diff options
Diffstat (limited to 'drivers/gpu/drm')
136 files changed, 2278 insertions, 619 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/Kconfig b/drivers/gpu/drm/amd/amdgpu/Kconfig index 058e3b3ad520..1acfed2f92ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/Kconfig +++ b/drivers/gpu/drm/amd/amdgpu/Kconfig @@ -76,7 +76,7 @@ config DRM_AMDGPU_USERPTR config DRM_AMD_ISP bool "Enable AMD Image Signal Processor IP support" - depends on DRM_AMDGPU + depends on DRM_AMDGPU && ACPI select MFD_CORE select PM_GENERIC_DOMAINS if PM help diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 836ea081088a..a5ccd0ada16a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -1713,6 +1713,10 @@ static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { retu static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; } #endif +#if defined(CONFIG_DRM_AMD_ISP) +int amdgpu_acpi_get_isp4_dev_hid(u8 (*hid)[ACPI_ID_LEN]); +#endif + void amdgpu_register_gpu_instance(struct amdgpu_device *adev); void amdgpu_unregister_gpu_instance(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c index 707e131f89d2..f5466c592d94 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -1532,5 +1532,35 @@ bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) return true; #endif /* CONFIG_AMD_PMC */ } - #endif /* CONFIG_SUSPEND */ + +#if IS_ENABLED(CONFIG_DRM_AMD_ISP) +static const struct acpi_device_id isp_sensor_ids[] = { + { "OMNI5C10" }, + { } +}; + +static int isp_match_acpi_device_ids(struct device *dev, const void *data) +{ + return acpi_match_device(data, dev) ? 1 : 0; +} + +int amdgpu_acpi_get_isp4_dev_hid(u8 (*hid)[ACPI_ID_LEN]) +{ + struct device *pdev __free(put_device) = NULL; + struct acpi_device *acpi_pdev; + + pdev = bus_find_device(&platform_bus_type, NULL, isp_sensor_ids, + isp_match_acpi_device_ids); + if (!pdev) + return -EINVAL; + + acpi_pdev = ACPI_COMPANION(pdev); + if (!acpi_pdev) + return -ENODEV; + + strscpy(*hid, acpi_device_hid(acpi_pdev)); + + return 0; +} +#endif /* CONFIG_DRM_AMD_ISP */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c index 4cec3a873995..d8ac4b1051a8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c @@ -368,6 +368,9 @@ void amdgpu_amdkfd_free_gtt_mem(struct amdgpu_device *adev, void **mem_obj) { struct amdgpu_bo **bo = (struct amdgpu_bo **) mem_obj; + if (!bo || !*bo) + return; + (void)amdgpu_bo_reserve(*bo, true); amdgpu_bo_kunmap(*bo); amdgpu_bo_unpin(*bo); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c index c43d1b6e5d66..85567d0d9545 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c @@ -919,7 +919,7 @@ long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout) return timeout; } -void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr) +static void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr) { struct amdgpu_ctx *ctx; struct idr *idp; @@ -949,19 +949,7 @@ void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr) void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr) { - struct amdgpu_ctx *ctx; - struct idr *idp; - uint32_t id; - amdgpu_ctx_mgr_entity_fini(mgr); - - idp = &mgr->ctx_handles; - - idr_for_each_entry(idp, ctx, id) { - if (kref_put(&ctx->refcount, amdgpu_ctx_fini) != 1) - DRM_ERROR("ctx %p is still alive\n", ctx); - } - idr_destroy(&mgr->ctx_handles); mutex_destroy(&mgr->lock); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h index 85376baaa92f..090dfe86f75b 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h @@ -92,7 +92,6 @@ int amdgpu_ctx_wait_prev_fence(struct amdgpu_ctx *ctx, void amdgpu_ctx_mgr_init(struct amdgpu_ctx_mgr *mgr, struct amdgpu_device *adev); -void amdgpu_ctx_mgr_entity_fini(struct amdgpu_ctx_mgr *mgr); long amdgpu_ctx_mgr_entity_flush(struct amdgpu_ctx_mgr *mgr, long timeout); void amdgpu_ctx_mgr_fini(struct amdgpu_ctx_mgr *mgr); void amdgpu_ctx_mgr_usage(struct amdgpu_ctx_mgr *mgr, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 4d1b54f58495..e1bab6a96cb6 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -512,12 +512,13 @@ void amdgpu_device_detect_runtime_pm_mode(struct amdgpu_device *adev) break; case CHIP_VEGA10: /* enable BACO as runpm mode if noretry=0 */ - if (!adev->gmc.noretry) + if (!adev->gmc.noretry && !amdgpu_passthrough(adev)) adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; break; default: /* enable BACO as runpm mode on CI+ */ - adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; + if (!amdgpu_passthrough(adev)) + adev->pm.rpm_mode = AMDGPU_RUNPM_BACO; break; } @@ -4728,7 +4729,7 @@ fence_driver_init: amdgpu_fru_sysfs_init(adev); amdgpu_reg_state_sysfs_init(adev); - amdgpu_xcp_cfg_sysfs_init(adev); + amdgpu_xcp_sysfs_init(adev); if (IS_ENABLED(CONFIG_PERF_EVENTS)) r = amdgpu_pmu_init(adev); @@ -4858,7 +4859,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) amdgpu_fru_sysfs_fini(adev); amdgpu_reg_state_sysfs_fini(adev); - amdgpu_xcp_cfg_sysfs_fini(adev); + amdgpu_xcp_sysfs_fini(adev); /* disable ras feature must before hw fini */ amdgpu_ras_pre_fini(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 9e738fae2b74..a0e9bf9b2710 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -270,9 +270,10 @@ static int amdgpu_discovery_read_binary_from_sysmem(struct amdgpu_device *adev, static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, uint8_t *binary) { + bool sz_valid = true; uint64_t vram_size; - u32 msg; int i, ret = 0; + u32 msg; if (!amdgpu_sriov_vf(adev)) { /* It can take up to a second for IFWI init to complete on some dGPUs, @@ -291,9 +292,13 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, } } - vram_size = (uint64_t)RREG32(mmRCC_CONFIG_MEMSIZE) << 20; + vram_size = RREG32(mmRCC_CONFIG_MEMSIZE); + if (!vram_size || vram_size == U32_MAX) + sz_valid = false; + else + vram_size <<= 20; - if (vram_size) { + if (sz_valid) { uint64_t pos = vram_size - DISCOVERY_TMR_OFFSET; amdgpu_device_vram_access(adev, pos, (uint32_t *)binary, adev->mman.discovery_tmr_size, false); @@ -301,6 +306,11 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, ret = amdgpu_discovery_read_binary_from_sysmem(adev, binary); } + if (ret) + dev_err(adev->dev, + "failed to read discovery info from memory, vram size read: %llx", + vram_size); + return ret; } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 4ddd08ce8885..4db92e0a60da 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -2913,8 +2913,8 @@ static int amdgpu_drm_release(struct inode *inode, struct file *filp) if (fpriv) { fpriv->evf_mgr.fd_closing = true; - amdgpu_userq_mgr_fini(&fpriv->userq_mgr); amdgpu_eviction_fence_destroy(&fpriv->evf_mgr); + amdgpu_userq_mgr_fini(&fpriv->userq_mgr); } return drm_release(inode, filp); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c index 73b629b5f56f..8b919ad3af29 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_eviction_fence.c @@ -108,13 +108,22 @@ amdgpu_eviction_fence_suspend_worker(struct work_struct *work) struct amdgpu_eviction_fence *ev_fence; mutex_lock(&uq_mgr->userq_mutex); + spin_lock(&evf_mgr->ev_fence_lock); ev_fence = evf_mgr->ev_fence; - if (!ev_fence) + if (ev_fence) + dma_fence_get(&ev_fence->base); + else goto unlock; + spin_unlock(&evf_mgr->ev_fence_lock); amdgpu_userq_evict(uq_mgr, ev_fence); + mutex_unlock(&uq_mgr->userq_mutex); + dma_fence_put(&ev_fence->base); + return; + unlock: + spin_unlock(&evf_mgr->ev_fence_lock); mutex_unlock(&uq_mgr->userq_mutex); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c index 2c68118fe9fd..0ecc88df7208 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c @@ -58,7 +58,7 @@ amdgpu_gem_add_input_fence(struct drm_file *filp, return 0; syncobj_handles = memdup_user(u64_to_user_ptr(syncobj_handles_array), - sizeof(uint32_t) * num_syncobj_handles); + size_mul(sizeof(uint32_t), num_syncobj_handles)); if (IS_ERR(syncobj_handles)) return PTR_ERR(syncobj_handles); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index 1db1e6ec0184..c5646af055ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -2228,6 +2228,9 @@ void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring) enum PP_SMC_POWER_PROFILE profile; int r; + if (amdgpu_dpm_is_overdrive_enabled(adev)) + return; + if (adev->gfx.num_gfx_rings) profile = PP_SMC_POWER_PROFILE_FULLSCREEN3D; else @@ -2258,6 +2261,11 @@ void amdgpu_gfx_profile_ring_begin_use(struct amdgpu_ring *ring) void amdgpu_gfx_profile_ring_end_use(struct amdgpu_ring *ring) { + struct amdgpu_device *adev = ring->adev; + + if (amdgpu_dpm_is_overdrive_enabled(adev)) + return; + atomic_dec(&ring->adev->gfx.total_submission_cnt); schedule_delayed_work(&ring->adev->gfx.idle_work, GFX_PROFILE_IDLE_TIMEOUT); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c index 9fbb04aee97b..d2ce7d86dbc8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c @@ -1502,11 +1502,6 @@ void amdgpu_driver_postclose_kms(struct drm_device *dev, amdgpu_bo_unreserve(pd); } - if (!fpriv->evf_mgr.fd_closing) { - fpriv->evf_mgr.fd_closing = true; - amdgpu_userq_mgr_fini(&fpriv->userq_mgr); - amdgpu_eviction_fence_destroy(&fpriv->evf_mgr); - } amdgpu_ctx_mgr_fini(&fpriv->ctx_mgr); amdgpu_vm_fini(adev, &fpriv->vm); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index 2febb63ab232..6fa9fa11c8f3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -300,7 +300,9 @@ int amdgpu_mes_map_legacy_queue(struct amdgpu_device *adev, queue_input.mqd_addr = amdgpu_bo_gpu_offset(ring->mqd_obj); queue_input.wptr_addr = ring->wptr_gpu_addr; + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->map_legacy_queue(&adev->mes, &queue_input); + amdgpu_mes_unlock(&adev->mes); if (r) DRM_ERROR("failed to map legacy queue\n"); @@ -323,7 +325,9 @@ int amdgpu_mes_unmap_legacy_queue(struct amdgpu_device *adev, queue_input.trail_fence_addr = gpu_addr; queue_input.trail_fence_data = seq; + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->unmap_legacy_queue(&adev->mes, &queue_input); + amdgpu_mes_unlock(&adev->mes); if (r) DRM_ERROR("failed to unmap legacy queue\n"); @@ -353,7 +357,9 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev, if (ring->funcs->type == AMDGPU_RING_TYPE_GFX) queue_input.legacy_gfx = true; + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->reset_hw_queue(&adev->mes, &queue_input); + amdgpu_mes_unlock(&adev->mes); if (r) DRM_ERROR("failed to reset legacy queue\n"); @@ -383,7 +389,9 @@ uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg) goto error; } + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->misc_op(&adev->mes, &op_input); + amdgpu_mes_unlock(&adev->mes); if (r) dev_err(adev->dev, "failed to read reg (0x%x)\n", reg); else @@ -411,7 +419,9 @@ int amdgpu_mes_wreg(struct amdgpu_device *adev, goto error; } + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->misc_op(&adev->mes, &op_input); + amdgpu_mes_unlock(&adev->mes); if (r) dev_err(adev->dev, "failed to write reg (0x%x)\n", reg); @@ -438,32 +448,9 @@ int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev, goto error; } + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->misc_op(&adev->mes, &op_input); - if (r) - dev_err(adev->dev, "failed to reg_write_reg_wait\n"); - -error: - return r; -} - -int amdgpu_mes_reg_wait(struct amdgpu_device *adev, uint32_t reg, - uint32_t val, uint32_t mask) -{ - struct mes_misc_op_input op_input; - int r; - - op_input.op = MES_MISC_OP_WRM_REG_WAIT; - op_input.wrm_reg.reg0 = reg; - op_input.wrm_reg.ref = val; - op_input.wrm_reg.mask = mask; - - if (!adev->mes.funcs->misc_op) { - dev_err(adev->dev, "mes reg wait is not supported!\n"); - r = -EINVAL; - goto error; - } - - r = adev->mes.funcs->misc_op(&adev->mes, &op_input); + amdgpu_mes_unlock(&adev->mes); if (r) dev_err(adev->dev, "failed to reg_write_reg_wait\n"); @@ -539,42 +526,6 @@ int amdgpu_mes_flush_shader_debugger(struct amdgpu_device *adev, return r; } -#define DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(_eng) \ -do { \ - if (id_offs < AMDGPU_MES_CTX_MAX_OFFS) \ - return offsetof(struct amdgpu_mes_ctx_meta_data, \ - _eng[ring->idx].slots[id_offs]); \ - else if (id_offs == AMDGPU_MES_CTX_RING_OFFS) \ - return offsetof(struct amdgpu_mes_ctx_meta_data, \ - _eng[ring->idx].ring); \ - else if (id_offs == AMDGPU_MES_CTX_IB_OFFS) \ - return offsetof(struct amdgpu_mes_ctx_meta_data, \ - _eng[ring->idx].ib); \ - else if (id_offs == AMDGPU_MES_CTX_PADDING_OFFS) \ - return offsetof(struct amdgpu_mes_ctx_meta_data, \ - _eng[ring->idx].padding); \ -} while(0) - -int amdgpu_mes_ctx_get_offs(struct amdgpu_ring *ring, unsigned int id_offs) -{ - switch (ring->funcs->type) { - case AMDGPU_RING_TYPE_GFX: - DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(gfx); - break; - case AMDGPU_RING_TYPE_COMPUTE: - DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(compute); - break; - case AMDGPU_RING_TYPE_SDMA: - DEFINE_AMDGPU_MES_CTX_GET_OFFS_ENG(sdma); - break; - default: - break; - } - - WARN_ON(1); - return -EINVAL; -} - uint32_t amdgpu_mes_get_aggregated_doorbell_index(struct amdgpu_device *adev, enum amdgpu_mes_priority_level prio) { @@ -694,7 +645,9 @@ static int amdgpu_mes_set_enforce_isolation(struct amdgpu_device *adev, goto error; } + amdgpu_mes_lock(&adev->mes); r = adev->mes.funcs->misc_op(&adev->mes, &op_input); + amdgpu_mes_unlock(&adev->mes); if (r) dev_err(adev->dev, "failed to change_config.\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h index a41f65b4f733..c0d2c195fe2e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.h @@ -372,8 +372,6 @@ struct amdgpu_mes_funcs { #define amdgpu_mes_kiq_hw_init(adev) (adev)->mes.kiq_hw_init((adev)) #define amdgpu_mes_kiq_hw_fini(adev) (adev)->mes.kiq_hw_fini((adev)) -int amdgpu_mes_ctx_get_offs(struct amdgpu_ring *ring, unsigned int id_offs); - int amdgpu_mes_init_microcode(struct amdgpu_device *adev, int pipe); int amdgpu_mes_init(struct amdgpu_device *adev); void amdgpu_mes_fini(struct amdgpu_device *adev); @@ -395,8 +393,6 @@ int amdgpu_mes_reset_legacy_queue(struct amdgpu_device *adev, uint32_t amdgpu_mes_rreg(struct amdgpu_device *adev, uint32_t reg); int amdgpu_mes_wreg(struct amdgpu_device *adev, uint32_t reg, uint32_t val); -int amdgpu_mes_reg_wait(struct amdgpu_device *adev, uint32_t reg, - uint32_t val, uint32_t mask); int amdgpu_mes_reg_write_reg_wait(struct amdgpu_device *adev, uint32_t reg0, uint32_t reg1, uint32_t ref, uint32_t mask); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index dc07936d2fcb..de0944947eaf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -2859,6 +2859,15 @@ static int __amdgpu_ras_convert_rec_array_from_rom(struct amdgpu_device *adev, return -EINVAL; } } else { + if (bps[0].address == 0) { + /* for specific old eeprom data, mca address is not stored, + * calc it from pa + */ + if (amdgpu_umc_pa2mca(adev, bps[0].retired_page << AMDGPU_GPU_PAGE_SHIFT, + &(bps[0].address), AMDGPU_NPS1_PARTITION_MODE)) + return -EINVAL; + } + if (amdgpu_ras_mca2pa(adev, &bps[0], err_data)) { if (nps == AMDGPU_NPS1_PARTITION_MODE) memcpy(err_data->err_addr, bps, @@ -2886,8 +2895,20 @@ static int __amdgpu_ras_convert_rec_from_rom(struct amdgpu_device *adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT)) return -EINVAL; } else { - if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data)) - return -EINVAL; + if (bps->address) { + if (amdgpu_ras_mca2pa_by_idx(adev, bps, err_data)) + return -EINVAL; + } else { + /* for specific old eeprom data, mca address is not stored, + * calc it from pa + */ + if (amdgpu_umc_pa2mca(adev, bps->retired_page << AMDGPU_GPU_PAGE_SHIFT, + &(bps->address), AMDGPU_NPS1_PARTITION_MODE)) + return -EINVAL; + + if (amdgpu_ras_mca2pa(adev, bps, err_data)) + return -EOPNOTSUPP; + } } return __amdgpu_ras_restore_bad_pages(adev, err_data->err_addr, @@ -3708,7 +3729,8 @@ static void amdgpu_ras_query_ras_capablity_from_vbios(struct amdgpu_device *adev */ if (amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(2, 6, 0) || amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 0) || - amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3)) + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(4, 0, 3) || + amdgpu_ip_version(adev, VCN_HWIP, 0) == IP_VERSION(5, 0, 1)) adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__VCN | 1 << AMDGPU_RAS_BLOCK__JPEG); else diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h index 5605921212f0..e5f8951bbb6f 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sdma.h @@ -113,6 +113,7 @@ struct amdgpu_sdma { struct amdgpu_sdma_instance instance[AMDGPU_MAX_SDMA_INSTANCES]; struct amdgpu_irq_src trap_irq; struct amdgpu_irq_src illegal_inst_irq; + struct amdgpu_irq_src fence_irq; struct amdgpu_irq_src ecc_irq; struct amdgpu_irq_src vm_hole_irq; struct amdgpu_irq_src doorbell_invalid_irq; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c index 3939761be31c..d45ebfb642ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_seq64.c @@ -139,7 +139,7 @@ void amdgpu_seq64_unmap(struct amdgpu_device *adev, struct amdgpu_fpriv *fpriv) vm = &fpriv->vm; - drm_exec_init(&exec, DRM_EXEC_INTERRUPTIBLE_WAIT, 0); + drm_exec_init(&exec, 0, 0); drm_exec_until_all_locked(&exec) { r = amdgpu_vm_lock_pd(vm, &exec, 0); if (likely(!r)) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c index 4a72c2bbd49e..2505c46a9c3d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c @@ -765,6 +765,7 @@ FW_VERSION_ATTR(sdma_fw_version, 0444, sdma.instance[0].fw_version); FW_VERSION_ATTR(sdma2_fw_version, 0444, sdma.instance[1].fw_version); FW_VERSION_ATTR(vcn_fw_version, 0444, vcn.fw_version); FW_VERSION_ATTR(dmcu_fw_version, 0444, dm.dmcu_fw_version); +FW_VERSION_ATTR(dmcub_fw_version, 0444, dm.dmcub_fw_version); FW_VERSION_ATTR(mes_fw_version, 0444, mes.sched_version & AMDGPU_MES_VERSION_MASK); FW_VERSION_ATTR(mes_kiq_fw_version, 0444, mes.kiq_version & AMDGPU_MES_VERSION_MASK); FW_VERSION_ATTR(pldm_fw_version, 0444, firmware.pldm_version); @@ -780,9 +781,10 @@ static struct attribute *fw_attrs[] = { &dev_attr_ta_ras_fw_version.attr, &dev_attr_ta_xgmi_fw_version.attr, &dev_attr_smc_fw_version.attr, &dev_attr_sdma_fw_version.attr, &dev_attr_sdma2_fw_version.attr, &dev_attr_vcn_fw_version.attr, - &dev_attr_dmcu_fw_version.attr, &dev_attr_imu_fw_version.attr, - &dev_attr_mes_fw_version.attr, &dev_attr_mes_kiq_fw_version.attr, - &dev_attr_pldm_fw_version.attr, NULL + &dev_attr_dmcu_fw_version.attr, &dev_attr_dmcub_fw_version.attr, + &dev_attr_imu_fw_version.attr, &dev_attr_mes_fw_version.attr, + &dev_attr_mes_kiq_fw_version.attr, &dev_attr_pldm_fw_version.attr, + NULL }; #define to_dev_attr(x) container_of(x, struct device_attribute, attr) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index 8c6e55b5b967..c92b8794aa73 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -562,3 +562,26 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, return 0; } + +int amdgpu_umc_pa2mca(struct amdgpu_device *adev, + uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps) +{ + struct ta_ras_query_address_input addr_in; + struct ta_ras_query_address_output addr_out; + int ret; + + /* nps: the pa belongs to */ + addr_in.pa.pa = pa | ((uint64_t)nps << 58); + addr_in.addr_type = TA_RAS_PA_TO_MCA; + ret = psp_ras_query_address(&adev->psp, &addr_in, &addr_out); + if (ret) { + dev_warn(adev->dev, "Failed to query RAS MCA address for 0x%llx", + pa); + + return ret; + } + + *mca = addr_out.ma.err_addr; + + return 0; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h index 29ce6b1d214a..ec203f9e5ffa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.h @@ -189,4 +189,6 @@ int amdgpu_umc_mca_to_addr(struct amdgpu_device *adev, uint64_t err_addr, uint32_t ch, uint32_t umc, uint32_t node, uint32_t socket, struct ta_ras_query_address_output *addr_out, bool dump_addr); +int amdgpu_umc_pa2mca(struct amdgpu_device *adev, + uint64_t pa, uint64_t *mca, enum amdgpu_memory_partition nps); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c index fc4d0d42e223..a86616c6deef 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_userq_fence.c @@ -430,7 +430,7 @@ int amdgpu_userq_signal_ioctl(struct drm_device *dev, void *data, num_syncobj_handles = args->num_syncobj_handles; syncobj_handles = memdup_user(u64_to_user_ptr(args->syncobj_handles), - sizeof(u32) * num_syncobj_handles); + size_mul(sizeof(u32), num_syncobj_handles)); if (IS_ERR(syncobj_handles)) return PTR_ERR(syncobj_handles); @@ -612,13 +612,13 @@ int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data, num_read_bo_handles = wait_info->num_bo_read_handles; bo_handles_read = memdup_user(u64_to_user_ptr(wait_info->bo_read_handles), - sizeof(u32) * num_read_bo_handles); + size_mul(sizeof(u32), num_read_bo_handles)); if (IS_ERR(bo_handles_read)) return PTR_ERR(bo_handles_read); num_write_bo_handles = wait_info->num_bo_write_handles; bo_handles_write = memdup_user(u64_to_user_ptr(wait_info->bo_write_handles), - sizeof(u32) * num_write_bo_handles); + size_mul(sizeof(u32), num_write_bo_handles)); if (IS_ERR(bo_handles_write)) { r = PTR_ERR(bo_handles_write); goto free_bo_handles_read; @@ -626,7 +626,7 @@ int amdgpu_userq_wait_ioctl(struct drm_device *dev, void *data, num_syncobj = wait_info->num_syncobj_handles; syncobj_handles = memdup_user(u64_to_user_ptr(wait_info->syncobj_handles), - sizeof(u32) * num_syncobj); + size_mul(sizeof(u32), num_syncobj)); if (IS_ERR(syncobj_handles)) { r = PTR_ERR(syncobj_handles); goto free_bo_handles_write; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c index 2d7f82e98df9..abdc52b0895a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vram_mgr.c @@ -463,7 +463,7 @@ static int amdgpu_vram_mgr_new(struct ttm_resource_manager *man, int r; lpfn = (u64)place->lpfn << PAGE_SHIFT; - if (!lpfn) + if (!lpfn || lpfn > man->size) lpfn = man->size; fpfn = (u64)place->fpfn << PAGE_SHIFT; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c index b03c3895897b..322816805bfb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c @@ -27,6 +27,9 @@ #include <drm/drm_drv.h> #include "../amdxcp/amdgpu_xcp_drv.h" +static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr); +static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr); + static int __amdgpu_xcp_run(struct amdgpu_xcp_mgr *xcp_mgr, struct amdgpu_xcp_ip *xcp_ip, int xcp_state) { @@ -189,7 +192,7 @@ static int __amdgpu_xcp_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr, goto out; } - + amdgpu_xcp_sysfs_entries_update(xcp_mgr); out: mutex_unlock(&xcp_mgr->xcp_lock); @@ -263,9 +266,10 @@ static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev) if (ret == -ENOSPC) { dev_warn(adev->dev, "Skip xcp node #%d when out of drm node resource.", i); - return 0; + ret = 0; + goto out; } else if (ret) { - return ret; + goto out; } /* Redirect all IOCTLs to the primary device */ @@ -278,9 +282,14 @@ static int amdgpu_xcp_dev_alloc(struct amdgpu_device *adev) p_ddev->vma_offset_manager = ddev->vma_offset_manager; p_ddev->driver = &amdgpu_partition_driver; adev->xcp_mgr->xcp[i].ddev = p_ddev; + + dev_set_drvdata(p_ddev->dev, &adev->xcp_mgr->xcp[i]); } + ret = 0; +out: + amdgpu_xcp_sysfs_entries_init(adev->xcp_mgr); - return 0; + return ret; } int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode, @@ -288,6 +297,7 @@ int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode, struct amdgpu_xcp_mgr_funcs *xcp_funcs) { struct amdgpu_xcp_mgr *xcp_mgr; + int i; if (!xcp_funcs || !xcp_funcs->get_ip_details) return -EINVAL; @@ -306,6 +316,8 @@ int amdgpu_xcp_mgr_init(struct amdgpu_device *adev, int init_mode, amdgpu_xcp_init(xcp_mgr, init_num_xcps, init_mode); adev->xcp_mgr = xcp_mgr; + for (i = 0; i < MAX_XCP; ++i) + xcp_mgr->xcp[i].xcp_mgr = xcp_mgr; return amdgpu_xcp_dev_alloc(adev); } @@ -433,6 +445,7 @@ void amdgpu_xcp_release_sched(struct amdgpu_device *adev, } } +/*====================== xcp sysfs - configuration ======================*/ #define XCP_CFG_SYSFS_RES_ATTR_SHOW(_name) \ static ssize_t amdgpu_xcp_res_sysfs_##_name##_show( \ struct amdgpu_xcp_res_details *xcp_res, char *buf) \ @@ -635,7 +648,7 @@ static const struct attribute *xcp_attrs[] = { NULL, }; -void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev) +static void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev) { struct amdgpu_xcp_res_details *xcp_res; struct amdgpu_xcp_cfg *xcp_cfg; @@ -703,7 +716,7 @@ err1: kobject_put(&xcp_cfg->kobj); } -void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev) +static void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev) { struct amdgpu_xcp_res_details *xcp_res; struct amdgpu_xcp_cfg *xcp_cfg; @@ -722,3 +735,124 @@ void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev) sysfs_remove_files(&xcp_cfg->kobj, xcp_attrs); kobject_put(&xcp_cfg->kobj); } + +/*====================== xcp sysfs - data entries ======================*/ + +#define to_xcp(x) container_of(x, struct amdgpu_xcp, kobj) + +static ssize_t xcp_metrics_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + struct amdgpu_xcp *xcp = to_xcp(kobj); + struct amdgpu_xcp_mgr *xcp_mgr; + ssize_t size; + + xcp_mgr = xcp->xcp_mgr; + size = amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, NULL); + if (size <= 0) + return size; + + if (size > PAGE_SIZE) + return -ENOSPC; + + return amdgpu_dpm_get_xcp_metrics(xcp_mgr->adev, xcp->id, buf); +} + +static umode_t amdgpu_xcp_attrs_is_visible(struct kobject *kobj, + struct attribute *attr, int n) +{ + struct amdgpu_xcp *xcp = to_xcp(kobj); + + if (!xcp || !xcp->valid) + return 0; + + return attr->mode; +} + +static struct kobj_attribute xcp_sysfs_metrics = __ATTR_RO(xcp_metrics); + +static struct attribute *amdgpu_xcp_attrs[] = { + &xcp_sysfs_metrics.attr, + NULL, +}; + +static const struct attribute_group amdgpu_xcp_attrs_group = { + .attrs = amdgpu_xcp_attrs, + .is_visible = amdgpu_xcp_attrs_is_visible +}; + +static const struct kobj_type xcp_sysfs_ktype = { + .sysfs_ops = &kobj_sysfs_ops, +}; + +static void amdgpu_xcp_sysfs_entries_fini(struct amdgpu_xcp_mgr *xcp_mgr, int n) +{ + struct amdgpu_xcp *xcp; + + for (n--; n >= 0; n--) { + xcp = &xcp_mgr->xcp[n]; + if (!xcp->ddev || !xcp->valid) + continue; + sysfs_remove_group(&xcp->kobj, &amdgpu_xcp_attrs_group); + kobject_put(&xcp->kobj); + } +} + +static void amdgpu_xcp_sysfs_entries_init(struct amdgpu_xcp_mgr *xcp_mgr) +{ + struct amdgpu_xcp *xcp; + int i, r; + + for (i = 0; i < MAX_XCP; i++) { + /* Redirect all IOCTLs to the primary device */ + xcp = &xcp_mgr->xcp[i]; + if (!xcp->ddev) + break; + r = kobject_init_and_add(&xcp->kobj, &xcp_sysfs_ktype, + &xcp->ddev->dev->kobj, "xcp"); + if (r) + goto out; + + r = sysfs_create_group(&xcp->kobj, &amdgpu_xcp_attrs_group); + if (r) + goto out; + } + + return; +out: + kobject_put(&xcp->kobj); +} + +static void amdgpu_xcp_sysfs_entries_update(struct amdgpu_xcp_mgr *xcp_mgr) +{ + struct amdgpu_xcp *xcp; + int i; + + for (i = 0; i < MAX_XCP; i++) { + /* Redirect all IOCTLs to the primary device */ + xcp = &xcp_mgr->xcp[i]; + if (!xcp->ddev) + continue; + sysfs_update_group(&xcp->kobj, &amdgpu_xcp_attrs_group); + } + + return; +} + +void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev) +{ + if (!adev->xcp_mgr) + return; + + amdgpu_xcp_cfg_sysfs_init(adev); + + return; +} + +void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev) +{ + if (!adev->xcp_mgr) + return; + amdgpu_xcp_sysfs_entries_fini(adev->xcp_mgr, MAX_XCP); + amdgpu_xcp_cfg_sysfs_fini(adev); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h index b63f53242c57..454b33f889fb 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h @@ -108,6 +108,8 @@ struct amdgpu_xcp { struct drm_driver *driver; struct drm_vma_offset_manager *vma_offset_manager; struct amdgpu_sched gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX]; + struct amdgpu_xcp_mgr *xcp_mgr; + struct kobject kobj; }; struct amdgpu_xcp_mgr { @@ -175,8 +177,8 @@ int amdgpu_xcp_open_device(struct amdgpu_device *adev, void amdgpu_xcp_release_sched(struct amdgpu_device *adev, struct amdgpu_ctx_entity *entity); -void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev); -void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev); +void amdgpu_xcp_sysfs_init(struct amdgpu_device *adev); +void amdgpu_xcp_sysfs_fini(struct amdgpu_device *adev); #define amdgpu_xcp_select_scheds(adev, e, c, d, x, y) \ ((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c index f51ef4cf16e0..d9ad37711c3e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c @@ -294,6 +294,23 @@ static const struct amdgpu_pcs_ras_field xgmi3x16_pcs_ras_fields[] = { SOC15_REG_FIELD(PCS_XGMI3X16_PCS_ERROR_STATUS, RxCMDPktErr)}, }; +int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num) +{ + int link_map_6_4_x[8] = { 0, 3, 1, 2, 7, 6, 4, 5 }; + + switch (amdgpu_ip_version(adev, XGMI_HWIP, 0)) { + case IP_VERSION(6, 4, 0): + case IP_VERSION(6, 4, 1): + if (link_num < ARRAY_SIZE(link_map_6_4_x)) + return link_map_6_4_x[link_num]; + break; + default: + return -EINVAL; + } + + return -EINVAL; +} + static u32 xgmi_v6_4_get_link_status(struct amdgpu_device *adev, int global_link_num) { const u32 smn_xgmi_6_4_pcs_state_hist1[2] = { 0x11a00070, 0x11b00070 }; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h index 32dabba4062f..f994be985f42 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h @@ -125,6 +125,7 @@ int amdgpu_xgmi_request_nps_change(struct amdgpu_device *adev, int req_nps_mode); int amdgpu_get_xgmi_link_status(struct amdgpu_device *adev, int global_link_num); +int amdgpu_xgmi_get_ext_link(struct amdgpu_device *adev, int link_num); void amdgpu_xgmi_early_init(struct amdgpu_device *adev); uint32_t amdgpu_xgmi_get_max_bandwidth(struct amdgpu_device *adev); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h index 5255378af53c..f67569ccf9f6 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0_cleaner_shader.h @@ -43,9 +43,9 @@ static const u32 gfx_10_1_10_cleaner_shader_hex[] = { 0xd70f6a01, 0x000202ff, 0x00000400, 0x80828102, 0xbf84fff7, 0xbefc03ff, - 0x00000068, 0xbe803080, - 0xbe813080, 0xbe823080, - 0xbe833080, 0x80fc847c, + 0x00000068, 0xbe803000, + 0xbe813000, 0xbe823000, + 0xbe833000, 0x80fc847c, 0xbf84fffa, 0xbeea0480, 0xbeec0480, 0xbeee0480, 0xbef00480, 0xbef20480, diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm index 9ba3359253c9..54f7ed9e2801 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm @@ -40,7 +40,6 @@ shader main type(CS) wave_size(32) // Note: original source code from SQ team - // // Create 32 waves in a threadgroup (CS waves) // Each allocates 64 VGPRs @@ -71,8 +70,8 @@ label_0005: s_sub_u32 s2, s2, 8 s_cbranch_scc0 label_0005 // - s_mov_b32 s2, 0x80000000 // Bit31 is first_wave - s_and_b32 s2, s2, s0 // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set + s_mov_b32 s2, 0x80000000 // Bit31 is first_wave + s_and_b32 s2, s2, s1 // sgpr0 has tg_size (first_wave) term as in ucode only COMPUTE_PGM_RSRC2.tg_size_en is set s_cbranch_scc0 label_0023 // Clean LDS if its first wave of ThreadGroup/WorkGroup // CLEAR LDS // @@ -99,10 +98,10 @@ label_001F: label_0023: s_mov_b32 m0, 0x00000068 // Loop 108/4=27 times (loop unrolled for performance) label_sgpr_loop: - s_movreld_b32 s0, 0 - s_movreld_b32 s1, 0 - s_movreld_b32 s2, 0 - s_movreld_b32 s3, 0 + s_movreld_b32 s0, s0 + s_movreld_b32 s1, s0 + s_movreld_b32 s2, s0 + s_movreld_b32 s3, s0 s_sub_u32 m0, m0, 4 s_cbranch_scc0 label_sgpr_loop diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index f09d96bfee16..1234c8d64e20 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -36,7 +36,7 @@ #include "gc/gc_12_0_0_offset.h" #include "gc/gc_12_0_0_sh_mask.h" #include "soc24_enum.h" -#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h" +#include "ivsrcid/gfx/irqsrcs_gfx_12_0_0.h" #include "soc15.h" #include "clearstate_gfx12.h" @@ -1453,28 +1453,28 @@ static int gfx_v12_0_sw_init(struct amdgpu_ip_block *ip_block) /* EOP Event */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP, - GFX_11_0_0__SRCID__CP_EOP_INTERRUPT, + GFX_12_0_0__SRCID__CP_EOP_INTERRUPT, &adev->gfx.eop_irq); if (r) return r; /* Bad opcode Event */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP, - GFX_11_0_0__SRCID__CP_BAD_OPCODE_ERROR, + GFX_12_0_0__SRCID__CP_BAD_OPCODE_ERROR, &adev->gfx.bad_op_irq); if (r) return r; /* Privileged reg */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP, - GFX_11_0_0__SRCID__CP_PRIV_REG_FAULT, + GFX_12_0_0__SRCID__CP_PRIV_REG_FAULT, &adev->gfx.priv_reg_irq); if (r) return r; /* Privileged inst */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GRBM_CP, - GFX_11_0_0__SRCID__CP_PRIV_INSTR_FAULT, + GFX_12_0_0__SRCID__CP_PRIV_INSTR_FAULT, &adev->gfx.priv_inst_irq); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c index 69dd92f6e86d..574880d67009 100644 --- a/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c +++ b/drivers/gpu/drm/amd/amdgpu/isp_v4_1_1.c @@ -25,6 +25,7 @@ * */ +#include <linux/gpio/machine.h> #include "amdgpu.h" #include "isp_v4_1_1.h" @@ -39,15 +40,45 @@ static const unsigned int isp_4_1_1_int_srcid[MAX_ISP411_INT_SRC] = { ISP_4_1__SRCID__ISP_RINGBUFFER_WPT16 }; +static struct gpiod_lookup_table isp_gpio_table = { + .dev_id = "amd_isp_capture", + .table = { + GPIO_LOOKUP("AMDI0030:00", 85, "enable_isp", GPIO_ACTIVE_HIGH), + { } + }, +}; + +static struct gpiod_lookup_table isp_sensor_gpio_table = { + .dev_id = "i2c-ov05c10", + .table = { + GPIO_LOOKUP("amdisp-pinctrl", 0, "enable", GPIO_ACTIVE_HIGH), + { } + }, +}; + static int isp_v4_1_1_hw_init(struct amdgpu_isp *isp) { struct amdgpu_device *adev = isp->adev; int idx, int_idx, num_res, r; + u8 isp_dev_hid[ACPI_ID_LEN]; u64 isp_base; if (adev->rmmio_size == 0 || adev->rmmio_size < 0x5289) return -EINVAL; + r = amdgpu_acpi_get_isp4_dev_hid(&isp_dev_hid); + if (r) { + drm_dbg(&adev->ddev, "Invalid isp platform detected (%d)", r); + /* allow GPU init to progress */ + return 0; + } + + /* add GPIO resources required for OMNI5C10 sensor */ + if (!strcmp("OMNI5C10", isp_dev_hid)) { + gpiod_add_lookup_table(&isp_gpio_table); + gpiod_add_lookup_table(&isp_sensor_gpio_table); + } + isp_base = adev->rmmio_base; isp->isp_cell = kcalloc(3, sizeof(struct mfd_cell), GFP_KERNEL); diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c index a8ccae361ec7..79e342d5ab28 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -149,6 +149,18 @@ static int jpeg_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block) return r; } + /* JPEG DJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; + + /* JPEG EJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; + r = amdgpu_jpeg_sw_init(adev); if (r) return r; @@ -434,6 +446,9 @@ static int jpeg_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block) ret = jpeg_v4_0_3_set_powergating_state(ip_block, AMD_PG_STATE_GATE); } + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) + amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0); + return ret; } @@ -1041,6 +1056,14 @@ static int jpeg_v4_0_3_set_interrupt_state(struct amdgpu_device *adev, return 0; } +static int jpeg_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + return 0; +} + static int jpeg_v4_0_3_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -1200,6 +1223,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_irq_funcs = { .process = jpeg_v4_0_3_process_interrupt, }; +static const struct amdgpu_irq_src_funcs jpeg_v4_0_3_ras_irq_funcs = { + .set = jpeg_v4_0_3_set_ras_interrupt_state, + .process = amdgpu_jpeg_process_poison_irq, +}; + static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev) { int i; @@ -1208,6 +1236,9 @@ static void jpeg_v4_0_3_set_irq_funcs(struct amdgpu_device *adev) adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings; } adev->jpeg.inst->irq.funcs = &jpeg_v4_0_3_irq_funcs; + + adev->jpeg.inst->ras_poison_irq.num_types = 1; + adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v4_0_3_ras_irq_funcs; } const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block = { @@ -1304,9 +1335,47 @@ static void jpeg_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev) jpeg_v4_0_3_inst_reset_ras_error_count(adev, i); } +static uint32_t jpeg_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_JPEG_V4_0_3_JPEG0: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF); + break; + case AMDGPU_JPEG_V4_0_3_JPEG1: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool jpeg_v4_0_3_query_ras_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst = 0, sub = 0, poison_stat = 0; + + for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++) + for (sub = 0; sub < AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK; sub++) + poison_stat += + jpeg_v4_0_3_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + static const struct amdgpu_ras_block_hw_ops jpeg_v4_0_3_ras_hw_ops = { .query_ras_error_count = jpeg_v4_0_3_query_ras_error_count, .reset_ras_error_count = jpeg_v4_0_3_reset_ras_error_count, + .query_poison_status = jpeg_v4_0_3_query_ras_poison_status, }; static int jpeg_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, @@ -1383,6 +1452,13 @@ static int jpeg_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_comm if (r) return r; + if (amdgpu_ras_is_supported(adev, ras_block->block) && + adev->jpeg.inst->ras_poison_irq.funcs) { + r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0); + if (r) + goto late_fini; + } + r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG, &jpeg_v4_0_3_aca_info, NULL); if (r) diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h index a90bf370a002..2e110d04af84 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h @@ -46,6 +46,13 @@ #define JRBC_DEC_EXTERNAL_REG_WRITE_ADDR 0x18000 +enum amdgpu_jpeg_v4_0_3_sub_block { + AMDGPU_JPEG_V4_0_3_JPEG0 = 0, + AMDGPU_JPEG_V4_0_3_JPEG1, + + AMDGPU_JPEG_V4_0_3_MAX_SUB_BLOCK, +}; + extern const struct amdgpu_ip_block_version jpeg_v4_0_3_ip_block; void jpeg_v4_0_3_dec_ring_emit_ib(struct amdgpu_ring *ring, diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c index cb94bd71300f..3b6f65a25646 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c @@ -39,6 +39,7 @@ static void jpeg_v5_0_1_set_dec_ring_funcs(struct amdgpu_device *adev); static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev); static int jpeg_v5_0_1_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state); +static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev); static void jpeg_v5_0_1_dec_ring_set_wptr(struct amdgpu_ring *ring); static int amdgpu_ih_srcid_jpeg[] = { @@ -120,6 +121,7 @@ static int jpeg_v5_0_1_early_init(struct amdgpu_ip_block *ip_block) adev->jpeg.num_jpeg_rings = AMDGPU_MAX_JPEG_RINGS; jpeg_v5_0_1_set_dec_ring_funcs(adev); jpeg_v5_0_1_set_irq_funcs(adev); + jpeg_v5_0_1_set_ras_funcs(adev); return 0; } @@ -144,6 +146,17 @@ static int jpeg_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; } + /* JPEG DJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_5_0__SRCID_DJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; + + /* JPEG EJPEG POISON EVENT */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_5_0__SRCID_EJPEG0_POISON, &adev->jpeg.inst->ras_poison_irq); + if (r) + return r; r = amdgpu_jpeg_sw_init(adev); if (r) @@ -296,6 +309,9 @@ static int jpeg_v5_0_1_hw_fini(struct amdgpu_ip_block *ip_block) ret = jpeg_v5_0_1_set_powergating_state(ip_block, AMD_PG_STATE_GATE); } + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__JPEG)) + amdgpu_irq_put(adev, &adev->jpeg.inst->ras_poison_irq, 0); + return ret; } @@ -723,6 +739,16 @@ static int jpeg_v5_0_1_set_interrupt_state(struct amdgpu_device *adev, return 0; } +static int jpeg_v5_0_1_set_ras_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + return 0; +} + + + static int jpeg_v5_0_1_process_interrupt(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -892,6 +918,11 @@ static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_irq_funcs = { .process = jpeg_v5_0_1_process_interrupt, }; +static const struct amdgpu_irq_src_funcs jpeg_v5_0_1_ras_irq_funcs = { + .set = jpeg_v5_0_1_set_ras_interrupt_state, + .process = amdgpu_jpeg_process_poison_irq, +}; + static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev) { int i; @@ -900,6 +931,10 @@ static void jpeg_v5_0_1_set_irq_funcs(struct amdgpu_device *adev) adev->jpeg.inst->irq.num_types += adev->jpeg.num_jpeg_rings; adev->jpeg.inst->irq.funcs = &jpeg_v5_0_1_irq_funcs; + + adev->jpeg.inst->ras_poison_irq.num_types = 1; + adev->jpeg.inst->ras_poison_irq.funcs = &jpeg_v5_0_1_ras_irq_funcs; + } const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block = { @@ -909,3 +944,150 @@ const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block = { .rev = 1, .funcs = &jpeg_v5_0_1_ip_funcs, }; + +static uint32_t jpeg_v5_0_1_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_JPEG_V5_0_1_JPEG0: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG0_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG0_STATUS, POISONED_PF); + break; + case AMDGPU_JPEG_V5_0_1_JPEG1: + reg_value = RREG32_SOC15(JPEG, instance, regUVD_RAS_JPEG1_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_JPEG1_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in JPEG%d sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool jpeg_v5_0_1_query_ras_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst = 0, sub = 0, poison_stat = 0; + + for (inst = 0; inst < adev->jpeg.num_jpeg_inst; inst++) + for (sub = 0; sub < AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK; sub++) + poison_stat += + jpeg_v5_0_1_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + +static const struct amdgpu_ras_block_hw_ops jpeg_v5_0_1_ras_hw_ops = { + .query_poison_status = jpeg_v5_0_1_query_ras_poison_status, +}; + +static int jpeg_v5_0_1_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + struct aca_bank_info info; + u64 misc0; + int ret; + + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; + + misc0 = bank->regs[ACA_REG_IDX_MISC0]; + switch (type) { + case ACA_SMU_TYPE_UE: + bank->aca_err_type = ACA_ERROR_TYPE_UE; + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, + 1ULL); + break; + case ACA_SMU_TYPE_CE: + bank->aca_err_type = ACA_ERROR_TYPE_CE; + ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, + ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; + } + + return ret; +} + +/* reference to smu driver if header file */ +static int jpeg_v5_0_1_err_codes[] = { + 16, 17, 18, 19, 20, 21, 22, 23, /* JPEG[0-7][S|D] */ + 24, 25, 26, 27, 28, 29, 30, 31 +}; + +static bool jpeg_v5_0_1_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + u32 instlo; + + instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]); + instlo &= GENMASK(31, 1); + + if (instlo != mmSMNAID_AID0_MCA_SMU) + return false; + + if (aca_bank_check_error_codes(handle->adev, bank, + jpeg_v5_0_1_err_codes, + ARRAY_SIZE(jpeg_v5_0_1_err_codes))) + return false; + + return true; +} + +static const struct aca_bank_ops jpeg_v5_0_1_aca_bank_ops = { + .aca_bank_parser = jpeg_v5_0_1_aca_bank_parser, + .aca_bank_is_valid = jpeg_v5_0_1_aca_bank_is_valid, +}; + +static const struct aca_info jpeg_v5_0_1_aca_info = { + .hwip = ACA_HWIP_TYPE_SMU, + .mask = ACA_ERROR_UE_MASK, + .bank_ops = &jpeg_v5_0_1_aca_bank_ops, +}; + +static int jpeg_v5_0_1_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) +{ + int r; + + r = amdgpu_ras_block_late_init(adev, ras_block); + if (r) + return r; + + if (amdgpu_ras_is_supported(adev, ras_block->block) && + adev->jpeg.inst->ras_poison_irq.funcs) { + r = amdgpu_irq_get(adev, &adev->jpeg.inst->ras_poison_irq, 0); + if (r) + goto late_fini; + } + + r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__JPEG, + &jpeg_v5_0_1_aca_info, NULL); + if (r) + goto late_fini; + + return 0; + +late_fini: + amdgpu_ras_block_late_fini(adev, ras_block); + + return r; +} + +static struct amdgpu_jpeg_ras jpeg_v5_0_1_ras = { + .ras_block = { + .hw_ops = &jpeg_v5_0_1_ras_hw_ops, + .ras_late_init = jpeg_v5_0_1_ras_late_init, + }, +}; + +static void jpeg_v5_0_1_set_ras_funcs(struct amdgpu_device *adev) +{ + adev->jpeg.ras = &jpeg_v5_0_1_ras; +} diff --git a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h index efdab57324e4..a7e58d5fb246 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.h @@ -26,6 +26,9 @@ extern const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block; +#define regUVD_JRBC0_UVD_JRBC_SCRATCH0_INTERNAL_OFFSET 0x4094 +#define regUVD_JRBC_EXTERNAL_MCM_ADDR_INTERNAL_OFFSET 0x1bffe + #define regUVD_JRBC0_UVD_JRBC_RB_WPTR 0x0640 #define regUVD_JRBC0_UVD_JRBC_RB_WPTR_BASE_IDX 1 #define regUVD_JRBC0_UVD_JRBC_STATUS 0x0649 @@ -98,4 +101,11 @@ extern const struct amdgpu_ip_block_version jpeg_v5_0_1_ip_block; #define regVCN_RRMT_CNTL 0x0940 #define regVCN_RRMT_CNTL_BASE_IDX 1 +enum amdgpu_jpeg_v5_0_1_sub_block { + AMDGPU_JPEG_V5_0_1_JPEG0 = 0, + AMDGPU_JPEG_V5_0_1_JPEG1, + + AMDGPU_JPEG_V5_0_1_MAX_SUB_BLOCK, +}; + #endif /* __JPEG_V5_0_1_H__ */ diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index da5b5d64f137..5a70ae17be04 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -44,6 +44,7 @@ #include "sdma_v6_0.h" #include "v11_structs.h" #include "mes_userqueue.h" +#include "amdgpu_userq_fence.h" MODULE_FIRMWARE("amdgpu/sdma_6_0_0.bin"); MODULE_FIRMWARE("amdgpu/sdma_6_0_1.bin"); @@ -893,6 +894,9 @@ static int sdma_v6_0_mqd_init(struct amdgpu_device *adev, void *mqd, m->sdmax_rlcx_csa_addr_lo = lower_32_bits(prop->csa_addr); m->sdmax_rlcx_csa_addr_hi = upper_32_bits(prop->csa_addr); + m->sdmax_rlcx_f32_dbg0 = lower_32_bits(prop->fence_address); + m->sdmax_rlcx_f32_dbg1 = upper_32_bits(prop->fence_address); + return 0; } @@ -1315,6 +1319,13 @@ static int sdma_v6_0_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + /* SDMA user fence event */ + r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX, + GFX_11_0_0__SRCID__SDMA_FENCE, + &adev->sdma.fence_irq); + if (r) + return r; + for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; @@ -1575,25 +1586,9 @@ static int sdma_v6_0_process_trap_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { int instances, queue; - uint32_t mes_queue_id = entry->src_data[0]; DRM_DEBUG("IH: SDMA trap\n"); - if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) { - struct amdgpu_mes_queue *queue; - - mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK; - - spin_lock(&adev->mes.queue_id_lock); - queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id); - if (queue) { - DRM_DEBUG("process smda queue id = %d\n", mes_queue_id); - amdgpu_fence_process(queue->ring); - } - spin_unlock(&adev->mes.queue_id_lock); - return 0; - } - queue = entry->ring_id & 0xf; instances = (entry->ring_id & 0xf0) >> 4; if (instances > 1) { @@ -1615,6 +1610,29 @@ static int sdma_v6_0_process_trap_irq(struct amdgpu_device *adev, return 0; } +static int sdma_v6_0_process_fence_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + u32 doorbell_offset = entry->src_data[0]; + + if (adev->enable_mes && doorbell_offset) { + struct amdgpu_userq_fence_driver *fence_drv = NULL; + struct xarray *xa = &adev->userq_xa; + unsigned long flags; + + doorbell_offset >>= SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT; + + xa_lock_irqsave(xa, flags); + fence_drv = xa_load(xa, doorbell_offset); + if (fence_drv) + amdgpu_userq_fence_driver_process(fence_drv); + xa_unlock_irqrestore(xa, flags); + } + + return 0; +} + static int sdma_v6_0_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -1751,6 +1769,10 @@ static const struct amdgpu_irq_src_funcs sdma_v6_0_trap_irq_funcs = { .process = sdma_v6_0_process_trap_irq, }; +static const struct amdgpu_irq_src_funcs sdma_v6_0_fence_irq_funcs = { + .process = sdma_v6_0_process_fence_irq, +}; + static const struct amdgpu_irq_src_funcs sdma_v6_0_illegal_inst_irq_funcs = { .process = sdma_v6_0_process_illegal_inst_irq, }; @@ -1760,6 +1782,7 @@ static void sdma_v6_0_set_irq_funcs(struct amdgpu_device *adev) adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 + adev->sdma.num_instances; adev->sdma.trap_irq.funcs = &sdma_v6_0_trap_irq_funcs; + adev->sdma.fence_irq.funcs = &sdma_v6_0_fence_irq_funcs; adev->sdma.illegal_inst_irq.funcs = &sdma_v6_0_illegal_inst_irq_funcs; } diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c index befe013b11a7..ad47d0bdf777 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c @@ -33,7 +33,7 @@ #include "gc/gc_12_0_0_offset.h" #include "gc/gc_12_0_0_sh_mask.h" #include "hdp/hdp_6_0_0_offset.h" -#include "ivsrcid/gfx/irqsrcs_gfx_11_0_0.h" +#include "ivsrcid/gfx/irqsrcs_gfx_12_0_0.h" #include "soc15_common.h" #include "soc15.h" @@ -43,6 +43,7 @@ #include "sdma_v7_0.h" #include "v12_structs.h" #include "mes_userqueue.h" +#include "amdgpu_userq_fence.h" MODULE_FIRMWARE("amdgpu/sdma_7_0_0.bin"); MODULE_FIRMWARE("amdgpu/sdma_7_0_1.bin"); @@ -910,6 +911,9 @@ static int sdma_v7_0_mqd_init(struct amdgpu_device *adev, void *mqd, m->sdmax_rlcx_csa_addr_lo = lower_32_bits(prop->csa_addr); m->sdmax_rlcx_csa_addr_hi = upper_32_bits(prop->csa_addr); + m->sdmax_rlcx_mcu_dbg0 = lower_32_bits(prop->fence_address); + m->sdmax_rlcx_mcu_dbg1 = upper_32_bits(prop->fence_address); + return 0; } @@ -1296,11 +1300,18 @@ static int sdma_v7_0_sw_init(struct amdgpu_ip_block *ip_block) /* SDMA trap event */ r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX, - GFX_11_0_0__SRCID__SDMA_TRAP, + GFX_12_0_0__SRCID__SDMA_TRAP, &adev->sdma.trap_irq); if (r) return r; + /* SDMA user fence event */ + r = amdgpu_irq_add_id(adev, SOC21_IH_CLIENTID_GFX, + GFX_12_0_0__SRCID__SDMA_FENCE, + &adev->sdma.fence_irq); + if (r) + return r; + for (i = 0; i < adev->sdma.num_instances; i++) { ring = &adev->sdma.instance[i].ring; ring->ring_obj = NULL; @@ -1526,25 +1537,9 @@ static int sdma_v7_0_process_trap_irq(struct amdgpu_device *adev, struct amdgpu_iv_entry *entry) { int instances, queue; - uint32_t mes_queue_id = entry->src_data[0]; DRM_DEBUG("IH: SDMA trap\n"); - if (adev->enable_mes && (mes_queue_id & AMDGPU_FENCE_MES_QUEUE_FLAG)) { - struct amdgpu_mes_queue *queue; - - mes_queue_id &= AMDGPU_FENCE_MES_QUEUE_ID_MASK; - - spin_lock(&adev->mes.queue_id_lock); - queue = idr_find(&adev->mes.queue_id_idr, mes_queue_id); - if (queue) { - DRM_DEBUG("process smda queue id = %d\n", mes_queue_id); - amdgpu_fence_process(queue->ring); - } - spin_unlock(&adev->mes.queue_id_lock); - return 0; - } - queue = entry->ring_id & 0xf; instances = (entry->ring_id & 0xf0) >> 4; if (instances > 1) { @@ -1566,6 +1561,29 @@ static int sdma_v7_0_process_trap_irq(struct amdgpu_device *adev, return 0; } +static int sdma_v7_0_process_fence_irq(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + struct amdgpu_iv_entry *entry) +{ + u32 doorbell_offset = entry->src_data[0]; + + if (adev->enable_mes && doorbell_offset) { + struct amdgpu_userq_fence_driver *fence_drv = NULL; + struct xarray *xa = &adev->userq_xa; + unsigned long flags; + + doorbell_offset >>= SDMA0_QUEUE0_DOORBELL_OFFSET__OFFSET__SHIFT; + + xa_lock_irqsave(xa, flags); + fence_drv = xa_load(xa, doorbell_offset); + if (fence_drv) + amdgpu_userq_fence_driver_process(fence_drv); + xa_unlock_irqrestore(xa, flags); + } + + return 0; +} + static int sdma_v7_0_process_illegal_inst_irq(struct amdgpu_device *adev, struct amdgpu_irq_src *source, struct amdgpu_iv_entry *entry) @@ -1703,6 +1721,10 @@ static const struct amdgpu_irq_src_funcs sdma_v7_0_trap_irq_funcs = { .process = sdma_v7_0_process_trap_irq, }; +static const struct amdgpu_irq_src_funcs sdma_v7_0_fence_irq_funcs = { + .process = sdma_v7_0_process_fence_irq, +}; + static const struct amdgpu_irq_src_funcs sdma_v7_0_illegal_inst_irq_funcs = { .process = sdma_v7_0_process_illegal_inst_irq, }; @@ -1712,6 +1734,7 @@ static void sdma_v7_0_set_irq_funcs(struct amdgpu_device *adev) adev->sdma.trap_irq.num_types = AMDGPU_SDMA_IRQ_INSTANCE0 + adev->sdma.num_instances; adev->sdma.trap_irq.funcs = &sdma_v7_0_trap_irq_funcs; + adev->sdma.fence_irq.funcs = &sdma_v7_0_fence_irq_funcs; adev->sdma.illegal_inst_irq.funcs = &sdma_v7_0_illegal_inst_irq_funcs; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index 21b57c29bf7d..c74947705d77 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -1009,6 +1009,11 @@ static int vcn_v1_0_start_spg_mode(struct amdgpu_vcn_inst *vinst) jpeg_v1_0_start(adev, 0); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1154,6 +1159,11 @@ static int vcn_v1_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst) jpeg_v1_0_start(adev, 1); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1216,6 +1226,12 @@ static int vcn_v1_0_stop_spg_mode(struct amdgpu_vcn_inst *vinst) vcn_v1_0_enable_clock_gating(vinst); vcn_1_0_enable_static_power_gating(vinst); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1250,6 +1266,11 @@ static int vcn_v1_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index b8d835c9e17e..148b651be7ca 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -978,6 +978,12 @@ static int vcn_v2_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect) /* Unstall DPG */ WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1152,6 +1158,11 @@ static int vcn_v2_0_start(struct amdgpu_vcn_inst *vinst) WREG32_SOC15(UVD, 0, mmUVD_RB_SIZE2, ring->ring_size / 4); fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET; + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1183,6 +1194,11 @@ static int vcn_v2_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) WREG32_P(SOC15_REG_OFFSET(UVD, 0, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(UVD, 0, mmUVD_STATUS); + return 0; } @@ -1248,6 +1264,11 @@ static int vcn_v2_0_stop(struct amdgpu_vcn_inst *vinst) vcn_v2_0_enable_clock_gating(vinst); vcn_v2_0_enable_static_power_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, 0, mmUVD_STATUS); + power_off: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, 0); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 3eec1b8feaee..58b527a6b795 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -1158,6 +1158,11 @@ static int vcn_v2_5_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect) WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS); + return 0; } @@ -1343,6 +1348,11 @@ static int vcn_v2_5_start(struct amdgpu_vcn_inst *vinst) WREG32_SOC15(VCN, i, mmUVD_RB_SIZE2, ring->ring_size / 4); fw_shared->multi_queue.encode_lowlatency_queue_mode &= ~FW_QUEUE_RING_RESET; + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, mmUVD_STATUS); + return 0; } @@ -1569,6 +1579,11 @@ static int vcn_v2_5_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS); + return 0; } @@ -1635,6 +1650,10 @@ static int vcn_v2_5_stop(struct amdgpu_vcn_inst *vinst) UVD_POWER_STATUS__UVD_POWER_STATUS_MASK, ~UVD_POWER_STATUS__UVD_POWER_STATUS_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, mmUVD_STATUS); done: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, i); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index 0b19f0ab4480..9fb0d5380589 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -1173,6 +1173,11 @@ static int vcn_v3_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect) WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__STALL_DPG_POWER_UP_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS); + return 0; } @@ -1360,6 +1365,11 @@ static int vcn_v3_0_start(struct amdgpu_vcn_inst *vinst) fw_shared->multi_queue.encode_lowlatency_queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); } + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, mmUVD_STATUS); + return 0; } @@ -1602,6 +1612,11 @@ static int vcn_v3_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, mmUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, mmUVD_STATUS); + return 0; } @@ -1674,6 +1689,11 @@ static int vcn_v3_0_stop(struct amdgpu_vcn_inst *vinst) /* enable VCN power gating */ vcn_v3_0_enable_static_power_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, mmUVD_STATUS); + done: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, i); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 8fff470bce87..b5071f77f78d 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -1122,6 +1122,11 @@ static int vcn_v4_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, bool indirect) ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT | VCN_RB1_DB_CTRL__EN_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, regUVD_STATUS); + return 0; } @@ -1303,6 +1308,11 @@ static int vcn_v4_0_start(struct amdgpu_vcn_inst *vinst) WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp); fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, regUVD_STATUS); + return 0; } @@ -1583,6 +1593,11 @@ static void vcn_v4_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) /* disable dynamic power gating mode */ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, regUVD_STATUS); } /** @@ -1666,6 +1681,11 @@ static int vcn_v4_0_stop(struct amdgpu_vcn_inst *vinst) /* enable VCN power gating */ vcn_v4_0_enable_static_power_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, regUVD_STATUS); + done: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, i); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index 712e1fba33ce..5a33140f5723 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -169,6 +169,10 @@ static int vcn_v4_0_3_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + /* VCN POISON TRAP */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_4_0__SRCID_UVD_POISON, &adev->vcn.inst->ras_poison_irq); + for (i = 0; i < adev->vcn.num_vcn_inst; i++) { r = amdgpu_vcn_sw_init(adev, i); @@ -387,6 +391,9 @@ static int vcn_v4_0_3_hw_fini(struct amdgpu_ip_block *ip_block) vinst->set_pg_state(vinst, AMD_PG_STATE_GATE); } + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) + amdgpu_irq_put(adev, &adev->vcn.inst->ras_poison_irq, 0); + return 0; } @@ -970,6 +977,11 @@ static int vcn_v4_0_3_start_dpg_mode(struct amdgpu_vcn_inst *vinst, /*resetting done, fw can check RB ring */ fw_shared->sq.queue_mode &= cpu_to_le32(~FW_QUEUE_RING_RESET); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); + return 0; } @@ -1363,6 +1375,12 @@ static int vcn_v4_0_3_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) /* disable dynamic power gating mode */ WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); + return 0; } @@ -1446,6 +1464,11 @@ static int vcn_v4_0_3_stop(struct amdgpu_vcn_inst *vinst) /* apply HW clock gating */ vcn_v4_0_3_enable_clock_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); + Done: return 0; } @@ -1814,11 +1837,24 @@ static int vcn_v4_0_3_process_interrupt(struct amdgpu_device *adev, return 0; } +static int vcn_v4_0_3_set_ras_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + return 0; +} + static const struct amdgpu_irq_src_funcs vcn_v4_0_3_irq_funcs = { .set = vcn_v4_0_3_set_interrupt_state, .process = vcn_v4_0_3_process_interrupt, }; +static const struct amdgpu_irq_src_funcs vcn_v4_0_3_ras_irq_funcs = { + .set = vcn_v4_0_3_set_ras_interrupt_state, + .process = amdgpu_vcn_process_poison_irq, +}; + /** * vcn_v4_0_3_set_irq_funcs - set VCN block interrupt irq functions * @@ -1834,6 +1870,9 @@ static void vcn_v4_0_3_set_irq_funcs(struct amdgpu_device *adev) adev->vcn.inst->irq.num_types++; } adev->vcn.inst->irq.funcs = &vcn_v4_0_3_irq_funcs; + + adev->vcn.inst->ras_poison_irq.num_types = 1; + adev->vcn.inst->ras_poison_irq.funcs = &vcn_v4_0_3_ras_irq_funcs; } static void vcn_v4_0_3_print_ip_state(struct amdgpu_ip_block *ip_block, struct drm_printer *p) @@ -1981,9 +2020,44 @@ static void vcn_v4_0_3_reset_ras_error_count(struct amdgpu_device *adev) vcn_v4_0_3_inst_reset_ras_error_count(adev, i); } +static uint32_t vcn_v4_0_3_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_VCN_V4_0_3_VCPU_VCODEC: + reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool vcn_v4_0_3_query_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst, sub; + uint32_t poison_stat = 0; + + for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++) + for (sub = 0; sub < AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK; sub++) + poison_stat += + vcn_v4_0_3_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + static const struct amdgpu_ras_block_hw_ops vcn_v4_0_3_ras_hw_ops = { .query_ras_error_count = vcn_v4_0_3_query_ras_error_count, .reset_ras_error_count = vcn_v4_0_3_reset_ras_error_count, + .query_poison_status = vcn_v4_0_3_query_poison_status, }; static int vcn_v4_0_3_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, @@ -2059,6 +2133,13 @@ static int vcn_v4_0_3_ras_late_init(struct amdgpu_device *adev, struct ras_commo if (r) return r; + if (amdgpu_ras_is_supported(adev, ras_block->block) && + adev->vcn.inst->ras_poison_irq.funcs) { + r = amdgpu_irq_get(adev, &adev->vcn.inst->ras_poison_irq, 0); + if (r) + goto late_fini; + } + r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__VCN, &vcn_v4_0_3_aca_info, NULL); if (r) diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h index 03572a1d0c9c..aeab89853a92 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h @@ -24,6 +24,12 @@ #ifndef __VCN_V4_0_3_H__ #define __VCN_V4_0_3_H__ +enum amdgpu_vcn_v4_0_3_sub_block { + AMDGPU_VCN_V4_0_3_VCPU_VCODEC = 0, + + AMDGPU_VCN_V4_0_3_MAX_SUB_BLOCK, +}; + extern const struct amdgpu_ip_block_version vcn_v4_0_3_ip_block; void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c index a09f9a2dd471..16ade84facc7 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -1254,6 +1254,11 @@ static void vcn_v4_0_5_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) /* disable dynamic power gating mode */ WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, regUVD_STATUS); } /** @@ -1337,6 +1342,11 @@ static int vcn_v4_0_5_stop(struct amdgpu_vcn_inst *vinst) /* enable VCN power gating */ vcn_v4_0_5_enable_static_power_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, regUVD_STATUS); + done: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, i); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c index 27dcc6f37a73..f8e3f0b882da 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c @@ -794,6 +794,11 @@ static int vcn_v5_0_0_start_dpg_mode(struct amdgpu_vcn_inst *vinst, ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT | VCN_RB1_DB_CTRL__EN_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, regUVD_STATUS); + return 0; } @@ -946,6 +951,11 @@ static int vcn_v5_0_0_start(struct amdgpu_vcn_inst *vinst) WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp); fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, regUVD_STATUS); + return 0; } @@ -977,6 +987,11 @@ static void vcn_v5_0_0_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) WREG32_P(SOC15_REG_OFFSET(VCN, inst_idx, regUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, inst_idx, regUVD_STATUS); + return; } @@ -1058,6 +1073,11 @@ static int vcn_v5_0_0_stop(struct amdgpu_vcn_inst *vinst) /* enable VCN power gating */ vcn_v5_0_0_enable_static_power_gating(vinst); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, i, regUVD_STATUS); + done: if (adev->pm.dpm_enabled) amdgpu_dpm_enable_vcn(adev, false, i); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c index 8e843011703c..338cf43c45fe 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c @@ -46,7 +46,7 @@ static void vcn_v5_0_1_set_irq_funcs(struct amdgpu_device *adev); static int vcn_v5_0_1_set_pg_state(struct amdgpu_vcn_inst *vinst, enum amd_powergating_state state); static void vcn_v5_0_1_unified_ring_set_wptr(struct amdgpu_ring *ring); - +static void vcn_v5_0_1_set_ras_funcs(struct amdgpu_device *adev); /** * vcn_v5_0_1_early_init - set function pointers and load microcode * @@ -66,6 +66,7 @@ static int vcn_v5_0_1_early_init(struct amdgpu_ip_block *ip_block) vcn_v5_0_1_set_unified_ring_funcs(adev); vcn_v5_0_1_set_irq_funcs(adev); + vcn_v5_0_1_set_ras_funcs(adev); for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { adev->vcn.inst[i].set_pg_state = vcn_v5_0_1_set_pg_state; @@ -113,6 +114,10 @@ static int vcn_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; + /* VCN POISON TRAP */ + r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VCN, + VCN_5_0__SRCID_UVD_POISON, &adev->vcn.inst->ras_poison_irq); + for (i = 0; i < adev->vcn.num_vcn_inst; i++) { vcn_inst = GET_INST(VCN, i); @@ -279,6 +284,9 @@ static int vcn_v5_0_1_hw_fini(struct amdgpu_ip_block *ip_block) vinst->set_pg_state(vinst, AMD_PG_STATE_GATE); } + if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__VCN)) + amdgpu_irq_put(adev, &adev->vcn.inst->ras_poison_irq, 0); + return 0; } @@ -1030,6 +1038,11 @@ static int vcn_v5_0_1_start(struct amdgpu_vcn_inst *vinst) WREG32_SOC15(VCN, vcn_inst, regVCN_RB_ENABLE, tmp); fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); + return 0; } @@ -1064,6 +1077,11 @@ static void vcn_v5_0_1_stop_dpg_mode(struct amdgpu_vcn_inst *vinst) /* disable dynamic power gating mode */ WREG32_P(SOC15_REG_OFFSET(VCN, vcn_inst, regUVD_POWER_STATUS), 0, ~UVD_POWER_STATUS__UVD_PG_MODE_MASK); + + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); } /** @@ -1139,6 +1157,11 @@ static int vcn_v5_0_1_stop(struct amdgpu_vcn_inst *vinst) /* clear status */ WREG32_SOC15(VCN, vcn_inst, regUVD_STATUS, 0); + /* Keeping one read-back to ensure all register writes are done, + * otherwise it may introduce race conditions. + */ + RREG32_SOC15(VCN, vcn_inst, regUVD_STATUS); + return 0; } @@ -1391,10 +1414,24 @@ static int vcn_v5_0_1_process_interrupt(struct amdgpu_device *adev, struct amdgp return 0; } +static int vcn_v5_0_1_set_ras_interrupt_state(struct amdgpu_device *adev, + struct amdgpu_irq_src *source, + unsigned int type, + enum amdgpu_interrupt_state state) +{ + return 0; +} + static const struct amdgpu_irq_src_funcs vcn_v5_0_1_irq_funcs = { .process = vcn_v5_0_1_process_interrupt, }; +static const struct amdgpu_irq_src_funcs vcn_v5_0_1_ras_irq_funcs = { + .set = vcn_v5_0_1_set_ras_interrupt_state, + .process = amdgpu_vcn_process_poison_irq, +}; + + /** * vcn_v5_0_1_set_irq_funcs - set VCN block interrupt irq functions * @@ -1408,7 +1445,12 @@ static void vcn_v5_0_1_set_irq_funcs(struct amdgpu_device *adev) for (i = 0; i < adev->vcn.num_vcn_inst; ++i) adev->vcn.inst->irq.num_types++; + adev->vcn.inst->irq.funcs = &vcn_v5_0_1_irq_funcs; + + adev->vcn.inst->ras_poison_irq.num_types = 1; + adev->vcn.inst->ras_poison_irq.funcs = &vcn_v5_0_1_ras_irq_funcs; + } static const struct amd_ip_funcs vcn_v5_0_1_ip_funcs = { @@ -1440,3 +1482,139 @@ const struct amdgpu_ip_block_version vcn_v5_0_1_ip_block = { .rev = 1, .funcs = &vcn_v5_0_1_ip_funcs, }; + +static uint32_t vcn_v5_0_1_query_poison_by_instance(struct amdgpu_device *adev, + uint32_t instance, uint32_t sub_block) +{ + uint32_t poison_stat = 0, reg_value = 0; + + switch (sub_block) { + case AMDGPU_VCN_V5_0_1_VCPU_VCODEC: + reg_value = RREG32_SOC15(VCN, instance, regUVD_RAS_VCPU_VCODEC_STATUS); + poison_stat = REG_GET_FIELD(reg_value, UVD_RAS_VCPU_VCODEC_STATUS, POISONED_PF); + break; + default: + break; + } + + if (poison_stat) + dev_info(adev->dev, "Poison detected in VCN%d, sub_block%d\n", + instance, sub_block); + + return poison_stat; +} + +static bool vcn_v5_0_1_query_poison_status(struct amdgpu_device *adev) +{ + uint32_t inst, sub; + uint32_t poison_stat = 0; + + for (inst = 0; inst < adev->vcn.num_vcn_inst; inst++) + for (sub = 0; sub < AMDGPU_VCN_V5_0_1_MAX_SUB_BLOCK; sub++) + poison_stat += + vcn_v5_0_1_query_poison_by_instance(adev, inst, sub); + + return !!poison_stat; +} + +static const struct amdgpu_ras_block_hw_ops vcn_v5_0_1_ras_hw_ops = { + .query_poison_status = vcn_v5_0_1_query_poison_status, +}; + +static int vcn_v5_0_1_aca_bank_parser(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + struct aca_bank_info info; + u64 misc0; + int ret; + + ret = aca_bank_info_decode(bank, &info); + if (ret) + return ret; + + misc0 = bank->regs[ACA_REG_IDX_MISC0]; + switch (type) { + case ACA_SMU_TYPE_UE: + bank->aca_err_type = ACA_ERROR_TYPE_UE; + ret = aca_error_cache_log_bank_error(handle, &info, ACA_ERROR_TYPE_UE, + 1ULL); + break; + case ACA_SMU_TYPE_CE: + bank->aca_err_type = ACA_ERROR_TYPE_CE; + ret = aca_error_cache_log_bank_error(handle, &info, bank->aca_err_type, + ACA_REG__MISC0__ERRCNT(misc0)); + break; + default: + return -EINVAL; + } + + return ret; +} + +/* reference to smu driver if header file */ +static int vcn_v5_0_1_err_codes[] = { + 14, 15, /* VCN */ +}; + +static bool vcn_v5_0_1_aca_bank_is_valid(struct aca_handle *handle, struct aca_bank *bank, + enum aca_smu_type type, void *data) +{ + u32 instlo; + + instlo = ACA_REG__IPID__INSTANCEIDLO(bank->regs[ACA_REG_IDX_IPID]); + instlo &= GENMASK(31, 1); + + if (instlo != mmSMNAID_AID0_MCA_SMU) + return false; + + if (aca_bank_check_error_codes(handle->adev, bank, + vcn_v5_0_1_err_codes, + ARRAY_SIZE(vcn_v5_0_1_err_codes))) + return false; + + return true; +} + +static const struct aca_bank_ops vcn_v5_0_1_aca_bank_ops = { + .aca_bank_parser = vcn_v5_0_1_aca_bank_parser, + .aca_bank_is_valid = vcn_v5_0_1_aca_bank_is_valid, +}; + +static const struct aca_info vcn_v5_0_1_aca_info = { + .hwip = ACA_HWIP_TYPE_SMU, + .mask = ACA_ERROR_UE_MASK, + .bank_ops = &vcn_v5_0_1_aca_bank_ops, +}; + +static int vcn_v5_0_1_ras_late_init(struct amdgpu_device *adev, struct ras_common_if *ras_block) +{ + int r; + + r = amdgpu_ras_block_late_init(adev, ras_block); + if (r) + return r; + + r = amdgpu_ras_bind_aca(adev, AMDGPU_RAS_BLOCK__VCN, + &vcn_v5_0_1_aca_info, NULL); + if (r) + goto late_fini; + + return 0; + +late_fini: + amdgpu_ras_block_late_fini(adev, ras_block); + + return r; +} + +static struct amdgpu_vcn_ras vcn_v5_0_1_ras = { + .ras_block = { + .hw_ops = &vcn_v5_0_1_ras_hw_ops, + .ras_late_init = vcn_v5_0_1_ras_late_init, + }, +}; + +static void vcn_v5_0_1_set_ras_funcs(struct amdgpu_device *adev) +{ + adev->vcn.ras = &vcn_v5_0_1_ras; +} diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h index 8fd90bd10807..b72e4da68317 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.h @@ -27,6 +27,13 @@ #define regVCN_RRMT_CNTL 0x0940 #define regVCN_RRMT_CNTL_BASE_IDX 1 + +enum amdgpu_vcn_v5_0_1_sub_block { + AMDGPU_VCN_V5_0_1_VCPU_VCODEC = 0, + + AMDGPU_VCN_V5_0_1_MAX_SUB_BLOCK, +}; + extern const struct amdgpu_ip_block_version vcn_v5_0_1_ip_block; #endif /* __VCN_v5_0_1_H__ */ diff --git a/drivers/gpu/drm/amd/amdkfd/Kconfig b/drivers/gpu/drm/amd/amdkfd/Kconfig index d3c3d3ab7225..62e88e5362e9 100644 --- a/drivers/gpu/drm/amd/amdkfd/Kconfig +++ b/drivers/gpu/drm/amd/amdkfd/Kconfig @@ -5,7 +5,7 @@ config HSA_AMD bool "HSA kernel driver for AMD GPU devices" - depends on DRM_AMDGPU && (X86_64 || ARM64 || PPC64) + depends on DRM_AMDGPU && (X86_64 || ARM64 || PPC64 || (RISCV && 64BIT)) select HMM_MIRROR select MMU_NOTIFIER select DRM_AMDGPU_USERPTR diff --git a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c index 981d9adcc5e1..73acbe0b7c21 100644 --- a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c +++ b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c @@ -91,7 +91,6 @@ static void cik_event_interrupt_wq(struct kfd_node *dev, const struct cik_ih_ring_entry *ihre = (const struct cik_ih_ring_entry *)ih_ring_entry; uint32_t context_id = ihre->data & 0xfffffff; - unsigned int vmid = (ihre->ring_id & 0x0000ff00) >> 8; u32 pasid = (ihre->ring_id & 0xffff0000) >> 16; if (pasid == 0) @@ -125,11 +124,7 @@ static void cik_event_interrupt_wq(struct kfd_node *dev, return; } - if (info.vmid == vmid) - kfd_signal_vm_fault_event(pdd, &info, NULL); - else - kfd_signal_vm_fault_event(pdd, &info, NULL); - + kfd_signal_vm_fault_event(pdd, &info, NULL); kfd_unref_process(p); } } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 1e9dd00620bf..a2149afa5803 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -2039,9 +2039,7 @@ static int criu_get_process_object_info(struct kfd_process *p, num_events = kfd_get_num_events(p); - ret = svm_range_get_info(p, &num_svm_ranges, &svm_priv_data_size); - if (ret) - return ret; + svm_range_get_info(p, &num_svm_ranges, &svm_priv_data_size); *num_objects = num_queues + num_events + num_svm_ranges; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index e54e708ed82d..2b294ada3ec0 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -1350,6 +1350,7 @@ void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid) user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id); if (unlikely(user_gpu_id == -EINVAL)) { WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id); + kfd_unref_process(p); return; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c index 6d5fa57d4a23..c643e0ccec52 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c @@ -279,20 +279,17 @@ static int init_user_queue(struct process_queue_manager *pqm, /* Starting with GFX11, wptr BOs must be mapped to GART for MES to determine work * on unmapped queues for usermode queue oversubscription (no aggregated doorbell) */ - if (((dev->adev->mes.sched_version & AMDGPU_MES_API_VERSION_MASK) - >> AMDGPU_MES_API_VERSION_SHIFT) >= 2) { - if (dev->adev != amdgpu_ttm_adev(q_properties->wptr_bo->tbo.bdev)) { - pr_err("Queue