diff options
106 files changed, 2438 insertions, 1134 deletions
diff --git a/MAINTAINERS b/MAINTAINERS index 5d59ff76cfc6..b91e5b933be8 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -1080,7 +1080,7 @@ M: Austin Zheng <austin.zheng@amd.com> M: Jun Lei <jun.lei@amd.com> S: Supported F: drivers/gpu/drm/amd/display/dc/dml/ -F: drivers/gpu/drm/amd/display/dc/dml2/ +F: drivers/gpu/drm/amd/display/dc/dml2_0/ AMD FAM15H PROCESSOR POWER MONITORING DRIVER M: Huang Rui <ray.huang@amd.com> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 50079209c472..9f9774f58ce1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1316,6 +1316,7 @@ struct amdgpu_device { bool apu_prefer_gtt; bool userq_halt_for_enforce_isolation; + struct work_struct userq_reset_work; struct amdgpu_uid *uid_info; /* KFD @@ -1539,11 +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))) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 96ccd5ade031..b1c24c8fa686 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -1274,6 +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); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 654f4844b7ad..86255c13fbb7 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, @@ -2494,6 +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) @@ -3784,7 +3790,7 @@ 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); @@ -3807,10 +3813,23 @@ static int amdgpu_device_ip_suspend_phase1(struct amdgpu_device *adev) 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; } /** @@ -3826,7 +3845,7 @@ 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); @@ -3889,7 +3908,7 @@ static int amdgpu_device_ip_suspend_phase2(struct amdgpu_device *adev) r = amdgpu_ip_block_suspend(&adev->ip_blocks[i]); if (r) - return r; + goto unwind; /* handle putting the SMC in the appropriate state */ if (!amdgpu_sriov_vf(adev)) { @@ -3899,13 +3918,40 @@ 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; } /** @@ -4607,6 +4653,7 @@ 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; @@ -5229,7 +5276,7 @@ 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; @@ -5245,8 +5292,9 @@ 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)); @@ -5257,16 +5305,16 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) r = amdgpu_device_ip_suspend_phase1(adev); if (r) - return r; + goto unwind_smartshift; amdgpu_amdkfd_suspend(adev, !amdgpu_sriov_vf(adev) && !adev->in_runpm); r = amdgpu_userq_suspend(adev); if (r) - return 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); @@ -5274,16 +5322,62 @@ int amdgpu_device_suspend(struct drm_device *dev, bool notify_clients) r = amdgpu_device_ip_suspend_phase2(adev); if (r) - return 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) @@ -5989,6 +6083,10 @@ int amdgpu_device_reinit_after_reset(struct amdgpu_reset_context *reset_context) if (r) goto out; + r = amdgpu_userq_post_reset(tmp_adev, vram_lost); + if (r) + goto out; + drm_client_dev_resume(adev_to_drm(tmp_adev)); /* @@ -6211,6 +6309,7 @@ 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); @@ -6331,6 +6430,8 @@ static void amdgpu_device_halt_activities(struct amdgpu_device *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]; @@ -6560,6 +6661,9 @@ 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); @@ -6590,6 +6694,7 @@ 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); @@ -7327,7 +7432,7 @@ void amdgpu_device_flush_hdp(struct amdgpu_device *adev, return; } - amdgpu_asic_flush_hdp(adev, ring); + amdgpu_hdp_flush(adev, ring); } void amdgpu_device_invalidate_hdp(struct amdgpu_device *adev, @@ -7340,7 +7445,7 @@ 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_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index 8561ad7f6180..268d69d862e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -81,13 +81,20 @@ 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; 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 3776901bbb1b..cb522d6272d6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2625,9 +2625,14 @@ 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; } 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 518ca3f4db2b..90d26d820bac 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c @@ -287,9 +287,7 @@ void amdgpu_hmm_range_free(struct amdgpu_hmm_range *range) if (!range) return; - if (range->hmm_range.hmm_pfns) - kvfree(range->hmm_range.hmm_pfns); - + kvfree(range->hmm_range.hmm_pfns); amdgpu_bo_unref(&range->bo); kfree(range); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c index 39229ece83f8..586a58facca1 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ib.c @@ -176,18 +176,21 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, 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 * @@ -196,7 +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; @@ -302,6 +305,9 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, 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); @@ -328,6 +334,11 @@ int amdgpu_ib_schedule(struct amdgpu_ring *ring, unsigned int num_ibs, amdgpu_ring_commit(ring); return 0; + +free_fence: + if (!job) + kfree(af); + return r; } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index e08d837668f1..7d8ef7ae10c2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -130,11 +130,9 @@ 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); @@ -295,6 +293,15 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job) amdgpu_sync_free(&job->explicit_sync); + 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 + kfree(job->hw_vm_fence); + kfree(job); } @@ -324,6 +331,15 @@ 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) + 