, + {"CONFIG", ACA_REG_IDX__CONFG}, + {"IPID", ACA_REG_IDX__IPID}, + {"SYND", ACA_REG_IDX__SYND}, + {"DESTAT", ACA_REG_IDX__DESTAT}, + {"DEADDR", ACA_REG_IDX__DEADDR}, + {"CONTROL_MASK", ACA_REG_IDX__CTL_MASK}, +}; + + +static void aca_report_ecc_info(struct ras_core_context *ras_core, + u64 seq_no, u32 blk, u32 skt, u32 aid, + struct aca_aid_ecc *aid_ecc, + struct aca_bank_ecc *new_ecc) +{ + struct aca_ecc_count ecc_count = {0}; + + ecc_count.new_ue_count = new_ecc->ue_count; + ecc_count.new_de_count = new_ecc->de_count; + ecc_count.new_ce_count = new_ecc->ce_count; + if (blk == RAS_BLOCK_ID__GFX) { + struct aca_ecc_count *xcd_ecc; + int xcd_id; + + for (xcd_id = 0; xcd_id < aid_ecc->xcd.xcd_num; xcd_id++) { + xcd_ecc = &aid_ecc->xcd.xcd[xcd_id].ecc_err; + ecc_count.total_ue_count += xcd_ecc->total_ue_count; + ecc_count.total_de_count += xcd_ecc->total_de_count; + ecc_count.total_ce_count += xcd_ecc->total_ce_count; + } + } else { + ecc_count.total_ue_count = aid_ecc->ecc_err.total_ue_count; + ecc_count.total_de_count = aid_ecc->ecc_err.total_de_count; + ecc_count.total_ce_count = aid_ecc->ecc_err.total_ce_count; + } + + if (ecc_count.new_ue_count) { + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u new uncorrectable hardware errors detected in %s block\n", + seq_no, skt, aid, ecc_count.new_ue_count, blk_name(blk)); + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u uncorrectable hardware errors detected in total in %s block\n", + seq_no, skt, aid, ecc_count.total_ue_count, blk_name(blk)); + } + + if (ecc_count.new_de_count) { + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u new %s detected in %s block\n", + seq_no, skt, aid, ecc_count.new_de_count, + (blk == RAS_BLOCK_ID__UMC) ? + "deferred hardware errors" : "poison consumption", + blk_name(blk)); + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u %s detected in total in %s block\n", + seq_no, skt, aid, ecc_count.total_de_count, + (blk == RAS_BLOCK_ID__UMC) ? + "deferred hardware errors" : "poison consumption", + blk_name(blk)); + } + + if (ecc_count.new_ce_count) { + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u new correctable hardware errors detected in %s block\n", + seq_no, skt, aid, ecc_count.new_ce_count, blk_name(blk)); + RAS_DEV_INFO(ras_core->dev, + "{%llu} socket: %d, die: %d, %u correctable hardware errors detected in total in %s block\n", + seq_no, skt, aid, ecc_count.total_ce_count, blk_name(blk)); + } +} + +static void aca_bank_log(struct ras_core_context *ras_core, + int idx, int total, struct aca_bank_reg *bank, + struct aca_bank_ecc *bank_ecc) +{ + int i; + + RAS_DEV_INFO(ras_core->dev, + "{%llu}" RAS_HW_ERR "Accelerator Check Architecture events logged\n", + bank->seq_no); + /* plus 1 for output format, e.g: ACA[08/08]: xxxx */ + for (i = 0; i < ARRAY_SIZE(aca_regs); i++) + RAS_DEV_INFO(ras_core->dev, + "{%llu}" RAS_HW_ERR "ACA[%02d/%02d].%s=0x%016llx\n", + bank->seq_no, idx + 1, total, + aca_regs[i].name, bank->regs[aca_regs[i].reg_idx]); +} + +static void aca_log_bank_data(struct ras_core_context *ras_core, + struct aca_bank_reg *bank, struct aca_bank_ecc *bank_ecc, + struct ras_log_batch_tag *batch) +{ + if (bank_ecc->ue_count) + ras_log_ring_add_log_event(ras_core, RAS_LOG_EVENT_UE, bank->regs, batch); + else if (bank_ecc->de_count) + ras_log_ring_add_log_event(ras_core, RAS_LOG_EVENT_DE, bank->regs, batch); + else + ras_log_ring_add_log_event(ras_core, RAS_LOG_EVENT_CE, bank->regs, batch); +} + +static int aca_get_bank_count(struct ras_core_context *ras_core, + enum ras_err_type type, u32 *count) +{ + return ras_mp1_get_bank_count(ras_core, type, count); +} + +static bool aca_match_bank(struct aca_block *aca_blk, struct aca_bank_reg *bank) +{ + const struct aca_bank_hw_ops *bank_ops; + + if (!aca_blk->blk_info) + return false; + + bank_ops = &aca_blk->blk_info->bank_ops; + if (!bank_ops->bank_match) + return false; + + return bank_ops->bank_match(aca_blk, bank); +} + +static int aca_parse_bank(struct ras_core_context *ras_core, + struct aca_block *aca_blk, + struct aca_bank_reg *bank, + struct aca_bank_ecc *ecc) +{ + const struct aca_bank_hw_ops *bank_ops = &aca_blk->blk_info->bank_ops; + + if (!bank_ops || !bank_ops->bank_parse) + return -RAS_CORE_NOT_SUPPORTED; + + return bank_ops->bank_parse(ras_core, aca_blk, bank, ecc); +} + +static int aca_check_block_ecc_info(struct ras_core_context *ras_core, + struct aca_block *aca_blk, struct aca_ecc_info *info) +{ + if (info->socket_id >= aca_blk->ecc.socket_num_per_hive) { + RAS_DEV_ERR(ras_core->dev, + "Socket id (%d) is out of config! max:%u\n", + info->socket_id, aca_blk->ecc.socket_num_per_hive); + return -ENODATA; + } + + if (info->die_id >= aca_blk->ecc.socket[info->socket_id].aid_num) { + RAS_DEV_ERR(ras_core->dev, + "Die id (%d) is out of config! max:%u\n", + info->die_id, aca_blk->ecc.socket[info->socket_id].aid_num); + return -ENODATA; + } + + if ((aca_blk->blk_info->ras_block_id == RAS_BLOCK_ID__GFX) && + (info->xcd_id >= + aca_blk->ecc.socket[info->socket_id].aid[info->die_id].xcd.xcd_num)) { + RAS_DEV_ERR(ras_core->dev, + "Xcd id (%d) is out of config! max:%u\n", + info->xcd_id, + aca_blk->ecc.socket[info->socket_id].aid[info->die_id].xcd.xcd_num); + return -ENODATA; + } + + return 0; +} + +static int aca_log_bad_bank(struct ras_core_context *ras_core, + struct aca_block *aca_blk, struct aca_bank_reg *bank, + struct aca_bank_ecc *bank_ecc) +{ + struct aca_ecc_info *info; + struct aca_ecc_count *ecc_err; + struct aca_aid_ecc *aid_ecc; + int ret; + + info = &bank_ecc->bank_info; + + ret = aca_check_block_ecc_info(ras_core, aca_blk, info); + if (ret) + return ret; + + mutex_lock(&ras_core->ras_aca.aca_lock); + aid_ecc = &aca_blk->ecc.socket[info->socket_id].aid[info->die_id]; + if (aca_blk->blk_info->ras_block_id == RAS_BLOCK_ID__GFX) + ecc_err = &aid_ecc->xcd.xcd[info->xcd_id].ecc_err; + else + ecc_err = &aid_ecc->ecc_err; + + ecc_err->new_ce_count += bank_ecc->ce_count; + ecc_err->total_ce_count += bank_ecc->ce_count; + ecc_err->new_ue_count += bank_ecc->ue_count; + ecc_err->total_ue_count += bank_ecc->ue_count; + ecc_err->new_de_count += bank_ecc->de_count; + ecc_err->total_de_count += bank_ecc->de_count; + mutex_unlock(&ras_core->ras_aca.aca_lock); + + if ((aca_blk->blk_info->ras_block_id == RAS_BLOCK_ID__UMC) && + bank_ecc->de_count) { + struct ras_bank_ecc ras_ecc = {0}; + + ras_ecc.nps = ras_core_get_curr_nps_mode(ras_core); + ras_ecc.addr = bank_ecc->bank_info.addr; + ras_ecc.ipid = bank_ecc->bank_info.ipid; + ras_ecc.status = bank_ecc->bank_info.status; + ras_ecc.seq_no = bank->seq_no; + + if (ras_core_gpu_in_reset(ras_core)) + ras_umc_log_bad_bank_pending(ras_core, &ras_ecc); + else + ras_umc_log_bad_bank(ras_core, &ras_ecc); + } + + aca_report_ecc_info(ras_core, + bank->seq_no, aca_blk->blk_info->ras_block_id, info->socket_id, info->die_id, + &aca_blk->ecc.socket[info->socket_id].aid[info->die_id], bank_ecc); + + return 0; +} + +static struct aca_block *aca_get_bank_aca_block(struct ras_core_context *ras_core, + struct aca_bank_reg *bank) +{ + int i = 0; + + for (i = 0; i < RAS_BLOCK_ID__LAST; i++) + if (aca_match_bank(&ras_core->ras_aca.aca_blk[i], bank)) + return &ras_core->ras_aca.aca_blk[i]; + + return NULL; +} + +static int aca_dump_bank(struct ras_core_context *ras_core, u32 ecc_type, + int idx, void *data) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + int i, ret, reg_cnt; + + reg_cnt = min_t(int, 16, ARRAY_SIZE(bank->regs)); + for (i = 0; i < reg_cnt; i++) { + ret = ras_mp1_dump_bank(ras_core, ecc_type, idx, i, &bank->regs[i]); + if (ret) + return ret; + } + + return 0; +} + +static uint64_t aca_get_bank_seqno(struct ras_core_context *ras_core, + enum ras_err_type err_type, struct aca_block *aca_blk, + struct aca_bank_ecc *bank_ecc) +{ + uint64_t seq_no = 0; + + if (bank_ecc->de_count) { + if (aca_blk->blk_info->ras_block_id == RAS_BLOCK_ID__UMC) + seq_no = ras_core_get_seqno(ras_core, RAS_SEQNO_TYPE_DE, true); + else + seq_no = ras_core_get_seqno(ras_core, + RAS_SEQNO_TYPE_POISON_CONSUMPTION, true); + } else if (bank_ecc->ue_count) { + seq_no = ras_core_get_seqno(ras_core, RAS_SEQNO_TYPE_UE, true); + } else { + seq_no = ras_core_get_seqno(ras_core, RAS_SEQNO_TYPE_CE, true); + } + + return seq_no; +} + +static bool aca_dup_update_ue_in_fatal(struct ras_core_context *ras_core, + u32 ecc_type) +{ + struct ras_aca *aca = &ras_core->ras_aca; + + if (ecc_type != RAS_ERR_TYPE__UE) + return false; + + if (aca->ue_updated_mark & ACA_MARK_FATAL_FLAG) { + if (aca->ue_updated_mark & ACA_MARK_UE_READ_FLAG) + return true; + + aca->ue_updated_mark |= ACA_MARK_UE_READ_FLAG; + } + + return false; +} + +void ras_aca_mark_fatal_flag(struct ras_core_context *ras_core) +{ + struct ras_aca *aca = &ras_core->ras_aca; + + if (!aca) + return; + + aca->ue_updated_mark |= ACA_MARK_FATAL_FLAG; +} + +void ras_aca_clear_fatal_flag(struct ras_core_context *ras_core) +{ + struct ras_aca *aca = &ras_core->ras_aca; + + if (!aca) + return; + + if ((aca->ue_updated_mark & ACA_MARK_FATAL_FLAG) && + (aca->ue_updated_mark & ACA_MARK_UE_READ_FLAG)) + aca->ue_updated_mark = 0; +} + +static int aca_banks_update(struct ras_core_context *ras_core, + u32 ecc_type, void *data) +{ + struct aca_bank_reg bank; + struct aca_block *aca_blk; + struct aca_bank_ecc bank_ecc; + struct ras_log_batch_tag *batch_tag = NULL; + u32 count = 0; + int ret = 0; + int i; + + mutex_lock(&ras_core->ras_aca.bank_op_lock); + + if (aca_dup_update_ue_in_fatal(ras_core, ecc_type)) + goto out; + + ret = aca_get_bank_count(ras_core, ecc_type, &count); + if (ret) + goto out; + + if (!count) + goto out; + + batch_tag = ras_log_ring_create_batch_tag(ras_core); + for (i = 0; i < count; i++) { + memset(&bank, 0, sizeof(bank)); + ret = aca_dump_bank(ras_core, ecc_type, i, &bank); + if (ret) + break; + + bank.ecc_type = ecc_type; + + memset(&bank_ecc, 0, sizeof(bank_ecc)); + aca_blk = aca_get_bank_aca_block(ras_core, &bank); + if (aca_blk) + ret = aca_parse_bank(ras_core, aca_blk, &bank, &bank_ecc); + + bank.seq_no = aca_get_bank_seqno(ras_core, ecc_type, aca_blk, &bank_ecc); + + aca_log_bank_data(ras_core, &bank, &bank_ecc, batch_tag); + aca_bank_log(ras_core, i, count, &bank, &bank_ecc); + + if (!ret && aca_blk) + ret = aca_log_bad_bank(ras_core, aca_blk, &bank, &bank_ecc); + + if (ret) + break; + } + ras_log_ring_destroy_batch_tag(ras_core, batch_tag); + +out: + mutex_unlock(&ras_core->ras_aca.bank_op_lock); + return ret; +} + +int ras_aca_update_ecc(struct ras_core_context *ras_core, u32 type, void *data) +{ + /* Update aca bank to aca source error_cache first */ + return aca_banks_update(ras_core, type, data); +} + +static struct aca_block *ras_aca_get_block_handle(struct ras_core_context *ras_core, uint32_t blk) +{ + return &ras_core->ras_aca.aca_blk[blk]; +} + +static int ras_aca_clear_block_ecc_count(struct ras_core_context *ras_core, u32 blk) +{ + struct aca_block *aca_blk; + struct aca_aid_ecc *aid_ecc; + int skt, aid, xcd; + + mutex_lock(&ras_core->ras_aca.aca_lock); + aca_blk = ras_aca_get_block_handle(ras_core, blk); + for (skt = 0; skt < aca_blk->ecc.socket_num_per_hive; skt++) { + for (aid = 0; aid < aca_blk->ecc.socket[skt].aid_num; aid++) { + aid_ecc = &aca_blk->ecc.socket[skt].aid[aid]; + if (blk == RAS_BLOCK_ID__GFX) { + for (xcd = 0; xcd < aid_ecc->xcd.xcd_num; xcd++) + memset(&aid_ecc->xcd.xcd[xcd], + 0, sizeof(struct aca_xcd_ecc)); + } else { + memset(&aid_ecc->ecc_err, 0, sizeof(aid_ecc->ecc_err)); + } + } + } + mutex_unlock(&ras_core->ras_aca.aca_lock); + + return 0; +} + +int ras_aca_clear_all_blocks_ecc_count(struct ras_core_context *ras_core) +{ + enum ras_block_id blk; + int ret; + + for (blk = RAS_BLOCK_ID__UMC; blk < RAS_BLOCK_ID__LAST; blk++) { + ret = ras_aca_clear_block_ecc_count(ras_core, blk); + if (ret) + break; + } + + return ret; +} + +int ras_aca_clear_block_new_ecc_count(struct ras_core_context *ras_core, u32 blk) +{ + struct aca_block *aca_blk; + int skt, aid, xcd; + struct aca_ecc_count *ecc_err; + struct aca_aid_ecc *aid_ecc; + + mutex_lock(&ras_core->ras_aca.aca_lock); + aca_blk = ras_aca_get_block_handle(ras_core, blk); + for (skt = 0; skt < aca_blk->ecc.socket_num_per_hive; skt++) { + for (aid = 0; aid < aca_blk->ecc.socket[skt].aid_num; aid++) { + aid_ecc = &aca_blk->ecc.socket[skt].aid[aid]; + if (blk == RAS_BLOCK_ID__GFX) { + for (xcd = 0; xcd < aid_ecc->xcd.xcd_num; xcd++) { + ecc_err = &aid_ecc->xcd.xcd[xcd].ecc_err; + ecc_err->new_ce_count = 0; + ecc_err->new_ue_count = 0; + ecc_err->new_de_count = 0; + } + } else { + ecc_err = &aid_ecc->ecc_err; + ecc_err->new_ce_count = 0; + ecc_err->new_ue_count = 0; + ecc_err->new_de_count = 0; + } + } + } + mutex_unlock(&ras_core->ras_aca.aca_lock); + + return 0; +} + +static int ras_aca_get_block_each_aid_ecc_count(struct ras_core_context *ras_core, + u32 blk, u32 skt, u32 aid, u32 xcd, + struct aca_ecc_count *ecc_count) +{ + struct aca_block *aca_blk; + struct aca_ecc_count *ecc_err; + + aca_blk = ras_aca_get_block_handle(ras_core, blk); + if (blk == RAS_BLOCK_ID__GFX) + ecc_err = &aca_blk->ecc.socket[skt].aid[aid].xcd.xcd[xcd].ecc_err; + else + ecc_err = &aca_blk->ecc.socket[skt].aid[aid].ecc_err; + + ecc_count->new_ce_count = ecc_err->new_ce_count; + ecc_count->total_ce_count = ecc_err->total_ce_count; + ecc_count->new_ue_count = ecc_err->new_ue_count; + ecc_count->total_ue_count = ecc_err->total_ue_count; + ecc_count->new_de_count = ecc_err->new_de_count; + ecc_count->total_de_count = ecc_err->total_de_count; + + return 0; +} + +static inline void _add_ecc_count(struct aca_ecc_count *des, struct aca_ecc_count *src) +{ + des->new_ce_count += src->new_ce_count; + des->total_ce_count += src->total_ce_count; + des->new_ue_count += src->new_ue_count; + des->total_ue_count += src->total_ue_count; + des->new_de_count += src->new_de_count; + des->total_de_count += src->total_de_count; +} + +static const struct ras_aca_ip_func *aca_get_ip_func( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(1, 0, 0): + return &ras_aca_func_v1_0; + default: + RAS_DEV_ERR(ras_core->dev, + "ACA ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +int ras_aca_get_block_ecc_count(struct ras_core_context *ras_core, + u32 blk, void *data) +{ + struct ras_ecc_count *err_data = (struct ras_ecc_count *)data; + struct aca_block *aca_blk; + int skt, aid, xcd; + struct aca_ecc_count ecc_xcd; + struct aca_ecc_count ecc_aid; + struct aca_ecc_count ecc; + + if (blk >= RAS_BLOCK_ID__LAST) + return -EINVAL; + + if (!err_data) + return -EINVAL; + + aca_blk = ras_aca_get_block_handle(ras_core, blk); + memset(&ecc, 0, sizeof(ecc)); + + mutex_lock(&ras_core->ras_aca.aca_lock); + if (blk == RAS_BLOCK_ID__GFX) { + for (skt = 0; skt < aca_blk->ecc.socket_num_per_hive; skt++) { + for (aid = 0; aid < aca_blk->ecc.socket[skt].aid_num; aid++) { + memset(&ecc_aid, 0, sizeof(ecc_aid)); + for (xcd = 0; + xcd < aca_blk->ecc.socket[skt].aid[aid].xcd.xcd_num; + xcd++) { + memset(&ecc_xcd, 0, sizeof(ecc_xcd)); + if (ras_aca_get_block_each_aid_ecc_count(ras_core, + blk, skt, aid, xcd, &ecc_xcd)) + continue; + _add_ecc_count(&ecc_aid, &ecc_xcd); + } + _add_ecc_count(&ecc, &ecc_aid); + } + } + } else { + for (skt = 0; skt < aca_blk->ecc.socket_num_per_hive; skt++) { + for (aid = 0; aid < aca_blk->ecc.socket[skt].aid_num; aid++) { + memset(&ecc_aid, 0, sizeof(ecc_aid)); + if (ras_aca_get_block_each_aid_ecc_count(ras_core, + blk, skt, aid, 0, &ecc_aid)) + continue; + _add_ecc_count(&ecc, &ecc_aid); + } + } + } + + err_data->new_ce_count = ecc.new_ce_count; + err_data->total_ce_count = ecc.total_ce_count; + err_data->new_ue_count = ecc.new_ue_count; + err_data->total_ue_count = ecc.total_ue_count; + err_data->new_de_count = ecc.new_de_count; + err_data->total_de_count = ecc.total_de_count; + mutex_unlock(&ras_core->ras_aca.aca_lock); + + return 0; +} + +int ras_aca_sw_init(struct ras_core_context *ras_core) +{ + struct ras_aca *ras_aca = &ras_core->ras_aca; + struct ras_aca_config *aca_cfg = &ras_core->config->aca_cfg; + struct aca_block *aca_blk; + uint32_t socket_num_per_hive; + uint32_t aid_num_per_socket; + uint32_t xcd_num_per_aid; + int blk, skt, aid; + + socket_num_per_hive = aca_cfg->socket_num_per_hive; + aid_num_per_socket = aca_cfg->aid_num_per_socket; + xcd_num_per_aid = aca_cfg->xcd_num_per_aid; + + if (!xcd_num_per_aid || !aid_num_per_socket || + (socket_num_per_hive > MAX_SOCKET_NUM_PER_HIVE) || + (aid_num_per_socket > MAX_AID_NUM_PER_SOCKET) || + (xcd_num_per_aid > MAX_XCD_NUM_PER_AID)) { + RAS_DEV_ERR(ras_core->dev, "Invalid ACA system configuration: %d, %d, %d\n", + socket_num_per_hive, aid_num_per_socket, xcd_num_per_aid); + return -EINVAL; + } + + memset(ras_aca, 0, sizeof(*ras_aca)); + + for (blk = 0; blk < RAS_BLOCK_ID__LAST; blk++) { + aca_blk = &ras_aca->aca_blk[blk]; + aca_blk->ecc.socket_num_per_hive = socket_num_per_hive; + for (skt = 0; skt < aca_blk->ecc.socket_num_per_hive; skt++) { + aca_blk->ecc.socket[skt].aid_num = aid_num_per_socket; + if (blk == RAS_BLOCK_ID__GFX) { + for (aid = 0; aid < aca_blk->ecc.socket[skt].aid_num; aid++) + aca_blk->ecc.socket[skt].aid[aid].xcd.xcd_num = + xcd_num_per_aid; + } + } + } + + mutex_init(&ras_aca->aca_lock); + mutex_init(&ras_aca->bank_op_lock); + + return 0; +} + +int ras_aca_sw_fini(struct ras_core_context *ras_core) +{ + struct ras_aca *ras_aca = &ras_core->ras_aca; + + mutex_destroy(&ras_aca->aca_lock); + mutex_destroy(&ras_aca->bank_op_lock); + + return 0; +} + +int ras_aca_hw_init(struct ras_core_context *ras_core) +{ + struct ras_aca *ras_aca = &ras_core->ras_aca; + struct aca_block *aca_blk; + const struct ras_aca_ip_func *ip_func; + int i; + + ras_aca->aca_ip_version = ras_core->config->aca_ip_version; + ip_func = aca_get_ip_func(ras_core, ras_aca->aca_ip_version); + if (!ip_func) + return -EINVAL; + + for (i = 0; i < ip_func->block_num; i++) { + aca_blk = &ras_aca->aca_blk[ip_func->block_info[i]->ras_block_id]; + aca_blk->blk_info = ip_func->block_info[i]; + } + + ras_aca->ue_updated_mark = 0; + + return 0; +} + +int ras_aca_hw_fini(struct ras_core_context *ras_core) +{ + struct ras_aca *ras_aca = &ras_core->ras_aca; + + ras_aca->ue_updated_mark = 0; + + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_aca.h b/drivers/gpu/drm/amd/ras/rascore/ras_aca.h new file mode 100644 index 000000000000..f61b02a5f0fc --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_aca.h @@ -0,0 +1,164 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_ACA_H__ +#define __RAS_ACA_H__ +#include "ras.h" + +#define MAX_SOCKET_NUM_PER_HIVE 8 +#define MAX_AID_NUM_PER_SOCKET 4 +#define MAX_XCD_NUM_PER_AID 2 +#define MAX_ACA_RAS_BLOCK 20 + +#define ACA_ERROR__UE_MASK (0x1 << RAS_ERR_TYPE__UE) +#define ACA_ERROR__CE_MASK (0x1 << RAS_ERR_TYPE__CE) +#define ACA_ERROR__DE_MASK (0x1 << RAS_ERR_TYPE__DE) + +enum ras_aca_reg_idx { + ACA_REG_IDX__CTL = 0, + ACA_REG_IDX__STATUS = 1, + ACA_REG_IDX__ADDR = 2, + ACA_REG_IDX__MISC0 = 3, + ACA_REG_IDX__CONFG = 4, + ACA_REG_IDX__IPID = 5, + ACA_REG_IDX__SYND = 6, + ACA_REG_IDX__DESTAT = 8, + ACA_REG_IDX__DEADDR = 9, + ACA_REG_IDX__CTL_MASK = 10, + ACA_REG_MAX_COUNT = 16, +}; + +struct ras_core_context; +struct aca_block; + +struct aca_bank_reg { + u32 ecc_type; + u64 seq_no; + u64 regs[ACA_REG_MAX_COUNT]; +}; + +enum aca_ecc_hwip { + ACA_ECC_HWIP__UNKNOWN = -1, + ACA_ECC_HWIP__PSP = 0, + ACA_ECC_HWIP__UMC, + ACA_ECC_HWIP__SMU, + ACA_ECC_HWIP__PCS_XGMI, + ACA_ECC_HWIP_COUNT, +}; + +struct aca_ecc_info { + int die_id; + int socket_id; + int xcd_id; + int hwid; + int mcatype; + uint64_t status; + uint64_t ipid; + uint64_t addr; +}; + +struct aca_bank_ecc { + struct aca_ecc_info bank_info; + u32 ce_count; + u32 ue_count; + u32 de_count; +}; + +struct aca_ecc_count { + u32 new_ce_count; + u32 total_ce_count; + u32 new_ue_count; + u32 total_ue_count; + u32 new_de_count; + u32 total_de_count; +}; + +struct aca_xcd_ecc { + struct aca_ecc_count ecc_err; +}; + +struct aca_aid_ecc { + union { + struct aca_xcd { + struct aca_xcd_ecc xcd[MAX_XCD_NUM_PER_AID]; + u32 xcd_num; + } xcd; + struct aca_ecc_count ecc_err; + }; +}; + +struct aca_socket_ecc { + struct aca_aid_ecc aid[MAX_AID_NUM_PER_SOCKET]; + u32 aid_num; +}; + +struct aca_block_ecc { + struct aca_socket_ecc socket[MAX_SOCKET_NUM_PER_HIVE]; + u32 socket_num_per_hive; +}; + +struct aca_bank_hw_ops { + bool (*bank_match)(struct aca_block *ras_blk, void *data); + int (*bank_parse)(struct ras_core_context *ras_core, + struct aca_block *aca_blk, void *data, void *buf); +}; + +struct aca_block_info { + char name[32]; + u32 ras_block_id; + enum aca_ecc_hwip hwip; + struct aca_bank_hw_ops bank_ops; + u32 mask; +}; + +struct aca_block { + const struct aca_block_info *blk_info; + struct aca_block_ecc ecc; +}; + +struct ras_aca_ip_func { + uint32_t block_num; + const struct aca_block_info **block_info; +}; + +struct ras_aca { + uint32_t aca_ip_version; + const struct ras_aca_ip_func *ip_func; + struct mutex aca_lock; + struct mutex bank_op_lock; + struct aca_block aca_blk[MAX_ACA_RAS_BLOCK]; + uint32_t ue_updated_mark; +}; + +int ras_aca_sw_init(struct ras_core_context *ras_core); +int ras_aca_sw_fini(struct ras_core_context *ras_core); +int ras_aca_hw_init(struct ras_core_context *ras_core); +int ras_aca_hw_fini(struct ras_core_context *ras_core); +int ras_aca_get_block_ecc_count(struct ras_core_context *ras_core, u32 blk, void *data); +int ras_aca_clear_block_new_ecc_count(struct ras_core_context *ras_core, u32 blk); +int ras_aca_clear_all_blocks_ecc_count(struct ras_core_context *ras_core); +int ras_aca_update_ecc(struct ras_core_context *ras_core, u32 ecc_type, void *data); +void ras_aca_mark_fatal_flag(struct ras_core_context *ras_core); +void ras_aca_clear_fatal_flag(struct ras_core_context *ras_core); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.c b/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.c new file mode 100644 index 000000000000..29df98948703 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.c @@ -0,0 +1,379 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_aca.h" +#include "ras_core_status.h" +#include "ras_aca_v1_0.h" + +struct ras_aca_hwip { + int hwid; + int mcatype; +}; + +static struct ras_aca_hwip aca_hwid_mcatypes[ACA_ECC_HWIP_COUNT] = { + [ACA_ECC_HWIP__SMU] = {0x01, 0x01}, + [ACA_ECC_HWIP__PCS_XGMI] = {0x50, 0x00}, + [ACA_ECC_HWIP__UMC] = {0x96, 0x00}, +}; + +static int aca_decode_bank_info(struct aca_block *aca_blk, + struct aca_bank_reg *bank, struct aca_ecc_info *info) +{ + u64 ipid; + u32 instidhi, instidlo; + + ipid = bank->regs[ACA_REG_IDX__IPID]; + info->hwid = ACA_REG_IPID_HARDWAREID(ipid); + info->mcatype = ACA_REG_IPID_MCATYPE(ipid); + /* + * Unified DieID Format: SAASS. A:AID, S:Socket. + * Unified DieID[4:4] = InstanceId[0:0] + * Unified DieID[0:3] = InstanceIdHi[0:3] + */ + instidhi = ACA_REG_IPID_INSTANCEIDHI(ipid); + instidlo = ACA_REG_IPID_INSTANCEIDLO(ipid); + info->die_id = ((instidhi >> 2) & 0x03); + info->socket_id = ((instidlo & 0x1) << 2) | (instidhi & 0x03); + + if ((aca_blk->blk_info->hwip == ACA_ECC_HWIP__SMU) && + (aca_blk->blk_info->ras_block_id == RAS_BLOCK_ID__GFX)) + info->xcd_id = + ((instidlo & GENMASK_ULL(31, 1)) == mmSMNAID_XCD0_MCA_SMU) ? 0 : 1; + + return 0; +} + +static bool aca_check_bank_hwip(struct aca_bank_reg *bank, enum aca_ecc_hwip type) +{ + struct ras_aca_hwip *hwip; + int hwid, mcatype; + u64 ipid; + + if (!bank || (type == ACA_ECC_HWIP__UNKNOWN)) + return false; + + hwip = &aca_hwid_mcatypes[type]; + if (!hwip->hwid) + return false; + + ipid = bank->regs[ACA_REG_IDX__IPID]; + hwid = ACA_REG_IPID_HARDWAREID(ipid); + mcatype = ACA_REG_IPID_MCATYPE(ipid); + + return hwip->hwid == hwid && hwip->mcatype == mcatype; +} + +static bool aca_match_bank_default(struct aca_block *aca_blk, void *data) +{ + return aca_check_bank_hwip((struct aca_bank_reg *)data, aca_blk->blk_info->hwip); +} + +static bool aca_match_gfx_bank(struct aca_block *aca_blk, void *data) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + u32 instlo; + + if (!aca_check_bank_hwip(bank, aca_blk->blk_info->hwip)) + return false; + + instlo = ACA_REG_IPID_INSTANCEIDLO(bank->regs[ACA_REG_IDX__IPID]); + instlo &= GENMASK_ULL(31, 1); + switch (instlo) { + case mmSMNAID_XCD0_MCA_SMU: + case mmSMNAID_XCD1_MCA_SMU: + case mmSMNXCD_XCD0_MCA_SMU: + return true; + default: + break; + } + + return false; +} + +static bool aca_match_sdma_bank(struct aca_block *aca_blk, void *data) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + /* CODE_SDMA0 - CODE_SDMA4, reference to smu driver if header file */ + static int sdma_err_codes[] = { 33, 34, 35, 36 }; + u32 instlo; + int errcode, i; + + if (!aca_check_bank_hwip(bank, aca_blk->blk_info->hwip)) + return false; + + instlo = ACA_REG_IPID_INSTANCEIDLO(bank->regs[ACA_REG_IDX__IPID]); + instlo &= GENMASK_ULL(31, 1); + if (instlo != mmSMNAID_AID0_MCA_SMU) + return false; + + errcode = ACA_REG_SYND_ERRORINFORMATION(bank->regs[ACA_REG_IDX__SYND]); + errcode &= 0xff; + + /* Check SDMA error codes */ + for (i = 0; i < ARRAY_SIZE(sdma_err_codes); i++) { + if (errcode == sdma_err_codes[i]) + return true; + } + + return false; +} + +static bool aca_match_mmhub_bank(struct aca_block *aca_blk, void *data) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + /* reference to smu driver if header file */ + const int mmhub_err_codes[] = { + 0, 1, 2, 3, 4, /* CODE_DAGB0 - 4 */ + 5, 6, 7, 8, 9, /* CODE_EA0 - 4 */ + 10, /* CODE_UTCL2_ROUTER */ + 11, /* CODE_VML2 */ + 12, /* CODE_VML2_WALKER */ + 13, /* CODE_MMCANE */ + }; + u32 instlo; + int errcode, i; + + if (!aca_check_bank_hwip(bank, aca_blk->blk_info->hwip)) + return false; + + instlo = ACA_REG_IPID_INSTANCEIDLO(bank->regs[ACA_REG_IDX__IPID]); + instlo &= GENMASK_ULL(31, 1); + if (instlo != mmSMNAID_AID0_MCA_SMU) + return false; + + errcode = ACA_REG_SYND_ERRORINFORMATION(bank->regs[ACA_REG_IDX__SYND]); + errcode &= 0xff; + + /* Check MMHUB error codes */ + for (i = 0; i < ARRAY_SIZE(mmhub_err_codes); i++) { + if (errcode == mmhub_err_codes[i]) + return true; + } + + return false; +} + +static bool aca_check_umc_de(struct ras_core_context *ras_core, uint64_t mc_umc_status) +{ + return (ras_core->poison_supported && + ACA_REG_STATUS_VAL(mc_umc_status) && + ACA_REG_STATUS_DEFERRED(mc_umc_status)); +} + +static bool aca_check_umc_ue(struct ras_core_context *ras_core, uint64_t mc_umc_status) +{ + if (aca_check_umc_de(ras_core, mc_umc_status)) + return false; + + return (ACA_REG_STATUS_VAL(mc_umc_status) && + (ACA_REG_STATUS_PCC(mc_umc_status) || + ACA_REG_STATUS_UC(mc_umc_status) || + ACA_REG_STATUS_TCC(mc_umc_status))); +} + +static bool aca_check_umc_ce(struct ras_core_context *ras_core, uint64_t mc_umc_status) +{ + if (aca_check_umc_de(ras_core, mc_umc_status)) + return false; + + return (ACA_REG_STATUS_VAL(mc_umc_status) && + (ACA_REG_STATUS_CECC(mc_umc_status) || + (ACA_REG_STATUS_UECC(mc_umc_status) && + ACA_REG_STATUS_UC(mc_umc_status) == 0) || + /* Identify data parity error in replay mode */ + ((ACA_REG_STATUS_ERRORCODEEXT(mc_umc_status) == 0x5 || + ACA_REG_STATUS_ERRORCODEEXT(mc_umc_status) == 0xb) && + !(aca_check_umc_ue(ras_core, mc_umc_status))))); +} + +static int aca_parse_umc_bank(struct ras_core_context *ras_core, + struct aca_block *ras_blk, void *data, void *buf) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + struct aca_bank_ecc *ecc = (struct aca_bank_ecc *)buf; + struct aca_ecc_info bank_info; + uint32_t ext_error_code; + uint64_t status0; + + status0 = bank->regs[ACA_REG_IDX__STATUS]; + if (!ACA_REG_STATUS_VAL(status0)) + return 0; + + memset(&bank_info, 0, sizeof(bank_info)); + aca_decode_bank_info(ras_blk, bank, &bank_info); + memcpy(&ecc->bank_info, &bank_info, sizeof(bank_info)); + ecc->bank_info.status = bank->regs[ACA_REG_IDX__STATUS]; + ecc->bank_info.ipid = bank->regs[ACA_REG_IDX__IPID]; + ecc->bank_info.addr = bank->regs[ACA_REG_IDX__ADDR]; + + ext_error_code = ACA_REG_STATUS_ERRORCODEEXT(status0); + + if (aca_check_umc_de(ras_core, status0)) + ecc->de_count = 1; + else if (aca_check_umc_ue(ras_core, status0)) + ecc->ue_count = ext_error_code ? + 1 : ACA_REG_MISC0_ERRCNT(bank->regs[ACA_REG_IDX__MISC0]); + else if (aca_check_umc_ce(ras_core, status0)) + ecc->ce_count = ext_error_code ? + 1 : ACA_REG_MISC0_ERRCNT(bank->regs[ACA_REG_IDX__MISC0]); + + return 0; +} + +static bool aca_check_bank_is_de(struct ras_core_context *ras_core, + uint64_t status) +{ + return (ACA_REG_STATUS_POISON(status) || + ACA_REG_STATUS_DEFERRED(status)); +} + +static int aca_parse_bank_default(struct ras_core_context *ras_core, + struct aca_block *ras_blk, + void *data, void *buf) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + struct aca_bank_ecc *ecc = (struct aca_bank_ecc *)buf; + struct aca_ecc_info bank_info; + u64 misc0 = bank->regs[ACA_REG_IDX__MISC0]; + u64 status = bank->regs[ACA_REG_IDX__STATUS]; + + memset(&bank_info, 0, sizeof(bank_info)); + aca_decode_bank_info(ras_blk, bank, &bank_info); + memcpy(&ecc->bank_info, &bank_info, sizeof(bank_info)); + ecc->bank_info.status = status; + ecc->bank_info.ipid = bank->regs[ACA_REG_IDX__IPID]; + ecc->bank_info.addr = bank->regs[ACA_REG_IDX__ADDR]; + + if (aca_check_bank_is_de(ras_core, status)) { + ecc->de_count = 1; + } else { + if (bank->ecc_type == RAS_ERR_TYPE__UE) + ecc->ue_count = 1; + else if (bank->ecc_type == RAS_ERR_TYPE__CE) + ecc->ce_count = ACA_REG_MISC0_ERRCNT(misc0); + } + + return 0; +} + +static int aca_parse_xgmi_bank(struct ras_core_context *ras_core, + struct aca_block *ras_blk, + void *data, void *buf) +{ + struct aca_bank_reg *bank = (struct aca_bank_reg *)data; + struct aca_bank_ecc *ecc = (struct aca_bank_ecc *)buf; + struct aca_ecc_info bank_info; + u64 status, count; + int ext_error_code; + + memset(&bank_info, 0, sizeof(bank_info)); + aca_decode_bank_info(ras_blk, bank, &bank_info); + memcpy(&ecc->bank_info, &bank_info, sizeof(bank_info)); + ecc->bank_info.status = bank->regs[ACA_REG_IDX__STATUS]; + ecc->bank_info.ipid = bank->regs[ACA_REG_IDX__IPID]; + ecc->bank_info.addr = bank->regs[ACA_REG_IDX__ADDR]; + + status = bank->regs[ACA_REG_IDX__STATUS]; + ext_error_code = ACA_REG_STATUS_ERRORCODEEXT(status); + + count = ACA_REG_MISC0_ERRCNT(bank->regs[ACA_REG_IDX__MISC0]); + if (bank->ecc_type == RAS_ERR_TYPE__UE) { + if (ext_error_code != 0 && ext_error_code != 9) + count = 0ULL; + ecc->ue_count = count; + } else if (bank->ecc_type == RAS_ERR_TYPE__CE) { + count = ext_error_code == 6 ? count : 0ULL; + ecc->ce_count = count; + } + + return 0; +} + +static const struct aca_block_info aca_v1_0_umc = { + .name = "umc", + .ras_block_id = RAS_BLOCK_ID__UMC, + .hwip = ACA_ECC_HWIP__UMC, + .mask = ACA_ERROR__UE_MASK | ACA_ERROR__CE_MASK | ACA_ERROR__DE_MASK, + .bank_ops = { + .bank_match = aca_match_bank_default, + .bank_parse = aca_parse_umc_bank, + }, +}; + +static const struct aca_block_info aca_v1_0_gfx = { + .name = "gfx", + .ras_block_id = RAS_BLOCK_ID__GFX, + .hwip = ACA_ECC_HWIP__SMU, + .mask = ACA_ERROR__UE_MASK | ACA_ERROR__CE_MASK, + .bank_ops = { + .bank_match = aca_match_gfx_bank, + .bank_parse = aca_parse_bank_default, + }, +}; + +static const struct aca_block_info aca_v1_0_sdma = { + .name = "sdma", + .ras_block_id = RAS_BLOCK_ID__SDMA, + .hwip = ACA_ECC_HWIP__SMU, + .mask = ACA_ERROR__UE_MASK, + .bank_ops = { + .bank_match = aca_match_sdma_bank, + .bank_parse = aca_parse_bank_default, + }, +}; + +static const struct aca_block_info aca_v1_0_mmhub = { + .name = "mmhub", + .ras_block_id = RAS_BLOCK_ID__MMHUB, + .hwip = ACA_ECC_HWIP__SMU, + .mask = ACA_ERROR__UE_MASK, + .bank_ops = { + .bank_match = aca_match_mmhub_bank, + .bank_parse = aca_parse_bank_default, + }, +}; + +static const struct aca_block_info aca_v1_0_xgmi = { + .name = "xgmi", + .ras_block_id = RAS_BLOCK_ID__XGMI_WAFL, + .hwip = ACA_ECC_HWIP__PCS_XGMI, + .mask = ACA_ERROR__UE_MASK | ACA_ERROR__CE_MASK, + .bank_ops = { + .bank_match = aca_match_bank_default, + .bank_parse = aca_parse_xgmi_bank, + }, +}; + +static const struct aca_block_info *aca_block_info_v1_0[] = { + &aca_v1_0_umc, + &aca_v1_0_gfx, + &aca_v1_0_sdma, + &aca_v1_0_mmhub, + &aca_v1_0_xgmi, +}; + +const struct ras_aca_ip_func ras_aca_func_v1_0 = { + .block_num = ARRAY_SIZE(aca_block_info_v1_0), + .block_info = aca_block_info_v1_0, +}; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.h b/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.h new file mode 100644 index 000000000000..40e5d94b037f --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_aca_v1_0.h @@ -0,0 +1,71 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_ACA_V1_0_H__ +#define __RAS_ACA_V1_0_H__ +#include "ras.h" + +#define ACA__REG__FIELD(x, h, l) (((x) & GENMASK_ULL(h, l)) >> l) +#define ACA_REG_STATUS_VAL(x) ACA__REG__FIELD(x, 63, 63) +#define ACA_REG_STATUS_OVERFLOW(x) ACA__REG__FIELD(x, 62, 62) +#define ACA_REG_STATUS_UC(x) ACA__REG__FIELD(x, 61, 61) +#define ACA_REG_STATUS_EN(x) ACA__REG__FIELD(x, 60, 60) +#define ACA_REG_STATUS_MISCV(x) ACA__REG__FIELD(x, 59, 59) +#define ACA_REG_STATUS_ADDRV(x) ACA__REG__FIELD(x, 58, 58) +#define ACA_REG_STATUS_PCC(x) ACA__REG__FIELD(x, 57, 57) +#define ACA_REG_STATUS_ERRCOREIDVAL(x) ACA__REG__FIELD(x, 56, 56) +#define ACA_REG_STATUS_TCC(x) ACA__REG__FIELD(x, 55, 55) +#define ACA_REG_STATUS_SYNDV(x) ACA__REG__FIELD(x, 53, 53) +#define ACA_REG_STATUS_CECC(x) ACA__REG__FIELD(x, 46, 46) +#define ACA_REG_STATUS_UECC(x) ACA__REG__FIELD(x, 45, 45) +#define ACA_REG_STATUS_DEFERRED(x) ACA__REG__FIELD(x, 44, 44) +#define ACA_REG_STATUS_POISON(x) ACA__REG__FIELD(x, 43, 43) +#define ACA_REG_STATUS_SCRUB(x) ACA__REG__FIELD(x, 40, 40) +#define ACA_REG_STATUS_ERRCOREID(x) ACA__REG__FIELD(x, 37, 32) +#define ACA_REG_STATUS_ADDRLSB(x) ACA__REG__FIELD(x, 29, 24) +#define ACA_REG_STATUS_ERRORCODEEXT(x) ACA__REG__FIELD(x, 21, 16) +#define ACA_REG_STATUS_ERRORCODE(x) ACA__REG__FIELD(x, 15, 0) + +#define ACA_REG_IPID_MCATYPE(x) ACA__REG__FIELD(x, 63, 48) +#define ACA_REG_IPID_INSTANCEIDHI(x) ACA__REG__FIELD(x, 47, 44) +#define ACA_REG_IPID_HARDWAREID(x) ACA__REG__FIELD(x, 43, 32) +#define ACA_REG_IPID_INSTANCEIDLO(x) ACA__REG__FIELD(x, 31, 0) + +#define ACA_REG_MISC0_VALID(x) ACA__REG__FIELD(x, 63, 63) +#define ACA_REG_MISC0_OVRFLW(x) ACA__REG__FIELD(x, 48, 48) +#define ACA_REG_MISC0_ERRCNT(x) ACA__REG__FIELD(x, 43, 32) + +#define ACA_REG_SYND_ERRORINFORMATION(x) ACA__REG__FIELD(x, 17, 0) + +/* NOTE: The following codes refers to the smu header file */ +#define ACA_EXTERROR_CODE_CE 0x3a +#define ACA_EXTERROR_CODE_FAULT 0x3b + +#define mmSMNAID_XCD0_MCA_SMU 0x36430400 /* SMN AID XCD0 */ +#define mmSMNAID_XCD1_MCA_SMU 0x38430400 /* SMN AID XCD1 */ +#define mmSMNXCD_XCD0_MCA_SMU 0x40430400 /* SMN XCD XCD0 */ +#define mmSMNAID_AID0_MCA_SMU 0x03b30400 /* SMN AID AID0 */ + +extern const struct ras_aca_ip_func ras_aca_func_v1_0; +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_cmd.c b/drivers/gpu/drm/amd/ras/rascore/ras_cmd.c new file mode 100644 index 000000000000..94e6d7420d94 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_cmd.c @@ -0,0 +1,522 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_cmd.h" + +#define RAS_CMD_MAJOR_VERSION 6 +#define RAS_CMD_MINOR_VERSION 0 +#define RAS_CMD_VERSION (((RAS_CMD_MAJOR_VERSION) << 10) | (RAS_CMD_MINOR_VERSION)) + +static int ras_cmd_add_device(struct ras_core_context *ras_core) +{ + INIT_LIST_HEAD(&ras_core->ras_cmd.head); + ras_core->ras_cmd.ras_core = ras_core; + ras_core->ras_cmd.dev_handle = (uintptr_t)ras_core ^ RAS_CMD_DEV_HANDLE_MAGIC; + return 0; +} + +static int ras_cmd_remove_device(struct ras_core_context *ras_core) +{ + memset(&ras_core->ras_cmd, 0, sizeof(ras_core->ras_cmd)); + return 0; +} + +static int ras_get_block_ecc_info(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_block_ecc_info_req *input_data = + (struct ras_cmd_block_ecc_info_req *)cmd->input_buff_raw; + struct ras_cmd_block_ecc_info_rsp *output_data = + (struct ras_cmd_block_ecc_info_rsp *)cmd->output_buff_raw; + struct ras_ecc_count err_data; + int ret; + + if (cmd->input_size != sizeof(struct ras_cmd_block_ecc_info_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + memset(&err_data, 0, sizeof(err_data)); + ret = ras_aca_get_block_ecc_count(ras_core, input_data->block_id, &err_data); + if (ret) + return RAS_CMD__ERROR_GENERIC; + + output_data->ce_count = err_data.total_ce_count; + output_data->ue_count = err_data.total_ue_count; + output_data->de_count = err_data.total_de_count; + + cmd->output_size = sizeof(struct ras_cmd_block_ecc_info_rsp); + return RAS_CMD__SUCCESS; +} + +static void ras_cmd_update_bad_page_info(struct ras_cmd_bad_page_record *ras_cmd_record, + struct eeprom_umc_record *record) +{ + ras_cmd_record->retired_page = record->cur_nps_retired_row_pfn; + ras_cmd_record->ts = record->ts; + ras_cmd_record->err_type = record->err_type; + ras_cmd_record->mem_channel = record->mem_channel; + ras_cmd_record->mcumc_id = record->mcumc_id; + ras_cmd_record->address = record->address; + ras_cmd_record->bank = record->bank; + ras_cmd_record->valid = 1; +} + +static int ras_cmd_get_group_bad_pages(struct ras_core_context *ras_core, + uint32_t group_index, struct ras_cmd_bad_pages_info_rsp *output_data) +{ + struct eeprom_umc_record record; + struct ras_cmd_bad_page_record *ras_cmd_record; + uint32_t i = 0, bp_cnt = 0, group_cnt = 0; + + output_data->bp_in_group = 0; + output_data->group_index = 0; + + bp_cnt = ras_umc_get_badpage_count(ras_core); + if (bp_cnt) { + output_data->group_index = group_index; + group_cnt = bp_cnt / RAS_CMD_MAX_BAD_PAGES_PER_GROUP + + ((bp_cnt % RAS_CMD_MAX_BAD_PAGES_PER_GROUP) ? 