memory allocated to wrong device\n"); - retval = -EINVAL; - goto free_gang_ctx_bo; - } + if (dev->adev != amdgpu_ttm_adev(q_properties->wptr_bo->tbo.bdev)) { + pr_err("Queue memory allocated to wrong device\n"); + retval = -EINVAL; + goto free_gang_ctx_bo; + } - retval = amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo, - &(*q)->wptr_bo_gart); - if (retval) { - pr_err("Failed to map wptr bo to GART\n"); - goto free_gang_ctx_bo; - } + retval = amdgpu_amdkfd_map_gtt_bo_to_gart(q_properties->wptr_bo, + &(*q)->wptr_bo_gart); + if (retval) { + pr_err("Failed to map wptr bo to GART\n"); + goto free_gang_ctx_bo; } } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index 72be6e152e88..865dca2547de 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -4075,8 +4075,8 @@ exit: return ret; } -int svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, - uint64_t *svm_priv_data_size) +void svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, + uint64_t *svm_priv_data_size) { uint64_t total_size, accessibility_size, common_attr_size; int nattr_common = 4, nattr_accessibility = 1; @@ -4088,8 +4088,6 @@ int svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, *svm_priv_data_size = 0; svms = &p->svms; - if (!svms) - return -EINVAL; mutex_lock(&svms->lock); list_for_each_entry(prange, &svms->list, list) { @@ -4131,7 +4129,6 @@ int svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, pr_debug("num_svm_ranges %u total_priv_size %llu\n", *num_svm_ranges, *svm_priv_data_size); - return 0; } int kfd_criu_checkpoint_svm(struct kfd_process *p, @@ -4148,8 +4145,6 @@ int kfd_criu_checkpoint_svm(struct kfd_process *p, struct mm_struct *mm; svms = &p->svms; - if (!svms) - return -EINVAL; mm = get_task_mm(p->lead_thread); if (!mm) { diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h index 6ea23c78009c..01c7a4877904 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.h @@ -184,8 +184,8 @@ void schedule_deferred_list_work(struct svm_range_list *svms); void svm_range_dma_unmap_dev(struct device *dev, dma_addr_t *dma_addr, unsigned long offset, unsigned long npages); void svm_range_dma_unmap(struct svm_range *prange); -int svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, - uint64_t *svm_priv_data_size); +void svm_range_get_info(struct kfd_process *p, uint32_t *num_svm_ranges, + uint64_t *svm_priv_data_size); int kfd_criu_checkpoint_svm(struct kfd_process *p, uint8_t __user *user_priv_data, uint64_t *priv_offset); @@ -237,13 +237,12 @@ static inline int svm_range_schedule_evict_svm_bo( return -EINVAL; } -static inline int svm_range_get_info(struct kfd_process *p, - uint32_t *num_svm_ranges, - uint64_t *svm_priv_data_size) +static inline void svm_range_get_info(struct kfd_process *p, + uint32_t *num_svm_ranges, + uint64_t *svm_priv_data_size) { *num_svm_ranges = 0; *svm_priv_data_size = 0; - return 0; } static inline int kfd_criu_checkpoint_svm(struct kfd_process *p, 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 742b10881112..d3100f641ac6 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -676,21 +676,15 @@ static void dm_crtc_high_irq(void *interrupt_params) spin_lock_irqsave(&adev_to_drm(adev)->event_lock, flags); if (acrtc->dm_irq_params.stream && - acrtc->dm_irq_params.vrr_params.supported) { - bool replay_en = acrtc->dm_irq_params.stream->link->replay_settings.replay_feature_enabled; - bool psr_en = acrtc->dm_irq_params.stream->link->psr_settings.psr_feature_enabled; - bool fs_active_var_en = acrtc->dm_irq_params.freesync_config.state == VRR_STATE_ACTIVE_VARIABLE; - + acrtc->dm_irq_params.vrr_params.supported && + acrtc->dm_irq_params.freesync_config.state == + VRR_STATE_ACTIVE_VARIABLE) { mod_freesync_handle_v_update(adev->dm.freesync_module, acrtc->dm_irq_params.stream, &acrtc->dm_irq_params.vrr_params); - /* update vmin_vmax only if freesync is enabled, or only if PSR and REPLAY are disabled */ - if (fs_active_var_en || (!fs_active_var_en && !replay_en && !psr_en)) { - dc_stream_adjust_vmin_vmax(adev->dm.dc, - acrtc->dm_irq_params.stream, - &acrtc->dm_irq_params.vrr_params.adjust); - } + dc_stream_adjust_vmin_vmax(adev->dm.dc, acrtc->dm_irq_params.stream, + &acrtc->dm_irq_params.vrr_params.adjust); } /* @@ -2006,8 +2000,10 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (amdgpu_dc_debug_mask & DC_FORCE_SUBVP_MCLK_SWITCH) adev->dm.dc->debug.force_subvp_mclk_switch = true; - if (amdgpu_dc_debug_mask & DC_DISABLE_SUBVP) + if (amdgpu_dc_debug_mask & DC_DISABLE_SUBVP_FAMS) { adev->dm.dc->debug.force_disable_subvp = true; + adev->dm.dc->debug.fams2_config.bits.enable = false; + } if (amdgpu_dc_debug_mask & DC_ENABLE_DML2) { adev->dm.dc->debug.using_dml2 = true; @@ -2020,6 +2016,9 @@ static int amdgpu_dm_init(struct amdgpu_device *adev) if (amdgpu_dc_debug_mask & DC_HDCP_LC_ENABLE_SW_FALLBACK) adev->dm.dc->debug.hdcp_lc_enable_sw_fallback = true; + if (amdgpu_dc_debug_mask & DC_SKIP_DETECTION_LT) + adev->dm.dc->debug.skip_detection_link_training = true; + adev->dm.dc->debug.visual_confirm = amdgpu_dc_visual_confirm; /* TODO: Remove after DP2 receiver gets proper support of Cable ID feature */ @@ -4911,6 +4910,7 @@ amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector) struct backlight_properties props = { 0 }; struct amdgpu_dm_backlight_caps caps = { 0 }; char bl_name[16]; + int min, max; if (aconnector->bl_idx == -1) return; @@ -4923,11 +4923,15 @@ amdgpu_dm_register_backlight_device(struct amdgpu_dm_connector *aconnector) } amdgpu_acpi_get_backlight_caps(&caps); - if (caps.caps_valid) { + if (caps.caps_valid && get_brightness_range(&caps, &min, &max)) { if (power_supply_is_system_supplied() > 0) - props.brightness = caps.ac_level; + props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps.ac_level, 100); else - props.brightness = caps.dc_level; + props.brightness = (max - min) * DIV_ROUND_CLOSEST(caps.dc_level, 100); + /* min is zero, so max needs to be adjusted */ + props.max_brightness = max - min; + drm_dbg(drm, "Backlight caps: min: %d, max: %d, ac %d, dc %d\n", min, max, + caps.ac_level, caps.dc_level); } else props.brightness = AMDGPU_MAX_BL_LEVEL; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c index e8bdd7f0c460..87058271b00c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_crtc.c @@ -246,8 +246,6 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) struct vblank_control_work *vblank_work = container_of(work, struct vblank_control_work, work); struct amdgpu_display_manager *dm = vblank_work->dm; - struct amdgpu_device *adev = drm_to_adev(dm->ddev); - int r; mutex_lock(&dm->dc_lock); @@ -275,15 +273,8 @@ static void amdgpu_dm_crtc_vblank_control_worker(struct work_struct *work) vblank_work->acrtc->dm_irq_params.allow_sr_entry); } - if (dm->active_vblank_irq_count == 0) { - r = amdgpu_dpm_pause_power_profile(adev, true); - if (r) - dev_warn(adev->dev, "failed to set default power profile mode\n"); + if (dm->active_vblank_irq_count == 0) dc_allow_idle_optimizations(dm->dc, true); - r = amdgpu_dpm_pause_power_profile(adev, false); - if (r) - dev_warn(adev->dev, "failed to restore the power profile mode\n"); - } mutex_unlock(&dm->dc_lock); diff --git a/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c b/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c index 681799468487..d897f8a30ede 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/basics/dce_calcs.c @@ -1393,7 +1393,7 @@ static void calculate_bandwidth( if ((bw_mtn(data->dram_speed_change_margin, bw_int_to_fixed(0)) && bw_ltn(data->dram_speed_change_margin, bw_int_to_fixed(9999)))) { /*determine the minimum dram clock change margin for each set of clock frequencies*/ data->min_dram_speed_change_margin[i][j] = bw_min2(data->min_dram_speed_change_margin[i][j], data->dram_speed_change_margin); - /*compute the maximum clock frequuency required for the dram clock change at each set of clock frequencies*/ + /*compute the maximum clock frequency required for the dram clock change at each set of clock frequencies*/ data->dispclk_required_for_dram_speed_change_pipe[i][j] = bw_max2(bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->active_time[k])))); if ((bw_ltn(data->dispclk_required_for_dram_speed_change_pipe[i][j], vbios->high_voltage_max_dispclk))) { data->display_pstate_change_enable[k] = 1; @@ -1407,7 +1407,7 @@ static void calculate_bandwidth( if ((bw_mtn(data->dram_speed_change_margin, bw_int_to_fixed(0)) && bw_ltn(data->dram_speed_change_margin, bw_int_to_fixed(9999)))) { /*determine the minimum dram clock change margin for each display pipe*/ data->min_dram_speed_change_margin[i][j] = bw_min2(data->min_dram_speed_change_margin[i][j], data->dram_speed_change_margin); - /*compute the maximum clock frequuency required for the dram clock change at each set of clock frequencies*/ + /*compute the maximum clock frequency required for the dram clock change at each set of clock frequencies*/ data->dispclk_required_for_dram_speed_change_pipe[i][j] = bw_max2(bw_div(bw_div(bw_mul(data->src_pixels_for_first_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_sub(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->mcifwr_burst_time[i][j]))), bw_div(bw_div(bw_mul(data->src_pixels_for_last_output_pixel[k], dceip->display_pipe_throughput_factor), dceip->lb_write_pixels_per_dispclk), (bw_add(bw_sub(bw_sub(bw_sub(bw_sub(data->maximum_latency_hiding_with_cursor[k], vbios->nbp_state_change_latency), data->dmif_burst_time[i][j]), data->dram_speed_change_line_source_transfer_time[k][i][j]), data->mcifwr_burst_time[i][j]), data->active_time[k])))); if ((bw_ltn(data->dispclk_required_for_dram_speed_change_pipe[i][j], vbios->high_voltage_max_dispclk))) { data->display_pstate_change_enable[k] = 1; diff --git a/drivers/gpu/drm/amd/display/dc/gpio/hw_hpd.c b/drivers/gpu/drm/amd/display/dc/gpio/hw_hpd.c index 3f13a744d07d..01ec451004f7 100644 --- a/drivers/gpu/drm/amd/display/dc/gpio/hw_hpd.c +++ b/drivers/gpu/drm/amd/display/dc/gpio/hw_hpd.c @@ -62,7 +62,7 @@ static void dal_hw_hpd_destroy( *ptr = NULL; } -static enum gpio_result get_value( +static enum gpio_result dal_hw_hpd_get_value( const struct hw_gpio_pin *ptr, uint32_t *value) { @@ -85,7 +85,7 @@ static enum gpio_result get_value( return dal_hw_gpio_get_value(ptr, value); } -static enum gpio_result set_config( +static enum gpio_result dal_hw_hpd_set_config( struct hw_gpio_pin *ptr, const struct gpio_config_data *config_data) { @@ -104,9 +104,9 @@ static enum gpio_result set_config( static const struct hw_gpio_pin_funcs funcs = { .destroy = dal_hw_hpd_destroy, .open = dal_hw_gpio_open, - .get_value = get_value, + .get_value = dal_hw_hpd_get_value, .set_value = dal_hw_gpio_set_value, - .set_config = set_config, + .set_config = dal_hw_hpd_set_config, .change_mode = dal_hw_gpio_change_mode, .close = dal_hw_gpio_close, }; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c index 23bec5d25ed6..e8730cc40edb 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dce110/dce110_hwseq.c @@ -952,8 +952,8 @@ void dce110_edp_backlight_control( struct dc_context *ctx = link->ctx; struct bp_transmitter_control cntl = { 0 }; uint8_t pwrseq_instance = 0; - unsigned int pre_T11_delay = OLED_PRE_T11_DELAY; - unsigned int post_T7_delay = OLED_POST_T7_DELAY; + unsigned int pre_T11_delay = (link->dpcd_sink_ext_caps.bits.oled ? OLED_PRE_T11_DELAY : 0); + unsigned int post_T7_delay = (link->dpcd_sink_ext_caps.bits.oled ? OLED_POST_T7_DELAY : 0); if (dal_graphics_object_id_get_connector_id(link->link_enc->connector) != CONNECTOR_ID_EDP) { @@ -1069,7 +1069,8 @@ void dce110_edp_backlight_control( if (!enable) { /*follow oem panel config's requirement*/ pre_T11_delay += link->panel_config.pps.extra_pre_t11_ms; - msleep(pre_T11_delay); + if (pre_T11_delay) + msleep(pre_T11_delay); } } @@ -1220,6 +1221,9 @@ void dce110_blank_stream(struct pipe_ctx *pipe_ctx) struct dc_link *link = stream->link; struct dce_hwseq *hws = link->dc->hwseq; + if (hws && hws->wa_state.skip_blank_stream) + return; + if (link->local_sink && link->local_sink->sink_signal == SIGNAL_TYPE_EDP) { if (!link->skip_implict_edp_power_control) hws->funcs.edp_backlight_control(link, false); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index 858288c3b1ac..c277df12c817 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -76,6 +76,7 @@ void dcn20_log_color_state(struct dc *dc, { struct dc_context *dc_ctx = dc->ctx; struct resource_pool *pool = dc->res_pool; + bool is_gamut_remap_available = false; int i; DTN_INFO("DPP: DGAM mode SHAPER mode 3DLUT mode 3DLUT bit depth" @@ -89,15 +90,15 @@ void dcn20_log_color_state(struct dc *dc, struct dcn_dpp_state s = {0}; dpp->funcs->dpp_read_state(dpp, &s); - dpp->funcs->dpp_get_gamut_remap(dpp, &s.gamut_remap); + if (dpp->funcs->dpp_get_gamut_remap) { + dpp->funcs->dpp_get_gamut_remap(dpp, &s.gamut_remap); + is_gamut_remap_available = true; + } if (!s.is_enabled) continue; - DTN_INFO("[%2d]: %8s %11s %10s %15s %10s %9s %12s " - "%010lld %010lld %010lld %010lld " - "%010lld %010lld %010lld %010lld " - "%010lld %010lld %010lld %010lld", + DTN_INFO("[%2d]: %8s %11s %10s %15s %10s %9s", dpp->inst, (s.dgam_lut_mode == 0) ? "Bypass" : ((s.dgam_lut_mode == 1) ? "sRGB" : @@ -114,10 +115,17 @@ void dcn20_log_color_state(struct dc *dc, (s.lut3d_bit_depth <= 0) ? "12-bit" : "10-bit", (s.lut3d_size == 0) ? "17x17x17" : "9x9x9", (s.rgam_lut_mode == 1) ? "RAM A" : - ((s.rgam_lut_mode == 1) ? "RAM B" : "Bypass"), + ((s.rgam_lut_mode == 1) ? "RAM B" : "Bypass")); + + if (is_gamut_remap_available) { + DTN_INFO(" %12s " + "%010lld %010lld %010lld %010lld " + "%010lld %010lld %010lld %010lld " + "%010lld %010lld %010lld %010lld", + (s.gamut_remap.gamut_adjust_type == 0) ? "Bypass" : - ((s.gamut_remap.gamut_adjust_type == 1) ? "HW" : - "SW"), + ((s.gamut_remap.gamut_adjust_type == 1) ? "HW" : + "SW"), s.gamut_remap.temperature_matrix[0].value, s.gamut_remap.temperature_matrix[1].value, s.gamut_remap.temperature_matrix[2].value, @@ -130,6 +138,8 @@ void dcn20_log_color_state(struct dc *dc, s.gamut_remap.temperature_matrix[9].value, s.gamut_remap.temperature_matrix[10].value, s.gamut_remap.temperature_matrix[11].value); + } + DTN_INFO("\n"); } DTN_INFO("\n"); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index e89ebfda4873..37a239219dfe 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -74,6 +74,7 @@ void dcn30_log_color_state(struct dc *dc, { struct dc_context *dc_ctx = dc->ctx; struct resource_pool *pool = dc->res_pool; + bool is_gamut_remap_available = false; int i; DTN_INFO("DPP: DGAM ROM DGAM ROM type DGAM LUT SHAPER mode" @@ -88,16 +89,16 @@ void dcn30_log_color_state(struct dc *dc, struct dcn_dpp_state s = {0}; dpp->funcs->dpp_read_state(dpp, &s); - dpp->funcs->dpp_get_gamut_remap(dpp, &s.gamut_remap); + + if (dpp->funcs->dpp_get_gamut_remap) { + dpp->funcs->dpp_get_gamut_remap(dpp, &s.gamut_remap); + is_gamut_remap_available = true; + } if (!s.is_enabled) continue; - DTN_INFO("[%2d]: %7x %13s %8s %11s %10s %15s %10s %9s" - " %12s " - "%010lld %010lld %010lld %010lld " - "%010lld %010lld %010lld %010lld " - "%010lld %010lld %010lld %010lld", + DTN_INFO("[%2d]: %7x %13s %8s %11s %10s %15s %10s %9s", dpp->inst, s.pre_dgam_mode, (s.pre_dgam_select == 0) ? "sRGB" : @@ -121,7 +122,14 @@ void dcn30_log_color_state(struct dc *dc, (s.lut3d_size == 0) ? "17x17x17" : "9x9x9", (s.rgam_lut_mode == 0) ? "Bypass" : ((s.rgam_lut_mode == 1) ? "RAM A" : - "RAM B"), + "RAM B")); + + if (is_gamut_remap_available) { + DTN_INFO(" %12s " + "%010lld %010lld %010lld %010lld " + "%010lld %010lld %010lld %010lld " + "%010lld %010lld %010lld %010lld", + (s.gamut_remap.gamut_adjust_type == 0) ? "Bypass" : ((s.gamut_remap.gamut_adjust_type == 1) ? "HW" : "SW"), @@ -137,6 +145,8 @@ void dcn30_log_color_state(struct dc *dc, s.gamut_remap.temperature_matrix[9].value, s.gamut_remap.temperature_matrix[10].value, s.gamut_remap.temperature_matrix[11].value); + } + DTN_INFO("\n"); } DTN_INFO("\n"); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c index f38340aa3f15..5ba3999991b0 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c @@ -526,9 +526,15 @@ static void dcn31_reset_back_end_for_pipe( link = pipe_ctx->stream->link; + if (dc->hwseq) + dc->hwseq->wa_state.skip_blank_stream = false; + if ((!pipe_ctx->stream->dpms_off || link->link_status.link_active) && - (link->connector_signal == SIGNAL_TYPE_EDP)) + (link->connector_signal == SIGNAL_TYPE_EDP)) { dc->hwss.blank_stream(pipe_ctx); + if (dc->hwseq) + dc->hwseq->wa_state.skip_blank_stream = true; + } pipe_ctx->stream_res.tg->funcs->set_dsc_config( pipe_ctx->stream_res.tg, @@ -570,7 +576,8 @@ static void dcn31_reset_back_end_for_pipe( pipe_ctx->stream_res.audio = NULL; } } - + if (dc->hwseq) + dc->hwseq->wa_state.skip_blank_stream = false; pipe_ctx->stream = NULL; DC_LOG_DEBUG("Reset back end for pipe %d, tg:%d\n", pipe_ctx->pipe_idx, pipe_ctx->stream_res.tg->inst); diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h index 09bc65c2fa23..1e2d247fbbac 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer_private.h @@ -49,6 +49,7 @@ struct hwseq_wa_state { bool DEGVIDCN10_253_applied; bool disallow_self_refresh_during_multi_plane_transition_applied; unsigned int disallow_self_refresh_during_multi_plane_transition_applied_on_frame; + bool skip_blank_stream; }; struct pipe_ctx; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index 8f79881ad9f1..a5127c2d47ef 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -2023,7 +2023,7 @@ static bool retrieve_link_cap(struct dc_link *link) /* Read DP tunneling information. */ status = dpcd_get_tunneling_device_data(link); if (status != DC_OK) - dm_error("%s: Read DP tunneling device data failed.\n", __func__); + DC_LOG_DP2("%s: Read DP tunneling device data failed.\n", __func__); retrieve_cable_id(link); dpcd_write_cable_id_to_dprx(link); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.c index 81857ce6d68d..e7a90a437fff 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.c @@ -502,7 +502,7 @@ void optc2_get_last_used_drr_vtotal(struct timing_generator *optc, uint32_t *ref REG_GET(OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, refresh_rate); } -static struct timing_generator_funcs dcn20_tg_funcs = { +static const struct timing_generator_funcs dcn20_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn201/dcn201_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn201/dcn201_optc.c index f2415eebdc09..772a8bfb949c 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn201/dcn201_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn201/dcn201_optc.c @@ -129,7 +129,7 @@ static void optc201_get_optc_source(struct timing_generator *optc, *num_of_src_opp = 1; } -static struct timing_generator_funcs dcn201_tg_funcs = { +static const struct timing_generator_funcs dcn201_tg_funcs = { .validate_timing = optc201_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c index 78b58a449fa4..ee4665aa49e9 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c @@ -357,7 +357,7 @@ void optc3_tg_init(struct timing_generator *optc) optc1_clear_optc_underflow(optc); } -static struct timing_generator_funcs dcn30_tg_funcs = { +static const struct timing_generator_funcs dcn30_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c index 65e9089b7f31..38f85bc2681a 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c @@ -109,7 +109,7 @@ void optc301_setup_manual_trigger(struct timing_generator *optc) OTG_TRIGA_CLEAR, 1); } -static struct timing_generator_funcs dcn30_tg_funcs = { +static const struct timing_generator_funcs dcn30_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c index ef536f37b4ed..4f1830ba619f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.c @@ -315,7 +315,7 @@ void optc31_read_otg_state(struct timing_generator *optc, s->otg_double_buffer_control = REG_READ(OTG_DOUBLE_BUFFER_CONTROL); } -static struct timing_generator_funcs dcn31_tg_funcs = { +static const struct timing_generator_funcs dcn31_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c index 0e603bad0d12..4a2caca37255 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.c @@ -192,7 +192,7 @@ static void optc314_set_h_timing_div_manual_mode(struct timing_generator *optc, } -static struct timing_generator_funcs dcn314_tg_funcs = { +static const struct timing_generator_funcs dcn314_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c index 2cdd19ba634b..b2b226bcd871 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c @@ -297,7 +297,7 @@ static void optc32_set_drr( optc32_setup_manual_trigger(optc); } -static struct timing_generator_funcs dcn32_tg_funcs = { +static const struct timing_generator_funcs dcn32_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c index 4cfc6c0fa147..72bff94cb57d 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.c @@ -428,7 +428,7 @@ static void optc35_set_long_vtotal( } } -static struct timing_generator_funcs dcn35_tg_funcs = { +static const struct timing_generator_funcs dcn35_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index 382ac18e7854..ff79c38287df 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -459,7 +459,7 @@ bool optc401_wait_update_lock_status(struct timing_generator *tg, bool locked) return true; } -static struct timing_generator_funcs dcn401_tg_funcs = { +static const struct timing_generator_funcs dcn401_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, .setup_vertical_interrupt0 = optc1_setup_vertical_interrupt0, 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 e0e32975ca34..f420c4dafa03 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 @@ -1938,8 +1938,8 @@ static bool dcn401_resource_construct( dc->caps.color.dpp.gamma_corr = 1; dc->caps.color.dpp.dgam_rom_for_yuv = 0; - dc->caps.color.dpp.hw_3d_lut = 1; - dc->caps.color.dpp.ogam_ram = 1; + dc->caps.color.dpp.hw_3d_lut = 0; + dc->caps.color.dpp.ogam_ram = 0; // no OGAM ROM on DCN2 and later ASICs dc->caps.color.dpp.ogam_rom_caps.srgb = 0; dc->caps.color.dpp.ogam_rom_caps.bt2020 = 0; 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 57fa05bddb45..b66bd10cdc9b 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -2139,11 +2139,6 @@ union dmub_cmd_fams2_config { } stream_v1; //v1 }; -struct dmub_fams2_config_v2 { - struct dmub_cmd_fams2_global_config global; - struct dmub_fams2_stream_static_state_v1 stream_v1[DMUB_MAX_STREAMS]; //v1 -}; - /** * DMUB rb command definition for FAMS2 (merged SubVP, FPO, Legacy) */ @@ -2153,22 +2148,6 @@ struct dmub_rb_cmd_fams2 { }; /** - * Indirect buffer descriptor - */ -struct dmub_ib_data { - union dmub_addr src; // location of indirect buffer in memory - uint16_t size; // indirect buffer size in bytes -}; - -/** - * DMUB rb command definition for commands passed over indirect buffer - */ -struct dmub_rb_cmd_ib { - struct dmub_cmd_header header; - struct dmub_ib_data ib_data; -}; - -/** * enum dmub_cmd_idle_opt_type - Idle optimization command type. */ enum dmub_cmd_idle_opt_type { @@ -2191,11 +2170,6 @@ enum dmub_cmd_idle_opt_type { * DCN hardware notify power state. */ DMUB_CMD__IDLE_OPT_SET_DC_POWER_STATE = 3, - - /** - * DCN notify to release HW. - */ - DMUB_CMD__IDLE_OPT_RELEASE_HW = 4, }; /** @@ -2957,9 +2931,8 @@ enum dmub_cmd_fams_type { */ DMUB_CMD__FAMS_SET_MANUAL_TRIGGER = 3, DMUB_CMD__FAMS2_CONFIG = 4, - DMUB_CMD__FAMS2_IB_CONFIG = 5, - DMUB_CMD__FAMS2_DRR_UPDATE = 6, - DMUB_CMD__FAMS2_FLIP = 7, + DMUB_CMD__FAMS2_DRR_UPDATE = 5, + DMUB_CMD__FAMS2_FLIP = 6, }; /** @@ -5953,11 +5926,8 @@ union dmub_rb_cmd { * Definition of a DMUB_CMD__PSP_ASSR_ENABLE command. */ struct dmub_rb_cmd_assr_enable assr_enable; - struct dmub_rb_cmd_fams2 fams2_config; - struct dmub_rb_cmd_ib ib_fams2_config; - struct dmub_rb_cmd_fams2_drr_update fams2_drr_update; struct dmub_rb_cmd_fams2_flip fams2_flip; diff --git a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h index 813463ffe15c..cc467031651d 100644 --- a/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h +++ b/drivers/gpu/drm/amd/display/include/grph_object_ctrl_defs.h @@ -424,7 +424,7 @@ struct integrated_info { /* * DFS-bypass flag */ -/* Copy of SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS from atombios.h */ +/* Copy of SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS from atombios.h */ enum { DFS_BYPASS_ENABLE = 0x10 }; diff --git a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.c b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.c index 8c137d7c032e..e58e7b93810b 100644 --- a/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.c +++ b/drivers/gpu/drm/amd/display/modules/hdcp/hdcp_psp.c @@ -368,6 +368,9 @@ enum mod_hdcp_status mod_hdcp_hdcp1_enable_encryption(struct mod_hdcp *hdcp) struct mod_hdcp_display *display = get_first_active_display(hdcp); enum mod_hdcp_status status = MOD_HDCP_STATUS_SUCCESS; + if (!display) + return MOD_HDCP_STATUS_DISPLAY_NOT_FOUND; + mutex_lock(&psp->hdcp_context.mutex); hdcp_cmd = (struct ta_hdcp_shared_memory *)psp->hdcp_context.context.mem_context.shared_buf; memset(hdcp_cmd, 0, sizeof(struct ta_hdcp_shared_memory)); diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h index c8eccee9b023..11374a2cbab8 100644 --- a/drivers/gpu/drm/amd/include/amd_shared.h +++ b/drivers/gpu/drm/amd/include/amd_shared.h @@ -351,9 +351,10 @@ enum DC_DEBUG_MASK { DC_DISABLE_HDMI_CEC = 0x10000, /** - * @DC_DISABLE_SUBVP: If set, disable DCN Sub-Viewport feature in amdgpu driver. + * @DC_DISABLE_SUBVP_FAMS: If set, disable DCN Sub-Viewport & Firmware Assisted + * Memory Clock Switching (FAMS) feature in amdgpu driver. */ - DC_DISABLE_SUBVP = 0x20000, + DC_DISABLE_SUBVP_FAMS = 0x20000, /** * @DC_DISABLE_CUSTOM_BRIGHTNESS_CURVE: If set, disable support for custom brightness curves */ @@ -370,6 +371,11 @@ enum DC_DEBUG_MASK { * path failure, retry using legacy SW path. */ DC_HDCP_LC_ENABLE_SW_FALLBACK = 0x100000, + + /** + * @DC_SKIP_DETECTION_LT: If set, skip detection link training + */ + DC_SKIP_DETECTION_LT = 0x200000, }; enum amd_dpm_forced_level; diff --git a/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_offset.h index 15e5a65cf492..70ee6be94a9b 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_offset.h @@ -9776,6 +9776,14 @@ #define regDIG0_DIG_BE_CNTL_BASE_IDX 2 #define regDIG0_DIG_BE_EN_CNTL 0x20bd #define regDIG0_DIG_BE_EN_CNTL_BASE_IDX 2 +#define regDIG0_HDCP_INT_CONTROL 0x20c0 +#define regDIG0_HDCP_INT_CONTROL_BASE_IDX 2 +#define regDIG0_HDCP_LINK0_STATUS 0x20c1 +#define regDIG0_HDCP_LINK0_STATUS_BASE_IDX 2 +#define regDIG0_HDCP_I2C_CONTROL_0 0x20c2 +#define regDIG0_HDCP_I2C_CONTROL_0_BASE_IDX 2 +#define regDIG0_HDCP_I2C_CONTROL_1 0x20c3 +#define regDIG0_HDCP_I2C_CONTROL_1_BASE_IDX 2 #define regDIG0_TMDS_CNTL 0x20e4 #define regDIG0_TMDS_CNTL_BASE_IDX 2 #define regDIG0_TMDS_CONTROL_CHAR 0x20e5 @@ -10081,6 +10089,12 @@ #define regDIG1_DIG_BE_CNTL_BASE_IDX 2 #define regDIG1_DIG_BE_EN_CNTL 0x21e1 #define regDIG1_DIG_BE_EN_CNTL_BASE_IDX 2 +#define regDIG1_HDCP_INT_CONTROL 0x21e4 +#define regDIG1_HDCP_INT_CONTROL_BASE_IDX 2 +#define regDIG1_HDCP_I2C_CONTROL_0 0x21e6 +#define regDIG1_HDCP_I2C_CONTROL_0_BASE_IDX 2 +#define regDIG1_HDCP_I2C_CONTROL_1 0x21e7 +#define regDIG1_HDCP_I2C_CONTROL_1_BASE_IDX 2 #define regDIG1_TMDS_CNTL 0x2208 #define regDIG1_TMDS_CNTL_BASE_IDX 2 #define regDIG1_TMDS_CONTROL_CHAR 0x2209 @@ -10386,6 +10400,12 @@ #define regDIG2_DIG_BE_CNTL_BASE_IDX 2 #define regDIG2_DIG_BE_EN_CNTL 0x2305 #define regDIG2_DIG_BE_EN_CNTL_BASE_IDX 2 +#define regDIG2_HDCP_INT_CONTROL 0x2308 +#define regDIG2_HDCP_INT_CONTROL_BASE_IDX 2 +#define regDIG2_HDCP_I2C_CONTROL_0 0x230a +#define regDIG2_HDCP_I2C_CONTROL_0_BASE_IDX 2 +#define regDIG2_HDCP_I2C_CONTROL_1 0x230b +#define regDIG2_HDCP_I2C_CONTROL_1_BASE_IDX 2 #define regDIG2_TMDS_CNTL 0x232c #define regDIG2_TMDS_CNTL_BASE_IDX 2 #define regDIG2_TMDS_CONTROL_CHAR 0x232d @@ -10691,6 +10711,12 @@ #define regDIG3_DIG_BE_CNTL_BASE_IDX 2 #define regDIG3_DIG_BE_EN_CNTL 0x2429 #define regDIG3_DIG_BE_EN_CNTL_BASE_IDX 2 +#define regDIG3_HDCP_INT_CONTROL 0x242c +#define regDIG3_HDCP_INT_CONTROL_BASE_IDX 2 +#define regDIG3_HDCP_I2C_CONTROL_0 0x242e +#define regDIG3_HDCP_I2C_CONTROL_0_BASE_IDX 2 +#define regDIG3_HDCP_I2C_CONTROL_1 0x242f +#define regDIG3_HDCP_I2C_CONTROL_1_BASE_IDX 2 #define regDIG3_TMDS_CNTL 0x2450 #define regDIG3_TMDS_CNTL_BASE_IDX 2 #define regDIG3_TMDS_CONTROL_CHAR 0x2451 diff --git a/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_sh_mask.h index 5d9d5fea6e06..e3d841b2e9af 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/dcn/dcn_4_1_0_sh_mask.h @@ -2847,6 +2847,14 @@ #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_AUTH_FAIL_INTERRUPT_DEST__SHIFT 0x1 #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_I2C_XFER_REQ_INTERRUPT_DEST__SHIFT 0x2 #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_I2C_XFER_DONE_INTERRUPT_DEST__SHIFT 0x3 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_AUTH_SUCCESS_INTERRUPT_DEST__SHIFT 0x4 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_AUTH_FAIL_INTERRUPT_DEST__SHIFT 0x5 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_I2C_XFER_REQ_INTERRUPT_DEST__SHIFT 0x6 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_I2C_XFER_DONE_INTERRUPT_DEST__SHIFT 0x7 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_AUTH_SUCCESS_INTERRUPT_DEST__SHIFT 0x8 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_AUTH_FAIL_INTERRUPT_DEST__SHIFT 0x9 +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_I2C_XFER_REQ_INTERRUPT_DEST__SHIFT 0xa +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_I2C_XFER_DONE_INTERRUPT_DEST__SHIFT 0xb #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_AUTH_SUCCESS_INTERRUPT_DEST__SHIFT 0xc #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_AUTH_FAIL_INTERRUPT_DEST__SHIFT 0xd #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_I2C_XFER_REQ_INTERRUPT_DEST__SHIFT 0xe @@ -2871,6 +2879,14 @@ #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_AUTH_FAIL_INTERRUPT_DEST_MASK 0x00000002L #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_I2C_XFER_REQ_INTERRUPT_DEST_MASK 0x00000004L #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP0_I2C_XFER_DONE_INTERRUPT_DEST_MASK 0x00000008L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_AUTH_SUCCESS_INTERRUPT_DEST_MASK 0x00000010L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_AUTH_FAIL_INTERRUPT_DEST_MASK 0x00000020L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_I2C_XFER_REQ_INTERRUPT_DEST_MASK 0x00000040L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP1_I2C_XFER_DONE_INTERRUPT_DEST_MASK 0x00000080L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_AUTH_SUCCESS_INTERRUPT_DEST_MASK 0x00000100L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_AUTH_FAIL_INTERRUPT_DEST_MASK 0x00000200L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_I2C_XFER_REQ_INTERRUPT_DEST_MASK 0x00000400L +#define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP2_I2C_XFER_DONE_INTERRUPT_DEST_MASK 0x00000800L #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_AUTH_SUCCESS_INTERRUPT_DEST_MASK 0x00001000L #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_AUTH_FAIL_INTERRUPT_DEST_MASK 0x00002000L #define HDCP_INTERRUPT_DEST__DOUT_IHC_HDCP3_I2C_XFER_REQ_INTERRUPT_DEST_MASK 0x00004000L diff --git a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_offset.h b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_offset.h index c4aaa86a95e2..72a118b2af69 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_offset.h @@ -1067,7 +1067,13 @@ #define regVCN_FEATURES_BASE_IDX 1 #define regUVD_GPUIOV_STATUS 0x0055 #define regUVD_GPUIOV_STATUS_BASE_IDX 1 +#define regUVD_RAS_VCPU_VCODEC_STATUS 0x0057 +#define regUVD_RAS_VCPU_VCODEC_STATUS_BASE_IDX 1 #define regUVD_SCRATCH15 0x005c +#define regUVD_RAS_JPEG0_STATUS 0x0059 +#define regUVD_RAS_JPEG0_STATUS_BASE_IDX 1 +#define regUVD_RAS_JPEG1_STATUS 0x005a +#define regUVD_RAS_JPEG1_STATUS_BASE_IDX 1 #define regUVD_SCRATCH15_BASE_IDX 1 #define regUVD_VERSION 0x005d #define regUVD_VERSION_BASE_IDX 1 diff --git a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_sh_mask.h index bd7242e4e9c6..c78b09d6fbae 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/vcn/vcn_5_0_0_sh_mask.h @@ -5714,6 +5714,22 @@ //UVD_GPUIOV_STATUS #define UVD_GPUIOV_STATUS__UVD_GPUIOV_STATUS_VF_ENABLE__SHIFT 0x0 #define UVD_GPUIOV_STATUS__UVD_GPUIOV_STATUS_VF_ENABLE_MASK 0x00000001L +//UVD_RAS_VCPU_VCODEC_STATUS +#define UVD_RAS_VCPU_VCODEC_STATUS__POISONED_VF__SHIFT 0x0 +#define UVD_RAS_VCPU_VCODEC_STATUS__POISONED_PF__SHIFT 0x1f +#define UVD_RAS_VCPU_VCODEC_STATUS__POISONED_VF_MASK 0x7FFFFFFFL +#define UVD_RAS_VCPU_VCODEC_STATUS__POISONED_PF_MASK 0x80000000L + +//UVD_RAS_JPEG0_STATUS +#define UVD_RAS_JPEG0_STATUS__POISONED_VF__SHIFT 0x0 +#define UVD_RAS_JPEG0_STATUS__POISONED_PF__SHIFT 0x1f +#define UVD_RAS_JPEG0_STATUS__POISONED_VF_MASK 0x7FFFFFFFL +#define UVD_RAS_JPEG0_STATUS__POISONED_PF_MASK 0x80000000L +//UVD_RAS_JPEG1_STATUS +#define UVD_RAS_JPEG1_STATUS__POISONED_VF__SHIFT 0x0 +#define UVD_RAS_JPEG1_STATUS__POISONED_PF__SHIFT 0x1f +#define UVD_RAS_JPEG1_STATUS__POISONED_VF_MASK 0x7FFFFFFFL +#define UVD_RAS_JPEG1_STATUS__POISONED_PF_MASK 0x80000000L //UVD_SCRATCH15 #define UVD_SCRATCH15__SCRATCH15_DATA__SHIFT 0x0 #define UVD_SCRATCH15__SCRATCH15_DATA_MASK 0xFFFFFFFFL diff --git a/drivers/gpu/drm/amd/include/atombios.h b/drivers/gpu/drm/amd/include/atombios.h index 52bac19fb404..b344acefc606 100644 --- a/drivers/gpu/drm/amd/include/atombios.h +++ b/drivers/gpu/drm/amd/include/atombios.h @@ -6017,7 +6017,7 @@ typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 #define SYS_INFO_GPUCAPS__TMDSHDMI_COHERENT_SINGLEPLL_MODE 0x01 #define SYS_INFO_GPUCAPS__DP_SINGLEPLL_MODE 0x02 #define SYS_INFO_GPUCAPS__DISABLE_AUX_MODE_DETECT 0x08 -#define SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS 0x10 +#define SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS 0x10 //ulGPUCapInfo[16]=1 indicate SMC firmware is able to support GNB fast resume function, so that driver can call SMC to program most of GNB register during resuming, from ML #define SYS_INFO_GPUCAPS__GNB_FAST_RESUME_CAPABLE 0x00010000 @@ -6460,7 +6460,7 @@ typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_9 // ulGPUCapInfo #define SYS_INFO_V1_9_GPUCAPSINFO_DISABLE_AUX_MODE_DETECT 0x08 -#define SYS_INFO_V1_9_GPUCAPSINFO_ENABEL_DFS_BYPASS 0x10 +#define SYS_INFO_V1_9_GPUCAPSINFO_ENABLE_DFS_BYPASS 0x10 //ulGPUCapInfo[16]=1 indicate SMC firmware is able to support GNB fast resume function, so that driver can call SMC to program most of GNB register during resuming, from ML #define SYS_INFO_V1_9_GPUCAPSINFO_GNB_FAST_RESUME_CAPABLE 0x00010000 //ulGPUCapInfo[18]=1 indicate the IOMMU is not available diff --git a/drivers/gpu/drm/amd/include/atomfirmware.h b/drivers/gpu/drm/amd/include/atomfirmware.h index 2d1135bdc4b9..5c86423c2e92 100644 --- a/drivers/gpu/drm/amd/include/atomfirmware.h +++ b/drivers/gpu/drm/amd/include/atomfirmware.h @@ -1714,7 +1714,7 @@ enum atom_system_vbiosmisc_def{ // gpucapinfo enum atom_system_gpucapinf_def{ - SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS = 0x10, + SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS = 0x10, }; //dpphy_override diff --git a/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_11_0_0.h b/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_11_0_0.h index 3a4670bc4449..b98b7ae551b5 100644 --- a/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_11_0_0.h +++ b/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_11_0_0.h @@ -48,6 +48,7 @@ #define GFX_11_0_0__SRCID__SDMA_SRAM_ECC 64 // 0x40 SRAM ECC Error #define GFX_11_0_0__SRCID__SDMA_SEM_INCOMPLETE_TIMEOUT 65 // 0x41 GPF(Sem incomplete timeout) #define GFX_11_0_0__SRCID__SDMA_SEM_WAIT_FAIL_TIMEOUT 66 // 0x42 Semaphore wait fail timeout +#define GFX_11_0_0__SRCID__SDMA_FENCE 67 // 0x43 User fence #define GFX_11_0_0__SRCID__RLC_GC_FED_INTERRUPT 128 // 0x80 FED Interrupt (for data poisoning) diff --git a/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_12_0_0.h b/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_12_0_0.h new file mode 100644 index 000000000000..467897ec2e65 --- /dev/null +++ b/drivers/gpu/drm/amd/include/ivsrcid/gfx/irqsrcs_gfx_12_0_0.h @@ -0,0 +1,74 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2024 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + * + */ +#ifndef __IRQSRCS_GFX_12_0_0_H__ +#define __IRQSRCS_GFX_12_0_0_H__ + +#define GFX_12_0_0__SRCID__UTCL2_FAULT 0 // UTCL2 has encountered a fault or retry scenario +#define GFX_12_0_0__SRCID__UTCL2_DATA_POISONING 1 // UTCL2 for data poisoning +#define GFX_12_0_0__SRCID__MEM_ACCES_MON 10 // 0x0A EA memory access monitor interrupt +#define GFX_12_0_0__SRCID__SDMA_ATOMIC_RTN_DONE 48 // 0x30 SDMA atomic*_rtn ops complete +#define GFX_12_0_0__SRCID__SDMA_TRAP 49 // 0x31 Trap +#define GFX_12_0_0__SRCID__SDMA_SRBMWRITE 50 // 0x32 SRBM write Protection +#define GFX_12_0_0__SRCID__SDMA_CTXEMPTY 51 // 0x33 Context Empty +#define GFX_12_0_0__SRCID__SDMA_PREEMPT 52 // 0x34 SDMA New Run List +#define GFX_12_0_0__SRCID__SDMA_IB_PREEMPT 53 // 0x35 sdma mid - command buffer preempt interrupt +#define GFX_12_0_0__SRCID__SDMA_DOORBELL_INVALID 54 // 0x36 Doorbell BE invalid +#define GFX_12_0_0__SRCID__SDMA_QUEUE_HANG 55 // 0x37 Queue hang or Command timeout +#define GFX_12_0_0__SRCID__SDMA_ATOMIC_TIMEOUT 56 // 0x38 SDMA atomic CMPSWAP loop timeout +#define GFX_12_0_0__SRCID__SDMA_POLL_TIMEOUT 57 // 0x39 SRBM read poll timeout +#define GFX_12_0_0__SRCID__SDMA_PAGE_TIMEOUT 58 // 0x3A Page retry timeout after UTCL2 return nack = 1 +#define GFX_12_0_0__SRCID__SDMA_PAGE_NULL 59 // 0x3B Page Null from UTCL2 when nack = 2 +#define GFX_12_0_0__SRCID__SDMA_PAGE_FAULT 60 // 0x3C Page Fault Error from UTCL2 when nack = 3 +#define GFX_12_0_0__SRCID__SDMA_VM_HOLE 61 // 0x3D MC or SEM address in VM hole +#define GFX_12_0_0__SRCID__SDMA_ECC 62 // 0x3E ECC Error +#define GFX_12_0_0__SRCID__SDMA_FROZEN 63 // 0x3F SDMA Frozen +#define GFX_12_0_0__SRCID__SDMA_SRAM_ECC 64 // 0x40 SRAM ECC Error +#define GFX_12_0_0__SRCID__SDMA_SEM_INCOMPLETE_TIMEOUT 65 // 0x41 GPF(Sem incomplete timeout) +#define GFX_12_0_0__SRCID__SDMA_SEM_WAIT_FAIL_TIMEOUT 66 // 0x42 Semaphore wait fail timeout +#define GFX_12_0_0__SRCID__SDMA_FENCE 70 // 0x46 User fence +#define GFX_12_0_0__SRCID__RLC_GC_FED_INTERRUPT 128 // 0x80 FED Interrupt (for data poisoning) +#define GFX_12_0_0__SRCID__CP_GENERIC_INT 177 // 0xB1 CP_GENERIC int +#define GFX_12_0_0__SRCID__CP_PM4_PKT_RSVD_BIT_ERROR 180 // 0xB4 PM4 Pkt Rsvd Bits Error +#define GFX_12_0_0__SRCID__CP_EOP_INTERRUPT 181 // 0xB5 End-of-Pipe Interrupt +#define GFX_12_0_0__SRCID__CP_BAD_OPCODE_ERROR 183 // 0xB7 Bad Opcode Error +#define GFX_12_0_0__SRCID__CP_PRIV_REG_FAULT 184 // 0xB8 Privileged Register Fault +#define GFX_12_0_0__SRCID__CP_PRIV_INSTR_FAULT 185 // 0xB9 Privileged Instr Fault +#define GFX_12_0_0__SRCID__CP_WAIT_MEM_SEM_FAULT 186 // 0xBA Wait Memory Semaphore Fault (Sync Object Fault) +#define GFX_12_0_0__SRCID__CP_CTX_EMPTY_INTERRUPT 187 // 0xBB Context Empty Interrupt +#define GFX_12_0_0__SRCID__CP_CTX_BUSY_INTERRUPT 188 // 0xBC Context Busy Interrupt +#define GFX_12_0_0__SRCID__CP_ME_WAIT_REG_MEM_POLL_TIMEOUT 192 // 0xC0 CP.ME Wait_Reg_Mem Poll Timeout +#define GFX_12_0_0__SRCID__CP_SIG_INCOMPLETE 193 // 0xC1 "Surface Probe Fault Signal Incomplete" +#define GFX_12_0_0__SRCID__CP_PREEMPT_ACK 194 // 0xC2 Preemption Ack-wledge +#define GFX_12_0_0__SRCID__CP_GPF 195 // 0xC3 General Protection Fault (GPF) +#define GFX_12_0_0__SRCID__CP_GDS_ALLOC_ERROR 196 // 0xC4 GDS Alloc Error +#define GFX_12_0_0__SRCID__CP_ECC_ERROR 197 // 0xC5 ECC Error +#define GFX_12_0_0__SRCID__CP_COMPUTE_QUERY_STATUS 199 // 0xC7 Compute query status +#define GFX_12_0_0__SRCID__CP_VM_DOORBELL 200 // 0xC8 Unattached VM Doorbell Received +#define GFX_12_0_0__SRCID__CP_FUE_ERROR 201 // 0xC9 ECC FUE Error +#define GFX_12_0_0__SRCID__RLC_STRM_PERF_MONITOR_INTERRUPT 202 // 0xCA Streaming Perf Monitor Interrupt +#define GFX_12_0_0__SRCID__GRBM_RD_TIMEOUT_ERROR 232 // 0xE8 CRead timeout error +#define GFX_12_0_0__SRCID__GRBM_REG_GUI_IDLE 233 // 0xE9 Register GUI Idle +#define GFX_12_0_0__SRCID__SQ_INTERRUPT_ID 239 // 0xEF SQ Interrupt (ttrace wrap, errors) + +#endif diff --git a/drivers/gpu/drm/amd/include/kgd_pp_interface.h b/drivers/gpu/drm/amd/include/kgd_pp_interface.h index 0f7542d7074b..f4d914dc731f 100644 --- a/drivers/gpu/drm/amd/include/kgd_pp_interface.h +++ b/drivers/gpu/drm/amd/include/kgd_pp_interface.h @@ -494,6 +494,7 @@ struct amd_pm_funcs { int (*set_df_cstate)(void *handle, enum pp_df_cstate state); int (*set_xgmi_pstate)(void *handle, uint32_t pstate); ssize_t (*get_gpu_metrics)(void *handle, void **table); + ssize_t (*get_xcp_metrics)(void *handle, int xcp_id, void *table); ssize_t (*get_pm_metrics)(void *handle, void *pmmetrics, size_t size); int (*set_watermarks_for_clock_ranges)(void *handle, struct pp_smu_wm_range_sets *ranges); @@ -1592,4 +1593,27 @@ struct amdgpu_pm_metrics { uint8_t data[]; }; +struct amdgpu_partition_metrics_v1_0 { + struct metrics_table_header common_header; + /* Current clocks (Mhz) */ + uint16_t current_gfxclk[MAX_XCC]; + uint16_t current_socclk[MAX_CLKS]; + uint16_t current_vclk0[MAX_CLKS]; + uint16_t current_dclk0[MAX_CLKS]; + uint16_t current_uclk; + uint16_t padding; + + /* Utilization Instantaneous (%) */ + uint32_t gfx_busy_inst[MAX_XCC]; + uint16_t jpeg_busy[NUM_JPEG_ENG_V1]; + uint16_t vcn_busy[NUM_VCN]; + /* Utilization Accumulated (%) */ + uint64_t gfx_busy_acc[MAX_XCC]; + /* Total App Clock Counter Accumulated */ + uint64_t gfx_below_host_limit_ppt_acc[MAX_XCC]; + uint64_t gfx_below_host_limit_thm_acc[MAX_XCC]; + uint64_t gfx_low_utilization_acc[MAX_XCC]; + uint64_t gfx_below_host_limit_total_acc[MAX_XCC]; +}; + #endif diff --git a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c index 2148c8db5a59..5c1cbdc122d2 100644 --- a/drivers/gpu/drm/amd/pm/amdgpu_dpm.c +++ b/drivers/gpu/drm/amd/pm/amdgpu_dpm.c @@ -1697,6 +1697,28 @@ int amdgpu_dpm_is_overdrive_supported(struct amdgpu_device *adev) } } +int amdgpu_dpm_is_overdrive_enabled(struct amdgpu_device *adev) +{ + if (is_support_sw_smu(adev)) { + struct smu_context *smu = adev->powerplay.pp_handle; + + return smu->od_enabled; + } else { + struct pp_hwmgr *hwmgr; + + /* + * dpm on some legacy asics don't carry od_enabled member + * as its pp_handle is casted directly from adev. + */ + if (amdgpu_dpm_is_legacy_dpm(adev)) + return false; + + hwmgr = (struct pp_hwmgr *)adev->powerplay.pp_handle; + + return hwmgr->od_enabled; + } +} + int amdgpu_dpm_set_pp_table(struct amdgpu_device *adev, const char *buf, size_t size) @@ -2019,3 +2041,35 @@ int amdgpu_dpm_get_dpm_clock_table(struct amdgpu_device *adev, return ret; } + +/** + * amdgpu_dpm_get_xcp_metrics - Retrieve metrics for a specific compute + * partition + * @adev: Pointer to the device. + * @xcp_id: Identifier of the XCP for which metrics are to be retrieved. + * @table: Pointer to a buffer where the metrics will be stored. If NULL, the + * function returns the size of the metrics structure. + * + * This function retrieves metrics for a specific XCP, including details such as + * VCN/JPEG activity, clock frequencies, and other performance metrics. If the + * table parameter is NULL, the function returns the size of the metrics + * structure without populating it. + * + * Return: Size of the metrics structure on success, or a negative error code on failure. + */ +ssize_t amdgpu_dpm_get_xcp_metrics(struct amdgpu_device *adev, int xcp_id, + void *table) +{ + const struct amd_pm_funcs *pp_funcs = adev->powerplay.pp_funcs; + int ret = 0; + + if (!pp_funcs->get_xcp_metrics) + return 0; + + mutex_lock(&adev->pm.mutex); + ret = pp_funcs->get_xcp_metrics(adev->powerplay.pp_handle, xcp_id, + table); + mutex_unlock(&adev->pm.mutex); + + return ret; +} diff --git a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h index 2c3c97587dd5..768317ee1486 100644 --- a/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h +++ b/drivers/gpu/drm/amd/pm/inc/amdgpu_dpm.h @@ -524,6 +524,8 @@ int amdgpu_dpm_get_power_profile_mode(struct amdgpu_device *adev, int amdgpu_dpm_set_power_profile_mode(struct amdgpu_device *adev, long *input, uint32_t size); int amdgpu_dpm_get_gpu_metrics(struct amdgpu_device *adev, void **table); +ssize_t amdgpu_dpm_get_xcp_metrics(struct amdgpu_device *adev, int xcp_id, + void *table); /** * @get_pm_metrics: Get one snapshot of power management metrics from PMFW. The @@ -561,6 +563,7 @@ int amdgpu_dpm_get_smu_prv_buf_details(struct amdgpu_device *adev, void **addr, size_t *size); int amdgpu_dpm_is_overdrive_supported(struct amdgpu_device *adev); +int amdgpu_dpm_is_overdrive_enabled(struct amdgpu_device *adev); int amdgpu_dpm_set_pp_table(struct amdgpu_device *adev, const char *buf, size_t size); diff --git a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c index 59fae668dc3f..34e71727b27d 100644 --- a/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c +++ b/drivers/gpu/drm/amd/pm/legacy-dpm/kv_dpm.c @@ -2594,7 +2594,7 @@ static int kv_parse_sys_info_table(struct amdgpu_device *adev) le32_to_cpu(igp_info->info_8.ulNbpStateNClkFreq[i]); } if (le32_to_cpu(igp_info->info_8.ulGPUCapInfo) & - SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS) + SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS) pi->caps_enable_dfs_bypass = true; sumo_construct_sclk_voltage_mapping_table(adev, diff --git a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu8_hwmgr.c b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu8_hwmgr.c index 9d3b33446adc..9b20076e26c0 100644 --- a/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu8_hwmgr.c +++ b/drivers/gpu/drm/amd/pm/powerplay/hwmgr/smu8_hwmgr.c @@ -394,7 +394,7 @@ static int smu8_get_system_info_data(struct pp_hwmgr *hwmgr) } if (le32_to_cpu(info->ulGPUCapInfo) & - SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS) { + SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS) { phm_cap_set(hwmgr->platform_descriptor.platformCaps, PHM_PlatformCaps_EnableDFSBypass); } diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index f24a1d8c77db..d79a1d94661a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -3758,6 +3758,19 @@ int smu_set_pm_policy(struct smu_context *smu, enum pp_pm_policy p_type, return ret; } +static ssize_t smu_sys_get_xcp_metrics(void *handle, int xcp_id, void *table) +{ + struct smu_context *smu = handle; + + if (!smu->pm_enabled || !smu->adev->pm.dpm_enabled) + return -EOPNOTSUPP; + + if (!smu->adev->xcp_mgr || !smu->ppt_funcs->get_xcp_metrics) + return -EOPNOTSUPP; + + return smu->ppt_funcs->get_xcp_metrics(smu, xcp_id, table); +} + static const struct amd_pm_funcs swsmu_pm_funcs = { /* export for sysfs */ .set_fan_control_mode = smu_set_fan_control_mode, @@ -3816,6 +3829,7 @@ static const struct amd_pm_funcs swsmu_pm_funcs = { .get_uclk_dpm_states = smu_get_uclk_dpm_states, .get_dpm_clock_table = smu_get_dpm_clock_table, .get_smu_prv_buf_details = smu_get_prv_buffer_details, + .get_xcp_metrics = smu_sys_get_xcp_metrics, }; int smu_wait_for_event(struct smu_context *smu, enum smu_event_type event, 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 d47e32ae4671..9aacc7bc1c69 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/amdgpu_smu.h @@ -1466,6 +1466,12 @@ struct pptable_funcs { */ int (*set_wbrf_exclusion_ranges)(struct smu_context *smu, struct freq_band_range *exclusion_ranges); + /** + * @get_xcp_metrics: Get a copy of the partition metrics table from SMU. + * Return: Size of table + */ + ssize_t (*get_xcp_metrics)(struct smu_context *smu, int xcp_id, + void *table); }; typedef enum { diff --git a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_pmfw.h b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_pmfw.h index 3d9e5e967c94..01790a927930 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_pmfw.h +++ b/drivers/gpu/drm/amd/pm/swsmu/inc/pmfw_if/smu_v13_0_6_pmfw.h @@ -127,7 +127,7 @@ typedef enum { VOLTAGE_GUARDBAND_COUNT } GFX_GUARDBAND_e; -#define SMU_METRICS_TABLE_VERSION 0x10 +#define SMU_METRICS_TABLE_VERSION 0x11 // Unified metrics table for smu_v13_0_6 typedef struct __attribute__((packed, aligned(4))) { @@ -463,6 +463,8 @@ typedef struct __attribute__((packed, aligned(4))) { typedef struct { // Telemetry uint32_t InputTelemetryVoltageInmV; + // General info + uint32_t pldmVersion[2]; } StaticMetricsTable_t; #pragma pack(pop) 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 533d58e57d05..e0d356f93ab0 