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 + kfree(job->hw_vm_fence); + kfree(job); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index 8c0e5d03de50..aa7987d0806c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -2355,8 +2355,11 @@ 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_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index c8b4dd3ea5c3..055a9bbabbdb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1661,7 +1661,7 @@ static int amdgpu_uniras_error_inject(struct amdgpu_device *adev, inject_req.address = info->address; inject_req.error_type = info->head.type; inject_req.instance_mask = info->instance_mask; - inject_req.value = info->value; + 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)); @@ -2921,8 +2921,12 @@ 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); + } } } @@ -3154,8 +3158,12 @@ 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 { + save_nps = nps; + } if (save_nps == nps) { if (amdgpu_umc_pages_in_a_row(adev, err_data, @@ -3221,7 +3229,8 @@ 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) && @@ -3352,7 +3361,8 @@ 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) && @@ -3767,6 +3777,8 @@ int amdgpu_ras_init_badpage_info(struct amdgpu_device *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; @@ -5671,3 +5683,25 @@ 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 556cf4d7b5ef..ff44190d7d98 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.h @@ -503,6 +503,32 @@ 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 */ @@ -591,6 +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 { @@ -1009,4 +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 5a7bf0661dbf..670c0dedf4e9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -444,40 +444,57 @@ 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; @@ -770,7 +787,8 @@ 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) && @@ -853,6 +871,18 @@ Out: return res; } +int amdgpu_ras_eeprom_update_record_num(struct amdgpu_ras_eeprom_control *control) +{ + struct amdgpu_device *adev = to_amdgpu_device(control); + + if (!amdgpu_ras_smu_eeprom_supported(adev)) + return 0; + + control->ras_num_recs_old = control->ras_num_recs; + return amdgpu_ras_smu_get_badpage_count(adev, + &(control->ras_num_recs), 12); +} + /** * amdgpu_ras_eeprom_append -- append records to the EEPROM RAS table * @control: pointer to control structure @@ -871,11 +901,17 @@ int amdgpu_ras_eeprom_append(struct amdgpu_ras_eeprom_control *control, const u32 num) { struct amdgpu_device *adev = to_amdgpu_device(control); + struct amdgpu_ras *con = amdgpu_ras_get_context(adev); int res, i; uint64_t nps = AMDGPU_NPS1_PARTITION_MODE; - if (!__is_ras_eeprom_supported(adev)) + if (!__is_ras_eeprom_supported(adev) || !con) + return 0; + + if (amdgpu_ras_smu_eeprom_supported(adev)) { + control->ras_num_bad_pages = con->bad_page_num; return 0; + } if (num == 0) { dev_err(adev->dev, "will not append 0 records\n"); @@ -952,6 +988,50 @@ 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; + record[i - rec_idx].cu = 0; + + adev->umc.ras->mca_ipid_parse(adev, ipid, NULL, + (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 @@ -973,6 +1053,9 @@ 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; @@ -1144,6 +1227,10 @@ 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 @@ -1374,6 +1461,42 @@ 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); @@ -1382,6 +1505,9 @@ 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)) @@ -1448,6 +1574,47 @@ 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); @@ -1455,6 +1622,9 @@ 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; @@ -1545,7 +1715,8 @@ 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) @@ -1565,4 +1736,143 @@ 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 43f769fed810..cd8873c6931a 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" /* @@ -468,9 +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) @@ -498,6 +496,61 @@ 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)) { + if (copy_to_user(buf, ring_header, ring_header_size)) + return -EFAULT; + buf += 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 @@ -514,6 +567,9 @@ 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; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h index 87b962df5460..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 { 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 13cc5a686dfd..836a14ef0052 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.c @@ -25,8 +25,10 @@ #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" @@ -46,6 +48,107 @@ u32 amdgpu_userq_get_supported_ip_mask(struct amdgpu_device *adev) return userq_ip_mask; } +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) { @@ -176,17 +279,22 @@ 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; } @@ -218,16 +326,23 @@ 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; } @@ -244,10 +359,12 @@ 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; } @@ -475,10 +592,11 @@ 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)) { @@ -701,6 +819,7 @@ 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); @@ -943,6 +1062,8 @@ retry_lock: /* 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; @@ -1045,6 +1166,7 @@ amdgpu_userq_evict_all(struct amdgpu_userq_mgr *uq_mgr) 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 */ xa_for_each(&uq_mgr->userq_mgr_xa, queue_id, queue) { r = amdgpu_userq_preempt_helper(uq_mgr, queue); @@ -1057,6 +1179,23 @@ 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) { @@ -1084,22 +1223,19 @@ 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); @@ -1133,6 +1269,7 @@ void amdgpu_userq_mgr_fini(struct amdgpu_userq_mgr *userq_mgr) cancel_delayed_work_sync(&userq_mgr->resume_work); mutex_lock(&userq_mgr->userq_mutex); + 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); @@ -1159,6 +1296,7 @@ int amdgpu_userq_suspend(struct amdgpu_device *adev) uqm = queue->userq_mgr; cancel_delayed_work_sync(&uqm->resume_work); guard(mutex)(&uqm->userq_mutex); + amdgpu_userq_detect_and_reset_queues(uqm); if (adev->in_s0ix) r = amdgpu_userq_preempt_helper(uqm, queue); else @@ -1217,6 +1355,7 @@ int amdgpu_userq_stop_sched_for_enforce_isolation(struct amdgpu_device *adev, 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; @@ -1289,3 +1428,56 @@ int amdgpu_userq_gem_va_unmap_validate(struct amdgpu_device *adev, 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 09da0617bfa2..c37444427a14 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq.h @@ -106,6 +106,7 @@ struct amdgpu_userq_mgr { struct amdgpu_device *adev; struct delayed_work resume_work; struct drm_file *file; + atomic_t userq_count[AMDGPU_RING_TYPE_MAX]; }; struct amdgpu_db_info { @@ -148,6 +149,10 @@ 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); +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, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c index 2aeeaa954882..99ae1d19b751 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c @@ -386,6 +386,7 @@ 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; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 9309830821b7..700b4a776532 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); @@ -845,6 +849,8 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, if (r) return r; fence = &job->hw_vm_fence->base; + /* get a ref for the job */ + dma_fence_get(fence); } if (vm_flush_needed) { 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/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 751732f3e883..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; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 252517ce5d5a..02d7cfae22bd 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; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index 35d5a7e99a7c..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; } 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 d3d0a4b0380c..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; } diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index f1a2efc2a8d0..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); 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 e0b50c690f8c..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; } @@ -2289,7 +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( 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..0ceeb19df2e5 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: 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/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/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 4fbe865ff279..0904c36192c7 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -2094,7 +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)) { 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_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index ffb7b36e577c..c30dfb8ec236 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -1744,11 +1744,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm, else r = -ENOMEM; WRITE_ONCE(p->svms.faulting_task, NULL); - if (r) { - amdgpu_hmm_range_free(range); - range = NULL; + if (r) pr_debug("failed %d to get svm range pages\n", r); - } } else { r = -EFAULT; } @@ -1771,10 +1768,9 @@ static int svm_range_validate_and_map(struct mm_struct *mm, pr_debug("hmm update the range, need validate again\n"); r = -EAGAIN; } - /* Free the hmm range */ - if (range) - amdgpu_hmm_range_free(range); + /* 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"); 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 bb0fe91a1601..18f1cf16ec18 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -3624,6 +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; @@ -3633,7 +3634,14 @@ 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); @@ -7389,7 +7397,7 @@ int amdgpu_dm_connector_atomic_set_property(struct drm_connector *connector, default: dm_new_state->abm_sysfs_forbidden = true; dm_new_state->abm_level = val; - }; + } ret = 0; } @@ -8216,7 +8224,7 @@ 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; } @@ -8690,7 +8698,7 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector) amdgpu_dm_connector->num_modes += drm_add_modes_noedid(connector, 1920, 1080); - if (amdgpu_dm_connector->dc_sink->edid_caps.analog) { + if (amdgpu_dm_connector->dc_sink && amdgpu_dm_connector->dc_sink->edid_caps.analog) { /* Analog monitor connected by DAC load detection. * Add common modes. It will be up to the user to select one that works. */ 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 cb4bb67289a4..