1 : 0); + + if (group_index >= group_cnt) + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + + i = group_index * RAS_CMD_MAX_BAD_PAGES_PER_GROUP; + for (; + i < bp_cnt && output_data->bp_in_group < RAS_CMD_MAX_BAD_PAGES_PER_GROUP; + i++) { + if (ras_umc_get_badpage_record(ras_core, i, &record)) + return RAS_CMD__ERROR_GENERIC; + + ras_cmd_record = &output_data->records[i % RAS_CMD_MAX_BAD_PAGES_PER_GROUP]; + + memset(ras_cmd_record, 0, sizeof(*ras_cmd_record)); + ras_cmd_update_bad_page_info(ras_cmd_record, &record); + output_data->bp_in_group++; + } + } + output_data->bp_total_cnt = bp_cnt; + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_get_bad_pages(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_bad_pages_info_req *input_data = + (struct ras_cmd_bad_pages_info_req *)cmd->input_buff_raw; + struct ras_cmd_bad_pages_info_rsp *output_data = + (struct ras_cmd_bad_pages_info_rsp *)cmd->output_buff_raw; + int ret; + + if (cmd->input_size != sizeof(struct ras_cmd_bad_pages_info_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + ret = ras_cmd_get_group_bad_pages(ras_core, input_data->group_index, output_data); + if (ret) + return RAS_CMD__ERROR_GENERIC; + + output_data->version = 0; + + cmd->output_size = sizeof(struct ras_cmd_bad_pages_info_rsp); + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_clear_bad_page_info(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + if (cmd->input_size != sizeof(struct ras_cmd_dev_handle)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + if (ras_eeprom_reset_table(ras_core)) + return RAS_CMD__ERROR_GENERIC; + + if (ras_umc_clean_badpage_data(ras_core)) + return RAS_CMD__ERROR_GENERIC; + + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_reset_all_error_counts(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + if (cmd->input_size != sizeof(struct ras_cmd_dev_handle)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + if (ras_aca_clear_all_blocks_ecc_count(ras_core)) + return RAS_CMD__ERROR_GENERIC; + + if (ras_umc_clear_logged_ecc(ras_core)) + return RAS_CMD__ERROR_GENERIC; + + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_get_cper_snapshot(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_cper_snapshot_rsp *output_data = + (struct ras_cmd_cper_snapshot_rsp *)cmd->output_buff_raw; + struct ras_log_batch_overview overview; + + if (cmd->input_size != sizeof(struct ras_cmd_cper_snapshot_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + ras_log_ring_get_batch_overview(ras_core, &overview); + + output_data->total_cper_num = overview.logged_batch_count; + output_data->start_cper_id = overview.first_batch_id; + output_data->latest_cper_id = overview.last_batch_id; + + output_data->version = 0; + + cmd->output_size = sizeof(struct ras_cmd_cper_snapshot_rsp); + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_get_cper_records(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_cper_record_req *req = + (struct ras_cmd_cper_record_req *)cmd->input_buff_raw; + struct ras_cmd_cper_record_rsp *rsp = + (struct ras_cmd_cper_record_rsp *)cmd->output_buff_raw; + struct ras_log_info *trace[MAX_RECORD_PER_BATCH] = {0}; + struct ras_log_batch_overview overview; + uint32_t offset = 0, real_data_len = 0; + uint64_t batch_id; + uint8_t *buffer; + int ret = 0, i, count; + + if (cmd->input_size != sizeof(struct ras_cmd_cper_record_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + if (!req->buf_size || !req->buf_ptr || !req->cper_num) + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + + buffer = kzalloc(req->buf_size, GFP_KERNEL); + if (!buffer) + return RAS_CMD__ERROR_GENERIC; + + ras_log_ring_get_batch_overview(ras_core, &overview); + for (i = 0; i < req->cper_num; i++) { + batch_id = req->cper_start_id + i; + if (batch_id >= overview.last_batch_id) + break; + + count = ras_log_ring_get_batch_records(ras_core, batch_id, trace, + ARRAY_SIZE(trace)); + if (count > 0) { + ret = ras_cper_generate_cper(ras_core, trace, count, + &buffer[offset], req->buf_size - offset, &real_data_len); + if (ret) + break; + + offset += real_data_len; + } + } + + if ((ret && (ret != -ENOMEM)) || + copy_to_user(u64_to_user_ptr(req->buf_ptr), buffer, offset)) { + kfree(buffer); + return RAS_CMD__ERROR_GENERIC; + } + + rsp->real_data_size = offset; + rsp->real_cper_num = i; + rsp->remain_num = (ret == -ENOMEM) ? (req->cper_num - i) : 0; + rsp->version = 0; + + cmd->output_size = sizeof(struct ras_cmd_cper_record_rsp); + + kfree(buffer); + + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_get_batch_trace_snapshot(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_batch_trace_snapshot_rsp *rsp = + (struct ras_cmd_batch_trace_snapshot_rsp *)cmd->output_buff_raw; + struct ras_log_batch_overview overview; + + + if (cmd->input_size != sizeof(struct ras_cmd_batch_trace_snapshot_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + ras_log_ring_get_batch_overview(ras_core, &overview); + + rsp->total_batch_num = overview.logged_batch_count; + rsp->start_batch_id = overview.first_batch_id; + rsp->latest_batch_id = overview.last_batch_id; + rsp->version = 0; + + cmd->output_size = sizeof(struct ras_cmd_batch_trace_snapshot_rsp); + return RAS_CMD__SUCCESS; +} + +static int ras_cmd_get_batch_trace_records(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_batch_trace_record_req *input_data = + (struct ras_cmd_batch_trace_record_req *)cmd->input_buff_raw; + struct ras_cmd_batch_trace_record_rsp *output_data = + (struct ras_cmd_batch_trace_record_rsp *)cmd->output_buff_raw; + struct ras_log_batch_overview overview; + struct ras_log_info *trace_arry[MAX_RECORD_PER_BATCH] = {0}; + struct ras_log_info *record; + int i, j, count = 0, offset = 0; + uint64_t id; + bool completed = false; + + if (cmd->input_size != sizeof(struct ras_cmd_batch_trace_record_req)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + if ((!input_data->batch_num) || (input_data->batch_num > RAS_CMD_MAX_BATCH_NUM)) + return RAS_CMD__ERROR_INVALID_INPUT_DATA; + + ras_log_ring_get_batch_overview(ras_core, &overview); + if ((input_data->start_batch_id < overview.first_batch_id) || + (input_data->start_batch_id >= overview.last_batch_id)) + return RAS_CMD__ERROR_INVALID_INPUT_SIZE; + + for (i = 0; i < input_data->batch_num; i++) { + id = input_data->start_batch_id + i; + if (id >= overview.last_batch_id) { + completed = true; + break; + } + + count = ras_log_ring_get_batch_records(ras_core, + id, trace_arry, ARRAY_SIZE(trace_arry)); + if (count > 0) { + if ((offset + count) > RAS_CMD_MAX_TRACE_NUM) + break; + for (j = 0; j < count; j++) { + record = &output_data->records[offset + j]; + record->seqno = trace_arry[j]->seqno; + record->timestamp = trace_arry[j]->timestamp; + record->event = trace_arry[j]->event; + memcpy(&record->aca_reg, + &trace_arry[j]->aca_reg, sizeof(trace_arry[j]->aca_reg)); + } + } else { + count = 0; + } + + output_data->batchs[i].batch_id = id; + output_data->batchs[i].offset = offset; + output_data->batchs[i].trace_num = count; + offset += count; + } + + output_data->start_batch_id = input_data->start_batch_id; + output_data->real_batch_num = i; + output_data->remain_num = completed ? 0 : (input_data->batch_num - i); + output_data->version = 0; + + cmd->output_size = sizeof(struct ras_cmd_batch_trace_record_rsp); + + return RAS_CMD__SUCCESS; +} + +static enum ras_ta_block __get_ras_ta_block(enum ras_block_id block) +{ + switch (block) { + case RAS_BLOCK_ID__UMC: + return RAS_TA_BLOCK__UMC; + case RAS_BLOCK_ID__SDMA: + return RAS_TA_BLOCK__SDMA; + case RAS_BLOCK_ID__GFX: + return RAS_TA_BLOCK__GFX; + case RAS_BLOCK_ID__MMHUB: + return RAS_TA_BLOCK__MMHUB; + case RAS_BLOCK_ID__ATHUB: + return RAS_TA_BLOCK__ATHUB; + case RAS_BLOCK_ID__PCIE_BIF: + return RAS_TA_BLOCK__PCIE_BIF; + case RAS_BLOCK_ID__HDP: + return RAS_TA_BLOCK__HDP; + case RAS_BLOCK_ID__XGMI_WAFL: + return RAS_TA_BLOCK__XGMI_WAFL; + case RAS_BLOCK_ID__DF: + return RAS_TA_BLOCK__DF; + case RAS_BLOCK_ID__SMN: + return RAS_TA_BLOCK__SMN; + case RAS_BLOCK_ID__SEM: + return RAS_TA_BLOCK__SEM; + case RAS_BLOCK_ID__MP0: + return RAS_TA_BLOCK__MP0; + case RAS_BLOCK_ID__MP1: + return RAS_TA_BLOCK__MP1; + case RAS_BLOCK_ID__FUSE: + return RAS_TA_BLOCK__FUSE; + case RAS_BLOCK_ID__MCA: + return RAS_TA_BLOCK__MCA; + case RAS_BLOCK_ID__VCN: + return RAS_TA_BLOCK__VCN; + case RAS_BLOCK_ID__JPEG: + return RAS_TA_BLOCK__JPEG; + default: + return RAS_TA_BLOCK__UMC; + } +} + +static enum ras_ta_error_type __get_ras_ta_err_type(enum ras_ecc_err_type error) +{ + switch (error) { + case RAS_ECC_ERR__NONE: + return RAS_TA_ERROR__NONE; + case RAS_ECC_ERR__PARITY: + return RAS_TA_ERROR__PARITY; + case RAS_ECC_ERR__SINGLE_CORRECTABLE: + return RAS_TA_ERROR__SINGLE_CORRECTABLE; + case RAS_ECC_ERR__MULTI_UNCORRECTABLE: + return RAS_TA_ERROR__MULTI_UNCORRECTABLE; + case RAS_ECC_ERR__POISON: + return RAS_TA_ERROR__POISON; + default: + return RAS_TA_ERROR__NONE; + } +} + +static int ras_cmd_inject_error(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_inject_error_req *req = + (struct ras_cmd_inject_error_req *)cmd->input_buff_raw; + struct ras_cmd_inject_error_rsp *output_data = + (struct ras_cmd_inject_error_rsp *)cmd->output_buff_raw; + int ret = 0; + struct ras_ta_trigger_error_input block_info = { + .block_id = __get_ras_ta_block(req->block_id), + .sub_block_index = req->subblock_id, + .inject_error_type = __get_ras_ta_err_type(req->error_type), + .address = req->address, + .value = req->method, + }; + + ret = ras_psp_trigger_error(ras_core, &block_info, req->instance_mask); + if (!ret) { + output_data->version = 0; + output_data->address = block_info.address; + cmd->output_size = sizeof(struct ras_cmd_inject_error_rsp); + } else { + RAS_DEV_ERR(ras_core->dev, "ras inject block %u failed %d\n", req->block_id, ret); + ret = RAS_CMD__ERROR_ACCESS_DENIED; + } + + return ret; +} + +static struct ras_cmd_func_map ras_cmd_maps[] = { + {RAS_CMD__INJECT_ERROR, ras_cmd_inject_error}, + {RAS_CMD__GET_BLOCK_ECC_STATUS, ras_get_block_ecc_info}, + {RAS_CMD__GET_BAD_PAGES, ras_cmd_get_bad_pages}, + {RAS_CMD__CLEAR_BAD_PAGE_INFO, ras_cmd_clear_bad_page_info}, + {RAS_CMD__RESET_ALL_ERROR_COUNTS, ras_cmd_reset_all_error_counts}, + {RAS_CMD__GET_CPER_SNAPSHOT, ras_cmd_get_cper_snapshot}, + {RAS_CMD__GET_CPER_RECORD, ras_cmd_get_cper_records}, + {RAS_CMD__GET_BATCH_TRACE_SNAPSHOT, ras_cmd_get_batch_trace_snapshot}, + {RAS_CMD__GET_BATCH_TRACE_RECORD, ras_cmd_get_batch_trace_records}, +}; + +int rascore_handle_cmd(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data) +{ + struct ras_cmd_func_map *ras_cmd = NULL; + int i; + + for (i = 0; i < ARRAY_SIZE(ras_cmd_maps); i++) { + if (cmd->cmd_id == ras_cmd_maps[i].cmd_id) { + ras_cmd = &ras_cmd_maps[i]; + break; + } + } + + if (!ras_cmd) + return RAS_CMD__ERROR_UKNOWN_CMD; + + return ras_cmd->func(ras_core, cmd, data); +} + +int ras_cmd_init(struct ras_core_context *ras_core) +{ + return ras_cmd_add_device(ras_core); +} + +int ras_cmd_fini(struct ras_core_context *ras_core) +{ + ras_cmd_remove_device(ras_core); + return 0; +} + +int ras_cmd_query_interface_info(struct ras_core_context *ras_core, + struct ras_query_interface_info_rsp *rsp) +{ + rsp->ras_cmd_major_ver = RAS_CMD_MAJOR_VERSION; + rsp->ras_cmd_minor_ver = RAS_CMD_MINOR_VERSION; + + return 0; +} + +int ras_cmd_translate_soc_pa_to_bank(struct ras_core_context *ras_core, + uint64_t soc_pa, struct ras_fb_bank_addr *bank_addr) +{ + struct umc_bank_addr umc_bank = {0}; + int ret; + + ret = ras_umc_translate_soc_pa_and_bank(ras_core, &soc_pa, &umc_bank, false); + if (ret) + return RAS_CMD__ERROR_GENERIC; + + bank_addr->stack_id = umc_bank.stack_id; + bank_addr->bank_group = umc_bank.bank_group; + bank_addr->bank = umc_bank.bank; + bank_addr->row = umc_bank.row; + bank_addr->column = umc_bank.column; + bank_addr->channel = umc_bank.channel; + bank_addr->subchannel = umc_bank.subchannel; + + return 0; +} + +int ras_cmd_translate_bank_to_soc_pa(struct ras_core_context *ras_core, + struct ras_fb_bank_addr bank_addr, uint64_t *soc_pa) +{ + struct umc_bank_addr umc_bank = {0}; + + umc_bank.stack_id = bank_addr.stack_id; + umc_bank.bank_group = bank_addr.bank_group; + umc_bank.bank = bank_addr.bank; + umc_bank.row = bank_addr.row; + umc_bank.column = bank_addr.column; + umc_bank.channel = bank_addr.channel; + umc_bank.subchannel = bank_addr.subchannel; + + return ras_umc_translate_soc_pa_and_bank(ras_core, soc_pa, &umc_bank, true); +} + +uint64_t ras_cmd_get_dev_handle(struct ras_core_context *ras_core) +{ + return ras_core->ras_cmd.dev_handle; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_cmd.h b/drivers/gpu/drm/amd/ras/rascore/ras_cmd.h new file mode 100644 index 000000000000..48a0715eb821 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_cmd.h @@ -0,0 +1,426 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_CMD_H__ +#define __RAS_CMD_H__ +#include "ras.h" +#include "ras_eeprom.h" +#include "ras_log_ring.h" +#include "ras_cper.h" + +#define RAS_CMD_DEV_HANDLE_MAGIC 0xFEEDAD00UL + +#define RAS_CMD_MAX_IN_SIZE 256 +#define RAS_CMD_MAX_GPU_NUM 32 +#define RAS_CMD_MAX_BAD_PAGES_PER_GROUP 32 + +/* position of instance value in sub_block_index of + * ta_ras_trigger_error_input, the sub block uses lower 12 bits + */ +#define RAS_TA_INST_MASK 0xfffff000 +#define RAS_TA_INST_SHIFT 0xc + +enum ras_cmd_interface_type { + RAS_CMD_INTERFACE_TYPE_NONE, + RAS_CMD_INTERFACE_TYPE_AMDGPU, + RAS_CMD_INTERFACE_TYPE_VF, + RAS_CMD_INTERFACE_TYPE_PF, +}; + +enum ras_cmd_id_range { + RAS_CMD_ID_COMMON_START = 0, + RAS_CMD_ID_COMMON_END = 0x10000, + RAS_CMD_ID_AMDGPU_START = RAS_CMD_ID_COMMON_END, + RAS_CMD_ID_AMDGPU_END = 0x20000, + RAS_CMD_ID_MXGPU_START = RAS_CMD_ID_AMDGPU_END, + RAS_CMD_ID_MXGPU_END = 0x30000, + RAS_CMD_ID_MXGPU_VF_START = RAS_CMD_ID_MXGPU_END, + RAS_CMD_ID_MXGPU_VF_END = 0x40000, +}; + +enum ras_cmd_id { + RAS_CMD__BEGIN = RAS_CMD_ID_COMMON_START, + RAS_CMD__QUERY_INTERFACE_INFO, + RAS_CMD__GET_DEVICES_INFO, + RAS_CMD__GET_BLOCK_ECC_STATUS, + RAS_CMD__INJECT_ERROR, + RAS_CMD__GET_BAD_PAGES, + RAS_CMD__CLEAR_BAD_PAGE_INFO, + RAS_CMD__RESET_ALL_ERROR_COUNTS, + RAS_CMD__GET_SAFE_FB_ADDRESS_RANGES, + RAS_CMD__TRANSLATE_FB_ADDRESS, + RAS_CMD__GET_LINK_TOPOLOGY, + RAS_CMD__GET_CPER_SNAPSHOT, + RAS_CMD__GET_CPER_RECORD, + RAS_CMD__GET_BATCH_TRACE_SNAPSHOT, + RAS_CMD__GET_BATCH_TRACE_RECORD, + RAS_CMD__SUPPORTED_MAX = RAS_CMD_ID_COMMON_END, +}; + +enum ras_cmd_response { + RAS_CMD__SUCCESS = 0, + RAS_CMD__SUCCESS_EXEED_BUFFER, + RAS_CMD__ERROR_UKNOWN_CMD, + RAS_CMD__ERROR_INVALID_CMD, + RAS_CMD__ERROR_VERSION, + RAS_CMD__ERROR_INVALID_INPUT_SIZE, + RAS_CMD__ERROR_INVALID_INPUT_DATA, + RAS_CMD__ERROR_DRV_INIT_FAIL, + RAS_CMD__ERROR_ACCESS_DENIED, + RAS_CMD__ERROR_GENERIC, + RAS_CMD__ERROR_TIMEOUT, +}; + +enum ras_error_type { + RAS_TYPE_ERROR__NONE = 0, + RAS_TYPE_ERROR__PARITY = 1, + RAS_TYPE_ERROR__SINGLE_CORRECTABLE = 2, + RAS_TYPE_ERROR__MULTI_UNCORRECTABLE = 4, + RAS_TYPE_ERROR__POISON = 8, +}; + +struct ras_core_context; +struct ras_cmd_ctx; + +struct ras_cmd_mgr { + struct list_head head; + struct ras_core_context *ras_core; + uint64_t dev_handle; +}; + +struct ras_cmd_func_map { + uint32_t cmd_id; + int (*func)(struct ras_core_context *ras_core, + struct ras_cmd_ctx *cmd, void *data); +}; + +struct ras_device_bdf { + union { + struct { + uint32_t function : 3; + uint32_t device : 5; + uint32_t bus : 8; + uint32_t domain : 16; + }; + uint32_t u32_all; + }; +}; + +struct ras_cmd_param { + uint32_t idx_vf; + void *data; +}; + +#pragma pack(push, 8) +struct ras_cmd_ctx { + uint32_t magic; + union { + struct { + uint16_t ras_cmd_minor_ver : 10; + uint16_t ras_cmd_major_ver : 6; + }; + uint16_t ras_cmd_ver; + }; + union { + struct { + uint16_t plat_major_ver : 10; + uint16_t plat_minor_ver : 6; + }; + uint16_t plat_ver; + }; + uint32_t cmd_id; + uint32_t cmd_res; + uint32_t input_size; + uint32_t output_size; + uint32_t output_buf_size; + uint32_t reserved[5]; + uint8_t input_buff_raw[RAS_CMD_MAX_IN_SIZE]; + uint8_t output_buff_raw[]; +}; + +struct ras_cmd_dev_handle { + uint64_t dev_handle; +}; + +struct ras_cmd_block_ecc_info_req { + struct ras_cmd_dev_handle dev; + uint32_t block_id; + uint32_t subblock_id; + uint32_t reserved[4]; +}; + +struct ras_cmd_block_ecc_info_rsp { + uint32_t version; + uint32_t ce_count; + uint32_t ue_count; + uint32_t de_count; + uint32_t reserved[6]; +}; + +struct ras_cmd_inject_error_req { + struct ras_cmd_dev_handle dev; + uint32_t block_id; + uint32_t subblock_id; + uint64_t address; + uint32_t error_type; + uint32_t instance_mask; + union { + struct { + /* vf index */ + uint64_t vf_idx : 6; + /* method of error injection. i.e persistent, coherent etc */ + uint64_t method : 10; + uint64_t rsv : 48; + }; + uint64_t value; + }; + uint32_t reserved[8]; +}; + +struct ras_cmd_inject_error_rsp { + uint32_t version; + uint32_t reserved[5]; + uint64_t address; +}; + +struct ras_cmd_dev_info { + uint64_t dev_handle; + uint32_t location_id; + uint32_t ecc_enabled; + uint32_t ecc_supported; + uint32_t vf_num; + uint32_t asic_type; + uint32_t oam_id; + uint32_t reserved[8]; +}; + +struct ras_cmd_devices_info_rsp { + uint32_t version; + uint32_t dev_num; + uint32_t reserved[6]; + struct ras_cmd_dev_info devs[RAS_CMD_MAX_GPU_NUM]; +}; + +struct ras_cmd_bad_page_record { + union { + uint64_t address; + uint64_t offset; + }; + uint64_t retired_page; + uint64_t ts; + + uint32_t err_type; + + union { + unsigned char bank; + unsigned char cu; + }; + + unsigned char mem_channel; + unsigned char mcumc_id; + + unsigned char valid; + unsigned char reserved[8]; +}; + +struct ras_cmd_bad_pages_info_req { + struct ras_cmd_dev_handle device; + uint32_t group_index; + uint32_t reserved[5]; +}; + +struct ras_cmd_bad_pages_info_rsp { + uint32_t version; + uint32_t group_index; + uint32_t bp_in_group; + uint32_t bp_total_cnt; + uint32_t reserved[4]; + struct ras_cmd_bad_page_record records[RAS_CMD_MAX_BAD_PAGES_PER_GROUP]; +}; + +struct ras_query_interface_info_req { + uint32_t reserved[8]; +}; + +struct ras_query_interface_info_rsp { + uint32_t version; + uint32_t ras_cmd_major_ver; + uint32_t ras_cmd_minor_ver; + uint32_t plat_major_ver; + uint32_t plat_minor_ver; + uint8_t interface_type; + uint8_t rsv[3]; + uint32_t reserved[8]; +}; + +#define RAS_MAX_NUM_SAFE_RANGES 64 +struct ras_cmd_ras_safe_fb_address_ranges_rsp { + uint32_t version; + uint32_t num_ranges; + uint32_t reserved[4]; + struct { + uint64_t start; + uint64_t size; + uint32_t idx; + uint32_t reserved[3]; + } range[RAS_MAX_NUM_SAFE_RANGES]; +}; + +enum ras_fb_addr_type { + RAS_FB_ADDR_SOC_PHY, /* SPA */ + RAS_FB_ADDR_BANK, + RAS_FB_ADDR_VF_PHY, /* GPA */ + RAS_FB_ADDR_UNKNOWN +}; + +struct ras_fb_bank_addr { + uint32_t stack_id; /* SID */ + uint32_t bank_group; + uint32_t bank; + uint32_t row; + uint32_t column; + uint32_t channel; + uint32_t subchannel; /* Also called Pseudochannel (PC) */ + uint32_t reserved[3]; +}; + +struct ras_fb_vf_phy_addr { + uint32_t vf_idx; + uint32_t reserved; + uint64_t addr; +}; + +union ras_translate_fb_address { + struct ras_fb_bank_addr bank_addr; + uint64_t soc_phy_addr; + struct ras_fb_vf_phy_addr vf_phy_addr; +}; + +struct ras_cmd_translate_fb_address_req { + struct ras_cmd_dev_handle dev; + enum ras_fb_addr_type src_addr_type; + enum ras_fb_addr_type dest_addr_type; + union ras_translate_fb_address trans_addr; +}; + +struct ras_cmd_translate_fb_address_rsp { + uint32_t version; + uint32_t reserved[5]; + union ras_translate_fb_address trans_addr; +}; + +struct ras_dev_link_topology_req { + struct ras_cmd_dev_handle src; + struct ras_cmd_dev_handle dst; +}; + +struct ras_dev_link_topology_rsp { + uint32_t version; + uint32_t link_status; /* HW status of the link */ + uint32_t link_type; /* type of the link */ + uint32_t num_hops; /* number of hops */ + uint32_t reserved[8]; +}; + +struct ras_cmd_cper_snapshot_req { + struct ras_cmd_dev_handle dev; +}; + +struct ras_cmd_cper_snapshot_rsp { + uint32_t version; + uint32_t reserved[4]; + uint32_t total_cper_num; + uint64_t start_cper_id; + uint64_t latest_cper_id; +}; + +struct ras_cmd_cper_record_req { + struct ras_cmd_dev_handle dev; + uint64_t cper_start_id; + uint32_t cper_num; + uint32_t buf_size; + uint64_t buf_ptr; + uint32_t reserved[4]; +}; + +struct ras_cmd_cper_record_rsp { + uint32_t version; + uint32_t real_data_size; + uint32_t real_cper_num; + uint32_t remain_num; + uint32_t reserved[4]; +}; + +struct ras_cmd_batch_trace_snapshot_req { + struct ras_cmd_dev_handle dev; +}; + +struct ras_cmd_batch_trace_snapshot_rsp { + uint32_t version; + uint32_t reserved[4]; + uint32_t total_batch_num; + uint64_t start_batch_id; + uint64_t latest_batch_id; +}; + +struct ras_cmd_batch_trace_record_req { + struct ras_cmd_dev_handle dev; + uint64_t start_batch_id; + uint32_t batch_num; + uint32_t reserved[5]; +}; + +struct batch_ras_trace_info { + uint64_t batch_id; + uint16_t offset; + uint8_t trace_num; + uint8_t rsv; + uint32_t reserved; +}; + +#define RAS_CMD_MAX_BATCH_NUM 300 +#define RAS_CMD_MAX_TRACE_NUM 300 +struct ras_cmd_batch_trace_record_rsp { + uint32_t version; + uint16_t real_batch_num; + uint16_t remain_num; + uint64_t start_batch_id; + uint32_t reserved[2]; + struct batch_ras_trace_info batchs[RAS_CMD_MAX_BATCH_NUM]; + struct ras_log_info records[RAS_CMD_MAX_TRACE_NUM]; +}; + +#pragma pack(pop) + +int ras_cmd_init(struct ras_core_context *ras_core); +int ras_cmd_fini(struct ras_core_context *ras_core); +int rascore_handle_cmd(struct ras_core_context *ras_core, struct ras_cmd_ctx *cmd, void *data); +uint64_t ras_cmd_get_dev_handle(struct ras_core_context *ras_core); +int ras_cmd_query_interface_info(struct ras_core_context *ras_core, + struct ras_query_interface_info_rsp *rsp); +int ras_cmd_translate_soc_pa_to_bank(struct ras_core_context *ras_core, + uint64_t soc_pa, struct ras_fb_bank_addr *bank_addr); +int ras_cmd_translate_bank_to_soc_pa(struct ras_core_context *ras_core, + struct ras_fb_bank_addr bank_addr, uint64_t *soc_pa); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_core.c b/drivers/gpu/drm/amd/ras/rascore/ras_core.c new file mode 100644 index 000000000000..01122b55c98a --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_core.c @@ -0,0 +1,603 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_core_status.h" + +#define RAS_SEQNO_FIFO_SIZE (128 * sizeof(uint64_t)) + +#define IS_LEAP_YEAR(x) ((x % 4 == 0 && x % 100 != 0) || x % 400 == 0) + +static const char * const ras_block_name[] = { + "umc", + "sdma", + "gfx", + "mmhub", + "athub", + "pcie_bif", + "hdp", + "xgmi_wafl", + "df", + "smn", + "sem", + "mp0", + "mp1", + "fuse", + "mca", + "vcn", + "jpeg", + "ih", + "mpio", +}; + +const char *ras_core_get_ras_block_name(enum ras_block_id block_id) +{ + if (block_id >= ARRAY_SIZE(ras_block_name)) + return ""; + + return ras_block_name[block_id]; +} + +int ras_core_convert_timestamp_to_time(struct ras_core_context *ras_core, + uint64_t timestamp, struct ras_time *tm) +{ + int days_in_month[] = {31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31}; + uint64_t month = 0, day = 0, hour = 0, minute = 0, second = 0; + uint32_t year = 0; + int seconds_per_day = 24 * 60 * 60; + int seconds_per_hour = 60 * 60; + int seconds_per_minute = 60; + int days, remaining_seconds; + + days = div64_u64_rem(timestamp, seconds_per_day, (uint64_t *)&remaining_seconds); + + /* utc_timestamp follows the Unix epoch */ + year = 1970; + while (days >= 365) { + if (IS_LEAP_YEAR(year)) { + if (days < 366) + break; + days -= 366; + } else { + days -= 365; + } + year++; + } + + days_in_month[1] += IS_LEAP_YEAR(year); + + month = 0; + while (days >= days_in_month[month]) { + days -= days_in_month[month]; + month++; + } + month++; + day = days + 1; + + if (remaining_seconds) { + hour = remaining_seconds / seconds_per_hour; + minute = (remaining_seconds % seconds_per_hour) / seconds_per_minute; + second = remaining_seconds % seconds_per_minute; + } + + tm->tm_year = year; + tm->tm_mon = month; + tm->tm_mday = day; + tm->tm_hour = hour; + tm->tm_min = minute; + tm->tm_sec = second; + + return 0; +} + +bool ras_core_gpu_in_reset(struct ras_core_context *ras_core) +{ + uint32_t status = 0; + + if (ras_core->sys_fn && + ras_core->sys_fn->check_gpu_status) + ras_core->sys_fn->check_gpu_status(ras_core, &status); + + return (status & RAS_GPU_STATUS__IN_RESET) ? true : false; +} + +bool ras_core_gpu_is_vf(struct ras_core_context *ras_core) +{ + uint32_t status = 0; + + if (ras_core->sys_fn && + ras_core->sys_fn->check_gpu_status) + ras_core->sys_fn->check_gpu_status(ras_core, &status); + + return (status & RAS_GPU_STATUS__IS_VF) ? true : false; +} + +bool ras_core_gpu_is_rma(struct ras_core_context *ras_core) +{ + if (!ras_core) + return false; + + return ras_core->is_rma; +} + +static int ras_core_seqno_fifo_write(struct ras_core_context *ras_core, + enum ras_seqno_fifo fifo_type, uint64_t seqno) +{ + int ret = 0; + struct kfifo *seqno_fifo = NULL; + + if (fifo_type == SEQNO_FIFO_POISON_CREATION) + seqno_fifo = &ras_core->de_seqno_fifo; + else if (fifo_type == SEQNO_FIFO_POISON_CONSUMPTION) + seqno_fifo = &ras_core->consumption_seqno_fifo; + + if (seqno_fifo) + ret = kfifo_in_spinlocked(seqno_fifo, + &seqno, sizeof(seqno), &ras_core->seqno_lock); + + return ret ? 0 : -EINVAL; +} + +static int ras_core_seqno_fifo_read(struct ras_core_context *ras_core, + enum ras_seqno_fifo fifo_type, uint64_t *seqno, bool pop) +{ + int ret = 0; + struct kfifo *seqno_fifo = NULL; + + if (fifo_type == SEQNO_FIFO_POISON_CREATION) + seqno_fifo = &ras_core->de_seqno_fifo; + else if (fifo_type == SEQNO_FIFO_POISON_CONSUMPTION) + seqno_fifo = &ras_core->consumption_seqno_fifo; + + if (seqno_fifo) { + if (pop) + ret = kfifo_out_spinlocked(seqno_fifo, + seqno, sizeof(*seqno), &ras_core->seqno_lock); + else + ret = kfifo_out_peek(seqno_fifo, seqno, sizeof(*seqno)); + } + + return ret ? 0 : -EINVAL; +} + +uint64_t ras_core_gen_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type type) +{ + uint64_t seqno = 0; + + if (ras_core->sys_fn && + ras_core->sys_fn->gen_seqno) + ras_core->sys_fn->gen_seqno(ras_core, type, &seqno); + + return seqno; +} + +int ras_core_put_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, uint64_t seqno) +{ + int ret = 0; + + if (seqno_type >= RAS_SEQNO_TYPE_COUNT_MAX) + return -EINVAL; + + if (seqno_type == RAS_SEQNO_TYPE_DE) + ret = ras_core_seqno_fifo_write(ras_core, + SEQNO_FIFO_POISON_CREATION, seqno); + else if (seqno_type == RAS_SEQNO_TYPE_POISON_CONSUMPTION) + ret = ras_core_seqno_fifo_write(ras_core, + SEQNO_FIFO_POISON_CONSUMPTION, seqno); + else + ret = -EINVAL; + + return ret; +} + +uint64_t ras_core_get_seqno(struct ras_core_context *ras_core, + enum ras_seqno_type seqno_type, bool pop) +{ + uint64_t seq_no; + int ret = -ENODATA; + + if (seqno_type >= RAS_SEQNO_TYPE_COUNT_MAX) + return 0; + + if (seqno_type == RAS_SEQNO_TYPE_DE) + ret = ras_core_seqno_fifo_read(ras_core, + SEQNO_FIFO_POISON_CREATION, &seq_no, pop); + else if (seqno_type == RAS_SEQNO_TYPE_POISON_CONSUMPTION) + ret = ras_core_seqno_fifo_read(ras_core, + SEQNO_FIFO_POISON_CONSUMPTION, &seq_no, pop); + + if (ret) + seq_no = ras_core_gen_seqno(ras_core, seqno_type); + + return seq_no; +} + +static int ras_core_eeprom_recovery(struct ras_core_context *ras_core) +{ + int count; + int ret; + + count = ras_eeprom_get_record_count(ras_core); + if (!count) + return 0; + + /* Avoid bad page to be loaded again after gpu reset */ + if (ras_umc_get_saved_eeprom_count(ras_core) >= count) + return 0; + + ret = ras_umc_load_bad_pages(ras_core); + if (ret) { + RAS_DEV_ERR(ras_core->dev, "ras_umc_load_bad_pages failed: %d\n", ret); + return ret; + } + + ras_eeprom_sync_info(ras_core); + + return ret; +} + +struct ras_core_context *ras_core_create(struct ras_core_config *init_config) +{ + struct ras_core_context *ras_core; + struct ras_core_config *config; + + ras_core = kzalloc(sizeof(*ras_core), GFP_KERNEL); + if (!ras_core) + return NULL; + + config = kzalloc(sizeof(*config), GFP_KERNEL); + if (!config) { + kfree(ras_core); + return NULL; + } + + memcpy(config, init_config, sizeof(*config)); + ras_core->config = config; + + return ras_core; +} + +void ras_core_destroy(struct ras_core_context *ras_core) +{ + if (ras_core) + kfree(ras_core->config); + + kfree(ras_core); +} + +int ras_core_sw_init(struct ras_core_context *ras_core) +{ + int ret; + + if (!ras_core->config) { + RAS_DEV_ERR(ras_core->dev, "No ras core config!\n"); + return -EINVAL; + } + + ras_core->sys_fn = ras_core->config->sys_fn; + if (!ras_core->sys_fn) + return -EINVAL; + + ret = kfifo_alloc(&ras_core->de_seqno_fifo, + RAS_SEQNO_FIFO_SIZE, GFP_KERNEL); + if (ret) + return ret; + + ret = kfifo_alloc(&ras_core->consumption_seqno_fifo, + RAS_SEQNO_FIFO_SIZE, GFP_KERNEL); + if (ret) + return ret; + + spin_lock_init(&ras_core->seqno_lock); + + ret = ras_aca_sw_init(ras_core); + if (ret) + return ret; + + ret = ras_umc_sw_init(ras_core); + if (ret) + return ret; + + ret = ras_cmd_init(ras_core); + if (ret) + return ret; + + ret = ras_log_ring_sw_init(ras_core); + if (ret) + return ret; + + ret = ras_psp_sw_init(ras_core); + if (ret) + return ret; + + return 0; +} + +int ras_core_sw_fini(struct ras_core_context *ras_core) +{ + kfifo_free(&ras_core->de_seqno_fifo); + kfifo_free(&ras_core->consumption_seqno_fifo); + + ras_psp_sw_fini(ras_core); + ras_log_ring_sw_fini(ras_core); + ras_cmd_fini(ras_core); + ras_umc_sw_fini(ras_core); + ras_aca_sw_fini(ras_core); + + return 0; +} + +int ras_core_hw_init(struct ras_core_context *ras_core) +{ + int ret; + + ras_core->ras_eeprom_supported = + ras_core->config->ras_eeprom_supported; + + ras_core->poison_supported = ras_core->config->poison_supported; + + ret = ras_psp_hw_init(ras_core); + if (ret) + return ret; + + ret = ras_aca_hw_init(ras_core); + if (ret) + goto init_err1; + + ret = ras_mp1_hw_init(ras_core); + if (ret) + goto init_err2; + + ret = ras_nbio_hw_init(ras_core); + if (ret) + goto init_err3; + + ret = ras_umc_hw_init(ras_core); + if (ret) + goto init_err4; + + ret = ras_gfx_hw_init(ras_core); + if (ret) + goto init_err5; + + ret = ras_eeprom_hw_init(ras_core); + if (ret) + goto init_err6; + + ret = ras_core_eeprom_recovery(ras_core); + if (ret) { + RAS_DEV_ERR(ras_core->dev, + "Failed to recovery ras core, ret:%d\n", ret); + goto init_err6; + } + + ret = ras_eeprom_check_storage_status(ras_core); + if (ret) + goto init_err6; + + ret = ras_process_init(ras_core); + if (ret) + goto init_err7; + + ras_core->is_initialized = true; + + return 0; + +init_err7: + ras_eeprom_hw_fini(ras_core); +init_err6: + ras_gfx_hw_fini(ras_core); +init_err5: + ras_umc_hw_fini(ras_core); +init_err4: + ras_nbio_hw_fini(ras_core); +init_err3: + ras_mp1_hw_fini(ras_core); +init_err2: + ras_aca_hw_fini(ras_core); +init_err1: + ras_psp_hw_fini(ras_core); + return ret; +} + +int ras_core_hw_fini(struct ras_core_context *ras_core) +{ + ras_core->is_initialized = false; + + ras_process_fini(ras_core); + ras_eeprom_hw_fini(ras_core); + ras_gfx_hw_fini(ras_core); + ras_nbio_hw_fini(ras_core); + ras_umc_hw_fini(ras_core); + ras_mp1_hw_fini(ras_core); + ras_aca_hw_fini(ras_core); + ras_psp_hw_fini(ras_core); + + return 0; +} + +bool ras_core_handle_nbio_irq(struct ras_core_context *ras_core, void *data) +{ + return ras_nbio_handle_irq_error(ras_core, data); +} + +int ras_core_handle_fatal_error(struct ras_core_context *ras_core) +{ + int ret = 0; + + ras_aca_mark_fatal_flag(ras_core); + + ret = ras_core_event_notify(ras_core, + RAS_EVENT_ID__FATAL_ERROR_DETECTED, NULL); + + return ret; +} + +uint32_t ras_core_get_curr_nps_mode(struct ras_core_context *ras_core) +{ + if (ras_core->ras_nbio.ip_func && + ras_core->ras_nbio.ip_func->get_memory_partition_mode) + return ras_core->ras_nbio.ip_func->get_memory_partition_mode(ras_core); + + RAS_DEV_ERR(ras_core->dev, "Failed to get gpu memory nps mode!