diff options
Diffstat (limited to 'drivers/gpu/drm')
253 files changed, 3337 insertions, 1729 deletions
diff --git a/drivers/gpu/drm/Kconfig b/drivers/gpu/drm/Kconfig index fbef3f471bd0..bd228dc77e99 100644 --- a/drivers/gpu/drm/Kconfig +++ b/drivers/gpu/drm/Kconfig @@ -188,7 +188,7 @@ config DRM_DEBUG_DP_MST_TOPOLOGY_REFS bool "Enable refcount backtrace history in the DP MST helpers" depends on STACKTRACE_SUPPORT select STACKDEPOT - depends on DRM_KMS_HELPER + select DRM_KMS_HELPER depends on DEBUG_KERNEL depends on EXPERT help diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h index 69895fccb474..90f688b3d9d3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h @@ -352,7 +352,6 @@ enum amdgpu_kiq_irq { AMDGPU_CP_KIQ_IRQ_DRIVER0 = 0, AMDGPU_CP_KIQ_IRQ_LAST }; -#define SRIOV_USEC_TIMEOUT 1200000 /* wait 12 * 100ms for SRIOV */ #define MAX_KIQ_REG_WAIT 5000 /* in usecs, 5ms */ #define MAX_KIQ_REG_BAILOUT_INTERVAL 5 /* in msecs, 5ms */ #define MAX_KIQ_REG_TRY 1000 @@ -1119,6 +1118,7 @@ struct amdgpu_device { bool in_s3; bool in_s4; bool in_s0ix; + suspend_state_t last_suspend_state; enum pp_mp1_state mp1_state; struct amdgpu_doorbell_index doorbell_index; @@ -1187,9 +1187,15 @@ struct amdgpu_device { bool debug_enable_ras_aca; bool debug_exp_resets; - bool enforce_isolation[MAX_XCP]; - /* Added this mutex for cleaner shader isolation between GFX and compute processes */ + /* Protection for the following isolation structure */ struct mutex enforce_isolation_mutex; + bool enforce_isolation[MAX_XCP]; + struct amdgpu_isolation { + void *owner; + struct dma_fence *spearhead; + struct amdgpu_sync active; + struct amdgpu_sync prev; + } isolation[MAX_XCP]; struct amdgpu_init_level *init_lvl; }; @@ -1470,6 +1476,9 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev, struct dma_fence *amdgpu_device_get_gang(struct amdgpu_device *adev); struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev, struct dma_fence *gang); +struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_job *job); bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev); ssize_t amdgpu_get_soft_full_reset_mask(struct amdgpu_ring *ring); ssize_t amdgpu_show_reset_mask(char *buf, uint32_t supported_reset); @@ -1593,11 +1602,9 @@ static inline void amdgpu_acpi_get_backlight_caps(struct amdgpu_dm_backlight_cap #if defined(CONFIG_ACPI) && defined(CONFIG_SUSPEND) bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev); bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev); -void amdgpu_choose_low_power_state(struct amdgpu_device *adev); #else static inline bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) { return false; } static inline bool amdgpu_acpi_is_s3_active(struct amdgpu_device *adev) { return false; } -static inline void amdgpu_choose_low_power_state(struct amdgpu_device *adev) { } #endif void amdgpu_register_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 b8d4e07d2043..bebfbc1497d8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c @@ -1533,22 +1533,4 @@ bool amdgpu_acpi_is_s0ix_active(struct amdgpu_device *adev) #endif /* CONFIG_AMD_PMC */ } -/** - * amdgpu_choose_low_power_state - * - * @adev: amdgpu_device_pointer - * - * Choose the target low power state for the GPU - */ -void amdgpu_choose_low_power_state(struct amdgpu_device *adev) -{ - if (adev->in_runpm) - return; - - if (amdgpu_acpi_is_s0ix_active(adev)) - adev->in_s0ix = true; - else if (amdgpu_acpi_is_s3_active(adev)) - adev->in_s3 = true; -} - #endif /* CONFIG_SUSPEND */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h index 8af67f18500a..55d539967695 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h @@ -47,6 +47,7 @@ enum TLB_FLUSH_TYPE { }; struct amdgpu_device; +struct kfd_process_device; struct amdgpu_reset_context; enum kfd_mem_attachment_type { @@ -192,7 +193,7 @@ int kfd_debugfs_kfd_mem_limits(struct seq_file *m, void *data); #if IS_ENABLED(CONFIG_HSA_AMD) bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm); struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f); -int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo); +void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo); int amdgpu_amdkfd_evict_userptr(struct mmu_interval_notifier *mni, unsigned long cur_seq, struct kgd_mem *mem); int amdgpu_amdkfd_bo_validate_and_fence(struct amdgpu_bo *bo, @@ -212,9 +213,8 @@ struct amdgpu_amdkfd_fence *to_amdgpu_amdkfd_fence(struct dma_fence *f) } static inline -int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo) +void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo) { - return 0; } static inline @@ -299,14 +299,10 @@ bool amdgpu_amdkfd_compute_active(struct amdgpu_device *adev, uint32_t node_id); (&((struct amdgpu_fpriv *) \ ((struct drm_file *)(drm_priv))->driver_priv)->vm) -int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev, - struct amdgpu_vm *avm, u32 pasid); int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, struct amdgpu_vm *avm, void **process_info, struct dma_fence **ef); -void amdgpu_amdkfd_gpuvm_release_process_vm(struct amdgpu_device *adev, - void *drm_priv); uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv); size_t amdgpu_amdkfd_get_available_memory(struct amdgpu_device *adev, uint8_t xcp_id); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c index 1e998f972c30..b3c8eae46042 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c @@ -370,40 +370,32 @@ static int amdgpu_amdkfd_remove_eviction_fence(struct amdgpu_bo *bo, return 0; } -int amdgpu_amdkfd_remove_fence_on_pt_pd_bos(struct amdgpu_bo *bo) +/** + * amdgpu_amdkfd_remove_all_eviction_fences - Remove all eviction fences + * @bo: the BO where to remove the evictions fences from. + * + * This functions should only be used on release when all references to the BO + * are already dropped. We remove the eviction fence from the private copy of + * the dma_resv object here since that is what is used during release to + * determine of the BO is idle or not. + */ +void amdgpu_amdkfd_remove_all_eviction_fences(struct amdgpu_bo *bo) { - struct amdgpu_bo *root = bo; - struct amdgpu_vm_bo_base *vm_bo; - struct amdgpu_vm *vm; - struct amdkfd_process_info *info; - struct amdgpu_amdkfd_fence *ef; - int ret; + struct dma_resv *resv = &bo->tbo.base._resv; + struct dma_fence *fence, *stub; + struct dma_resv_iter cursor; - /* we can always get vm_bo from root PD bo.*/ - while (root->parent) - root = root->parent; + dma_resv_assert_held(resv); - vm_bo = root->vm_bo; - if (!vm_bo) - return 0; - - vm = vm_bo->vm; - if (!vm) - return 0; - - info = vm->process_info; - if (!info || !info->eviction_fence) - return 0; - - ef = container_of(dma_fence_get(&info->eviction_fence->base), - struct amdgpu_amdkfd_fence, base); - - BUG_ON(!dma_resv_trylock(bo->tbo.base.resv)); - ret = amdgpu_amdkfd_remove_eviction_fence(bo, ef); - dma_resv_unlock(bo->tbo.base.resv); + stub = dma_fence_get_stub(); + dma_resv_for_each_fence(&cursor, resv, DMA_RESV_USAGE_BOOKKEEP, fence) { + if (!to_amdgpu_amdkfd_fence(fence)) + continue; - dma_fence_put(&ef->base); - return ret; + dma_resv_replace_fences(resv, fence->context, stub, + DMA_RESV_USAGE_BOOKKEEP); + } + dma_fence_put(stub); } static int amdgpu_amdkfd_bo_validate(struct amdgpu_bo *bo, uint32_t domain, @@ -499,7 +491,7 @@ static int vm_update_pds(struct amdgpu_vm *vm, struct amdgpu_sync *sync) if (ret) return ret; - return amdgpu_sync_fence(sync, vm->last_update); + return amdgpu_sync_fence(sync, vm->last_update, GFP_KERNEL); } static uint64_t get_pte_flags(struct amdgpu_device *adev, struct kgd_mem *mem) @@ -1263,7 +1255,7 @@ static int unmap_bo_from_gpuvm(struct kgd_mem *mem, (void)amdgpu_vm_clear_freed(adev, vm, &bo_va->last_pt_update); - (void)amdgpu_sync_fence(sync, bo_va->last_pt_update); + (void)amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL); return 0; } @@ -1287,7 +1279,7 @@ static int update_gpuvm_pte(struct kgd_mem *mem, return ret; } - return amdgpu_sync_fence(sync, bo_va->last_pt_update); + return amdgpu_sync_fence(sync, bo_va->last_pt_update, GFP_KERNEL); } static int map_bo_to_gpuvm(struct kgd_mem *mem, @@ -1529,27 +1521,6 @@ static void amdgpu_amdkfd_gpuvm_unpin_bo(struct amdgpu_bo *bo) amdgpu_bo_unreserve(bo); } -int amdgpu_amdkfd_gpuvm_set_vm_pasid(struct amdgpu_device *adev, - struct amdgpu_vm *avm, u32 pasid) - -{ - int ret; - - /* Free the original amdgpu allocated pasid, - * will be replaced with kfd allocated pasid. - */ - if (avm->pasid) { - amdgpu_pasid_free(avm->pasid); - amdgpu_vm_set_pasid(adev, avm, 0); - } - - ret = amdgpu_vm_set_pasid(adev, avm, pasid); - if (ret) - return ret; - - return 0; -} - int amdgpu_amdkfd_gpuvm_acquire_process_vm(struct amdgpu_device *adev, struct amdgpu_vm *avm, void **process_info, @@ -1607,27 +1578,6 @@ void amdgpu_amdkfd_gpuvm_destroy_cb(struct amdgpu_device *adev, } } -void amdgpu_amdkfd_gpuvm_release_process_vm(struct amdgpu_device *adev, - void *drm_priv) -{ - struct amdgpu_vm *avm; - - if (WARN_ON(!adev || !drm_priv)) - return; - - avm = drm_priv_to_vm(drm_priv); - - pr_debug("Releasing process vm %p\n", avm); - - /* The original pasid of amdgpu vm has already been - * released during making a amdgpu vm to a compute vm - * The current pasid is managed by kfd and will be - * released on kfd process destroy. Set amdgpu pasid - * to 0 to avoid duplicate release. - */ - amdgpu_vm_release_compute(adev, avm); -} - uint64_t amdgpu_amdkfd_gpuvm_get_process_page_dir(void *drm_priv) { struct amdgpu_vm *avm = drm_priv_to_vm(drm_priv); @@ -2969,7 +2919,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence __rcu * } dma_resv_for_each_fence(&cursor, bo->tbo.base.resv, DMA_RESV_USAGE_KERNEL, fence) { - ret = amdgpu_sync_fence(&sync_obj, fence); + ret = amdgpu_sync_fence(&sync_obj, fence, GFP_KERNEL); if (ret) { pr_debug("Memory eviction: Sync BO fence failed. Try again\n"); goto validate_map_fail; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c index 5cc5f59e3018..4a5b406601fa 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c @@ -428,7 +428,7 @@ static int amdgpu_cs_p2_dependencies(struct amdgpu_cs_parser *p, dma_fence_put(old); } - r = amdgpu_sync_fence(&p->sync, fence); + r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL); dma_fence_put(fence); if (r) return r; @@ -450,7 +450,7 @@ static int amdgpu_syncobj_lookup_and_add(struct amdgpu_cs_parser *p, return r; } - r = amdgpu_sync_fence(&p->sync, fence); + r = amdgpu_sync_fence(&p->sync, fence, GFP_KERNEL); dma_fence_put(fence); return r; } @@ -1124,7 +1124,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p) if (r) return r; - r = amdgpu_sync_fence(&p->sync, fpriv->prt_va->last_pt_update); + r = amdgpu_sync_fence(&p->sync, fpriv->prt_va->last_pt_update, + GFP_KERNEL); if (r) return r; @@ -1135,7 +1136,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p) if (r) return r; - r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update); + r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update, + GFP_KERNEL); if (r) return r; } @@ -1154,7 +1156,8 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p) if (r) return r; - r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update); + r = amdgpu_sync_fence(&p->sync, bo_va->last_pt_update, + GFP_KERNEL); if (r) return r; } @@ -1167,7 +1170,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p) if (r) return r; - r = amdgpu_sync_fence(&p->sync, vm->last_update); + r = amdgpu_sync_fence(&p->sync, vm->last_update, GFP_KERNEL); if (r) return r; @@ -1248,7 +1251,8 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p) continue; } - r = amdgpu_sync_fence(&p->gang_leader->explicit_sync, fence); + r = amdgpu_sync_fence(&p->gang_leader->explicit_sync, fence, + GFP_KERNEL); dma_fence_put(fence); if (r) return r; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c index cfdf558b48b6..02138aa55793 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c @@ -109,7 +109,7 @@ int amdgpu_unmap_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm, struct drm_exec exec; int r; - 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_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 71e8a76180ad..28190b0ac5fc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -224,6 +224,24 @@ static ssize_t amdgpu_device_get_pcie_replay_count(struct device *dev, static DEVICE_ATTR(pcie_replay_count, 0444, amdgpu_device_get_pcie_replay_count, NULL); +static int amdgpu_device_attr_sysfs_init(struct amdgpu_device *adev) +{ + int ret = 0; + + if (!amdgpu_sriov_vf(adev)) + ret = sysfs_create_file(&adev->dev->kobj, + &dev_attr_pcie_replay_count.attr); + + return ret; +} + +static void amdgpu_device_attr_sysfs_fini(struct amdgpu_device *adev) +{ + if (!amdgpu_sriov_vf(adev)) + sysfs_remove_file(&adev->dev->kobj, + &dev_attr_pcie_replay_count.attr); +} + static ssize_t amdgpu_sysfs_reg_state_get(struct file *f, struct kobject *kobj, struct bin_attribute *attr, char *buf, loff_t ppos, size_t count) @@ -4123,11 +4141,6 @@ static bool amdgpu_device_check_iommu_remap(struct amdgpu_device *adev) } #endif -static const struct attribute *amdgpu_dev_attributes[] = { - &dev_attr_pcie_replay_count.attr, - NULL -}; - static void amdgpu_device_set_mcbp(struct amdgpu_device *adev) { if (amdgpu_mcbp == 1) @@ -4232,6 +4245,11 @@ int amdgpu_device_init(struct amdgpu_device *adev, mutex_init(&adev->gfx.reset_sem_mutex); /* Initialize the mutex for cleaner shader isolation between GFX and compute processes */ mutex_init(&adev->enforce_isolation_mutex); + for (i = 0; i < MAX_XCP; ++i) { + adev->isolation[i].spearhead = dma_fence_get_stub(); + amdgpu_sync_create(&adev->isolation[i].active); + amdgpu_sync_create(&adev->isolation[i].prev); + } mutex_init(&adev->gfx.kfd_sch_mutex); amdgpu_device_init_apu_flags(adev); @@ -4352,10 +4370,17 @@ int amdgpu_device_init(struct amdgpu_device *adev, if (r) return r; - /* Get rid of things like offb */ - r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name); - if (r) - return r; + /* + * No need to remove conflicting FBs for non-display class devices. + * This prevents the sysfb from being freed accidently. + */ + if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA || + (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) { + /* Get rid of things like offb */ + r = aperture_remove_conflicting_pci_devices(adev->pdev, amdgpu_kms_driver.name); + if (r) + return r; + } /* Enable TMZ based on IP_VERSION */ amdgpu_gmc_tmz_set(adev); @@ -4567,7 +4592,7 @@ fence_driver_init: } else adev->ucode_sysfs_en = true; - r = sysfs_create_files(&adev->dev->kobj, amdgpu_dev_attributes); + r = amdgpu_device_attr_sysfs_init(adev); if (r) dev_err(adev->dev, "Could not create amdgpu device attr\n"); @@ -4704,7 +4729,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) amdgpu_pm_sysfs_fini(adev); if (adev->ucode_sysfs_en) amdgpu_ucode_sysfs_fini(adev); - sysfs_remove_files(&adev->dev->kobj, amdgpu_dev_attributes); + amdgpu_device_attr_sysfs_fini(adev); amdgpu_fru_sysfs_fini(adev); amdgpu_reg_state_sysfs_fini(adev); @@ -4731,7 +4756,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) void amdgpu_device_fini_sw(struct amdgpu_device *adev) { - int idx; + int i, idx; bool px; amdgpu_device_ip_fini(adev); @@ -4739,6 +4764,11 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev) amdgpu_ucode_release(&adev->firmware.gpu_info_fw); adev->accel_working = false; dma_fence_put(rcu_dereference_protected(adev->gang_submit, true)); + for (i = 0; i < MAX_XCP; ++i) { + dma_fence_put(adev->isolation[i].spearhead); + amdgpu_sync_free(&adev->isolation[i].active); + amdgpu_sync_free(&adev->isolation[i].prev); + } amdgpu_reset_fini(adev); @@ -4755,6 +4785,9 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev) kfree(adev->fru_info); adev->fru_info = NULL; + kfree(adev->xcp_mgr); + adev->xcp_mgr = NULL; + px = amdgpu_device_supports_px(adev_to_drm(adev)); if (px || (!dev_is_removable(&adev->pdev->dev) && @@ -4819,28 +4852,20 @@ static int amdgpu_device_evict_resources(struct amdgpu_device *adev) * @data: data * * This function is called when the system is about to suspend or hibernate. - * It is used to evict resources from the device before the system goes to - * sleep while there is still access to swap. + * It is used to set the appropriate flags so that eviction can be optimized + * in the pm prepare callback. */ static int amdgpu_device_pm_notifier(struct notifier_block *nb, unsigned long mode, void *data) { struct amdgpu_device *adev = container_of(nb, struct amdgpu_device, pm_nb); - int r; switch (mode) { case PM_HIBERNATION_PREPARE: adev->in_s4 = true; - fallthrough; - case PM_SUSPEND_PREPARE: - r = amdgpu_device_evict_resources(adev); - /* - * This is considered non-fatal at this time because - * amdgpu_device_prepare() will also fatally evict resources. - * See https://gitlab.freedesktop.org/drm/amd/-/issues/3781 - */ - if (r) - drm_warn(adev_to_drm(adev), "Failed to evict resources, freeze active processes if problems occur: %d\n", r); + break; + case PM_POST_HIBERNATION: + adev->in_s4 = false; break; } @@ -4861,15 +4886,13 @@ int amdgpu_device_prepare(struct drm_device *dev) struct amdgpu_device *adev = drm_to_adev(dev); int i, r; - amdgpu_choose_low_power_state(adev); - if (dev->switch_power_state == DRM_SWITCH_POWER_OFF) return 0; /* Evict the majority of BOs before starting suspend sequence */ r = amdgpu_device_evict_resources(adev); if (r) - goto unprepare; + return r; flush_delayed_work(&adev->gfx.gfx_off_delay_work); @@ -4880,15 +4903,10 @@ int amdgpu_device_prepare(struct drm_device *dev) continue; r = adev->ip_blocks[i].version->funcs->prepare_suspend(&adev->ip_blocks[i]); if (r) - goto unprepare; + return r; } return 0; - -unprepare: - adev->in_s0ix = adev->in_s3 = adev->in_s4 = false; - - return r; } /** @@ -6875,6 +6893,92 @@ struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev, return NULL; } +/** + * amdgpu_device_enforce_isolation - enforce HW isolation + * @adev: the amdgpu device pointer + * @ring: the HW ring the job is supposed to run on + * @job: the job which is about to be pushed to the HW ring + * + * Makes sure that only one client at a time can use the GFX block. + * Returns: The dependency to wait on before the job can be pushed to the HW. + * The function is called multiple times until NULL is returned. + */ +struct dma_fence *amdgpu_device_enforce_isolation(struct amdgpu_device *adev, + struct amdgpu_ring *ring, + struct amdgpu_job *job) +{ + struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id]; + struct drm_sched_fence *f = job->base.s_fence; + struct dma_fence *dep; + void *owner; + int r; + + /* + * For now enforce isolation only for the GFX block since we only need + * the cleaner shader on those rings. + */ + if (ring->funcs->type != AMDGPU_RING_TYPE_GFX && + ring->funcs->type != AMDGPU_RING_TYPE_COMPUTE) + return NULL; + + /* + * All submissions where enforce isolation is false are handled as if + * they come from a single client. Use ~0l as the owner to distinct it + * from kernel submissions where the owner is NULL. + */ + owner = job->enforce_isolation ? f->owner : (void *)~0l; + + mutex_lock(&adev->enforce_isolation_mutex); + + /* + * The "spearhead" submission is the first one which changes the + * ownership to its client. We always need to wait for it to be + * pushed to the HW before proceeding with anything. + */ + if (&f->scheduled != isolation->spearhead && + !dma_fence_is_signaled(isolation->spearhead)) { + dep = isolation->spearhead; + goto out_grab_ref; + } + + if (isolation->owner != owner) { + + /* + * Wait for any gang to be assembled before switching to a + * different owner or otherwise we could deadlock the + * submissions. + */ + if (!job->gang_submit) { + dep = amdgpu_device_get_gang(adev); + if (!dma_fence_is_signaled(dep)) + goto out_return_dep; + dma_fence_put(dep); + } + + dma_fence_put(isolation->spearhead); + isolation->spearhead = dma_fence_get(&f->scheduled); + amdgpu_sync_move(&isolation->active, &isolation->prev); + isolation->owner = owner; + } + + /* + * Specifying the ring here helps to pipeline submissions even when + * isolation is enabled. If that is not desired for testing NULL can be + * used instead of the ring to enforce a CPU round trip while switching + * between clients. + */ + dep = amdgpu_sync_peek_fence(&isolation->prev, ring); + r = amdgpu_sync_fence(&isolation->active, &f->finished, GFP_NOWAIT); + if (r) + DRM_WARN("OOM tracking isolation\n"); + +out_grab_ref: + dma_fence_get(dep); +out_return_dep: + mutex_unlock(&adev->enforce_isolation_mutex); + return dep; +} + bool amdgpu_device_has_display_hardware(struct amdgpu_device *adev) { switch (adev->asic_type) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c index 949d74eff294..6a6dc15273dc 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c @@ -113,8 +113,7 @@ #include "amdgpu_isp.h" #endif -#define FIRMWARE_IP_DISCOVERY "amdgpu/ip_discovery.bin" -MODULE_FIRMWARE(FIRMWARE_IP_DISCOVERY); +MODULE_FIRMWARE("amdgpu/ip_discovery.bin"); #define mmIP_DISCOVERY_VERSION 0x16A00 #define mmRCC_CONFIG_MEMSIZE 0xde3 @@ -297,21 +296,13 @@ static int amdgpu_discovery_read_binary_from_mem(struct amdgpu_device *adev, return ret; } -static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, uint8_t *binary) +static int amdgpu_discovery_read_binary_from_file(struct amdgpu_device *adev, + uint8_t *binary, + const char *fw_name) { const struct firmware *fw; - const char *fw_name; int r; - switch (amdgpu_discovery) { - case 2: - fw_name = FIRMWARE_IP_DISCOVERY; - break; - default: - dev_warn(adev->dev, "amdgpu_discovery is not set properly\n"); - return -EINVAL; - } - r = request_firmware(&fw, fw_name, adev->dev); if (r) { dev_err(adev->dev, "can't load firmware \"%s\"\n", @@ -404,10 +395,19 @@ static int amdgpu_discovery_verify_npsinfo(struct amdgpu_device *adev, return 0; } +static const char *amdgpu_discovery_get_fw_name(struct amdgpu_device *adev) +{ + if (amdgpu_discovery == 2) + return "amdgpu/ip_discovery.bin"; + + return NULL; +} + static int amdgpu_discovery_init(struct amdgpu_device *adev) { struct table_info *info; struct binary_header *bhdr; + const char *fw_name; uint16_t offset; uint16_t size; uint16_t checksum; @@ -419,9 +419,10 @@ static int amdgpu_discovery_init(struct amdgpu_device *adev) return -ENOMEM; /* Read from file if it is the preferred option */ - if (amdgpu_discovery == 2) { + fw_name = amdgpu_discovery_get_fw_name(adev); + if (fw_name != NULL) { dev_info(adev->dev, "use ip discovery information from file"); - r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin); + r = amdgpu_discovery_read_binary_from_file(adev, adev->mman.discovery_bin, fw_name); if (r) { dev_err(adev->dev, "failed to read ip discovery binary from file\n"); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c index c9842a0e2a1c..cb043296f9ae 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_dma_buf.c @@ -43,6 +43,29 @@ #include <linux/dma-fence-array.h> #include <linux/pci-p2pdma.h> +static const struct dma_buf_attach_ops amdgpu_dma_buf_attach_ops; + +/** + * dma_buf_attach_adev - Helper to get adev of an attachment + * + * @attach: attachment + * + * Returns: + * A struct amdgpu_device * if the attaching device is an amdgpu device or + * partition, NULL otherwise. + */ +static struct amdgpu_device *dma_buf_attach_adev(struct dma_buf_attachment *attach) +{ + if (attach->importer_ops == &amdgpu_dma_buf_attach_ops) { + struct drm_gem_object *obj = attach->importer_priv; + struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); + + return amdgpu_ttm_adev(bo->tbo.bdev); + } + + return NULL; +} + /** * amdgpu_dma_buf_attach - &dma_buf_ops.attach implementation * @@ -54,11 +77,13 @@ static int amdgpu_dma_buf_attach(struct dma_buf *dmabuf, struct dma_buf_attachment *attach) { + struct amdgpu_device *attach_adev = dma_buf_attach_adev(attach); struct drm_gem_object *obj = dmabuf->priv; struct amdgpu_bo *bo = gem_to_amdgpu_bo(obj); struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev); - if (pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0) + if (!amdgpu_dmabuf_is_xgmi_accessible(attach_adev, bo) && + pci_p2pdma_distance(adev->pdev, attach->dev, false) < 0) attach->peer2peer = false; amdgpu_vm_bo_update_shared(bo); @@ -459,6 +484,9 @@ bool amdgpu_dmabuf_is_xgmi_accessible(struct amdgpu_device *adev, struct drm_gem_object *obj = &bo->tbo.base; struct drm_gem_object *gobj; + if (!adev) + return false; + if (obj->import_attach) { struct dma_buf *dma_buf = obj->import_attach->dmabuf; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c index 24c255e05079..e4ce33e69a48 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c @@ -173,6 +173,7 @@ uint amdgpu_sdma_phase_quantum = 32; char *amdgpu_disable_cu; char *amdgpu_virtual_display; bool enforce_isolation; +int amdgpu_modeset = -1; /* Specifies the default granularity for SVM, used in buffer * migration and restoration of backing memory when handling @@ -1034,6 +1035,13 @@ module_param(enforce_isolation, bool, 0444); MODULE_PARM_DESC(enforce_isolation, "enforce process isolation between graphics and compute . enforce_isolation = on"); /** + * DOC: modeset (int) + * Override nomodeset (1 = override, -1 = auto). The default is -1 (auto). + */ +MODULE_PARM_DESC(modeset, "Override nomodeset (1 = enable, -1 = auto)"); +module_param_named(modeset, amdgpu_modeset, int, 0444); + +/** * DOC: seamless (int) * Seamless boot will keep the image on the screen during the boot process. */ @@ -2244,6 +2252,12 @@ static int amdgpu_pci_probe(struct pci_dev *pdev, int ret, retry = 0, i; bool supports_atomic = false; + if ((pdev->class >> 8) == PCI_CLASS_DISPLAY_VGA || + (pdev->class >> 8) == PCI_CLASS_DISPLAY_OTHER) { + if (drm_firmware_drivers_only() && amdgpu_modeset == -1) + return -EINVAL; + } + /* skip devices which are owned by radeon */ for (i = 0; i < ARRAY_SIZE(amdgpu_unsupported_pciidlist); i++) { if (amdgpu_unsupported_pciidlist[i] == pdev->device) @@ -2515,8 +2529,20 @@ static int amdgpu_pmops_suspend(struct device *dev) adev->in_s0ix = true; else if (amdgpu_acpi_is_s3_active(adev)) adev->in_s3 = true; - if (!adev->in_s0ix && !adev->in_s3) + if (!adev->in_s0ix && !adev->in_s3) { + /* don't allow going deep first time followed by s2idle the next time */ + if (adev->last_suspend_state != PM_SUSPEND_ON && + adev->last_suspend_state != pm_suspend_target_state) { + drm_err_once(drm_dev, "Unsupported suspend state %d\n", + pm_suspend_target_state); + return -EINVAL; + } return 0; + } + + /* cache the state last used for suspend */ + adev->last_suspend_state = pm_suspend_target_state; + return amdgpu_device_suspend(drm_dev, true); } @@ -2570,13 +2596,8 @@ static int amdgpu_pmops_freeze(struct device *dev) static int amdgpu_pmops_thaw(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); - struct amdgpu_device *adev = drm_to_adev(drm_dev); - int r; - - r = amdgpu_device_resume(drm_dev, true); - adev->in_s4 = false; - return r; + return amdgpu_device_resume(drm_dev, true); } static int amdgpu_pmops_poweroff(struct device *dev) @@ -2589,9 +2610,6 @@ static int amdgpu_pmops_poweroff(struct device *dev) static int amdgpu_pmops_restore(struct device *dev) { struct drm_device *drm_dev = dev_get_drvdata(dev); - struct amdgpu_device *adev = drm_to_adev(drm_dev); - - adev->in_s4 = false; return amdgpu_device_resume(drm_dev, true); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c index c1f35ded684e..506786784e32 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.c @@ -1411,9 +1411,11 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring) struct amdgpu_device *adev = ring->adev; struct drm_gpu_scheduler *sched = &ring->sched; struct drm_sched_entity entity; + static atomic_t counter; struct dma_fence *f; struct amdgpu_job *job; struct amdgpu_ib *ib; + void *owner; int i, r; /* Initialize the scheduler entity */ @@ -1424,9 +1426,15 @@ static int amdgpu_gfx_run_cleaner_shader_job(struct amdgpu_ring *ring) goto err; } - r = amdgpu_job_alloc_with_ib(ring->adev, &entity, NULL, - 64, 0, - &job); + /* + * Use some unique dummy value as the owner to make sure we execute + * the cleaner shader on each submission. The value just need to change + * for each submission and is otherwise meaningless. + */ + owner = (void *)(unsigned long)atomic_inc_return(&counter); + + r = amdgpu_job_alloc_with_ib(ring->adev, &entity, owner, + 64, 0, &job); if (r) goto err; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c index 1c19a65e6553..ef74259c448d 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.c @@ -678,12 +678,10 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid, uint32_t flush_type, bool all_hub, uint32_t inst) { - u32 usec_timeout = amdgpu_sriov_vf(adev) ? SRIOV_USEC_TIMEOUT : - adev->usec_timeout; struct amdgpu_ring *ring = &adev->gfx.kiq[inst].ring; struct amdgpu_kiq *kiq = &adev->gfx.kiq[inst]; unsigned int ndw; - int r; + int r, cnt = 0; uint32_t seq; /* @@ -740,10 +738,21 @@ int amdgpu_gmc_flush_gpu_tlb_pasid(struct amdgpu_device *adev, uint16_t pasid, amdgpu_ring_commit(ring); spin_unlock(&adev->gfx.kiq[inst].ring_lock); - if (amdgpu_fence_wait_polling(ring, seq, usec_timeout) < 1) { + + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + + might_sleep(); + while (r < 1 && cnt++ < MAX_KIQ_REG_TRY && + !amdgpu_reset_pending(adev->reset_domain)) { + msleep(MAX_KIQ_REG_BAILOUT_INTERVAL); + r = amdgpu_fence_wait_polling(ring, seq, MAX_KIQ_REG_WAIT); + } + + if (cnt > MAX_KIQ_REG_TRY) { dev_err(adev->dev, "timeout waiting for kiq fence\n"); r = -ETIME; - } + } else + r = 0; } error_unlock_reset: diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c index 8e712a11aba5..92ab821afc06 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ids.c @@ -209,7 +209,7 @@ static int amdgpu_vmid_grab_idle(struct amdgpu_ring *ring, return 0; } - fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_KERNEL); + fences = kmalloc_array(id_mgr->num_ids, sizeof(void *), GFP_NOWAIT); if (!fences) return -ENOMEM; @@ -287,46 +287,34 @@ static int amdgpu_vmid_grab_reserved(struct amdgpu_vm *vm, (*id)->flushed_updates < updates || !