diff options
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c')
| -rw-r--r-- | drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c | 193 |
1 files changed, 108 insertions, 85 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c index 85967a5570cb..27c76bd424cf 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v2_0.c @@ -32,43 +32,40 @@ #include "vcn/vcn_2_0_0_sh_mask.h" #include "ivsrcid/vcn/irqsrcs_vcn_2_0.h" -#define mmUVD_JRBC_EXTERNAL_REG_INTERNAL_OFFSET 0x1bfff -#define mmUVD_JPEG_GPCOM_CMD_INTERNAL_OFFSET 0x4029 -#define mmUVD_JPEG_GPCOM_DATA0_INTERNAL_OFFSET 0x402a -#define mmUVD_JPEG_GPCOM_DATA1_INTERNAL_OFFSET 0x402b -#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ea -#define mmUVD_LMI_JRBC_RB_MEM_WR_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40eb -#define mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET 0x40cf -#define mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET 0x40d1 -#define mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40e8 -#define mmUVD_LMI_JRBC_IB_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40e9 -#define mmUVD_JRBC_IB_SIZE_INTERNAL_OFFSET 0x4082 -#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_LOW_INTERNAL_OFFSET 0x40ec -#define mmUVD_LMI_JRBC_RB_MEM_RD_64BIT_BAR_HIGH_INTERNAL_OFFSET 0x40ed -#define mmUVD_JRBC_RB_COND_RD_TIMER_INTERNAL_OFFSET 0x4085 -#define mmUVD_JRBC_RB_REF_DATA_INTERNAL_OFFSET 0x4084 -#define mmUVD_JRBC_STATUS_INTERNAL_OFFSET 0x4089 -#define mmUVD_JPEG_PITCH_INTERNAL_OFFSET 0x401f - -#define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR 0x18000 +static const struct amdgpu_hwip_reg_entry jpeg_reg_list_2_0[] = { + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_POWER_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_INT_STAT), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_RPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_WPTR), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_CNTL), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_RB_SIZE), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JRBC_STATUS), + SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_ADDR_MODE), + SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_GFX10_ADDR_CONFIG), + SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_Y_GFX10_TILING_SURFACE), + SOC15_REG_ENTRY_STR(JPEG, 0, mmJPEG_DEC_UV_GFX10_TILING_SURFACE), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_PITCH), + SOC15_REG_ENTRY_STR(JPEG, 0, mmUVD_JPEG_UV_PITCH), +}; static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev); static void jpeg_v2_0_set_irq_funcs(struct amdgpu_device *adev); -static int jpeg_v2_0_set_powergating_state(void *handle, +static int jpeg_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state); /** * jpeg_v2_0_early_init - set function pointers * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Set ring and irq function pointers */ -static int jpeg_v2_0_early_init(void *handle) +static int jpeg_v2_0_early_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; adev->jpeg.num_jpeg_inst = 1; + adev->jpeg.num_jpeg_rings = 1; jpeg_v2_0_set_dec_ring_funcs(adev); jpeg_v2_0_set_irq_funcs(adev); @@ -79,13 +76,13 @@ static int jpeg_v2_0_early_init(void *handle) /** * jpeg_v2_0_sw_init - sw init for JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Load firmware and sw initialization */ -static int jpeg_v2_0_sw_init(void *handle) +static int jpeg_v2_0_sw_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; struct amdgpu_ring *ring; int r; @@ -103,37 +100,50 @@ static int jpeg_v2_0_sw_init(void *handle) if (r) return