\n"); + return 0; +} + +int ras_core_update_ecc_info(struct ras_core_context *ras_core) +{ + int ret; + + ret = ras_aca_update_ecc(ras_core, RAS_ERR_TYPE__CE, NULL); + if (!ret) + ret = ras_aca_update_ecc(ras_core, RAS_ERR_TYPE__UE, NULL); + + return ret; +} + +int ras_core_query_block_ecc_data(struct ras_core_context *ras_core, + enum ras_block_id block, struct ras_ecc_count *ecc_count) +{ + int ret; + + if (!ecc_count || (block >= RAS_BLOCK_ID__LAST) || !ras_core) + return -EINVAL; + + ret = ras_aca_get_block_ecc_count(ras_core, block, ecc_count); + if (!ret) + ras_aca_clear_block_new_ecc_count(ras_core, block); + + return ret; +} + +int ras_core_set_status(struct ras_core_context *ras_core, bool enable) +{ + ras_core->ras_core_enabled = enable; + + return 0; +} + +bool ras_core_is_enabled(struct ras_core_context *ras_core) +{ + return ras_core->ras_core_enabled; +} + +uint64_t ras_core_get_utc_second_timestamp(struct ras_core_context *ras_core) +{ + if (ras_core && ras_core->sys_fn && + ras_core->sys_fn->get_utc_second_timestamp) + return ras_core->sys_fn->get_utc_second_timestamp(ras_core); + + RAS_DEV_ERR(ras_core->dev, "Failed to get system timestamp!\n"); + return 0; +} + +int ras_core_translate_soc_pa_and_bank(struct ras_core_context *ras_core, + uint64_t *soc_pa, struct umc_bank_addr *bank_addr, bool bank_to_pa) +{ + if (!ras_core || !soc_pa || !bank_addr) + return -EINVAL; + + return ras_umc_translate_soc_pa_and_bank(ras_core, soc_pa, bank_addr, bank_to_pa); +} + +bool ras_core_ras_interrupt_detected(struct ras_core_context *ras_core) +{ + if (ras_core && ras_core->sys_fn && + ras_core->sys_fn->detect_ras_interrupt) + return ras_core->sys_fn->detect_ras_interrupt(ras_core); + + RAS_DEV_ERR(ras_core->dev, "Failed to detect ras interrupt!\n"); + return false; +} + +int ras_core_get_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem) +{ + if (ras_core->sys_fn && ras_core->sys_fn->get_gpu_mem) + return ras_core->sys_fn->get_gpu_mem(ras_core, mem_type, gpu_mem); + + RAS_DEV_ERR(ras_core->dev, "Not config get gpu memory API!\n"); + return -EACCES; +} + +int ras_core_put_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type, struct gpu_mem_block *gpu_mem) +{ + if (ras_core->sys_fn && ras_core->sys_fn->put_gpu_mem) + return ras_core->sys_fn->put_gpu_mem(ras_core, mem_type, gpu_mem); + + RAS_DEV_ERR(ras_core->dev, "Not config put gpu memory API!!\n"); + return -EACCES; +} + +bool ras_core_is_ready(struct ras_core_context *ras_core) +{ + return ras_core ? ras_core->is_initialized : false; +} + +bool ras_core_check_safety_watermark(struct ras_core_context *ras_core) +{ + return ras_eeprom_check_safety_watermark(ras_core); +} + +int ras_core_down_trylock_gpu_reset_lock(struct ras_core_context *ras_core) +{ + if (ras_core->sys_fn && ras_core->sys_fn->gpu_reset_lock) + return ras_core->sys_fn->gpu_reset_lock(ras_core, true, true); + + return 1; +} + +void ras_core_down_gpu_reset_lock(struct ras_core_context *ras_core) +{ + if (ras_core->sys_fn && ras_core->sys_fn->gpu_reset_lock) + ras_core->sys_fn->gpu_reset_lock(ras_core, true, false); +} + +void ras_core_up_gpu_reset_lock(struct ras_core_context *ras_core) +{ + if (ras_core->sys_fn && ras_core->sys_fn->gpu_reset_lock) + ras_core->sys_fn->gpu_reset_lock(ras_core, false, false); +} + +int ras_core_event_notify(struct ras_core_context *ras_core, + enum ras_notify_event event_id, void *data) +{ + if (ras_core && ras_core->sys_fn && + ras_core->sys_fn->ras_notifier) + return ras_core->sys_fn->ras_notifier(ras_core, event_id, data); + + return -RAS_CORE_NOT_SUPPORTED; +} + +int ras_core_get_device_system_info(struct ras_core_context *ras_core, + struct device_system_info *dev_info) +{ + if (ras_core && ras_core->sys_fn && + ras_core->sys_fn->get_device_system_info) + return ras_core->sys_fn->get_device_system_info(ras_core, dev_info); + + return -RAS_CORE_NOT_SUPPORTED; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_cper.c b/drivers/gpu/drm/amd/ras/rascore/ras_cper.c new file mode 100644 index 000000000000..0fc7522b7ab6 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_cper.c @@ -0,0 +1,315 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_core_status.h" +#include "ras_log_ring.h" +#include "ras_cper.h" + +static const struct ras_cper_guid MCE = CPER_NOTIFY__MCE; +static const struct ras_cper_guid CMC = CPER_NOTIFY__CMC; +static const struct ras_cper_guid BOOT = BOOT__TYPE; + +static const struct ras_cper_guid CRASHDUMP = GPU__CRASHDUMP; +static const struct ras_cper_guid RUNTIME = GPU__NONSTANDARD_ERROR; + +static void cper_get_timestamp(struct ras_core_context *ras_core, + struct ras_cper_timestamp *timestamp, uint64_t utc_second_timestamp) +{ + struct ras_time tm = {0}; + + ras_core_convert_timestamp_to_time(ras_core, utc_second_timestamp, &tm); + timestamp->seconds = tm.tm_sec; + timestamp->minutes = tm.tm_min; + timestamp->hours = tm.tm_hour; + timestamp->flag = 0; + timestamp->day = tm.tm_mday; + timestamp->month = tm.tm_mon; + timestamp->year = tm.tm_year % 100; + timestamp->century = tm.tm_year / 100; +} + +static void fill_section_hdr(struct ras_core_context *ras_core, + struct cper_section_hdr *hdr, enum ras_cper_type type, + enum ras_cper_severity sev, struct ras_log_info *trace) +{ + struct device_system_info dev_info = {0}; + char record_id[32]; + + hdr->signature[0] = 'C'; + hdr->signature[1] = 'P'; + hdr->signature[2] = 'E'; + hdr->signature[3] = 'R'; + hdr->revision = CPER_HDR__REV_1; + hdr->signature_end = 0xFFFFFFFF; + hdr->error_severity = (sev == RAS_CPER_SEV_RMA ? RAS_CPER_SEV_FATAL_UE : sev); + + hdr->valid_bits.platform_id = 1; + hdr->valid_bits.timestamp = 1; + + ras_core_get_device_system_info(ras_core, &dev_info); + + cper_get_timestamp(ras_core, &hdr->timestamp, trace->timestamp); + + snprintf(record_id, sizeof(record_id), "%d:%llX", dev_info.socket_id, + RAS_LOG_SEQNO_TO_BATCH_IDX(trace->seqno)); + memcpy(hdr->record_id, record_id, 8); + + snprintf(hdr->platform_id, 16, "0x%04X:0x%04X", + dev_info.vendor_id, dev_info.device_id); + /* pmfw version should be part of creator_id according to CPER spec */ + snprintf(hdr->creator_id, 16, "%s", CPER_CREATOR_ID__AMDGPU); + + switch (type) { + case RAS_CPER_TYPE_BOOT: + hdr->notify_type = BOOT; + break; + case RAS_CPER_TYPE_FATAL: + case RAS_CPER_TYPE_RMA: + hdr->notify_type = MCE; + break; + case RAS_CPER_TYPE_RUNTIME: + if (sev == RAS_CPER_SEV_NON_FATAL_CE) + hdr->notify_type = CMC; + else + hdr->notify_type = MCE; + break; + default: + RAS_DEV_ERR(ras_core->dev, "Unknown CPER Type\n"); + break; + } +} + +static int fill_section_descriptor(struct ras_core_context *ras_core, + struct cper_section_descriptor *descriptor, + enum ras_cper_severity sev, + struct ras_cper_guid sec_type, + uint32_t section_offset, + uint32_t section_length) +{ + struct device_system_info dev_info = {0}; + + descriptor->revision_minor = CPER_SEC__MINOR_REV_1; + descriptor->revision_major = CPER_SEC__MAJOR_REV_22; + descriptor->sec_offset = section_offset; + descriptor->sec_length = section_length; + descriptor->valid_bits.fru_text = 1; + descriptor->flag_bits.primary = 1; + descriptor->severity = (sev == RAS_CPER_SEV_RMA ? RAS_CPER_SEV_FATAL_UE : sev); + descriptor->sec_type = sec_type; + + ras_core_get_device_system_info(ras_core, &dev_info); + + snprintf(descriptor->fru_text, 20, "OAM%d", dev_info.socket_id); + + if (sev == RAS_CPER_SEV_RMA) + descriptor->flag_bits.exceed_err_threshold = 1; + + if (sev == RAS_CPER_SEV_NON_FATAL_UE) + descriptor->flag_bits.latent_err = 1; + + return 0; +} + +static int fill_section_fatal(struct ras_core_context *ras_core, + struct cper_section_fatal *fatal, struct ras_log_info *trace) +{ + fatal->data.reg_ctx_type = CPER_CTX_TYPE__CRASH; + fatal->data.reg_arr_size = sizeof(fatal->data.reg); + + fatal->data.reg.status = trace->aca_reg.regs[RAS_CPER_ACA_REG_STATUS]; + fatal->data.reg.addr = trace->aca_reg.regs[RAS_CPER_ACA_REG_ADDR]; + fatal->data.reg.ipid = trace->aca_reg.regs[RAS_CPER_ACA_REG_IPID]; + fatal->data.reg.synd = trace->aca_reg.regs[RAS_CPER_ACA_REG_SYND]; + + return 0; +} + +static int fill_section_runtime(struct ras_core_context *ras_core, + struct cper_section_runtime *runtime, struct ras_log_info *trace, + enum ras_cper_severity sev) +{ + runtime->hdr.valid_bits.err_info_cnt = 1; + runtime->hdr.valid_bits.err_context_cnt = 1; + + runtime->descriptor.error_type = RUNTIME; + runtime->descriptor.ms_chk_bits.err_type_valid = 1; + if (sev == RAS_CPER_SEV_RMA) { + runtime->descriptor.valid_bits.ms_chk = 1; + runtime->descriptor.ms_chk_bits.err_type = 1; + runtime->descriptor.ms_chk_bits.pcc = 1; + } + + runtime->reg.reg_ctx_type = CPER_CTX_TYPE__CRASH; + runtime->reg.reg_arr_size = sizeof(runtime->reg.reg_dump); + + runtime->reg.reg_dump[RAS_CPER_ACA_REG_CTL] = trace->aca_reg.regs[ACA_REG_IDX__CTL]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_STATUS] = trace->aca_reg.regs[ACA_REG_IDX__STATUS]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_ADDR] = trace->aca_reg.regs[ACA_REG_IDX__ADDR]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_MISC0] = trace->aca_reg.regs[ACA_REG_IDX__MISC0]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_CONFIG] = trace->aca_reg.regs[ACA_REG_IDX__CONFG]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_IPID] = trace->aca_reg.regs[ACA_REG_IDX__IPID]; + runtime->reg.reg_dump[RAS_CPER_ACA_REG_SYND] = trace->aca_reg.regs[ACA_REG_IDX__SYND]; + + return 0; +} + +static int cper_generate_runtime_record(struct ras_core_context *ras_core, + struct cper_section_hdr *hdr, struct ras_log_info **trace_arr, uint32_t arr_num, + enum ras_cper_severity sev) +{ + struct cper_section_descriptor *descriptor; + struct cper_section_runtime *runtime; + int i; + + fill_section_hdr(ras_core, hdr, RAS_CPER_TYPE_RUNTIME, sev, trace_arr[0]); + hdr->record_length = RAS_HDR_LEN + ((RAS_SEC_DESC_LEN + RAS_NONSTD_SEC_LEN) * arr_num); + hdr->sec_cnt = arr_num; + for (i = 0; i < arr_num; i++) { + descriptor = (struct cper_section_descriptor *)((uint8_t *)hdr + + RAS_SEC_DESC_OFFSET(i)); + runtime = (struct cper_section_runtime *)((uint8_t *)hdr + + RAS_NONSTD_SEC_OFFSET(hdr->sec_cnt, i)); + + fill_section_descriptor(ras_core, descriptor, sev, RUNTIME, + RAS_NONSTD_SEC_OFFSET(hdr->sec_cnt, i), + sizeof(struct cper_section_runtime)); + fill_section_runtime(ras_core, runtime, trace_arr[i], sev); + } + + return 0; +} + +static int cper_generate_fatal_record(struct ras_core_context *ras_core, + uint8_t *buffer, struct ras_log_info **trace_arr, uint32_t arr_num) +{ + struct ras_cper_fatal_record record = {0}; + int i = 0; + + for (i = 0; i < arr_num; i++) { + fill_section_hdr(ras_core, &record.hdr, RAS_CPER_TYPE_FATAL, + RAS_CPER_SEV_FATAL_UE, trace_arr[i]); + record.hdr.record_length = RAS_HDR_LEN + RAS_SEC_DESC_LEN + RAS_FATAL_SEC_LEN; + record.hdr.sec_cnt = 1; + + fill_section_descriptor(ras_core, &record.descriptor, RAS_CPER_SEV_FATAL_UE, + CRASHDUMP, offsetof(struct ras_cper_fatal_record, fatal), + sizeof(struct cper_section_fatal)); + + fill_section_fatal(ras_core, &record.fatal, trace_arr[i]); + + memcpy(buffer + (i * record.hdr.record_length), + &record, record.hdr.record_length); + } + + return 0; +} + +static int cper_get_record_size(enum ras_cper_type type, uint16_t section_count) +{ + int size = 0; + + size += RAS_HDR_LEN; + size += (RAS_SEC_DESC_LEN * section_count); + + switch (type) { + case RAS_CPER_TYPE_RUNTIME: + case RAS_CPER_TYPE_RMA: + size += (RAS_NONSTD_SEC_LEN * section_count); + break; + case RAS_CPER_TYPE_FATAL: + size += (RAS_FATAL_SEC_LEN * section_count); + size += (RAS_HDR_LEN * (section_count - 1)); + break; + case RAS_CPER_TYPE_BOOT: + size += (RAS_BOOT_SEC_LEN * section_count); + break; + default: + /* should never reach here */ + break; + } + + return size; +} + +static enum ras_cper_type cper_ras_log_event_to_cper_type(enum ras_log_event event) +{ + switch (event) { + case RAS_LOG_EVENT_UE: + return RAS_CPER_TYPE_FATAL; + case RAS_LOG_EVENT_DE: + case RAS_LOG_EVENT_CE: + case RAS_LOG_EVENT_POISON_CREATION: + case RAS_LOG_EVENT_POISON_CONSUMPTION: + return RAS_CPER_TYPE_RUNTIME; + case RAS_LOG_EVENT_RMA: + return RAS_CPER_TYPE_RMA; + default: + /* should never reach here */ + return RAS_CPER_TYPE_RUNTIME; + } +} + +int ras_cper_generate_cper(struct ras_core_context *ras_core, + struct ras_log_info **trace_list, uint32_t count, + uint8_t *buf, uint32_t buf_len, uint32_t *real_data_len) +{ + uint8_t *buffer = buf; + uint64_t buf_size = buf_len; + int record_size, saved_size = 0; + struct cper_section_hdr *hdr; + + /* All the batch traces share the same event */ + record_size = cper_get_record_size( + cper_ras_log_event_to_cper_type(trace_list[0]->event), count); + + if ((record_size + saved_size) > buf_size) + return -ENOMEM; + + hdr = (struct cper_section_hdr *)(buffer + saved_size); + + switch (trace_list[0]->event) { + case RAS_LOG_EVENT_RMA: + cper_generate_runtime_record(ras_core, hdr, trace_list, count, RAS_CPER_SEV_RMA); + break; + case RAS_LOG_EVENT_DE: + cper_generate_runtime_record(ras_core, + hdr, trace_list, count, RAS_CPER_SEV_NON_FATAL_UE); + break; + case RAS_LOG_EVENT_CE: + cper_generate_runtime_record(ras_core, + hdr, trace_list, count, RAS_CPER_SEV_NON_FATAL_CE); + break; + case RAS_LOG_EVENT_UE: + cper_generate_fatal_record(ras_core, buffer + saved_size, trace_list, count); + break; + default: + RAS_DEV_WARN(ras_core->dev, "Unprocessed trace event: %d\n", trace_list[0]->event); + break; + } + + saved_size += record_size; + + *real_data_len = saved_size; + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_cper.h b/drivers/gpu/drm/amd/ras/rascore/ras_cper.h new file mode 100644 index 000000000000..076c1883c1ce --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_cper.h @@ -0,0 +1,304 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_CPER_H__ +#define __RAS_CPER_H__ + +#define CPER_UUID_MAX_SIZE 16 +struct ras_cper_guid { + uint8_t b[CPER_UUID_MAX_SIZE]; +}; + +#define CPER_GUID__INIT(a, b, c, d0, d1, d2, d3, d4, d5, d6, d7) \ + ((struct ras_cper_guid) \ + {{ (a) & 0xff, ((a) >> 8) & 0xff, ((a) >> 16) & 0xff, ((a) >> 24) & 0xff, \ + (b) & 0xff, ((b) >> 8) & 0xff, \ + (c) & 0xff, ((c) >> 8) & 0xff, \ + (d0), (d1), (d2), (d3), (d4), (d5), (d6), (d7) }}) + +#define CPER_HDR__REV_1 (0x100) +#define CPER_SEC__MINOR_REV_1 (0x01) +#define CPER_SEC__MAJOR_REV_22 (0x22) +#define CPER_OAM_MAX_COUNT (8) + +#define CPER_CTX_TYPE__CRASH (1) +#define CPER_CTX_TYPE__BOOT (9) + +#define CPER_CREATOR_ID__AMDGPU "amdgpu" + +#define CPER_NOTIFY__MCE \ + CPER_GUID__INIT(0xE8F56FFE, 0x919C, 0x4cc5, 0xBA, 0x88, 0x65, 0xAB, \ + 0xE1, 0x49, 0x13, 0xBB) +#define CPER_NOTIFY__CMC \ + CPER_GUID__INIT(0x2DCE8BB1, 0xBDD7, 0x450e, 0xB9, 0xAD, 0x9C, 0xF4, \ + 0xEB, 0xD4, 0xF8, 0x90) +#define BOOT__TYPE \ + CPER_GUID__INIT(0x3D61A466, 0xAB40, 0x409a, 0xA6, 0x98, 0xF3, 0x62, \ + 0xD4, 0x64, 0xB3, 0x8F) + +#define GPU__CRASHDUMP \ + CPER_GUID__INIT(0x32AC0C78, 0x2623, 0x48F6, 0xB0, 0xD0, 0x73, 0x65, \ + 0x72, 0x5F, 0xD6, 0xAE) +#define GPU__NONSTANDARD_ERROR \ + CPER_GUID__INIT(0x32AC0C78, 0x2623, 0x48F6, 0x81, 0xA2, 0xAC, 0x69, \ + 0x17, 0x80, 0x55, 0x1D) +#define PROC_ERR__SECTION_TYPE \ + CPER_GUID__INIT(0xDC3EA0B0, 0xA144, 0x4797, 0xB9, 0x5B, 0x53, 0xFA, \ + 0x24, 0x2B, 0x6E, 0x1D) + +enum ras_cper_type { + RAS_CPER_TYPE_RUNTIME, + RAS_CPER_TYPE_FATAL, + RAS_CPER_TYPE_BOOT, + RAS_CPER_TYPE_RMA, +}; + +enum ras_cper_severity { + RAS_CPER_SEV_NON_FATAL_UE = 0, + RAS_CPER_SEV_FATAL_UE = 1, + RAS_CPER_SEV_NON_FATAL_CE = 2, + RAS_CPER_SEV_RMA = 3, + + RAS_CPER_SEV_UNUSED = 10, +}; + +enum ras_cper_aca_reg { + RAS_CPER_ACA_REG_CTL = 0, + RAS_CPER_ACA_REG_STATUS = 1, + RAS_CPER_ACA_REG_ADDR = 2, + RAS_CPER_ACA_REG_MISC0 = 3, + RAS_CPER_ACA_REG_CONFIG = 4, + RAS_CPER_ACA_REG_IPID = 5, + RAS_CPER_ACA_REG_SYND = 6, + RAS_CPER_ACA_REG_DESTAT = 8, + RAS_CPER_ACA_REG_DEADDR = 9, + RAS_CPER_ACA_REG_MASK = 10, + + RAS_CPER_ACA_REG_COUNT = 16, +}; + +#pragma pack(push, 1) + +struct ras_cper_timestamp { + uint8_t seconds; + uint8_t minutes; + uint8_t hours; + uint8_t flag; + uint8_t day; + uint8_t month; + uint8_t year; + uint8_t century; +}; + +struct cper_section_hdr { + char signature[4]; /* "CPER" */ + uint16_t revision; + uint32_t signature_end; /* 0xFFFFFFFF */ + uint16_t sec_cnt; + enum ras_cper_severity error_severity; + union { + struct { + uint32_t platform_id : 1; + uint32_t timestamp : 1; + uint32_t partition_id : 1; + uint32_t reserved : 29; + } valid_bits; + uint32_t valid_mask; + }; + uint32_t record_length; /* Total size of CPER Entry */ + struct ras_cper_timestamp timestamp; + char platform_id[16]; + struct ras_cper_guid partition_id; /* Reserved */ + char creator_id[16]; + struct ras_cper_guid notify_type; /* CMC, MCE */ + char record_id[8]; /* Unique CPER Entry ID */ + uint32_t flags; /* Reserved */ + uint64_t persistence_info; /* Reserved */ + uint8_t reserved[12]; /* Reserved */ +}; + +struct cper_section_descriptor { + uint32_t sec_offset; /* Offset from the start of CPER entry */ + uint32_t sec_length; + uint8_t revision_minor; /* CPER_SEC_MINOR_REV_1 */ + uint8_t revision_major; /* CPER_SEC_MAJOR_REV_22 */ + union { + struct { + uint8_t fru_id : 1; + uint8_t fru_text : 1; + uint8_t reserved : 6; + } valid_bits; + uint8_t valid_mask; + }; + uint8_t reserved; + union { + struct { + uint32_t primary : 1; + uint32_t reserved1 : 2; + uint32_t exceed_err_threshold : 1; + uint32_t latent_err : 1; + uint32_t reserved2 : 27; + } flag_bits; + uint32_t flag_mask; + }; + struct ras_cper_guid sec_type; + char fru_id[16]; + enum ras_cper_severity severity; + char fru_text[20]; +}; + +struct runtime_hdr { + union { + struct { + uint64_t apic_id : 1; + uint64_t fw_id : 1; + uint64_t err_info_cnt : 6; + uint64_t err_context_cnt : 6; + } valid_bits; + uint64_t valid_mask; + }; + uint64_t apic_id; + char fw_id[48]; +}; + +struct runtime_descriptor { + struct ras_cper_guid error_type; + union { + struct { + uint64_t ms_chk : 1; + uint64_t target_addr_id : 1; + uint64_t req_id : 1; + uint64_t resp_id : 1; + uint64_t instr_ptr : 1; + uint64_t reserved : 59; + } valid_bits; + uint64_t valid_mask; + }; + union { + struct { + uint64_t err_type_valid : 1; + uint64_t pcc_valid : 1; + uint64_t uncorr_valid : 1; + uint64_t precise_ip_valid : 1; + uint64_t restartable_ip_valid : 1; + uint64_t overflow_valid : 1; + uint64_t reserved1 : 10; + uint64_t err_type : 2; + uint64_t pcc : 1; + uint64_t uncorr : 1; + uint64_t precised_ip : 1; + uint64_t restartable_ip : 1; + uint64_t overflow : 1; + uint64_t reserved2 : 41; + } ms_chk_bits; + uint64_t ms_chk_mask; + }; + uint64_t target_addr_id; + uint64_t req_id; + uint64_t resp_id; + uint64_t instr_ptr; +}; + +struct runtime_error_reg { + uint16_t reg_ctx_type; + uint16_t reg_arr_size; + uint32_t msr_addr; + uint64_t mm_reg_addr; + uint64_t reg_dump[RAS_CPER_ACA_REG_COUNT]; +}; + +struct cper_section_runtime { + struct runtime_hdr hdr; + struct runtime_descriptor descriptor; + struct runtime_error_reg reg; +}; + +struct crashdump_hdr { + uint64_t reserved1; + uint64_t reserved2; + char fw_id[48]; + uint64_t reserved3[8]; +}; + +struct fatal_reg_info { + uint64_t status; + uint64_t addr; + uint64_t ipid; + uint64_t synd; +}; + +struct crashdump_fatal { + uint16_t reg_ctx_type; + uint16_t reg_arr_size; + uint32_t reserved1; + uint64_t reserved2; + struct fatal_reg_info reg; +}; + +struct crashdump_boot { + uint16_t reg_ctx_type; + uint16_t reg_arr_size; + uint32_t reserved1; + uint64_t reserved2; + uint64_t msg[CPER_OAM_MAX_COUNT]; +}; + +struct cper_section_fatal { + struct crashdump_hdr hdr; + struct crashdump_fatal data; +}; + +struct cper_section_boot { + struct crashdump_hdr hdr; + struct crashdump_boot data; +}; + +struct ras_cper_fatal_record { + struct cper_section_hdr hdr; + struct cper_section_descriptor descriptor; + struct cper_section_fatal fatal; +}; +#pragma pack(pop) + +#define RAS_HDR_LEN (sizeof(struct cper_section_hdr)) +#define RAS_SEC_DESC_LEN (sizeof(struct cper_sec_desc)) + +#define RAS_BOOT_SEC_LEN (sizeof(struct cper_sec_crashdump_boot)) +#define RAS_FATAL_SEC_LEN (sizeof(struct cper_sec_crashdump_fatal)) +#define RAS_NONSTD_SEC_LEN (sizeof(struct cper_sec_nonstd_err)) + +#define RAS_SEC_DESC_OFFSET(idx) (RAS_HDR_LEN + (RAS_SEC_DESC_LEN * idx)) + +#define RAS_BOOT_SEC_OFFSET(count, idx) \ + (RAS_HDR_LEN + (RAS_SEC_DESC_LEN * count) + (RAS_BOOT_SEC_LEN * idx)) +#define RAS_FATAL_SEC_OFFSET(count, idx) \ + (RAS_HDR_LEN + (RAS_SEC_DESC_LEN * count) + (RAS_FATAL_SEC_LEN * idx)) +#define RAS_NONSTD_SEC_OFFSET(count, idx) \ + (RAS_HDR_LEN + (RAS_SEC_DESC_LEN * count) + (RAS_NONSTD_SEC_LEN * idx)) + +struct ras_core_context; +struct ras_log_info; +int ras_cper_generate_cper(struct ras_core_context *ras_core, + struct ras_log_info **trace_list, uint32_t count, + uint8_t *buf, uint32_t buf_len, uint32_t *real_data_len); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.c b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.c new file mode 100644 index 000000000000..cd6b057bdaf3 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.c @@ -0,0 +1,1339 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras_eeprom.h" +#include "ras.h" + +/* These are memory addresses as would be seen by one or more EEPROM + * chips strung on the I2C bus, usually by manipulating pins 1-3 of a + * set of EEPROM devices. They form a continuous memory space. + * + * The I2C device address includes the device type identifier, 1010b, + * which is a reserved value and indicates that this is an I2C EEPROM + * device. It also includes the top 3 bits of the 19 bit EEPROM memory + * address, namely bits 18, 17, and 16. This makes up the 7 bit + * address sent on the I2C bus with bit 0 being the direction bit, + * which is not represented here, and sent by the hardware directly. + * + * For instance, + * 50h = 1010000b => device type identifier 1010b, bits 18:16 = 000b, address 0. + * 54h = 1010100b => --"--, bits 18:16 = 100b, address 40000h. + * 56h = 1010110b => --"--, bits 18:16 = 110b, address 60000h. + * Depending on the size of the I2C EEPROM device(s), bits 18:16 may + * address memory in a device or a device on the I2C bus, depending on + * the status of pins 1-3. + * + * The RAS table lives either at address 0 or address 40000h of EEPROM. + */ +#define EEPROM_I2C_MADDR_0 0x0 +#define EEPROM_I2C_MADDR_4 0x40000 + +#define EEPROM_PAGE_BITS 8 +#define EEPROM_PAGE_SIZE (1U << EEPROM_PAGE_BITS) +#define EEPROM_PAGE_MASK (EEPROM_PAGE_SIZE - 1) + +#define EEPROM_OFFSET_SIZE 2 +#define MAKE_I2C_ADDR(_aa) ((0xA << 3) | (((_aa) >> 16) & 0xF)) + +/* + * The 2 macros bellow represent the actual size in bytes that + * those entities occupy in the EEPROM memory. + * RAS_TABLE_RECORD_SIZE is different than sizeof(eeprom_umc_record) which + * uses uint64 to store 6b fields such as retired_page. + */ +#define RAS_TABLE_HEADER_SIZE 20 +#define RAS_TABLE_RECORD_SIZE 24 + +/* Table hdr is 'AMDR' */ +#define RAS_TABLE_HDR_VAL 0x414d4452 + +/* Bad GPU tag ‘BADG’ */ +#define RAS_TABLE_HDR_BAD 0x42414447 + +/* + * EEPROM Table structure v1 + * --------------------------------- + * | | + * | EEPROM TABLE HEADER | + * | ( size 20 Bytes ) | + * | | + * --------------------------------- + * | | + * | BAD PAGE RECORD AREA | + * | | + * --------------------------------- + */ + +/* Assume 2-Mbit size EEPROM and take up the whole space. */ +#define RAS_TBL_SIZE_BYTES (256 * 1024) +#define RAS_TABLE_START 0 +#define RAS_HDR_START RAS_TABLE_START +#define RAS_RECORD_START (RAS_HDR_START + RAS_TABLE_HEADER_SIZE) +#define RAS_MAX_RECORD_COUNT ((RAS_TBL_SIZE_BYTES - RAS_TABLE_HEADER_SIZE) \ + / RAS_TABLE_RECORD_SIZE) + +/* + * EEPROM Table structrue v2.1 + * --------------------------------- + * | | + * | EEPROM TABLE HEADER | + * | ( size 20 Bytes ) | + * | | + * --------------------------------- + * | | + * | EEPROM TABLE RAS INFO | + * | (available info size 4 Bytes) | + * | ( reserved size 252 Bytes ) | + * | | + * --------------------------------- + * | | + * | BAD PAGE RECORD AREA | + * | | + * --------------------------------- + */ + +/* EEPROM Table V2_1 */ +#define RAS_TABLE_V2_1_INFO_SIZE 256 +#define RAS_TABLE_V2_1_INFO_START RAS_TABLE_HEADER_SIZE +#define RAS_RECORD_START_V2_1 (RAS_HDR_START + RAS_TABLE_HEADER_SIZE + \ + RAS_TABLE_V2_1_INFO_SIZE) +#define RAS_MAX_RECORD_COUNT_V2_1 ((RAS_TBL_SIZE_BYTES - RAS_TABLE_HEADER_SIZE - \ + RAS_TABLE_V2_1_INFO_SIZE) \ + / RAS_TABLE_RECORD_SIZE) + +/* Given a zero-based index of an EEPROM RAS record, yields the EEPROM + * offset off of RAS_TABLE_START. That is, this is something you can + * add to control->i2c_address, and then tell I2C layer to read + * from/write to there. _N is the so called absolute index, + * because it starts right after the table header. + */ +#define RAS_INDEX_TO_OFFSET(_C, _N) ((_C)->ras_record_offset + \ + (_N) * RAS_TABLE_RECORD_SIZE) + +#define RAS_OFFSET_TO_INDEX(_C, _O) (((_O) - \ + (_C)->ras_record_offset) / RAS_TABLE_RECORD_SIZE) + +/* Given a 0-based relative record index, 0, 1, 2, ..., etc., off + * of "fri", return the absolute record index off of the end of + * the table header. + */ +#define RAS_RI_TO_AI(_C, _I) (((_I) + (_C)->ras_fri) % \ + (_C)->ras_max_record_count) + +#define RAS_NUM_RECS(_tbl_hdr) (((_tbl_hdr)->tbl_size - \ + RAS_TABLE_HEADER_SIZE) / RAS_TABLE_RECORD_SIZE) + +#define RAS_NUM_RECS_V2_1(_tbl_hdr) (((_tbl_hdr)->tbl_size - \ + RAS_TABLE_HEADER_SIZE - \ + RAS_TABLE_V2_1_INFO_SIZE) / RAS_TABLE_RECORD_SIZE) + +#define to_ras_core_context(x) (container_of(x, struct ras_core_context, ras_eeprom)) + +static bool __is_ras_eeprom_supported(struct ras_core_context *ras_core) +{ + return ras_core->ras_eeprom_supported; +} + +static bool __get_eeprom_i2c_addr(struct ras_core_context *ras_core, + struct ras_eeprom_control *control) +{ + int ret = -EINVAL; + + if (control->sys_func && + control->sys_func->update_eeprom_i2c_config) + ret = control->sys_func->update_eeprom_i2c_config(ras_core); + else + RAS_DEV_WARN(ras_core->dev, + "No eeprom i2c system config!\n"); + + return !ret ? true : false; +} + +static int __ras_eeprom_xfer(struct ras_core_context *ras_core, u32 eeprom_addr, + u8 *eeprom_buf, u32 buf_size, bool read) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + int ret; + + if (control->sys_func && control->sys_func->eeprom_i2c_xfer) { + ret = control->sys_func->eeprom_i2c_xfer(ras_core, + eeprom_addr, eeprom_buf, buf_size, read); + + if ((ret > 0) && !read) { + /* According to EEPROM specs the length of the + * self-writing cycle, tWR (tW), is 10 ms. + * + * TODO: Use polling on ACK, aka Acknowledge + * Polling, to minimize waiting for the + * internal write cycle to complete, as it is + * usually smaller than tWR (tW). + */ + msleep(10); + } + + return ret; + } + + RAS_DEV_ERR(ras_core->dev, "Error: No eeprom i2c system xfer function!\n"); + return -EINVAL; +} + +static int __eeprom_xfer(struct ras_core_context *ras_core, u32 eeprom_addr, + u8 *eeprom_buf, u32 buf_size, bool read) +{ + u16 limit; + u16 ps; /* Partial size */ + int res = 0, r; + + if (read) + limit = ras_core->ras_eeprom.max_read_len; + else + limit = ras_core->ras_eeprom.max_write_len; + + if (limit && (limit <= EEPROM_OFFSET_SIZE)) { + RAS_DEV_ERR(ras_core->dev, + "maddr:0x%04X size:0x%02X:quirk max_%s_len must be > %d", + eeprom_addr, buf_size, + read ? "read" : "write", EEPROM_OFFSET_SIZE); + return -EINVAL; + } + + ras_core_down_gpu_reset_lock(ras_core); + + if (limit == 0) { + res = __ras_eeprom_xfer(ras_core, eeprom_addr, + eeprom_buf, buf_size, read); + } else { + /* The "limit" includes all data bytes sent/received, + * which would include the EEPROM_OFFSET_SIZE bytes. + * Account for them here. + */ + limit -= EEPROM_OFFSET_SIZE; + for ( ; buf_size > 0; + buf_size -= ps, eeprom_addr += ps, eeprom_buf += ps) { + ps = (buf_size < limit) ? buf_size : limit; + + r = __ras_eeprom_xfer(ras_core, eeprom_addr, + eeprom_buf, ps, read); + if (r < 0) + break; + + res += r; + } + } + + ras_core_up_gpu_reset_lock(ras_core); + + return res; +} + +static int __eeprom_read(struct ras_core_context *ras_core, + u32 eeprom_addr, u8 *eeprom_buf, u32 bytes) +{ + return __eeprom_xfer(ras_core, eeprom_addr, + eeprom_buf, bytes, true); +} + +static int __eeprom_write(struct ras_core_context *ras_core, + u32 eeprom_addr, u8 *eeprom_buf, u32 bytes) +{ + return __eeprom_xfer(ras_core, eeprom_addr, + eeprom_buf, bytes, false); +} + +static void +__encode_table_header_to_buf(struct ras_eeprom_table_header *hdr, + unsigned char *buf) +{ + u32 *pp = (uint32_t *)buf; + + pp[0] = cpu_to_le32(hdr->header); + pp[1] = cpu_to_le32(hdr->version); + pp[2] = cpu_to_le32(hdr->first_rec_offset); + pp[3] = cpu_to_le32(hdr->tbl_size); + pp[4] = cpu_to_le32(hdr->checksum); +} + +static void +__decode_table_header_from_buf(struct ras_eeprom_table_header *hdr, + unsigned char *buf) +{ + u32 *pp = (uint32_t *)buf; + + hdr->header = le32_to_cpu(pp[0]); + hdr->version = le32_to_cpu(pp[1]); + hdr->first_rec_offset = le32_to_cpu(pp[2]); + hdr->tbl_size = le32_to_cpu(pp[3]); + hdr->checksum = le32_to_cpu(pp[4]); +} + +static int __write_table_header(struct ras_eeprom_control *control) +{ + u8 buf[RAS_TABLE_HEADER_SIZE]; + struct ras_core_context *ras_core = to_ras_core_context(control); + int res; + + memset(buf, 0, sizeof(buf)); + __encode_table_header_to_buf(&control->tbl_hdr, buf); + + /* i2c may be unstable in gpu reset */ + res = __eeprom_write(ras_core, + control->i2c_address + + control->ras_header_offset, + buf, RAS_TABLE_HEADER_SIZE); + + if (res < 0) { + RAS_DEV_ERR(ras_core->dev, + "Failed to write EEPROM table header:%d\n", res); + } else if (res < RAS_TABLE_HEADER_SIZE) { + RAS_DEV_ERR(ras_core->dev, + "Short write:%d out of %d\n", res, RAS_TABLE_HEADER_SIZE); + res = -EIO; + } else { + res = 0; + } + + return res; +} + +static void +__encode_table_ras_info_to_buf(struct ras_eeprom_table_ras_info *rai, + unsigned char *buf) +{ + u32 *pp = (uint32_t *)buf; + u32 tmp; + + tmp = ((uint32_t)(rai->rma_status) & 0xFF) | + (((uint32_t)(rai->health_percent) << 8) & 0xFF00) | + (((uint32_t)(rai->ecc_page_threshold) << 16) & 0xFFFF0000); + pp[0] = cpu_to_le32(tmp); +} + +static void +__decode_table_ras_info_from_buf(struct ras_eeprom_table_ras_info *rai, + unsigned char *buf) +{ + u32 *pp = (uint32_t *)buf; + u32 tmp; + + tmp = le32_to_cpu(pp[0]); + rai->rma_status = tmp & 0xFF; + rai->health_percent = (tmp >> 8) & 0xFF; + rai->ecc_page_threshold = (tmp >> 16) & 0xFFFF; +} + +static int __write_table_ras_info(struct ras_eeprom_control *control) +{ + struct ras_core_context *ras_core = to_ras_core_context(control); + u8 *buf; + int res; + + buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL); + if (!buf) { + RAS_DEV_ERR(ras_core->dev, + "Failed to alloc buf to write table ras info\n"); + return -ENOMEM; + } + + __encode_table_ras_info_to_buf(&control->tbl_rai, buf); + + /* i2c may be unstable in gpu reset */ + res = __eeprom_write(ras_core, + control->i2c_address + + control->ras_info_offset, + buf, RAS_TABLE_V2_1_INFO_SIZE); + + if (res < 0) { + RAS_DEV_ERR(ras_core->dev, + "Failed to write EEPROM table ras info:%d\n", res); + } else if (res < RAS_TABLE_V2_1_INFO_SIZE) { + RAS_DEV_ERR(ras_core->dev, + "Short write:%d out of %d\n", res, RAS_TABLE_V2_1_INFO_SIZE); + res = -EIO; + } else { + res = 0; + } + + kfree(buf); + + return res; +} + +static u8 __calc_hdr_byte_sum(const struct ras_eeprom_control *control) +{ + int ii; + u8 *pp, csum; + u32 sz; + + /* Header checksum, skip checksum field in the calculation */ + sz = sizeof(control->tbl_hdr) - sizeof(control->tbl_hdr.checksum); + pp = (u8 *) &control->tbl_hdr; + csum = 0; + for (ii = 0; ii < sz; ii++, pp++) + csum += *pp; + + return csum; +} + +static u8 __calc_ras_info_byte_sum(const struct ras_eeprom_control *control) +{ + int ii; + u8 *pp, csum; + u32 sz; + + sz = sizeof(control->tbl_rai); + pp = (u8 *) &control->tbl_rai; + csum = 0; + for (ii = 0; ii < sz; ii++, pp++) + csum += *pp; + + return csum; +} + +static int ras_eeprom_correct_header_tag( + struct ras_eeprom_control *control, + uint32_t header) +{ + struct ras_eeprom_table_header *hdr = &control->tbl_hdr; + u8 *hh; + int res; + u8 csum; + + csum = -hdr->checksum; + + hh = (void *) &hdr->header; + csum -= (hh[0] + hh[1] + hh[2] + hh[3]); + hh = (void *) &header; + csum += hh[0] + hh[1] + hh[2] + hh[3]; + csum = -csum; + mutex_lock(&control->ras_tbl_mutex); + hdr->header = header; + hdr->checksum = csum; + res = __write_table_header(control); + mutex_unlock(&control->ras_tbl_mutex); + + return res; +} + +static void ras_set_eeprom_table_version(struct ras_eeprom_control *control) +{ + struct ras_eeprom_table_header *hdr = &control->tbl_hdr; + + hdr->version = RAS_TABLE_VER_V3; +} + +int ras_eeprom_reset_table(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + struct ras_eeprom_table_header *hdr = &control->tbl_hdr; + struct ras_eeprom_table_ras_info *rai = &control->tbl_rai; + u8 csum; + int res; + + mutex_lock(&control->ras_tbl_mutex); + + hdr->header = RAS_TABLE_HDR_VAL; + ras_set_eeprom_table_version(control); + + if (hdr->version >= RAS_TABLE_VER_V2_1) { + hdr->first_rec_offset = RAS_RECORD_START_V2_1; + hdr->tbl_size = RAS_TABLE_HEADER_SIZE + + RAS_TABLE_V2_1_INFO_SIZE; + rai->rma_status = RAS_GPU_HEALTH_USABLE; + /** + * GPU health represented as a percentage. + * 0 means worst health, 100 means fully health. + */ + rai->health_percent = 100; + /* ecc_page_threshold = 0 means disable bad page retirement */ + rai->ecc_page_threshold = control->record_threshold_count; + } else { + hdr->first_rec_offset = RAS_RECORD_START; + hdr->tbl_size = RAS_TABLE_HEADER_SIZE; + } + + csum = __calc_hdr_byte_sum(control); + if (hdr->version >= RAS_TABLE_VER_V2_1) + csum += __calc_ras_info_byte_sum(control); + csum = -csum; + hdr->checksum = csum; + res = __write_table_header(control); + if (!res && hdr->version > RAS_TABLE_VER_V1) + res = __write_table_ras_info(control); + + control->ras_num_recs = 0; + control->ras_fri = 0; + + control->bad_channel_bitmap = 0; + ras_core_event_notify(ras_core, RAS_EVENT_ID__UPDATE_BAD_PAGE_NUM, + &control->ras_num_recs); + ras_core_event_notify(ras_core, RAS_EVENT_ID__UPDATE_BAD_CHANNEL_BITMAP, + &control->bad_channel_bitmap); + control->update_channel_flag = false; + + mutex_unlock(&control->ras_tbl_mutex); + + return res; +} + +static void +__encode_table_record_to_buf(struct ras_eeprom_control *control, + struct eeprom_umc_record *record, + unsigned char *buf) +{ + __le64 tmp = 0; + int i = 0; + + /* Next are all record fields according to EEPROM page spec in LE foramt */ + buf[i++] = record->err_type; + + buf[i++] = record->bank; + + tmp = cpu_to_le64(record->ts); + memcpy(buf + i, &tmp, 8); + i += 8; + + tmp = cpu_to_le64((record->offset & 0xffffffffffff)); + memcpy(buf + i, &tmp, 6); + i += 6; + + buf[i++] = record->mem_channel; + buf[i++] = record->mcumc_id; + + tmp = cpu_to_le64((record->retired_row_pfn & 0xffffffffffff)); + memcpy(buf + i, &tmp, 6); +} + +static void +__decode_table_record_from_buf(struct ras_eeprom_control *control, + struct eeprom_umc_record *record, + unsigned char *buf) +{ + __le64 tmp = 0; + int i = 0; + + /* Next are all record fields according to EEPROM page spec in LE foramt */ + record->err_type = buf[i++]; + + record->bank = buf[i++]; + + memcpy(&tmp, buf + i, 8); + record->ts = le64_to_cpu(tmp); + i += 8; + + memcpy(&tmp, buf + i, 6); + record->offset = (le64_to_cpu(tmp) & 0xffffffffffff); + i += 6; + + record->mem_channel = buf[i++]; + record->mcumc_id = buf[i++]; + + memcpy(&tmp, buf + i, 6); + record->retired_row_pfn = (le64_to_cpu(tmp) & 0xffffffffffff); +} + +bool ras_eeprom_check_safety_watermark(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + bool ret = false; + int bad_page_count; + + if (!