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 @@ -1303,7 +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/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index b720e007c654..f519e5893a68 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3389,7 +3389,11 @@ 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; } /** 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 f95cb0cf4b8a..a7ec633b26c0 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 @@ -38,6 +38,7 @@ #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 @@ -764,7 +765,9 @@ void hwss_build_fast_sequence(struct dc *dc, 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); + 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)++; } diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 010d9315b96b..75b25b2506a8 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -63,7 +63,7 @@ struct dcn_dsc_reg_state; struct dcn_optc_reg_state; struct dcn_dccg_reg_state; -#define DC_VER "3.2.356" +#define DC_VER "3.2.357" /** * MAX_SURFACES - representative of the upper bound of surfaces that can be piped to a single CRTC 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 39f5fa73c43e..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 @@ -61,31 +61,49 @@ 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_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) { - /* ASIC doesn't support DMUB */ - if (!dc->ctx->dmub_srv) + if (!link) return false; - if (link) { + if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) + return true; - if (link->psr_settings.psr_version == DC_PSR_VERSION_SU_1) - return true; + if (link->replay_settings.replay_feature_enabled) + return true; - if (link->replay_settings.replay_feature_enabled) - return true; + if (link->psr_settings.psr_version == DC_PSR_VERSION_1) { + struct dc_link *edp_links[MAX_NUM_EDP]; + int edp_num; - /* 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(dc, edp_links, &edp_num); + if (edp_num == 1) + return true; + } + return false; +} - dc_get_edp_links(dc, edp_links, &edp_num); +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 (edp_num == 1) - return true; - } + 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 9f53d2ea5fa5..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 @@ -46,5 +46,7 @@ void dmub_hw_lock_mgr_inbox0_cmd(struct dc_dmub_srv *dmub_srv, * 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/dml2_0/dml21/inc/dml_top_dchub_registers.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h index 8e5a30287220..bf57df42d1d9 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_dchub_registers.h @@ -121,6 +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_0/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 13749c9fcf18..35aa954248cd 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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 @@ -87,6 +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, @@ -393,8 +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 vactive_latency_to_hide_for_pstate_admissibility_us; + 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; @@ -423,7 +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 { @@ -489,7 +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_0/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 4a9a0d5a09b7..1fbc520c2540 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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; @@ -191,7 +191,7 @@ 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_ns; + 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 */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h index 8646ce5f1c01..452e4a2e72c0 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2_0/dml21/inc/dml_top_types.h @@ -195,14 +195,6 @@ struct dml2_mcache_surface_allocation { } informative; }; -enum dml2_pstate_type { - dml2_pstate_type_uclk, - dml2_pstate_type_ppt, - dml2_pstate_type_temp_read, - dml2_pstate_type_dummy_pstate, - dml2_pstate_type_count -}; - enum dml2_pstate_method { dml2_pstate_method_na = 0, /* hw exclusive modes */ @@ -684,6 +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_0/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 f809c4073b43..df81bd963bb8 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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 @@ -6972,7 +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)); @@ -7069,9 +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]; } } } @@ -9051,11 +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); @@ -9063,8 +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); @@ -9506,8 +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, @@ -9515,7 +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); @@ -11009,11 +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); @@ -11021,8 +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); @@ -12943,7 +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[plane_idx][dml2_pstate_type_uclk]); } 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) @@ -13024,7 +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; diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/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 051c31ec2f0e..1087a8c926ff 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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 @@ -264,8 +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; @@ -336,7 +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; @@ -591,7 +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]; @@ -621,8 +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]; @@ -646,7 +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. @@ -837,6 +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 // ------------------- @@ -963,11 +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]; @@ -1134,8 +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]; @@ -1226,8 +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]; @@ -1313,7 +1317,7 @@ struct dml2_core_calcs_CalculateVMRowAndSwath_params { unsigned int HostVMMinPageSize; unsigned int DCCMetaBufferSizeBytes; bool mrq_present; - enum dml2_pstate_method *pstate_switch_modes; + enum dml2_pstate_method *uclk_pstate_switch_modes; // Output bool *PTEBufferSizeNotExceeded; @@ -1740,10 +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; @@ -1754,10 +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; @@ -2247,7 +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; diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/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_0/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_0/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 5769c2638f9a..c26e100fcaf2 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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 @@ -1087,7 +1087,7 @@ static bool all_timings_support_drr(const struct dml2_pmo_instance *pmo, /* check required stretch is allowed */ if (stream_descriptor->timing.drr_config.max_instant_vtotal_delta > 0 && - stream_pstate_meta->method_drr.stretched_vtotal - stream_pstate_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; } } @@ -1669,15 +1669,15 @@ 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]; } } @@ -1962,9 +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++) { @@ -1997,7 +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; - } } } @@ -2063,7 +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; - } } } @@ -2078,6 +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; @@ -2099,7 +2095,7 @@ 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 = + 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); } } @@ -2120,7 +2116,7 @@ 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 = + 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); } } diff --git a/drivers/gpu/drm/amd/display/dc/dml2_0/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 9f562f0c4797..1a6c0727cd2a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2_0/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 { @@ -257,8 +258,8 @@ struct dml2_implicit_svp_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; @@ -268,39 +269,44 @@ struct dml2_pstate_per_method_common_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 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; + 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_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_pstate_per_method_common_meta common; } method_subvp; struct { - unsigned int programming_delay_otg_vlines; - unsigned int stretched_vtotal; + int programming_delay_otg_vlines; + int stretched_vtotal; struct dml2_pstate_per_method_common_meta common; } method_drr; }; 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 c417780f37bc..5d287874c125 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c @@ -1257,6 +1257,7 @@ static bool detect_link_and_local_sink(struct dc_link *link, !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; 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 81e64e17d0cb..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 @@ -1847,7 +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) { @@ -2200,7 +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; 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 130058d7a70c..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 @@ -1671,7 +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) { 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 772e07a1a959..9bc512a522e0 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -599,6 +599,104 @@ 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; @@ -1732,9 +1830,15 @@ enum dmub_cmd_type { DMUB_CMD__CURSOR_OFFLOAD = 92, /** - * Command type used for all SMART_POWER_HDR commands. + * 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__SMART_POWER_HDR = 93, + DMUB_CMD__PR = 94, + /** * Command type use for VBIOS shared commands. @@ -4148,6 +4252,33 @@ 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 { @@ -4197,6 +4328,25 @@ 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. */ @@ -4349,17 +4499,13 @@ 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]; }; /** @@ -4405,11 +4551,11 @@ enum replay_enable { }; /** - * Data passed from driver to FW in a DMUB_CMD__SMART_POWER_HDR_ENABLE command. + * Data passed from driver to FW in a DMUB_CMD__SMART_POWER_OLED_ENABLE command. */ -struct dmub_rb_cmd_smart_power_hdr_enable_data { +struct dmub_rb_cmd_smart_power_oled_enable_data { /** - * SMART_POWER_HDR enable or disable. + * SMART_POWER_OLED enable or disable. */ uint8_t enable; /** @@ -4777,53 +4923,53 @@ union dmub_replay_cmd_set { }; /** - * SMART POWER HDR command sub-types. + * SMART POWER OLED command sub-types. */ -enum dmub_cmd_smart_power_hdr_type { +enum dmub_cmd_smart_power_oled_type { /** - * Enable/Disable SMART_POWER_HDR. + * Enable/Disable SMART_POWER_OLED. */ - DMUB_CMD__SMART_POWER_HDR_ENABLE = 1, + DMUB_CMD__SMART_POWER_OLED_ENABLE = 1, /** - * Get current MaxCLL value if SMART POWER HDR is enabled. + * Get current MaxCLL value if SMART POWER OLED is enabled. */ - DMUB_CMD__SMART_POWER_HDR_GETMAXCLL = 2, + DMUB_CMD__SMART_POWER_OLED_GETMAXCLL = 2, }; /** - * Definition of a DMUB_CMD__SMART_POWER_HDR command. + * Definition of a DMUB_CMD__SMART_POWER_OLED command. */ -struct dmub_rb_cmd_smart_power_hdr_enable { +struct dmub_rb_cmd_smart_power_oled_enable { /** * Command header. */ struct dmub_cmd_header header; - struct dmub_rb_cmd_smart_power_hdr_enable_data data; + struct dmub_rb_cmd_smart_power_oled_enable_data data; }; -struct dmub_cmd_smart_power_hdr_getmaxcll_input { +struct dmub_cmd_smart_power_oled_getmaxcll_input { uint8_t panel_inst; uint8_t pad[3]; }; -struct