r; - ring = &adev->jpeg.inst->ring_dec; + ring = adev->jpeg.inst->ring_dec; ring->use_doorbell = true; ring->doorbell_index = (adev->doorbell_index.vcn.vcn_ring0_1 << 1) + 1; + ring->vm_hub = AMDGPU_MMHUB0(0); sprintf(ring->name, "jpeg_dec"); r = amdgpu_ring_init(adev, ring, 512, &adev->jpeg.inst->irq, 0, AMDGPU_RING_PRIO_DEFAULT, NULL); if (r) return r; - adev->jpeg.internal.jpeg_pitch = mmUVD_JPEG_PITCH_INTERNAL_OFFSET; - adev->jpeg.inst->external.jpeg_pitch = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH); + adev->jpeg.internal.jpeg_pitch[0] = mmUVD_JPEG_PITCH_INTERNAL_OFFSET; + adev->jpeg.inst->external.jpeg_pitch[0] = SOC15_REG_OFFSET(JPEG, 0, mmUVD_JPEG_PITCH); - return 0; + r = amdgpu_jpeg_reg_dump_init(adev, jpeg_reg_list_2_0, ARRAY_SIZE(jpeg_reg_list_2_0)); + if (r) + return r; + + adev->jpeg.supported_reset = + amdgpu_get_soft_full_reset_mask(adev->jpeg.inst[0].ring_dec); + if (!amdgpu_sriov_vf(adev)) + adev->jpeg.supported_reset |= AMDGPU_RESET_TYPE_PER_QUEUE; + r = amdgpu_jpeg_sysfs_reset_mask_init(adev); + + return r; } /** * jpeg_v2_0_sw_fini - sw fini for JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * JPEG suspend and free up sw allocation */ -static int jpeg_v2_0_sw_fini(void *handle) +static int jpeg_v2_0_sw_fini(struct amdgpu_ip_block *ip_block) { int r; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; r = amdgpu_jpeg_suspend(adev); if (r) return r; + amdgpu_jpeg_sysfs_reset_mask_fini(adev); + r = amdgpu_jpeg_sw_fini(adev); return r; @@ -142,41 +152,36 @@ static int jpeg_v2_0_sw_fini(void *handle) /** * jpeg_v2_0_hw_init - start and test JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * */ -static int jpeg_v2_0_hw_init(void *handle) +static int jpeg_v2_0_hw_init(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - struct amdgpu_ring *ring = &adev->jpeg.inst->ring_dec; - int r; + struct amdgpu_device *adev = ip_block->adev; + struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec; adev->nbio.funcs->vcn_doorbell_range(adev, ring->use_doorbell, (adev->doorbell_index.vcn.vcn_ring0_1 << 1), 0); - r = amdgpu_ring_test_helper(ring); - if (!r) - DRM_INFO("JPEG decode initialized successfully.\n"); - - return r; + return amdgpu_ring_test_helper(ring); } /** * jpeg_v2_0_hw_fini - stop the hardware block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Stop the JPEG block, mark ring as not ready any more */ -static int jpeg_v2_0_hw_fini(void *handle) +static int jpeg_v2_0_hw_fini(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; - cancel_delayed_work_sync(&adev->vcn.idle_work); + cancel_delayed_work_sync(&adev->jpeg.idle_work); if (adev->jpeg.cur_state != AMD_PG_STATE_GATE && RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS)) - jpeg_v2_0_set_powergating_state(adev, AMD_PG_STATE_GATE); + jpeg_v2_0_set_powergating_state(ip_block, AMD_PG_STATE_GATE); return 0; } @@ -184,20 +189,19 @@ static int jpeg_v2_0_hw_fini(void *handle) /** * jpeg_v2_0_suspend - suspend JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * HW fini and suspend JPEG block */ -static int jpeg_v2_0_suspend(void *handle) +static int jpeg_v2_0_suspend(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; int r; - r = jpeg_v2_0_hw_fini(adev); + r = jpeg_v2_0_hw_fini(ip_block); if (r) return r; - r = amdgpu_jpeg_suspend(adev); + r = amdgpu_jpeg_suspend(ip_block->adev); return r; } @@ -205,20 +209,19 @@ static int jpeg_v2_0_suspend(void *handle) /** * jpeg_v2_0_resume - resume JPEG block * - * @handle: amdgpu_device pointer + * @ip_block: Pointer to the amdgpu_ip_block for this hw instance. * * Resume firmware