__is_ras_eeprom_supported(ras_core) || + !control->record_threshold_config) + return false; + + bad_page_count = ras_umc_get_badpage_count(ras_core); + if (control->tbl_hdr.header == RAS_TABLE_HDR_BAD) { + if (bad_page_count > control->record_threshold_count) + RAS_DEV_WARN(ras_core->dev, "RAS records:%d exceed threshold:%d", + bad_page_count, control->record_threshold_count); + + if ((control->record_threshold_config == WARN_NONSTOP_OVER_THRESHOLD) || + (control->record_threshold_config == NONSTOP_OVER_THRESHOLD)) { + RAS_DEV_WARN(ras_core->dev, + "Please consult AMD Service Action Guide (SAG) for appropriate service procedures.\n"); + ret = false; + } else { + ras_core->is_rma = true; + RAS_DEV_WARN(ras_core->dev, + "Please consider adjusting the customized threshold.\n"); + ret = true; + } + } + + return ret; +} + +/** + * __ras_eeprom_write -- write indexed from buffer to EEPROM + * @control: pointer to control structure + * @buf: pointer to buffer containing data to write + * @fri: start writing at this index + * @num: number of records to write + * + * The caller must hold the table mutex in @control. + * Return 0 on success, -errno otherwise. + */ +static int __ras_eeprom_write(struct ras_eeprom_control *control, + u8 *buf, const u32 fri, const u32 num) +{ + struct ras_core_context *ras_core = to_ras_core_context(control); + u32 buf_size; + int res; + + /* i2c may be unstable in gpu reset */ + buf_size = num * RAS_TABLE_RECORD_SIZE; + res = __eeprom_write(ras_core, + control->i2c_address + RAS_INDEX_TO_OFFSET(control, fri), + buf, buf_size); + if (res < 0) { + RAS_DEV_ERR(ras_core->dev, + "Writing %d EEPROM table records error:%d\n", num, res); + } else if (res < buf_size) { + /* Short write, return error.*/ + RAS_DEV_ERR(ras_core->dev, + "Wrote %d records out of %d\n", + (res/RAS_TABLE_RECORD_SIZE), num); + res = -EIO; + } else { + res = 0; + } + + return res; +} + +static int ras_eeprom_append_table(struct ras_eeprom_control *control, + struct eeprom_umc_record *record, + const u32 num) +{ + u32 a, b, i; + u8 *buf, *pp; + int res; + + buf = kcalloc(num, RAS_TABLE_RECORD_SIZE, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + /* Encode all of them in one go. + */ + pp = buf; + for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) { + __encode_table_record_to_buf(control, &record[i], pp); + + /* update bad channel bitmap */ + if ((record[i].mem_channel < BITS_PER_TYPE(control->bad_channel_bitmap)) && + !(control->bad_channel_bitmap & (1 << record[i].mem_channel))) { + control->bad_channel_bitmap |= 1 << record[i].mem_channel; + control->update_channel_flag = true; + } + } + + /* a, first record index to write into. + * b, last record index to write into. + * a = first index to read (fri) + number of records in the table, + * b = a + @num - 1. + * Let N = control->ras_max_num_record_count, then we have, + * case 0: 0 <= a <= b < N, + * just append @num records starting at a; + * case 1: 0 <= a < N <= b, + * append (N - a) records starting at a, and + * append the remainder, b % N + 1, starting at 0. + * case 2: 0 <= fri < N <= a <= b, then modulo N we get two subcases, + * case 2a: 0 <= a <= b < N + * append num records starting at a; and fix fri if b overwrote it, + * and since a <= b, if b overwrote it then a must've also, + * and if b didn't overwrite it, then a didn't also. + * case 2b: 0 <= b < a < N + * write num records starting at a, which wraps around 0=N + * and overwrite fri unconditionally. Now from case 2a, + * this means that b eclipsed fri to overwrite it and wrap + * around 0 again, i.e. b = 2N+r pre modulo N, so we unconditionally + * set fri = b + 1 (mod N). + * Now, since fri is updated in every case, except the trivial case 0, + * the number of records present in the table after writing, is, + * num_recs - 1 = b - fri (mod N), and we take the positive value, + * by adding an arbitrary multiple of N before taking the modulo N + * as shown below. + */ + a = control->ras_fri + control->ras_num_recs; + b = a + num - 1; + if (b < control->ras_max_record_count) { + res = __ras_eeprom_write(control, buf, a, num); + } else if (a < control->ras_max_record_count) { + u32 g0, g1; + + g0 = control->ras_max_record_count - a; + g1 = b % control->ras_max_record_count + 1; + res = __ras_eeprom_write(control, buf, a, g0); + if (res) + goto Out; + res = __ras_eeprom_write(control, + buf + g0 * RAS_TABLE_RECORD_SIZE, + 0, g1); + if (res) + goto Out; + if (g1 > control->ras_fri) + control->ras_fri = g1 % control->ras_max_record_count; + } else { + a %= control->ras_max_record_count; + b %= control->ras_max_record_count; + + if (a <= b) { + /* Note that, b - a + 1 = num. */ + res = __ras_eeprom_write(control, buf, a, num); + if (res) + goto Out; + if (b >= control->ras_fri) + control->ras_fri = (b + 1) % control->ras_max_record_count; + } else { + u32 g0, g1; + + /* b < a, which means, we write from + * a to the end of the table, and from + * the start of the table to b. + */ + g0 = control->ras_max_record_count - a; + g1 = b + 1; + res = __ras_eeprom_write(control, buf, a, g0); + if (res) + goto Out; + res = __ras_eeprom_write(control, + buf + g0 * RAS_TABLE_RECORD_SIZE, 0, g1); + if (res) + goto Out; + control->ras_fri = g1 % control->ras_max_record_count; + } + } + control->ras_num_recs = 1 + + (control->ras_max_record_count + b - control->ras_fri) + % control->ras_max_record_count; +Out: + kfree(buf); + return res; +} + +static int ras_eeprom_update_header(struct ras_eeprom_control *control) +{ + struct ras_core_context *ras_core = to_ras_core_context(control); + int threshold_config = control->record_threshold_config; + u8 *buf, *pp, csum; + u32 buf_size; + int bad_page_count; + int res; + + bad_page_count = ras_umc_get_badpage_count(ras_core); + /* Modify the header if it exceeds. + */ + if (threshold_config != 0 && + bad_page_count > control->record_threshold_count) { + RAS_DEV_WARN(ras_core->dev, + "Saved bad pages %d reaches threshold value %d\n", + bad_page_count, control->record_threshold_count); + control->tbl_hdr.header = RAS_TABLE_HDR_BAD; + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) { + control->tbl_rai.rma_status = RAS_GPU_RETIRED__ECC_REACH_THRESHOLD; + control->tbl_rai.health_percent = 0; + } + + if ((threshold_config != WARN_NONSTOP_OVER_THRESHOLD) && + (threshold_config != NONSTOP_OVER_THRESHOLD)) + ras_core->is_rma = true; + + /* ignore the -ENOTSUPP return value */ + ras_core_event_notify(ras_core, RAS_EVENT_ID__DEVICE_RMA, NULL); + } + + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) + control->tbl_hdr.tbl_size = RAS_TABLE_HEADER_SIZE + + RAS_TABLE_V2_1_INFO_SIZE + + control->ras_num_recs * RAS_TABLE_RECORD_SIZE; + else + control->tbl_hdr.tbl_size = RAS_TABLE_HEADER_SIZE + + control->ras_num_recs * RAS_TABLE_RECORD_SIZE; + control->tbl_hdr.checksum = 0; + + buf_size = control->ras_num_recs * RAS_TABLE_RECORD_SIZE; + buf = kcalloc(control->ras_num_recs, RAS_TABLE_RECORD_SIZE, GFP_KERNEL); + if (!buf) { + RAS_DEV_ERR(ras_core->dev, + "allocating memory for table of size %d bytes failed\n", + control->tbl_hdr.tbl_size); + res = -ENOMEM; + goto Out; + } + + res = __eeprom_read(ras_core, + control->i2c_address + + control->ras_record_offset, + buf, buf_size); + if (res < 0) { + RAS_DEV_ERR(ras_core->dev, + "EEPROM failed reading records:%d\n", res); + goto Out; + } else if (res < buf_size) { + RAS_DEV_ERR(ras_core->dev, + "EEPROM read %d out of %d bytes\n", res, buf_size); + res = -EIO; + goto Out; + } + + /** + * bad page records have been stored in eeprom, + * now calculate gpu health percent + */ + if (threshold_config != 0 && + control->tbl_hdr.version >= RAS_TABLE_VER_V2_1 && + bad_page_count <= control->record_threshold_count) + control->tbl_rai.health_percent = ((control->record_threshold_count - + bad_page_count) * 100) / control->record_threshold_count; + + /* Recalc the checksum. + */ + csum = 0; + for (pp = buf; pp < buf + buf_size; pp++) + csum += *pp; + + csum += __calc_hdr_byte_sum(control); + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) + csum += __calc_ras_info_byte_sum(control); + /* avoid sign extension when assigning to "checksum" */ + csum = -csum; + control->tbl_hdr.checksum = csum; + res = __write_table_header(control); + if (!res && control->tbl_hdr.version > RAS_TABLE_VER_V1) + res = __write_table_ras_info(control); +Out: + kfree(buf); + return res; +} + +/** + * ras_core_eeprom_append -- append records to the EEPROM RAS table + * @control: pointer to control structure + * @record: array of records to append + * @num: number of records in @record array + * + * Append @num records to the table, calculate the checksum and write + * the table back to EEPROM. The maximum number of records that + * can be appended is between 1 and control->ras_max_record_count, + * regardless of how many records are already stored in the table. + * + * Return 0 on success or if EEPROM is not supported, -errno on error. + */ +int ras_eeprom_append(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, const u32 num) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + int res; + + if (!__is_ras_eeprom_supported(ras_core)) + return 0; + + if (num == 0) { + RAS_DEV_ERR(ras_core->dev, "will not append 0 records\n"); + return -EINVAL; + } else if ((num + control->ras_num_recs) > control->ras_max_record_count) { + RAS_DEV_ERR(ras_core->dev, + "cannot append %d records than the size of table %d\n", + num, control->ras_max_record_count); + return -EINVAL; + } + + mutex_lock(&control->ras_tbl_mutex); + res = ras_eeprom_append_table(control, record, num); + if (!res) + res = ras_eeprom_update_header(control); + + mutex_unlock(&control->ras_tbl_mutex); + + return res; +} + +/** + * __ras_eeprom_read -- read indexed from EEPROM into buffer + * @control: pointer to control structure + * @buf: pointer to buffer to read into + * @fri: first record index, start reading at this index, absolute index + * @num: number of records to read + * + * The caller must hold the table mutex in @control. + * Return 0 on success, -errno otherwise. + */ +static int __ras_eeprom_read(struct ras_eeprom_control *control, + u8 *buf, const u32 fri, const u32 num) +{ + struct ras_core_context *ras_core = to_ras_core_context(control); + u32 buf_size; + int res; + + /* i2c may be unstable in gpu reset */ + buf_size = num * RAS_TABLE_RECORD_SIZE; + res = __eeprom_read(ras_core, + control->i2c_address + + RAS_INDEX_TO_OFFSET(control, fri), + buf, buf_size); + if (res < 0) { + RAS_DEV_ERR(ras_core->dev, + "Reading %d EEPROM table records error:%d\n", num, res); + } else if (res < buf_size) { + /* Short read, return error. + */ + RAS_DEV_ERR(ras_core->dev, + "Read %d records out of %d\n", + (res/RAS_TABLE_RECORD_SIZE), num); + res = -EIO; + } else { + res = 0; + } + + return res; +} + +int ras_eeprom_read(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, const u32 num) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + int i, res; + u8 *buf, *pp; + u32 g0, g1; + + if (!__is_ras_eeprom_supported(ras_core)) + return 0; + + if (num == 0) { + RAS_DEV_ERR(ras_core->dev, "will not read 0 records\n"); + return -EINVAL; + } else if (num > control->ras_num_recs) { + RAS_DEV_ERR(ras_core->dev, + "too many records to read:%d available:%d\n", + num, control->ras_num_recs); + return -EINVAL; + } + + buf = kcalloc(num, RAS_TABLE_RECORD_SIZE, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + /* Determine how many records to read, from the first record + * index, fri, to the end of the table, and from the beginning + * of the table, such that the total number of records is + * @num, and we handle wrap around when fri > 0 and + * fri + num > RAS_MAX_RECORD_COUNT. + * + * First we compute the index of the last element + * which would be fetched from each region, + * g0 is in [fri, fri + num - 1], and + * g1 is in [0, RAS_MAX_RECORD_COUNT - 1]. + * Then, if g0 < RAS_MAX_RECORD_COUNT, the index of + * the last element to fetch, we set g0 to _the number_ + * of elements to fetch, @num, since we know that the last + * indexed to be fetched does not exceed the table. + * + * If, however, g0 >= RAS_MAX_RECORD_COUNT, then + * we set g0 to the number of elements to read + * until the end of the table, and g1 to the number of + * elements to read from the beginning of the table. + */ + g0 = control->ras_fri + num - 1; + g1 = g0 % control->ras_max_record_count; + if (g0 < control->ras_max_record_count) { + g0 = num; + g1 = 0; + } else { + g0 = control->ras_max_record_count - control->ras_fri; + g1 += 1; + } + + mutex_lock(&control->ras_tbl_mutex); + res = __ras_eeprom_read(control, buf, control->ras_fri, g0); + if (res) + goto Out; + if (g1) { + res = __ras_eeprom_read(control, + buf + g0 * RAS_TABLE_RECORD_SIZE, 0, g1); + if (res) + goto Out; + } + + res = 0; + + /* Read up everything? Then transform. + */ + pp = buf; + for (i = 0; i < num; i++, pp += RAS_TABLE_RECORD_SIZE) { + __decode_table_record_from_buf(control, &record[i], pp); + + /* update bad channel bitmap */ + if ((record[i].mem_channel < BITS_PER_TYPE(control->bad_channel_bitmap)) && + !(control->bad_channel_bitmap & (1 << record[i].mem_channel))) { + control->bad_channel_bitmap |= 1 << record[i].mem_channel; + control->update_channel_flag = true; + } + } +Out: + kfree(buf); + mutex_unlock(&control->ras_tbl_mutex); + + return res; +} + +uint32_t ras_eeprom_max_record_count(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + + /* get available eeprom table version first before eeprom table init */ + ras_set_eeprom_table_version(control); + + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) + return RAS_MAX_RECORD_COUNT_V2_1; + else + return RAS_MAX_RECORD_COUNT; +} + +/** + * __verify_ras_table_checksum -- verify the RAS EEPROM table checksum + * @control: pointer to control structure + * + * Check the checksum of the stored in EEPROM RAS table. + * + * Return 0 if the checksum is correct, + * positive if it is not correct, and + * -errno on I/O error. + */ +static int __verify_ras_table_checksum(struct ras_eeprom_control *control) +{ + struct ras_core_context *ras_core = to_ras_core_context(control); + int buf_size, res; + u8 csum, *buf, *pp; + + if (control->tbl_hdr.version >= RAS_TABLE_VER_V2_1) + buf_size = RAS_TABLE_HEADER_SIZE + + RAS_TABLE_V2_1_INFO_SIZE + + control->ras_num_recs * RAS_TABLE_RECORD_SIZE; + else + buf_size = RAS_TABLE_HEADER_SIZE + + control->ras_num_recs * RAS_TABLE_RECORD_SIZE; + + buf = kzalloc(buf_size, GFP_KERNEL); + if (!buf) { + RAS_DEV_ERR(ras_core->dev, + "Out of memory checking RAS table checksum.\n"); + return -ENOMEM; + } + + res = __eeprom_read(ras_core, + control->i2c_address + + control->ras_header_offset, + buf, buf_size); + if (res < buf_size) { + RAS_DEV_ERR(ras_core->dev, + "Partial read for checksum, res:%d\n", res); + /* On partial reads, return -EIO. + */ + if (res >= 0) + res = -EIO; + goto Out; + } + + csum = 0; + for (pp = buf; pp < buf + buf_size; pp++) + csum += *pp; +Out: + kfree(buf); + return res < 0 ? res : csum; +} + +static int __read_table_ras_info(struct ras_eeprom_control *control) +{ + struct ras_eeprom_table_ras_info *rai = &control->tbl_rai; + struct ras_core_context *ras_core = to_ras_core_context(control); + unsigned char *buf; + int res; + + buf = kzalloc(RAS_TABLE_V2_1_INFO_SIZE, GFP_KERNEL); + if (!buf) { + RAS_DEV_ERR(ras_core->dev, + "Failed to alloc buf to read EEPROM table ras info\n"); + return -ENOMEM; + } + + /** + * EEPROM table V2_1 supports ras info, + * read EEPROM table ras info + */ + res = __eeprom_read(ras_core, + control->i2c_address + control->ras_info_offset, + buf, RAS_TABLE_V2_1_INFO_SIZE); + if (res < RAS_TABLE_V2_1_INFO_SIZE) { + RAS_DEV_ERR(ras_core->dev, + "Failed to read EEPROM table ras info, res:%d\n", res); + res = res >= 0 ? -EIO : res; + goto Out; + } + + __decode_table_ras_info_from_buf(rai, buf); + +Out: + kfree(buf); + return res == RAS_TABLE_V2_1_INFO_SIZE ? 0 : res; +} + +static int __check_ras_table_status(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + unsigned char buf[RAS_TABLE_HEADER_SIZE] = { 0 }; + struct ras_eeprom_table_header *hdr; + int res; + + hdr = &control->tbl_hdr; + + if (!__is_ras_eeprom_supported(ras_core)) + return 0; + + if (!__get_eeprom_i2c_addr(ras_core, control)) + return -EINVAL; + + control->ras_header_offset = RAS_HDR_START; + control->ras_info_offset = RAS_TABLE_V2_1_INFO_START; + mutex_init(&control->ras_tbl_mutex); + + /* Read the table header from EEPROM address */ + res = __eeprom_read(ras_core, + control->i2c_address + control->ras_header_offset, + buf, RAS_TABLE_HEADER_SIZE); + if (res < RAS_TABLE_HEADER_SIZE) { + RAS_DEV_ERR(ras_core->dev, + "Failed to read EEPROM table header, res:%d\n", res); + return res >= 0 ? -EIO : res; + } + + __decode_table_header_from_buf(hdr, buf); + + if (hdr->header != RAS_TABLE_HDR_VAL && + hdr->header != RAS_TABLE_HDR_BAD) { + RAS_DEV_INFO(ras_core->dev, "Creating a new EEPROM table"); + return ras_eeprom_reset_table(ras_core); + } + + switch (hdr->version) { + case RAS_TABLE_VER_V2_1: + case RAS_TABLE_VER_V3: + control->ras_num_recs = RAS_NUM_RECS_V2_1(hdr); + control->ras_record_offset = RAS_RECORD_START_V2_1; + control->ras_max_record_count = RAS_MAX_RECORD_COUNT_V2_1; + break; + case RAS_TABLE_VER_V1: + control->ras_num_recs = RAS_NUM_RECS(hdr); + control->ras_record_offset = RAS_RECORD_START; + control->ras_max_record_count = RAS_MAX_RECORD_COUNT; + break; + default: + RAS_DEV_ERR(ras_core->dev, + "RAS header invalid, unsupported version: %u", + hdr->version); + return -EINVAL; + } + + if (control->ras_num_recs > control->ras_max_record_count) { + RAS_DEV_ERR(ras_core->dev, + "RAS header invalid, records in header: %u max allowed :%u", + control->ras_num_recs, control->ras_max_record_count); + return -EINVAL; + } + + control->ras_fri = RAS_OFFSET_TO_INDEX(control, hdr->first_rec_offset); + + return 0; +} + +int ras_eeprom_check_storage_status(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + struct ras_eeprom_table_header *hdr; + int bad_page_count; + int res = 0; + + if (!__is_ras_eeprom_supported(ras_core)) + return 0; + + if (!__get_eeprom_i2c_addr(ras_core, control)) + return -EINVAL; + + hdr = &control->tbl_hdr; + + bad_page_count = ras_umc_get_badpage_count(ras_core); + if (hdr->header == RAS_TABLE_HDR_VAL) { + RAS_DEV_INFO(ras_core->dev, + "Found existing EEPROM table with %d records\n", + bad_page_count); + + if (hdr->version >= RAS_TABLE_VER_V2_1) { + res = __read_table_ras_info(control); + if (res) + return res; + } + + res = __verify_ras_table_checksum(control); + if (res) + RAS_DEV_ERR(ras_core->dev, + "RAS table incorrect checksum or error:%d\n", res); + + /* Warn if we are at 90% of the threshold or above + */ + if (10 * bad_page_count >= 9 * control->record_threshold_count) + RAS_DEV_WARN(ras_core->dev, + "RAS records:%u exceeds 90%% of threshold:%d\n", + bad_page_count, + control->record_threshold_count); + + } else if (hdr->header == RAS_TABLE_HDR_BAD && + control->record_threshold_config != 0) { + if (hdr->version >= RAS_TABLE_VER_V2_1) { + res = __read_table_ras_info(control); + if (res) + return res; + } + + res = __verify_ras_table_checksum(control); + if (res) + RAS_DEV_ERR(ras_core->dev, + "RAS Table incorrect checksum or error:%d\n", res); + + if (control->record_threshold_count >= bad_page_count) { + /* This means that, the threshold was increased since + * the last time the system was booted, and now, + * ras->record_threshold_count - control->num_recs > 0, + * so that at least one more record can be saved, + * before the page count threshold is reached. + */ + RAS_DEV_INFO(ras_core->dev, + "records:%d threshold:%d, resetting RAS table header signature", + bad_page_count, + control->record_threshold_count); + res = ras_eeprom_correct_header_tag(control, RAS_TABLE_HDR_VAL); + } else { + RAS_DEV_ERR(ras_core->dev, "RAS records:%d exceed threshold:%d", + bad_page_count, control->record_threshold_count); + if ((control->record_threshold_config == WARN_NONSTOP_OVER_THRESHOLD) || + (control->record_threshold_config == NONSTOP_OVER_THRESHOLD)) { + RAS_DEV_WARN(ras_core->dev, + "Please consult AMD Service Action Guide (SAG) for appropriate service procedures\n"); + res = 0; + } else { + ras_core->is_rma = true; + RAS_DEV_ERR(ras_core->dev, + "User defined threshold is set, runtime service will be halt when threshold is reached\n"); + } + } + } + + return res < 0 ? res : 0; +} + +int ras_eeprom_hw_init(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control; + struct ras_eeprom_config *eeprom_cfg; + + if (!ras_core) + return -EINVAL; + + ras_core->is_rma = false; + + control = &ras_core->ras_eeprom; + + memset(control, 0, sizeof(*control)); + + eeprom_cfg = &ras_core->config->eeprom_cfg; + control->record_threshold_config = + eeprom_cfg->eeprom_record_threshold_config; + + control->record_threshold_count = ras_eeprom_max_record_count(ras_core); + if (eeprom_cfg->eeprom_record_threshold_count < + control->record_threshold_count) + control->record_threshold_count = + eeprom_cfg->eeprom_record_threshold_count; + + control->sys_func = eeprom_cfg->eeprom_sys_fn; + control->max_read_len = eeprom_cfg->max_i2c_read_len; + control->max_write_len = eeprom_cfg->max_i2c_write_len; + control->i2c_adapter = eeprom_cfg->eeprom_i2c_adapter; + control->i2c_port = eeprom_cfg->eeprom_i2c_port; + control->i2c_address = eeprom_cfg->eeprom_i2c_addr; + + control->update_channel_flag = false; + + return __check_ras_table_status(ras_core); +} + +int ras_eeprom_hw_fini(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control; + + if (!ras_core) + return -EINVAL; + + control = &ras_core->ras_eeprom; + mutex_destroy(&control->ras_tbl_mutex); + + return 0; +} + +uint32_t ras_eeprom_get_record_count(struct ras_core_context *ras_core) +{ + if (!ras_core) + return 0; + + return ras_core->ras_eeprom.ras_num_recs; +} + +void ras_eeprom_sync_info(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control; + + if (!ras_core) + return; + + control = &ras_core->ras_eeprom; + ras_core_event_notify(ras_core, RAS_EVENT_ID__UPDATE_BAD_PAGE_NUM, + &control->ras_num_recs); + ras_core_event_notify(ras_core, RAS_EVENT_ID__UPDATE_BAD_CHANNEL_BITMAP, + &control->bad_channel_bitmap); +} + +enum ras_gpu_health_status + ras_eeprom_check_gpu_status(struct ras_core_context *ras_core) +{ + struct ras_eeprom_control *control = &ras_core->ras_eeprom; + struct ras_eeprom_table_ras_info *rai = &control->tbl_rai; + + if (!__is_ras_eeprom_supported(ras_core) || + !control->record_threshold_config) + return RAS_GPU_HEALTH_NONE; + + if (control->tbl_hdr.header == RAS_TABLE_HDR_BAD) + return RAS_GPU_IN_BAD_STATUS; + + return rai->rma_status; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.h b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.h new file mode 100644 index 000000000000..2abe566c18b6 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_eeprom.h @@ -0,0 +1,197 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_EEPROM_H__ +#define __RAS_EEPROM_H__ +#include "ras_sys.h" + +#define RAS_TABLE_VER_V1 0x00010000 +#define RAS_TABLE_VER_V2_1 0x00021000 +#define RAS_TABLE_VER_V3 0x00030000 + +#define NONSTOP_OVER_THRESHOLD -2 +#define WARN_NONSTOP_OVER_THRESHOLD -1 +#define DISABLE_RETIRE_PAGE 0 + +/* + * Bad address pfn : eeprom_umc_record.retired_row_pfn[39:0], + * nps mode: eeprom_umc_record.retired_row_pfn[47:40] + */ +#define EEPROM_RECORD_UMC_ADDR_MASK 0xFFFFFFFFFFULL +#define EEPROM_RECORD_UMC_NPS_MASK 0xFF0000000000ULL +#define EEPROM_RECORD_UMC_NPS_SHIFT 40 + +#define EEPROM_RECORD_UMC_NPS_MODE(RECORD) \ + (((RECORD)->retired_row_pfn & EEPROM_RECORD_UMC_NPS_MASK) >> \ + EEPROM_RECORD_UMC_NPS_SHIFT) + +#define EEPROM_RECORD_UMC_ADDR_PFN(RECORD) \ + ((RECORD)->retired_row_pfn & EEPROM_RECORD_UMC_ADDR_MASK) + +#define EEPROM_RECORD_SETUP_UMC_ADDR_AND_NPS(RECORD, ADDR, NPS) \ +do { \ + uint64_t tmp = (NPS); \ + tmp = ((tmp << EEPROM_RECORD_UMC_NPS_SHIFT) & EEPROM_RECORD_UMC_NPS_MASK); \ + tmp |= (ADDR) & EEPROM_RECORD_UMC_ADDR_MASK; \ + (RECORD)->retired_row_pfn = tmp; \ +} while (0) + +enum ras_gpu_health_status { + RAS_GPU_HEALTH_NONE = 0, + RAS_GPU_HEALTH_USABLE = 1, + RAS_GPU_RETIRED__ECC_REACH_THRESHOLD = 2, + RAS_GPU_IN_BAD_STATUS = 3, +}; + +enum ras_eeprom_err_type { + RAS_EEPROM_ERR_NA, + RAS_EEPROM_ERR_RECOVERABLE, + RAS_EEPROM_ERR_NON_RECOVERABLE, + RAS_EEPROM_ERR_COUNT, +}; + +struct ras_eeprom_table_header { + uint32_t header; + uint32_t version; + uint32_t first_rec_offset; + uint32_t tbl_size; + uint32_t checksum; +} __packed; + +struct ras_eeprom_table_ras_info { + u8 rma_status; + u8 health_percent; + u16 ecc_page_threshold; + u32 padding[64 - 1]; +} __packed; + +struct ras_eeprom_control { + struct ras_eeprom_table_header tbl_hdr; + struct ras_eeprom_table_ras_info tbl_rai; + + /* record threshold */ + int record_threshold_config; + uint32_t record_threshold_count; + bool update_channel_flag; + + const struct ras_eeprom_sys_func *sys_func; + void *i2c_adapter; + u32 i2c_port; + u16 max_read_len; + u16 max_write_len; + + /* Base I2C EEPPROM 19-bit memory address, + * where the table is located. For more information, + * see top of amdgpu_eeprom.c. + */ + u32 i2c_address; + + /* The byte offset off of @i2c_address + * where the table header is found, + * and where the records start--always + * right after the header. + */ + u32 ras_header_offset; + u32 ras_info_offset; + u32 ras_record_offset; + + /* Number of records in the table. + */ + u32 ras_num_recs; + + /* First record index to read, 0-based. + * Range is [0, num_recs-1]. This is + * an absolute index, starting right after + * the table header. + */ + u32 ras_fri; + + /* Maximum possible number of records + * we could store, i.e. the maximum capacity + * of the table. + */ + u32 ras_max_record_count; + + /* Protect table access via this mutex. + */ + struct mutex ras_tbl_mutex; + + /* Record channel info which occurred bad pages + */ + u32 bad_channel_bitmap; +}; + +/* + * Represents single table record. Packed to be easily serialized into byte + * stream. + */ +struct eeprom_umc_record { + + union { + uint64_t address; + uint64_t offset; + }; + + uint64_t retired_row_pfn; + uint64_t ts; + + enum ras_eeprom_err_type err_type; + + union { + unsigned char bank; + unsigned char cu; + }; + + unsigned char mem_channel; + unsigned char mcumc_id; + + /* The following variables will not be saved to eeprom. + */ + uint64_t cur_nps_retired_row_pfn; + uint32_t cur_nps_bank; + uint32_t cur_nps; +}; + +struct ras_core_context; +int ras_eeprom_hw_init(struct ras_core_context *ras_core); +int ras_eeprom_hw_fini(struct ras_core_context *ras_core); + +int ras_eeprom_reset_table(struct ras_core_context *ras_core); + +bool ras_eeprom_check_safety_watermark(struct ras_core_context *ras_core); + +int ras_eeprom_read(struct ras_core_context *ras_core, + struct eeprom_umc_record *records, const u32 num); + +int ras_eeprom_append(struct ras_core_context *ras_core, + struct eeprom_umc_record *records, const u32 num); + +uint32_t ras_eeprom_max_record_count(struct ras_core_context *ras_core); +uint32_t ras_eeprom_get_record_count(struct ras_core_context *ras_core); +void ras_eeprom_sync_info(struct ras_core_context *ras_core); + +int ras_eeprom_check_storage_status(struct ras_core_context *ras_core); +enum ras_gpu_health_status + ras_eeprom_check_gpu_status(struct ras_core_context *ras_core); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_gfx.c b/drivers/gpu/drm/amd/ras/rascore/ras_gfx.c new file mode 100644 index 000000000000..f5ce28777705 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_gfx.c @@ -0,0 +1,70 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras.h" +#include "ras_gfx_v9_0.h" +#include "ras_gfx.h" +#include "ras_core_status.h" + +static const struct ras_gfx_ip_func *ras_gfx_get_ip_funcs( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(9, 4, 3): + case IP_VERSION(9, 4, 4): + case IP_VERSION(9, 5, 0): + return &gfx_ras_func_v9_0; + default: + RAS_DEV_ERR(ras_core->dev, + "GFX ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +int ras_gfx_get_ta_subblock(struct ras_core_context *ras_core, + uint32_t error_type, uint32_t subblock, uint32_t *ta_subblock) +{ + struct ras_gfx *gfx = &ras_core->ras_gfx; + + return gfx->ip_func->get_ta_subblock(ras_core, + error_type, subblock, ta_subblock); +} + +int ras_gfx_hw_init(struct ras_core_context *ras_core) +{ + struct ras_gfx *gfx = &ras_core->ras_gfx; + + gfx->gfx_ip_version = ras_core->config->gfx_ip_version; + + gfx->ip_func = ras_gfx_get_ip_funcs(ras_core, gfx->gfx_ip_version); + + return gfx->ip_func ? RAS_CORE_OK : -EINVAL; +} + +int ras_gfx_hw_fini(struct ras_core_context *ras_core) +{ + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_gfx.h b/drivers/gpu/drm/amd/ras/rascore/ras_gfx.h new file mode 100644 index 000000000000..8a42d69fb0ad --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_gfx.h @@ -0,0 +1,43 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_GFX_H__ +#define __RAS_GFX_H__ + +struct ras_gfx_ip_func { + int (*get_ta_subblock)(struct ras_core_context *ras_core, + uint32_t error_type, uint32_t subblock, uint32_t *ta_subblock); +}; + +struct ras_gfx { + uint32_t gfx_ip_version; + const struct ras_gfx_ip_func *ip_func; +}; + +int ras_gfx_hw_init(struct ras_core_context *ras_core); +int ras_gfx_hw_fini(struct ras_core_context *ras_core); + +int ras_gfx_get_ta_subblock(struct ras_core_context *ras_core, + uint32_t error_type, uint32_t subblock, uint32_t *ta_subblock); + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.c b/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.c new file mode 100644 index 000000000000..6213d3f125be --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.c @@ -0,0 +1,426 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_gfx_v9_0.h" +#include "ras_core_status.h" + +enum ta_gfx_v9_subblock { + /*CPC*/ + TA_GFX_V9__GFX_CPC_INDEX_START = 0, + TA_GFX_V9__GFX_CPC_SCRATCH = TA_GFX_V9__GFX_CPC_INDEX_START, + TA_GFX_V9__GFX_CPC_UCODE, + TA_GFX_V9__GFX_DC_STATE_ME1, + TA_GFX_V9__GFX_DC_CSINVOC_ME1, + TA_GFX_V9__GFX_DC_RESTORE_ME1, + TA_GFX_V9__GFX_DC_STATE_ME2, + TA_GFX_V9__GFX_DC_CSINVOC_ME2, + TA_GFX_V9__GFX_DC_RESTORE_ME2, + TA_GFX_V9__GFX_CPC_INDEX_END = TA_GFX_V9__GFX_DC_RESTORE_ME2, + /* CPF*/ + TA_GFX_V9__GFX_CPF_INDEX_START, + TA_GFX_V9__GFX_CPF_ROQ_ME2 = TA_GFX_V9__GFX_CPF_INDEX_START, + TA_GFX_V9__GFX_CPF_ROQ_ME1, + TA_GFX_V9__GFX_CPF_TAG, + TA_GFX_V9__GFX_CPF_INDEX_END = TA_GFX_V9__GFX_CPF_TAG, + /* CPG*/ + TA_GFX_V9__GFX_CPG_INDEX_START, + TA_GFX_V9__GFX_CPG_DMA_ROQ = TA_GFX_V9__GFX_CPG_INDEX_START, + TA_GFX_V9__GFX_CPG_DMA_TAG, + TA_GFX_V9__GFX_CPG_TAG, + TA_GFX_V9__GFX_CPG_INDEX_END = TA_GFX_V9__GFX_CPG_TAG, + /* GDS*/ + TA_GFX_V9__GFX_GDS_INDEX_START, + TA_GFX_V9__GFX_GDS_MEM = TA_GFX_V9__GFX_GDS_INDEX_START, + TA_GFX_V9__GFX_GDS_INPUT_QUEUE, + TA_GFX_V9__GFX_GDS_OA_PHY_CMD_RAM_MEM, + TA_GFX_V9__GFX_GDS_OA_PHY_DATA_RAM_MEM, + TA_GFX_V9__GFX_GDS_OA_PIPE_MEM, + TA_GFX_V9__GFX_GDS_INDEX_END = TA_GFX_V9__GFX_GDS_OA_PIPE_MEM, + /* SPI*/ + TA_GFX_V9__GFX_SPI_SR_MEM, + /* SQ*/ + TA_GFX_V9__GFX_SQ_INDEX_START, + TA_GFX_V9__GFX_SQ_SGPR = TA_GFX_V9__GFX_SQ_INDEX_START, + TA_GFX_V9__GFX_SQ_LDS_D, + TA_GFX_V9__GFX_SQ_LDS_I, + TA_GFX_V9__GFX_SQ_VGPR, /* VGPR = SP*/ + TA_GFX_V9__GFX_SQ_INDEX_END = TA_GFX_V9__GFX_SQ_VGPR, + /* SQC (3 ranges)*/ + TA_GFX_V9__GFX_SQC_INDEX_START, + /* SQC range 0*/ + TA_GFX_V9__GFX_SQC_INDEX0_START = TA_GFX_V9__GFX_SQC_INDEX_START, + TA_GFX_V9__GFX_SQC_INST_UTCL1_LFIFO = + TA_GFX_V9__GFX_SQC_INDEX0_START, + TA_GFX_V9__GFX_SQC_DATA_CU0_WRITE_DATA_BUF, + TA_GFX_V9__GFX_SQC_DATA_CU0_UTCL1_LFIFO, + TA_GFX_V9__GFX_SQC_DATA_CU1_WRITE_DATA_BUF, + TA_GFX_V9__GFX_SQC_DATA_CU1_UTCL1_LFIFO, + TA_GFX_V9__GFX_SQC_DATA_CU2_WRITE_DATA_BUF, + TA_GFX_V9__GFX_SQC_DATA_CU2_UTCL1_LFIFO, + TA_GFX_V9__GFX_SQC_INDEX0_END = + TA_GFX_V9__GFX_SQC_DATA_CU2_UTCL1_LFIFO, + /* SQC range 1*/ + TA_GFX_V9__GFX_SQC_INDEX1_START, + TA_GFX_V9__GFX_SQC_INST_BANKA_TAG_RAM = + TA_GFX_V9__GFX_SQC_INDEX1_START, + TA_GFX_V9__GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO, + TA_GFX_V9__GFX_SQC_INST_BANKA_MISS_FIFO, + TA_GFX_V9__GFX_SQC_INST_BANKA_BANK_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKA_TAG_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKA_HIT_FIFO, + TA_GFX_V9__GFX_SQC_DATA_BANKA_MISS_FIFO, + TA_GFX_V9__GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKA_BANK_RAM, + TA_GFX_V9__GFX_SQC_INDEX1_END = + TA_GFX_V9__GFX_SQC_DATA_BANKA_BANK_RAM, + /* SQC range 2*/ + TA_GFX_V9__GFX_SQC_INDEX2_START, + TA_GFX_V9__GFX_SQC_INST_BANKB_TAG_RAM = + TA_GFX_V9__GFX_SQC_INDEX2_START, + TA_GFX_V9__GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO, + TA_GFX_V9__GFX_SQC_INST_BANKB_MISS_FIFO, + TA_GFX_V9__GFX_SQC_INST_BANKB_BANK_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKB_TAG_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKB_HIT_FIFO, + TA_GFX_V9__GFX_SQC_DATA_BANKB_MISS_FIFO, + TA_GFX_V9__GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM, + TA_GFX_V9__GFX_SQC_DATA_BANKB_BANK_RAM, + TA_GFX_V9__GFX_SQC_INDEX2_END = + TA_GFX_V9__GFX_SQC_DATA_BANKB_BANK_RAM, + TA_GFX_V9__GFX_SQC_INDEX_END = TA_GFX_V9__GFX_SQC_INDEX2_END, + /* TA*/ + TA_GFX_V9__GFX_TA_INDEX_START, + TA_GFX_V9__GFX_TA_FS_DFIFO = TA_GFX_V9__GFX_TA_INDEX_START, + TA_GFX_V9__GFX_TA_FS_AFIFO, + TA_GFX_V9__GFX_TA_FL_LFIFO, + TA_GFX_V9__GFX_TA_FX_LFIFO, + TA_GFX_V9__GFX_TA_FS_CFIFO, + TA_GFX_V9__GFX_TA_INDEX_END = TA_GFX_V9__GFX_TA_FS_CFIFO, + /* TCA*/ + TA_GFX_V9__GFX_TCA_INDEX_START, + TA_GFX_V9__GFX_TCA_HOLE_FIFO = TA_GFX_V9__GFX_TCA_INDEX_START, + TA_GFX_V9__GFX_TCA_REQ_FIFO, + TA_GFX_V9__GFX_TCA_INDEX_END = TA_GFX_V9__GFX_TCA_REQ_FIFO, + /* TCC (5 sub-ranges)*/ + TA_GFX_V9__GFX_TCC_INDEX_START, + /* TCC range 0*/ + TA_GFX_V9__GFX_TCC_INDEX0_START = TA_GFX_V9__GFX_TCC_INDEX_START, + TA_GFX_V9__GFX_TCC_CACHE_DATA = TA_GFX_V9__GFX_TCC_INDEX0_START, + TA_GFX_V9__GFX_TCC_CACHE_DATA_BANK_0_1, + TA_GFX_V9__GFX_TCC_CACHE_DATA_BANK_1_0, + TA_GFX_V9__GFX_TCC_CACHE_DATA_BANK_1_1, + TA_GFX_V9__GFX_TCC_CACHE_DIRTY_BANK_0, + TA_GFX_V9__GFX_TCC_CACHE_DIRTY_BANK_1, + TA_GFX_V9__GFX_TCC_HIGH_RATE_TAG, + TA_GFX_V9__GFX_TCC_LOW_RATE_TAG, + TA_GFX_V9__GFX_TCC_INDEX0_END = TA_GFX_V9__GFX_TCC_LOW_RATE_TAG, + /* TCC range 1*/ + TA_GFX_V9__GFX_TCC_INDEX1_START, + TA_GFX_V9__GFX_TCC_IN_USE_DEC = TA_GFX_V9__GFX_TCC_INDEX1_START, + TA_GFX_V9__GFX_TCC_IN_USE_TRANSFER, + TA_GFX_V9__GFX_TCC_INDEX1_END = + TA_GFX_V9__GFX_TCC_IN_USE_TRANSFER, + /* TCC range 2*/ + TA_GFX_V9__GFX_TCC_INDEX2_START, + TA_GFX_V9__GFX_TCC_RETURN_DATA = TA_GFX_V9__GFX_TCC_INDEX2_START, + TA_GFX_V9__GFX_TCC_RETURN_CONTROL, + TA_GFX_V9__GFX_TCC_UC_ATOMIC_FIFO, + TA_GFX_V9__GFX_TCC_WRITE_RETURN, + TA_GFX_V9__GFX_TCC_WRITE_CACHE_READ, + TA_GFX_V9__GFX_TCC_SRC_FIFO, + TA_GFX_V9__GFX_TCC_SRC_FIFO_NEXT_RAM, + TA_GFX_V9__GFX_TCC_CACHE_TAG_PROBE_FIFO, + TA_GFX_V9__GFX_TCC_INDEX2_END = + TA_GFX_V9__GFX_TCC_CACHE_TAG_PROBE_FIFO, + /* TCC range 3*/ + TA_GFX_V9__GFX_TCC_INDEX3_START, + TA_GFX_V9__GFX_TCC_LATENCY_FIFO = TA_GFX_V9__GFX_TCC_INDEX3_START, + TA_GFX_V9__GFX_TCC_LATENCY_FIFO_NEXT_RAM, + TA_GFX_V9__GFX_TCC_INDEX3_END = + TA_GFX_V9__GFX_TCC_LATENCY_FIFO_NEXT_RAM, + /* TCC range 4*/ + TA_GFX_V9__GFX_TCC_INDEX4_START, + TA_GFX_V9__GFX_TCC_WRRET_TAG_WRITE_RETURN = + TA_GFX_V9__GFX_TCC_INDEX4_START, + TA_GFX_V9__GFX_TCC_ATOMIC_RETURN_BUFFER, + TA_GFX_V9__GFX_TCC_INDEX4_END = + TA_GFX_V9__GFX_TCC_ATOMIC_RETURN_BUFFER, + TA_GFX_V9__GFX_TCC_INDEX_END = TA_GFX_V9__GFX_TCC_INDEX4_END, + /* TCI*/ + TA_GFX_V9__GFX_TCI_WRITE_RAM, + /* TCP*/ + TA_GFX_V9__GFX_TCP_INDEX_START, + TA_GFX_V9__GFX_TCP_CACHE_RAM = TA_GFX_V9__GFX_TCP_INDEX_START, + TA_GFX_V9__GFX_TCP_LFIFO_RAM, + TA_GFX_V9__GFX_TCP_CMD_FIFO, + TA_GFX_V9__GFX_TCP_VM_FIFO, + TA_GFX_V9__GFX_TCP_DB_RAM, + TA_GFX_V9__GFX_TCP_UTCL1_LFIFO0, + TA_GFX_V9__GFX_TCP_UTCL1_LFIFO1, + TA_GFX_V9__GFX_TCP_INDEX_END = TA_GFX_V9__GFX_TCP_UTCL1_LFIFO1, + /* TD*/ + TA_GFX_V9__GFX_TD_INDEX_START, + TA_GFX_V9__GFX_TD_SS_FIFO_LO = TA_GFX_V9__GFX_TD_INDEX_START, + TA_GFX_V9__GFX_TD_SS_FIFO_HI, + TA_GFX_V9__GFX_TD_CS_FIFO, + TA_GFX_V9__GFX_TD_INDEX_END = TA_GFX_V9__GFX_TD_CS_FIFO, + /* EA (3 sub-ranges)*/ + TA_GFX_V9__GFX_EA_INDEX_START, + /* EA range 0*/ + TA_GFX_V9__GFX_EA_INDEX0_START = TA_GFX_V9__GFX_EA_INDEX_START, + TA_GFX_V9__GFX_EA_DRAMRD_CMDMEM = TA_GFX_V9__GFX_EA_INDEX0_START, + TA_GFX_V9__GFX_EA_DRAMWR_CMDMEM, + TA_GFX_V9__GFX_EA_DRAMWR_DATAMEM, + TA_GFX_V9__GFX_EA_RRET_TAGMEM, + TA_GFX_V9__GFX_EA_WRET_TAGMEM, + TA_GFX_V9__GFX_EA_GMIRD_CMDMEM, + TA_GFX_V9__GFX_EA_GMIWR_CMDMEM, + TA_GFX_V9__GFX_EA_GMIWR_DATAMEM, + TA_GFX_V9__GFX_EA_INDEX0_END = TA_GFX_V9__GFX_EA_GMIWR_DATAMEM, + /* EA range 1*/ + TA_GFX_V9__GFX_EA_INDEX1_START, + TA_GFX_V9__GFX_EA_DRAMRD_PAGEMEM = TA_GFX_V9__GFX_EA_INDEX1_START, + TA_GFX_V9__GFX_EA_DRAMWR_PAGEMEM, + TA_GFX_V9__GFX_EA_IORD_CMDMEM, + TA_GFX_V9__GFX_EA_IOWR_CMDMEM, + TA_GFX_V9__GFX_EA_IOWR_DATAMEM, + TA_GFX_V9__GFX_EA_GMIRD_PAGEMEM, + TA_GFX_V9__GFX_EA_GMIWR_PAGEMEM, + TA_GFX_V9__GFX_EA_INDEX1_END = TA_GFX_V9__GFX_EA_GMIWR_PAGEMEM, + /* EA range 2*/ + TA_GFX_V9__GFX_EA_INDEX2_START, + TA_GFX_V9__GFX_EA_MAM_D0MEM = TA_GFX_V9__GFX_EA_INDEX2_START, + TA_GFX_V9__GFX_EA_MAM_D1MEM, + TA_GFX_V9__GFX_EA_MAM_D2MEM, + TA_GFX_V9__GFX_EA_MAM_D3MEM, + TA_GFX_V9__GFX_EA_INDEX2_END = TA_GFX_V9__GFX_EA_MAM_D3MEM, + TA_GFX_V9__GFX_EA_INDEX_END = TA_GFX_V9__GFX_EA_INDEX2_END, + /* UTC VM L2 bank*/ + TA_GFX_V9__UTC_VML2_BANK_CACHE, + /* UTC VM walker*/ + TA_GFX_V9__UTC_VML2_WALKER, + /* UTC ATC L2 2MB cache*/ + TA_GFX_V9__UTC_ATCL2_CACHE_2M_BANK, + /* UTC ATC L2 4KB cache*/ + TA_GFX_V9__UTC_ATCL2_CACHE_4K_BANK, + TA_GFX_V9__GFX_MAX +}; + +struct ras_gfx_subblock_t { + unsigned char *name; + int ta_subblock; + int hw_supported_error_type; + int sw_supported_error_type; +}; + +#define RAS_GFX_SUB_BLOCK(subblock, a, b, c, d, e, f, g, h) \ + [RAS_GFX_V9__##subblock] = { \ + #subblock, \ + TA_GFX_V9__##subblock, \ + ((a) | ((b) << 1) | ((c) << 2) | ((d) << 3)), \ + (((e) << 1) | ((f) << 3) | (g) | ((h) << 2)), \ + } + +const struct ras_gfx_subblock_t ras_gfx_v9_0_subblocks[] = { + RAS_GFX_SUB_BLOCK(GFX_CPC_SCRATCH, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_CPC_UCODE, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_DC_STATE_ME1, 1, 0, 0, 1, 0, 0, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_DC_CSINVOC_ME1, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_DC_RESTORE_ME1, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_DC_STATE_ME2, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_DC_CSINVOC_ME2, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_DC_RESTORE_ME2, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_CPF_ROQ_ME2, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_CPF_ROQ_ME1, 1, 0, 0, 1, 0, 0, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_CPF_TAG, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_CPG_DMA_ROQ, 1, 0, 0, 1, 0, 0, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_CPG_DMA_TAG, 0, 1, 1, 1, 0, 1, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_CPG_TAG, 0, 1, 1, 1, 1, 1, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_GDS_MEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_GDS_INPUT_QUEUE, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_GDS_OA_PHY_CMD_RAM_MEM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_GDS_OA_PHY_DATA_RAM_MEM, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_GDS_OA_PIPE_MEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SPI_SR_MEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQ_SGPR, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQ_LDS_D, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_SQ_LDS_I, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQ_VGPR, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU0_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU0_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU1_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU1_UTCL1_LFIFO, 0, 1, 1, 1, 1, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU2_WRITE_DATA_BUF, 0, 1, 1, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_CU2_UTCL1_LFIFO, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKA_TAG_RAM, 0, 1, 1, 1, 1, 0, 0, + 1), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO, 1, 0, 0, 1, 0, + 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKA_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKA_BANK_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKA_TAG_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKA_HIT_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKA_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM, 1, 0, 0, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKA_BANK_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKB_TAG_RAM, 0, 1, 1, 1, 1, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO, 1, 0, 0, 1, 0, + 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKB_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_INST_BANKB_BANK_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKB_TAG_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKB_HIT_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKB_MISS_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM, 1, 0, 0, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_SQC_DATA_BANKB_BANK_RAM, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TA_FS_DFIFO, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_TA_FS_AFIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TA_FL_LFIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TA_FX_LFIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TA_FS_CFIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCA_HOLE_FIFO, 1, 0, 0, 1, 0, 1, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_TCA_REQ_FIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DATA, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_0_1, 0, 1, 1, 1, 1, 0, 0, + 1), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_1_0, 0, 1, 1, 1, 1, 0, 0, + 1), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DATA_BANK_1_1, 0, 1, 1, 1, 1, 0, 0, + 1), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DIRTY_BANK_0, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_DIRTY_BANK_1, 0, 1, 1, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_HIGH_RATE_TAG, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_LOW_RATE_TAG, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_IN_USE_DEC, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_IN_USE_TRANSFER, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_RETURN_DATA, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_RETURN_CONTROL, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_UC_ATOMIC_FIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_WRITE_RETURN, 1, 0, 0, 1, 0, 1, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_WRITE_CACHE_READ, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_SRC_FIFO, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_SRC_FIFO_NEXT_RAM, 1, 0, 0, 1, 0, 0, 1, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_CACHE_TAG_PROBE_FIFO, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_LATENCY_FIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_LATENCY_FIFO_NEXT_RAM, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_WRRET_TAG_WRITE_RETURN, 1, 0, 0, 1, 0, 0, + 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCC_ATOMIC_RETURN_BUFFER, 1, 0, 0, 1, 0, 0, 0, + 0), + RAS_GFX_SUB_BLOCK(GFX_TCI_WRITE_RAM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_CACHE_RAM, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_TCP_LFIFO_RAM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_CMD_FIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_VM_FIFO, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_DB_RAM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_UTCL1_LFIFO0, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TCP_UTCL1_LFIFO1, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TD_SS_FIFO_LO, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_TD_SS_FIFO_HI, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_TD_CS_FIFO, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_DRAMRD_CMDMEM, 0, 1, 1, 1, 1, 0, 0, 1), + RAS_GFX_SUB_BLOCK(GFX_EA_DRAMWR_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_DRAMWR_DATAMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_RRET_TAGMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_WRET_TAGMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_GMIRD_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_GMIWR_CMDMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_GMIWR_DATAMEM, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_DRAMRD_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_DRAMWR_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_IORD_CMDMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_IOWR_CMDMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_IOWR_DATAMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_GMIRD_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_GMIWR_PAGEMEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_MAM_D0MEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_MAM_D1MEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_MAM_D2MEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(GFX_EA_MAM_D3MEM, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(UTC_VML2_BANK_CACHE, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(UTC_VML2_WALKER, 0, 1, 1, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(UTC_ATCL2_CACHE_2M_BANK, 1, 0, 0, 1, 0, 0, 0, 0), + RAS_GFX_SUB_BLOCK(UTC_ATCL2_CACHE_4K_BANK, 0, 1, 1, 1, 0, 0, 0, 0), +}; + +static int gfx_v9_0_get_ta_subblock(struct ras_core_context *ras_core, + uint32_t error_type, uint32_t subblock, uint32_t *ta_subblock) +{ + const struct ras_gfx_subblock_t *gfx_subblock; + + if (subblock >= ARRAY_SIZE(ras_gfx_v9_0_subblocks)) + return -EINVAL; + + gfx_subblock = &ras_gfx_v9_0_subblocks[subblock]; + if (!gfx_subblock->name) + return -EPERM; + + if (!(gfx_subblock->hw_supported_error_type & error_type)) { + RAS_DEV_ERR(ras_core->dev, "GFX Subblock %s, hardware do not support type 0x%x\n", + gfx_subblock->name, error_type); + return -EPERM; + } + + if (!(gfx_subblock->sw_supported_error_type & error_type)) { + RAS_DEV_ERR(ras_core->dev, "GFX Subblock %s, driver do not support type 0x%x\n", + gfx_subblock->name, error_type); + return -EPERM; + } + + *ta_subblock = gfx_subblock->ta_subblock; + + return 0; +} + +const struct ras_gfx_ip_func gfx_ras_func_v9_0 = { + .get_ta_subblock = gfx_v9_0_get_ta_subblock, +}; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.h b/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.h new file mode 100644 index 000000000000..659b56619747 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_gfx_v9_0.h @@ -0,0 +1,259 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_GFX_V9_0_H__ +#define __RAS_GFX_V9_0_H__ + +enum ras_gfx_v9_subblock { + /* CPC */ + RAS_GFX_V9__GFX_CPC_INDEX_START = 0, + RAS_GFX_V9__GFX_CPC_SCRATCH = + RAS_GFX_V9__GFX_CPC_INDEX_START, + RAS_GFX_V9__GFX_CPC_UCODE, + RAS_GFX_V9__GFX_DC_STATE_ME1, + RAS_GFX_V9__GFX_DC_CSINVOC_ME1, + RAS_GFX_V9__GFX_DC_RESTORE_ME1, + RAS_GFX_V9__GFX_DC_STATE_ME2, + RAS_GFX_V9__GFX_DC_CSINVOC_ME2, + RAS_GFX_V9__GFX_DC_RESTORE_ME2, + RAS_GFX_V9__GFX_CPC_INDEX_END = + RAS_GFX_V9__GFX_DC_RESTORE_ME2, + /* CPF */ + RAS_GFX_V9__GFX_CPF_INDEX_START, + RAS_GFX_V9__GFX_CPF_ROQ_ME2 = + RAS_GFX_V9__GFX_CPF_INDEX_START, + RAS_GFX_V9__GFX_CPF_ROQ_ME1, + RAS_GFX_V9__GFX_CPF_TAG, + RAS_GFX_V9__GFX_CPF_INDEX_END = RAS_GFX_V9__GFX_CPF_TAG, + /* CPG */ + RAS_GFX_V9__GFX_CPG_INDEX_START, + RAS_GFX_V9__GFX_CPG_DMA_ROQ = + RAS_GFX_V9__GFX_CPG_INDEX_START, + RAS_GFX_V9__GFX_CPG_DMA_TAG, + RAS_GFX_V9__GFX_CPG_TAG, + RAS_GFX_V9__GFX_CPG_INDEX_END = RAS_GFX_V9__GFX_CPG_TAG, + /* GDS */ + RAS_GFX_V9__GFX_GDS_INDEX_START, + RAS_GFX_V9__GFX_GDS_MEM = RAS_GFX_V9__GFX_GDS_INDEX_START, + RAS_GFX_V9__GFX_GDS_INPUT_QUEUE, + RAS_GFX_V9__GFX_GDS_OA_PHY_CMD_RAM_MEM, + RAS_GFX_V9__GFX_GDS_OA_PHY_DATA_RAM_MEM, + RAS_GFX_V9__GFX_GDS_OA_PIPE_MEM, + RAS_GFX_V9__GFX_GDS_INDEX_END = + RAS_GFX_V9__GFX_GDS_OA_PIPE_MEM, + /* SPI */ + RAS_GFX_V9__GFX_SPI_SR_MEM, + /* SQ */ + RAS_GFX_V9__GFX_SQ_INDEX_START, + RAS_GFX_V9__GFX_SQ_SGPR = RAS_GFX_V9__GFX_SQ_INDEX_START, + RAS_GFX_V9__GFX_SQ_LDS_D, + RAS_GFX_V9__GFX_SQ_LDS_I, + RAS_GFX_V9__GFX_SQ_VGPR, + RAS_GFX_V9__GFX_SQ_INDEX_END = RAS_GFX_V9__GFX_SQ_VGPR, + /* SQC (3 ranges) */ + RAS_GFX_V9__GFX_SQC_INDEX_START, + /* SQC range 0 */ + RAS_GFX_V9__GFX_SQC_INDEX0_START = + RAS_GFX_V9__GFX_SQC_INDEX_START, + RAS_GFX_V9__GFX_SQC_INST_UTCL1_LFIFO = + RAS_GFX_V9__GFX_SQC_INDEX0_START, + RAS_GFX_V9__GFX_SQC_DATA_CU0_WRITE_DATA_BUF, + RAS_GFX_V9__GFX_SQC_DATA_CU0_UTCL1_LFIFO, + RAS_GFX_V9__GFX_SQC_DATA_CU1_WRITE_DATA_BUF, + RAS_GFX_V9__GFX_SQC_DATA_CU1_UTCL1_LFIFO, + RAS_GFX_V9__GFX_SQC_DATA_CU2_WRITE_DATA_BUF, + RAS_GFX_V9__GFX_SQC_DATA_CU2_UTCL1_LFIFO, + RAS_GFX_V9__GFX_SQC_INDEX0_END = + RAS_GFX_V9__GFX_SQC_DATA_CU2_UTCL1_LFIFO, + /* SQC range 1 */ + RAS_GFX_V9__GFX_SQC_INDEX1_START, + RAS_GFX_V9__GFX_SQC_INST_BANKA_TAG_RAM = + RAS_GFX_V9__GFX_SQC_INDEX1_START, + RAS_GFX_V9__GFX_SQC_INST_BANKA_UTCL1_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_INST_BANKA_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_INST_BANKA_BANK_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKA_TAG_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKA_HIT_FIFO, + RAS_GFX_V9__GFX_SQC_DATA_BANKA_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_DATA_BANKA_DIRTY_BIT_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKA_BANK_RAM, + RAS_GFX_V9__GFX_SQC_INDEX1_END = + RAS_GFX_V9__GFX_SQC_DATA_BANKA_BANK_RAM, + /* SQC range 2 */ + RAS_GFX_V9__GFX_SQC_INDEX2_START, + RAS_GFX_V9__GFX_SQC_INST_BANKB_TAG_RAM = + RAS_GFX_V9__GFX_SQC_INDEX2_START, + RAS_GFX_V9__GFX_SQC_INST_BANKB_UTCL1_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_INST_BANKB_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_INST_BANKB_BANK_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKB_TAG_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKB_HIT_FIFO, + RAS_GFX_V9__GFX_SQC_DATA_BANKB_MISS_FIFO, + RAS_GFX_V9__GFX_SQC_DATA_BANKB_DIRTY_BIT_RAM, + RAS_GFX_V9__GFX_SQC_DATA_BANKB_BANK_RAM, + RAS_GFX_V9__GFX_SQC_INDEX2_END = + RAS_GFX_V9__GFX_SQC_DATA_BANKB_BANK_RAM, + RAS_GFX_V9__GFX_SQC_INDEX_END = + RAS_GFX_V9__GFX_SQC_INDEX2_END, + /* TA */ + RAS_GFX_V9__GFX_TA_INDEX_START, + RAS_GFX_V9__GFX_TA_FS_DFIFO = + RAS_GFX_V9__GFX_TA_INDEX_START, + RAS_GFX_V9__GFX_TA_FS_AFIFO, + RAS_GFX_V9__GFX_TA_FL_LFIFO, + RAS_GFX_V9__GFX_TA_FX_LFIFO, + RAS_GFX_V9__GFX_TA_FS_CFIFO, + RAS_GFX_V9__GFX_TA_INDEX_END = RAS_GFX_V9__GFX_TA_FS_CFIFO, + /* TCA */ + RAS_GFX_V9__GFX_TCA_INDEX_START, + RAS_GFX_V9__GFX_TCA_HOLE_FIFO = + RAS_GFX_V9__GFX_TCA_INDEX_START, + RAS_GFX_V9__GFX_TCA_REQ_FIFO, + RAS_GFX_V9__GFX_TCA_INDEX_END = + RAS_GFX_V9__GFX_TCA_REQ_FIFO, + /* TCC (5 sub-ranges) */ + RAS_GFX_V9__GFX_TCC_INDEX_START, + /* TCC range 0 */ + RAS_GFX_V9__GFX_TCC_INDEX0_START = + RAS_GFX_V9__GFX_TCC_INDEX_START, + RAS_GFX_V9__GFX_TCC_CACHE_DATA = + RAS_GFX_V9__GFX_TCC_INDEX0_START, + RAS_GFX_V9__GFX_TCC_CACHE_DATA_BANK_0_1, + RAS_GFX_V9__GFX_TCC_CACHE_DATA_BANK_1_0, + RAS_GFX_V9__GFX_TCC_CACHE_DATA_BANK_1_1, + RAS_GFX_V9__GFX_TCC_CACHE_DIRTY_BANK_0, + RAS_GFX_V9__GFX_TCC_CACHE_DIRTY_BANK_1, + RAS_GFX_V9__GFX_TCC_HIGH_RATE_TAG, + RAS_GFX_V9__GFX_TCC_LOW_RATE_TAG, + RAS_GFX_V9__GFX_TCC_INDEX0_END = + RAS_GFX_V9__GFX_TCC_LOW_RATE_TAG, + /* TCC range 1 */ + RAS_GFX_V9__GFX_TCC_INDEX1_START, + RAS_GFX_V9__GFX_TCC_IN_USE_DEC = + RAS_GFX_V9__GFX_TCC_INDEX1_START, + RAS_GFX_V9__GFX_TCC_IN_USE_TRANSFER, + RAS_GFX_V9__GFX_TCC_INDEX1_END = + RAS_GFX_V9__GFX_TCC_IN_USE_TRANSFER, + /* TCC range 2 */ + RAS_GFX_V9__GFX_TCC_INDEX2_START, + RAS_GFX_V9__GFX_TCC_RETURN_DATA = + RAS_GFX_V9__GFX_TCC_INDEX2_START, + RAS_GFX_V9__GFX_TCC_RETURN_CONTROL, + RAS_GFX_V9__GFX_TCC_UC_ATOMIC_FIFO, + RAS_GFX_V9__GFX_TCC_WRITE_RETURN, + RAS_GFX_V9__GFX_TCC_WRITE_CACHE_READ, + RAS_GFX_V9__GFX_TCC_SRC_FIFO, + RAS_GFX_V9__GFX_TCC_SRC_FIFO_NEXT_RAM, + RAS_GFX_V9__GFX_TCC_CACHE_TAG_PROBE_FIFO, + RAS_GFX_V9__GFX_TCC_INDEX2_END = + RAS_GFX_V9__GFX_TCC_CACHE_TAG_PROBE_FIFO, + /* TCC range 3 */ + RAS_GFX_V9__GFX_TCC_INDEX3_START, + RAS_GFX_V9__GFX_TCC_LATENCY_FIFO = + RAS_GFX_V9__GFX_TCC_INDEX3_START, + RAS_GFX_V9__GFX_TCC_LATENCY_FIFO_NEXT_RAM, + RAS_GFX_V9__GFX_TCC_INDEX3_END = + RAS_GFX_V9__GFX_TCC_LATENCY_FIFO_NEXT_RAM, + /* TCC range 4 */ + RAS_GFX_V9__GFX_TCC_INDEX4_START, + RAS_GFX_V9__GFX_TCC_WRRET_TAG_WRITE_RETURN = + RAS_GFX_V9__GFX_TCC_INDEX4_START, + RAS_GFX_V9__GFX_TCC_ATOMIC_RETURN_BUFFER, + RAS_GFX_V9__GFX_TCC_INDEX4_END = + RAS_GFX_V9__GFX_TCC_ATOMIC_RETURN_BUFFER, + RAS_GFX_V9__GFX_TCC_INDEX_END = + RAS_GFX_V9__GFX_TCC_INDEX4_END, + /* TCI */ + RAS_GFX_V9__GFX_TCI_WRITE_RAM, + /* TCP */ + RAS_GFX_V9__GFX_TCP_INDEX_START, + RAS_GFX_V9__GFX_TCP_CACHE_RAM = + RAS_GFX_V9__GFX_TCP_INDEX_START, + RAS_GFX_V9__GFX_TCP_LFIFO_RAM, + RAS_GFX_V9__GFX_TCP_CMD_FIFO, + RAS_GFX_V9__GFX_TCP_VM_FIFO, + RAS_GFX_V9__GFX_TCP_DB_RAM, + RAS_GFX_V9__GFX_TCP_UTCL1_LFIFO0, + RAS_GFX_V9__GFX_TCP_UTCL1_LFIFO1, + RAS_GFX_V9__GFX_TCP_INDEX_END = + RAS_GFX_V9__GFX_TCP_UTCL1_LFIFO1, + /* TD */ + RAS_GFX_V9__GFX_TD_INDEX_START, + RAS_GFX_V9__GFX_TD_SS_FIFO_LO = + RAS_GFX_V9__GFX_TD_INDEX_START, + RAS_GFX_V9__GFX_TD_SS_FIFO_HI, + RAS_GFX_V9__GFX_TD_CS_FIFO, + RAS_GFX_V9__GFX_TD_INDEX_END = RAS_GFX_V9__GFX_TD_CS_FIFO, + /* EA (3 sub-ranges) */ + RAS_GFX_V9__GFX_EA_INDEX_START, + /* EA range 0 */ + RAS_GFX_V9__GFX_EA_INDEX0_START = + RAS_GFX_V9__GFX_EA_INDEX_START, + RAS_GFX_V9__GFX_EA_DRAMRD_CMDMEM = + RAS_GFX_V9__GFX_EA_INDEX0_START, + RAS_GFX_V9__GFX_EA_DRAMWR_CMDMEM, + RAS_GFX_V9__GFX_EA_DRAMWR_DATAMEM, + RAS_GFX_V9__GFX_EA_RRET_TAGMEM, + RAS_GFX_V9__GFX_EA_WRET_TAGMEM, + RAS_GFX_V9__GFX_EA_GMIRD_CMDMEM, + RAS_GFX_V9__GFX_EA_GMIWR_CMDMEM, + RAS_GFX_V9__GFX_EA_GMIWR_DATAMEM, + RAS_GFX_V9__GFX_EA_INDEX0_END = + RAS_GFX_V9__GFX_EA_GMIWR_DATAMEM, + /* EA range 1 */ + RAS_GFX_V9__GFX_EA_INDEX1_START, + RAS_GFX_V9__GFX_EA_DRAMRD_PAGEMEM = + RAS_GFX_V9__GFX_EA_INDEX1_START, + RAS_GFX_V9__GFX_EA_DRAMWR_PAGEMEM, + RAS_GFX_V9__GFX_EA_IORD_CMDMEM, + RAS_GFX_V9__GFX_EA_IOWR_CMDMEM, + RAS_GFX_V9__GFX_EA_IOWR_DATAMEM, + RAS_GFX_V9__GFX_EA_GMIRD_PAGEMEM, + RAS_GFX_V9__GFX_EA_GMIWR_PAGEMEM, + RAS_GFX_V9__GFX_EA_INDEX1_END = + RAS_GFX_V9__GFX_EA_GMIWR_PAGEMEM, + /* EA range 2 */ + RAS_GFX_V9__GFX_EA_INDEX2_START, + RAS_GFX_V9__GFX_EA_MAM_D0MEM = + RAS_GFX_V9__GFX_EA_INDEX2_START, + RAS_GFX_V9__GFX_EA_MAM_D1MEM, + RAS_GFX_V9__GFX_EA_MAM_D2MEM, + RAS_GFX_V9__GFX_EA_MAM_D3MEM, + RAS_GFX_V9__GFX_EA_INDEX2_END = + RAS_GFX_V9__GFX_EA_MAM_D3MEM, + RAS_GFX_V9__GFX_EA_INDEX_END = + RAS_GFX_V9__GFX_EA_INDEX2_END, + /* UTC VM L2 bank */ + RAS_GFX_V9__UTC_VML2_BANK_CACHE, + /* UTC VM walker */ + RAS_GFX_V9__UTC_VML2_WALKER, + /* UTC ATC L2 2MB cache */ + RAS_GFX_V9__UTC_ATCL2_CACHE_2M_BANK, + /* UTC ATC L2 4KB cache */ + RAS_GFX_V9__UTC_ATCL2_CACHE_4K_BANK, + RAS_GFX_V9__GFX_MAX +}; + +extern const struct ras_gfx_ip_func gfx_ras_func_v9_0; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c new file mode 100644 index 000000000000..0a838fdcb2f6 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c @@ -0,0 +1,317 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_core_status.h" +#include "ras_log_ring.h" + +#define RAS_LOG_MAX_QUERY_SIZE 0xC000 +#define RAS_LOG_MEM_TEMP_SIZE 0x200 +#define RAS_LOG_MEMPOOL_SIZE \ + (RAS_LOG_MAX_QUERY_SIZE + RAS_LOG_MEM_TEMP_SIZE) + +#define BATCH_IDX_TO_TREE_IDX(batch_idx, sn) (((batch_idx) << 8) | (sn)) + +static const uint64_t ras_rma_aca_reg[ACA_REG_MAX_COUNT] = { + [ACA_REG_IDX__CTL] = 0x1, + [ACA_REG_IDX__STATUS] = 0xB000000000000137, + [ACA_REG_IDX__ADDR] = 0x0, + [ACA_REG_IDX__MISC0] = 0x0, + [ACA_REG_IDX__CONFG] = 0x1ff00000002, + [ACA_REG_IDX__IPID] = 0x9600000000, + [ACA_REG_IDX__SYND] = 0x0, +}; + +static uint64_t ras_log_ring_get_logged_ecc_count(struct ras_core_context *ras_core) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + uint64_t count = 0; + + if (log_ring->logged_ecc_count < 0) { + RAS_DEV_WARN(ras_core->dev, + "Error: the logged ras count should not less than 0!\n"); + count = 0; + } else { + count = log_ring->logged_ecc_count; + } + + if (count > RAS_LOG_MEMPOOL_SIZE) + RAS_DEV_WARN(ras_core->dev, + "Error: the logged ras count is out of range!\n"); + + return count; +} + +static int ras_log_ring_add_data(struct ras_core_context *ras_core, + struct ras_log_info *log, struct ras_log_batch_tag *batch_tag) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + unsigned long flags = 0; + int ret = 0; + + if (batch_tag && (batch_tag->sub_seqno >= MAX_RECORD_PER_BATCH)) { + RAS_DEV_ERR(ras_core->dev, + "Invalid batch sub seqno:%d, batch:0x%llx\n", + batch_tag->sub_seqno, batch_tag->batch_id); + return -EINVAL; + } + + spin_lock_irqsave(&log_ring->spin_lock, flags); + if (batch_tag) { + log->seqno = + BATCH_IDX_TO_TREE_IDX(batch_tag->batch_id, batch_tag->sub_seqno); + batch_tag->sub_seqno++; + } else { + log->seqno = BATCH_IDX_TO_TREE_IDX(log_ring->mono_upward_batch_id, 0); + log_ring->mono_upward_batch_id++; + } + ret = radix_tree_insert(&log_ring->ras_log_root, log->seqno, log); + if (!ret) + log_ring->logged_ecc_count++; + spin_unlock_irqrestore(&log_ring->spin_lock, flags); + + if (ret) { + RAS_DEV_ERR(ras_core->dev, + "Failed to add ras log! seqno:0x%llx, ret:%d\n", + log->seqno, ret); + mempool_free(log, log_ring->ras_log_mempool); + } + + return ret; +} + +static int ras_log_ring_delete_data(struct ras_core_context *ras_core, uint32_t count) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + unsigned long flags = 0; + uint32_t i = 0, j = 0; + uint64_t batch_id, idx; + void *data; + int ret = -ENODATA; + + if (count > ras_log_ring_get_logged_ecc_count(ras_core)) + return -EINVAL; + + spin_lock_irqsave(&log_ring->spin_lock, flags); + batch_id = log_ring->last_del_batch_id; + while (batch_id < log_ring->mono_upward_batch_id) { + for (j = 0; j < MAX_RECORD_PER_BATCH; j++) { + idx = BATCH_IDX_TO_TREE_IDX(batch_id, j); + data = radix_tree_delete(&log_ring->ras_log_root, idx); + if (data) { + mempool_free(data, log_ring->ras_log_mempool); + log_ring->logged_ecc_count--; + i++; + } + } + batch_id = ++log_ring->last_del_batch_id; + if (i >= count) { + ret = 0; + break; + } + } + spin_unlock_irqrestore(&log_ring->spin_lock, flags); + + return ret; +} + +static void ras_log_ring_clear_log_tree(struct ras_core_context *ras_core) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + uint64_t batch_id, idx; + unsigned long flags = 0; + void *data; + int j; + + if ((log_ring->mono_upward_batch_id <= log_ring->last_del_batch_id) && + !log_ring->logged_ecc_count) + return; + + spin_lock_irqsave(&log_ring->spin_lock, flags); + batch_id = log_ring->last_del_batch_id; + while (batch_id < log_ring->mono_upward_batch_id) { + for (j = 0; j < MAX_RECORD_PER_BATCH; j++) { + idx = BATCH_IDX_TO_TREE_IDX(batch_id, j); + data = radix_tree_delete(&log_ring->ras_log_root, idx); + if (data) { + mempool_free(data, log_ring->ras_log_mempool); + log_ring->logged_ecc_count--; + } + } + batch_id++; + } + spin_unlock_irqrestore(&log_ring->spin_lock, flags); + +} + +int ras_log_ring_sw_init(struct ras_core_context *ras_core) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + + memset(log_ring, 0, sizeof(*log_ring)); + + log_ring->ras_log_mempool = mempool_create_kmalloc_pool( + RAS_LOG_MEMPOOL_SIZE, sizeof(struct ras_log_info)); + if (!log_ring->ras_log_mempool) + return -ENOMEM; + + INIT_RADIX_TREE(&log_ring->ras_log_root, GFP_KERNEL); + + spin_lock_init(&log_ring->spin_lock); + + return 0; +} + +int ras_log_ring_sw_fini(struct ras_core_context *ras_core) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + + ras_log_ring_clear_log_tree(ras_core); + log_ring->logged_ecc_count = 0; + log_ring->last_del_batch_id = 0; + log_ring->mono_upward_batch_id = 0; + + mempool_destroy(log_ring->ras_log_mempool); + + return 0; +} + +struct ras_log_batch_tag *ras_log_ring_create_batch_tag(struct ras_core_context *ras_core) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + struct ras_log_batch_tag *batch_tag; + unsigned long flags = 0; + + batch_tag = kzalloc(sizeof(*batch_tag), GFP_KERNEL); + if (!batch_tag) + return NULL; + + spin_lock_irqsave(&log_ring->spin_lock, flags); + batch_tag->batch_id = log_ring->mono_upward_batch_id; + log_ring->mono_upward_batch_id++; + spin_unlock_irqrestore(&log_ring->spin_lock, flags); + + batch_tag->sub_seqno = 0; + batch_tag->timestamp = ras_core_get_utc_second_timestamp(ras_core); + return batch_tag; +} + +void ras_log_ring_destroy_batch_tag(struct ras_core_context *ras_core, + struct ras_log_batch_tag *batch_tag) +{ + kfree(batch_tag); +} + +void ras_log_ring_add_log_event(struct ras_core_context *ras_core, + enum ras_log_event event, void *data, struct ras_log_batch_tag *batch_tag) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + struct device_system_info dev_info = {0}; + struct ras_log_info *log; + uint64_t socket_id; + void *obj; + + obj = mempool_alloc_preallocated(log_ring->ras_log_mempool); + if (!obj || + (ras_log_ring_get_logged_ecc_count(ras_core) >= RAS_LOG_MEMPOOL_SIZE)) { + ras_log_ring_delete_data(ras_core, RAS_LOG_MEM_TEMP_SIZE); + if (!obj) + obj = mempool_alloc_preallocated(log_ring->ras_log_mempool); + } + + if (!obj) { + RAS_DEV_ERR(ras_core->dev, "ERROR: Failed to alloc ras log buffer!\n"); + return; + } + + log = (struct ras_log_info *)obj; + + memset(log, 0, sizeof(*log)); + log->timestamp = + batch_tag ? batch_tag->timestamp : ras_core_get_utc_second_timestamp(ras_core); + log->event = event; + + if (data) + memcpy(&log->aca_reg, data, sizeof(log->aca_reg)); + + if (event == RAS_LOG_EVENT_RMA) { + memcpy(&log->aca_reg, ras_rma_aca_reg, sizeof(log->aca_reg)); + ras_core_get_device_system_info(ras_core, &dev_info); + socket_id = dev_info.socket_id; + log->aca_reg.regs[ACA_REG_IDX__IPID] |= ((socket_id / 4) & 0x01); + log->aca_reg.regs[ACA_REG_IDX__IPID] |= (((socket_id % 4) & 0x3) << 44); + } + + ras_log_ring_add_data(ras_core, log, batch_tag); +} + +static struct ras_log_info *ras_log_ring_lookup_data(struct ras_core_context *ras_core, + uint64_t idx) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + unsigned long flags = 0; + void *data; + + spin_lock_irqsave(&log_ring->spin_lock, flags); + data = radix_tree_lookup(&log_ring->ras_log_root, idx); + spin_unlock_irqrestore(&log_ring->spin_lock, flags); + + return (struct ras_log_info *)data; +} + +int ras_log_ring_get_batch_records(struct ras_core_context *ras_core, uint64_t batch_id, + struct ras_log_info **log_arr, uint32_t arr_num) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + uint32_t i, idx, count = 0; + void *data; + + if ((batch_id >= log_ring->mono_upward_batch_id) || + (batch_id < log_ring->last_del_batch_id)) + return -EINVAL; + + for (i = 0; i < MAX_RECORD_PER_BATCH; i++) { + idx = BATCH_IDX_TO_TREE_IDX(batch_id, i); + data = ras_log_ring_lookup_data(ras_core, idx); + if (data) { + log_arr[count++] = data; + if (count >= arr_num) + break; + } + } + + return count; +} + +int ras_log_ring_get_batch_overview(struct ras_core_context *ras_core, + struct ras_log_batch_overview *overview) +{ + struct ras_log_ring *log_ring = &ras_core->ras_log_ring; + + overview->logged_batch_count = + log_ring->mono_upward_batch_id - log_ring->last_del_batch_id; + overview->last_batch_id = log_ring->mono_upward_batch_id; + overview->first_batch_id = log_ring->last_del_batch_id; + + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.h b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.h new file mode 100644 index 000000000000..0ff6cc35678d --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.h @@ -0,0 +1,93 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_LOG_RING_H__ +#define __RAS_LOG_RING_H__ +#include "ras_aca.h" + +#define MAX_RECORD_PER_BATCH 32 + +#define RAS_LOG_SEQNO_TO_BATCH_IDX(seqno) ((seqno) >> 8) + +enum ras_log_event { + RAS_LOG_EVENT_NONE, + RAS_LOG_EVENT_UE, + RAS_LOG_EVENT_DE, + RAS_LOG_EVENT_CE, + RAS_LOG_EVENT_POISON_CREATION, + RAS_LOG_EVENT_POISON_CONSUMPTION, + RAS_LOG_EVENT_RMA, + RAS_LOG_EVENT_COUNT_MAX, +}; + +struct ras_aca_reg { + uint64_t regs[ACA_REG_MAX_COUNT]; +}; + +struct ras_log_info { + uint64_t seqno; + uint64_t timestamp; + enum ras_log_event event; + union { + struct ras_aca_reg aca_reg; + }; +}; + +struct ras_log_batch_tag { + uint64_t batch_id; + uint64_t timestamp; + uint32_t sub_seqno; +}; + +struct ras_log_ring { + void *ras_log_mempool; + struct radix_tree_root ras_log_root; + spinlock_t spin_lock; + uint64_t mono_upward_batch_id; + uint64_t last_del_batch_id; + int logged_ecc_count; +}; + +struct ras_log_batch_overview { + uint64_t first_batch_id; + uint64_t last_batch_id; + uint32_t logged_batch_count; +}; + +struct ras_core_context; + +int ras_log_ring_sw_init(struct ras_core_context *ras_core); +int ras_log_ring_sw_fini(struct ras_core_context *ras_core); + +struct ras_log_batch_tag *ras_log_ring_create_batch_tag(struct ras_core_context *ras_core); +void ras_log_ring_destroy_batch_tag(struct ras_core_context *ras_core, + struct ras_log_batch_tag *tag); +void ras_log_ring_add_log_event(struct ras_core_context *ras_core, + enum ras_log_event event, void *data, struct ras_log_batch_tag *tag); + +int ras_log_ring_get_batch_records(struct ras_core_context *ras_core, uint64_t batch_idx, + struct ras_log_info **log_arr, uint32_t arr_num); + +int ras_log_ring_get_batch_overview(struct ras_core_context *ras_core, + struct ras_log_batch_overview *overview); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_mp1.c b/drivers/gpu/drm/amd/ras/rascore/ras_mp1.c new file mode 100644 index 000000000000..f3321df85021 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_mp1.c @@ -0,0 +1,81 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras.h" +#include "ras_mp1.h" +#include "ras_mp1_v13_0.h" + +static const struct ras_mp1_ip_func *ras_mp1_get_ip_funcs( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 14): + case IP_VERSION(13, 0, 12): + return &mp1_ras_func_v13_0; + default: + RAS_DEV_ERR(ras_core->dev, + "MP1 ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +int ras_mp1_get_bank_count(struct ras_core_context *ras_core, + enum ras_err_type type, u32 *count) +{ + struct ras_mp1 *mp1 = &ras_core->ras_mp1; + + return mp1->ip_func->get_valid_bank_count(ras_core, type, count); +} + +int ras_mp1_dump_bank(struct ras_core_context *ras_core, + u32 type, u32 idx, u32 reg_idx, u64 *val) +{ + struct ras_mp1 *mp1 = &ras_core->ras_mp1; + + return mp1->ip_func->dump_valid_bank(ras_core, type, idx, reg_idx, val); +} + +int ras_mp1_hw_init(struct ras_core_context *ras_core) +{ + struct ras_mp1 *mp1 = &ras_core->ras_mp1; + + mp1->mp1_ip_version = ras_core->config->mp1_ip_version; + mp1->sys_func = ras_core->config->mp1_cfg.mp1_sys_fn; + if (!mp1->sys_func) { + RAS_DEV_ERR(ras_core->dev, "RAS mp1 sys function not configured!