dmub_cmd_smart_power_hdr_getmaxcll_output { +struct dmub_cmd_smart_power_oled_getmaxcll_output { uint16_t current_max_cll; uint8_t pad[2]; }; /** - * Definition of a DMUB_CMD__SMART_POWER_HDR command. + * Definition of a DMUB_CMD__SMART_POWER_OLED command. */ -struct dmub_rb_cmd_smart_power_hdr_getmaxcll { +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_HDR_GETMAXCLL command. + * Data passed from driver to FW in a DMUB_CMD__SMART_POWER_OLED_GETMAXCLL command. */ - union dmub_cmd_smart_power_hdr_getmaxcll_data { - struct dmub_cmd_smart_power_hdr_getmaxcll_input input; /**< Input */ - struct dmub_cmd_smart_power_hdr_getmaxcll_output output; /**< Output */ + 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; }; @@ -6357,6 +6503,223 @@ struct dmub_rb_cmd_cursor_offload_stream_cntl { }; /** + * 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 { @@ -6698,13 +7061,25 @@ union dmub_rb_cmd { */ struct dmub_rb_cmd_cursor_offload_stream_cntl cursor_offload_stream_ctnl; /** - * Definition of a DMUB_CMD__SMART_POWER_HDR_ENABLE command. + * Definition of a DMUB_CMD__SMART_POWER_OLED_ENABLE command. */ - struct dmub_rb_cmd_smart_power_hdr_enable smart_power_hdr_enable; + struct dmub_rb_cmd_smart_power_oled_enable smart_power_oled_enable; /** - * Definition of a DMUB_CMD__DMUB_CMD__SMART_POWER_HDR_GETMAXCLL command. + * Definition of a DMUB_CMD__DMUB_CMD__SMART_POWER_OLED_GETMAXCLL command. */ - struct dmub_rb_cmd_smart_power_hdr_getmaxcll smart_power_hdr_getmaxcll; + 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_srv.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_srv.c index a657efda89ce..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) diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp.h index b883d626f1c3..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; 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 5628f0ef73fd..27500abf9fee 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp2_execution.c @@ -465,54 +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; @@ -524,9 +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..ac44ee1532fd 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_log.c @@ -248,6 +248,8 @@ char *mod_hdcp_status_to_str(int32_t status) 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"; + case MOD_HDCP_STATUS_HDCP2_LOCALITY_COMBO_READ_FAILURE: + return "MOD_HDCP_STATUS_HDCP2_LOCALITY_COMBO_READ_FAILURE"; default: return "MOD_HDCP_STATUS_UNKNOWN"; } 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 46e52fb3a118..264348989e9b 100644 --- a/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h +++ b/drivers/gpu/drm/amd/display/modules/inc/mod_hdcp.h @@ -98,6 +98,7 @@ enum mod_hdcp_status { 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_HDCP2_LOCALITY_COMBO_READ_FAILURE, }; struct mod_hdcp_displayport { @@ -214,8 +215,9 @@ 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 { @@ -317,10 +319,6 @@ 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/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index f92f78d5d330..2366e68262e6 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -1820,4 +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/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c index 5d08dc3b7110..c6f55d3522cd 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; @@ -2140,3 +2122,10 @@ 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 c88a76cce401..40ffaced74fd 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_pm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_pm.c @@ -4723,14 +4723,14 @@ 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 3bce74f8bb0a..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); @@ -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 3a9522c17fee..020e05c137e4 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; } } @@ -7509,8 +7504,6 @@ 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; @@ -7521,7 +7514,6 @@ 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); 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/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 4317da6f7c38..c0e7c45ac0e6 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -2054,6 +2054,12 @@ 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. @@ -2797,6 +2803,17 @@ 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; 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 c48028abc8c4..8815fc70b63b 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h @@ -1531,6 +1531,12 @@ struct pptable_funcs { 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 { @@ -1795,6 +1801,7 @@ 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); 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..fa43d2e229a0 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; 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..fe1b3ac50a75 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 @@ -105,19 +105,15 @@ #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 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..3a3930ef7ed9 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 +279,14 @@ __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), #undef __SMU_DUMMY_MAP #define __SMU_DUMMY_MAP(type) SMU_MSG_##type @@ -458,7 +465,8 @@ 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/navi10_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c index bbf09aec9152..7c9f77124ab2 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/navi10_ppt.c @@ -2889,7 +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/vangogh_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c