and hw init JPEG block */ -static int jpeg_v2_0_resume(void *handle) +static int jpeg_v2_0_resume(struct amdgpu_ip_block *ip_block) { int r; - struct amdgpu_device *adev = (struct amdgpu_device *)handle; - r = amdgpu_jpeg_resume(adev); + r = amdgpu_jpeg_resume(ip_block->adev); if (r) return r; - r = jpeg_v2_0_hw_init(adev); + r = jpeg_v2_0_hw_init(ip_block); return r; } @@ -331,7 +334,7 @@ static void jpeg_v2_0_enable_clock_gating(struct amdgpu_device *adev) */ static int jpeg_v2_0_start(struct amdgpu_device *adev) { - struct amdgpu_ring *ring = &adev->jpeg.inst->ring_dec; + struct amdgpu_ring *ring = adev->jpeg.inst->ring_dec; int r; if (adev->pm.dpm_enabled) @@ -427,7 +430,7 @@ static uint64_t jpeg_v2_0_dec_ring_get_wptr(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; if (ring->use_doorbell) - return adev->wb.wb[ring->wptr_offs]; + return *ring->wptr_cpu_addr; else return RREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR); } @@ -444,7 +447,7 @@ static void jpeg_v2_0_dec_ring_set_wptr(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; if (ring->use_doorbell) { - adev->wb.wb[ring->wptr_offs] = lower_32_bits(ring->wptr); + *ring->wptr_cpu_addr = lower_32_bits(ring->wptr); WDOORBELL32(ring->doorbell_index, lower_32_bits(ring->wptr)); } else { WREG32_SOC15(JPEG, 0, mmUVD_JRBC_RB_WPTR, lower_32_bits(ring->wptr)); @@ -555,13 +558,21 @@ void jpeg_v2_0_dec_ring_emit_ib(struct amdgpu_ring *ring, { unsigned vmid = AMDGPU_JOB_GET_VMID(job); + amdgpu_ring_write(ring, PACKETJ(mmUVD_JPEG_IH_CTRL_INTERNAL_OFFSET, + 0, 0, PACKETJ_TYPE0)); + amdgpu_ring_write(ring, (vmid << JPEG_IH_CTRL__IH_VMID__SHIFT)); + amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_VMID_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, (vmid | (vmid << 4))); + + if (ring->funcs->parse_cs) + amdgpu_ring_write(ring, 0); + else + amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8))); amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JPEG_VMID_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); - amdgpu_ring_write(ring, (vmid | (vmid << 4))); + amdgpu_ring_write(ring, (vmid | (vmid << 4) | (vmid << 8))); amdgpu_ring_write(ring, PACKETJ(mmUVD_LMI_JRBC_IB_64BIT_BAR_LOW_INTERNAL_OFFSET, 0, 0, PACKETJ_TYPE0)); @@ -629,7 +640,7 @@ void jpeg_v2_0_dec_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, void jpeg_v2_0_dec_ring_emit_vm_flush(struct amdgpu_ring *ring, unsigned vmid, uint64_t pd_addr) { - struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->funcs->vmhub]; + struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub]; uint32_t data0, data1, mask; pd_addr = amdgpu_gmc_emit_flush_gpu_tlb(ring, vmid, pd_addr); @@ -671,18 +682,18 @@ void jpeg_v2_0_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count) } } -static bool jpeg_v2_0_is_idle(void *handle) +static bool jpeg_v2_0_is_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; return ((RREG32_SOC15(JPEG, 0, mmUVD_JRBC_STATUS) & UVD_JRBC_STATUS__RB_JOB_DONE_MASK) == UVD_JRBC_STATUS__RB_JOB_DONE_MASK); } -static int jpeg_v2_0_wait_for_idle(void *handle) +static int jpeg_v2_0_wait_for_idle(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int ret; ret = SOC15_WAIT_ON_RREG(JPEG, 0, mmUVD_JRBC_STATUS, UVD_JRBC_STATUS__RB_JOB_DONE_MASK, @@ -691,14 +702,14 @@ static int jpeg_v2_0_wait_for_idle(void *handle) return ret; } -static int jpeg_v2_0_set_clockgating_state(void *handle, +static int jpeg_v2_0_set_clockgating_state(struct amdgpu_ip_block *ip_block, enum amd_clockgating_state state) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; bool enable = (state == AMD_CG_STATE_GATE); if (enable) { - if (!jpeg_v2_0_is_idle(handle)) + if (!jpeg_v2_0_is_idle(ip_block)) return -EBUSY; jpeg_v2_0_enable_clock_gating(adev); } else { @@ -708,10 +719,10 @@ static int jpeg_v2_0_set_clockgating_state(void *handle, return 0; } -static int jpeg_v2_0_set_powergating_state(void *handle, +static int jpeg_v2_0_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int ret; if (state == adev->jpeg.cur_state) @@ -744,7 +755,7 @@ static int jpeg_v2_0_process_interrupt(struct amdgpu_device *adev, switch (entry->src_id) { case VCN_2_0__SRCID__JPEG_DECODE: - amdgpu_fence_process(&adev->jpeg.inst->ring_dec); + amdgpu_fence_process(adev->jpeg.inst->ring_dec); break; default: DRM_ERROR("Unhandled interrupt: %d %d\n", @@ -755,10 +766,25 @@ static int jpeg_v2_0_process_interrupt(struct amdgpu_device *adev, return 0; } +static int jpeg_v2_0_ring_reset(struct amdgpu_ring *ring, + unsigned int vmid, + struct amdgpu_fence *timedout_fence) +{ + int r; + + amdgpu_ring_reset_helper_begin(ring, timedout_fence); + r = jpeg_v2_0_stop(ring->adev); + if (r) + return r; + r = jpeg_v2_0_start(ring->adev); + if (r) + return r; + return amdgpu_ring_reset_helper_end(ring, timedout_fence); +} + static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = { .name = "jpeg_v2_0", .early_init = jpeg_v2_0_early_init, - .late_init = NULL, .sw_init = jpeg_v2_0_sw_init, .sw_fini = jpeg_v2_0_sw_fini, .hw_init = jpeg_v2_0_hw_init, @@ -767,28 +793,26 @@ static const struct amd_ip_funcs jpeg_v2_0_ip_funcs = { .resume = jpeg_v2_0_resume, .is_idle = jpeg_v2_0_is_idle, .wait_for_idle = jpeg_v2_0_wait_for_idle, - .check_soft_reset = NULL, - .pre_soft_reset = NULL, - .soft_reset = NULL, - .post_soft_reset = NULL, .set_clockgating_state = jpeg_v2_0_set_clockgating_state, .set_powergating_state = jpeg_v2_0_set_powergating_state, + .dump_ip_state = amdgpu_jpeg_dump_ip_state, + .print_ip_state = amdgpu_jpeg_print_ip_state, }; static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = { .type = AMDGPU_RING_TYPE_VCN_JPEG, .align_mask = 0xf, - .vmhub = AMDGPU_MMHUB_0, .get_rptr = jpeg_v2_0_dec_ring_get_rptr, .get_wptr = jpeg_v2_0_dec_ring_get_wptr, .set_wptr = jpeg_v2_0_dec_ring_set_wptr, + .parse_cs = amdgpu_jpeg_dec_parse_cs, .emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 6 + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 8 + 8 + /* jpeg_v2_0_dec_ring_emit_vm_flush */ 18 + 18 + /* jpeg_v2_0_dec_ring_emit_fence x2 vm fence */ 8 + 16, - .emit_ib_size = 22, /* jpeg_v2_0_dec_ring_emit_ib */ + .emit_ib_size = 24, /* jpeg_v2_0_dec_ring_emit_ib */ .emit_ib = jpeg_v2_0_dec_ring_emit_ib, .emit_fence = jpeg_v2_0_dec_ring_emit_fence, .emit_vm_flush = jpeg_v2_0_dec_ring_emit_vm_flush, @@ -803,12 +827,12 @@ static const struct amdgpu_ring_funcs jpeg_v2_0_dec_ring_vm_funcs = { .emit_wreg = jpeg_v2_0_dec_ring_emit_wreg, .emit_reg_wait = jpeg_v2_0_dec_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, + .reset = jpeg_v2_0_ring_reset, }; static void jpeg_v2_0_set_dec_ring_funcs(struct amdgpu_device *adev) { - adev->jpeg.inst->ring_dec.funcs = &jpeg_v2_0_dec_ring_vm_funcs; - DRM_INFO("JPEG decode is enabled in VM mode\n"); + adev->jpeg.inst->ring_dec->funcs = &jpeg_v2_0_dec_ring_vm_funcs; } static const struct amdgpu_irq_src_funcs jpeg_v2_0_irq_funcs = { @@ -822,8 +846,7 @@ static void jpeg_v2_0_set_irq_funcs(struct amdgpu_device *adev) adev->jpeg.inst->irq.funcs = &jpeg_v2_0_irq_funcs; } -const struct amdgpu_ip_block_version jpeg_v2_0_ip_block = -{ +const struct amdgpu_ip_block_version jpeg_v2_0_ip_block = { .type = AMD_IP_BLOCK_TYPE_JPEG, .major = 2, .minor = 0, |