\n"); + return -EINVAL; + } + + mp1->ip_func = ras_mp1_get_ip_funcs(ras_core, mp1->mp1_ip_version); + + return mp1->ip_func ? RAS_CORE_OK : -EINVAL; +} + +int ras_mp1_hw_fini(struct ras_core_context *ras_core) +{ + return 0; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_mp1.h b/drivers/gpu/drm/amd/ras/rascore/ras_mp1.h new file mode 100644 index 000000000000..de1d08286f41 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_mp1.h @@ -0,0 +1,50 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_MP1_H__ +#define __RAS_MP1_H__ +#include "ras.h" + +enum ras_err_type; +struct ras_mp1_ip_func { + int (*get_valid_bank_count)(struct ras_core_context *ras_core, + enum ras_err_type type, u32 *count); + int (*dump_valid_bank)(struct ras_core_context *ras_core, + enum ras_err_type type, u32 idx, u32 reg_idx, u64 *val); +}; + +struct ras_mp1 { + uint32_t mp1_ip_version; + const struct ras_mp1_ip_func *ip_func; + const struct ras_mp1_sys_func *sys_func; +}; + +int ras_mp1_hw_init(struct ras_core_context *ras_core); +int ras_mp1_hw_fini(struct ras_core_context *ras_core); + +int ras_mp1_get_bank_count(struct ras_core_context *ras_core, + enum ras_err_type type, u32 *count); + +int ras_mp1_dump_bank(struct ras_core_context *ras_core, + u32 ecc_type, u32 idx, u32 reg_idx, u64 *val); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.c b/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.c new file mode 100644 index 000000000000..310d39fc816b --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.c @@ -0,0 +1,105 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_mp1.h" +#include "ras_core_status.h" +#include "ras_mp1_v13_0.h" + +#define RAS_MP1_MSG_QueryValidMcaCount 0x36 +#define RAS_MP1_MSG_McaBankDumpDW 0x37 +#define RAS_MP1_MSG_ClearMcaOnRead 0x39 +#define RAS_MP1_MSG_QueryValidMcaCeCount 0x3A +#define RAS_MP1_MSG_McaBankCeDumpDW 0x3B + +#define MAX_UE_BANKS_PER_QUERY 12 +#define MAX_CE_BANKS_PER_QUERY 12 + +static int mp1_v13_0_get_bank_count(struct ras_core_context *ras_core, + enum ras_err_type type, u32 *count) +{ + struct ras_mp1 *mp1 = &ras_core->ras_mp1; + const struct ras_mp1_sys_func *sys_func = mp1->sys_func; + uint32_t bank_count = 0; + u32 msg; + int ret; + + if (!count) + return -EINVAL; + + if (!sys_func || !sys_func->mp1_get_valid_bank_count) + return -RAS_CORE_NOT_SUPPORTED; + + switch (type) { + case RAS_ERR_TYPE__UE: + msg = RAS_MP1_MSG_QueryValidMcaCount; + break; + case RAS_ERR_TYPE__CE: + case RAS_ERR_TYPE__DE: + msg = RAS_MP1_MSG_QueryValidMcaCeCount; + break; + default: + return -EINVAL; + } + + ret = sys_func->mp1_get_valid_bank_count(ras_core, msg, &bank_count); + if (!ret) { + if (((type == RAS_ERR_TYPE__UE) && (bank_count >= MAX_UE_BANKS_PER_QUERY)) || + ((type == RAS_ERR_TYPE__CE) && (bank_count >= MAX_CE_BANKS_PER_QUERY))) + return -EINVAL; + + *count = bank_count; + } + + return ret; +} + +static int mp1_v13_0_dump_bank(struct ras_core_context *ras_core, + enum ras_err_type type, u32 idx, u32 reg_idx, u64 *val) +{ + struct ras_mp1 *mp1 = &ras_core->ras_mp1; + const struct ras_mp1_sys_func *sys_func = mp1->sys_func; + u32 msg; + + if (!sys_func || !sys_func->mp1_dump_valid_bank) + return -RAS_CORE_NOT_SUPPORTED; + + switch (type) { + case RAS_ERR_TYPE__UE: + msg = RAS_MP1_MSG_McaBankDumpDW; + break; + case RAS_ERR_TYPE__CE: + case RAS_ERR_TYPE__DE: + msg = RAS_MP1_MSG_McaBankCeDumpDW; + break; + default: + return -EINVAL; + } + + return sys_func->mp1_dump_valid_bank(ras_core, msg, idx, reg_idx, val); +} + +const struct ras_mp1_ip_func mp1_ras_func_v13_0 = { + .get_valid_bank_count = mp1_v13_0_get_bank_count, + .dump_valid_bank = mp1_v13_0_dump_bank, +}; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.h b/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.h new file mode 100644 index 000000000000..2edfdb5f6a75 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_mp1_v13_0.h @@ -0,0 +1,30 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_MP1_V13_0_H__ +#define __RAS_MP1_V13_0_H__ +#include "ras_mp1.h" + +extern const struct ras_mp1_ip_func mp1_ras_func_v13_0; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c new file mode 100644 index 000000000000..bfddd104d548 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c @@ -0,0 +1,96 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras.h" +#include "ras_nbio.h" +#include "ras_nbio_v7_9.h" + +static const struct ras_nbio_ip_func *ras_nbio_get_ip_funcs( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(7, 9, 0): + case IP_VERSION(7, 9, 1): + return &ras_nbio_v7_9; + default: + RAS_DEV_ERR(ras_core->dev, + "NBIO ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +int ras_nbio_hw_init(struct ras_core_context *ras_core) +{ + struct ras_nbio *nbio = &ras_core->ras_nbio; + + nbio->nbio_ip_version = ras_core->config->nbio_ip_version; + nbio->sys_func = ras_core->config->nbio_cfg.nbio_sys_fn; + if (!nbio->sys_func) { + RAS_DEV_ERR(ras_core->dev, "RAS nbio sys function not configured!\n"); + return -EINVAL; + } + + nbio->ip_func = ras_nbio_get_ip_funcs(ras_core, nbio->nbio_ip_version); + if (!nbio->ip_func) + return -EINVAL; + + if (nbio->sys_func) { + if (nbio->sys_func->set_ras_controller_irq_state) + nbio->sys_func->set_ras_controller_irq_state(ras_core, true); + if (nbio->sys_func->set_ras_err_event_athub_irq_state) + nbio->sys_func->set_ras_err_event_athub_irq_state(ras_core, true); + } + + return 0; +} + +int ras_nbio_hw_fini(struct ras_core_context *ras_core) +{ + struct ras_nbio *nbio = &ras_core->ras_nbio; + + if (nbio->sys_func) { + if (nbio->sys_func->set_ras_controller_irq_state) + nbio->sys_func->set_ras_controller_irq_state(ras_core, false); + if (nbio->sys_func->set_ras_err_event_athub_irq_state) + nbio->sys_func->set_ras_err_event_athub_irq_state(ras_core, false); + } + + return 0; +} + +bool ras_nbio_handle_irq_error(struct ras_core_context *ras_core, void *data) +{ + struct ras_nbio *nbio = &ras_core->ras_nbio; + + if (nbio->ip_func) { + if (nbio->ip_func->handle_ras_controller_intr_no_bifring) + nbio->ip_func->handle_ras_controller_intr_no_bifring(ras_core); + if (nbio->ip_func->handle_ras_err_event_athub_intr_no_bifring) + nbio->ip_func->handle_ras_err_event_athub_intr_no_bifring(ras_core); + } + + return true; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_nbio.h b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.h new file mode 100644 index 000000000000..0a1313e59a02 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.h @@ -0,0 +1,46 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_NBIO_H__ +#define __RAS_NBIO_H__ +#include "ras.h" + +struct ras_core_context; + +struct ras_nbio_ip_func { + int (*handle_ras_controller_intr_no_bifring)(struct ras_core_context *ras_core); + int (*handle_ras_err_event_athub_intr_no_bifring)(struct ras_core_context *ras_core); + uint32_t (*get_memory_partition_mode)(struct ras_core_context *ras_core); +}; + +struct ras_nbio { + uint32_t nbio_ip_version; + const struct ras_nbio_ip_func *ip_func; + const struct ras_nbio_sys_func *sys_func; +}; + +int ras_nbio_hw_init(struct ras_core_context *ras_core); +int ras_nbio_hw_fini(struct ras_core_context *ras_core); +bool ras_nbio_handle_irq_error(struct ras_core_context *ras_core, void *data); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.c b/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.c new file mode 100644 index 000000000000..f17d708ec668 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.c @@ -0,0 +1,123 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras.h" +#include "ras_nbio_v7_9.h" + +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR__SHIFT 0x12 +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR_MASK 0x00040000L +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS__SHIFT 0x2 +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS_MASK 0x00000004L +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_CNTLR_INTERRUPT_CLEAR__SHIFT 0x11 +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_CNTLR_INTERRUPT_CLEAR_MASK 0x00020000L +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_CNTLR_INTERRUPT_STATUS__SHIFT 0x1 +#define BIF_BX0_BIF_DOORBELL_INT_CNTL__RAS_CNTLR_INTERRUPT_STATUS_MASK 0x00000002L + +#define regBIF_BX0_BIF_DOORBELL_INT_CNTL_BASE_IDX 2 +#define regBIF_BX0_BIF_DOORBELL_INT_CNTL 0x00fe + +#define regBIF_BX0_BIF_INTR_CNTL 0x0101 +#define regBIF_BX0_BIF_INTR_CNTL_BASE_IDX 2 + +/* BIF_BX0_BIF_INTR_CNTL */ +#define BIF_BX0_BIF_INTR_CNTL__RAS_INTR_VEC_SEL__SHIFT 0x0 +#define BIF_BX0_BIF_INTR_CNTL__RAS_INTR_VEC_SEL_MASK 0x00000001L + +#define regBIF_BX_PF0_PARTITION_MEM_STATUS 0x0164 +#define regBIF_BX_PF0_PARTITION_MEM_STATUS_BASE_IDX 2 +/* BIF_BX_PF0_PARTITION_MEM_STATUS */ +#define BIF_BX_PF0_PARTITION_MEM_STATUS__CHANGE_STATUE__SHIFT 0x0 +#define BIF_BX_PF0_PARTITION_MEM_STATUS__NPS_MODE__SHIFT 0x4 +#define BIF_BX_PF0_PARTITION_MEM_STATUS__CHANGE_STATUE_MASK 0x0000000FL +#define BIF_BX_PF0_PARTITION_MEM_STATUS__NPS_MODE_MASK 0x00000FF0L + + +static int nbio_v7_9_handle_ras_controller_intr_no_bifring(struct ras_core_context *ras_core) +{ + uint32_t bif_doorbell_intr_cntl = 0; + + bif_doorbell_intr_cntl = + RAS_DEV_RREG32_SOC15(ras_core->dev, NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL); + + if (REG_GET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_CNTLR_INTERRUPT_STATUS)) { + /* driver has to clear the interrupt status when bif ring is disabled */ + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, + RAS_CNTLR_INTERRUPT_CLEAR, 1); + + RAS_DEV_WREG32_SOC15(ras_core->dev, + NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + /* TODO: handle ras controller interrupt */ + } + + return 0; +} + +static int nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring(struct ras_core_context *ras_core) +{ + uint32_t bif_doorbell_intr_cntl = 0; + int ret = 0; + + bif_doorbell_intr_cntl = + RAS_DEV_RREG32_SOC15(ras_core->dev, NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL); + + if (REG_GET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, RAS_ATHUB_ERR_EVENT_INTERRUPT_STATUS)) { + /* driver has to clear the interrupt status when bif ring is disabled */ + bif_doorbell_intr_cntl = REG_SET_FIELD(bif_doorbell_intr_cntl, + BIF_BX0_BIF_DOORBELL_INT_CNTL, + RAS_ATHUB_ERR_EVENT_INTERRUPT_CLEAR, 1); + + RAS_DEV_WREG32_SOC15(ras_core->dev, + NBIO, 0, regBIF_BX0_BIF_DOORBELL_INT_CNTL, bif_doorbell_intr_cntl); + + ret = ras_core_handle_fatal_error(ras_core); + } + + return ret; +} + +static uint32_t nbio_v7_9_get_memory_partition_mode(struct ras_core_context *ras_core) +{ + uint32_t mem_status; + uint32_t mem_mode; + + mem_status = + RAS_DEV_RREG32_SOC15(ras_core->dev, NBIO, 0, regBIF_BX_PF0_PARTITION_MEM_STATUS); + + /* Each bit represents a mode 1-8*/ + mem_mode = REG_GET_FIELD(mem_status, BIF_BX_PF0_PARTITION_MEM_STATUS, NPS_MODE); + + return ffs(mem_mode); +} + +const struct ras_nbio_ip_func ras_nbio_v7_9 = { + .handle_ras_controller_intr_no_bifring = + nbio_v7_9_handle_ras_controller_intr_no_bifring, + .handle_ras_err_event_athub_intr_no_bifring = + nbio_v7_9_handle_ras_err_event_athub_intr_no_bifring, + .get_memory_partition_mode = nbio_v7_9_get_memory_partition_mode, +}; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.h b/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.h new file mode 100644 index 000000000000..8711c82a927f --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_nbio_v7_9.h @@ -0,0 +1,31 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_NBIO_V7_9_H__ +#define __RAS_NBIO_V7_9_H__ +#include "ras_nbio.h" + +extern const struct ras_nbio_ip_func ras_nbio_v7_9; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_process.c b/drivers/gpu/drm/amd/ras/rascore/ras_process.c new file mode 100644 index 000000000000..3267dcdb169c --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_process.c @@ -0,0 +1,322 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_process.h" + +#define RAS_EVENT_FIFO_SIZE (128 * sizeof(struct ras_event_req)) + +#define RAS_POLLING_ECC_TIMEOUT 300 + +static int ras_process_put_event(struct ras_core_context *ras_core, + struct ras_event_req *req) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + int ret; + + ret = kfifo_in_spinlocked(&ras_proc->event_fifo, + req, sizeof(*req), &ras_proc->fifo_spinlock); + if (!ret) { + RAS_DEV_ERR(ras_core->dev, "Poison message fifo is full!\n"); + return -ENOSPC; + } + + return 0; +} + +static int ras_process_add_reset_gpu_event(struct ras_core_context *ras_core, + uint32_t reset_cause) +{ + struct ras_event_req req = {0}; + + req.reset = reset_cause; + + return ras_process_put_event(ras_core, &req); +} + +static int ras_process_get_event(struct ras_core_context *ras_core, + struct ras_event_req *req) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + + return kfifo_out_spinlocked(&ras_proc->event_fifo, + req, sizeof(*req), &ras_proc->fifo_spinlock); +} + +static void ras_process_clear_event_fifo(struct ras_core_context *ras_core) +{ + struct ras_event_req req; + int ret; + + do { + ret = ras_process_get_event(ras_core, &req); + } while (ret); +} + +#define AMDGPU_RAS_WAITING_DATA_READY 200 +static int ras_process_umc_event(struct ras_core_context *ras_core, + uint32_t event_count) +{ + struct ras_ecc_count ecc_data; + int ret = 0; + uint32_t timeout = 0; + uint32_t detected_de_count = 0; + + do { + memset(&ecc_data, 0, sizeof(ecc_data)); + ret = ras_core_update_ecc_info(ras_core); + if (ret) + return ret; + + ret = ras_core_query_block_ecc_data(ras_core, RAS_BLOCK_ID__UMC, &ecc_data); + if (ret) + return ret; + + if (ecc_data.new_de_count) { + detected_de_count += ecc_data.new_de_count; + timeout = 0; + } else { + if (!timeout && event_count) + timeout = AMDGPU_RAS_WAITING_DATA_READY; + + if (timeout) { + if (!--timeout) + break; + + msleep(1); + } + } + } while (detected_de_count < event_count); + + if (detected_de_count && ras_core_gpu_is_rma(ras_core)) + ras_process_add_reset_gpu_event(ras_core, GPU_RESET_CAUSE_RMA); + + return 0; +} + +static int ras_process_non_umc_event(struct ras_core_context *ras_core) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + struct ras_event_req req; + uint32_t event_count = kfifo_len(&ras_proc->event_fifo); + uint32_t reset_flags = 0; + int ret = 0, i; + + for (i = 0; i < event_count; i++) { + memset(&req, 0, sizeof(req)); + ret = ras_process_get_event(ras_core, &req); + if (!ret) + continue; + + ras_core_event_notify(ras_core, + RAS_EVENT_ID__POISON_CONSUMPTION, &req); + + reset_flags |= req.reset; + + if (req.reset == GPU_RESET_CAUSE_RMA) + continue; + + if (req.reset) + RAS_DEV_INFO(ras_core->dev, + "{%llu} GPU reset for %s RAS poison consumption is issued!\n", + req.seqno, ras_core_get_ras_block_name(req.block)); + else + RAS_DEV_INFO(ras_core->dev, + "{%llu} %s RAS poison consumption is issued!\n", + req.seqno, ras_core_get_ras_block_name(req.block)); + } + + if (reset_flags) { + ret = ras_core_event_notify(ras_core, + RAS_EVENT_ID__RESET_GPU, &reset_flags); + if (!ret && (reset_flags & GPU_RESET_CAUSE_RMA)) + return -RAS_CORE_GPU_IN_MODE1_RESET; + } + + return ret; +} + +int ras_process_handle_ras_event(struct ras_core_context *ras_core) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + uint32_t umc_event_count; + int ret; + + ret = ras_core_event_notify(ras_core, + RAS_EVENT_ID__RAS_EVENT_PROC_BEGIN, NULL); + if (ret) + return ret; + + ras_aca_clear_fatal_flag(ras_core); + ras_umc_log_pending_bad_bank(ras_core); + + do { + umc_event_count = atomic_read(&ras_proc->umc_interrupt_count); + ret = ras_process_umc_event(ras_core, umc_event_count); + if (ret == -RAS_CORE_GPU_IN_MODE1_RESET) + break; + + if (umc_event_count) + atomic_sub(umc_event_count, &ras_proc->umc_interrupt_count); + } while (atomic_read(&ras_proc->umc_interrupt_count)); + + if ((ret != -RAS_CORE_GPU_IN_MODE1_RESET) && + (kfifo_len(&ras_proc->event_fifo))) + ret = ras_process_non_umc_event(ras_core); + + if (ret == -RAS_CORE_GPU_IN_MODE1_RESET) { + /* Clear poison fifo */ + ras_process_clear_event_fifo(ras_core); + atomic_set(&ras_proc->umc_interrupt_count, 0); + } + + ras_core_event_notify(ras_core, + RAS_EVENT_ID__RAS_EVENT_PROC_END, NULL); + return ret; +} + +static int thread_wait_condition(void *param) +{ + struct ras_process *ras_proc = (struct ras_process *)param; + + return (kthread_should_stop() || + atomic_read(&ras_proc->ras_interrupt_req)); +} + +static int ras_process_thread(void *context) +{ + struct ras_core_context *ras_core = (struct ras_core_context *)context; + struct ras_process *ras_proc = &ras_core->ras_proc; + + while (!kthread_should_stop()) { + ras_wait_event_interruptible_timeout(&ras_proc->ras_process_wq, + thread_wait_condition, ras_proc, + msecs_to_jiffies(RAS_POLLING_ECC_TIMEOUT)); + + if (kthread_should_stop()) + break; + + if (!ras_core->is_initialized) + continue; + + atomic_set(&ras_proc->ras_interrupt_req, 0); + + if (ras_core_gpu_in_reset(ras_core)) + continue; + + if (ras_core->sys_fn && ras_core->sys_fn->async_handle_ras_event) + ras_core->sys_fn->async_handle_ras_event(ras_core, NULL); + else + ras_process_handle_ras_event(ras_core); + } + + return 0; +} + +int ras_process_init(struct ras_core_context *ras_core) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + int ret; + + ret = kfifo_alloc(&ras_proc->event_fifo, RAS_EVENT_FIFO_SIZE, GFP_KERNEL); + if (ret) + return ret; + + spin_lock_init(&ras_proc->fifo_spinlock); + + init_waitqueue_head(&ras_proc->ras_process_wq); + + ras_proc->ras_process_thread = kthread_run(ras_process_thread, + (void *)ras_core, "ras_process_thread"); + if (!ras_proc->ras_process_thread) { + RAS_DEV_ERR(ras_core->dev, "Failed to create ras_process_thread.\n"); + ret = -ENOMEM; + goto err; + } + + return 0; + +err: + ras_process_fini(ras_core); + return ret; +} + +int ras_process_fini(struct ras_core_context *ras_core) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + + if (ras_proc->ras_process_thread) { + kthread_stop(ras_proc->ras_process_thread); + ras_proc->ras_process_thread = NULL; + } + + kfifo_free(&ras_proc->event_fifo); + + return 0; +} + +static int ras_process_add_umc_interrupt_req(struct ras_core_context *ras_core, + struct ras_event_req *req) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + + atomic_inc(&ras_proc->umc_interrupt_count); + atomic_inc(&ras_proc->ras_interrupt_req); + + wake_up(&ras_proc->ras_process_wq); + return 0; +} + +static int ras_process_add_non_umc_interrupt_req(struct ras_core_context *ras_core, + struct ras_event_req *req) +{ + struct ras_process *ras_proc = &ras_core->ras_proc; + int ret; + + ret = ras_process_put_event(ras_core, req); + if (!ret) { + atomic_inc(&ras_proc->ras_interrupt_req); + wake_up(&ras_proc->ras_process_wq); + } + + return ret; +} + +int ras_process_add_interrupt_req(struct ras_core_context *ras_core, + struct ras_event_req *req, bool is_umc) +{ + int ret; + + if (!ras_core) + return -EINVAL; + + if (!ras_core->is_initialized) + return -EPERM; + + if (is_umc) + ret = ras_process_add_umc_interrupt_req(ras_core, req); + else + ret = ras_process_add_non_umc_interrupt_req(ras_core, req); + + return ret; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_process.h b/drivers/gpu/drm/amd/ras/rascore/ras_process.h new file mode 100644 index 000000000000..28458b50510e --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_process.h @@ -0,0 +1,53 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_PROCESS_H__ +#define __RAS_PROCESS_H__ + +struct ras_event_req { + uint64_t seqno; + uint32_t idx_vf; + uint32_t block; + uint16_t pasid; + uint32_t reset; + void *pasid_fn; + void *data; +}; + +struct ras_process { + void *dev; + void *ras_process_thread; + wait_queue_head_t ras_process_wq; + atomic_t ras_interrupt_req; + atomic_t umc_interrupt_count; + struct kfifo event_fifo; + spinlock_t fifo_spinlock; +}; + +struct ras_core_context; +int ras_process_init(struct ras_core_context *ras_core); +int ras_process_fini(struct ras_core_context *ras_core); +int ras_process_handle_ras_event(struct ras_core_context *ras_core); +int ras_process_add_interrupt_req(struct ras_core_context *ras_core, + struct ras_event_req *req, bool is_umc); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_psp.c b/drivers/gpu/drm/amd/ras/rascore/ras_psp.c new file mode 100644 index 000000000000..ccdb42d2dd60 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_psp.c @@ -0,0 +1,750 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_ta_if.h" +#include "ras_psp.h" +#include "ras_psp_v13_0.h" + +/* position of instance value in sub_block_index of + * ta_ras_trigger_error_input, the sub block uses lower 12 bits + */ +#define RAS_TA_INST_MASK 0xfffff000 +#define RAS_TA_INST_SHIFT 0xc + +static const struct ras_psp_ip_func *ras_psp_get_ip_funcs( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 14): + case IP_VERSION(13, 0, 12): + return &ras_psp_v13_0; + default: + RAS_DEV_ERR(ras_core->dev, + "psp ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +static int ras_psp_sync_system_ras_psp_status(struct ras_core_context *ras_core) +{ + struct ras_psp *psp = &ras_core->ras_psp; + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + struct ras_psp_ctx *psp_ctx = &ras_core->ras_psp.psp_ctx; + struct ras_psp_sys_status status = {0}; + int ret; + + if (psp->sys_func && psp->sys_func->get_ras_psp_system_status) { + ret = psp->sys_func->get_ras_psp_system_status(ras_core, &status); + if (ret) + return ret; + + if (status.initialized) { + ta_ctx->preload_ras_ta_enabled = true; + ta_ctx->ras_ta_initialized = status.initialized; + ta_ctx->session_id = status.session_id; + } + + psp_ctx->external_mutex = status.psp_cmd_mutex; + } + + return 0; +} + +static int ras_psp_get_ras_ta_init_param(struct ras_core_context *ras_core, + struct ras_ta_init_param *ras_ta_param) +{ + struct ras_psp *psp = &ras_core->ras_psp; + + if (psp->sys_func && psp->sys_func->get_ras_ta_init_param) + return psp->sys_func->get_ras_ta_init_param(ras_core, ras_ta_param); + + RAS_DEV_ERR(ras_core->dev, "Not config get_ras_ta_init_param API!!\n"); + return -EACCES; +} + +static struct gpu_mem_block *ras_psp_get_gpu_mem(struct ras_core_context *ras_core, + enum gpu_mem_type mem_type) +{ + struct ras_psp *psp = &ras_core->ras_psp; + struct gpu_mem_block *gpu_mem = NULL; + int ret; + + switch (mem_type) { + case GPU_MEM_TYPE_RAS_PSP_RING: + gpu_mem = &psp->psp_ring.ras_ring_gpu_mem; + break; + case GPU_MEM_TYPE_RAS_PSP_CMD: + gpu_mem = &psp->psp_ctx.psp_cmd_gpu_mem; + break; + case GPU_MEM_TYPE_RAS_PSP_FENCE: + gpu_mem = &psp->psp_ctx.out_fence_gpu_mem; + break; + case GPU_MEM_TYPE_RAS_TA_FW: + gpu_mem = &psp->ta_ctx.fw_gpu_mem; + break; + case GPU_MEM_TYPE_RAS_TA_CMD: + gpu_mem = &psp->ta_ctx.cmd_gpu_mem; + break; + default: + return NULL; + } + + if (!gpu_mem->ref_count) { + ret = ras_core_get_gpu_mem(ras_core, mem_type, gpu_mem); + if (ret) + return NULL; + gpu_mem->mem_type = mem_type; + } + + gpu_mem->ref_count++; + + return gpu_mem; +} + +static int ras_psp_put_gpu_mem(struct ras_core_context *ras_core, + struct gpu_mem_block *gpu_mem) +{ + if (!gpu_mem) + return 0; + + gpu_mem->ref_count--; + + if (gpu_mem->ref_count > 0) { + return 0; + } else if (gpu_mem->ref_count < 0) { + RAS_DEV_WARN(ras_core->dev, + "Duplicate free gpu memory %u\n", gpu_mem->mem_type); + } else { + ras_core_put_gpu_mem(ras_core, gpu_mem->mem_type, gpu_mem); + memset(gpu_mem, 0, sizeof(*gpu_mem)); + } + + return 0; +} + +static void __acquire_psp_cmd_lock(struct ras_core_context *ras_core) +{ + struct ras_psp_ctx *psp_ctx = &ras_core->ras_psp.psp_ctx; + + if (psp_ctx->external_mutex) + mutex_lock(psp_ctx->external_mutex); + else + mutex_lock(&psp_ctx->internal_mutex); +} + +static void __release_psp_cmd_lock(struct ras_core_context *ras_core) +{ + struct ras_psp_ctx *psp_ctx = &ras_core->ras_psp.psp_ctx; + + if (psp_ctx->external_mutex) + mutex_unlock(psp_ctx->external_mutex); + else + mutex_unlock(&psp_ctx->internal_mutex); +} + +static uint32_t __get_ring_frame_slot(struct ras_core_context *ras_core) +{ + struct ras_psp *psp = &ras_core->ras_psp; + uint32_t ras_ring_wptr_dw; + + ras_ring_wptr_dw = psp->ip_func->psp_ras_ring_wptr_get(ras_core); + + return div64_u64((ras_ring_wptr_dw << 2), sizeof(struct psp_gfx_rb_frame)); +} + +static int __set_ring_frame_slot(struct ras_core_context *ras_core, + uint32_t slot) +{ + struct ras_psp *psp = &ras_core->ras_psp; + + return psp->ip_func->psp_ras_ring_wptr_set(ras_core, + (slot * sizeof(struct psp_gfx_rb_frame)) >> 2); +} + +static int write_frame_to_ras_psp_ring(struct ras_core_context *ras_core, + struct psp_gfx_rb_frame *frame) +{ + struct gpu_mem_block *ring_mem; + struct psp_gfx_rb_frame *rb_frame; + uint32_t max_frame_slot; + uint32_t slot_idx; + uint32_t write_flush_read_back = 0; + int ret = 0; + + ring_mem = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_PSP_RING); + if (!ring_mem) + return -ENOMEM; + + max_frame_slot = + div64_u64(ring_mem->mem_size, sizeof(struct psp_gfx_rb_frame)); + + rb_frame = + (struct psp_gfx_rb_frame *)ring_mem->mem_cpu_addr; + + slot_idx = __get_ring_frame_slot(ras_core); + if (slot_idx >= max_frame_slot) + slot_idx = 0; + + memcpy(&rb_frame[slot_idx], frame, sizeof(*frame)); + + /* Do a read to force the write of the frame before writing + * write pointer. + */ + write_flush_read_back = rb_frame[slot_idx].fence_value; + if (write_flush_read_back != frame->fence_value) { + RAS_DEV_ERR(ras_core->dev, + "Failed to submit ring cmd! cmd:0x%x:0x%x, fence:0x%x:0x%x value:%u, expected:%u\n", + rb_frame[slot_idx].cmd_buf_addr_hi, + rb_frame[slot_idx].cmd_buf_addr_lo, + rb_frame[slot_idx].fence_addr_hi, + rb_frame[slot_idx].fence_addr_lo, + write_flush_read_back, frame->fence_value); + ret = -EACCES; + goto err; + } + + slot_idx++; + + if (slot_idx >= max_frame_slot) + slot_idx = 0; + + __set_ring_frame_slot(ras_core, slot_idx); + +err: + ras_psp_put_gpu_mem(ras_core, ring_mem); + return ret; +} + +static int send_psp_cmd(struct ras_core_context *ras_core, + enum psp_gfx_cmd_id gfx_cmd_id, void *cmd_data, + uint32_t cmd_size, struct psp_cmd_resp *resp) +{ + struct ras_psp_ctx *psp_ctx = &ras_core->ras_psp.psp_ctx; + struct gpu_mem_block *psp_cmd_buf = NULL; + struct gpu_mem_block *psp_fence_buf = NULL; + struct psp_gfx_cmd_resp *gfx_cmd; + struct psp_gfx_rb_frame rb_frame; + int ret = 0; + int timeout = 1000; + + if (!cmd_data || (cmd_size > sizeof(union psp_gfx_commands)) || !resp) { + RAS_DEV_ERR(ras_core->dev, "Invalid RAS PSP command, id: %u\n", gfx_cmd_id); + return -EINVAL; + } + + __acquire_psp_cmd_lock(ras_core); + + psp_cmd_buf = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_PSP_CMD); + if (!psp_cmd_buf) { + ret = -ENOMEM; + goto exit; + } + + psp_fence_buf = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_PSP_FENCE); + if (!psp_fence_buf) { + ret = -ENOMEM; + goto exit; + } + + gfx_cmd = (struct psp_gfx_cmd_resp *)psp_cmd_buf->mem_cpu_addr; + memset(gfx_cmd, 0, sizeof(*gfx_cmd)); + gfx_cmd->cmd_id = gfx_cmd_id; + memcpy(&gfx_cmd->cmd, cmd_data, cmd_size); + + psp_ctx->in_fence_value++; + + memset(&rb_frame, 0, sizeof(rb_frame)); + rb_frame.cmd_buf_addr_hi = upper_32_bits(psp_cmd_buf->mem_mc_addr); + rb_frame.cmd_buf_addr_lo = lower_32_bits(psp_cmd_buf->mem_mc_addr); + rb_frame.fence_addr_hi = upper_32_bits(psp_fence_buf->mem_mc_addr); + rb_frame.fence_addr_lo = lower_32_bits(psp_fence_buf->mem_mc_addr); + rb_frame.fence_value = psp_ctx->in_fence_value; + + ret = write_frame_to_ras_psp_ring(ras_core, &rb_frame); + if (ret) { + psp_ctx->in_fence_value--; + goto exit; + } + + while (*((uint64_t *)psp_fence_buf->mem_cpu_addr) != + psp_ctx->in_fence_value) { + if (--timeout == 0) + break; + /* + * Shouldn't wait for timeout when err_event_athub occurs, + * because gpu reset thread triggered and lock resource should + * be released for psp resume sequence. + */ + if (ras_core_ras_interrupt_detected(ras_core)) + break; + + msleep(2); + } + + resp->status = gfx_cmd->resp.status; + resp->session_id = gfx_cmd->resp.session_id; + +exit: + ras_psp_put_gpu_mem(ras_core, psp_cmd_buf); + ras_psp_put_gpu_mem(ras_core, psp_fence_buf); + + __release_psp_cmd_lock(ras_core); + + return ret; +} + +static void __check_ras_ta_cmd_resp(struct ras_core_context *ras_core, + struct ras_ta_cmd *ras_cmd) +{ + + if (ras_cmd->ras_out_message.flags.err_inject_switch_disable_flag) { + RAS_DEV_WARN(ras_core->dev, "ECC switch disabled\n"); + ras_cmd->ras_status = RAS_TA_STATUS__ERROR_RAS_NOT_AVAILABLE; + } else if (ras_cmd->ras_out_message.flags.reg_access_failure_flag) + RAS_DEV_WARN(ras_core->dev, "RAS internal register access blocked\n"); + + switch (ras_cmd->ras_status) { + case RAS_TA_STATUS__ERROR_UNSUPPORTED_IP: + RAS_DEV_WARN(ras_core->dev, + "RAS WARNING: cmd failed due to unsupported ip\n"); + break; + case RAS_TA_STATUS__ERROR_UNSUPPORTED_ERROR_INJ: + RAS_DEV_WARN(ras_core->dev, + "RAS WARNING: cmd failed due to unsupported error injection\n"); + break; + case RAS_TA_STATUS__SUCCESS: + break; + case RAS_TA_STATUS__TEE_ERROR_ACCESS_DENIED: + if (ras_cmd->cmd_id == RAS_TA_CMD_ID__TRIGGER_ERROR) + RAS_DEV_WARN(ras_core->dev, + "RAS WARNING: Inject error to critical region is not allowed\n"); + break; + default: + RAS_DEV_WARN(ras_core->dev, + "RAS WARNING: ras status = 0x%X\n", ras_cmd->ras_status); + break; + } +} + +static int send_ras_ta_runtime_cmd(struct ras_core_context *ras_core, + enum ras_ta_cmd_id cmd_id, void *in, uint32_t in_size, + void *out, uint32_t out_size) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + struct gpu_mem_block *cmd_mem; + struct ras_ta_cmd *ras_cmd; + struct psp_gfx_cmd_invoke_cmd invoke_cmd = {0}; + struct psp_cmd_resp resp = {0}; + int ret = 0; + + if (!in || (in_size > sizeof(union ras_ta_cmd_input)) || + (cmd_id >= MAX_RAS_TA_CMD_ID)) { + RAS_DEV_ERR(ras_core->dev, "Invalid RAS TA command, id: %u\n", cmd_id); + return -EINVAL; + } + + ras_psp_sync_system_ras_psp_status(ras_core); + + cmd_mem = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_TA_CMD); + if (!cmd_mem) + return -ENOMEM; + + if (!ras_core_down_trylock_gpu_reset_lock(ras_core)) { + ret = -EACCES; + goto out; + } + + ras_cmd = (struct ras_ta_cmd *)cmd_mem->mem_cpu_addr; + + mutex_lock(&ta_ctx->ta_mutex); + + memset(ras_cmd, 0, sizeof(*ras_cmd)); + ras_cmd->cmd_id = cmd_id; + memcpy(&ras_cmd->ras_in_message, in, in_size); + + invoke_cmd.ta_cmd_id = cmd_id; + invoke_cmd.session_id = ta_ctx->session_id; + + ret = send_psp_cmd(ras_core, GFX_CMD_ID_INVOKE_CMD, + &invoke_cmd, sizeof(invoke_cmd), &resp); + + /* If err_event_athub occurs error inject was successful, however + * return status from TA is no long reliable + */ + if (ras_core_ras_interrupt_detected(ras_core)) { + ret = 0; + goto unlock; + } + + if (ret || resp.status) { + RAS_DEV_ERR(ras_core->dev, + "RAS: Failed to send psp cmd! ret:%d, status:%u\n", + ret, resp.status); + ret = -ESTRPIPE; + goto unlock; + } + + if (ras_cmd->if_version > RAS_TA_HOST_IF_VER) { + RAS_DEV_WARN(ras_core->dev, "RAS: Unsupported Interface\n"); + ret = -EINVAL; + goto unlock; + } + + if (!ras_cmd->ras_status && out && out_size) + memcpy(out, &ras_cmd->ras_out_message, out_size); + + __check_ras_ta_cmd_resp(ras_core, ras_cmd); + +unlock: + mutex_unlock(&ta_ctx->ta_mutex); + ras_core_up_gpu_reset_lock(ras_core); +out: + ras_psp_put_gpu_mem(ras_core, cmd_mem); + return ret; +} + +static int trigger_ras_ta_error(struct ras_core_context *ras_core, + struct ras_ta_trigger_error_input *info, uint32_t instance_mask) +{ + uint32_t dev_mask = 0; + + switch (info->block_id) { + case RAS_TA_BLOCK__GFX: + if (ras_gfx_get_ta_subblock(ras_core, info->inject_error_type, + info->sub_block_index, &info->sub_block_index)) + return -EINVAL; + + dev_mask = RAS_GET_MASK(ras_core->dev, GC, instance_mask); + break; + case RAS_TA_BLOCK__SDMA: + dev_mask = RAS_GET_MASK(ras_core->dev, SDMA0, instance_mask); + break; + case RAS_TA_BLOCK__VCN: + case RAS_TA_BLOCK__JPEG: + dev_mask = RAS_GET_MASK(ras_core->dev, VCN, instance_mask); + break; + default: + dev_mask = instance_mask; + break; + } + + /* reuse sub_block_index for backward compatibility */ + dev_mask <<= RAS_TA_INST_SHIFT; + dev_mask &= RAS_TA_INST_MASK; + info->sub_block_index |= dev_mask; + + return send_ras_ta_runtime_cmd(ras_core, RAS_TA_CMD_ID__TRIGGER_ERROR, + info, sizeof(*info), NULL, 0); +} + +static int send_load_ta_fw_cmd(struct ras_core_context *ras_core, + struct ras_ta_ctx *ta_ctx) +{ + struct ras_ta_fw_bin *fw_bin = &ta_ctx->fw_bin; + struct gpu_mem_block *fw_mem; + struct gpu_mem_block *cmd_mem; + struct ras_ta_cmd *ta_cmd; + struct ras_ta_init_flags *ta_init_flags; + struct psp_gfx_cmd_load_ta psp_load_ta_cmd; + struct psp_cmd_resp resp = {0}; + struct ras_ta_image_header *fw_hdr = NULL; + int ret; + + fw_mem = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_TA_FW); + if (!fw_mem) + return -ENOMEM; + + cmd_mem = ras_psp_get_gpu_mem(ras_core, GPU_MEM_TYPE_RAS_TA_CMD); + if (!cmd_mem) { + ret = -ENOMEM; + goto err; + } + + ret = ras_psp_get_ras_ta_init_param(ras_core, &ta_ctx->init_param); + if (ret) + goto err; + + if (!ras_core_down_trylock_gpu_reset_lock(ras_core)) { + ret = -EACCES; + goto err; + } + + /* copy ras ta binary to shared gpu memory */ + memcpy(fw_mem->mem_cpu_addr, fw_bin->bin_addr, fw_bin->bin_size); + fw_mem->mem_size = fw_bin->bin_size; + + /* Initialize ras ta startup parameter */ + ta_cmd = (struct ras_ta_cmd *)cmd_mem->mem_cpu_addr; + ta_init_flags = &ta_cmd->ras_in_message.init_flags; + + ta_init_flags->poison_mode_en = ta_ctx->init_param.poison_mode_en; + ta_init_flags->dgpu_mode = ta_ctx->init_param.dgpu_mode; + ta_init_flags->xcc_mask = ta_ctx->init_param.xcc_mask; + ta_init_flags->channel_dis_num = ta_ctx->init_param.channel_dis_num; + ta_init_flags->nps_mode = ta_ctx->init_param.nps_mode; + ta_init_flags->active_umc_mask = ta_ctx->init_param.active_umc_mask; + + /* Setup load ras ta command */ + memset(&psp_load_ta_cmd, 0, sizeof(psp_load_ta_cmd)); + psp_load_ta_cmd.app_phy_addr_lo = lower_32_bits(fw_mem->mem_mc_addr); + psp_load_ta_cmd.app_phy_addr_hi = upper_32_bits(fw_mem->mem_mc_addr); + psp_load_ta_cmd.app_len = fw_mem->mem_size; + psp_load_ta_cmd.cmd_buf_phy_addr_lo = lower_32_bits(cmd_mem->mem_mc_addr); + psp_load_ta_cmd.cmd_buf_phy_addr_hi = upper_32_bits(cmd_mem->mem_mc_addr); + psp_load_ta_cmd.cmd_buf_len = cmd_mem->mem_size; + + ret = send_psp_cmd(ras_core, GFX_CMD_ID_LOAD_TA, + &psp_load_ta_cmd, sizeof(psp_load_ta_cmd), &resp); + if (!ret && !resp.status) { + /* Read TA version at FW offset 0x60 if TA version not found*/ + fw_hdr = (struct ras_ta_image_header *)fw_bin->bin_addr; + RAS_DEV_INFO(ras_core->dev, "PSP: RAS TA(version:%X.