index 53579208cffb..9626da2dba58 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/vangogh_ppt.c @@ -2219,6 +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)) { 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..0ce8cff27bf9 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 +141,13 @@ 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), }; int smu_v13_0_12_tables_init(struct smu_context *smu) @@ -449,7 +458,7 @@ 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 +728,14 @@ 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 +780,17 @@ 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 +880,184 @@ 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); + + 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_6_ppt.c b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_6_ppt.c index 0a7d2cea7dc6..dd8c7b98ce7e 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 @@ -549,7 +549,7 @@ 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; @@ -579,24 +579,28 @@ 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); @@ -732,6 +736,7 @@ 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); } @@ -766,7 +771,7 @@ 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; @@ -845,7 +850,7 @@ 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; @@ -1394,7 +1399,7 @@ 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 / @@ -1514,9 +1519,13 @@ 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); @@ -1528,9 +1537,13 @@ 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); @@ -1542,9 +1555,13 @@ 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); @@ -1556,9 +1573,13 @@ 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); @@ -1570,9 +1591,13 @@ 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; } @@ -1847,7 +1872,7 @@ 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; @@ -2385,7 +2410,7 @@ 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); @@ -2629,7 +2654,7 @@ 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; @@ -2649,8 +2674,8 @@ 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) @@ -2742,18 +2767,16 @@ 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); @@ -2761,16 +2784,20 @@ 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 */ @@ -2891,55 +2918,49 @@ 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]); } } } @@ -2949,7 +2970,8 @@ 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); } @@ -3883,6 +3905,29 @@ 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, @@ -3942,6 +3987,7 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { .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..367102cdbf09 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 @@ -72,9 +72,17 @@ enum smu_v13_0_6_caps { SMU_CAP(PLDM_VERSION), SMU_CAP(TEMP_METRICS), SMU_CAP(NPM_METRICS), + SMU_CAP(RAS_EEPROM), 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 +95,6 @@ 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 +106,156 @@ 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/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index a8961a8f5c42..3b98065dac1d 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -980,7 +980,7 @@ 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 +992,7 @@ 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/ras_mgr/amdgpu_ras_eeprom_i2c.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_eeprom_i2c.c index 1bb7b7001ec7..3ed3ff42b7e1 100644 --- 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 @@ -85,6 +85,7 @@ static int ras_eeprom_i2c_config(struct ras_core_context *ras_core) 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; 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 index 8007e49951d8..afe8135b6258 100644 --- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.c @@ -37,7 +37,7 @@ #define MAX_XCD_NUM_PER_AID 2 /* typical ECC bad page rate is 1 bad page per 100MB VRAM */ -#define ESTIMATE_BAD_PAGE_THRESHOLD(size) ((size)/(100 * 1024 * 1024ULL)) +#define TYPICAL_ECC_BAD_PAGE_RATE (100ULL * SZ_1M) #define COUNT_BAD_PAGE_THRESHOLD(size) (((size) >> 21) << 4) @@ -129,7 +129,7 @@ static int amdgpu_ras_mgr_init_eeprom_config(struct amdgpu_device *adev, */ if (amdgpu_bad_page_threshold == NONSTOP_OVER_THRESHOLD) eeprom_cfg->eeprom_record_threshold_count = - ESTIMATE_BAD_PAGE_THRESHOLD(adev->gmc.mc_vram_size); + 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); @@ -172,12 +172,13 @@ static int amdgpu_ras_mgr_init_nbio_config(struct amdgpu_device *adev, 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->mp1_ip_version); + config->nbio_ip_version); ret = -EINVAL; break; } @@ -258,7 +259,8 @@ static struct ras_core_context *amdgpu_ras_mgr_create_ras_core(struct amdgpu_dev 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)) + 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; @@ -283,6 +285,18 @@ static int amdgpu_ras_mgr_sw_init(struct amdgpu_ip_block *ip_block) 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; @@ -315,6 +329,9 @@ static int amdgpu_ras_mgr_sw_fini(struct amdgpu_ip_block *ip_block) 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; @@ -332,12 +349,11 @@ static int amdgpu_ras_mgr_sw_fini(struct amdgpu_ip_block *ip_block) 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; - /* Currently only debug mode can enable the ras module - */ - if (!adev->debug_enable_ras_aca) + if (!con->uniras_enabled) return 0; if (!ras_mgr || !ras_mgr->ras_core) @@ -360,11 +376,10 @@ static int amdgpu_ras_mgr_hw_init(struct amdgpu_ip_block *ip_block) 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); - /* Currently