(*id)->last_flush || ((*id)->last_flush->context != fence_context && - !dma_fence_is_signaled((*id)->last_flush))) { + !dma_fence_is_signaled((*id)->last_flush))) + needs_flush = true; + + if ((*id)->owner != vm->immediate.fence_context || + (!adev->vm_manager.concurrent_flush && needs_flush)) { struct dma_fence *tmp; - /* Wait for the gang to be assembled before using a - * reserved VMID or otherwise the gang could deadlock. + /* Don't use per engine and per process VMID at the + * same time */ - tmp = amdgpu_device_get_gang(adev); - if (!dma_fence_is_signaled(tmp) && tmp != job->gang_submit) { + if (adev->vm_manager.concurrent_flush) + ring = NULL; + + /* to prevent one context starved by another context */ + (*id)->pd_gpu_addr = 0; + tmp = amdgpu_sync_peek_fence(&(*id)->active, ring); + if (tmp) { *id = NULL; - *fence = tmp; + *fence = dma_fence_get(tmp); return 0; } - dma_fence_put(tmp); - - /* Make sure the id is owned by the gang before proceeding */ - if (!job->gang_submit || - (*id)->owner != vm->immediate.fence_context) { - - /* Don't use per engine and per process VMID at the - * same time - */ - if (adev->vm_manager.concurrent_flush) - ring = NULL; - - /* to prevent one context starved by another context */ - (*id)->pd_gpu_addr = 0; - tmp = amdgpu_sync_peek_fence(&(*id)->active, ring); - if (tmp) { - *id = NULL; - *fence = dma_fence_get(tmp); - return 0; - } - } - needs_flush = true; } /* Good we can use this VMID. Remember this submission as * user of the VMID. */ - r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished); + r = amdgpu_sync_fence(&(*id)->active, &job->base.s_fence->finished, + GFP_NOWAIT); if (r) return r; @@ -385,7 +373,8 @@ static int amdgpu_vmid_grab_used(struct amdgpu_vm *vm, * user of the VMID. */ r = amdgpu_sync_fence(&(*id)->active, - &job->base.s_fence->finished); + &job->base.s_fence->finished, + GFP_NOWAIT); if (r) return r; @@ -437,7 +426,8 @@ int amdgpu_vmid_grab(struct amdgpu_vm *vm, struct amdgpu_ring *ring, /* Remember this submission as user of the VMID */ r = amdgpu_sync_fence(&id->active, - &job->base.s_fence->finished); + &job->base.s_fence->finished, + GFP_NOWAIT); if (r) goto error; diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h index 7d4395a5d8ac..b0a88f92cd82 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h @@ -78,6 +78,9 @@ struct amdgpu_ih_ring { #define amdgpu_ih_ts_after(t1, t2) \ (((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) > 0LL) +#define amdgpu_ih_ts_after_or_equal(t1, t2) \ + (((int64_t)((t2) << 16) - (int64_t)((t1) << 16)) >= 0LL) + /* provided by the ih block */ struct amdgpu_ih_funcs { /* ring read/write ptr handling, called from interrupt context */ diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 100f04475943..685c61a05af8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -342,17 +342,24 @@ amdgpu_job_prepare_job(struct drm_sched_job *sched_job, { struct amdgpu_ring *ring = to_amdgpu_ring(s_entity->rq->sched); struct amdgpu_job *job = to_amdgpu_job(sched_job); - struct dma_fence *fence = NULL; + struct dma_fence *fence; int r; r = drm_sched_entity_error(s_entity); if (r) goto error; - if (job->gang_submit) + if (job->gang_submit) { fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit); + if (fence) + return fence; + } + + fence = amdgpu_device_enforce_isolation(ring->adev, ring, job); + if (fence) + return fence; - if (!fence && job->vm && !job->vmid) { + if (job->vm && !job->vmid) { r = amdgpu_vmid_grab(job->vm, ring, job, &fence); if (r) { dev_err(ring->adev->dev, "Error getting VM ID (%d)\n", r); @@ -365,9 +372,10 @@ amdgpu_job_prepare_job(struct drm_sched_job *sched_job, */ if (!fence) job->vm = NULL; + return fence; } - return fence; + return NULL; error: dma_fence_set_error(&job->base.s_fence->finished, r); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c index 6fa20980a0b1..e4251d0691c9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_mes.c @@ -1335,14 +1335,14 @@ int amdgpu_mes_ctx_map_meta_data(struct amdgpu_device *adev, DRM_ERROR("failed to do vm_bo_update on meta data\n"); goto error_del_bo_va; } - amdgpu_sync_fence(&sync, bo_va->last_pt_update); + amdgpu_sync_fence(&sync, bo_va->last_pt_update, GFP_KERNEL); r = amdgpu_vm_update_pdes(adev, vm, false); if (r) { DRM_ERROR("failed to update pdes on meta data\n"); goto error_del_bo_va; } - amdgpu_sync_fence(&sync, vm->last_update); + amdgpu_sync_fence(&sync, vm->last_update, GFP_KERNEL); amdgpu_sync_wait(&sync, false); drm_exec_fini(&exec); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c index 00752e3f9d8a..0b9987781f76 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c @@ -1295,28 +1295,36 @@ void amdgpu_bo_release_notify(struct ttm_buffer_object *bo) if (abo->kfd_bo) amdgpu_amdkfd_release_notify(abo); - /* We only remove the fence if the resv has individualized. */ - WARN_ON_ONCE(bo->type == ttm_bo_type_kernel - && bo->base.resv != &bo->base._resv); - if (bo->base.resv == &bo->base._resv) - amdgpu_amdkfd_remove_fence_on_pt_pd_bos(abo); + /* + * We lock the private dma_resv object here and since the BO is about to + * be released nobody else should have a pointer to it. + * So when this locking here fails something is wrong with the reference + * counting. + */ + if (WARN_ON_ONCE(!dma_resv_trylock(&bo->base._resv))) + return; + + amdgpu_amdkfd_remove_all_eviction_fences(abo); if (!bo->resource || bo->resource->mem_type != TTM_PL_VRAM || !(abo->flags & AMDGPU_GEM_CREATE_VRAM_WIPE_ON_RELEASE) || adev->in_suspend || drm_dev_is_unplugged(adev_to_drm(adev))) - return; + goto out; - if (WARN_ON_ONCE(!dma_resv_trylock(bo->base.resv))) - return; + r = dma_resv_reserve_fences(&bo->base._resv, 1); + if (r) + goto out; - r = amdgpu_fill_buffer(abo, 0, bo->base.resv, &fence, true); - if (!WARN_ON(r)) { - amdgpu_vram_mgr_set_cleared(bo->resource); - amdgpu_bo_fence(abo, fence, false); - dma_fence_put(fence); - } + r = amdgpu_fill_buffer(abo, 0, &bo->base._resv, &fence, true); + if (WARN_ON(r)) + goto out; + + amdgpu_vram_mgr_set_cleared(bo->resource); + dma_resv_add_fence(&bo->base._resv, fence, DMA_RESV_USAGE_KERNEL); + dma_fence_put(fence); - dma_resv_unlock(bo->base.resv); +out: + dma_resv_unlock(&bo->base._resv); } /** diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c index e5fc80ed06ea..6dded11a23ac 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c @@ -44,7 +44,7 @@ #include "amdgpu_securedisplay.h" #include "amdgpu_atomfirmware.h" -#define AMD_VBIOS_FILE_MAX_SIZE_B (1024*1024*3) +#define AMD_VBIOS_FILE_MAX_SIZE_B (1024*1024*16) static int psp_load_smu_fw(struct psp_context *psp); static int psp_rap_terminate(struct psp_context *psp); @@ -533,7 +533,6 @@ static int psp_sw_fini(struct amdgpu_ip_block *ip_block) { struct amdgpu_device *adev = ip_block->adev; struct psp_context *psp = &adev->psp; - struct psp_gfx_cmd_resp *cmd = psp->cmd; psp_memory_training_fini(psp); @@ -543,8 +542,8 @@ static int psp_sw_fini(struct amdgpu_ip_block *ip_block) amdgpu_ucode_release(&psp->cap_fw); amdgpu_ucode_release(&psp->toc_fw); - kfree(cmd); - cmd = NULL; + kfree(psp->cmd); + psp->cmd = NULL; psp_free_shared_bufs(psp); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c index f0924aa3f4e4..0c338dcdde48 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c @@ -1864,6 +1864,9 @@ int amdgpu_ras_sysfs_create(struct amdgpu_device *adev, if (!obj || obj->attr_inuse) return -EINVAL; + if (amdgpu_sriov_vf(adev) && !amdgpu_virt_ras_telemetry_block_en(adev, head->block)) + return 0; + get_obj(obj); snprintf(obj->fs_data.sysfs_name, sizeof(obj->fs_data.sysfs_name), diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c index 52c16bfeccaa..12ffe4a963d3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c @@ -748,7 +748,7 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) /* Modify the header if it exceeds. */ if (amdgpu_bad_page_threshold != 0 && - control->ras_num_bad_pages >= ras->bad_page_cnt_threshold) { + control->ras_num_bad_pages > ras->bad_page_cnt_threshold) { dev_warn(adev->dev, "Saved bad pages %d reaches threshold value %d\n", control->ras_num_bad_pages, ras->bad_page_cnt_threshold); @@ -806,7 +806,7 @@ amdgpu_ras_eeprom_update_header(struct amdgpu_ras_eeprom_control *control) */ if (amdgpu_bad_page_threshold != 0 && control->tbl_hdr.version == RAS_TABLE_VER_V2_1 && - control->ras_num_bad_pages < ras->bad_page_cnt_threshold) + control->ras_num_bad_pages <= ras->bad_page_cnt_threshold) control->tbl_rai.health_percent = ((ras->bad_page_cnt_threshold - control->ras_num_bad_pages) * 100) / ras->bad_page_cnt_threshold; @@ -1451,7 +1451,7 @@ int amdgpu_ras_eeprom_check(struct amdgpu_ras_eeprom_control *control) res); return -EINVAL; } - if (ras->bad_page_cnt_threshold > control->ras_num_bad_pages) { + if (ras->bad_page_cnt_threshold >= control->ras_num_bad_pages) { /* This means that, the threshold was increased since * the last time the system was booted, and now, * ras->bad_page_cnt_threshold - control->num_recs > 0, diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c index c586ab4c911b..34fc742fda91 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.c @@ -152,7 +152,8 @@ static bool amdgpu_sync_add_later(struct amdgpu_sync *sync, struct dma_fence *f) * * Add the fence to the sync object. */ -int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f) +int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f, + gfp_t flags) { struct amdgpu_sync_entry *e; @@ -162,7 +163,7 @@ int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f) if (amdgpu_sync_add_later(sync, f)) return 0; - e = kmem_cache_alloc(amdgpu_sync_slab, GFP_KERNEL); + e = kmem_cache_alloc(amdgpu_sync_slab, flags); if (!e) return -ENOMEM; @@ -249,7 +250,7 @@ int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync, struct dma_fence *tmp = dma_fence_chain_contained(f); if (amdgpu_sync_test_fence(adev, mode, owner, tmp)) { - r = amdgpu_sync_fence(sync, f); + r = amdgpu_sync_fence(sync, f, GFP_KERNEL); dma_fence_put(f); if (r) return r; @@ -281,7 +282,7 @@ int amdgpu_sync_kfd(struct amdgpu_sync *sync, struct dma_resv *resv) if (fence_owner != AMDGPU_FENCE_OWNER_KFD) continue; - r = amdgpu_sync_fence(sync, f); + r = amdgpu_sync_fence(sync, f, GFP_KERNEL); if (r) break; } @@ -388,7 +389,7 @@ int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone) hash_for_each_safe(source->fences, i, tmp, e, node) { f = e->fence; if (!dma_fence_is_signaled(f)) { - r = amdgpu_sync_fence(clone, f); + r = amdgpu_sync_fence(clone, f, GFP_KERNEL); if (r) return r; } else { @@ -400,6 +401,25 @@ int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone) } /** + * amdgpu_sync_move - move all fences from src to dst + * + * @src: source of the fences, empty after function + * @dst: destination for the fences + * + * Moves all fences from source to destination. All fences in destination are + * freed and source is empty after the function call. + */ +void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst) +{ + unsigned int i; + + amdgpu_sync_free(dst); + + for (i = 0; i < HASH_SIZE(src->fences); ++i) + hlist_move_list(&src->fences[i], &dst->fences[i]); +} + +/** * amdgpu_sync_push_to_job - push fences into job * @sync: sync object to get the fences from * @job: job to push the fences into diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h index e3272dce798d..51eb4382c91e 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_sync.h @@ -47,7 +47,8 @@ struct amdgpu_sync { }; void amdgpu_sync_create(struct amdgpu_sync *sync); -int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f); +int amdgpu_sync_fence(struct amdgpu_sync *sync, struct dma_fence *f, + gfp_t flags); int amdgpu_sync_resv(struct amdgpu_device *adev, struct amdgpu_sync *sync, struct dma_resv *resv, enum amdgpu_sync_mode mode, void *owner); @@ -56,6 +57,7 @@ struct dma_fence *amdgpu_sync_peek_fence(struct amdgpu_sync *sync, struct amdgpu_ring *ring); struct dma_fence *amdgpu_sync_get_fence(struct amdgpu_sync *sync); int amdgpu_sync_clone(struct amdgpu_sync *source, struct amdgpu_sync *clone); +void amdgpu_sync_move(struct amdgpu_sync *src, struct amdgpu_sync *dst); int amdgpu_sync_push_to_job(struct amdgpu_sync *sync, struct amdgpu_job *job); int amdgpu_sync_wait(struct amdgpu_sync *sync, bool intr); void amdgpu_sync_free(struct amdgpu_sync *sync); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c index eafe20d8fe0b..0a1ef95b2866 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c @@ -387,6 +387,45 @@ int amdgpu_umc_fill_error_record(struct ras_err_data *err_data, return 0; } +static int amdgpu_umc_loop_all_aid(struct amdgpu_device *adev, umc_func func, + void *data) +{ + uint32_t umc_node_inst; + uint32_t node_inst; + uint32_t umc_inst; + uint32_t ch_inst; + int ret; + + /* + * This loop is done based on the following - + * umc.active mask = mask of active umc instances across all nodes + * umc.umc_inst_num = maximum number of umc instancess per node + * umc.node_inst_num = maximum number of node instances + * Channel instances are not assumed to be harvested. + */ + dev_dbg(adev->dev, "active umcs :%lx umc_inst per node: %d", + adev->umc.active_mask, adev->umc.umc_inst_num); + for_each_set_bit(umc_node_inst, &(adev->umc.active_mask), + adev->umc.node_inst_num * adev->umc.umc_inst_num) { + node_inst = umc_node_inst / adev->umc.umc_inst_num; + umc_inst = umc_node_inst % adev->umc.umc_inst_num; + LOOP_UMC_CH_INST(ch_inst) { + dev_dbg(adev->dev, + "node_inst :%d umc_inst: %d ch_inst: %d", + node_inst, umc_inst, ch_inst); + ret = func(adev, node_inst, umc_inst, ch_inst, data); + if (ret) { + dev_err(adev->dev, + "Node %d umc %d ch %d func returns %d\n", + node_inst, umc_inst, ch_inst, ret); + return ret; + } + } + } + + return 0; +} + int amdgpu_umc_loop_channels(struct amdgpu_device *adev, umc_func func, void *data) { @@ -395,6 +434,9 @@ int amdgpu_umc_loop_channels(struct amdgpu_device *adev, uint32_t ch_inst = 0; int ret = 0; + if (adev->aid_mask) + return amdgpu_umc_loop_all_aid(adev, func, data); + if (adev->umc.node_inst_num) { LOOP_UMC_EACH_NODE_INST_AND_CH(node_inst, umc_inst, ch_inst) { ret = func(adev, node_inst, umc_inst, ch_inst, data); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h index adaf4388ad28..ce66a938f41a 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.h @@ -66,7 +66,6 @@ #define VCN_ENC_CMD_REG_WAIT 0x0000000c #define VCN_AON_SOC_ADDRESS_2_0 0x1f800 -#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 #define VCN_VID_IP_ADDRESS_2_0 0x0 #define VCN_AON_IP_ADDRESS_2_0 0x30000 diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c index 13e5709ea1ca..e6f0152e5b08 