%X.%X.%X) is loaded.\n", + (fw_hdr->image_version >> 24) & 0xFF, (fw_hdr->image_version >> 16) & 0xFF, + (fw_hdr->image_version >> 8) & 0xFF, fw_hdr->image_version & 0xFF); + ta_ctx->ta_version = fw_hdr->image_version; + ta_ctx->session_id = resp.session_id; + ta_ctx->ras_ta_initialized = true; + } else { + RAS_DEV_ERR(ras_core->dev, + "Failed to load RAS TA! ret:%d, status:%d\n", ret, resp.status); + } + + ras_core_up_gpu_reset_lock(ras_core); + +err: + ras_psp_put_gpu_mem(ras_core, fw_mem); + ras_psp_put_gpu_mem(ras_core, cmd_mem); + return ret; +} + +static int load_ras_ta_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_load *ras_ta_load) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + struct ras_ta_fw_bin *fw_bin = &ta_ctx->fw_bin; + int ret; + + fw_bin->bin_addr = ras_ta_load->bin_addr; + fw_bin->bin_size = ras_ta_load->bin_size; + fw_bin->fw_version = ras_ta_load->fw_version; + fw_bin->feature_version = ras_ta_load->feature_version; + + ret = send_load_ta_fw_cmd(ras_core, ta_ctx); + if (!ret) { + ras_ta_load->out_session_id = ta_ctx->session_id; + ras_ta_load->out_loaded_ta_version = ta_ctx->ta_version; + } + + return ret; +} + +static int unload_ras_ta_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_unload *ras_ta_unload) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + struct psp_gfx_cmd_unload_ta cmd_unload_ta = {0}; + struct psp_cmd_resp resp = {0}; + int ret; + + if (!ras_core_down_trylock_gpu_reset_lock(ras_core)) + return -EACCES; + + cmd_unload_ta.session_id = ta_ctx->session_id; + ret = send_psp_cmd(ras_core, GFX_CMD_ID_UNLOAD_TA, + &cmd_unload_ta, sizeof(cmd_unload_ta), &resp); + if (ret || resp.status) { + RAS_DEV_ERR(ras_core->dev, + "Failed to unload RAS TA! ret:%d, status:%u\n", + ret, resp.status); + goto unlock; + } + + kfree(ta_ctx->fw_bin.bin_addr); + memset(&ta_ctx->fw_bin, 0, sizeof(ta_ctx->fw_bin)); + ta_ctx->ta_version = 0; + ta_ctx->ras_ta_initialized = false; + ta_ctx->session_id = 0; + +unlock: + ras_core_up_gpu_reset_lock(ras_core); + + return ret; +} + +int ras_psp_load_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_load *ras_ta_load) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + struct ras_psp_ta_unload ras_ta_unload = {0}; + int ret; + + if (ta_ctx->preload_ras_ta_enabled) + return 0; + + if (!ras_ta_load) + return -EINVAL; + + if (ta_ctx->ras_ta_initialized) { + ras_ta_unload.ras_session_id = ta_ctx->session_id; + ret = unload_ras_ta_firmware(ras_core, &ras_ta_unload); + if (ret) + return ret; + } + + return load_ras_ta_firmware(ras_core, ras_ta_load); +} + +int ras_psp_unload_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_unload *ras_ta_unload) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + + if (ta_ctx->preload_ras_ta_enabled) + return 0; + + if ((!ras_ta_unload) || + (ras_ta_unload->ras_session_id != ta_ctx->session_id)) + return -EINVAL; + + return unload_ras_ta_firmware(ras_core, ras_ta_unload); +} + +int ras_psp_trigger_error(struct ras_core_context *ras_core, + struct ras_ta_trigger_error_input *info, uint32_t instance_mask) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + + if (!ta_ctx->preload_ras_ta_enabled && !ta_ctx->ras_ta_initialized) { + RAS_DEV_ERR(ras_core->dev, "RAS: ras firmware not initialized!"); + return -ENOEXEC; + } + + if (!info) + return -EINVAL; + + return trigger_ras_ta_error(ras_core, info, instance_mask); +} + +int ras_psp_query_address(struct ras_core_context *ras_core, + struct ras_ta_query_address_input *addr_in, + struct ras_ta_query_address_output *addr_out) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + + if (!ta_ctx->preload_ras_ta_enabled && + !ta_ctx->ras_ta_initialized) { + RAS_DEV_ERR(ras_core->dev, "RAS: ras firmware not initialized!"); + return -ENOEXEC; + } + + if (!addr_in || !addr_out) + return -EINVAL; + + return send_ras_ta_runtime_cmd(ras_core, RAS_TA_CMD_ID__QUERY_ADDRESS, + addr_in, sizeof(*addr_in), addr_out, sizeof(*addr_out)); +} + +int ras_psp_sw_init(struct ras_core_context *ras_core) +{ + struct ras_psp *psp = &ras_core->ras_psp; + + memset(psp, 0, sizeof(*psp)); + + psp->sys_func = ras_core->config->psp_cfg.psp_sys_fn; + if (!psp->sys_func) { + RAS_DEV_ERR(ras_core->dev, "RAS psp sys function not configured!\n"); + return -EINVAL; + } + + mutex_init(&psp->psp_ctx.internal_mutex); + mutex_init(&psp->ta_ctx.ta_mutex); + + return 0; +} + +int ras_psp_sw_fini(struct ras_core_context *ras_core) +{ + struct ras_psp *psp = &ras_core->ras_psp; + + mutex_destroy(&psp->psp_ctx.internal_mutex); + mutex_destroy(&psp->ta_ctx.ta_mutex); + + memset(psp, 0, sizeof(*psp)); + + return 0; +} + +int ras_psp_hw_init(struct ras_core_context *ras_core) +{ + struct ras_psp *psp = &ras_core->ras_psp; + + psp->psp_ip_version = ras_core->config->psp_ip_version; + + psp->ip_func = ras_psp_get_ip_funcs(ras_core, psp->psp_ip_version); + if (!psp->ip_func) + return -EINVAL; + + /* After GPU reset, the system RAS PSP status may change. + * therefore, it is necessary to synchronize the system status again. + */ + ras_psp_sync_system_ras_psp_status(ras_core); + + return 0; +} + +int ras_psp_hw_fini(struct ras_core_context *ras_core) +{ + return 0; +} + +bool ras_psp_check_supported_cmd(struct ras_core_context *ras_core, + enum ras_ta_cmd_id cmd_id) +{ + struct ras_ta_ctx *ta_ctx = &ras_core->ras_psp.ta_ctx; + bool ret = false; + + if (!ta_ctx->preload_ras_ta_enabled && !ta_ctx->ras_ta_initialized) + return false; + + switch (cmd_id) { + case RAS_TA_CMD_ID__QUERY_ADDRESS: + /* Currently, querying the address from RAS TA is only supported + * when the RAS TA firmware is loaded during driver installation. + */ + if (ta_ctx->preload_ras_ta_enabled) + ret = true; + break; + case RAS_TA_CMD_ID__TRIGGER_ERROR: + ret = true; + break; + default: + ret = false; + break; + } + + return ret; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_psp.h b/drivers/gpu/drm/amd/ras/rascore/ras_psp.h new file mode 100644 index 000000000000..71776fecfd66 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_psp.h @@ -0,0 +1,145 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_PSP_H__ +#define __RAS_PSP_H__ +#include "ras.h" +#include "ras_ta_if.h" + +struct ras_core_context; +struct ras_ta_trigger_error_input; +struct ras_ta_query_address_input; +struct ras_ta_query_address_output; +enum ras_ta_cmd_id; + +struct ras_ta_image_header { + uint32_t reserved1[24]; + uint32_t image_version; /* [0x60] Off Chip Firmware Version */ + uint32_t reserved2[39]; +}; + +struct ras_psp_sys_status { + bool initialized; + uint32_t session_id; + void *psp_cmd_mutex; +}; + +struct ras_ta_init_param { + uint8_t poison_mode_en; + uint8_t dgpu_mode; + uint16_t xcc_mask; + uint8_t channel_dis_num; + uint8_t nps_mode; + uint32_t active_umc_mask; +}; + +struct gpu_mem_block { + uint32_t mem_type; + void *mem_bo; + uint64_t mem_mc_addr; + void *mem_cpu_addr; + uint32_t mem_size; + int ref_count; + void *private; +}; + +struct ras_psp_ip_func { + uint32_t (*psp_ras_ring_wptr_get)(struct ras_core_context *ras_core); + int (*psp_ras_ring_wptr_set)(struct ras_core_context *ras_core, uint32_t wptr); +}; + +struct ras_psp_ring { + struct gpu_mem_block ras_ring_gpu_mem; +}; + +struct psp_cmd_resp { + uint32_t status; + uint32_t session_id; +}; + +struct ras_psp_ctx { + void *external_mutex; + struct mutex internal_mutex; + uint64_t in_fence_value; + struct gpu_mem_block psp_cmd_gpu_mem; + struct gpu_mem_block out_fence_gpu_mem; +}; + +struct ras_ta_fw_bin { + uint32_t fw_version; + uint32_t feature_version; + uint32_t bin_size; + uint8_t *bin_addr; +}; + +struct ras_ta_ctx { + bool preload_ras_ta_enabled; + bool ras_ta_initialized; + uint32_t session_id; + uint32_t resp_status; + uint32_t ta_version; + struct mutex ta_mutex; + struct ras_ta_fw_bin fw_bin; + struct ras_ta_init_param init_param; + struct gpu_mem_block fw_gpu_mem; + struct gpu_mem_block cmd_gpu_mem; +}; + +struct ras_psp { + uint32_t psp_ip_version; + struct ras_psp_ring psp_ring; + struct ras_psp_ctx psp_ctx; + struct ras_ta_ctx ta_ctx; + const struct ras_psp_ip_func *ip_func; + const struct ras_psp_sys_func *sys_func; +}; + +struct ras_psp_ta_load { + uint32_t fw_version; + uint32_t feature_version; + uint32_t bin_size; + uint8_t *bin_addr; + uint64_t out_session_id; + uint32_t out_loaded_ta_version; +}; + +struct ras_psp_ta_unload { + uint64_t ras_session_id; +}; + +int ras_psp_sw_init(struct ras_core_context *ras_core); +int ras_psp_sw_fini(struct ras_core_context *ras_core); +int ras_psp_hw_init(struct ras_core_context *ras_core); +int ras_psp_hw_fini(struct ras_core_context *ras_core); +int ras_psp_load_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_load *ras_ta_load); +int ras_psp_unload_firmware(struct ras_core_context *ras_core, + struct ras_psp_ta_unload *ras_ta_unload); +int ras_psp_trigger_error(struct ras_core_context *ras_core, + struct ras_ta_trigger_error_input *info, uint32_t instance_mask); +int ras_psp_query_address(struct ras_core_context *ras_core, + struct ras_ta_query_address_input *addr_in, + struct ras_ta_query_address_output *addr_out); +bool ras_psp_check_supported_cmd(struct ras_core_context *ras_core, + enum ras_ta_cmd_id cmd_id); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.c b/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.c new file mode 100644 index 000000000000..626cf39b75ac --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.c @@ -0,0 +1,46 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#include "ras.h" +#include "ras_psp_v13_0.h" + +#define regMP0_SMN_C2PMSG_67 0x0083 +#define regMP0_SMN_C2PMSG_67_BASE_IDX 0 + +static uint32_t ras_psp_v13_0_ring_wptr_get(struct ras_core_context *ras_core) +{ + return RAS_DEV_RREG32_SOC15(ras_core->dev, MP0, 0, regMP0_SMN_C2PMSG_67); +} + +static int ras_psp_v13_0_ring_wptr_set(struct ras_core_context *ras_core, uint32_t value) +{ + RAS_DEV_WREG32_SOC15(ras_core->dev, MP0, 0, regMP0_SMN_C2PMSG_67, value); + + return 0; +} + +const struct ras_psp_ip_func ras_psp_v13_0 = { + .psp_ras_ring_wptr_get = ras_psp_v13_0_ring_wptr_get, + .psp_ras_ring_wptr_set = ras_psp_v13_0_ring_wptr_set, +}; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.h b/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.h new file mode 100644 index 000000000000..b705ffe38a12 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_psp_v13_0.h @@ -0,0 +1,31 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_PSP_V13_0_H__ +#define __RAS_PSP_V13_0_H__ +#include "ras_psp.h" + +extern const struct ras_psp_ip_func ras_psp_v13_0; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_ta_if.h b/drivers/gpu/drm/amd/ras/rascore/ras_ta_if.h new file mode 100644 index 000000000000..0921e36d3274 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_ta_if.h @@ -0,0 +1,231 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef _RAS_TA_IF_H +#define _RAS_TA_IF_H +#include "ras.h" + +#define RAS_TA_HOST_IF_VER 0 + +/* Responses have bit 31 set */ +#define RSP_ID_MASK (1U << 31) +#define RSP_ID(cmdId) (((uint32_t)(cmdId)) | RSP_ID_MASK) + +/* invalid node instance value */ +#define RAS_TA_INV_NODE 0xffff + +/* RAS related enumerations */ +/**********************************************************/ +enum ras_ta_cmd_id { + RAS_TA_CMD_ID__ENABLE_FEATURES = 0, + RAS_TA_CMD_ID__DISABLE_FEATURES, + RAS_TA_CMD_ID__TRIGGER_ERROR, + RAS_TA_CMD_ID__QUERY_BLOCK_INFO, + RAS_TA_CMD_ID__QUERY_SUB_BLOCK_INFO, + RAS_TA_CMD_ID__QUERY_ADDRESS, + MAX_RAS_TA_CMD_ID +}; + +enum ras_ta_status { + RAS_TA_STATUS__SUCCESS = 0x0000, + RAS_TA_STATUS__RESET_NEEDED = 0xA001, + RAS_TA_STATUS__ERROR_INVALID_PARAMETER = 0xA002, + RAS_TA_STATUS__ERROR_RAS_NOT_AVAILABLE = 0xA003, + RAS_TA_STATUS__ERROR_RAS_DUPLICATE_CMD = 0xA004, + RAS_TA_STATUS__ERROR_INJECTION_FAILED = 0xA005, + RAS_TA_STATUS__ERROR_ASD_READ_WRITE = 0xA006, + RAS_TA_STATUS__ERROR_TOGGLE_DF_CSTATE = 0xA007, + RAS_TA_STATUS__ERROR_TIMEOUT = 0xA008, + RAS_TA_STATUS__ERROR_BLOCK_DISABLED = 0XA009, + RAS_TA_STATUS__ERROR_GENERIC = 0xA00A, + RAS_TA_STATUS__ERROR_RAS_MMHUB_INIT = 0xA00B, + RAS_TA_STATUS__ERROR_GET_DEV_INFO = 0xA00C, + RAS_TA_STATUS__ERROR_UNSUPPORTED_DEV = 0xA00D, + RAS_TA_STATUS__ERROR_NOT_INITIALIZED = 0xA00E, + RAS_TA_STATUS__ERROR_TEE_INTERNAL = 0xA00F, + RAS_TA_STATUS__ERROR_UNSUPPORTED_FUNCTION = 0xA010, + RAS_TA_STATUS__ERROR_SYS_DRV_REG_ACCESS = 0xA011, + RAS_TA_STATUS__ERROR_RAS_READ_WRITE = 0xA012, + RAS_TA_STATUS__ERROR_NULL_PTR = 0xA013, + RAS_TA_STATUS__ERROR_UNSUPPORTED_IP = 0xA014, + RAS_TA_STATUS__ERROR_PCS_STATE_QUIET = 0xA015, + RAS_TA_STATUS__ERROR_PCS_STATE_ERROR = 0xA016, + RAS_TA_STATUS__ERROR_PCS_STATE_HANG = 0xA017, + RAS_TA_STATUS__ERROR_PCS_STATE_UNKNOWN = 0xA018, + RAS_TA_STATUS__ERROR_UNSUPPORTED_ERROR_INJ = 0xA019, + RAS_TA_STATUS__TEE_ERROR_ACCESS_DENIED = 0xA01A +}; + +enum ras_ta_block { + RAS_TA_BLOCK__UMC = 0, + RAS_TA_BLOCK__SDMA, + RAS_TA_BLOCK__GFX, + RAS_TA_BLOCK__MMHUB, + RAS_TA_BLOCK__ATHUB, + RAS_TA_BLOCK__PCIE_BIF, + RAS_TA_BLOCK__HDP, + RAS_TA_BLOCK__XGMI_WAFL, + RAS_TA_BLOCK__DF, + RAS_TA_BLOCK__SMN, + RAS_TA_BLOCK__SEM, + RAS_TA_BLOCK__MP0, + RAS_TA_BLOCK__MP1, + RAS_TA_BLOCK__FUSE, + RAS_TA_BLOCK__MCA, + RAS_TA_BLOCK__VCN, + RAS_TA_BLOCK__JPEG, + RAS_TA_BLOCK__IH, + RAS_TA_BLOCK__MPIO, + RAS_TA_BLOCK__MMSCH, + RAS_TA_NUM_BLOCK_MAX +}; + +enum ras_ta_mca_block { + RAS_TA_MCA_BLOCK__MP0 = 0, + RAS_TA_MCA_BLOCK__MP1 = 1, + RAS_TA_MCA_BLOCK__MPIO = 2, + RAS_TA_MCA_BLOCK__IOHC = 3, + RAS_TA_MCA_NUM_BLOCK_MAX +}; + +enum ras_ta_error_type { + RAS_TA_ERROR__NONE = 0, + RAS_TA_ERROR__PARITY = 1, + RAS_TA_ERROR__SINGLE_CORRECTABLE = 2, + RAS_TA_ERROR__MULTI_UNCORRECTABLE = 4, + RAS_TA_ERROR__POISON = 8, +}; + +enum ras_ta_address_type { + RAS_TA_MCA_TO_PA, + RAS_TA_PA_TO_MCA, +}; + +enum ras_ta_nps_mode { + RAS_TA_UNKNOWN_MODE = 0, + RAS_TA_NPS1_MODE = 1, + RAS_TA_NPS2_MODE = 2, + RAS_TA_NPS4_MODE = 4, + RAS_TA_NPS8_MODE = 8, +}; + +/* Input/output structures for RAS commands */ +/**********************************************************/ + +struct ras_ta_enable_features_input { + enum ras_ta_block block_id; + enum ras_ta_error_type error_type; +}; + +struct ras_ta_disable_features_input { + enum ras_ta_block block_id; + enum ras_ta_error_type error_type; +}; + +struct ras_ta_trigger_error_input { + /* ras-block. i.e. umc, gfx */ + enum ras_ta_block block_id; + + /* type of error. i.e. single_correctable */ + enum ras_ta_error_type inject_error_type; + + /* mem block. i.e. hbm, sram etc. */ + uint32_t sub_block_index; + + /* explicit address of error */ + uint64_t address; + + /* method if error injection. i.e persistent, coherent etc. */ + uint64_t value; +}; + +struct ras_ta_init_flags { + uint8_t poison_mode_en; + uint8_t dgpu_mode; + uint16_t xcc_mask; + uint8_t channel_dis_num; + uint8_t nps_mode; + uint32_t active_umc_mask; +}; + +struct ras_ta_mca_addr { + uint64_t err_addr; + uint32_t ch_inst; + uint32_t umc_inst; + uint32_t node_inst; + uint32_t socket_id; +}; + +struct ras_ta_phy_addr { + uint64_t pa; + uint32_t bank; + uint32_t channel_idx; +}; + +struct ras_ta_query_address_input { + enum ras_ta_address_type addr_type; + struct ras_ta_mca_addr ma; + struct ras_ta_phy_addr pa; +}; + +struct ras_ta_output_flags { + uint8_t ras_init_success_flag; + uint8_t err_inject_switch_disable_flag; + uint8_t reg_access_failure_flag; +}; + +struct ras_ta_query_address_output { + /* don't use the flags here */ + struct ras_ta_output_flags flags; + struct ras_ta_mca_addr ma; + struct ras_ta_phy_addr pa; +}; + +/* Common input structure for RAS callbacks */ +/**********************************************************/ +union ras_ta_cmd_input { + struct ras_ta_init_flags init_flags; + struct ras_ta_enable_features_input enable_features; + struct ras_ta_disable_features_input disable_features; + struct ras_ta_trigger_error_input trigger_error; + struct ras_ta_query_address_input address; + uint32_t reserve_pad[256]; +}; + +union ras_ta_cmd_output { + struct ras_ta_output_flags flags; + struct ras_ta_query_address_output address; + uint32_t reserve_pad[256]; +}; + +struct ras_ta_cmd { + uint32_t cmd_id; + uint32_t resp_id; + uint32_t ras_status; + uint32_t if_version; + union ras_ta_cmd_input ras_in_message; + union ras_ta_cmd_output ras_out_message; +}; + +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_umc.c b/drivers/gpu/drm/amd/ras/rascore/ras_umc.c new file mode 100644 index 000000000000..4dae64c424a2 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_umc.c @@ -0,0 +1,707 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_umc.h" +#include "ras_umc_v12_0.h" + +#define MAX_ECC_NUM_PER_RETIREMENT 16 + +/* bad page timestamp format + * yy[31:27] mm[26:23] day[22:17] hh[16:12] mm[11:6] ss[5:0] + */ +#define EEPROM_TIMESTAMP_MINUTE 6 +#define EEPROM_TIMESTAMP_HOUR 12 +#define EEPROM_TIMESTAMP_DAY 17 +#define EEPROM_TIMESTAMP_MONTH 23 +#define EEPROM_TIMESTAMP_YEAR 27 + +static uint64_t ras_umc_get_eeprom_timestamp(struct ras_core_context *ras_core) +{ + struct ras_time tm = {0}; + uint64_t utc_timestamp = 0; + uint64_t eeprom_timestamp = 0; + + utc_timestamp = ras_core_get_utc_second_timestamp(ras_core); + if (!utc_timestamp) + return utc_timestamp; + + ras_core_convert_timestamp_to_time(ras_core, utc_timestamp, &tm); + + /* the year range is 2000 ~ 2031, set the year if not in the range */ + if (tm.tm_year < 2000) + tm.tm_year = 2000; + if (tm.tm_year > 2031) + tm.tm_year = 2031; + + tm.tm_year -= 2000; + + eeprom_timestamp = tm.tm_sec + (tm.tm_min << EEPROM_TIMESTAMP_MINUTE) + + (tm.tm_hour << EEPROM_TIMESTAMP_HOUR) + + (tm.tm_mday << EEPROM_TIMESTAMP_DAY) + + (tm.tm_mon << EEPROM_TIMESTAMP_MONTH) + + (tm.tm_year << EEPROM_TIMESTAMP_YEAR); + eeprom_timestamp &= 0xffffffff; + + return eeprom_timestamp; +} + +static const struct ras_umc_ip_func *ras_umc_get_ip_func( + struct ras_core_context *ras_core, uint32_t ip_version) +{ + switch (ip_version) { + case IP_VERSION(12, 0, 0): + case IP_VERSION(12, 5, 0): + return &ras_umc_func_v12_0; + default: + RAS_DEV_ERR(ras_core->dev, + "UMC ip version(0x%x) is not supported!\n", ip_version); + break; + } + + return NULL; +} + +int ras_umc_psp_convert_ma_to_pa(struct ras_core_context *ras_core, + struct umc_mca_addr *in, struct umc_phy_addr *out, + uint32_t nps) +{ + struct ras_ta_query_address_input addr_in; + struct ras_ta_query_address_output addr_out; + int ret; + + if (!in) + return -EINVAL; + + memset(&addr_in, 0, sizeof(addr_in)); + memset(&addr_out, 0, sizeof(addr_out)); + + addr_in.ma.err_addr = in->err_addr; + addr_in.ma.ch_inst = in->ch_inst; + addr_in.ma.umc_inst = in->umc_inst; + addr_in.ma.node_inst = in->node_inst; + addr_in.ma.socket_id = in->socket_id; + + addr_in.addr_type = RAS_TA_MCA_TO_PA; + + ret = ras_psp_query_address(ras_core, &addr_in, &addr_out); + if (ret) { + RAS_DEV_WARN(ras_core->dev, + "Failed to query RAS physical address for 0x%llx, ret:%d", + in->err_addr, ret); + return -EREMOTEIO; + } + + if (out) { + out->pa = addr_out.pa.pa; + out->bank = addr_out.pa.bank; + out->channel_idx = addr_out.pa.channel_idx; + } + + return 0; +} + +static int ras_umc_log_ecc(struct ras_core_context *ras_core, + unsigned long idx, void *data) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + int ret; + + mutex_lock(&ras_umc->tree_lock); + ret = radix_tree_insert(&ras_umc->root, idx, data); + if (!ret) + radix_tree_tag_set(&ras_umc->root, idx, UMC_ECC_NEW_DETECTED_TAG); + mutex_unlock(&ras_umc->tree_lock); + + return ret; +} + +int ras_umc_clear_logged_ecc(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + uint64_t buf[8] = {0}; + void **slot; + void *data; + void *iter = buf; + + mutex_lock(&ras_umc->tree_lock); + radix_tree_for_each_slot(slot, &ras_umc->root, iter, 0) { + data = ras_radix_tree_delete_iter(&ras_umc->root, iter); + kfree(data); + } + mutex_unlock(&ras_umc->tree_lock); + + return 0; +} + +static void ras_umc_reserve_eeprom_record(struct ras_core_context *ras_core, + struct eeprom_umc_record *record) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + uint64_t page_pfn[16]; + int count = 0, i; + + memset(page_pfn, 0, sizeof(page_pfn)); + if (ras_umc->ip_func && ras_umc->ip_func->eeprom_record_to_nps_pages) { + count = ras_umc->ip_func->eeprom_record_to_nps_pages(ras_core, + record, record->cur_nps, page_pfn, ARRAY_SIZE(page_pfn)); + if (count <= 0) { + RAS_DEV_ERR(ras_core->dev, + "Fail to convert error address! count:%d\n", count); + return; + } + } + + /* Reserve memory */ + for (i = 0; i < count; i++) + ras_core_event_notify(ras_core, + RAS_EVENT_ID__RESERVE_BAD_PAGE, &page_pfn[i]); +} + +/* When gpu reset is ongoing, ecc logging operations will be pended. + */ +int ras_umc_log_bad_bank_pending(struct ras_core_context *ras_core, struct ras_bank_ecc *bank) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct ras_bank_ecc_node *ecc_node; + + ecc_node = kzalloc(sizeof(*ecc_node), GFP_KERNEL); + if (!ecc_node) + return -ENOMEM; + + memcpy(&ecc_node->ecc, bank, sizeof(ecc_node->ecc)); + + mutex_lock(&ras_umc->pending_ecc_lock); + list_add_tail(&ecc_node->node, &ras_umc->pending_ecc_list); + mutex_unlock(&ras_umc->pending_ecc_lock); + + return 0; +} + +/* After gpu reset is complete, re-log the pending error banks. + */ +int ras_umc_log_pending_bad_bank(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct ras_bank_ecc_node *ecc_node, *tmp; + + mutex_lock(&ras_umc->pending_ecc_lock); + list_for_each_entry_safe(ecc_node, + tmp, &ras_umc->pending_ecc_list, node){ + if (ecc_node && !ras_umc_log_bad_bank(ras_core, &ecc_node->ecc)) { + list_del(&ecc_node->node); + kfree(ecc_node); + } + } + mutex_unlock(&ras_umc->pending_ecc_lock); + + return 0; +} + +int ras_umc_log_bad_bank(struct ras_core_context *ras_core, struct ras_bank_ecc *bank) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_umc_record umc_rec; + struct eeprom_umc_record *err_rec; + int ret; + + memset(&umc_rec, 0, sizeof(umc_rec)); + + mutex_lock(&ras_umc->bank_log_lock); + ret = ras_umc->ip_func->bank_to_eeprom_record(ras_core, bank, &umc_rec); + if (ret) + goto out; + + err_rec = kzalloc(sizeof(*err_rec), GFP_KERNEL); + if (!err_rec) { + ret = -ENOMEM; + goto out; + } + + memcpy(err_rec, &umc_rec, sizeof(umc_rec)); + ret = ras_umc_log_ecc(ras_core, err_rec->cur_nps_retired_row_pfn, err_rec); + if (ret) { + if (ret == -EEXIST) { + RAS_DEV_INFO(ras_core->dev, "The bad pages have been logged before.\n"); + ret = 0; + } + + kfree(err_rec); + goto out; + } + + ras_umc_reserve_eeprom_record(ras_core, err_rec); + + ret = ras_core_event_notify(ras_core, + RAS_EVENT_ID__BAD_PAGE_DETECTED, NULL); + +out: + mutex_unlock(&ras_umc->bank_log_lock); + return ret; +} + +static int ras_umc_get_new_records(struct ras_core_context *ras_core, + struct eeprom_umc_record *records, u32 num) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_umc_record *entries[MAX_ECC_NUM_PER_RETIREMENT]; + u32 entry_num = num < MAX_ECC_NUM_PER_RETIREMENT ? num : MAX_ECC_NUM_PER_RETIREMENT; + int count = 0; + int new_detected, i; + + mutex_lock(&ras_umc->tree_lock); + new_detected = radix_tree_gang_lookup_tag(&ras_umc->root, (void **)entries, + 0, entry_num, UMC_ECC_NEW_DETECTED_TAG); + for (i = 0; i < new_detected; i++) { + if (!entries[i]) + continue; + + memcpy(&records[i], entries[i], sizeof(struct eeprom_umc_record)); + count++; + radix_tree_tag_clear(&ras_umc->root, + entries[i]->cur_nps_retired_row_pfn, UMC_ECC_NEW_DETECTED_TAG); + } + mutex_unlock(&ras_umc->tree_lock); + + return count; +} + +static bool ras_umc_check_retired_record(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, bool from_eeprom) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_store_record *data = &ras_umc->umc_err_data.rom_data; + uint32_t nps = 0; + int i, ret; + + if (from_eeprom) { + nps = ras_umc->umc_err_data.umc_nps_mode; + if (ras_umc->ip_func && ras_umc->ip_func->eeprom_record_to_nps_record) { + ret = ras_umc->ip_func->eeprom_record_to_nps_record(ras_core, record, nps); + if (ret) + RAS_DEV_WARN(ras_core->dev, + "Failed to adjust eeprom record, ret:%d", ret); + } + return false; + } + + for (i = 0; i < data->count; i++) { + if ((data->bps[i].retired_row_pfn == record->retired_row_pfn) && + (data->bps[i].cur_nps_retired_row_pfn == record->cur_nps_retired_row_pfn)) + return true; + } + + return false; +} + +/* alloc/realloc bps array */ +static int ras_umc_realloc_err_data_space(struct ras_core_context *ras_core, + struct eeprom_store_record *data, int pages) +{ + unsigned int old_space = data->count + data->space_left; + unsigned int new_space = old_space + pages; + unsigned int align_space = ALIGN(new_space, 512); + void *bps = kzalloc(align_space * sizeof(*data->bps), GFP_KERNEL); + + if (!bps) + return -ENOMEM; + + if (data->bps) { + memcpy(bps, data->bps, + data->count * sizeof(*data->bps)); + kfree(data->bps); + } + + data->bps = bps; + data->space_left += align_space - old_space; + return 0; +} + +static int ras_umc_update_eeprom_rom_data(struct ras_core_context *ras_core, + struct eeprom_umc_record *bps) +{ + struct eeprom_store_record *data = &ras_core->ras_umc.umc_err_data.rom_data; + + if (!data->space_left && + ras_umc_realloc_err_data_space(ras_core, data, 256)) { + return -ENOMEM; + } + + memcpy(&data->bps[data->count], bps, sizeof(*data->bps)); + data->count++; + data->space_left--; + return 0; +} + +static int ras_umc_update_eeprom_ram_data(struct ras_core_context *ras_core, + struct eeprom_umc_record *bps) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_store_record *data = &ras_umc->umc_err_data.ram_data; + uint64_t page_pfn[16]; + int count = 0, j; + + if (!data->space_left && + ras_umc_realloc_err_data_space(ras_core, data, 256)) { + return -ENOMEM; + } + + memset(page_pfn, 0, sizeof(page_pfn)); + if (ras_umc->ip_func && ras_umc->ip_func->eeprom_record_to_nps_pages) + count = ras_umc->ip_func->eeprom_record_to_nps_pages(ras_core, + bps, bps->cur_nps, page_pfn, ARRAY_SIZE(page_pfn)); + + if (count > 0) { + for (j = 0; j < count; j++) { + bps->cur_nps_retired_row_pfn = page_pfn[j]; + memcpy(&data->bps[data->count], bps, sizeof(*data->bps)); + data->count++; + data->space_left--; + } + } else { + memcpy(&data->bps[data->count], bps, sizeof(*data->bps)); + data->count++; + data->space_left--; + } + + return 0; +} + +/* it deal with vram only. */ +static int ras_umc_add_bad_pages(struct ras_core_context *ras_core, + struct eeprom_umc_record *bps, + int pages, bool from_eeprom) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct ras_umc_err_data *data = &ras_umc->umc_err_data; + int i, ret = 0; + + if (!bps || pages <= 0) + return 0; + + mutex_lock(&ras_umc->umc_lock); + for (i = 0; i < pages; i++) { + if (ras_umc_check_retired_record(ras_core, &bps[i], from_eeprom)) + continue; + + ret = ras_umc_update_eeprom_rom_data(ras_core, &bps[i]); + if (ret) + goto out; + + if (data->last_retired_pfn == bps[i].cur_nps_retired_row_pfn) + continue; + + data->last_retired_pfn = bps[i].cur_nps_retired_row_pfn; + + if (from_eeprom) + ras_umc_reserve_eeprom_record(ras_core, &bps[i]); + + ret = ras_umc_update_eeprom_ram_data(ras_core, &bps[i]); + if (ret) + goto out; + } +out: + mutex_unlock(&ras_umc->umc_lock); + + return ret; +} + +/* + * read error record array in eeprom and reserve enough space for + * storing new bad pages + */ +int ras_umc_load_bad_pages(struct ras_core_context *ras_core) +{ + struct eeprom_umc_record *bps; + uint32_t ras_num_recs; + int ret; + + ras_num_recs = ras_eeprom_get_record_count(ras_core); + /* no bad page record, skip eeprom access */ + if (!ras_num_recs || + ras_core->ras_eeprom.record_threshold_config == DISABLE_RETIRE_PAGE) + return 0; + + bps = kcalloc(ras_num_recs, sizeof(*bps), GFP_KERNEL); + if (!bps) + return -ENOMEM; + + ret = ras_eeprom_read(ras_core, bps, ras_num_recs); + if (ret) { + RAS_DEV_ERR(ras_core->dev, "Failed to load EEPROM table records!"); + } else { + ras_core->ras_umc.umc_err_data.last_retired_pfn = UMC_INV_MEM_PFN; + ret = ras_umc_add_bad_pages(ras_core, bps, ras_num_recs, true); + } + + kfree(bps); + return ret; +} + +/* + * write error record array to eeprom, the function should be + * protected by recovery_lock + * new_cnt: new added UE count, excluding reserved bad pages, can be NULL + */ +static int ras_umc_save_bad_pages(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_store_record *data = &ras_umc->umc_err_data.rom_data; + uint32_t eeprom_record_num; + int save_count; + int ret = 0; + + if (!data->bps) + return 0; + + eeprom_record_num = ras_eeprom_get_record_count(ras_core); + mutex_lock(&ras_umc->umc_lock); + save_count = data->count - eeprom_record_num; + /* only new entries are saved */ + if (save_count > 0) { + if (ras_eeprom_append(ras_core, + &data->bps[eeprom_record_num], + save_count)) { + RAS_DEV_ERR(ras_core->dev, "Failed to save EEPROM table data!"); + ret = -EIO; + goto exit; + } + + RAS_DEV_INFO(ras_core->dev, "Saved %d pages to EEPROM table.\n", save_count); + } + +exit: + mutex_unlock(&ras_umc->umc_lock); + return ret; +} + +int ras_umc_handle_bad_pages(struct ras_core_context *ras_core, void *data) +{ + struct eeprom_umc_record records[MAX_ECC_NUM_PER_RETIREMENT]; + int count, ret; + + memset(records, 0, sizeof(records)); + count = ras_umc_get_new_records(ras_core, records, ARRAY_SIZE(records)); + if (count <= 0) + return -ENODATA; + + ret = ras_umc_add_bad_pages(ras_core, records, count, false); + if (ret) { + RAS_DEV_ERR(ras_core->dev, "Failed to add ras bad page!\n"); + return -EINVAL; + } + + ret = ras_umc_save_bad_pages(ras_core); + if (ret) { + RAS_DEV_ERR(ras_core->dev, "Failed to save ras bad page\n"); + return -EINVAL; + } + + return 0; +} + +int ras_umc_sw_init(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + + memset(ras_umc, 0, sizeof(*ras_umc)); + + INIT_LIST_HEAD(&ras_umc->pending_ecc_list); + + INIT_RADIX_TREE(&ras_umc->root, GFP_KERNEL); + + mutex_init(&ras_umc->tree_lock); + mutex_init(&ras_umc->pending_ecc_lock); + mutex_init(&ras_umc->umc_lock); + mutex_init(&ras_umc->bank_log_lock); + + return 0; +} + +int ras_umc_sw_fini(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct ras_umc_err_data *umc_err_data = &ras_umc->umc_err_data; + struct ras_bank_ecc_node *ecc_node, *tmp; + + mutex_destroy(&ras_umc->umc_lock); + mutex_destroy(&ras_umc->bank_log_lock); + + if (umc_err_data->rom_data.bps) { + umc_err_data->rom_data.count = 0; + kfree(umc_err_data->rom_data.bps); + umc_err_data->rom_data.bps = NULL; + umc_err_data->rom_data.space_left = 0; + } + + if (umc_err_data->ram_data.bps) { + umc_err_data->ram_data.count = 0; + kfree(umc_err_data->ram_data.bps); + umc_err_data->ram_data.bps = NULL; + umc_err_data->ram_data.space_left = 0; + } + + ras_umc_clear_logged_ecc(ras_core); + + mutex_lock(&ras_umc->pending_ecc_lock); + list_for_each_entry_safe(ecc_node, + tmp, &ras_umc->pending_ecc_list, node){ + list_del(&ecc_node->node); + kfree(ecc_node); + } + mutex_unlock(&ras_umc->pending_ecc_lock); + + mutex_destroy(&ras_umc->tree_lock); + mutex_destroy(&ras_umc->pending_ecc_lock); + + return 0; +} + +int ras_umc_hw_init(struct ras_core_context *ras_core) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + uint32_t nps; + + nps = ras_core_get_curr_nps_mode(ras_core); + + if (!nps || (nps >= UMC_MEMORY_PARTITION_MODE_UNKNOWN)) { + RAS_DEV_ERR(ras_core->dev, "Invalid memory NPS mode: %u!\n", nps); + return -ENODATA; + } + + ras_umc->umc_err_data.umc_nps_mode = nps; + + ras_umc->umc_vram_type = ras_core->config->umc_cfg.umc_vram_type; + if (!ras_umc->umc_vram_type) { + RAS_DEV_ERR(ras_core->dev, "Invalid UMC VRAM Type: %u!