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 @@ -322,7 +322,63 @@ int smu_v13_0_12_get_smu_metrics_data(struct smu_context *smu, return ret; } -ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table) +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 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); + 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 instance */ + inst = GET_INST(VCN, k); + for (j = 0; j < num_jpeg_rings; ++j) { + xcp_metrics->jpeg_busy[(idx * num_jpeg_rings) + j] = + SMUQ10_ROUND(metrics-> + JpegBusy[(inst * num_jpeg_rings) + j]); + } + xcp_metrics->vcn_busy[idx] = + SMUQ10_ROUND(metrics->VcnBusy[inst]); + xcp_metrics->current_vclk0[idx] = SMUQ10_ROUND( + metrics->VclkFrequency[inst]); + xcp_metrics->current_dclk0[idx] = SMUQ10_ROUND( + metrics->DclkFrequency[inst]); + xcp_metrics->current_socclk[idx] = SMUQ10_ROUND( + metrics->SocclkFrequency[inst]); + + idx++; + } + + xcp_metrics->current_uclk = + SMUQ10_ROUND(metrics->UclkFrequency); + + amdgpu_xcp_get_inst_details(xcp, AMDGPU_XCP_GFX, &inst_mask); + idx = 0; + for_each_inst(k, inst_mask) { + inst = GET_INST(GC, k); + xcp_metrics->current_gfxclk[idx] = SMUQ10_ROUND(metrics->GfxclkFrequency[inst]); + xcp_metrics->gfx_busy_inst[idx] = SMUQ10_ROUND(metrics->GfxBusy[inst]); + xcp_metrics->gfx_busy_acc[idx] = SMUQ10_ROUND(metrics->GfxBusyAcc[inst]); + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(HST_LIMIT_METRICS))) { + xcp_metrics->gfx_below_host_limit_ppt_acc[idx] = SMUQ10_ROUND(metrics->GfxclkBelowHostLimitPptAcc[inst]); + xcp_metrics->gfx_below_host_limit_thm_acc[idx] = SMUQ10_ROUND(metrics->GfxclkBelowHostLimitThmAcc[inst]); + xcp_metrics->gfx_low_utilization_acc[idx] = SMUQ10_ROUND(metrics->GfxclkLowUtilizationAcc[inst]); + xcp_metrics->gfx_below_host_limit_total_acc[idx] = SMUQ10_ROUND(metrics->GfxclkBelowHostLimitTotalAcc[inst]); + } + idx++; + } + + return sizeof(*xcp_metrics); +} + +ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table, void *smu_metrics) { struct smu_table_context *smu_table = &smu->smu_table; struct gpu_metrics_v1_8 *gpu_metrics = @@ -334,8 +390,7 @@ ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table) struct amdgpu_xcp *xcp; u32 inst_mask; - metrics = kzalloc(sizeof(MetricsTable_t), GFP_KERNEL); - memcpy(metrics, smu_table->metrics_table, sizeof(MetricsTable_t)); + metrics = (MetricsTable_t *)smu_metrics; smu_cmn_init_soft_gpu_metrics(gpu_metrics, 1, 8); @@ -416,13 +471,16 @@ ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table) gpu_metrics->mem_activity_acc = SMUQ10_ROUND(metrics->DramBandwidthUtilizationAcc); for (i = 0; i < NUM_XGMI_LINKS; i++) { - gpu_metrics->xgmi_read_data_acc[i] = + j = amdgpu_xgmi_get_ext_link(adev, i); + if (j < 0 || j >= NUM_XGMI_LINKS) + continue; + gpu_metrics->xgmi_read_data_acc[j] = SMUQ10_ROUND(metrics->XgmiReadDataSizeAcc[i]); - gpu_metrics->xgmi_write_data_acc[i] = + gpu_metrics->xgmi_write_data_acc[j] = SMUQ10_ROUND(metrics->XgmiWriteDataSizeAcc[i]); ret = amdgpu_get_xgmi_link_status(adev, i); if (ret >= 0) - gpu_metrics->xgmi_link_status[i] = ret; + gpu_metrics->xgmi_link_status[j] = ret; } gpu_metrics->num_partition = adev->xcp_mgr->num_xcps; @@ -474,7 +532,6 @@ ssize_t smu_v13_0_12_get_gpu_metrics(struct smu_context *smu, void **table) gpu_metrics->firmware_timestamp = metrics->Timestamp; *table = (void *)gpu_metrics; - kfree(metrics); return sizeof(*gpu_metrics); } 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 7d4ff09be7e8..f00ef7f3f355 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 @@ -312,6 +312,11 @@ static void smu_v13_0_14_init_caps(struct smu_context *smu) smu_v13_0_6_cap_set(smu, SMU_CAP(PER_INST_METRICS)); if (fw_ver >= 0x5551200) smu_v13_0_6_cap_set(smu, SMU_CAP(SDMA_RESET)); + if (fw_ver >= 0x5551600) { + smu_v13_0_6_cap_set(smu, SMU_CAP(STATIC_METRICS)); + smu_v13_0_6_cap_set(smu, SMU_CAP(BOARD_VOLTAGE)); + smu_v13_0_6_cap_set(smu, SMU_CAP(PLDM_VERSION)); + } } static void smu_v13_0_12_init_caps(struct smu_context *smu) @@ -392,10 +397,14 @@ static void smu_v13_0_6_init_caps(struct smu_context *smu) if ((pgm == 7 && fw_ver >= 0x7550E00) || (pgm == 0 && fw_ver >= 0x00557E00)) smu_v13_0_6_cap_set(smu, SMU_CAP(HST_LIMIT_METRICS)); - if (fw_ver >= 0x00557F01) { + if ((pgm == 0 && fw_ver >= 0x00557F01) || + (pgm == 7 && fw_ver >= 0x7551000)) { smu_v13_0_6_cap_set(smu, SMU_CAP(STATIC_METRICS)); smu_v13_0_6_cap_set(smu, SMU_CAP(BOARD_VOLTAGE)); } + if ((pgm == 0 && fw_ver >= 0x00558000) || + (pgm == 7 && fw_ver >= 0x7551000)) + smu_v13_0_6_cap_set(smu, SMU_CAP(PLDM_VERSION)); } if (((pgm == 7) && (fw_ver >= 0x7550700)) || ((pgm == 0) && (fw_ver >= 0x00557900)) || @@ -752,6 +761,11 @@ static void smu_v13_0_6_fill_static_metrics_table(struct smu_context *smu, } dpm_context->board_volt = static_metrics->InputTelemetryVoltageInmV; + + if (smu_v13_0_6_cap_supported(smu, SMU_CAP(PLDM_VERSION)) && + static_metrics->pldmVersion[0] != 0xFFFFFFFF) + smu->adev->firmware.pldm_version = + static_metrics->pldmVersion[0]; } int smu_v13_0_6_get_static_metrics_table(struct smu_context *smu) @@ -2529,6 +2543,126 @@ static int smu_v13_0_6_get_current_pcie_link_speed(struct smu_context *smu) return pcie_gen_to_speed(speed_level + 1); } +static ssize_t smu_v13_0_6_get_xcp_metrics(struct smu_context *smu, int xcp_id, + void *table) +{ + 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 amdgpu_device *adev = smu->adev; + int ret, inst, i, j, k, idx; + MetricsTableV0_t *metrics_v0; + MetricsTableV1_t *metrics_v1; + MetricsTableV2_t *metrics_v2; + struct amdgpu_xcp *xcp; + u32 inst_mask; + bool per_inst; + + if (!table) + return sizeof(*xcp_metrics); + + for_each_xcp(adev->xcp_mgr, xcp, i) { + if (xcp->id == xcp_id) + break; + } + 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); + + metrics_v0 = kzalloc(METRICS_TABLE_SIZE, GFP_KERNEL); + if (!metrics_v0) + return -ENOMEM; + + ret = smu_v13_0_6_get_metrics_table(smu, metrics_v0, false); + if (ret) { + kfree(metrics_v0); + 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))) { + ret = smu_v13_0_12_get_xcp_metrics(smu, xcp, table, metrics_v0); + goto out; + } + + metrics_v1 = (MetricsTableV1_t *)metrics_v0; + metrics_v2 = (MetricsTableV2_t *)metrics_v0; + + per_inst = smu_v13_0_6_cap_supported(smu, SMU_CAP(PER_INST_METRICS)); + + 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) { + xcp_metrics->jpeg_busy[(idx * num_jpeg_rings) + j] = + SMUQ10_ROUND(GET_METRIC_FIELD( + JpegBusy, + version)[(inst * num_jpeg_rings) + j]); + } + xcp_metrics->vcn_busy[idx] = + SMUQ10_ROUND(GET_METRIC_FIELD(VcnBusy, version)[inst]); + + xcp_metrics->current_vclk0[idx] = SMUQ10_ROUND( + GET_METRIC_FIELD(VclkFrequency, version)[inst]); + xcp_metrics->current_dclk0[idx] = SMUQ10_ROUND( + GET_METRIC_FIELD(DclkFrequency, version)[inst]); + xcp_metrics->current_socclk[idx] = SMUQ10_ROUND( + GET_METRIC_FIELD(SocclkFrequency, version)[inst]); + + idx++; + } + + xcp_metrics->current_uclk = + SMUQ10_ROUND(GET_METRIC_FIELD(UclkFrequency, version)); + + 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); + xcp_metrics->current_gfxclk[idx] = + SMUQ10_ROUND(GET_METRIC_FIELD(GfxclkFrequency, + version)[inst]); + + xcp_metrics->gfx_busy_inst[idx] = SMUQ10_ROUND( + GET_GPU_METRIC_FIELD(GfxBusy, version)[inst]); + xcp_metrics->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))) { + xcp_metrics->gfx_below_host_limit_ppt_acc + [idx] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitPptAcc + [inst]); + xcp_metrics->gfx_below_host_limit_thm_acc + [idx] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitThmAcc + [inst]); + xcp_metrics->gfx_low_utilization_acc + [idx] = SMUQ10_ROUND( + metrics_v0 + ->GfxclkLowUtilizationAcc[inst]); + xcp_metrics->gfx_below_host_limit_total_acc + [idx] = SMUQ10_ROUND( + metrics_v0->GfxclkBelowHostLimitTotalAcc + [inst]); + } + idx++; + } + } +out: + kfree(metrics_v0); + + return sizeof(*xcp_metrics); +} + 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; @@ -2542,6 +2676,7 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table MetricsTableV2_t *metrics_v2; struct amdgpu_xcp *xcp; u16 link_width_level; + ssize_t num_bytes; u8 num_jpeg_rings; u32 inst_mask; bool per_inst; @@ -2554,8 +2689,11 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table } 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); + smu_v13_0_6_cap_supported(smu, SMU_CAP(STATIC_METRICS))) { + num_bytes = smu_v13_0_12_get_gpu_metrics(smu, table, metrics_v0); + kfree(metrics_v0); + return num_bytes; + } metrics_v1 = (MetricsTableV1_t *)metrics_v0; metrics_v2 = (MetricsTableV2_t *)metrics_v0; @@ -2670,13 +2808,16 @@ static ssize_t smu_v13_0_6_get_gpu_metrics(struct smu_context *smu, void **table SMUQ10_ROUND(GET_METRIC_FIELD(DramBandwidthUtilizationAcc, version)); for (i = 0; i < NUM_XGMI_LINKS; i++) { - gpu_metrics->xgmi_read_data_acc[i] = - SMUQ10_ROUND(GET_METRIC_FIELD(XgmiReadDataSizeAcc, version)[i]); - gpu_metrics->xgmi_write_data_acc[i] = - SMUQ10_ROUND(GET_METRIC_FIELD(XgmiWriteDataSizeAcc, version)[i]); + j = amdgpu_xgmi_get_ext_link(adev, i); + if (j < 0 || j >= NUM_XGMI_LINKS) + continue; + gpu_metrics->xgmi_read_data_acc[j] = SMUQ10_ROUND( + GET_METRIC_FIELD(XgmiReadDataSizeAcc, version)[i]); + gpu_metrics->xgmi_write_data_acc[j] = SMUQ10_ROUND( + GET_METRIC_FIELD(XgmiWriteDataSizeAcc, version)[i]); ret = amdgpu_get_xgmi_link_status(adev, i); if (ret >= 0) - gpu_metrics->xgmi_link_status[i] = ret; + gpu_metrics->xgmi_link_status[j] = ret; } gpu_metrics->num_partition = adev->xcp_mgr->num_xcps; @@ -3673,6 +3814,7 @@ static const struct pptable_funcs smu_v13_0_6_ppt_funcs = { .get_pp_feature_mask = smu_cmn_get_pp_feature_mask, .get_gpu_metrics = smu_v13_0_6_get_gpu_metrics, .get_pm_metrics = smu_v13_0_6_get_pm_metrics, + .get_xcp_metrics = smu_v13_0_6_get_xcp_metrics, .get_thermal_temperature_range = smu_v13_0_6_get_thermal_temperature_range, .mode1_reset_is_support = smu_v13_0_6_is_mode1_reset_supported, .link_reset_is_support = smu_v13_0_6_is_link_reset_supported, 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 d151bcd0cca7..d38d6d76b1e7 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 @@ -67,6 +67,7 @@ enum smu_v13_0_6_caps { SMU_CAP(STATIC_METRICS), SMU_CAP(HST_LIMIT_METRICS), SMU_CAP(BOARD_VOLTAGE), + SMU_CAP(PLDM_VERSION), SMU_CAP(ALL), }; @@ -79,7 +80,10 @@ int smu_v13_0_12_get_max_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); +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); extern const struct cmn2asic_mapping smu_v13_0_12_feature_mask_map[]; extern const struct cmn2asic_msg_mapping smu_v13_0_12_message_map[]; #endif diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c index 80eb1a03b3ca..7eaf58fd7f9a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.c @@ -1051,73 +1051,6 @@ int smu_cmn_get_combo_pptable(struct smu_context *smu) false); } -void smu_cmn_init_soft_gpu_metrics(void *table, uint8_t frev, uint8_t crev) -{ - struct metrics_table_header *header = (struct metrics_table_header *)table; - uint16_t structure_size; - -#define METRICS_VERSION(a, b) ((a << 16) | b) - - switch (METRICS_VERSION(frev, crev)) { - case METRICS_VERSION(1, 0): - structure_size = sizeof(struct gpu_metrics_v1_0); - break; - case METRICS_VERSION(1, 1): - structure_size = sizeof(struct gpu_metrics_v1_1); - break; - case METRICS_VERSION(1, 2): - structure_size = sizeof(struct gpu_metrics_v1_2); - break; - case METRICS_VERSION(1, 3): - structure_size = sizeof(struct gpu_metrics_v1_3); - break; - case METRICS_VERSION(1, 4): - structure_size = sizeof(struct gpu_metrics_v1_4); - break; - case METRICS_VERSION(1, 5): - structure_size = sizeof(struct gpu_metrics_v1_5); - break; - case METRICS_VERSION(1, 6): - structure_size = sizeof(struct gpu_metrics_v1_6); - break; - case METRICS_VERSION(1, 7): - structure_size = sizeof(struct gpu_metrics_v1_7); - break; - case METRICS_VERSION(1, 8): - structure_size = sizeof(struct gpu_metrics_v1_8); - break; - case METRICS_VERSION(2, 0): - structure_size = sizeof(struct gpu_metrics_v2_0); - break; - case METRICS_VERSION(2, 1): - structure_size = sizeof(struct gpu_metrics_v2_1); - break; - case METRICS_VERSION(2, 2): - structure_size = sizeof(struct gpu_metrics_v2_2); - break; - case METRICS_VERSION(2, 3): - structure_size = sizeof(struct gpu_metrics_v2_3); - break; - case METRICS_VERSION(2, 4): - structure_size = sizeof(struct gpu_metrics_v2_4); - break; - case METRICS_VERSION(3, 0): - structure_size = sizeof(struct gpu_metrics_v3_0); - break; - default: - return; - } - -#undef METRICS_VERSION - - memset(header, 0xFF, structure_size); - - header->format_revision = frev; - header->content_revision = crev; - header->structure_size = structure_size; - -} - int smu_cmn_set_mp1_state(struct smu_context *smu, enum pp_mp1_state mp1_state) { diff --git a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h index a020277dec3e..7473672abd2a 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h +++ b/drivers/gpu/drm/amd/pm/swsmu/smu_cmn.h @@ -40,6 +40,30 @@ #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_ABNORMAL 0x8 #define SMU_IH_INTERRUPT_CONTEXT_ID_FAN_RECOVERY 0x9 +#define smu_cmn_init_soft_gpu_metrics(ptr, frev, crev) \ + do { \ + typecheck(struct gpu_metrics_v##frev##_##crev, \ + typeof(*(ptr))); \ + struct metrics_table_header *header = \ + (struct metrics_table_header *)(ptr); \ + memset(header, 0xFF, sizeof(*(ptr))); \ + header->format_revision = frev; \ + header->content_revision = crev; \ + header->structure_size = sizeof(*(ptr)); \ + } while (0) + +#define smu_cmn_init_partition_metrics(ptr, frev, crev) \ + do { \ + typecheck(struct amdgpu_partition_metrics_v##frev##_##crev, \ + typeof(*(ptr))); \ + struct metrics_table_header *header = \ + (struct metrics_table_header *)(ptr); \ + memset(header, 0xFF, sizeof(*(ptr))); \ + header->format_revision = frev; \ + header->content_revision = crev; \ + header->structure_size = sizeof(*(ptr)); \ + } while (0) + extern const int link_speed[]; /* Helper to Convert from PCIE Gen 1/2/3/4/5/6 to 0.1 GT/s speed units */ @@ -125,8 +149,6 @@ int smu_cmn_get_metrics_table(struct smu_context *smu, int smu_cmn_get_combo_pptable(struct smu_context *smu); -void smu_cmn_init_soft_gpu_metrics(void *table, uint8_t frev, uint8_t crev); - int smu_cmn_set_mp1_state(struct smu_context *smu, enum pp_mp1_state mp1_state); diff --git a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c index a761941bc3c2..505eec6b819b 100644 --- a/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c +++ b/drivers/gpu/drm/bridge/analogix/analogix_dp_core.c @@ -1531,10 +1531,8 @@ analogix_dp_probe(struct device *dev, struct analogix_dp_plat_data *plat_data) } dp->reg_base = devm_platform_ioremap_resource(pdev, 0); - if (IS_ERR(dp->reg_base)) { - ret = PTR_ERR(dp->reg_base); - goto err_disable_clk; - } + if (IS_ERR(dp->reg_base)) + return ERR_CAST(dp->reg_base); dp->force_hpd = of_property_read_bool(dev->of_node, "force-hpd"); @@ -1546,8 +1544,7 @@ analogix_dp_probe(struct device *dev, struct analogix_dp_plat_data *plat_data) if (IS_ERR(dp->hpd_gpiod)) { dev_err(dev, "error getting HDP GPIO: %ld\n", PTR_ERR(dp->hpd_gpiod)); - ret = PTR_ERR(dp->hpd_gpiod); - goto err_disable_clk; + return ERR_CAST(dp->hpd_gpiod); } if (dp->hpd_gpiod) { @@ -1567,8 +1564,7 @@ analogix_dp_probe(struct device *dev, struct analogix_dp_plat_data *plat_data) if (dp->irq == -ENXIO) { dev_err(&pdev->dev, "failed to get irq\n"); - ret = -ENODEV; - goto err_disable_clk; + return ERR_PTR(-ENODEV); } ret = devm_request_threaded_irq(&pdev->dev, dp->irq, @@ -1577,7 +1573,7 @@ analogix_dp_probe(struct device *dev, struct analogix_dp_plat_data *plat_data) irq_flags, "analogix-dp", dp); if (ret) { dev_err(&pdev->dev, "failed to request irq\n"); - goto err_disable_clk; + return ERR_PTR(ret); } dp->aux.name = "DP-AUX"; @@ -1590,13 +1586,9 @@ analogix_dp_probe(struct device *dev, struct analogix_dp_plat_data *plat_data) pm_runtime_set_autosuspend_delay(dp->dev, 100); ret = devm_pm_runtime_enable(dp->dev); if (ret) - goto err_disable_clk; + return ERR_PTR(ret); return dp; - -err_disable_clk: - clk_disable_unprepare(dp->clock); - return ERR_PTR(ret); } EXPORT_SYMBOL_GPL(analogix_dp_probe); diff --git a/drivers/gpu/drm/display/drm_hdmi_audio_helper.c b/drivers/gpu/drm/display/drm_hdmi_audio_helper.c index 05afc9f0bdd6..ae8a0cf595fc 100644 --- a/drivers/gpu/drm/display/drm_hdmi_audio_helper.c +++ b/drivers/gpu/drm/display/drm_hdmi_audio_helper.c @@ -103,7 +103,8 @@ static int drm_connector_hdmi_audio_hook_plugged_cb(struct device *dev, connector->hdmi_audio.plugged_cb = fn; connector->hdmi_audio.plugged_cb_dev = codec_dev; - fn(codec_dev, connector->hdmi_audio.last_state); + if (fn) + fn(codec_dev, connector->hdmi_audio.last_state); mutex_unlock(&connector->hdmi_audio.lock); diff --git a/drivers/gpu/drm/i915/display/intel_audio.c b/drivers/gpu/drm/i915/display/intel_audio.c index 40d8bbd8107d..55af3a553c58 100644 --- a/drivers/gpu/drm/i915/display/intel_audio.c +++ b/drivers/gpu/drm/i915/display/intel_audio.c @@ -397,6 +397,19 @@ hsw_audio_config_update(struct intel_encoder *encoder, hsw_hdmi_audio_config_update(encoder, crtc_state); } +static void intel_audio_sdp_split_update(const struct intel_crtc_state *crtc_state, + bool enable) +{ + struct intel_display *display = to_intel_display(crtc_state); + enum transcoder trans = crtc_state->cpu_transcoder; + + if (!HAS_DP20(display)) + return; + + intel_de_rmw(display, AUD_DP_2DOT0_CTRL(trans), AUD_ENABLE_SDP_SPLIT, + enable && crtc_state->sdp_split_enable ? AUD_ENABLE_SDP_SPLIT : 0); +} + static void hsw_audio_codec_disable(struct intel_encoder *encoder, const struct intel_crtc_state *old_crtc_state, const struct drm_connector_state *old_conn_state) @@ -430,6 +443,8 @@ static void hsw_audio_codec_disable(struct intel_encoder *encoder, if (needs_wa_14020863754(display)) intel_de_rmw(display, AUD_CHICKENBIT_REG3, DACBE_DISABLE_MIN_HBLANK_FIX, 0); + intel_audio_sdp_split_update(old_crtc_state, false); + mutex_unlock(&display->audio.mutex); } @@ -555,6 +570,8 @@ static void hsw_audio_codec_enable(struct intel_encoder *encoder, if (intel_crtc_has_type(crtc_state, INTEL_OUTPUT_DP)) enable_audio_dsc_wa(encoder, crtc_state); + intel_audio_sdp_split_update(crtc_state, true); + if (needs_wa_14020863754(display)) intel_de_rmw(display, AUD_CHICKENBIT_REG3, 0, DACBE_DISABLE_MIN_HBLANK_FIX); @@ -681,16 +698,6 @@ static void ibx_audio_codec_enable(struct intel_encoder *encoder, mutex_unlock(&display->audio.mutex); } -void intel_audio_sdp_split_update(const struct intel_crtc_state *crtc_state) -{ - struct intel_display *display = to_intel_display(crtc_state); - enum transcoder trans = crtc_state->cpu_transcoder; - - if (HAS_DP20(display)) - intel_de_rmw(display, AUD_DP_2DOT0_CTRL(trans), AUD_ENABLE_SDP_SPLIT, - crtc_state->sdp_split_enable ? AUD_ENABLE_SDP_SPLIT : 0); -} - bool intel_audio_compute_config(struct intel_encoder *encoder, struct intel_crtc_state *crtc_state, struct drm_connector_state *conn_state) diff --git a/drivers/gpu/drm/i915/display/intel_audio.h b/drivers/gpu/drm/i915/display/intel_audio.h index ad49eefa7182..42cf886f3d24 100644 --- a/drivers/gpu/drm/i915/display/intel_audio.h +++ b/drivers/gpu/drm/i915/display/intel_audio.h @@ -31,6 +31,5 @@ int intel_audio_min_cdclk(const struct intel_crtc_state *crtc_state); void intel_audio_init(struct intel_display *display); void intel_audio_register(struct intel_display *display); void intel_audio_deinit(struct intel_display *display); -void intel_audio_sdp_split_update(const struct intel_crtc_state *crtc_state); #endif /* __INTEL_AUDIO_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_ddi.c b/drivers/gpu/drm/i915/display/intel_ddi.c index 74132c1d6385..d58f8fc37326 100644 --- a/drivers/gpu/drm/i915/display/intel_ddi.c +++ b/drivers/gpu/drm/i915/display/intel_ddi.c @@ -3507,9 +3507,6 @@ static void intel_ddi_enable(struct intel_atomic_state *state, intel_vrr_transcoder_enable(crtc_state); - /* Enable/Disable DP2.0 SDP split config before transcoder */ - intel_audio_sdp_split_update(crtc_state); - /* 128b/132b SST */ if (!is_hdmi && intel_dp_is_uhbr(crtc_state)) { struct intel_dp *intel_dp = enc_to_intel_dp(encoder); diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index 06f4ad8de591..e1501b13f08f 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -1328,8 +1328,6 @@ static void mst_stream_enable(struct intel_atomic_state *state, FECSTALL_DIS_DPTSTREAM_DPTTG, pipe_config->fec_enable ? FECSTALL_DIS_DPTSTREAM_DPTTG : 0); - intel_audio_sdp_split_update(pipe_config); - intel_enable_transcoder(pipe_config); for_each_pipe_crtc_modeset_enable(display, pipe_crtc, pipe_config, i) { diff --git a/drivers/gpu/drm/i915/display/intel_psr_regs.h b/drivers/gpu/drm/i915/display/intel_psr_regs.h index 795e6b9cc575..248136456048 100644 --- a/drivers/gpu/drm/i915/display/intel_psr_regs.h +++ b/drivers/gpu/drm/i915/display/intel_psr_regs.h @@ -325,8 +325,8 @@ #define PORT_ALPM_LFPS_CTL_LFPS_HALF_CYCLE_DURATION_MASK REG_GENMASK(20, 16) #define PORT_ALPM_LFPS_CTL_LFPS_HALF_CYCLE_DURATION(val) REG_FIELD_PREP(PORT_ALPM_LFPS_CTL_LFPS_HALF_CYCLE_DURATION_MASK, val) #define PORT_ALPM_LFPS_CTL_FIRST_LFPS_HALF_CYCLE_DURATION_MASK REG_GENMASK(12, 8) -#define PORT_ALPM_LFPS_CTL_FIRST_LFPS_HALF_CYCLE_DURATION(val) REG_FIELD_PREP(PORT_ALPM_LFPS_CTL_LFPS_HALF_CYCLE_DURATION_MASK, val) +#define PORT_ALPM_LFPS_CTL_FIRST_LFPS_HALF_CYCLE_DURATION(val) REG_FIELD_PREP(PORT_ALPM_LFPS_CTL_FIRST_LFPS_HALF_CYCLE_DURATION_MASK, val) #define PORT_ALPM_LFPS_CTL_LAST_LFPS_HALF_CYCLE_DURATION_MASK REG_GENMASK(4, 0) -#define PORT_ALPM_LFPS_CTL_LAST_LFPS_HALF_CYCLE_DURATION(val) REG_FIELD_PREP(PORT_ALPM_LFPS_CTL_LFPS_HALF_CYCLE_DURATION_MASK, val) +#define PORT_ALPM_LFPS_CTL_LAST_LFPS_HALF_CYCLE_DURATION(val) REG_FIELD_PREP(PORT_ALPM_LFPS_CTL_LAST_LFPS_HALF_CYCLE_DURATION_MASK, val) #endif /* __INTEL_PSR_REGS_H__ */ diff --git a/drivers/gpu/drm/i915/display/intel_snps_hdmi_pll.c b/drivers/gpu/drm/i915/display/intel_snps_hdmi_pll.c index c6321dafef4f..74bb3bedf30f 100644 --- a/drivers/gpu/drm/i915/display/intel_snps_hdmi_pll.c +++ b/drivers/gpu/drm/i915/display/intel_snps_hdmi_pll.c @@ -41,12 +41,12 @@ static s64 interp(s64 x, s64 x1, s64 x2, s64 y1, s64 y2) { s64 dydx; - dydx = DIV_ROUND_UP_ULL((y2 - y1) * 100000, (x2 - x1)); + dydx = DIV64_U64_ROUND_UP((y2 - y1) * 100000, (x2 - x1)); - return (y1 + DIV_ROUND_UP_ULL(dydx * (x - x1), 100000)); + return (y1 + DIV64_U64_ROUND_UP(dydx * (x - x1), 100000)); } -static void get_ana_cp_int_prop(u32 vco_clk, +static void get_ana_cp_int_prop(u64 vco_clk, u32 refclk_postscalar, int mpll_ana_v2i, int c, int a, @@ -115,16 +115,16 @@ static void get_ana_cp_int_prop(u32 vco_clk, CURVE0_MULTIPLIER)); scaled_interpolated_sqrt = - int_sqrt(DIV_ROUND_UP_ULL(interpolated_product, vco_div_refclk_float) * + int_sqrt(DIV64_U64_ROUND_UP(interpolated_product, vco_div_refclk_float) * DIV_ROUND_DOWN_ULL(1000000000000ULL, 55)); /* Scale vco_div_refclk for ana_cp_int */ scaled_vco_div_refclk2 = DIV_ROUND_UP_ULL(vco_div_refclk_float, 1000000); - adjusted_vco_clk2 = 1460281 * DIV_ROUND_UP_ULL(scaled_interpolated_sqrt * + adjusted_vco_clk2 = 1460281 * DIV64_U64_ROUND_UP(scaled_interpolated_sqrt * scaled_vco_div_refclk2, curve_1_interpolated); - *ana_cp_prop = DIV_ROUND_UP_ULL(adjusted_vco_clk2, curve_2_scaled2); + *ana_cp_prop = DIV64_U64_ROUND_UP(adjusted_vco_clk2, curve_2_scaled2); *ana_cp_prop = max(1, min(*ana_cp_prop, 127)); } @@ -165,10 +165,10 @@ static void compute_hdmi_tmds_pll(u64 pixel_clock, u32 refclk, /* Select appropriate v2i point */ if (datarate <= INTEL_SNPS_PHY_HDMI_9999MHZ) { mpll_ana_v2i = 2; - tx_clk_div = ilog2(DIV_ROUND_DOWN_ULL(INTEL_SNPS_PHY_HDMI_9999MHZ, datarate)); + tx_clk_div = ilog2(div64_u64(INTEL_SNPS_PHY_HDMI_9999MHZ, datarate)); } else { mpll_ana_v2i = 3; - tx_clk_div = ilog2(DIV_ROUND_DOWN_ULL(INTEL_SNPS_PHY_HDMI_16GHZ, datarate)); + tx_clk_div = ilog2(div64_u64(INTEL_SNPS_PHY_HDMI_16GHZ, datarate)); } vco_clk = (datarate << tx_clk_div) >> 1; diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c index f8cb7c630d5b..127316d2c8aa 100644 --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c @@ -633,7 +633,7 @@ static int guc_submission_send_busy_loop(struct intel_guc *guc, atomic_inc(&guc->outstanding_submission_g2h); ret = intel_guc_send_busy_loop(guc, action, len, g2h_len_dw, loop); - if (ret) + if (ret && g2h_len_dw) atomic_dec(&guc->outstanding_submission_g2h); return ret; @@ -3443,18 +3443,29 @@ static inline int guc_lrc_desc_unpin(struct intel_context *ce) * GuC is active, lets destroy this context, but at this point we can still be racing * with suspend, so we undo everything if the H2G fails in deregister_context so * that GuC reset will find this context during clean up. + * + * There is a race condition where the reset code could have altered + * this context's state and done a wakeref put before we try to + * deregister it here. So check if the context is still set to be + * destroyed before undoing earlier changes, to avoid two wakeref puts + * on the same context. */ ret = deregister_context(ce, ce->guc_id.id); if (ret) { + bool pending_destroyed; spin_lock_irqsave(&ce->guc_state.lock, flags); - set_context_registered(ce); - clr_context_destroyed(ce); + pending_destroyed = context_destroyed(ce); + if (pending_destroyed) { + set_context_registered(ce); + clr_context_destroyed(ce); + } spin_unlock_irqrestore(&ce->guc_state.lock, flags); /* * As gt-pm is awake at function entry, intel_wakeref_put_async merely decrements * the wakeref immediately but per function spec usage call this after unlock. */ - intel_wakeref_put_async(>->wakeref); + if (pending_destroyed) + intel_wakeref_put_async(>->wakeref); } return ret; diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/r535.c index 9446049642e1..d294844d9eae 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/r535.