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.c @@ -1247,7 +1247,8 @@ amdgpu_ras_block_to_sriov(struct amdgpu_device *adev, enum amdgpu_ras_block bloc case AMDGPU_RAS_BLOCK__MPIO: return RAS_TELEMETRY_GPU_BLOCK_MPIO; default: - dev_err(adev->dev, "Unsupported SRIOV RAS telemetry block 0x%x\n", block); + DRM_WARN_ONCE("Unsupported SRIOV RAS telemetry block 0x%x\n", + block); return RAS_TELEMETRY_GPU_BLOCK_COUNT; } } @@ -1332,3 +1333,17 @@ int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev) return 0; } + +bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev, + enum amdgpu_ras_block block) +{ + enum amd_sriov_ras_telemetry_gpu_block sriov_block; + + sriov_block = amdgpu_ras_block_to_sriov(adev, block); + + if (sriov_block >= RAS_TELEMETRY_GPU_BLOCK_COUNT || + !amdgpu_sriov_ras_telemetry_block_en(adev, sriov_block)) + return false; + + return true; +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h index 0ca73343a768..0f3ccae5c1ab 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_virt.h @@ -407,4 +407,6 @@ bool amdgpu_virt_get_ras_capability(struct amdgpu_device *adev); int amdgpu_virt_req_ras_err_count(struct amdgpu_device *adev, enum amdgpu_ras_block block, struct ras_err_data *err_data); int amdgpu_virt_ras_telemetry_post_reset(struct amdgpu_device *adev); +bool amdgpu_virt_ras_telemetry_block_en(struct amdgpu_device *adev, + enum amdgpu_ras_block block); #endif diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 22aa4a8f1189..21be10d46cf9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -754,6 +754,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool need_pipe_sync) { struct amdgpu_device *adev = ring->adev; + struct amdgpu_isolation *isolation = &adev->isolation[ring->xcp_id]; unsigned vmhub = ring->vm_hub; struct amdgpu_vmid_mgr *id_mgr = &adev->vm_manager.id_mgr[vmhub]; struct amdgpu_vmid *id = &id_mgr->ids[job->vmid]; @@ -761,8 +762,9 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, bool gds_switch_needed = ring->funcs->emit_gds_switch && job->gds_switch_needed; bool vm_flush_needed = job->vm_needs_flush; - struct dma_fence *fence = NULL; + bool cleaner_shader_needed = false; bool pasid_mapping_needed = false; + struct dma_fence *fence = NULL; unsigned int patch; int r; @@ -785,8 +787,12 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, pasid_mapping_needed &= adev->gmc.gmc_funcs->emit_pasid_mapping && ring->funcs->emit_wreg; + cleaner_shader_needed = adev->gfx.enable_cleaner_shader && + ring->funcs->emit_cleaner_shader && job->base.s_fence && + &job->base.s_fence->scheduled == isolation->spearhead; + if (!vm_flush_needed && !gds_switch_needed && !need_pipe_sync && - !(job->enforce_isolation && !job->vmid)) + !cleaner_shader_needed) return 0; amdgpu_ring_ib_begin(ring); @@ -797,9 +803,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, if (need_pipe_sync) amdgpu_ring_emit_pipeline_sync(ring); - if (adev->gfx.enable_cleaner_shader && - ring->funcs->emit_cleaner_shader && - job->enforce_isolation) + if (cleaner_shader_needed) ring->funcs->emit_cleaner_shader(ring); if (vm_flush_needed) { @@ -821,7 +825,7 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, job->oa_size); } - if (vm_flush_needed || pasid_mapping_needed) { + if (vm_flush_needed || pasid_mapping_needed || cleaner_shader_needed) { r = amdgpu_fence_emit(ring, &fence, NULL, 0); if (r) return r; @@ -843,6 +847,17 @@ int amdgpu_vm_flush(struct amdgpu_ring *ring, struct amdgpu_job *job, id->pasid_mapping = dma_fence_get(fence); mutex_unlock(&id_mgr->lock); } + + /* + * Make sure that all other submissions wait for the cleaner shader to + * finish before we push them to the HW. + */ + if (cleaner_shader_needed) { + mutex_lock(&adev->enforce_isolation_mutex); + dma_fence_put(isolation->spearhead); + isolation->spearhead = dma_fence_get(fence); + mutex_unlock(&adev->enforce_isolation_mutex); + } dma_fence_put(fence); amdgpu_ring_patch_cond_exec(ring, patch); @@ -2672,20 +2687,6 @@ unreserve_bo: return r; } -/** - * amdgpu_vm_release_compute - release a compute vm - * @adev: amdgpu_device pointer - * @vm: a vm turned into compute vm by calling amdgpu_vm_make_compute - * - * This is a correspondant of amdgpu_vm_make_compute. It decouples compute - * pasid from vm. Compute should stop use of vm after this call. - */ -void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm) -{ - amdgpu_vm_set_pasid(adev, vm, 0); - vm->is_compute_context = false; -} - static int amdgpu_vm_stats_is_zero(struct amdgpu_vm *vm) { for (int i = 0; i < __AMDGPU_PL_NUM; ++i) { diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h index 5010a3107bf8..f3ad687125ad 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h @@ -489,7 +489,6 @@ int amdgpu_vm_set_pasid(struct amdgpu_device *adev, struct amdgpu_vm *vm, long amdgpu_vm_wait_idle(struct amdgpu_vm *vm, long timeout); int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm, int32_t xcp_id); int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm); -void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm); void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm); int amdgpu_vm_lock_pd(struct amdgpu_vm *vm, struct drm_exec *exec, unsigned int num_fences); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 5ba263fe5512..09178d56afbf 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -4739,6 +4739,20 @@ static int gfx_v10_0_sw_init(struct amdgpu_ip_block *ip_block) break; } switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { + case IP_VERSION(10, 1, 10): + adev->gfx.cleaner_shader_ptr = gfx_10_1_10_cleaner_shader_hex; + adev->gfx.cleaner_shader_size = sizeof(gfx_10_1_10_cleaner_shader_hex); + if (adev->gfx.me_fw_version >= 101 && + adev->gfx.pfp_fw_version >= 158 && + adev->gfx.mec_fw_version >= 152) { + adev->gfx.enable_cleaner_shader = true; + r = amdgpu_gfx_cleaner_shader_sw_init(adev, adev->gfx.cleaner_shader_size); + if (r) { + adev->gfx.enable_cleaner_shader = false; + dev_err(adev->dev, "Failed to initialize cleaner shader\n"); + } + } + break; case IP_VERSION(10, 3, 0): case IP_VERSION(10, 3, 2): case IP_VERSION(10, 3, 4): @@ -6044,7 +6058,7 @@ static int gfx_v10_0_cp_gfx_load_pfp_microcode(struct amdgpu_device *adev) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, mmCP_PFP_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0); @@ -6122,7 +6136,7 @@ static int gfx_v10_0_cp_gfx_load_ce_microcode(struct amdgpu_device *adev) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, mmCP_CE_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_CE_IC_BASE_CNTL, VMID, 0); @@ -6199,7 +6213,7 @@ static int gfx_v10_0_cp_gfx_load_me_microcode(struct amdgpu_device *adev) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, mmCP_ME_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0); @@ -6574,7 +6588,7 @@ static int gfx_v10_0_cp_compute_load_microcode(struct amdgpu_device *adev) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, mmCP_CPC_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0); 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 663c2572d440..5255378af53c 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 @@ -21,6 +21,41 @@ * OTHER DEALINGS IN THE SOFTWARE. */ +/* Define the cleaner shader gfx_10_1_10 */ +static const u32 gfx_10_1_10_cleaner_shader_hex[] = { + 0xb0804004, 0xbf8a0000, + 0xbf068100, 0xbf840023, + 0xbe8203b8, 0xbefc0380, + 0x7e008480, 0x7e028480, + 0x7e048480, 0x7e068480, + 0x7e088480, 0x7e0a8480, + 0x7e0c8480, 0x7e0e8480, + 0xbefc0302, 0x80828802, + 0xbf84fff5, 0xbe8203ff, + 0x80000000, 0x87020102, + 0xbf840012, 0xbefe03c1, + 0xbeff03c1, 0xd7650001, + 0x0001007f, 0xd7660001, + 0x0002027e, 0x16020288, + 0xbe8203bf, 0xbefc03c1, + 0xd9382000, 0x00020201, + 0xd9386040, 0x00040401, + 0xd70f6a01, 0x000202ff, + 0x00000400, 0x80828102, + 0xbf84fff7, 0xbefc03ff, + 0x00000068, 0xbe803080, + 0xbe813080, 0xbe823080, + 0xbe833080, 0x80fc847c, + 0xbf84fffa, 0xbeea0480, + 0xbeec0480, 0xbeee0480, + 0xbef00480, 0xbef20480, + 0xbef40480, 0xbef60480, + 0xbef80480, 0xbefa0480, + 0xbf810000, 0xbf9f0000, + 0xbf9f0000, 0xbf9f0000, + 0xbf9f0000, 0xbf9f0000, +}; + /* Define the cleaner shader gfx_10_3_0 */ static const u32 gfx_10_3_0_cleaner_shader_hex[] = { 0xb0804004, 0xbf8a0000, 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 new file mode 100644 index 000000000000..9ba3359253c9 --- /dev/null +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_1_10_cleaner_shader.asm @@ -0,0 +1,126 @@ +/* SPDX-License-Identifier: MIT */ +/* + * Copyright 2025 Advanced Micro Devices, Inc. + * + * Permission is hereby granted, free of charge, to any person obtaining a + * copy of this software and associated documentation files (the "Software"), + * to deal in the Software without restriction, including without limitation + * the rights to use, copy, modify, merge, publish, distribute, sublicense, + * and/or sell copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL + * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR + * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, + * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ + +// This shader is to clean LDS, SGPRs and VGPRs. It is first 64 Dwords or 256 bytes of 256 Dwords cleaner shader. + +// GFX10.1 : Clear SGPRs, VGPRs and LDS +// Launch 32 waves per CU (16 per SIMD) as a workgroup (threadgroup) to fill every wave slot +// Waves are "wave32" and have 64 VGPRs each, which uses all 1024 VGPRs per SIMD +// Waves are launched in "CU" mode, and the workgroup shares 64KB of LDS (half of the WGP's LDS) +// It takes 2 workgroups to use all of LDS: one on each CU of the WGP +// Each wave clears SGPRs 0 - 107 +// Each wave clears VGPRs 0 - 63 +// The first wave of the workgroup clears its 64KB of LDS +// The shader starts with "S_BARRIER" to ensure SPI has launched all waves of the workgroup +// before any wave in the workgroup could end. Without this, it is possible not all SGPRs get cleared. + + +shader main + asic(GFX10.1) + type(CS) + wave_size(32) +// Note: original source code from SQ team + +// +// Create 32 waves in a threadgroup (CS waves) +// Each allocates 64 VGPRs +// The workgroup allocates all of LDS (64kbytes) +// +// Takes about 2500 clocks to run. +// (theorhetical fastest = 1024clks vgpr + 640lds = 1660 clks) +// + S_BARRIER + s_cmp_eq_u32 s0, 1 // Bit0 is set, sgpr0 is set then clear VGPRS and LDS as FW set COMPUTE_USER_DATA_0 + s_cbranch_scc0 label_0023 // Clean VGPRs and LDS if sgpr0 of wave is set, scc = (s0 == 1) + + s_mov_b32 s2, 0x00000038 // Loop 64/8=8 times (loop unrolled for performance) + s_mov_b32 m0, 0 + // + // CLEAR VGPRs + // +label_0005: + v_movreld_b32 v0, 0 + v_movreld_b32 v1, 0 + v_movreld_b32 v2, 0 + v_movreld_b32 v3, 0 + v_movreld_b32 v4, 0 + v_movreld_b32 v5, 0 + v_movreld_b32 v6, 0 + v_movreld_b32 v7, 0 + s_mov_b32 m0, s2 + 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_cbranch_scc0 label_0023 // Clean LDS if its first wave of ThreadGroup/WorkGroup + // CLEAR LDS + // + s_mov_b32 exec_lo, 0xffffffff + s_mov_b32 exec_hi, 0xffffffff + v_mbcnt_lo_u32_b32 v1, exec_hi, 0 // Set V1 to thread-ID (0..63) + v_mbcnt_hi_u32_b32 v1, exec_lo, v1 // Set V1 to thread-ID (0..63) + v_mul_u32_u24 v1, 0x00000008, v1 // * 8, so each thread is a double-dword address (8byte) + s_mov_b32 s2, 0x00000003f // 64 loop iterations + s_mov_b32 m0, 0xffffffff + // Clear all of LDS space + // Each FirstWave of WorkGroup clears 64kbyte block + +label_001F: + ds_write2_b64 v1, v[2:3], v[2:3] offset1:32 + ds_write2_b64 v1, v[4:5], v[4:5] offset0:64 offset1:96 + v_add_co_u32 v1, vcc, 0x00000400, v1 + s_sub_u32 s2, s2, 1 + s_cbranch_scc0 label_001F + + // + // CLEAR SGPRs + // +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_sub_u32 m0, m0, 4 + s_cbranch_scc0 label_sgpr_loop + + //clear vcc + s_mov_b64 vcc, 0 //clear vcc + //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_lo, 0 //clear flat scratch lo SGPR + //s_setreg_imm32_b32 hw_reg_shader_flat_scratch_hi, 0 //clear flat scratch hi SGPR + s_mov_b64 ttmp0, 0 //Clear ttmp0 and ttmp1 + s_mov_b64 ttmp2, 0 //Clear ttmp2 and ttmp3 + s_mov_b64 ttmp4, 0 //Clear ttmp4 and ttmp5 + s_mov_b64 ttmp6, 0 //Clear ttmp6 and ttmp7 + s_mov_b64 ttmp8, 0 //Clear ttmp8 and ttmp9 + s_mov_b64 ttmp10, 0 //Clear ttmp10 and ttmp11 + s_mov_b64 ttmp12, 0 //Clear ttmp12 and ttmp13 + s_mov_b64 ttmp14, 0 //Clear ttmp14 and ttmp15 + + s_endpgm + +end + + diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index cfb51baa581a..e050c2e4ea73 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -64,6 +64,23 @@ #define regPC_CONFIG_CNTL_1 0x194d #define regPC_CONFIG_CNTL_1_BASE_IDX 1 +#define regCP_GFX_MQD_CONTROL_DEFAULT 0x00000100 +#define regCP_GFX_HQD_VMID_DEFAULT 0x00000000 +#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT 0x00000000 +#define regCP_GFX_HQD_QUANTUM_DEFAULT 0x00000a01 +#define regCP_GFX_HQD_CNTL_DEFAULT 0x00a00000 +#define regCP_RB_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_GFX_HQD_RPTR_DEFAULT 0x00000000 + +#define regCP_HQD_EOP_CONTROL_DEFAULT 0x00000006 +#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_MQD_CONTROL_DEFAULT 0x00000100 +#define regCP_HQD_PQ_CONTROL_DEFAULT 0x00308509 +#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_HQD_PQ_RPTR_DEFAULT 0x00000000 +#define regCP_HQD_PERSISTENT_STATE_DEFAULT 0x0be05501 +#define regCP_HQD_IB_CONTROL_DEFAULT 0x00300000 + MODULE_FIRMWARE("amdgpu/gc_11_0_0_pfp.