only debug mode can enable the ras module - */ - if (!adev->debug_enable_ras_aca) + if (!con->uniras_enabled) return 0; if (!ras_mgr || !ras_mgr->ras_core) @@ -609,3 +624,25 @@ int amdgpu_ras_mgr_handle_ras_cmd(struct amdgpu_device *adev, 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 index 42f190a8feb9..8fb7eb4b8f13 100644 --- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_mgr.h @@ -52,6 +52,9 @@ struct amdgpu_ras_mgr { 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; @@ -75,4 +78,6 @@ 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_process.c b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c index 6727fc9a2b9b..5782c007de71 100644 --- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.c @@ -29,6 +29,7 @@ #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) { @@ -57,6 +58,9 @@ 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; @@ -66,6 +70,7 @@ 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); @@ -124,3 +129,62 @@ int amdgpu_ras_process_handle_consumption_interrupt(struct amdgpu_device *adev, 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 index b9502bd21beb..d55cdaeac441 100644 --- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.h +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_process.h @@ -34,4 +34,8 @@ 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 index f21cd55a25be..45ed8c3b5563 100644 --- a/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_sys.c +++ b/drivers/gpu/drm/amd/ras/ras_mgr/amdgpu_ras_sys.c @@ -142,6 +142,12 @@ static int amdgpu_ras_sys_event_notifier(struct ras_core_context *ras_core, 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; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras.h b/drivers/gpu/drm/amd/ras/rascore/ras.h index fa224b36e3f2..3396b2e0949d 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras.h +++ b/drivers/gpu/drm/amd/ras/rascore/ras.h @@ -115,6 +115,8 @@ enum ras_notify_event { 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 { diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_cper.c b/drivers/gpu/drm/amd/ras/rascore/ras_cper.c index 2343991adccf..0fc7522b7ab6 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras_cper.c +++ b/drivers/gpu/drm/amd/ras/rascore/ras_cper.c @@ -54,7 +54,7 @@ static void fill_section_hdr(struct ras_core_context *ras_core, enum ras_cper_severity sev, struct ras_log_info *trace) { struct device_system_info dev_info = {0}; - char record_id[16]; + char record_id[32]; hdr->signature[0] = 'C'; hdr->signature[1] = 'P'; @@ -62,17 +62,16 @@ static void fill_section_hdr(struct ras_core_context *ras_core, hdr->signature[3] = 'R'; hdr->revision = CPER_HDR__REV_1; hdr->signature_end = 0xFFFFFFFF; - hdr->error_severity = sev; + hdr->error_severity = (sev == RAS_CPER_SEV_RMA ? RAS_CPER_SEV_FATAL_UE : sev); hdr->valid_bits.platform_id = 1; - hdr->valid_bits.partition_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, 9, "%d:%llX", dev_info.socket_id, + 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); @@ -116,7 +115,7 @@ static int fill_section_descriptor(struct ras_core_context *ras_core, descriptor->sec_length = section_length; descriptor->valid_bits.fru_text = 1; descriptor->flag_bits.primary = 1; - descriptor->severity = sev; + 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); @@ -147,13 +146,19 @@ static int fill_section_fatal(struct ras_core_context *ras_core, } static int fill_section_runtime(struct ras_core_context *ras_core, - struct cper_section_runtime *runtime, struct ras_log_info *trace) + 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); @@ -189,7 +194,7 @@ static int cper_generate_runtime_record(struct ras_core_context *ras_core, 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]); + fill_section_runtime(ras_core, runtime, trace_arr[i], sev); } return 0; diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c index d0621464f1a7..0a838fdcb2f6 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c +++ b/drivers/gpu/drm/amd/ras/rascore/ras_log_ring.c @@ -226,7 +226,9 @@ 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); @@ -252,8 +254,13 @@ void ras_log_ring_add_log_event(struct ras_core_context *ras_core, if (data) memcpy(&log->aca_reg, data, sizeof(log->aca_reg)); - if (event == RAS_LOG_EVENT_RMA) + 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); } diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c index 8bf1f35d595e..bfddd104d548 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c +++ b/drivers/gpu/drm/amd/ras/rascore/ras_nbio.c @@ -31,6 +31,7 @@ static const struct ras_nbio_ip_func *ras_nbio_get_ip_funcs( { 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, diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_process.c b/drivers/gpu/drm/amd/ras/rascore/ras_process.c index 02f0657f78a3..3267dcdb169c 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras_process.c +++ b/drivers/gpu/drm/amd/ras/rascore/ras_process.c @@ -162,6 +162,11 @@ int ras_process_handle_ras_event(struct ras_core_context *ras_core) 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); @@ -185,6 +190,8 @@ int ras_process_handle_ras_event(struct ras_core_context *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; } diff --git a/drivers/gpu/drm/amd/ras/rascore/ras_umc.c b/drivers/gpu/drm/amd/ras/rascore/ras_umc.c index 4067359bb299..4dae64c424a2 100644 --- a/drivers/gpu/drm/amd/ras/rascore/ras_umc.c +++ b/drivers/gpu/drm/amd/ras/rascore/ras_umc.c @@ -71,6 +71,7 @@ static const struct ras_umc_ip_func *ras_umc_get_ip_func( { 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, |