\n", + ras_umc->umc_vram_type); + return -ENODATA; + } + + ras_umc->umc_ip_version = ras_core->config->umc_ip_version; + ras_umc->ip_func = ras_umc_get_ip_func(ras_core, ras_umc->umc_ip_version); + if (!ras_umc->ip_func) + return -EINVAL; + + return 0; +} + +int ras_umc_hw_fini(struct ras_core_context *ras_core) +{ + return 0; +} + +int ras_umc_clean_badpage_data(struct ras_core_context *ras_core) +{ + struct ras_umc_err_data *data = &ras_core->ras_umc.umc_err_data; + + mutex_lock(&ras_core->ras_umc.umc_lock); + + kfree(data->rom_data.bps); + kfree(data->ram_data.bps); + + memset(data, 0, sizeof(*data)); + mutex_unlock(&ras_core->ras_umc.umc_lock); + + return 0; +} + +int ras_umc_fill_eeprom_record(struct ras_core_context *ras_core, + uint64_t err_addr, uint32_t umc_inst, struct umc_phy_addr *cur_nps_addr, + enum umc_memory_partition_mode cur_nps, struct eeprom_umc_record *record) +{ + struct eeprom_umc_record *err_rec = record; + + /* Set bad page pfn and nps mode */ + EEPROM_RECORD_SETUP_UMC_ADDR_AND_NPS(err_rec, + RAS_ADDR_TO_PFN(cur_nps_addr->pa), cur_nps); + + err_rec->address = err_addr; + err_rec->ts = ras_umc_get_eeprom_timestamp(ras_core); + err_rec->err_type = RAS_EEPROM_ERR_NON_RECOVERABLE; + err_rec->cu = 0; + err_rec->mem_channel = cur_nps_addr->channel_idx; + err_rec->mcumc_id = umc_inst; + err_rec->cur_nps_retired_row_pfn = RAS_ADDR_TO_PFN(cur_nps_addr->pa); + err_rec->cur_nps_bank = cur_nps_addr->bank; + err_rec->cur_nps = cur_nps; + return 0; +} + +int ras_umc_get_saved_eeprom_count(struct ras_core_context *ras_core) +{ + struct ras_umc_err_data *err_data = &ras_core->ras_umc.umc_err_data; + + return err_data->rom_data.count; +} + +int ras_umc_get_badpage_count(struct ras_core_context *ras_core) +{ + struct eeprom_store_record *data = &ras_core->ras_umc.umc_err_data.ram_data; + + return data->count; +} + +int ras_umc_get_badpage_record(struct ras_core_context *ras_core, uint32_t index, void *record) +{ + struct eeprom_store_record *data = &ras_core->ras_umc.umc_err_data.ram_data; + + if (index >= data->count) + return -EINVAL; + + memcpy(record, &data->bps[index], sizeof(struct eeprom_umc_record)); + return 0; +} + +bool ras_umc_check_retired_addr(struct ras_core_context *ras_core, uint64_t addr) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + struct eeprom_store_record *data = &ras_umc->umc_err_data.ram_data; + uint64_t page_pfn = RAS_ADDR_TO_PFN(addr); + int i, ret = false; + + mutex_lock(&ras_umc->umc_lock); + for (i = 0; i < data->count; i++) { + if (data->bps[i].cur_nps_retired_row_pfn == page_pfn) { + ret = true; + break; + } + } + mutex_unlock(&ras_umc->umc_lock); + + return ret; +} + +int ras_umc_translate_soc_pa_and_bank(struct ras_core_context *ras_core, + uint64_t *soc_pa, struct umc_bank_addr *bank_addr, bool bank_to_pa) +{ + struct ras_umc *ras_umc = &ras_core->ras_umc; + int ret = 0; + + if (bank_to_pa) + ret = ras_umc->ip_func->bank_to_soc_pa(ras_core, *bank_addr, soc_pa); + else + ret = ras_umc->ip_func->soc_pa_to_bank(ras_core, *soc_pa, bank_addr); + + return ret; +} diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_umc.h b/drivers/gpu/drm/amd/ras/rascore/ras_umc.h new file mode 100644 index 000000000000..7d9e779d8c4c --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_umc.h @@ -0,0 +1,166 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ + +#ifndef __RAS_UMC_H__ +#define __RAS_UMC_H__ +#include "ras.h" +#include "ras_eeprom.h" +#include "ras_cmd.h" + +#define UMC_VRAM_TYPE_UNKNOWN 0 +#define UMC_VRAM_TYPE_GDDR1 1 +#define UMC_VRAM_TYPE_DDR2 2 +#define UMC_VRAM_TYPE_GDDR3 3 +#define UMC_VRAM_TYPE_GDDR4 4 +#define UMC_VRAM_TYPE_GDDR5 5 +#define UMC_VRAM_TYPE_HBM 6 +#define UMC_VRAM_TYPE_DDR3 7 +#define UMC_VRAM_TYPE_DDR4 8 +#define UMC_VRAM_TYPE_GDDR6 9 +#define UMC_VRAM_TYPE_DDR5 10 +#define UMC_VRAM_TYPE_LPDDR4 11 +#define UMC_VRAM_TYPE_LPDDR5 12 +#define UMC_VRAM_TYPE_HBM3E 13 + +#define UMC_ECC_NEW_DETECTED_TAG 0x1 +#define UMC_INV_MEM_PFN (0xFFFFFFFFFFFFFFFF) + +/* three column bits and one row bit in MCA address flip + * in bad page retirement + */ +#define UMC_PA_FLIP_BITS_NUM 4 + +enum umc_memory_partition_mode { + UMC_MEMORY_PARTITION_MODE_NONE = 0, + UMC_MEMORY_PARTITION_MODE_NPS1 = 1, + UMC_MEMORY_PARTITION_MODE_NPS2 = 2, + UMC_MEMORY_PARTITION_MODE_NPS3 = 3, + UMC_MEMORY_PARTITION_MODE_NPS4 = 4, + UMC_MEMORY_PARTITION_MODE_NPS6 = 6, + UMC_MEMORY_PARTITION_MODE_NPS8 = 8, + UMC_MEMORY_PARTITION_MODE_UNKNOWN +}; + +struct ras_core_context; +struct ras_bank_ecc; + +struct umc_flip_bits { + uint32_t flip_bits_in_pa[UMC_PA_FLIP_BITS_NUM]; + uint32_t flip_row_bit; + uint32_t r13_in_pa; + uint32_t bit_num; +}; + +struct umc_mca_addr { + uint64_t err_addr; + uint32_t ch_inst; + uint32_t umc_inst; + uint32_t node_inst; + uint32_t socket_id; +}; + +struct umc_phy_addr { + uint64_t pa; + uint32_t bank; + uint32_t channel_idx; +}; + +struct umc_bank_addr { + uint32_t stack_id; /* SID */ + uint32_t bank_group; + uint32_t bank; + uint32_t row; + uint32_t column; + uint32_t channel; + uint32_t subchannel; /* Also called Pseudochannel (PC) */ +}; + +struct ras_umc_ip_func { + int (*bank_to_eeprom_record)(struct ras_core_context *ras_core, + struct ras_bank_ecc *bank, struct eeprom_umc_record *record); + int (*eeprom_record_to_nps_record)(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint32_t nps); + int (*eeprom_record_to_nps_pages)(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint32_t nps, + uint64_t *pfns, uint32_t num); + int (*bank_to_soc_pa)(struct ras_core_context *ras_core, + struct umc_bank_addr bank_addr, uint64_t *soc_pa); + int (*soc_pa_to_bank)(struct ras_core_context *ras_core, + uint64_t soc_pa, struct umc_bank_addr *bank_addr); +}; + +struct eeprom_store_record { + /* point to data records array */ + struct eeprom_umc_record *bps; + /* the count of entries */ + int count; + /* the space can place new entries */ + int space_left; +}; + +struct ras_umc_err_data { + struct eeprom_store_record rom_data; + struct eeprom_store_record ram_data; + enum umc_memory_partition_mode umc_nps_mode; + uint64_t last_retired_pfn; +}; + +struct ras_umc { + u32 umc_ip_version; + u32 umc_vram_type; + const struct ras_umc_ip_func *ip_func; + struct radix_tree_root root; + struct mutex tree_lock; + struct mutex umc_lock; + struct mutex bank_log_lock; + struct mutex pending_ecc_lock; + struct ras_umc_err_data umc_err_data; + struct list_head pending_ecc_list; +}; + +int ras_umc_sw_init(struct ras_core_context *ras); +int ras_umc_sw_fini(struct ras_core_context *ras); +int ras_umc_hw_init(struct ras_core_context *ras); +int ras_umc_hw_fini(struct ras_core_context *ras); +int ras_umc_psp_convert_ma_to_pa(struct ras_core_context *ras_core, + struct umc_mca_addr *in, struct umc_phy_addr *out, + uint32_t nps); +int ras_umc_handle_bad_pages(struct ras_core_context *ras_core, void *data); +int ras_umc_log_bad_bank(struct ras_core_context *ras, struct ras_bank_ecc *bank); +int ras_umc_log_bad_bank_pending(struct ras_core_context *ras_core, struct ras_bank_ecc *bank); +int ras_umc_log_pending_bad_bank(struct ras_core_context *ras_core); +int ras_umc_clear_logged_ecc(struct ras_core_context *ras_core); +int ras_umc_load_bad_pages(struct ras_core_context *ras_core); +int ras_umc_get_saved_eeprom_count(struct ras_core_context *ras_core); +int ras_umc_clean_badpage_data(struct ras_core_context *ras_core); +int ras_umc_fill_eeprom_record(struct ras_core_context *ras_core, + uint64_t err_addr, uint32_t umc_inst, struct umc_phy_addr *cur_nps_addr, + enum umc_memory_partition_mode cur_nps, struct eeprom_umc_record *record); + +int ras_umc_get_badpage_count(struct ras_core_context *ras_core); +int ras_umc_get_badpage_record(struct ras_core_context *ras_core, uint32_t index, void *record); +bool ras_umc_check_retired_addr(struct ras_core_context *ras_core, uint64_t addr); +int ras_umc_translate_soc_pa_and_bank(struct ras_core_context *ras_core, + uint64_t *soc_pa, struct umc_bank_addr *bank_addr, bool bank_to_pa); +#endif diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.c b/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.c new file mode 100644 index 000000000000..5d9a11c17a86 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.c @@ -0,0 +1,511 @@ +// SPDX-License-Identifier: MIT +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#include "ras.h" +#include "ras_umc.h" +#include "ras_core_status.h" +#include "ras_umc_v12_0.h" + +#define NumDieInterleaved 4 + +static const uint32_t umc_v12_0_channel_idx_tbl[] + [UMC_V12_0_UMC_INSTANCE_NUM][UMC_V12_0_CHANNEL_INSTANCE_NUM] = { + {{3, 7, 11, 15, 2, 6, 10, 14}, {1, 5, 9, 13, 0, 4, 8, 12}, + {19, 23, 27, 31, 18, 22, 26, 30}, {17, 21, 25, 29, 16, 20, 24, 28}}, + {{47, 43, 39, 35, 46, 42, 38, 34}, {45, 41, 37, 33, 44, 40, 36, 32}, + {63, 59, 55, 51, 62, 58, 54, 50}, {61, 57, 53, 49, 60, 56, 52, 48}}, + {{79, 75, 71, 67, 78, 74, 70, 66}, {77, 73, 69, 65, 76, 72, 68, 64}, + {95, 91, 87, 83, 94, 90, 86, 82}, {93, 89, 85, 81, 92, 88, 84, 80}}, + {{99, 103, 107, 111, 98, 102, 106, 110}, {97, 101, 105, 109, 96, 100, 104, 108}, + {115, 119, 123, 127, 114, 118, 122, 126}, {113, 117, 121, 125, 112, 116, 120, 124}} +}; + +/* mapping of MCA error address to normalized address */ +static const uint32_t umc_v12_0_ma2na_mapping[] = { + 0, 5, 6, 8, 9, 14, 12, 13, + 10, 11, 15, 16, 17, 18, 19, 20, + 21, 22, 23, 24, 25, 26, 27, 28, + 24, 7, 29, 30, +}; + +static bool umc_v12_0_bit_wise_xor(uint32_t val) +{ + bool result = 0; + int i; + + for (i = 0; i < 32; i++) + result = result ^ ((val >> i) & 0x1); + + return result; +} + +static void __get_nps_pa_flip_bits(struct ras_core_context *ras_core, + enum umc_memory_partition_mode nps, + struct umc_flip_bits *flip_bits) +{ + uint32_t vram_type = ras_core->ras_umc.umc_vram_type; + + /* default setting */ + flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_C2_BIT; + flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_C3_BIT; + flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_C4_BIT; + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R13_BIT; + flip_bits->flip_row_bit = 13; + flip_bits->bit_num = 4; + flip_bits->r13_in_pa = UMC_V12_0_PA_R13_BIT; + + if (nps == UMC_MEMORY_PARTITION_MODE_NPS2) { + flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_CH5_BIT; + flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_C2_BIT; + flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_B1_BIT; + flip_bits->r13_in_pa = UMC_V12_0_PA_R12_BIT; + } else if (nps == UMC_MEMORY_PARTITION_MODE_NPS4) { + flip_bits->flip_bits_in_pa[0] = UMC_V12_0_PA_CH4_BIT; + flip_bits->flip_bits_in_pa[1] = UMC_V12_0_PA_CH5_BIT; + flip_bits->flip_bits_in_pa[2] = UMC_V12_0_PA_B0_BIT; + flip_bits->r13_in_pa = UMC_V12_0_PA_R11_BIT; + } + + switch (vram_type) { + case UMC_VRAM_TYPE_HBM: + /* other nps modes are taken as nps1 */ + if (nps == UMC_MEMORY_PARTITION_MODE_NPS2) + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R12_BIT; + else if (nps == UMC_MEMORY_PARTITION_MODE_NPS4) + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R11_BIT; + + break; + case UMC_VRAM_TYPE_HBM3E: + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R12_BIT; + flip_bits->flip_row_bit = 12; + + if (nps == UMC_MEMORY_PARTITION_MODE_NPS2) + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R11_BIT; + else if (nps == UMC_MEMORY_PARTITION_MODE_NPS4) + flip_bits->flip_bits_in_pa[3] = UMC_V12_0_PA_R10_BIT; + + break; + default: + RAS_DEV_WARN(ras_core->dev, + "Unknown HBM type, set RAS retire flip bits to the value in NPS1 mode.\n"); + break; + } +} + +static uint64_t convert_nps_pa_to_row_pa(struct ras_core_context *ras_core, + uint64_t pa, enum umc_memory_partition_mode nps, bool zero_pfn_ok) +{ + struct umc_flip_bits flip_bits = {0}; + uint64_t row_pa; + int i; + + __get_nps_pa_flip_bits(ras_core, nps, &flip_bits); + + row_pa = pa; + /* clear loop bits in soc physical address */ + for (i = 0; i < flip_bits.bit_num; i++) + row_pa &= ~BIT_ULL(flip_bits.flip_bits_in_pa[i]); + + if (!zero_pfn_ok && !RAS_ADDR_TO_PFN(row_pa)) + row_pa |= BIT_ULL(flip_bits.flip_bits_in_pa[2]); + + return row_pa; +} + +static int lookup_bad_pages_in_a_row(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint32_t nps, + uint64_t *pfns, uint32_t num, + uint64_t seq_no, bool dump) +{ + uint32_t col, col_lower, row, row_lower, idx, row_high; + uint64_t soc_pa, row_pa, column, err_addr; + uint64_t retired_addr = RAS_PFN_TO_ADDR(record->cur_nps_retired_row_pfn); + struct umc_flip_bits flip_bits = {0}; + uint32_t retire_unit; + uint32_t i; + + __get_nps_pa_flip_bits(ras_core, nps, &flip_bits); + + row_pa = convert_nps_pa_to_row_pa(ras_core, retired_addr, nps, true); + + err_addr = record->address; + /* get column bit 0 and 1 in mca address */ + col_lower = (err_addr >> 1) & 0x3ULL; + /* MA_R13_BIT will be handled later */ + row_lower = (err_addr >> UMC_V12_0_MCA_R0_BIT) & 0x1fffULL; + row_lower &= ~BIT_ULL(flip_bits.flip_row_bit); + + if (ras_core->ras_gfx.gfx_ip_version >= IP_VERSION(9, 5, 0)) { + row_high = (row_pa >> flip_bits.r13_in_pa) & 0x3ULL; + /* it's 2.25GB in each channel, from MCA address to PA + * [R14 R13] is converted if the two bits value are 0x3, + * get them from PA instead of MCA address. + */ + row_lower |= (row_high << 13); + } + + idx = 0; + row = 0; + retire_unit = 0x1 << flip_bits.bit_num; + /* loop for all possibilities of retire bits */ + for (column = 0; column < retire_unit; column++) { + soc_pa = row_pa; + for (i = 0; i < flip_bits.bit_num; i++) + soc_pa |= (((column >> i) & 0x1ULL) << flip_bits.flip_bits_in_pa[i]); + + col = ((column & 0x7) << 2) | col_lower; + + /* add row bit 13 */ + if (flip_bits.bit_num == UMC_PA_FLIP_BITS_NUM) + row = ((column >> 3) << flip_bits.flip_row_bit) | row_lower; + + if (dump) + RAS_DEV_INFO(ras_core->dev, + "{%llu} Error Address(PA):0x%-10llx Row:0x%-4x Col:0x%-2x Bank:0x%x Channel:0x%x\n", + seq_no, soc_pa, row, col, + record->cur_nps_bank, record->mem_channel); + + + if (pfns && (idx < num)) + pfns[idx++] = RAS_ADDR_TO_PFN(soc_pa); + } + + return idx; +} + +static int umc_v12_convert_ma_to_pa(struct ras_core_context *ras_core, + struct umc_mca_addr *addr_in, struct umc_phy_addr *addr_out, + uint32_t nps) +{ + uint32_t i, na_shift; + uint64_t soc_pa, na, na_nps; + uint32_t bank_hash0, bank_hash1, bank_hash2, bank_hash3, col, row; + uint32_t bank0, bank1, bank2, bank3, bank; + uint32_t ch_inst = addr_in->ch_inst; + uint32_t umc_inst = addr_in->umc_inst; + uint32_t node_inst = addr_in->node_inst; + uint32_t socket_id = addr_in->socket_id; + uint32_t channel_index; + uint64_t err_addr = addr_in->err_addr; + + if (node_inst != UMC_INV_AID_NODE) { + if (ch_inst >= UMC_V12_0_CHANNEL_INSTANCE_NUM || + umc_inst >= UMC_V12_0_UMC_INSTANCE_NUM || + node_inst >= UMC_V12_0_AID_NUM_MAX || + socket_id >= UMC_V12_0_SOCKET_NUM_MAX) + return -EINVAL; + } else { + if (socket_id >= UMC_V12_0_SOCKET_NUM_MAX || + ch_inst >= UMC_V12_0_TOTAL_CHANNEL_NUM) + return -EINVAL; + } + + bank_hash0 = (err_addr >> UMC_V12_0_MCA_B0_BIT) & 0x1ULL; + bank_hash1 = (err_addr >> UMC_V12_0_MCA_B1_BIT) & 0x1ULL; + bank_hash2 = (err_addr >> UMC_V12_0_MCA_B2_BIT) & 0x1ULL; + bank_hash3 = (err_addr >> UMC_V12_0_MCA_B3_BIT) & 0x1ULL; + col = (err_addr >> 1) & 0x1fULL; + row = (err_addr >> 10) & 0x3fffULL; + + /* apply bank hash algorithm */ + bank0 = + bank_hash0 ^ (UMC_V12_0_XOR_EN0 & + (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR0) ^ + (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR0)))); + bank1 = + bank_hash1 ^ (UMC_V12_0_XOR_EN1 & + (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR1) ^ + (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR1)))); + bank2 = + bank_hash2 ^ (UMC_V12_0_XOR_EN2 & + (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR2) ^ + (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR2)))); + bank3 = + bank_hash3 ^ (UMC_V12_0_XOR_EN3 & + (umc_v12_0_bit_wise_xor(col & UMC_V12_0_COL_XOR3) ^ + (umc_v12_0_bit_wise_xor(row & UMC_V12_0_ROW_XOR3)))); + + bank = bank0 | (bank1 << 1) | (bank2 << 2) | (bank3 << 3); + err_addr &= ~0x3c0ULL; + err_addr |= (bank << UMC_V12_0_MCA_B0_BIT); + + na_nps = 0x0; + /* convert mca error address to normalized address */ + for (i = 1; i < ARRAY_SIZE(umc_v12_0_ma2na_mapping); i++) + na_nps |= ((err_addr >> i) & 0x1ULL) << umc_v12_0_ma2na_mapping[i]; + + if (nps == UMC_MEMORY_PARTITION_MODE_NPS1) + na_shift = 8; + else if (nps == UMC_MEMORY_PARTITION_MODE_NPS2) + na_shift = 9; + else if (nps == UMC_MEMORY_PARTITION_MODE_NPS4) + na_shift = 10; + else if (nps == UMC_MEMORY_PARTITION_MODE_NPS8) + na_shift = 11; + else + return -EINVAL; + + na = ((na_nps >> na_shift) << 8) | (na_nps & 0xff); + + if (node_inst != UMC_INV_AID_NODE) + channel_index = + umc_v12_0_channel_idx_tbl[node_inst][umc_inst][ch_inst]; + else { + channel_index = ch_inst; + node_inst = channel_index / + (UMC_V12_0_UMC_INSTANCE_NUM * UMC_V12_0_CHANNEL_INSTANCE_NUM); + } + + /* translate umc channel address to soc pa, 3 parts are included */ + soc_pa = ADDR_OF_32KB_BLOCK(na) | + ADDR_OF_256B_BLOCK(channel_index) | + OFFSET_IN_256B_BLOCK(na); + + /* calc channel hash based on absolute address */ + soc_pa += socket_id * SOCKET_LFB_SIZE; + /* the umc channel bits are not original values, they are hashed */ + UMC_V12_0_SET_CHANNEL_HASH(channel_index, soc_pa); + /* restore pa */ + soc_pa -= socket_id * SOCKET_LFB_SIZE; + + /* get some channel bits from na_nps directly and + * add nps section offset + */ + if (nps == UMC_MEMORY_PARTITION_MODE_NPS2) { + soc_pa &= ~(0x1ULL << UMC_V12_0_PA_CH5_BIT); + soc_pa |= ((na_nps & 0x100) << 5); + soc_pa += (node_inst >> 1) * (SOCKET_LFB_SIZE >> 1); + } else if (nps == UMC_MEMORY_PARTITION_MODE_NPS4) { + soc_pa &= ~(0x3ULL << UMC_V12_0_PA_CH4_BIT); + soc_pa |= ((na_nps & 0x300) << 4); + soc_pa += node_inst * (SOCKET_LFB_SIZE >> 2); + } else if (nps == UMC_MEMORY_PARTITION_MODE_NPS8) { + soc_pa &= ~(0x7ULL << UMC_V12_0_PA_CH4_BIT); + soc_pa |= ((na_nps & 0x700) << 4); + soc_pa += node_inst * (SOCKET_LFB_SIZE >> 2) + + (channel_index >> 4) * (SOCKET_LFB_SIZE >> 3); + } + + addr_out->pa = soc_pa; + addr_out->bank = bank; + addr_out->channel_idx = channel_index; + + return 0; +} + +static int convert_ma_to_pa(struct ras_core_context *ras_core, + struct umc_mca_addr *addr_in, struct umc_phy_addr *addr_out, + uint32_t nps) +{ + int ret; + + if (ras_psp_check_supported_cmd(ras_core, RAS_TA_CMD_ID__QUERY_ADDRESS)) + ret = ras_umc_psp_convert_ma_to_pa(ras_core, + addr_in, addr_out, nps); + else + ret = umc_v12_convert_ma_to_pa(ras_core, + addr_in, addr_out, nps); + + return ret; +} + +static int convert_bank_to_nps_addr(struct ras_core_context *ras_core, + struct ras_bank_ecc *bank, struct umc_phy_addr *pa_addr, uint32_t nps) +{ + struct umc_mca_addr addr_in; + struct umc_phy_addr addr_out; + int ret; + + memset(&addr_in, 0, sizeof(addr_in)); + memset(&addr_out, 0, sizeof(addr_out)); + + addr_in.err_addr = ACA_ADDR_2_ERR_ADDR(bank->addr); + addr_in.ch_inst = ACA_IPID_2_UMC_CH(bank->ipid); + addr_in.umc_inst = ACA_IPID_2_UMC_INST(bank->ipid); + addr_in.node_inst = ACA_IPID_2_DIE_ID(bank->ipid); + addr_in.socket_id = ACA_IPID_2_SOCKET_ID(bank->ipid); + + ret = convert_ma_to_pa(ras_core, &addr_in, &addr_out, nps); + if (!ret) { + pa_addr->pa = + convert_nps_pa_to_row_pa(ras_core, addr_out.pa, nps, false); + pa_addr->channel_idx = addr_out.channel_idx; + pa_addr->bank = addr_out.bank; + } + + return ret; +} + +static int umc_v12_0_bank_to_eeprom_record(struct ras_core_context *ras_core, + struct ras_bank_ecc *bank, struct eeprom_umc_record *record) +{ + struct umc_phy_addr nps_addr; + int ret; + + memset(&nps_addr, 0, sizeof(nps_addr)); + + ret = convert_bank_to_nps_addr(ras_core, bank, + &nps_addr, bank->nps); + if (ret) + return ret; + + ras_umc_fill_eeprom_record(ras_core, + ACA_ADDR_2_ERR_ADDR(bank->addr), ACA_IPID_2_UMC_INST(bank->ipid), + &nps_addr, bank->nps, record); + + lookup_bad_pages_in_a_row(ras_core, record, + bank->nps, NULL, 0, bank->seq_no, true); + + return 0; +} + +static int convert_eeprom_record_to_nps_addr(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint64_t *pa, uint32_t nps) +{ + struct device_system_info dev_info = {0}; + struct umc_mca_addr addr_in; + struct umc_phy_addr addr_out; + int ret; + + memset(&addr_in, 0, sizeof(addr_in)); + memset(&addr_out, 0, sizeof(addr_out)); + + ras_core_get_device_system_info(ras_core, &dev_info); + + addr_in.err_addr = record->address; + addr_in.ch_inst = record->mem_channel; + addr_in.umc_inst = record->mcumc_id; + addr_in.node_inst = UMC_INV_AID_NODE; + addr_in.socket_id = dev_info.socket_id; + + ret = convert_ma_to_pa(ras_core, &addr_in, &addr_out, nps); + if (ret) + return ret; + + *pa = convert_nps_pa_to_row_pa(ras_core, addr_out.pa, nps, false); + + return 0; +} + +static int umc_v12_0_eeprom_record_to_nps_record(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint32_t nps) +{ + uint64_t pa = 0; + int ret = 0; + + if (nps == EEPROM_RECORD_UMC_NPS_MODE(record)) { + record->cur_nps_retired_row_pfn = EEPROM_RECORD_UMC_ADDR_PFN(record); + } else { + ret = convert_eeprom_record_to_nps_addr(ras_core, + record, &pa, nps); + if (!ret) + record->cur_nps_retired_row_pfn = RAS_ADDR_TO_PFN(pa); + } + + record->cur_nps = nps; + + return ret; +} + +static int umc_v12_0_eeprom_record_to_nps_pages(struct ras_core_context *ras_core, + struct eeprom_umc_record *record, uint32_t nps, + uint64_t *pfns, uint32_t num) +{ + return lookup_bad_pages_in_a_row(ras_core, + record, nps, pfns, num, 0, false); +} + +static int umc_12_0_soc_pa_to_bank(struct ras_core_context *ras_core, + uint64_t soc_pa, + struct umc_bank_addr *bank_addr) +{ + + int channel_hashed = 0; + int channel_real = 0; + int channel_reversed = 0; + int i = 0; + + bank_addr->stack_id = UMC_V12_0_SOC_PA_TO_SID(soc_pa); + bank_addr->bank_group = 0; /* This is a combination of SID & Bank. Needed?? */ + bank_addr->bank = UMC_V12_0_SOC_PA_TO_BANK(soc_pa); + bank_addr->row = UMC_V12_0_SOC_PA_TO_ROW(soc_pa); + bank_addr->column = UMC_V12_0_SOC_PA_TO_COL(soc_pa); + + /* Channel bits 4-6 are hashed. Bruteforce reverse the hash */ + channel_hashed = (soc_pa >> UMC_V12_0_PA_CH4_BIT) & 0x7; + + for (i = 0; i < 8; i++) { + channel_reversed = 0; + channel_reversed |= UMC_V12_0_CHANNEL_HASH_CH4((i << 4), soc_pa); + channel_reversed |= (UMC_V12_0_CHANNEL_HASH_CH5((i << 4), soc_pa) << 1); + channel_reversed |= (UMC_V12_0_CHANNEL_HASH_CH6((i << 4), soc_pa) << 2); + if (channel_reversed == channel_hashed) + channel_real = ((i << 4)) | ((soc_pa >> UMC_V12_0_PA_CH0_BIT) & 0xf); + } + + bank_addr->channel = channel_real; + bank_addr->subchannel = UMC_V12_0_SOC_PA_TO_PC(soc_pa); + + return 0; +} + +static int umc_12_0_bank_to_soc_pa(struct ras_core_context *ras_core, + struct umc_bank_addr bank_addr, + uint64_t *soc_pa) +{ + uint64_t na = 0; + uint64_t tmp_pa = 0; + *soc_pa = 0; + + tmp_pa |= UMC_V12_0_SOC_SID_TO_PA(bank_addr.stack_id); + tmp_pa |= UMC_V12_0_SOC_BANK_TO_PA(bank_addr.bank); + tmp_pa |= UMC_V12_0_SOC_ROW_TO_PA(bank_addr.row); + tmp_pa |= UMC_V12_0_SOC_COL_TO_PA(bank_addr.column); + tmp_pa |= UMC_V12_0_SOC_CH_TO_PA(bank_addr.channel); + tmp_pa |= UMC_V12_0_SOC_PC_TO_PA(bank_addr.subchannel); + + /* Get the NA */ + na = ((tmp_pa >> UMC_V12_0_PA_C2_BIT) << UMC_V12_0_NA_C2_BIT); + na |= tmp_pa & 0xff; + + /* translate umc channel address to soc pa, 3 parts are included */ + tmp_pa = ADDR_OF_32KB_BLOCK(na) | + ADDR_OF_256B_BLOCK(bank_addr.channel) | + OFFSET_IN_256B_BLOCK(na); + + /* the umc channel bits are not original values, they are hashed */ + UMC_V12_0_SET_CHANNEL_HASH(bank_addr.channel, tmp_pa); + + *soc_pa = tmp_pa; + + return 0; +} + +const struct ras_umc_ip_func ras_umc_func_v12_0 = { + .bank_to_eeprom_record = umc_v12_0_bank_to_eeprom_record, + .eeprom_record_to_nps_record = umc_v12_0_eeprom_record_to_nps_record, + .eeprom_record_to_nps_pages = umc_v12_0_eeprom_record_to_nps_pages, + .bank_to_soc_pa = umc_12_0_bank_to_soc_pa, + .soc_pa_to_bank = umc_12_0_soc_pa_to_bank, +}; + diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.h b/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.h new file mode 100644 index 000000000000..8a35ad856165 --- /dev/null +++ b/drivers/gpu/drm/amd/ras/rascore/ras_umc_v12_0.h @@ -0,0 +1,314 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __RAS_UMC_V12_0_H__ +#define __RAS_UMC_V12_0_H__ +#include "ras.h" + +/* MCA_UMC_UMC0_MCUMC_ADDRT0 */ +#define MCA_UMC_UMC0_MCUMC_ADDRT0__ErrorAddr__SHIFT 0x0 +#define MCA_UMC_UMC0_MCUMC_ADDRT0__Reserved__SHIFT 0x38 +#define MCA_UMC_UMC0_MCUMC_ADDRT0__ErrorAddr_MASK 0x00FFFFFFFFFFFFFFL +#define MCA_UMC_UMC0_MCUMC_ADDRT0__Reserved_MASK 0xFF00000000000000L + +/* MCMP1_IPIDT0 */ +#define MCMP1_IPIDT0__InstanceIdLo__SHIFT 0x0 +#define MCMP1_IPIDT0__HardwareID__SHIFT 0x20 +#define MCMP1_IPIDT0__InstanceIdHi__SHIFT 0x2c +#define MCMP1_IPIDT0__McaType__SHIFT 0x30 + +#define MCMP1_IPIDT0__InstanceIdLo_MASK 0x00000000FFFFFFFFL +#define MCMP1_IPIDT0__HardwareID_MASK 0x00000FFF00000000L +#define MCMP1_IPIDT0__InstanceIdHi_MASK 0x0000F00000000000L +#define MCMP1_IPIDT0__McaType_MASK 0xFFFF000000000000L + +/* number of umc channel instance with memory map register access */ +#define UMC_V12_0_CHANNEL_INSTANCE_NUM 8 +/* number of umc instance with memory map register access */ +#define UMC_V12_0_UMC_INSTANCE_NUM 4 + +/* one piece of normalized address is mapped to 8 pieces of physical address */ +#define UMC_V12_0_NA_MAP_PA_NUM 8 + +/* bank bits in MCA error address */ +#define UMC_V12_0_MCA_B0_BIT 6 +#define UMC_V12_0_MCA_B1_BIT 7 +#define UMC_V12_0_MCA_B2_BIT 8 +#define UMC_V12_0_MCA_B3_BIT 9 + +/* row bits in MCA address */ +#define UMC_V12_0_MCA_R0_BIT 10 + +/* Stack ID bits in SOC physical address */ +#define UMC_V12_0_PA_SID1_BIT 37 +#define UMC_V12_0_PA_SID0_BIT 36 + +/* bank bits in SOC physical address */ +#define UMC_V12_0_PA_B3_BIT 18 +#define UMC_V12_0_PA_B2_BIT 17 +#define UMC_V12_0_PA_B1_BIT 20 +#define UMC_V12_0_PA_B0_BIT 19 + +/* row bits in SOC physical address */ +#define UMC_V12_0_PA_R13_BIT 35 +#define UMC_V12_0_PA_R12_BIT 34 +#define UMC_V12_0_PA_R11_BIT 33 +#define UMC_V12_0_PA_R10_BIT 32 +#define UMC_V12_0_PA_R9_BIT 31 +#define UMC_V12_0_PA_R8_BIT 30 +#define UMC_V12_0_PA_R7_BIT 29 +#define UMC_V12_0_PA_R6_BIT 28 +#define UMC_V12_0_PA_R5_BIT 27 +#define UMC_V12_0_PA_R4_BIT 26 +#define UMC_V12_0_PA_R3_BIT 25 +#define UMC_V12_0_PA_R2_BIT 24 +#define UMC_V12_0_PA_R1_BIT 23 +#define UMC_V12_0_PA_R0_BIT 22 + +/* column bits in SOC physical address */ +#define UMC_V12_0_PA_C4_BIT 21 +#define UMC_V12_0_PA_C3_BIT 16 +#define UMC_V12_0_PA_C2_BIT 15 +#define UMC_V12_0_PA_C1_BIT 6 +#define UMC_V12_0_PA_C0_BIT 5 + +/* channel index bits in SOC physical address */ +#define UMC_V12_0_PA_CH6_BIT 14 +#define UMC_V12_0_PA_CH5_BIT 13 +#define UMC_V12_0_PA_CH4_BIT 12 +#define UMC_V12_0_PA_CH3_BIT 11 +#define UMC_V12_0_PA_CH2_BIT 10 +#define UMC_V12_0_PA_CH1_BIT 9 +#define UMC_V12_0_PA_CH0_BIT 8 + +/* Pseudochannel index bits in SOC physical address */ +#define UMC_V12_0_PA_PC0_BIT 7 + +#define UMC_V12_0_NA_C2_BIT 8 + +#define UMC_V12_0_SOC_PA_TO_SID(pa) \ + ((((pa >> UMC_V12_0_PA_SID0_BIT) & 0x1ULL) << 0ULL) | \ + (((pa >> UMC_V12_0_PA_SID1_BIT) & 0x1ULL) << 1ULL)) + +#define UMC_V12_0_SOC_PA_TO_BANK(pa) \ + ((((pa >> UMC_V12_0_PA_B0_BIT) & 0x1ULL) << 0ULL) | \ + (((pa >> UMC_V12_0_PA_B1_BIT) & 0x1ULL) << 1ULL) | \ + (((pa >> UMC_V12_0_PA_B2_BIT) & 0x1ULL) << 2ULL) | \ + (((pa >> UMC_V12_0_PA_B3_BIT) & 0x1ULL) << 3ULL)) + +#define UMC_V12_0_SOC_PA_TO_ROW(pa) \ + ((((pa >> UMC_V12_0_PA_R0_BIT) & 0x1ULL) << 0ULL) | \ + (((pa >> UMC_V12_0_PA_R1_BIT) & 0x1ULL) << 1ULL) | \ + (((pa >> UMC_V12_0_PA_R2_BIT) & 0x1ULL) << 2ULL) | \ + (((pa >> UMC_V12_0_PA_R3_BIT) & 0x1ULL) << 3ULL) | \ + (((pa >> UMC_V12_0_PA_R4_BIT) & 0x1ULL) << 4ULL) | \ + (((pa >> UMC_V12_0_PA_R5_BIT) & 0x1ULL) << 5ULL) | \ + (((pa >> UMC_V12_0_PA_R6_BIT) & 0x1ULL) << 6ULL) | \ + (((pa >> UMC_V12_0_PA_R7_BIT) & 0x1ULL) << 7ULL) | \ + (((pa >> UMC_V12_0_PA_R8_BIT) & 0x1ULL) << 8ULL) | \ + (((pa >> UMC_V12_0_PA_R9_BIT) & 0x1ULL) << 9ULL) | \ + (((pa >> UMC_V12_0_PA_R10_BIT) & 0x1ULL) << 10ULL) | \ + (((pa >> UMC_V12_0_PA_R11_BIT) & 0x1ULL) << 11ULL) | \ + (((pa >> UMC_V12_0_PA_R12_BIT) & 0x1ULL) << 12ULL) | \ + (((pa >> UMC_V12_0_PA_R13_BIT) & 0x1ULL) << 13ULL)) + +#define UMC_V12_0_SOC_PA_TO_COL(pa) \ + ((((pa >> UMC_V12_0_PA_C0_BIT) & 0x1ULL) << 0ULL) | \ + (((pa >> UMC_V12_0_PA_C1_BIT) & 0x1ULL) << 1ULL) | \ + (((pa >> UMC_V12_0_PA_C2_BIT) & 0x1ULL) << 2ULL) | \ + (((pa >> UMC_V12_0_PA_C3_BIT) & 0x1ULL) << 3ULL) | \ + (((pa >> UMC_V12_0_PA_C4_BIT) & 0x1ULL) << 4ULL)) + +#define UMC_V12_0_SOC_PA_TO_CH(pa) \ + ((((pa >> UMC_V12_0_PA_CH0_BIT) & 0x1ULL) << 0ULL) | \ + (((pa >> UMC_V12_0_PA_CH1_BIT) & 0x1ULL) << 1ULL) | \ + (((pa >> UMC_V12_0_PA_CH2_BIT) & 0x1ULL) << 2ULL) | \ + (((pa >> UMC_V12_0_PA_CH3_BIT) & 0x1ULL) << 3ULL) | \ + (((pa >> UMC_V12_0_PA_CH4_BIT) & 0x1ULL) << 4ULL) | \ + (((pa >> UMC_V12_0_PA_CH5_BIT) & 0x1ULL) << 5ULL) | \ + (((pa >> UMC_V12_0_PA_CH6_BIT) & 0x1ULL) << 6ULL)) + +#define UMC_V12_0_SOC_PA_TO_PC(pa) (((pa >> UMC_V12_0_PA_PC0_BIT) & 0x1ULL) << 0ULL) + +#define UMC_V12_0_SOC_SID_TO_PA(sid) \ + ((((sid >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_SID0_BIT) | \ + (((sid >> 1ULL) & 0x1ULL) << UMC_V12_0_PA_SID1_BIT)) + +#define UMC_V12_0_SOC_BANK_TO_PA(bank) \ + ((((bank >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_B0_BIT) | \ + (((bank >> 1ULL) & 0x1ULL) << UMC_V12_0_PA_B1_BIT) | \ + (((bank >> 2ULL) & 0x1ULL) << UMC_V12_0_PA_B2_BIT) | \ + (((bank >> 3ULL) & 0x1ULL) << UMC_V12_0_PA_B3_BIT)) + +#define UMC_V12_0_SOC_ROW_TO_PA(row) \ + ((((row >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_R0_BIT) | \ + (((row >> 1ULL) & 0x1ULL) << UMC_V12_0_PA_R1_BIT) | \ + (((row >> 2ULL) & 0x1ULL) << UMC_V12_0_PA_R2_BIT) | \ + (((row >> 3ULL) & 0x1ULL) << UMC_V12_0_PA_R3_BIT) | \ + (((row >> 4ULL) & 0x1ULL) << UMC_V12_0_PA_R4_BIT) | \ + (((row >> 5ULL) & 0x1ULL) << UMC_V12_0_PA_R5_BIT) | \ + (((row >> 6ULL) & 0x1ULL) << UMC_V12_0_PA_R6_BIT) | \ + (((row >> 7ULL) & 0x1ULL) << UMC_V12_0_PA_R7_BIT) | \ + (((row >> 8ULL) & 0x1ULL) << UMC_V12_0_PA_R8_BIT) | \ + (((row >> 9ULL) & 0x1ULL) << UMC_V12_0_PA_R9_BIT) | \ + (((row >> 10ULL) & 0x1ULL) << UMC_V12_0_PA_R10_BIT) | \ + (((row >> 11ULL) & 0x1ULL) << UMC_V12_0_PA_R11_BIT) | \ + (((row >> 12ULL) & 0x1ULL) << UMC_V12_0_PA_R12_BIT) | \ + (((row >> 13ULL) & 0x1ULL) << UMC_V12_0_PA_R13_BIT)) + +#define UMC_V12_0_SOC_COL_TO_PA(col) \ + ((((col >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_C0_BIT) | \ + (((col >> 1ULL) & 0x1ULL) << UMC_V12_0_PA_C1_BIT) | \ + (((col >> 2ULL) & 0x1ULL) << UMC_V12_0_PA_C2_BIT) | \ + (((col >> 3ULL) & 0x1ULL) << UMC_V12_0_PA_C3_BIT) | \ + (((col >> 4ULL) & 0x1ULL) << UMC_V12_0_PA_C4_BIT)) + +#define UMC_V12_0_SOC_CH_TO_PA(ch) \ + ((((ch >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_CH0_BIT) | \ + (((ch >> 1ULL) & 0x1ULL) << UMC_V12_0_PA_CH1_BIT) | \ + (((ch >> 2ULL) & 0x1ULL) << UMC_V12_0_PA_CH2_BIT) | \ + (((ch >> 3ULL) & 0x1ULL) << UMC_V12_0_PA_CH3_BIT) | \ + (((ch >> 4ULL) & 0x1ULL) << UMC_V12_0_PA_CH4_BIT) | \ + (((ch >> 5ULL) & 0x1ULL) << UMC_V12_0_PA_CH5_BIT) | \ + (((ch >> 6ULL) & 0x1ULL) << UMC_V12_0_PA_CH6_BIT)) + +#define UMC_V12_0_SOC_PC_TO_PA(pc) (((pc >> 0ULL) & 0x1ULL) << UMC_V12_0_PA_PC0_BIT) + +/* bank hash settings */ +#define UMC_V12_0_XOR_EN0 1 +#define UMC_V12_0_XOR_EN1 1 +#define UMC_V12_0_XOR_EN2 1 +#define UMC_V12_0_XOR_EN3 1 +#define UMC_V12_0_COL_XOR0 0x0 +#define UMC_V12_0_COL_XOR1 0x0 +#define UMC_V12_0_COL_XOR2 0x800 +#define UMC_V12_0_COL_XOR3 0x1000 +#define UMC_V12_0_ROW_XOR0 0x11111 +#define UMC_V12_0_ROW_XOR1 0x22222 +#define UMC_V12_0_ROW_XOR2 0x4444 +#define UMC_V12_0_ROW_XOR3 0x8888 + +/* channel hash settings */ +#define UMC_V12_0_HASH_4K 0 +#define UMC_V12_0_HASH_64K 1 +#define UMC_V12_0_HASH_2M 1 +#define UMC_V12_0_HASH_1G 1 +#define UMC_V12_0_HASH_1T 1 + +/* XOR some bits of PA into CH4~CH6 bits (bits 12~14 of PA), + * hash bit is only effective when related setting is enabled + */ +#define UMC_V12_0_CHANNEL_HASH_CH4(channel_idx, pa) ((((channel_idx) >> 5) & 0x1) ^ \ + (((pa) >> 20) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ + (((pa) >> 27) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ + (((pa) >> 34) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ + (((pa) >> 41) & 0x1ULL & UMC_V12_0_HASH_1T)) +#define UMC_V12_0_CHANNEL_HASH_CH5(channel_idx, pa) ((((channel_idx) >> 6) & 0x1) ^ \ + (((pa) >> 21) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ + (((pa) >> 28) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ + (((pa) >> 35) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ + (((pa) >> 42) & 0x1ULL & UMC_V12_0_HASH_1T)) +#define UMC_V12_0_CHANNEL_HASH_CH6(channel_idx, pa) ((((channel_idx) >> 4) & 0x1) ^ \ + (((pa) >> 19) & 0x1ULL & UMC_V12_0_HASH_64K) ^ \ + (((pa) >> 26) & 0x1ULL & UMC_V12_0_HASH_2M) ^ \ + (((pa) >> 33) & 0x1ULL & UMC_V12_0_HASH_1G) ^ \ + (((pa) >> 40) & 0x1ULL & UMC_V12_0_HASH_1T) ^ \ + (((pa) >> 47) & 0x1ULL & UMC_V12_0_HASH_1T)) +#define UMC_V12_0_SET_CHANNEL_HASH(channel_idx, pa) do { \ + (pa) &= ~(0x7ULL << UMC_V12_0_PA_CH4_BIT); \ + (pa) |= (UMC_V12_0_CHANNEL_HASH_CH4(channel_idx, pa) << UMC_V12_0_PA_CH4_BIT); \ + (pa) |= (UMC_V12_0_CHANNEL_HASH_CH5(channel_idx, pa) << UMC_V12_0_PA_CH5_BIT); \ + (pa) |= (UMC_V12_0_CHANNEL_HASH_CH6(channel_idx, pa) << UMC_V12_0_PA_CH6_BIT); \ + } while (0) + + +/* + * (addr / 256) * 4096, the higher 26 bits in ErrorAddr + * is the index of 4KB block + */ +#define ADDR_OF_4KB_BLOCK(addr) (((addr) & ~0xffULL) << 4) +/* + * (addr / 256) * 8192, the higher 26 bits in ErrorAddr + * is the index of 8KB block + */ +#define ADDR_OF_8KB_BLOCK(addr) (((addr) & ~0xffULL) << 5) +/* + * (addr / 256) * 32768, the higher 26 bits in ErrorAddr + * is the index of 8KB block + */ +#define ADDR_OF_32KB_BLOCK(addr) (((addr) & ~0xffULL) << 7) +/* channel index is the index of 256B block */ +#define ADDR_OF_256B_BLOCK(channel_index) ((channel_index) << 8) +/* offset in 256B block */ +#define OFFSET_IN_256B_BLOCK(addr) ((addr) & 0xffULL) + + +#define UMC_V12_ADDR_MASK_BAD_COLS(addr) \ + ((addr) & ~((0x3ULL << UMC_V12_0_PA_C2_BIT) | \ + (0x1ULL << UMC_V12_0_PA_C4_BIT) | \ + (0x1ULL << UMC_V12_0_PA_R13_BIT))) + +#define ACA_IPID_HI_2_UMC_AID(_ipid_hi) (((_ipid_hi) >> 2) & 0x3) +#define ACA_IPID_LO_2_UMC_CH(_ipid_lo) \ + (((((_ipid_lo) >> 20) & 0x1) * 4) + (((_ipid_lo) >> 12) & 0xF)) +#define ACA_IPID_LO_2_UMC_INST(_ipid_lo) (((_ipid_lo) >> 21) & 0x7) + +#define ACA_IPID_2_DIE_ID(ipid) ((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) >> 2) & 0x03) +#define ACA_IPID_2_UMC_CH(ipid) \ + (ACA_IPID_LO_2_UMC_CH(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo))) + +#define ACA_IPID_2_UMC_INST(ipid) \ + (ACA_IPID_LO_2_UMC_INST(REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo))) + +#define ACA_IPID_2_SOCKET_ID(ipid) \ + (((REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdLo) & 0x1) << 2) | \ + (REG_GET_FIELD(ipid, MCMP1_IPIDT0, InstanceIdHi) & 0x03)) + +#define ACA_ADDR_2_ERR_ADDR(addr) \ + REG_GET_FIELD(addr, MCA_UMC_UMC0_MCUMC_ADDRT0, ErrorAddr) + +/* R13 bit shift should be considered, double the number */ +#define UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL (UMC_V12_0_NA_MAP_PA_NUM * 2) + + +/* C2, C3, C4, R13, four MCA bits are looped in page retirement */ +#define UMC_V12_0_RETIRE_LOOP_BITS 4 + +/* invalid node instance value */ +#define UMC_INV_AID_NODE 0xffff + +#define UMC_V12_0_AID_NUM_MAX 4 +#define UMC_V12_0_SOCKET_NUM_MAX 8 + +#define UMC_V12_0_TOTAL_CHANNEL_NUM \ + (UMC_V12_0_AID_NUM_MAX * UMC_V12_0_UMC_INSTANCE_NUM * UMC_V12_0_CHANNEL_INSTANCE_NUM) + +/* one device has 192GB HBM */ +#define SOCKET_LFB_SIZE 0x3000000000ULL + +extern const struct ras_umc_ip_func ras_umc_func_v12_0; + +int ras_umc_get_badpage_count(struct ras_core_context *ras_core); +int ras_umc_get_badpage_record(struct ras_core_context *ras_core, uint32_t index, void *record); +#endif + |