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_me.bin"); MODULE_FIRMWARE("amdgpu/gc_11_0_0_mec.bin"); @@ -2391,7 +2408,7 @@ static int gfx_v11_0_config_me_cache(struct amdgpu_device *adev, uint64_t addr) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, regCP_ME_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_ME_IC_BASE_CNTL, VMID, 0); @@ -2435,7 +2452,7 @@ static int gfx_v11_0_config_pfp_cache(struct amdgpu_device *adev, uint64_t addr) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_PFP_IC_BASE_CNTL, VMID, 0); @@ -2480,7 +2497,7 @@ static int gfx_v11_0_config_mec_cache(struct amdgpu_device *adev, uint64_t addr) } if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); tmp = RREG32_SOC15(GC, 0, regCP_CPC_IC_BASE_CNTL); tmp = REG_SET_FIELD(tmp, CP_CPC_IC_BASE_CNTL, CACHE_POLICY, 0); @@ -3115,7 +3132,7 @@ static int gfx_v11_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev) amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj); if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO, lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr)); @@ -3333,7 +3350,7 @@ static int gfx_v11_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev) amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj); if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO, lower_32_bits(adev->gfx.me.me_fw_gpu_addr)); @@ -3958,7 +3975,7 @@ static void gfx_v11_0_gfx_mqd_set_priority(struct amdgpu_device *adev, if (prop->hqd_pipe_priority == AMDGPU_GFX_PIPE_PRIO_HIGH) priority = 1; - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY); + tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, priority); mqd->cp_gfx_hqd_queue_priority = tmp; } @@ -3980,14 +3997,14 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr); /* set up mqd control */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_MQD_CONTROL); + tmp = regCP_GFX_MQD_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0); tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1); tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0); mqd->cp_gfx_mqd_control = tmp; /* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_VMID); + tmp = regCP_GFX_HQD_VMID_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0); mqd->cp_gfx_hqd_vmid = 0; @@ -3995,7 +4012,7 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, gfx_v11_0_gfx_mqd_set_priority(adev, mqd, prop); /* set up time quantum */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUANTUM); + tmp = regCP_GFX_HQD_QUANTUM_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1); mqd->cp_gfx_hqd_quantum = tmp; @@ -4017,7 +4034,7 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, /* set up the gfx_hqd_control, similar as CP_RB0_CNTL */ rb_bufsz = order_base_2(prop->queue_size / 4) - 1; - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_CNTL); + tmp = regCP_GFX_HQD_CNTL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz); tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2); #ifdef __BIG_ENDIAN @@ -4026,7 +4043,7 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_gfx_hqd_cntl = tmp; /* set up cp_doorbell_control */ - tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL); + tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT; if (prop->use_doorbell) { tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL, DOORBELL_OFFSET, prop->doorbell_index); @@ -4038,7 +4055,7 @@ static int gfx_v11_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_rb_doorbell_control = tmp; /* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */ - mqd->cp_gfx_hqd_rptr = RREG32_SOC15(GC, 0, regCP_GFX_HQD_RPTR); + mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT; /* active the queue */ mqd->cp_gfx_hqd_active = 1; @@ -4124,14 +4141,14 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr); /* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL); + tmp = regCP_HQD_EOP_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE, (order_base_2(GFX11_MEC_HPD_SIZE / 4) - 1)); mqd->cp_hqd_eop_control = tmp; /* enable doorbell? */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL); + tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT; if (prop->use_doorbell) { tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, @@ -4160,7 +4177,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr); /* set MQD vmid to 0 */ - tmp = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL); + tmp = regCP_MQD_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0); mqd->cp_mqd_control = tmp; @@ -4170,7 +4187,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr); /* set up the HQD, this is similar to CP_RB0_CNTL */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL); + tmp = regCP_HQD_PQ_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE, (order_base_2(prop->queue_size / 4) - 1)); tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE, @@ -4196,7 +4213,7 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m, tmp = 0; /* enable the doorbell if requested */ if (prop->use_doorbell) { - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL); + tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_OFFSET, prop->doorbell_index); @@ -4211,17 +4228,17 @@ static int gfx_v11_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_pq_doorbell_control = tmp; /* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */ - mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR); + mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT; /* set the vmid for the queue */ mqd->cp_hqd_vmid = 0; - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE); + tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55); mqd->cp_hqd_persistent_state = tmp; /* set MIN_IB_AVAIL_SIZE */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_IB_CONTROL); + tmp = regCP_HQD_IB_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3); mqd->cp_hqd_ib_control = tmp; @@ -4549,7 +4566,7 @@ static int gfx_v11_0_gfxhub_enable(struct amdgpu_device *adev) if (r) return r; - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index c21b168f75a7..2ec900d50d7f 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -52,6 +52,24 @@ #define RLCG_UCODE_LOADING_START_ADDRESS 0x00002000L +#define regCP_GFX_MQD_CONTROL_DEFAULT 0x00000100 +#define regCP_GFX_HQD_VMID_DEFAULT 0x00000000 +#define regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT 0x00000000 +#define regCP_GFX_HQD_QUANTUM_DEFAULT 0x00000a01 +#define regCP_GFX_HQD_CNTL_DEFAULT 0x00f00000 +#define regCP_RB_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_GFX_HQD_RPTR_DEFAULT 0x00000000 + +#define regCP_HQD_EOP_CONTROL_DEFAULT 0x00000006 +#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_MQD_CONTROL_DEFAULT 0x00000100 +#define regCP_HQD_PQ_CONTROL_DEFAULT 0x00308509 +#define regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT 0x00000000 +#define regCP_HQD_PQ_RPTR_DEFAULT 0x00000000 +#define regCP_HQD_PERSISTENT_STATE_DEFAULT 0x0be05501 +#define regCP_HQD_IB_CONTROL_DEFAULT 0x00300000 + + MODULE_FIRMWARE("amdgpu/gc_12_0_0_pfp.bin"); MODULE_FIRMWARE("amdgpu/gc_12_0_0_me.bin"); MODULE_FIRMWARE("amdgpu/gc_12_0_0_mec.bin"); @@ -2306,7 +2324,7 @@ static int gfx_v12_0_cp_gfx_load_pfp_microcode_rs64(struct amdgpu_device *adev) amdgpu_bo_unreserve(adev->gfx.pfp.pfp_fw_data_obj); if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); WREG32_SOC15(GC, 0, regCP_PFP_IC_BASE_LO, lower_32_bits(adev->gfx.pfp.pfp_fw_gpu_addr)); @@ -2450,7 +2468,7 @@ static int gfx_v12_0_cp_gfx_load_me_microcode_rs64(struct amdgpu_device *adev) amdgpu_bo_unreserve(adev->gfx.me.me_fw_data_obj); if (amdgpu_emu_mode == 1) - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); WREG32_SOC15(GC, 0, regCP_ME_IC_BASE_LO, lower_32_bits(adev->gfx.me.me_fw_gpu_addr)); @@ -2891,25 +2909,25 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr); /* set up mqd control */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_MQD_CONTROL); + tmp = regCP_GFX_MQD_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, VMID, 0); tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, PRIV_STATE, 1); tmp = REG_SET_FIELD(tmp, CP_GFX_MQD_CONTROL, CACHE_POLICY, 0); mqd->cp_gfx_mqd_control = tmp; /* set up gfx_hqd_vimd with 0x0 to indicate the ring buffer's vmid */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_VMID); + tmp = regCP_GFX_HQD_VMID_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_VMID, VMID, 0); mqd->cp_gfx_hqd_vmid = 0; /* set up default queue priority level * 0x0 = low priority, 0x1 = high priority */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUEUE_PRIORITY); + tmp = regCP_GFX_HQD_QUEUE_PRIORITY_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUEUE_PRIORITY, PRIORITY_LEVEL, 0); mqd->cp_gfx_hqd_queue_priority = tmp; /* set up time quantum */ - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_QUANTUM); + tmp = regCP_GFX_HQD_QUANTUM_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_QUANTUM, QUANTUM_EN, 1); mqd->cp_gfx_hqd_quantum = tmp; @@ -2931,7 +2949,7 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, /* set up the gfx_hqd_control, similar as CP_RB0_CNTL */ rb_bufsz = order_base_2(prop->queue_size / 4) - 1; - tmp = RREG32_SOC15(GC, 0, regCP_GFX_HQD_CNTL); + tmp = regCP_GFX_HQD_CNTL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BUFSZ, rb_bufsz); tmp = REG_SET_FIELD(tmp, CP_GFX_HQD_CNTL, RB_BLKSZ, rb_bufsz - 2); #ifdef __BIG_ENDIAN @@ -2940,7 +2958,7 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_gfx_hqd_cntl = tmp; /* set up cp_doorbell_control */ - tmp = RREG32_SOC15(GC, 0, regCP_RB_DOORBELL_CONTROL); + tmp = regCP_RB_DOORBELL_CONTROL_DEFAULT; if (prop->use_doorbell) { tmp = REG_SET_FIELD(tmp, CP_RB_DOORBELL_CONTROL, DOORBELL_OFFSET, prop->doorbell_index); @@ -2952,7 +2970,7 @@ static int gfx_v12_0_gfx_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_rb_doorbell_control = tmp; /* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */ - mqd->cp_gfx_hqd_rptr = RREG32_SOC15(GC, 0, regCP_GFX_HQD_RPTR); + mqd->cp_gfx_hqd_rptr = regCP_GFX_HQD_RPTR_DEFAULT; /* active the queue */ mqd->cp_gfx_hqd_active = 1; @@ -3047,14 +3065,14 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_eop_base_addr_hi = upper_32_bits(eop_base_addr); /* set the EOP size, register value is 2^(EOP_SIZE+1) dwords */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_EOP_CONTROL); + tmp = regCP_HQD_EOP_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_EOP_CONTROL, EOP_SIZE, (order_base_2(GFX12_MEC_HPD_SIZE / 4) - 1)); mqd->cp_hqd_eop_control = tmp; /* enable doorbell? */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL); + tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT; if (prop->use_doorbell) { tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, @@ -3083,7 +3101,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_mqd_base_addr_hi = upper_32_bits(prop->mqd_gpu_addr); /* set MQD vmid to 0 */ - tmp = RREG32_SOC15(GC, 0, regCP_MQD_CONTROL); + tmp = regCP_MQD_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_MQD_CONTROL, VMID, 0); mqd->cp_mqd_control = tmp; @@ -3093,7 +3111,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_pq_base_hi = upper_32_bits(hqd_gpu_addr); /* set up the HQD, this is similar to CP_RB0_CNTL */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_CONTROL); + tmp = regCP_HQD_PQ_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, QUEUE_SIZE, (order_base_2(prop->queue_size / 4) - 1)); tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_CONTROL, RPTR_BLOCK_SIZE, @@ -3118,7 +3136,7 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m, tmp = 0; /* enable the doorbell if requested */ if (prop->use_doorbell) { - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PQ_DOORBELL_CONTROL); + tmp = regCP_HQD_PQ_DOORBELL_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PQ_DOORBELL_CONTROL, DOORBELL_OFFSET, prop->doorbell_index); @@ -3133,17 +3151,17 @@ static int gfx_v12_0_compute_mqd_init(struct amdgpu_device *adev, void *m, mqd->cp_hqd_pq_doorbell_control = tmp; /* reset read and write pointers, similar to CP_RB0_WPTR/_RPTR */ - mqd->cp_hqd_pq_rptr = RREG32_SOC15(GC, 0, regCP_HQD_PQ_RPTR); + mqd->cp_hqd_pq_rptr = regCP_HQD_PQ_RPTR_DEFAULT; /* set the vmid for the queue */ mqd->cp_hqd_vmid = 0; - tmp = RREG32_SOC15(GC, 0, regCP_HQD_PERSISTENT_STATE); + tmp = regCP_HQD_PERSISTENT_STATE_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_PERSISTENT_STATE, PRELOAD_SIZE, 0x55); mqd->cp_hqd_persistent_state = tmp; /* set MIN_IB_AVAIL_SIZE */ - tmp = RREG32_SOC15(GC, 0, regCP_HQD_IB_CONTROL); + tmp = regCP_HQD_IB_CONTROL_DEFAULT; tmp = REG_SET_FIELD(tmp, CP_HQD_IB_CONTROL, MIN_IB_AVAIL_SIZE, 3); mqd->cp_hqd_ib_control = tmp; @@ -3469,7 +3487,7 @@ static int gfx_v12_0_gfxhub_enable(struct amdgpu_device *adev) if (r) return r; - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; diff --git a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c index 0e3ddea7b8e0..a7bfc9f41d0e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfxhub_v1_0.c @@ -92,12 +92,12 @@ static void gfxhub_v1_0_init_system_aperture_regs(struct amdgpu_device *adev) { uint64_t value; - /* Program the AGP BAR */ - WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BASE, 0); - WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24); - WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24); - if (!amdgpu_sriov_vf(adev) || adev->asic_type <= CHIP_VEGA10) { + /* Program the AGP BAR */ + WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BASE, 0); + WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_BOT, adev->gmc.agp_start >> 24); + WREG32_SOC15_RLC(GC, 0, mmMC_VM_AGP_TOP, adev->gmc.agp_end >> 24); + /* Program the system aperture low logical page number. */ WREG32_SOC15_RLC(GC, 0, mmMC_VM_SYSTEM_APERTURE_LOW_ADDR, min(adev->gmc.fb_start, adev->gmc.agp_start) >> 18); diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c index 9bedca9a79c6..a88ad9951d32 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c @@ -268,7 +268,7 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng; /* flush hdp cache */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); /* This is necessary for SRIOV as well as for GFXOFF to function * properly under bare metal @@ -969,7 +969,7 @@ static int gmc_v10_0_gart_enable(struct amdgpu_device *adev) adev->hdp.funcs->init_registers(adev); /* Flush HDP after it is initialized */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c index 72751ab4c766..0ba82eabca02 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v11_0.c @@ -229,7 +229,7 @@ static void gmc_v11_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, ack = hub->vm_inv_eng0_ack + hub->eng_distance * eng; /* flush hdp cache */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); /* This is necessary for SRIOV as well as for GFXOFF to function * properly under bare metal @@ -750,6 +750,18 @@ static int gmc_v11_0_sw_init(struct amdgpu_ip_block *ip_block) adev->gmc.vram_type = vram_type; adev->gmc.vram_vendor = vram_vendor; + /* The mall_size is already calculated as mall_size_per_umc * num_umc. + * However, for gfx1151, which features a 2-to-1 UMC mapping, + * the result must be multiplied by 2 to determine the actual mall size. + */ + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { + case IP_VERSION(11, 5, 1): + adev->gmc.mall_size *= 2; + break; + default: + break; + } + switch (amdgpu_ip_version(adev, GC_HWIP, 0)) { case IP_VERSION(11, 0, 0): case IP_VERSION(11, 0, 1): @@ -896,7 +908,7 @@ static int gmc_v11_0_gart_enable(struct amdgpu_device *adev) return r; /* Flush HDP after it is initialized */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c index c3c144a4f45e..0f136d6bbdc9 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v12_0.c @@ -297,7 +297,7 @@ static void gmc_v12_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid, return; /* flush hdp cache */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); /* This is necessary for SRIOV as well as for GFXOFF to function * properly under bare metal @@ -881,7 +881,7 @@ static int gmc_v12_0_gart_enable(struct amdgpu_device *adev) return r; /* Flush HDP after it is initialized */ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); value = (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) ? false : true; diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c index 291549765c38..f1dc9e50d67e 100644 --- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c @@ -1504,7 +1504,6 @@ static void gmc_v9_0_set_umc_funcs(struct amdgpu_device *adev) adev->umc.umc_inst_num = UMC_V12_0_UMC_INSTANCE_NUM; adev->umc.node_inst_num /= UMC_V12_0_UMC_INSTANCE_NUM; adev->umc.channel_offs = UMC_V12_0_PER_CHANNEL_OFFSET; - adev->umc.active_mask = adev->aid_mask; adev->umc.retire_unit = UMC_V12_0_BAD_PAGE_NUM_PER_CHANNEL; if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu) adev->umc.ras = &umc_v12_0_ras; @@ -2434,7 +2433,7 @@ static int gmc_v9_0_hw_init(struct amdgpu_ip_block *ip_block) adev->hdp.funcs->init_registers(adev); /* After HDP is initialized, flush HDP.