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/vfn/r535.c @@ -42,7 +42,7 @@ r535_vfn_new(const struct nvkm_vfn_func *hw, return -ENOMEM; rm->dtor = r535_vfn_dtor; - rm->intr = &tu102_vfn_intr, + rm->intr = &tu102_vfn_intr; rm->user.addr = 0x030000; rm->user.size = 0x010000; rm->user.base.minver = -1; diff --git a/drivers/gpu/drm/panel/Kconfig b/drivers/gpu/drm/panel/Kconfig index 721581d425b4..cfebb08e8a62 100644 --- a/drivers/gpu/drm/panel/Kconfig +++ b/drivers/gpu/drm/panel/Kconfig @@ -522,6 +522,8 @@ config DRM_PANEL_NOVATEK_NT37801 depends on OF depends on DRM_MIPI_DSI depends on BACKLIGHT_CLASS_DEVICE + select DRM_DISPLAY_DSC_HELPER + select DRM_DISPLAY_HELPER help Say Y here if you want to enable support for Novatek NT37801 (or NT37810) AMOLED DSI Video Mode LCD panel module with 1440x3200 diff --git a/drivers/gpu/drm/panel/panel-novatek-nt37801.c b/drivers/gpu/drm/panel/panel-novatek-nt37801.c index 84d367eab058..d6a37d7e0cc6 100644 --- a/drivers/gpu/drm/panel/panel-novatek-nt37801.c +++ b/drivers/gpu/drm/panel/panel-novatek-nt37801.c @@ -257,8 +257,8 @@ static int novatek_nt37801_probe(struct mipi_dsi_device *dsi) ctx = devm_drm_panel_alloc(dev, struct novatek_nt37801, panel, &novatek_nt37801_panel_funcs, DRM_MODE_CONNECTOR_DSI); - if (!ctx) - return -ENOMEM; + if (IS_ERR(ctx)) + return PTR_ERR(ctx); ret = devm_regulator_bulk_get_const(dev, ARRAY_SIZE(novatek_nt37801_supplies), diff --git a/drivers/gpu/drm/panel/panel-simple.c b/drivers/gpu/drm/panel/panel-simple.c index 82ee2f12b8d2..0a3b26bb4d73 100644 --- a/drivers/gpu/drm/panel/panel-simple.c +++ b/drivers/gpu/drm/panel/panel-simple.c @@ -2198,13 +2198,14 @@ static const struct display_timing evervision_vgg644804_timing = { static const struct panel_desc evervision_vgg644804 = { .timings = &evervision_vgg644804_timing, .num_timings = 1, - .bpc = 8, + .bpc = 6, .size = { .width = 115, .height = 86, }, .bus_format = MEDIA_BUS_FMT_RGB666_1X7X3_SPWG, - .bus_flags = DRM_BUS_FLAG_DE_HIGH | DRM_BUS_FLAG_PIXDATA_SAMPLE_NEGEDGE, + .bus_flags = DRM_BUS_FLAG_DE_HIGH, + .connector_type = DRM_MODE_CONNECTOR_LVDS, }; static const struct display_timing evervision_vgg804821_timing = { diff --git a/drivers/gpu/drm/radeon/atombios.h b/drivers/gpu/drm/radeon/atombios.h index a7caac5b8ac8..1afa70566985 100644 --- a/drivers/gpu/drm/radeon/atombios.h +++ b/drivers/gpu/drm/radeon/atombios.h @@ -5071,7 +5071,7 @@ typedef struct _ATOM_INTEGRATED_SYSTEM_INFO_V1_7 #define SYS_INFO_GPUCAPS__TMDSHDMI_COHERENT_SINGLEPLL_MODE 0x01 #define SYS_INFO_GPUCAPS__DP_SINGLEPLL_MODE 0x02 #define SYS_INFO_GPUCAPS__DISABLE_AUX_MODE_DETECT 0x08 -#define SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS 0x10 +#define SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS 0x10 /********************************************************************************************************************** ATOM_INTEGRATED_SYSTEM_INFO_V1_7 Description diff --git a/drivers/gpu/drm/radeon/kv_dpm.c b/drivers/gpu/drm/radeon/kv_dpm.c index 55dbf450bd9c..4aa050385284 100644 --- a/drivers/gpu/drm/radeon/kv_dpm.c +++ b/drivers/gpu/drm/radeon/kv_dpm.c @@ -2329,7 +2329,7 @@ static int kv_parse_sys_info_table(struct radeon_device *rdev) le32_to_cpu(igp_info->info_8.ulNbpStateNClkFreq[i]); } if (le32_to_cpu(igp_info->info_8.ulGPUCapInfo) & - SYS_INFO_GPUCAPS__ENABEL_DFS_BYPASS) + SYS_INFO_GPUCAPS__ENABLE_DFS_BYPASS) pi->caps_enable_dfs_bypass = true; sumo_construct_sclk_voltage_mapping_table(rdev, diff --git a/drivers/gpu/drm/scheduler/sched_entity.c b/drivers/gpu/drm/scheduler/sched_entity.c index bd39db7bb240..e671aa241720 100644 --- a/drivers/gpu/drm/scheduler/sched_entity.c +++ b/drivers/gpu/drm/scheduler/sched_entity.c @@ -176,6 +176,7 @@ static void drm_sched_entity_kill_jobs_work(struct work_struct *wrk) { struct drm_sched_job *job = container_of(wrk, typeof(*job), work); + drm_sched_fence_scheduled(job->s_fence, NULL); drm_sched_fence_finished(job->s_fence, -ESRCH); WARN_ON(job->s_fence->parent); job->sched->ops->free_job(job); diff --git a/drivers/gpu/drm/xe/Kconfig b/drivers/gpu/drm/xe/Kconfig index 2169bc969ea1..fcc2677a4229 100644 --- a/drivers/gpu/drm/xe/Kconfig +++ b/drivers/gpu/drm/xe/Kconfig @@ -2,6 +2,8 @@ config DRM_XE tristate "Intel Xe Graphics" depends on DRM && PCI && (m || (y && KUNIT=y)) + depends on INTEL_VSEC || !INTEL_VSEC + depends on X86_PLATFORM_DEVICES || !(X86 && ACPI) select INTERVAL_TREE # we need shmfs for the swappable backing store, and in particular # the shmem_readpage() which depends upon tmpfs @@ -27,7 +29,6 @@ config DRM_XE select BACKLIGHT_CLASS_DEVICE if ACPI select INPUT if ACPI select ACPI_VIDEO if X86 && ACPI - select X86_PLATFORM_DEVICES if X86 && ACPI select ACPI_WMI if X86 && ACPI select SYNC_FILE select IOSF_MBI diff --git a/drivers/gpu/drm/xe/regs/xe_mchbar_regs.h b/drivers/gpu/drm/xe/regs/xe_mchbar_regs.h index f5e5234857c1..5394a1373a6b 100644 --- a/drivers/gpu/drm/xe/regs/xe_mchbar_regs.h +++ b/drivers/gpu/drm/xe/regs/xe_mchbar_regs.h @@ -38,10 +38,10 @@ #define TEMP_MASK REG_GENMASK(7, 0) #define PCU_CR_PACKAGE_RAPL_LIMIT XE_REG(MCHBAR_MIRROR_BASE_SNB + 0x59a0) -#define PKG_PWR_LIM_1 REG_GENMASK(14, 0) -#define PKG_PWR_LIM_1_EN REG_BIT(15) -#define PKG_PWR_LIM_1_TIME REG_GENMASK(23, 17) -#define PKG_PWR_LIM_1_TIME_X REG_GENMASK(23, 22) -#define PKG_PWR_LIM_1_TIME_Y REG_GENMASK(21, 17) +#define PWR_LIM_VAL REG_GENMASK(14, 0) +#define PWR_LIM_EN REG_BIT(15) +#define PWR_LIM_TIME REG_GENMASK(23, 17) +#define PWR_LIM_TIME_X REG_GENMASK(23, 22) +#define PWR_LIM_TIME_Y REG_GENMASK(21, 17) #endif /* _XE_MCHBAR_REGS_H_ */ diff --git a/drivers/gpu/drm/xe/regs/xe_pcode_regs.h b/drivers/gpu/drm/xe/regs/xe_pcode_regs.h index c7d5d782e3f9..c556a04670ee 100644 --- a/drivers/gpu/drm/xe/regs/xe_pcode_regs.h +++ b/drivers/gpu/drm/xe/regs/xe_pcode_regs.h @@ -18,16 +18,12 @@ #define PVC_GT0_PLATFORM_ENERGY_STATUS XE_REG(0x28106c) #define PVC_GT0_PACKAGE_POWER_SKU XE_REG(0x281080) -#define BMG_PACKAGE_POWER_SKU XE_REG(0x138098) -#define BMG_PACKAGE_POWER_SKU_UNIT XE_REG(0x1380dc) #define BMG_PACKAGE_ENERGY_STATUS XE_REG(0x138120) #define BMG_FAN_1_SPEED XE_REG(0x138140) #define BMG_FAN_2_SPEED XE_REG(0x138170) #define BMG_FAN_3_SPEED XE_REG(0x1381a0) #define BMG_VRAM_TEMPERATURE XE_REG(0x1382c0) #define BMG_PACKAGE_TEMPERATURE XE_REG(0x138434) -#define BMG_PACKAGE_RAPL_LIMIT XE_REG(0x138440) #define BMG_PLATFORM_ENERGY_STATUS XE_REG(0x138458) -#define BMG_PLATFORM_POWER_LIMIT XE_REG(0x138460) #endif /* _XE_PCODE_REGS_H_ */ diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index d99d91fe8aa9..7aa2c17825da 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -841,21 +841,6 @@ static int xe_bo_move(struct ttm_buffer_object *ttm_bo, bool evict, goto out; } - /* Reject BO eviction if BO is bound to current VM. */ - if (evict && ctx->resv) { - struct drm_gpuvm_bo *vm_bo; - - drm_gem_for_each_gpuvm_bo(vm_bo, &bo->ttm.base) { - struct xe_vm *vm = gpuvm_to_vm(vm_bo->vm); - - if (xe_vm_resv(vm) == ctx->resv && - xe_vm_in_preempt_fence_mode(vm)) { - ret = -EBUSY; - goto out; - } - } - } - /* * Failed multi-hop where the old_mem is still marked as * TTM_PL_FLAG_TEMPORARY, should just be a dummy move. @@ -1013,6 +998,25 @@ static long xe_bo_shrink_purge(struct ttm_operation_ctx *ctx, return lret; } +static bool +xe_bo_eviction_valuable(struct ttm_buffer_object *bo, const struct ttm_place *place) +{ + struct drm_gpuvm_bo *vm_bo; + + if (!ttm_bo_eviction_valuable(bo, place)) + return false; + + if (!xe_bo_is_xe_bo(bo)) + return true; + + drm_gem_for_each_gpuvm_bo(vm_bo, &bo->base) { + if (xe_vm_is_validating(gpuvm_to_vm(vm_bo->vm))) + return false; + } + + return true; +} + /** * xe_bo_shrink() - Try to shrink an xe bo. * @ctx: The struct ttm_operation_ctx used for shrinking. @@ -1047,7 +1051,7 @@ long xe_bo_shrink(struct ttm_operation_ctx *ctx, struct ttm_buffer_object *bo, (flags.purge && !xe_tt->purgeable)) return -EBUSY; - if (!ttm_bo_eviction_valuable(bo, &place)) + if (!xe_bo_eviction_valuable(bo, &place)) return -EBUSY; if (!xe_bo_is_xe_bo(bo) || !xe_bo_get_unless_zero(xe_bo)) @@ -1588,7 +1592,7 @@ const struct ttm_device_funcs xe_ttm_funcs = { .io_mem_pfn = xe_ttm_io_mem_pfn, .access_memory = xe_ttm_access_memory, .release_notify = xe_ttm_bo_release_notify, - .eviction_valuable = ttm_bo_eviction_valuable, + .eviction_valuable = xe_bo_eviction_valuable, .delete_mem_notify = xe_ttm_bo_delete_mem_notify, .swap_notify = xe_ttm_bo_swap_notify, }; @@ -2431,6 +2435,8 @@ int xe_bo_validate(struct xe_bo *bo, struct xe_vm *vm, bool allow_res_evict) .no_wait_gpu = false, .gfp_retry_mayfail = true, }; + struct pin_cookie cookie; + int ret; if (vm) { lockdep_assert_held(&vm->lock); @@ -2440,8 +2446,12 @@ int xe_bo_validate(struct xe_bo *bo, struct xe_vm *vm, bool allow_res_evict) ctx.resv = xe_vm_resv(vm); } + cookie = xe_vm_set_validating(vm, allow_res_evict); trace_xe_bo_validate(bo); - return ttm_bo_validate(&bo->ttm, &bo->placement, &ctx); + ret = ttm_bo_validate(&bo->ttm, &bo->placement, &ctx); + xe_vm_clear_validating(vm, allow_res_evict, cookie); + + return ret; } bool xe_bo_is_xe_bo(struct ttm_buffer_object *bo) @@ -2557,7 +2567,7 @@ typedef int (*xe_gem_create_set_property_fn)(struct xe_device *xe, u64 value); static const xe_gem_create_set_property_fn gem_create_set_property_funcs[] = { - [DRM_XE_GEM_CREATE_EXTENSION_SET_PROPERTY] = gem_create_set_pxp_type, + [DRM_XE_GEM_CREATE_SET_PROPERTY_PXP_TYPE] = gem_create_set_pxp_type, }; static int gem_create_user_ext_set_property(struct xe_device *xe, diff --git a/drivers/gpu/drm/xe/xe_device_sysfs.c b/drivers/gpu/drm/xe/xe_device_sysfs.c index 2e657692e5b5..b9440f8c781e 100644 --- a/drivers/gpu/drm/xe/xe_device_sysfs.c +++ b/drivers/gpu/drm/xe/xe_device_sysfs.c @@ -115,7 +115,7 @@ auto_link_downgrade_capable_show(struct device *dev, struct device_attribute *at xe_pm_runtime_put(xe); cap = REG_FIELD_GET(LINK_DOWNGRADE, val); - return sysfs_emit(buf, "%u\n", cap == DOWNGRADE_CAPABLE ? true : false); + return sysfs_emit(buf, "%u\n", cap == DOWNGRADE_CAPABLE); } static DEVICE_ATTR_ADMIN_RO(auto_link_downgrade_capable); diff --git a/drivers/gpu/drm/xe/xe_device_types.h b/drivers/gpu/drm/xe/xe_device_types.h index c8fa2c011666..6383a1c0d478 100644 --- a/drivers/gpu/drm/xe/xe_device_types.h +++ b/drivers/gpu/drm/xe/xe_device_types.h @@ -325,6 +325,10 @@ struct xe_device { u8 has_heci_gscfi:1; /** @info.has_llc: Device has a shared CPU+GPU last level cache */ u8 has_llc:1; + /** @info.has_mbx_power_limits: Device has support to manage power limits using + * pcode mailbox commands. + */ + u8 has_mbx_power_limits:1; /** @info.has_pxp: Device has PXP support */ u8 has_pxp:1; /** @info.has_range_tlb_invalidation: Has range based TLB invalidations */ diff --git a/drivers/gpu/drm/xe/xe_exec_queue.c b/drivers/gpu/drm/xe/xe_exec_queue.c index ce78cee5dec6..fee22358cc09 100644 --- a/drivers/gpu/drm/xe/xe_exec_queue.c +++ b/drivers/gpu/drm/xe/xe_exec_queue.c @@ -114,7 +114,6 @@ static struct xe_exec_queue *__xe_exec_queue_alloc(struct xe_device *xe, static int __xe_exec_queue_init(struct xe_exec_queue *q) { - struct xe_vm *vm = q->vm; int i, err; u32 flags = 0; @@ -132,32 +131,20 @@ static int __xe_exec_queue_init(struct xe_exec_queue *q) flags |= XE_LRC_CREATE_RUNALONE; } - if (vm) { - err = xe_vm_lock(vm, true); - if (err) - return err; - } - for (i = 0; i < q->width; ++i) { q->lrc[i] = xe_lrc_create(q->hwe, q->vm, SZ_16K, q->msix_vec, flags); if (IS_ERR(q->lrc[i])) { err = PTR_ERR(q->lrc[i]); - goto err_unlock; + goto err_lrc; } } - if (vm) - xe_vm_unlock(vm); - err = q->ops->init(q); if (err) goto err_lrc; return 0; -err_unlock: - if (vm) - xe_vm_unlock(vm); err_lrc: for (i = i - 1; i >= 0; --i) xe_lrc_put(q->lrc[i]); diff --git a/drivers/gpu/drm/xe/xe_gpu_scheduler.h b/drivers/gpu/drm/xe/xe_gpu_scheduler.h index c250ea773491..308061f0cf37 100644 --- a/drivers/gpu/drm/xe/xe_gpu_scheduler.h +++ b/drivers/gpu/drm/xe/xe_gpu_scheduler.h @@ -51,7 +51,15 @@ static inline void xe_sched_tdr_queue_imm(struct xe_gpu_scheduler *sched) static inline void xe_sched_resubmit_jobs(struct xe_gpu_scheduler *sched) { - drm_sched_resubmit_jobs(&sched->base); + struct drm_sched_job *s_job; + + list_for_each_entry(s_job, &sched->base.pending_list, list) { + struct drm_sched_fence *s_fence = s_job->s_fence; + struct dma_fence *hw_fence = s_fence->parent; + + if (hw_fence && !dma_fence_is_signaled(hw_fence)) + sched->base.ops->run_job(s_job); + } } static inline bool diff --git a/drivers/gpu/drm/xe/xe_gt_freq.c b/drivers/gpu/drm/xe/xe_gt_freq.c index 868a5d2c1a52..60d9354e7dbf 100644 --- a/drivers/gpu/drm/xe/xe_gt_freq.c +++ b/drivers/gpu/drm/xe/xe_gt_freq.c @@ -32,13 +32,18 @@ * Xe's Freq provides a sysfs API for frequency management: * * device/tile#/gt#/freq0/<item>_freq *read-only* files: + * * - act_freq: The actual resolved frequency decided by PCODE. * - cur_freq: The current one requested by GuC PC to the PCODE. * - rpn_freq: The Render Performance (RP) N level, which is the minimal one. + * - rpa_freq: The Render Performance (RP) A level, which is the achiveable one. + * Calculated by PCODE at runtime based on multiple running conditions * - rpe_freq: The Render Performance (RP) E level, which is the efficient one. + * Calculated by PCODE at runtime based on multiple running conditions * - rp0_freq: The Render Performance (RP) 0 level, which is the maximum one. * * device/tile#/gt#/freq0/<item>_freq *read-write* files: + * * - min_freq: Min frequency request. * - max_freq: Max frequency request. * If max <= min, then freq_min becomes a fixed frequency request. diff --git a/drivers/gpu/drm/xe/xe_guc_submit.c b/drivers/gpu/drm/xe/xe_guc_submit.c index 2ad38f6b103e..6d84a52b660a 100644 --- a/drivers/gpu/drm/xe/xe_guc_submit.c +++ b/drivers/gpu/drm/xe/xe_guc_submit.c @@ -229,6 +229,17 @@ static bool exec_queue_killed_or_banned_or_wedged(struct xe_exec_queue *q) static void guc_submit_fini(struct drm_device *drm, void *arg) { struct xe_guc *guc = arg; + struct xe_device *xe = guc_to_xe(guc); + struct xe_gt *gt = guc_to_gt(guc); + int ret; + + ret = wait_event_timeout(guc->submission_state.fini_wq, + xa_empty(&guc->submission_state.exec_queue_lookup), + HZ * 5); + + drain_workqueue(xe->destroy_wq); + + xe_gt_assert(gt, ret); xa_destroy(&guc->submission_state.exec_queue_lookup); } diff --git a/drivers/gpu/drm/xe/xe_hwmon.c b/drivers/gpu/drm/xe/xe_hwmon.c index eb293aec36a0..74f31639b37f 100644 --- a/drivers/gpu/drm/xe/xe_hwmon.c +++ b/drivers/gpu/drm/xe/xe_hwmon.c @@ -52,6 +52,14 @@ enum xe_fan_channel { }; /* + * For platforms that support mailbox commands for power limits, REG_PKG_POWER_SKU_UNIT is + * not supported and below are SKU units to be used. + */ +#define PWR_UNIT 0x3 +#define ENERGY_UNIT 0xe +#define TIME_UNIT 0xa + +/* * SF_* - scale factors for particular quantities according to hwmon spec. */ #define SF_POWER 1000000 /* microwatts */ @@ -60,6 +68,18 @@ enum xe_fan_channel { #define SF_ENERGY 1000000 /* microjoules */ #define SF_TIME 1000 /* milliseconds */ +/* + * PL*_HWMON_ATTR - mapping of hardware power limits to corresponding hwmon power attribute. + */ +#define PL1_HWMON_ATTR hwmon_power_max + +#define PWR_ATTR_TO_STR(attr) (((attr) == hwmon_power_max) ? "PL1" : "Invalid") + +/* + * Timeout for power limit write mailbox command. + */ +#define PL_WRITE_MBX_TIMEOUT_MS (1) + /** * struct xe_hwmon_energy_info - to accumulate energy */ @@ -100,8 +120,80 @@ struct xe_hwmon { struct xe_hwmon_energy_info ei[CHANNEL_MAX]; /** @fi: Fan info for fanN_input */ struct xe_hwmon_fan_info fi[FAN_MAX]; + /** @boot_power_limit_read: is boot power limits read */ + bool boot_power_limit_read; + /** @pl1_on_boot: power limit PL1 on boot */ + u32 pl1_on_boot[CHANNEL_MAX]; }; +static int xe_hwmon_pcode_read_power_limit(const struct xe_hwmon *hwmon, u32 attr, int channel, + u32 *uval) +{ + struct xe_tile *root_tile = xe_device_get_root_tile(hwmon->xe); + u32 val0 = 0, val1 = 0; + int ret = 0; + + ret = xe_pcode_read(root_tile, PCODE_MBOX(PCODE_POWER_SETUP, + (channel == CHANNEL_CARD) ? + READ_PSYSGPU_POWER_LIMIT : + READ_PACKAGE_POWER_LIMIT, + hwmon->boot_power_limit_read ? + READ_PL_FROM_PCODE : READ_PL_FROM_FW), + &val0, &val1); + + if (ret) { + drm_dbg(&hwmon->xe->drm, "read failed ch %d val0 0x%08x, val1 0x%08x, ret %d\n", + channel, val0, val1, ret); + *uval = 0; + return ret; + } + + /* return the value only if limit is enabled */ + if (attr == PL1_HWMON_ATTR) + *uval = (val0 & PWR_LIM_EN) ? val0 : 0; + else if (attr == hwmon_power_label) + *uval = (val0 & PWR_LIM_EN) ? 1 : 0; + else + *uval = 0; + + return ret; +} + +static int xe_hwmon_pcode_write_power_limit(const struct xe_hwmon *hwmon, u32 attr, u8 channel, + u32 uval) +{ + struct xe_tile *root_tile = xe_device_get_root_tile(hwmon->xe); + u32 val0, val1; + int ret = 0; + + ret = xe_pcode_read(root_tile, PCODE_MBOX(PCODE_POWER_SETUP, + (channel == CHANNEL_CARD) ? + READ_PSYSGPU_POWER_LIMIT : + READ_PACKAGE_POWER_LIMIT, + hwmon->boot_power_limit_read ? + READ_PL_FROM_PCODE : READ_PL_FROM_FW), + &val0, &val1); + + if (ret) + drm_dbg(&hwmon->xe->drm, "read failed ch %d val0 0x%08x, val1 0x%08x, ret %d\n", + channel, val0, val1, ret); + + if (attr == PL1_HWMON_ATTR) + val0 = uval; + else + return -EIO; + + ret = xe_pcode_write64_timeout(root_tile, PCODE_MBOX(PCODE_POWER_SETUP, + (channel == CHANNEL_CARD) ? + WRITE_PSYSGPU_POWER_LIMIT : + WRITE_PACKAGE_POWER_LIMIT, 0), + val0, val1, PL_WRITE_MBX_TIMEOUT_MS); + if (ret) + drm_dbg(&hwmon->xe->drm, "write failed ch %d val0 0x%08x, val1 0x%08x, ret %d\n", + channel, val0, val1, ret); + return ret; +} + static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg hwmon_reg, int channel) { @@ -122,29 +214,19 @@ static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg } break; case REG_PKG_RAPL_LIMIT: - if (xe->info.platform == XE_BATTLEMAGE) { - if (channel == CHANNEL_PKG) - return BMG_PACKAGE_RAPL_LIMIT; - else - return BMG_PLATFORM_POWER_LIMIT; - } else if (xe->info.platform == XE_PVC && channel == CHANNEL_PKG) { + if (xe->info.platform == XE_PVC && channel == CHANNEL_PKG) return PVC_GT0_PACKAGE_RAPL_LIMIT; - } else if ((xe->info.platform == XE_DG2) && (channel == CHANNEL_PKG)) { + else if ((xe->info.platform == XE_DG2) && (channel == CHANNEL_PKG)) return PCU_CR_PACKAGE_RAPL_LIMIT; - } break; case REG_PKG_POWER_SKU: - if (xe->info.platform == XE_BATTLEMAGE) - return BMG_PACKAGE_POWER_SKU; - else if (xe->info.platform == XE_PVC && channel == CHANNEL_PKG) + if (xe->info.platform == XE_PVC && channel == CHANNEL_PKG) return PVC_GT0_PACKAGE_POWER_SKU; else if ((xe->info.platform == XE_DG2) && (channel == CHANNEL_PKG)) return PCU_CR_PACKAGE_POWER_SKU; break; case REG_PKG_POWER_SKU_UNIT: - if (xe->info.platform == XE_BATTLEMAGE) - return BMG_PACKAGE_POWER_SKU_UNIT; - else if (xe->info.platform == XE_PVC) + if (xe->info.platform == XE_PVC) return PVC_GT0_PACKAGE_POWER_SKU_UNIT; else if (xe->info.platform == XE_DG2) return PCU_CR_PACKAGE_POWER_SKU_UNIT; @@ -181,7 +263,7 @@ static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg return XE_REG(0); } -#define PL1_DISABLE 0 +#define PL_DISABLE 0 /* * HW allows arbitrary PL1 limits to be set but silently clamps these values to @@ -189,67 +271,83 @@ static struct xe_reg xe_hwmon_get_reg(struct xe_hwmon *hwmon, enum xe_hwmon_reg * same pattern for sysfs, allow arbitrary PL1 limits to be set but display * clamped values when read. */ -static void xe_hwmon_power_max_read(struct xe_hwmon *hwmon, int channel, long *value) +static void xe_hwmon_power_max_read(struct xe_hwmon *hwmon, u32 attr, int channel, long *value) { u64 reg_val, min, max; struct xe_device *xe = hwmon->xe; struct xe_reg rapl_limit, pkg_power_sku; struct xe_mmio *mmio = xe_root_tile_mmio(xe); - rapl_limit = xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel); - pkg_power_sku = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel); + mutex_lock(&hwmon->hwmon_lock); - /* - * Valid check of REG_PKG_RAPL_LIMIT is already done in xe_hwmon_power_is_visible. - * So not checking it again here. - */ - if (!xe_reg_is_valid(pkg_power_sku)) { - drm_warn(&xe->drm, "pkg_power_sku invalid\n"); - *value = 0; - return; + if (hwmon->xe->info.has_mbx_power_limits) { + xe_hwmon_pcode_read_power_limit(hwmon, attr, channel, (u32 *)®_val); + } else { + rapl_limit = xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel); + pkg_power_sku = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel); + + /* + * Valid check of REG_PKG_RAPL_LIMIT is already done in xe_hwmon_power_is_visible. + * So not checking it again here. + */ + if (!xe_reg_is_valid(pkg_power_sku)) { + drm_warn(&xe->drm, "pkg_power_sku invalid\n"); + *value = 0; + goto unlock; + } + reg_val = xe_mmio_read32(mmio, rapl_limit); } - mutex_lock(&hwmon->hwmon_lock); - - reg_val = xe_mmio_read32(mmio, rapl_limit); - /* Check if PL1 limit is disabled */ - if (!(reg_val & PKG_PWR_LIM_1_EN)) { - *value = PL1_DISABLE; + /* Check if PL limits are disabled. */ + if (!(reg_val & PWR_LIM_EN)) { + *value = PL_DISABLE; + drm_info(&hwmon->xe->drm, "%s disabled for channel %d, val 0x%016llx\n", + PWR_ATTR_TO_STR(attr), channel, reg_val); goto unlock; } - reg_val = REG_FIELD_GET(PKG_PWR_LIM_1, reg_val); + reg_val = REG_FIELD_GET(PWR_LIM_VAL, reg_val); *value = mul_u64_u32_shr(reg_val, SF_POWER, hwmon->scl_shift_power); - reg_val = xe_mmio_read64_2x32(mmio, pkg_power_sku); - min = REG_FIELD_GET(PKG_MIN_PWR, reg_val); - min = mul_u64_u32_shr(min, SF_POWER, hwmon->scl_shift_power); - max = REG_FIELD_GET(PKG_MAX_PWR, reg_val); - max = mul_u64_u32_shr(max, SF_POWER, hwmon->scl_shift_power); - - if (min && max) - *value = clamp_t(u64, *value, min, max); + /* For platforms with mailbox power limit support clamping would be done by pcode. */ + if (!hwmon->xe->info.has_mbx_power_limits) { + reg_val = xe_mmio_read64_2x32(mmio, pkg_power_sku); + min = REG_FIELD_GET(PKG_MIN_PWR, reg_val); + max = REG_FIELD_GET(PKG_MAX_PWR, reg_val); + min = mul_u64_u32_shr(min, SF_POWER, hwmon->scl_shift_power); + max = mul_u64_u32_shr(max, SF_POWER, hwmon->scl_shift_power); + if (min && max) + *value = clamp_t(u64, *value, min, max); + } unlock: mutex_unlock(&hwmon->hwmon_lock); } -static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, int channel, long value) +static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, u32 attr, int channel, long value) { struct xe_mmio *mmio = xe_root_tile_mmio(hwmon->xe); int ret = 0; - u64 reg_val; + u32 reg_val; struct xe_reg rapl_limit; + mutex_lock(&hwmon->hwmon_lock); + rapl_limit = xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel); - mutex_lock(&hwmon->hwmon_lock); + /* Disable Power Limit and verify, as limit cannot be disabled on all platforms. */ + if (value == PL_DISABLE) { + if (hwmon->xe->info.has_mbx_power_limits) { + drm_dbg(&hwmon->xe->drm, "disabling %s on channel %d\n", + PWR_ATTR_TO_STR(attr), channel); + xe_hwmon_pcode_write_power_limit(hwmon, attr, channel, 0); + xe_hwmon_pcode_read_power_limit(hwmon, attr, channel, ®_val); + } else { + reg_val = xe_mmio_rmw32(mmio, rapl_limit, PWR_LIM_EN, 0); + reg_val = xe_mmio_read32(mmio, rapl_limit); + } - /* Disable PL1 limit and verify, as limit cannot be disabled on all platforms */ - if (value == PL1_DISABLE) { - reg_val = xe_mmio_rmw32(mmio, rapl_limit, PKG_PWR_LIM_1_EN, 0); - reg_val = xe_mmio_read32(mmio, rapl_limit); - if (reg_val & PKG_PWR_LIM_1_EN) { - drm_warn(&hwmon->xe->drm, "PL1 disable is not supported!