*/ - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_ALWAYS) value = false; diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c index 194026e9be33..1ca1bbe7784e 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v4_0.c @@ -42,7 +42,12 @@ static void hdp_v4_0_flush_hdp(struct amdgpu_device *adev, { if (!ring || !ring->funcs->emit_wreg) { WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); - RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + /* We just need to read back a register to post the write. + * Reading back the remapped register causes problems on + * some platforms so just read back the memory size register. + */ + if (adev->nbio.funcs->get_memsize) + adev->nbio.funcs->get_memsize(adev); } else { amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); } diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c index d3962d469088..40705e13ca56 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_0.c @@ -33,7 +33,12 @@ static void hdp_v5_0_flush_hdp(struct amdgpu_device *adev, { if (!ring || !ring->funcs->emit_wreg) { WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); - RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + /* We just need to read back a register to post the write. + * Reading back the remapped register causes problems on + * some platforms so just read back the memory size register. + */ + if (adev->nbio.funcs->get_memsize) + adev->nbio.funcs->get_memsize(adev); } else { amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); } diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c index f52552c5fa27..6b9f2e1d9d69 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v5_2.c @@ -34,7 +34,17 @@ static void hdp_v5_2_flush_hdp(struct amdgpu_device *adev, if (!ring || !ring->funcs->emit_wreg) { WREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); - RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + if (amdgpu_sriov_vf(adev)) { + /* this is fine because SR_IOV doesn't remap the register */ + RREG32_NO_KIQ((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + } else { + /* We just need to read back a register to post the write. + * Reading back the remapped register causes problems on + * some platforms so just read back the memory size register. + */ + if (adev->nbio.funcs->get_memsize) + adev->nbio.funcs->get_memsize(adev); + } } else { amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c index 6948fe9956ce..20da813299f0 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v6_0.c @@ -36,7 +36,12 @@ static void hdp_v6_0_flush_hdp(struct amdgpu_device *adev, { if (!ring || !ring->funcs->emit_wreg) { WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); - RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + /* We just need to read back a register to post the write. + * Reading back the remapped register causes problems on + * some platforms so just read back the memory size register. + */ + if (adev->nbio.funcs->get_memsize) + adev->nbio.funcs->get_memsize(adev); } else { amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); } diff --git a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c index 63820329f67e..f7ecdd15d528 100644 --- a/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/hdp_v7_0.c @@ -33,7 +33,12 @@ static void hdp_v7_0_flush_hdp(struct amdgpu_device *adev, { if (!ring || !ring->funcs->emit_wreg) { WREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); - RREG32((adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2); + /* We just need to read back a register to post the write. + * Reading back the remapped register causes problems on + * some platforms so just read back the memory size register. + */ + if (adev->nbio.funcs->get_memsize) + adev->nbio.funcs->get_memsize(adev); } else { amdgpu_ring_emit_wreg(ring, (adev->rmmio_remap.reg_offset + KFD_MMIO_REMAP_HDP_MEM_FLUSH_CNTL) >> 2, 0); } 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 88f9771c1686..b2904ee494e0 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.c @@ -637,7 +637,7 @@ static uint64_t jpeg_v4_0_3_dec_ring_get_wptr(struct amdgpu_ring *ring) ring->pipe ? (0x40 * ring->pipe - 0xc80) : 0); } -static void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring) +void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring) { /* JPEG engine access for HDP flush doesn't work when RRMT is enabled. * This is a workaround to avoid any HDP flush through JPEG ring. 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 747a3e5f6856..a90bf370a002 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v4_0_3.h @@ -56,6 +56,7 @@ void jpeg_v4_0_3_dec_ring_emit_fence(struct amdgpu_ring *ring, u64 addr, u64 seq unsigned int flags); void jpeg_v4_0_3_dec_ring_emit_vm_flush(struct amdgpu_ring *ring, unsigned int vmid, uint64_t pd_addr); +void jpeg_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring); void jpeg_v4_0_3_dec_ring_nop(struct amdgpu_ring *ring, uint32_t count); void jpeg_v4_0_3_dec_ring_insert_start(struct amdgpu_ring *ring); void jpeg_v4_0_3_dec_ring_insert_end(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 40d4c32a8c2a..f2cc11b3fd68 100644 --- a/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/jpeg_v5_0_1.c @@ -655,6 +655,7 @@ static const struct amdgpu_ring_funcs jpeg_v5_0_1_dec_ring_vm_funcs = { .emit_ib = jpeg_v4_0_3_dec_ring_emit_ib, .emit_fence = jpeg_v4_0_3_dec_ring_emit_fence, .emit_vm_flush = jpeg_v4_0_3_dec_ring_emit_vm_flush, + .emit_hdp_flush = jpeg_v4_0_3_ring_emit_hdp_flush, .test_ring = amdgpu_jpeg_dec_ring_test_ring, .test_ib = amdgpu_jpeg_dec_ring_test_ib, .insert_nop = jpeg_v4_0_3_dec_ring_nop, diff --git a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c index 0f808ffcab94..68bb334393bb 100644 --- a/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/mes_v11_0.c @@ -730,7 +730,7 @@ static int mes_v11_0_set_hw_resources(struct amdgpu_mes *mes) static int mes_v11_0_set_hw_resources_1(struct amdgpu_mes *mes) { - int size = 128 * PAGE_SIZE; + int size = 128 * AMDGPU_GPU_PAGE_SIZE; int ret = 0; struct amdgpu_device *adev = mes->adev; union MESAPI_SET_HW_RESOURCES_1 mes_set_hw_res_pkt; diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c index 9689e2b5d4e5..2adee2b94c37 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_7.c @@ -172,6 +172,30 @@ static void mmhub_v1_7_init_tlb_regs(struct amdgpu_device *adev) WREG32_SOC15(MMHUB, 0, regMC_VM_MX_L1_TLB_CNTL, tmp); } +/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */ +static void mmhub_v1_7_init_snoop_override_regs(struct amdgpu_device *adev) +{ + uint32_t tmp; + int i; + uint32_t distance = regDAGB1_WRCLI_GPU_SNOOP_OVERRIDE - + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE; + + for (i = 0; i < 5; i++) { /* DAGB instances */ + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, i * distance); + tmp |= (1 << 15); /* SDMA client is BIT15 */ + WREG32_SOC15_OFFSET(MMHUB, 0, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, i * distance, tmp); + + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, i * distance); + tmp |= (1 << 15); + WREG32_SOC15_OFFSET(MMHUB, 0, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, i * distance, tmp); + } + +} + static void mmhub_v1_7_init_cache_regs(struct amdgpu_device *adev) { uint32_t tmp; @@ -337,6 +361,7 @@ static int mmhub_v1_7_gart_enable(struct amdgpu_device *adev) mmhub_v1_7_init_system_aperture_regs(adev); mmhub_v1_7_init_tlb_regs(adev); mmhub_v1_7_init_cache_regs(adev); + mmhub_v1_7_init_snoop_override_regs(adev); mmhub_v1_7_enable_system_domain(adev); mmhub_v1_7_disable_identity_aperture(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c index e646e5cef0a2..ce013a715b86 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v1_8.c @@ -213,6 +213,32 @@ static void mmhub_v1_8_init_tlb_regs(struct amdgpu_device *adev) } } +/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */ +static void mmhub_v1_8_init_snoop_override_regs(struct amdgpu_device *adev) +{ + uint32_t tmp, inst_mask; + int i, j; + uint32_t distance = regDAGB1_WRCLI_GPU_SNOOP_OVERRIDE - + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE; + + inst_mask = adev->aid_mask; + for_each_inst(i, inst_mask) { + for (j = 0; j < 5; j++) { /* DAGB instances */ + tmp = RREG32_SOC15_OFFSET(MMHUB, i, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, j * distance); + tmp |= (1 << 15); /* SDMA client is BIT15 */ + WREG32_SOC15_OFFSET(MMHUB, i, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, j * distance, tmp); + + tmp = RREG32_SOC15_OFFSET(MMHUB, i, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, j * distance); + tmp |= (1 << 15); + WREG32_SOC15_OFFSET(MMHUB, i, + regDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, j * distance, tmp); + } + } +} + static void mmhub_v1_8_init_cache_regs(struct amdgpu_device *adev) { uint32_t tmp, inst_mask; @@ -418,6 +444,7 @@ static int mmhub_v1_8_gart_enable(struct amdgpu_device *adev) mmhub_v1_8_init_system_aperture_regs(adev); mmhub_v1_8_init_tlb_regs(adev); mmhub_v1_8_init_cache_regs(adev); + mmhub_v1_8_init_snoop_override_regs(adev); mmhub_v1_8_enable_system_domain(adev); mmhub_v1_8_disable_identity_aperture(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c index ff1b58e44689..fe0710b55c3a 100644 --- a/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c +++ b/drivers/gpu/drm/amd/amdgpu/mmhub_v9_4.c @@ -198,6 +198,36 @@ static void mmhub_v9_4_init_tlb_regs(struct amdgpu_device *adev, int hubid) hubid * MMHUB_INSTANCE_REGISTER_OFFSET, tmp); } +/* Set snoop bit for SDMA so that SDMA writes probe-invalidates RW lines */ +static void mmhub_v9_4_init_snoop_override_regs(struct amdgpu_device *adev, int hubid) +{ + uint32_t tmp; + int i; + uint32_t distance = mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE - + mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE; + uint32_t huboffset = hubid * MMHUB_INSTANCE_REGISTER_OFFSET; + + for (i = 0; i < 5 - (2 * hubid); i++) { + /* DAGB instances 0 to 4 are in hub0 and 5 to 7 are in hub1 */ + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, + mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, + huboffset + i * distance); + tmp |= (1 << 15); /* SDMA client is BIT15 */ + WREG32_SOC15_OFFSET(MMHUB, 0, + mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE, + huboffset + i * distance, tmp); + + tmp = RREG32_SOC15_OFFSET(MMHUB, 0, + mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, + huboffset + i * distance); + tmp |= (1 << 15); + WREG32_SOC15_OFFSET(MMHUB, 0, + mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE, + huboffset + i * distance, tmp); + } + +} + static void mmhub_v9_4_init_cache_regs(struct amdgpu_device *adev, int hubid) { uint32_t tmp; @@ -392,6 +422,7 @@ static int mmhub_v9_4_gart_enable(struct amdgpu_device *adev) if (!amdgpu_sriov_vf(adev)) mmhub_v9_4_init_cache_regs(adev, i); + mmhub_v9_4_init_snoop_override_regs(adev, i); mmhub_v9_4_enable_system_domain(adev, i); if (!amdgpu_sriov_vf(adev)) mmhub_v9_4_disable_identity_aperture(adev, i); diff --git a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c index 41421da63a08..a11f556b3ff1 100644 --- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c +++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_11.c @@ -361,7 +361,7 @@ static void nbio_v7_11_get_clockgating_state(struct amdgpu_device *adev, *flags |= AMD_CG_SUPPORT_BIF_LS; } -#define MMIO_REG_HOLE_OFFSET (0x80000 - PAGE_SIZE) +#define MMIO_REG_HOLE_OFFSET 0x44000 static void nbio_v7_11_set_reg_remap(struct amdgpu_device *adev) { diff --git a/drivers/gpu/drm/amd/amdgpu/nv.c b/drivers/gpu/drm/amd/amdgpu/nv.c index 95c609317a8d..efca7dc27d68 100644 --- a/drivers/gpu/drm/amd/amdgpu/nv.c +++ b/drivers/gpu/drm/amd/amdgpu/nv.c @@ -141,23 +141,23 @@ static struct amdgpu_video_codec_info sriov_sc_video_codecs_encode_array[] = { }; static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn0[] = { - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 4096, 4096, 3)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 4096, 4096, 5)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; static struct amdgpu_video_codec_info sriov_sc_video_codecs_decode_array_vcn1[] = { - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 4096, 4096, 3)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 4096, 4096, 5)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 1920, 1088, 3)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 1920, 1088, 5)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 1920, 1088, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, }; diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c index 2395f1856962..e77a467af7ac 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c @@ -532,7 +532,7 @@ static int psp_v11_0_memory_training(struct psp_context *psp, uint32_t ops) } memcpy_toio(adev->mman.aper_base_kaddr, buf, sz); - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); vfree(buf); drm_dev_exit(idx); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c index cc621064610f..afdf8ce3b4c5 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c @@ -610,7 +610,7 @@ static int psp_v13_0_memory_training(struct psp_context *psp, uint32_t ops) } memcpy_toio(adev->mman.aper_base_kaddr, buf, sz); - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); vfree(buf); drm_dev_exit(idx); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c index 4d33c95a5116..89f6c06946c5 100644 --- a/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c +++ b/drivers/gpu/drm/amd/amdgpu/psp_v14_0.c @@ -488,7 +488,7 @@ static int psp_v14_0_memory_training(struct psp_context *psp, uint32_t ops) } memcpy_toio(adev->mman.aper_base_kaddr, buf, sz); - adev->hdp.funcs->flush_hdp(adev, NULL); + amdgpu_device_flush_hdp(adev, NULL); vfree(buf); drm_dev_exit(idx); } else { diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c index e98fb3fa36a8..6e09613de8cd 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc15.c +++ b/drivers/gpu/drm/amd/amdgpu/soc15.c @@ -604,12 +604,10 @@ soc15_asic_reset_method(struct amdgpu_device *adev) static bool soc15_need_reset_on_resume(struct amdgpu_device *adev) { /* Will reset for the following suspend abort cases. - * 1) Only reset on APU side, dGPU hasn't checked yet. - * 2) S3 suspend aborted in the normal S3 suspend or - * performing pm core test. + * 1) S3 suspend aborted in the normal S3 suspend + * 2) S3 suspend aborted in performing pm core test. */ - if (adev->flags & AMD_IS_APU && adev->in_s3 && - !pm_resume_via_firmware()) + if (adev->in_s3 && !pm_resume_via_firmware()) return true; else return false; diff --git a/drivers/gpu/drm/amd/amdgpu/soc21.c b/drivers/gpu/drm/amd/amdgpu/soc21.c index 62ad67d0b598..c66cff399f63 100644 --- a/drivers/gpu/drm/amd/amdgpu/soc21.c +++ b/drivers/gpu/drm/amd/amdgpu/soc21.c @@ -117,23 +117,17 @@ static struct amdgpu_video_codecs sriov_vcn_4_0_0_video_codecs_encode_vcn1 = { }; static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_decode_array_vcn0[] = { - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 4096, 4096, 3)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 4096, 4096, 5)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_AV1, 8192, 4352, 0)}, }; static struct amdgpu_video_codec_info sriov_vcn_4_0_0_video_codecs_decode_array_vcn1[] = { - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG2, 4096, 4096, 3)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4, 4096, 4096, 5)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_MPEG4_AVC, 4096, 4096, 52)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VC1, 4096, 4096, 4)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_HEVC, 8192, 4352, 186)}, - {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 4096, 4096, 0)}, + {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_JPEG, 16384, 16384, 0)}, {codec_info_build(AMDGPU_INFO_VIDEO_CAPS_CODEC_IDX_VP9, 8192, 4352, 0)}, }; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index e42cfc731ad8..f40737d27cb0 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -39,6 +39,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fa00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48200 +#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 #define mmUVD_CONTEXT_ID_INTERNAL_OFFSET 0x1fd #define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET 0x503 diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index b518202955ca..2431e1914a8f 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -39,6 +39,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fa00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48200 +#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 #define mmUVD_CONTEXT_ID_INTERNAL_OFFSET 0x27 #define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET 0x0f diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index 63ddd4cca910..02c2defcf91e 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -40,6 +40,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fa00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48200 +#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 #define mmUVD_CONTEXT_ID_INTERNAL_OFFSET 0x27 #define mmUVD_GPCOM_VCPU_CMD_INTERNAL_OFFSET 0x0f diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 00551d6f0370..090794457339 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -46,6 +46,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fb00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48300 +#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 #define VCN_HARVEST_MMSCH 0 @@ -582,7 +583,8 @@ static void vcn_v4_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_idx /* VCN global tiling registers */ WREG32_SOC15_DPG_MODE(inst_idx, SOC15_DPG_MODE_OFFSET( - VCN, 0, regUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect); + VCN, inst_idx, regUVD_GFX10_ADDR_CONFIG), + adev->gfx.config.gb_addr_config, 0, indirect); } /** 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 ecdc027f8220..855da1149c5c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -31,6 +31,7 @@ #include "soc15d.h" #include "soc15_hw_ip.h" #include "vcn_v2_0.h" +#include "vcn_v4_0_3.h" #include "mmsch_v4_0_3.h" #include "vcn/vcn_4_0_3_offset.h" @@ -44,6 +45,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fb00 #define VCN1_VID_SOC_ADDRESS_3_0 0x48300 +#define VCN1_AON_SOC_ADDRESS_3_0 0x48000 static const struct amdgpu_hwip_reg_entry vcn_reg_list_4_0_3[] = { SOC15_REG_ENTRY_STR(VCN, 0, regUVD_POWER_STATUS), @@ -1461,8 +1463,8 @@ static uint64_t vcn_v4_0_3_unified_ring_get_wptr(struct amdgpu_ring *ring) regUVD_RB_WPTR); } -static void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, - uint32_t val, uint32_t mask) +void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t reg, + uint32_t val, uint32_t mask) { /* Use normalized offsets when required */ if (vcn_v4_0_3_normalizn_reqd(ring->adev)) @@ -1474,7 +1476,8 @@ static void vcn_v4_0_3_enc_ring_emit_reg_wait(struct amdgpu_ring *ring, uint32_t amdgpu_ring_write(ring, val); } -static void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, uint32_t val) +void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, + uint32_t val) { /* Use normalized offsets when required */ if (vcn_v4_0_3_normalizn_reqd(ring->adev)) @@ -1485,8 +1488,8 @@ static void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg amdgpu_ring_write(ring, val); } -static void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring, - unsigned int vmid, uint64_t pd_addr) +void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring, + unsigned int vmid, uint64_t pd_addr) { struct amdgpu_vmhub *hub = &ring->adev->vmhub[ring->vm_hub]; @@ -1498,7 +1501,7 @@ static void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring, lower_32_bits(pd_addr), 0xffffffff); } -static void vcn_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring) +void vcn_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring) { /* VCN engine access for HDP flush doesn't work when RRMT is enabled. * This is a workaround to avoid any HDP flush through VCN ring. 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 0b046114373a..03572a1d0c9c 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.h @@ -26,4 +26,13 @@ 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, + uint32_t val, uint32_t mask); + +void vcn_v4_0_3_enc_ring_emit_wreg(struct amdgpu_ring *ring, uint32_t reg, + uint32_t val); +void vcn_v4_0_3_enc_ring_emit_vm_flush(struct amdgpu_ring *ring, + unsigned int vmid, uint64_t pd_addr); +void vcn_v4_0_3_ring_emit_hdp_flush(struct amdgpu_ring *ring); + #endif /* __VCN_V4_0_3_H__ */ 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 23d3c16c9d9f..c9761d27fd61 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -46,6 +46,7 @@ #define VCN_VID_SOC_ADDRESS_2_0 0x1fb00 #define VCN1_VID_SOC_ADDRESS_3_0 (0x48300 + 0x38000) +#define VCN1_AON_SOC_ADDRESS_3_0 (0x48000 + 0x38000) #define VCN_HARVEST_MMSCH 0 @@ -982,6 +983,10 @@ static int vcn_v4_0_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, b 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, regVCN_RB1_DB_CTRL); + return 0; } @@ -990,183 +995,182 @@ static int vcn_v4_0_5_start_dpg_mode(struct amdgpu_device *adev, int inst_idx, b * vcn_v4_0_5_start - VCN start * * @adev: amdgpu_device pointer + * @i: instance to start * * Start VCN block */ -static int vcn_v4_0_5_start(struct amdgpu_device *adev) +static int vcn_v4_0_5_start(struct amdgpu_device *adev, int i) { volatile struct amdgpu_vcn4_fw_shared *fw_shared; struct amdgpu_ring *ring; uint32_t tmp; - int i, j, k, r; + int j, k, r; - for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - if (adev->pm.dpm_enabled) - amdgpu_dpm_enable_vcn(adev, true, i); - } + if (adev->vcn.harvest_config & (1 << i)) + return 0; - for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - if (adev->vcn.harvest_config & (1 << i)) - continue; + if (adev->pm.dpm_enabled) + amdgpu_dpm_enable_vcn(adev, true, i); - fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; - if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { - r = vcn_v4_0_5_start_dpg_mode(adev, i, adev->vcn.indirect_sram); - continue; - } + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) + return vcn_v4_0_5_start_dpg_mode(adev, i, adev->vcn.indirect_sram); - /* disable VCN power gating */ - vcn_v4_0_5_disable_static_power_gating(adev, i); - - /* set VCN status busy */ - tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY; - WREG32_SOC15(VCN, i, regUVD_STATUS, tmp); - - /*SW clock gating */ - vcn_v4_0_5_disable_clock_gating(adev, i); - - /* enable VCPU clock */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), - UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK); - - /* disable master interrupt */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0, - ~UVD_MASTINT_EN__VCPU_EN_MASK); - - /* enable LMI MC and UMC channels */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0, - ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); - - tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); - tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK; - tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK; - WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); - - /* setup regUVD_LMI_CTRL */ - tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL); - WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp | - UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | - UVD_LMI_CTRL__MASK_MC_URGENT_MASK | - UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | - UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK); - - /* setup regUVD_MPC_CNTL */ - tmp = RREG32_SOC15(VCN, i, regUVD_MPC_CNTL); - tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK; - tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT; - WREG32_SOC15(VCN, i, regUVD_MPC_CNTL, tmp); - - /* setup UVD_MPC_SET_MUXA0 */ - WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXA0, - ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) | - (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) | - (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) | - (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT))); - - /* setup UVD_MPC_SET_MUXB0 */ - WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXB0, - ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) | - (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) | - (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) | - (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT))); - - /* setup UVD_MPC_SET_MUX */ - WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUX, - ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) | - (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) | - (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT))); - - vcn_v4_0_5_mc_resume(adev, i); - - /* VCN global tiling registers */ - WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG, - adev->gfx.config.gb_addr_config); - - /* unblock VCPU register access */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0, - ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK); - - /* release VCPU reset to boot */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, - ~UVD_VCPU_CNTL__BLK_RST_MASK); - - for (j = 0; j < 10; ++j) { - uint32_t status; - - for (k = 0; k < 100; ++k) { - status = RREG32_SOC15(VCN, i, regUVD_STATUS); - if (status & 2) - break; - mdelay(10); - if (amdgpu_emu_mode == 1) - msleep(1); - } + /* disable VCN power gating */ + vcn_v4_0_5_disable_static_power_gating(adev, i); + + /* set VCN status busy */ + tmp = RREG32_SOC15(VCN, i, regUVD_STATUS) | UVD_STATUS__UVD_BUSY; + WREG32_SOC15(VCN, i, regUVD_STATUS, tmp); + + /* SW clock gating */ + vcn_v4_0_5_disable_clock_gating(adev, i); + + /* enable VCPU clock */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), + UVD_VCPU_CNTL__CLK_EN_MASK, ~UVD_VCPU_CNTL__CLK_EN_MASK); + + /* disable master interrupt */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), 0, + ~UVD_MASTINT_EN__VCPU_EN_MASK); - if (amdgpu_emu_mode == 1) { - r = -1; - if (status & 2) { - r = 0; - break; - } - } else { + /* enable LMI MC and UMC channels */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_LMI_CTRL2), 0, + ~UVD_LMI_CTRL2__STALL_ARB_UMC_MASK); + + tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); + tmp &= ~UVD_SOFT_RESET__LMI_SOFT_RESET_MASK; + tmp &= ~UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK; + WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); + + /* setup regUVD_LMI_CTRL */ + tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL); + WREG32_SOC15(VCN, i, regUVD_LMI_CTRL, tmp | + UVD_LMI_CTRL__WRITE_CLEAN_TIMER_EN_MASK | + UVD_LMI_CTRL__MASK_MC_URGENT_MASK | + UVD_LMI_CTRL__DATA_COHERENCY_EN_MASK | + UVD_LMI_CTRL__VCPU_DATA_COHERENCY_EN_MASK); + + /* setup regUVD_MPC_CNTL */ + tmp = RREG32_SOC15(VCN, i, regUVD_MPC_CNTL); + tmp &= ~UVD_MPC_CNTL__REPLACEMENT_MODE_MASK; + tmp |= 0x2 << UVD_MPC_CNTL__REPLACEMENT_MODE__SHIFT; + WREG32_SOC15(VCN, i, regUVD_MPC_CNTL, tmp); + + /* setup UVD_MPC_SET_MUXA0 */ + WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXA0, + ((0x1 << UVD_MPC_SET_MUXA0__VARA_1__SHIFT) | + (0x2 << UVD_MPC_SET_MUXA0__VARA_2__SHIFT) | + (0x3 << UVD_MPC_SET_MUXA0__VARA_3__SHIFT) | + (0x4 << UVD_MPC_SET_MUXA0__VARA_4__SHIFT))); + + /* setup UVD_MPC_SET_MUXB0 */ + WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUXB0, + ((0x1 << UVD_MPC_SET_MUXB0__VARB_1__SHIFT) | + (0x2 << UVD_MPC_SET_MUXB0__VARB_2__SHIFT) | + (0x3 << UVD_MPC_SET_MUXB0__VARB_3__SHIFT) | + (0x4 << UVD_MPC_SET_MUXB0__VARB_4__SHIFT))); + + /* setup UVD_MPC_SET_MUX */ + WREG32_SOC15(VCN, i, regUVD_MPC_SET_MUX, + ((0x0 << UVD_MPC_SET_MUX__SET_0__SHIFT) | + (0x1 << UVD_MPC_SET_MUX__SET_1__SHIFT) | + (0x2 << UVD_MPC_SET_MUX__SET_2__SHIFT))); + + vcn_v4_0_5_mc_resume(adev, i); + + /* VCN global tiling registers */ + WREG32_SOC15(VCN, i, regUVD_GFX10_ADDR_CONFIG, + adev->gfx.config.gb_addr_config); + + /* unblock VCPU register access */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), 0, + ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK); + + /* release VCPU reset to boot */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, + ~UVD_VCPU_CNTL__BLK_RST_MASK); + + for (j = 0; j < 10; ++j) { + uint32_t status; + + for (k = 0; k < 100; ++k) { + status = RREG32_SOC15(VCN, i, regUVD_STATUS); + if (status & 2) + break; + mdelay(10); + if (amdgpu_emu_mode == 1) + msleep(1); + } + + if (amdgpu_emu_mode == 1) { + r = -1; + if (status & 2) { r = 0; - if (status & 2) - break; - - dev_err(adev->dev, - "VCN[%d] is not responding, trying to reset VCPU!!!\n", i); - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), - UVD_VCPU_CNTL__BLK_RST_MASK, - ~UVD_VCPU_CNTL__BLK_RST_MASK); - mdelay(10); - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, - ~UVD_VCPU_CNTL__BLK_RST_MASK); - - mdelay(10); - r = -1; + break; } + } else { + r = 0; + if (status & 2) + break; + + dev_err(adev->dev, + "VCN[%d] is not responding, trying to reset VCPU!!!