\n"); + if (reg_val & PWR_LIM_EN) { + drm_warn(&hwmon->xe->drm, "Power limit disable is not supported!\n"); ret = -EOPNOTSUPP; } goto unlock; @@ -257,26 +355,50 @@ static int xe_hwmon_power_max_write(struct xe_hwmon *hwmon, int channel, long va /* Computation in 64-bits to avoid overflow. Round to nearest. */ reg_val = DIV_ROUND_CLOSEST_ULL((u64)value << hwmon->scl_shift_power, SF_POWER); - reg_val = PKG_PWR_LIM_1_EN | REG_FIELD_PREP(PKG_PWR_LIM_1, reg_val); - reg_val = xe_mmio_rmw32(mmio, rapl_limit, PKG_PWR_LIM_1_EN | PKG_PWR_LIM_1, reg_val); + reg_val = PWR_LIM_EN | REG_FIELD_PREP(PWR_LIM_VAL, reg_val); + /* + * Clamp power limit to card-firmware default as maximum, as an additional protection to + * pcode clamp. + */ + if (hwmon->xe->info.has_mbx_power_limits) { + if (reg_val > REG_FIELD_GET(PWR_LIM_VAL, hwmon->pl1_on_boot[channel])) { + reg_val = REG_FIELD_GET(PWR_LIM_VAL, hwmon->pl1_on_boot[channel]); + drm_dbg(&hwmon->xe->drm, "Clamping power limit to firmware default 0x%x\n", + reg_val); + } + } + + if (hwmon->xe->info.has_mbx_power_limits) + ret = xe_hwmon_pcode_write_power_limit(hwmon, attr, channel, reg_val); + else + reg_val = xe_mmio_rmw32(mmio, rapl_limit, PWR_LIM_EN | PWR_LIM_VAL, + reg_val); unlock: mutex_unlock(&hwmon->hwmon_lock); return ret; } -static void xe_hwmon_power_rated_max_read(struct xe_hwmon *hwmon, int channel, long *value) +static void xe_hwmon_power_rated_max_read(struct xe_hwmon *hwmon, u32 attr, int channel, + long *value) { struct xe_mmio *mmio = xe_root_tile_mmio(hwmon->xe); - struct xe_reg reg = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel); - u64 reg_val; + u32 reg_val; + + if (hwmon->xe->info.has_mbx_power_limits) { + /* PL1 is rated max if supported. */ + xe_hwmon_pcode_read_power_limit(hwmon, PL1_HWMON_ATTR, channel, ®_val); + } else { + /* + * This sysfs file won't be visible if REG_PKG_POWER_SKU is invalid, so valid check + * for this register can be skipped. + * See xe_hwmon_power_is_visible. + */ + struct xe_reg reg = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, channel); + + reg_val = xe_mmio_read32(mmio, reg); + } - /* - * This sysfs file won't be visible if REG_PKG_POWER_SKU is invalid, so valid check - * for this register can be skipped. - * See xe_hwmon_power_is_visible. - */ - reg_val = xe_mmio_read32(mmio, reg); reg_val = REG_FIELD_GET(PKG_TDP, reg_val); *value = mul_u64_u32_shr(reg_val, SF_POWER, hwmon->scl_shift_power); } @@ -330,23 +452,35 @@ xe_hwmon_power_max_interval_show(struct device *dev, struct device_attribute *at struct xe_mmio *mmio = xe_root_tile_mmio(hwmon->xe); u32 x, y, x_w = 2; /* 2 bits */ u64 r, tau4, out; - int sensor_index = to_sensor_dev_attr(attr)->index; + int channel = to_sensor_dev_attr(attr)->index; + u32 power_attr = PL1_HWMON_ATTR; + int ret = 0; xe_pm_runtime_get(hwmon->xe); mutex_lock(&hwmon->hwmon_lock); - r = xe_mmio_read32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index)); + if (hwmon->xe->info.has_mbx_power_limits) { + ret = xe_hwmon_pcode_read_power_limit(hwmon, power_attr, channel, (u32 *)&r); + if (ret) { + drm_err(&hwmon->xe->drm, + "power interval read fail, ch %d, attr %d, r 0%llx, ret %d\n", + channel, power_attr, r, ret); + r = 0; + } + } else { + r = xe_mmio_read32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel)); + } mutex_unlock(&hwmon->hwmon_lock); xe_pm_runtime_put(hwmon->xe); - x = REG_FIELD_GET(PKG_PWR_LIM_1_TIME_X, r); - y = REG_FIELD_GET(PKG_PWR_LIM_1_TIME_Y, r); + x = REG_FIELD_GET(PWR_LIM_TIME_X, r); + y = REG_FIELD_GET(PWR_LIM_TIME_Y, r); /* - * tau = 1.x * power(2,y), x = bits(23:22), y = bits(21:17) + * tau = (1 + (x / 4)) * power(2,y), x = bits(23:22), y = bits(21:17) * = (4 | x) << (y - 2) * * Here (y - 2) ensures a 1.x fixed point representation of 1.x @@ -373,14 +507,15 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a u64 tau4, r, max_win; unsigned long val; int ret; - int sensor_index = to_sensor_dev_attr(attr)->index; + int channel = to_sensor_dev_attr(attr)->index; + u32 power_attr = PL1_HWMON_ATTR; ret = kstrtoul(buf, 0, &val); if (ret) return ret; /* - * Max HW supported tau in '1.x * power(2,y)' format, x = 0, y = 0x12. + * Max HW supported tau in '(1 + (x / 4)) * power(2,y)' format, x = 0, y = 0x12. * The hwmon->scl_shift_time default of 0xa results in a max tau of 256 seconds. * * The ideal scenario is for PKG_MAX_WIN to be read from the PKG_PWR_SKU register. @@ -400,11 +535,13 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a tau4 = (u64)((1 << x_w) | x) << y; max_win = mul_u64_u32_shr(tau4, SF_TIME, hwmon->scl_shift_time + x_w); - if (val > max_win) + if (val > max_win) { + drm_warn(&hwmon->xe->drm, "power_interval invalid val 0x%lx\n", val); return -EINVAL; + } /* val in hw units */ - val = DIV_ROUND_CLOSEST_ULL((u64)val << hwmon->scl_shift_time, SF_TIME); + val = DIV_ROUND_CLOSEST_ULL((u64)val << hwmon->scl_shift_time, SF_TIME) + 1; /* * Convert val to 1.x * power(2,y) @@ -419,14 +556,21 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a x = (val - (1ul << y)) << x_w >> y; } - rxy = REG_FIELD_PREP(PKG_PWR_LIM_1_TIME_X, x) | REG_FIELD_PREP(PKG_PWR_LIM_1_TIME_Y, y); + rxy = REG_FIELD_PREP(PWR_LIM_TIME_X, x) | + REG_FIELD_PREP(PWR_LIM_TIME_Y, y); xe_pm_runtime_get(hwmon->xe); mutex_lock(&hwmon->hwmon_lock); - r = xe_mmio_rmw32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, sensor_index), - PKG_PWR_LIM_1_TIME, rxy); + if (hwmon->xe->info.has_mbx_power_limits) { + ret = xe_hwmon_pcode_read_power_limit(hwmon, power_attr, channel, (u32 *)&r); + r = (r & ~PWR_LIM_TIME) | rxy; + xe_hwmon_pcode_write_power_limit(hwmon, power_attr, channel, r); + } else { + r = xe_mmio_rmw32(mmio, xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel), + PWR_LIM_TIME, rxy); + } mutex_unlock(&hwmon->hwmon_lock); @@ -435,6 +579,7 @@ xe_hwmon_power_max_interval_store(struct device *dev, struct device_attribute *a return count; } +/* PSYS PL1 */ static SENSOR_DEVICE_ATTR(power1_max_interval, 0664, xe_hwmon_power_max_interval_show, xe_hwmon_power_max_interval_store, CHANNEL_CARD); @@ -455,10 +600,19 @@ static umode_t xe_hwmon_attributes_visible(struct kobject *kobj, struct device *dev = kobj_to_dev(kobj); struct xe_hwmon *hwmon = dev_get_drvdata(dev); int ret = 0; + int channel = index ? CHANNEL_PKG : CHANNEL_CARD; + u32 power_attr = PL1_HWMON_ATTR; + u32 uval; xe_pm_runtime_get(hwmon->xe); - ret = xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, index)) ? attr->mode : 0; + if (hwmon->xe->info.has_mbx_power_limits) { + xe_hwmon_pcode_read_power_limit(hwmon, power_attr, channel, &uval); + ret = (uval & PWR_LIM_EN) ? attr->mode : 0; + } else { + ret = xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, + channel)) ? attr->mode : 0; + } xe_pm_runtime_put(hwmon->xe); @@ -478,8 +632,8 @@ static const struct attribute_group *hwmon_groups[] = { static const struct hwmon_channel_info * const hwmon_info[] = { HWMON_CHANNEL_INFO(temp, HWMON_T_LABEL, HWMON_T_INPUT | HWMON_T_LABEL, HWMON_T_INPUT | HWMON_T_LABEL), - HWMON_CHANNEL_INFO(power, HWMON_P_MAX | HWMON_P_RATED_MAX | HWMON_P_LABEL, - HWMON_P_MAX | HWMON_P_RATED_MAX | HWMON_P_CRIT | HWMON_P_LABEL), + HWMON_CHANNEL_INFO(power, HWMON_P_MAX | HWMON_P_RATED_MAX | HWMON_P_LABEL | HWMON_P_CRIT, + HWMON_P_MAX | HWMON_P_RATED_MAX | HWMON_P_LABEL), HWMON_CHANNEL_INFO(curr, HWMON_C_LABEL, HWMON_C_CRIT | HWMON_C_LABEL), HWMON_CHANNEL_INFO(in, HWMON_I_INPUT | HWMON_I_LABEL, HWMON_I_INPUT | HWMON_I_LABEL), HWMON_CHANNEL_INFO(energy, HWMON_E_INPUT | HWMON_E_LABEL, HWMON_E_INPUT | HWMON_E_LABEL), @@ -604,19 +758,27 @@ xe_hwmon_power_is_visible(struct xe_hwmon *hwmon, u32 attr, int channel) switch (attr) { case hwmon_power_max: - return xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, + if (hwmon->xe->info.has_mbx_power_limits) { + xe_hwmon_pcode_read_power_limit(hwmon, attr, channel, &uval); + return (uval) ? 0664 : 0; + } else { + return xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_RAPL_LIMIT, channel)) ? 0664 : 0; + } case hwmon_power_rated_max: - return xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, - channel)) ? 0444 : 0; + if (hwmon->xe->info.has_mbx_power_limits) + return 0; + else + return xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU, + channel)) ? 0444 : 0; case hwmon_power_crit: - if (channel == CHANNEL_PKG) - return (xe_hwmon_pcode_read_i1(hwmon, &uval) || - !(uval & POWER_SETUP_I1_WATTS)) ? 0 : 0644; - break; case hwmon_power_label: - return xe_reg_is_valid(xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU_UNIT, - channel)) ? 0444 : 0; + if (channel == CHANNEL_CARD) { + xe_hwmon_pcode_read_i1(hwmon, &uval); + return (uval & POWER_SETUP_I1_WATTS) ? (attr == hwmon_power_label) ? + 0444 : 0644 : 0; + } + break; default: return 0; } @@ -628,10 +790,10 @@ xe_hwmon_power_read(struct xe_hwmon *hwmon, u32 attr, int channel, long *val) { switch (attr) { case hwmon_power_max: - xe_hwmon_power_max_read(hwmon, channel, val); + xe_hwmon_power_max_read(hwmon, attr, channel, val); return 0; case hwmon_power_rated_max: - xe_hwmon_power_rated_max_read(hwmon, channel, val); + xe_hwmon_power_rated_max_read(hwmon, attr, channel, val); return 0; case hwmon_power_crit: return xe_hwmon_power_curr_crit_read(hwmon, channel, val, SF_POWER); @@ -645,7 +807,7 @@ xe_hwmon_power_write(struct xe_hwmon *hwmon, u32 attr, int channel, long val) { switch (attr) { case hwmon_power_max: - return xe_hwmon_power_max_write(hwmon, channel, val); + return xe_hwmon_power_max_write(hwmon, attr, channel, val); case hwmon_power_crit: return xe_hwmon_power_curr_crit_write(hwmon, channel, val, SF_POWER); default: @@ -965,18 +1127,42 @@ xe_hwmon_get_preregistration_info(struct xe_hwmon *hwmon) int channel; struct xe_reg pkg_power_sku_unit; - /* - * The contents of register PKG_POWER_SKU_UNIT do not change, - * so read it once and store the shift values. - */ - pkg_power_sku_unit = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU_UNIT, 0); - if (xe_reg_is_valid(pkg_power_sku_unit)) { - val_sku_unit = xe_mmio_read32(mmio, pkg_power_sku_unit); - hwmon->scl_shift_power = REG_FIELD_GET(PKG_PWR_UNIT, val_sku_unit); - hwmon->scl_shift_energy = REG_FIELD_GET(PKG_ENERGY_UNIT, val_sku_unit); - hwmon->scl_shift_time = REG_FIELD_GET(PKG_TIME_UNIT, val_sku_unit); + if (hwmon->xe->info.has_mbx_power_limits) { + /* Check if card firmware support mailbox power limits commands. */ + if (xe_hwmon_pcode_read_power_limit(hwmon, PL1_HWMON_ATTR, CHANNEL_CARD, + &hwmon->pl1_on_boot[CHANNEL_CARD]) | + xe_hwmon_pcode_read_power_limit(hwmon, PL1_HWMON_ATTR, CHANNEL_PKG, + &hwmon->pl1_on_boot[CHANNEL_PKG])) { + drm_warn(&hwmon->xe->drm, + "Failed to read power limits, check card firmware !\n"); + } else { + drm_info(&hwmon->xe->drm, "Using mailbox commands for power limits\n"); + /* Write default limits to read from pcode from now on. */ + xe_hwmon_pcode_write_power_limit(hwmon, PL1_HWMON_ATTR, + CHANNEL_CARD, + hwmon->pl1_on_boot[CHANNEL_CARD]); + xe_hwmon_pcode_write_power_limit(hwmon, PL1_HWMON_ATTR, + CHANNEL_PKG, + hwmon->pl1_on_boot[CHANNEL_PKG]); + hwmon->scl_shift_power = PWR_UNIT; + hwmon->scl_shift_energy = ENERGY_UNIT; + hwmon->scl_shift_time = TIME_UNIT; + hwmon->boot_power_limit_read = true; + } + } else { + drm_info(&hwmon->xe->drm, "Using register for power limits\n"); + /* + * The contents of register PKG_POWER_SKU_UNIT do not change, + * so read it once and store the shift values. + */ + pkg_power_sku_unit = xe_hwmon_get_reg(hwmon, REG_PKG_POWER_SKU_UNIT, 0); + if (xe_reg_is_valid(pkg_power_sku_unit)) { + val_sku_unit = xe_mmio_read32(mmio, pkg_power_sku_unit); + hwmon->scl_shift_power = REG_FIELD_GET(PKG_PWR_UNIT, val_sku_unit); + hwmon->scl_shift_energy = REG_FIELD_GET(PKG_ENERGY_UNIT, val_sku_unit); + hwmon->scl_shift_time = REG_FIELD_GET(PKG_TIME_UNIT, val_sku_unit); + } } - /* * Initialize 'struct xe_hwmon_energy_info', i.e. set fields to the * first value of the energy register read diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c index 61a2e87990a9..63d74e27f54c 100644 --- a/drivers/gpu/drm/xe/xe_lrc.c +++ b/drivers/gpu/drm/xe/xe_lrc.c @@ -909,10 +909,7 @@ static void xe_lrc_set_ppgtt(struct xe_lrc *lrc, struct xe_vm *vm) static void xe_lrc_finish(struct xe_lrc *lrc) { xe_hw_fence_ctx_finish(&lrc->fence_ctx); - xe_bo_lock(lrc->bo, false); - xe_bo_unpin(lrc->bo); - xe_bo_unlock(lrc->bo); - xe_bo_put(lrc->bo); + xe_bo_unpin_map_no_vm(lrc->bo); xe_bo_unpin_map_no_vm(lrc->bb_per_ctx_bo); } @@ -1007,7 +1004,7 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe, * FIXME: Perma-pinning LRC as we don't yet support moving GGTT address * via VM bind calls. */ - lrc->bo = xe_bo_create_pin_map(xe, tile, vm, lrc_size, + lrc->bo = xe_bo_create_pin_map(xe, tile, NULL, lrc_size, ttm_bo_type_kernel, bo_flags); if (IS_ERR(lrc->bo)) @@ -1795,9 +1792,6 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc) if (!snapshot) return NULL; - if (lrc->bo->vm) - xe_vm_get(lrc->bo->vm); - snapshot->context_desc = xe_lrc_ggtt_addr(lrc); snapshot->ring_addr = __xe_lrc_ring_ggtt_addr(lrc); snapshot->indirect_context_desc = xe_lrc_indirect_ring_ggtt_addr(lrc); @@ -1819,14 +1813,12 @@ struct xe_lrc_snapshot *xe_lrc_snapshot_capture(struct xe_lrc *lrc) void xe_lrc_snapshot_capture_delayed(struct xe_lrc_snapshot *snapshot) { struct xe_bo *bo; - struct xe_vm *vm; struct iosys_map src; if (!snapshot) return; bo = snapshot->lrc_bo; - vm = bo->vm; snapshot->lrc_bo = NULL; snapshot->lrc_snapshot = kvmalloc(snapshot->lrc_size, GFP_KERNEL); @@ -1846,8 +1838,6 @@ void xe_lrc_snapshot_capture_delayed(struct xe_lrc_snapshot *snapshot) xe_bo_unlock(bo); put_bo: xe_bo_put(bo); - if (vm) - xe_vm_put(vm); } void xe_lrc_snapshot_print(struct xe_lrc_snapshot *snapshot, struct drm_printer *p) @@ -1900,14 +1890,9 @@ void xe_lrc_snapshot_free(struct xe_lrc_snapshot *snapshot) return; kvfree(snapshot->lrc_snapshot); - if (snapshot->lrc_bo) { - struct xe_vm *vm; - - vm = snapshot->lrc_bo->vm; + if (snapshot->lrc_bo) xe_bo_put(snapshot->lrc_bo); - if (vm) - xe_vm_put(vm); - } + kfree(snapshot); } diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c index 024175cfe61e..ac4beaed58ff 100644 --- a/drivers/gpu/drm/xe/xe_pci.c +++ b/drivers/gpu/drm/xe/xe_pci.c @@ -66,6 +66,7 @@ struct xe_device_desc { u8 has_heci_gscfi:1; u8 has_heci_cscfi:1; u8 has_llc:1; + u8 has_mbx_power_limits:1; u8 has_pxp:1; u8 has_sriov:1; u8 needs_scratch:1; @@ -306,6 +307,7 @@ static const struct xe_device_desc dg2_desc = { DG2_FEATURES, .has_display = true, .has_fan_control = true, + .has_mbx_power_limits = false, }; static const __maybe_unused struct xe_device_desc pvc_desc = { @@ -317,6 +319,7 @@ static const __maybe_unused struct xe_device_desc pvc_desc = { .has_heci_gscfi = 1, .max_remote_tiles = 1, .require_force_probe = true, + .has_mbx_power_limits = false, }; static const struct xe_device_desc mtl_desc = { @@ -342,6 +345,7 @@ static const struct xe_device_desc bmg_desc = { .dma_mask_size = 46, .has_display = true, .has_fan_control = true, + .has_mbx_power_limits = true, .has_heci_cscfi = 1, .needs_scratch = true, }; @@ -584,6 +588,7 @@ static int xe_info_init_early(struct xe_device *xe, xe->info.dma_mask_size = desc->dma_mask_size; xe->info.is_dgfx = desc->is_dgfx; xe->info.has_fan_control = desc->has_fan_control; + xe->info.has_mbx_power_limits = desc->has_mbx_power_limits; xe->info.has_heci_gscfi = desc->has_heci_gscfi; xe->info.has_heci_cscfi = desc->has_heci_cscfi; xe->info.has_llc = desc->has_llc; diff --git a/drivers/gpu/drm/xe/xe_pcode.c b/drivers/gpu/drm/xe/xe_pcode.c index cf955b3ed52c..9189117fe825 100644 --- a/drivers/gpu/drm/xe/xe_pcode.c +++ b/drivers/gpu/drm/xe/xe_pcode.c @@ -109,6 +109,17 @@ int xe_pcode_write_timeout(struct xe_tile *tile, u32 mbox, u32 data, int timeout return err; } +int xe_pcode_write64_timeout(struct xe_tile *tile, u32 mbox, u32 data0, u32 data1, int timeout) +{ + int err; + + mutex_lock(&tile->pcode.lock); + err = pcode_mailbox_rw(tile, mbox, &data0, &data1, timeout, false, false); + mutex_unlock(&tile->pcode.lock); + + return err; +} + int xe_pcode_read(struct xe_tile *tile, u32 mbox, u32 *val, u32 *val1) { int err; diff --git a/drivers/gpu/drm/xe/xe_pcode.h b/drivers/gpu/drm/xe/xe_pcode.h index ba33991d72a7..de38f44f3201 100644 --- a/drivers/gpu/drm/xe/xe_pcode.h +++ b/drivers/gpu/drm/xe/xe_pcode.h @@ -18,6 +18,9 @@ int xe_pcode_init_min_freq_table(struct xe_tile *tile, u32 min_gt_freq, int xe_pcode_read(struct xe_tile *tile, u32 mbox, u32 *val, u32 *val1); int xe_pcode_write_timeout(struct xe_tile *tile, u32 mbox, u32 val, int timeout_ms); +int xe_pcode_write64_timeout(struct xe_tile *tile, u32 mbox, u32 data0, + u32 data1, int timeout); + #define xe_pcode_write(tile, mbox, val) \ xe_pcode_write_timeout(tile, mbox, val, 1) diff --git a/drivers/gpu/drm/xe/xe_pcode_api.h b/drivers/gpu/drm/xe/xe_pcode_api.h index 127d4d26c4cf..0befdea77db1 100644 --- a/drivers/gpu/drm/xe/xe_pcode_api.h +++ b/drivers/gpu/drm/xe/xe_pcode_api.h @@ -43,6 +43,13 @@ #define POWER_SETUP_I1_SHIFT 6 /* 10.6 fixed point format */ #define POWER_SETUP_I1_DATA_MASK REG_GENMASK(15, 0) +#define READ_PSYSGPU_POWER_LIMIT 0x6 +#define WRITE_PSYSGPU_POWER_LIMIT 0x7 +#define READ_PACKAGE_POWER_LIMIT 0x8 +#define WRITE_PACKAGE_POWER_LIMIT 0x9 +#define READ_PL_FROM_FW 0x1 +#define READ_PL_FROM_PCODE 0x0 + #define PCODE_FREQUENCY_CONFIG 0x6e /* Frequency Config Sub Commands (param1) */ #define PCODE_MBOX_FC_SC_READ_FUSED_P0 0x0 diff --git a/drivers/gpu/drm/xe/xe_pxp.c b/drivers/gpu/drm/xe/xe_pxp.c index 454ea7dc08ac..b5bc15f436fa 100644 --- a/drivers/gpu/drm/xe/xe_pxp.c +++ b/drivers/gpu/drm/xe/xe_pxp.c @@ -541,10 +541,14 @@ int xe_pxp_exec_queue_add(struct xe_pxp *pxp, struct xe_exec_queue *q) */ xe_pm_runtime_get(pxp->xe); - if (!pxp_prerequisites_done(pxp)) { - ret = -EBUSY; + /* get_readiness_status() returns 0 for in-progress and 1 for done */ + ret = xe_pxp_get_readiness_status(pxp); + if (ret <= 0) { + if (!ret) + ret = -EBUSY; goto out; } + ret = 0; wait_for_idle: /* diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 79323c78130f..861577746929 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -1678,13 +1678,21 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) * scheduler drops all the references of it, hence protecting the VM * for this case is necessary. */ - if (flags & XE_VM_FLAG_LR_MODE) + if (flags & XE_VM_FLAG_LR_MODE) { + INIT_WORK(&vm->preempt.rebind_work, preempt_rebind_work_func); xe_pm_runtime_get_noresume(xe); + } + + if (flags & XE_VM_FLAG_FAULT_MODE) { + err = xe_svm_init(vm); + if (err) + goto err_no_resv; + } vm_resv_obj = drm_gpuvm_resv_object_alloc(&xe->drm); if (!vm_resv_obj) { err = -ENOMEM; - goto err_no_resv; + goto err_svm_fini; } drm_gpuvm_init(&vm->gpuvm, "Xe VM", DRM_GPUVM_RESV_PROTECTED, &xe->drm, @@ -1724,10 +1732,8 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) vm->batch_invalidate_tlb = true; } - if (vm->flags & XE_VM_FLAG_LR_MODE) { - INIT_WORK(&vm->preempt.rebind_work, preempt_rebind_work_func); + if (vm->flags & XE_VM_FLAG_LR_MODE) vm->batch_invalidate_tlb = false; - } /* Fill pt_root after allocating scratch tables */ for_each_tile(tile, xe, id) { @@ -1757,12 +1763,6 @@ struct xe_vm *xe_vm_create(struct xe_device *xe, u32 flags) } } - if (flags & XE_VM_FLAG_FAULT_MODE) { - err = xe_svm_init(vm); - if (err) - goto err_close; - } - if (number_tiles > 1) vm->composite_fence_ctx = dma_fence_context_alloc(1); @@ -1776,6 +1776,11 @@ err_close: xe_vm_close_and_put(vm); return ERR_PTR(err); +err_svm_fini: + if (flags & XE_VM_FLAG_FAULT_MODE) { + vm->size = 0; /* close the vm */ + xe_svm_fini(vm); + } err_no_resv: mutex_destroy(&vm->snap_mutex); for_each_tile(tile, xe, id) diff --git a/drivers/gpu/drm/xe/xe_vm.h b/drivers/gpu/drm/xe/xe_vm.h index 0ef811fc2bde..494af6bdc646 100644 --- a/drivers/gpu/drm/xe/xe_vm.h +++ b/drivers/gpu/drm/xe/xe_vm.h @@ -301,6 +301,75 @@ void xe_vm_snapshot_capture_delayed(struct xe_vm_snapshot *snap); void xe_vm_snapshot_print(struct xe_vm_snapshot *snap, struct drm_printer *p); void xe_vm_snapshot_free(struct xe_vm_snapshot *snap); +/** + * xe_vm_set_validating() - Register this task as currently making bos resident + * @allow_res_evict: Allow eviction of buffer objects bound to @vm when + * validating. + * @vm: Pointer to the vm or NULL. + * + * Register this task as currently making bos resident for the vm. Intended + * to avoid eviction by the same task of shared bos bound to the vm. + * Call with the vm's resv lock held. + * + * Return: A pin cookie that should be used for xe_vm_clear_validating(). + */ +static inline struct pin_cookie xe_vm_set_validating(struct xe_vm *vm, + bool allow_res_evict) +{ + struct pin_cookie cookie = {}; + + if (vm && !allow_res_evict) { + xe_vm_assert_held(vm); + cookie = lockdep_pin_lock(&xe_vm_resv(vm)->lock.base); + /* Pairs with READ_ONCE in xe_vm_is_validating() */ + WRITE_ONCE(vm->validating, current); + } + + return cookie; +} + +/** + * xe_vm_clear_validating() - Unregister this task as currently making bos resident + * @vm: Pointer to the vm or NULL + * @allow_res_evict: Eviction from @vm was allowed. Must be set to the same + * value as for xe_vm_set_validation(). + * @cookie: Cookie obtained from xe_vm_set_validating(). + * + * Register this task as currently making bos resident for the vm. Intended + * to avoid eviction by the same task of shared bos bound to the vm. + * Call with the vm's resv lock held. + */ +static inline void xe_vm_clear_validating(struct xe_vm *vm, bool allow_res_evict, + struct pin_cookie cookie) +{ + if (vm && !allow_res_evict) { + lockdep_unpin_lock(&xe_vm_resv(vm)->lock.base, cookie); + /* Pairs with READ_ONCE in xe_vm_is_validating() */ + WRITE_ONCE(vm->validating, NULL); + } +} + +/** + * xe_vm_is_validating() - Whether bos bound to the vm are currently being made resident + * by the current task. + * @vm: Pointer to the vm. + * + * If this function returns %true, we should be in a vm resv locked region, since + * the current process is the same task that called xe_vm_set_validating(). + * The function asserts that that's indeed the case. + * + * Return: %true if the task is currently making bos resident, %false otherwise. + */ +static inline bool xe_vm_is_validating(struct xe_vm *vm) +{ + /* Pairs with WRITE_ONCE in xe_vm_is_validating() */ + if (READ_ONCE(vm->validating) == current) { + xe_vm_assert_held(vm); + return true; + } + return false; +} + #if IS_ENABLED(CONFIG_DRM_XE_USERPTR_INVAL_INJECT) void xe_vma_userptr_force_invalidate(struct xe_userptr_vma *uvma); #else diff --git a/drivers/gpu/drm/xe/xe_vm_types.h b/drivers/gpu/drm/xe/xe_vm_types.h index 1662604c4486..1979e9bdbdf3 100644 --- a/drivers/gpu/drm/xe/xe_vm_types.h +++ b/drivers/gpu/drm/xe/xe_vm_types.h @@ -310,6 +310,14 @@ struct xe_vm { * protected by the vm resv. */ u64 tlb_flush_seqno; + /** + * @validating: The task that is currently making bos resident for this vm. + * Protected by the VM's resv for writing. Opportunistic reading can be done + * using READ_ONCE. Note: This is a workaround for the + * TTM eviction_valuable() callback not being passed a struct + * ttm_operation_context(). Future work might want to address this. + */ + struct task_struct *validating; /** @batch_invalidate_tlb: Always invalidate TLB before batch start */ bool batch_invalidate_tlb; /** @xef: XE file handle for tracking this VM's drm client */ |