\n", i); + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), + UVD_VCPU_CNTL__BLK_RST_MASK, + ~UVD_VCPU_CNTL__BLK_RST_MASK); + mdelay(10); + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, + ~UVD_VCPU_CNTL__BLK_RST_MASK); + + mdelay(10); + r = -1; } + } - if (r) { - dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i); - return r; - } + if (r) { + dev_err(adev->dev, "VCN[%d] is not responding, giving up!!!\n", i); + return r; + } - /* enable master interrupt */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), - UVD_MASTINT_EN__VCPU_EN_MASK, - ~UVD_MASTINT_EN__VCPU_EN_MASK); + /* enable master interrupt */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_MASTINT_EN), + UVD_MASTINT_EN__VCPU_EN_MASK, + ~UVD_MASTINT_EN__VCPU_EN_MASK); - /* clear the busy bit of VCN_STATUS */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0, - ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT)); + /* clear the busy bit of VCN_STATUS */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_STATUS), 0, + ~(2 << UVD_STATUS__VCPU_REPORT__SHIFT)); - ring = &adev->vcn.inst[i].ring_enc[0]; - WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL, - ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT | - VCN_RB1_DB_CTRL__EN_MASK); - - WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr); - WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); - WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4); - - tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE); - tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK); - WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp); - fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET; - WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0); - WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0); - - tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR); - WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp); - ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR); - - tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE); - tmp |= VCN_RB_ENABLE__RB1_EN_MASK; - WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp); - fw_shared->sq.queue_mode &= ~(FW_QUEUE_RING_RESET | FW_QUEUE_DPG_HOLD_OFF); - } + ring = &adev->vcn.inst[i].ring_enc[0]; + WREG32_SOC15(VCN, i, regVCN_RB1_DB_CTRL, + ring->doorbell_index << VCN_RB1_DB_CTRL__OFFSET__SHIFT | + VCN_RB1_DB_CTRL__EN_MASK); + + WREG32_SOC15(VCN, i, regUVD_RB_BASE_LO, ring->gpu_addr); + WREG32_SOC15(VCN, i, regUVD_RB_BASE_HI, upper_32_bits(ring->gpu_addr)); + WREG32_SOC15(VCN, i, regUVD_RB_SIZE, ring->ring_size / 4); + + tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE); + tmp &= ~(VCN_RB_ENABLE__RB1_EN_MASK); + WREG32_SOC15(VCN, i, regVCN_RB_ENABLE, tmp); + fw_shared->sq.queue_mode |= FW_QUEUE_RING_RESET; + WREG32_SOC15(VCN, i, regUVD_RB_RPTR, 0); + WREG32_SOC15(VCN, i, regUVD_RB_WPTR, 0); + + tmp = RREG32_SOC15(VCN, i, regUVD_RB_RPTR); + WREG32_SOC15(VCN, i, regUVD_RB_WPTR, tmp); + ring->wptr = RREG32_SOC15(VCN, i, regUVD_RB_WPTR); + + tmp = RREG32_SOC15(VCN, i, regVCN_RB_ENABLE); + tmp |= VCN_RB_ENABLE__RB1_EN_MASK; + 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, regVCN_RB_ENABLE); return 0; } @@ -1203,88 +1207,87 @@ static void vcn_v4_0_5_stop_dpg_mode(struct amdgpu_device *adev, int inst_idx) * vcn_v4_0_5_stop - VCN stop * * @adev: amdgpu_device pointer + * @i: instance to stop * * Stop VCN block */ -static int vcn_v4_0_5_stop(struct amdgpu_device *adev) +static int vcn_v4_0_5_stop(struct amdgpu_device *adev, int i) { volatile struct amdgpu_vcn4_fw_shared *fw_shared; uint32_t tmp; - int i, r = 0; - - for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - if (adev->vcn.harvest_config & (1 << i)) - continue; + int r = 0; - fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; - fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF; + if (adev->vcn.harvest_config & (1 << i)) + return 0; - if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { - vcn_v4_0_5_stop_dpg_mode(adev, i); - continue; - } + fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; + fw_shared->sq.queue_mode |= FW_QUEUE_DPG_HOLD_OFF; - /* wait for vcn idle */ - r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7); - if (r) - return r; + if (adev->pg_flags & AMD_PG_SUPPORT_VCN_DPG) { + vcn_v4_0_5_stop_dpg_mode(adev, i); + r = 0; + goto done; + } - tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK | - UVD_LMI_STATUS__READ_CLEAN_MASK | - UVD_LMI_STATUS__WRITE_CLEAN_MASK | - UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK; - r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp); - if (r) - return r; + /* wait for vcn idle */ + r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_STATUS, UVD_STATUS__IDLE, 0x7); + if (r) + goto done; - /* disable LMI UMC channel */ - tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2); - tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK; - WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp); - tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK | - UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK; - r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp); - if (r) - return r; + tmp = UVD_LMI_STATUS__VCPU_LMI_WRITE_CLEAN_MASK | + UVD_LMI_STATUS__READ_CLEAN_MASK | + UVD_LMI_STATUS__WRITE_CLEAN_MASK | + UVD_LMI_STATUS__WRITE_CLEAN_RAW_MASK; + r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp); + if (r) + goto done; + + /* disable LMI UMC channel */ + tmp = RREG32_SOC15(VCN, i, regUVD_LMI_CTRL2); + tmp |= UVD_LMI_CTRL2__STALL_ARB_UMC_MASK; + WREG32_SOC15(VCN, i, regUVD_LMI_CTRL2, tmp); + tmp = UVD_LMI_STATUS__UMC_READ_CLEAN_RAW_MASK | + UVD_LMI_STATUS__UMC_WRITE_CLEAN_RAW_MASK; + r = SOC15_WAIT_ON_RREG(VCN, i, regUVD_LMI_STATUS, tmp, tmp); + if (r) + goto done; - /* block VCPU register access */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), - UVD_RB_ARB_CTRL__VCPU_DIS_MASK, - ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK); + /* block VCPU register access */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_RB_ARB_CTRL), + UVD_RB_ARB_CTRL__VCPU_DIS_MASK, + ~UVD_RB_ARB_CTRL__VCPU_DIS_MASK); - /* reset VCPU */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), - UVD_VCPU_CNTL__BLK_RST_MASK, - ~UVD_VCPU_CNTL__BLK_RST_MASK); + /* reset VCPU */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), + UVD_VCPU_CNTL__BLK_RST_MASK, + ~UVD_VCPU_CNTL__BLK_RST_MASK); - /* disable VCPU clock */ - WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, - ~(UVD_VCPU_CNTL__CLK_EN_MASK)); + /* disable VCPU clock */ + WREG32_P(SOC15_REG_OFFSET(VCN, i, regUVD_VCPU_CNTL), 0, + ~(UVD_VCPU_CNTL__CLK_EN_MASK)); - /* apply soft reset */ - tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); - tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK; - WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); - tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); - tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK; - WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); + /* apply soft reset */ + tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); + tmp |= UVD_SOFT_RESET__LMI_UMC_SOFT_RESET_MASK; + WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); + tmp = RREG32_SOC15(VCN, i, regUVD_SOFT_RESET); + tmp |= UVD_SOFT_RESET__LMI_SOFT_RESET_MASK; + WREG32_SOC15(VCN, i, regUVD_SOFT_RESET, tmp); - /* clear status */ - WREG32_SOC15(VCN, i, regUVD_STATUS, 0); + /* clear status */ + WREG32_SOC15(VCN, i, regUVD_STATUS, 0); - /* apply HW clock gating */ - vcn_v4_0_5_enable_clock_gating(adev, i); + /* apply HW clock gating */ + vcn_v4_0_5_enable_clock_gating(adev, i); - /* enable VCN power gating */ - vcn_v4_0_5_enable_static_power_gating(adev, i); - } + /* enable VCN power gating */ + vcn_v4_0_5_enable_static_power_gating(adev, i); - for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { - if (adev->pm.dpm_enabled) - amdgpu_dpm_enable_vcn(adev, false, i); - } +done: + if (adev->pm.dpm_enabled) + amdgpu_dpm_enable_vcn(adev, false, i); - return 0; + return r; } /** @@ -1536,15 +1539,17 @@ static int vcn_v4_0_5_set_powergating_state(struct amdgpu_ip_block *ip_block, enum amd_powergating_state state) { struct amdgpu_device *adev = ip_block->adev; - int ret; + int ret = 0, i; if (state == adev->vcn.cur_state) return 0; - if (state == AMD_PG_STATE_GATE) - ret = vcn_v4_0_5_stop(adev); - else - ret = vcn_v4_0_5_start(adev); + for (i = 0; i < adev->vcn.num_vcn_inst; ++i) { + if (state == AMD_PG_STATE_GATE) + ret |= vcn_v4_0_5_stop(adev, i); + else + ret |= vcn_v4_0_5_start(adev, i); + } if (!ret) adev->vcn.cur_state = state; 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 b6d78381ebfb..97fc3d5b1947 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c @@ -502,7 +502,8 @@ static void vcn_v5_0_0_mc_resume_dpg_mode(struct amdgpu_device *adev, int inst_i /* VCN global tiling registers */ WREG32_SOC24_DPG_MODE(inst_idx, SOC24_DPG_MODE_OFFSET( - VCN, 0, regUVD_GFX10_ADDR_CONFIG), adev->gfx.config.gb_addr_config, 0, indirect); + VCN, inst_idx, regUVD_GFX10_ADDR_CONFIG), + adev->gfx.config.gb_addr_config, 0, indirect); return; } 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 8b0b3739a537..f893a8428283 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_1.c @@ -29,6 +29,7 @@ #include "soc15d.h" #include "soc15_hw_ip.h" #include "vcn_v2_0.h" +#include "vcn_v4_0_3.h" #include "vcn/vcn_5_0_0_offset.h" #include "vcn/vcn_5_0_0_sh_mask.h" @@ -65,6 +66,22 @@ static int vcn_v5_0_1_early_init(struct amdgpu_ip_block *ip_block) return amdgpu_vcn_early_init(adev); } +static void vcn_v5_0_1_fw_shared_init(struct amdgpu_device *adev, int inst_idx) +{ + struct amdgpu_vcn5_fw_shared *fw_shared; + + fw_shared = adev->vcn.inst[inst_idx].fw_shared.cpu_addr; + + if (fw_shared->sq.is_enabled) + return; + fw_shared->present_flag_0 = + cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE); + fw_shared->sq.is_enabled = 1; + + if (amdgpu_vcnfw_log) + amdgpu_vcn_fwlog_init(&adev->vcn.inst[inst_idx]); +} + /** * vcn_v5_0_1_sw_init - sw init for VCN block * @@ -95,8 +112,6 @@ static int vcn_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block) return r; for (i = 0; i < adev->vcn.num_vcn_inst; i++) { - volatile struct amdgpu_vcn5_fw_shared *fw_shared; - vcn_inst = GET_INST(VCN, i); ring = &adev->vcn.inst[i].ring_enc[0]; @@ -111,12 +126,7 @@ static int vcn_v5_0_1_sw_init(struct amdgpu_ip_block *ip_block) if (r) return r; - fw_shared = adev->vcn.inst[i].fw_shared.cpu_addr; - fw_shared->present_flag_0 = cpu_to_le32(AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE); - fw_shared->sq.is_enabled = true; - - if (amdgpu_vcnfw_log) - amdgpu_vcn_fwlog_init(&adev->vcn.inst[i]); + vcn_v5_0_1_fw_shared_init(adev, i); } /* TODO: Add queue reset mask when FW fully supports it */ @@ -188,6 +198,9 @@ static int vcn_v5_0_1_hw_init(struct amdgpu_ip_block *ip_block) 9 * vcn_inst), adev->vcn.inst[i].aid_id); + /* Re-init fw_shared, if required */ + vcn_v5_0_1_fw_shared_init(adev, i); + r = amdgpu_ring_test_helper(ring); if (r) return r; @@ -893,16 +906,17 @@ static const struct amdgpu_ring_funcs vcn_v5_0_1_unified_ring_vm_funcs = { .get_rptr = vcn_v5_0_1_unified_ring_get_rptr, .get_wptr = vcn_v5_0_1_unified_ring_get_wptr, .set_wptr = vcn_v5_0_1_unified_ring_set_wptr, - .emit_frame_size = - SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + - SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 + - 4 + /* vcn_v2_0_enc_ring_emit_vm_flush */ - 5 + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */ - 1, /* vcn_v2_0_enc_ring_insert_end */ + .emit_frame_size = SOC15_FLUSH_GPU_TLB_NUM_WREG * 3 + + SOC15_FLUSH_GPU_TLB_NUM_REG_WAIT * 4 + + 4 + /* vcn_v2_0_enc_ring_emit_vm_flush */ + 5 + + 5 + /* vcn_v2_0_enc_ring_emit_fence x2 vm fence */ + 1, /* vcn_v2_0_enc_ring_insert_end */ .emit_ib_size = 5, /* vcn_v2_0_enc_ring_emit_ib */ .emit_ib = vcn_v2_0_enc_ring_emit_ib, .emit_fence = vcn_v2_0_enc_ring_emit_fence, - .emit_vm_flush = vcn_v2_0_enc_ring_emit_vm_flush, + .emit_vm_flush = vcn_v4_0_3_enc_ring_emit_vm_flush, + .emit_hdp_flush = vcn_v4_0_3_ring_emit_hdp_flush, .test_ring = amdgpu_vcn_enc_ring_test_ring, .test_ib = amdgpu_vcn_unified_ring_test_ib, .insert_nop = amdgpu_ring_insert_nop, @@ -910,8 +924,8 @@ static const struct amdgpu_ring_funcs vcn_v5_0_1_unified_ring_vm_funcs = { .pad_ib = amdgpu_ring_generic_pad_ib, .begin_use = amdgpu_vcn_ring_begin_use, .end_use = amdgpu_vcn_ring_end_use, - .emit_wreg = vcn_v2_0_enc_ring_emit_wreg, - .emit_reg_wait = vcn_v2_0_enc_ring_emit_reg_wait, + .emit_wreg = vcn_v4_0_3_enc_ring_emit_wreg, + .emit_reg_wait = vcn_v4_0_3_enc_ring_emit_reg_wait, .emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper, }; diff --git a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c index 795382b55e0a..981d9adcc5e1 100644 --- a/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c +++ b/drivers/gpu/drm/amd/amdkfd/cik_event_interrupt.c @@ -107,20 +107,30 @@ static void cik_event_interrupt_wq(struct kfd_node *dev, kfd_signal_hw_exception_event(pasid); else if (ihre->source_id == CIK_INTSRC_GFX_PAGE_INV_FAULT || ihre->source_id == CIK_INTSRC_GFX_MEM_PROT_FAULT) { + struct kfd_process_device *pdd = NULL; struct kfd_vm_fault_info info; + struct kfd_process *p; kfd_smi_event_update_vmfault(dev, pasid); - kfd_dqm_evict_pasid(dev->dqm, pasid); + p = kfd_lookup_process_by_pasid(pasid, &pdd); + if (!pdd) + return; + + kfd_evict_process_device(pdd); memset(&info, 0, sizeof(info)); amdgpu_amdkfd_gpuvm_get_vm_fault_info(dev->adev, &info); - if (!info.page_addr && !info.status) + if (!info.page_addr && !info.status) { + kfd_unref_process(p); return; + } if (info.vmid == vmid) - kfd_signal_vm_fault_event(dev, pasid, &info, NULL); + kfd_signal_vm_fault_event(pdd, &info, NULL); else - kfd_signal_vm_fault_event(dev, pasid, NULL, 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 33df35cab467..8c2e92378b49 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -155,8 +155,8 @@ static int kfd_open(struct inode *inode, struct file *filep) /* filep now owns the reference returned by kfd_create_process */ filep->private_data = process; - dev_dbg(kfd_device, "process %d opened, compat mode (32 bit) - %d\n", - process->pasid, process->is_32bit_user_mode); + dev_dbg(kfd_device, "process pid %d opened kfd node, compat mode (32 bit) - %d\n", + process->lead_thread->pid, process->is_32bit_user_mode); return 0; } @@ -366,8 +366,8 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p, goto err_acquire_queue_buf; } - pr_debug("Creating queue for PASID 0x%x on gpu 0x%x\n", - p->pasid, + pr_debug("Creating queue for process pid %d on gpu 0x%x\n", + p->lead_thread->pid, dev->id); err = pqm_create_queue(&p->pqm, dev, &q_properties, &queue_id, @@ -420,9 +420,9 @@ static int kfd_ioctl_destroy_queue(struct file *filp, struct kfd_process *p, int retval; struct kfd_ioctl_destroy_queue_args *args = data; - pr_debug("Destroying queue id %d for pasid 0x%x\n", + pr_debug("Destroying queue id %d for process pid %d\n", args->queue_id, - p->pasid); + p->lead_thread->pid); mutex_lock(&p->mutex); @@ -478,8 +478,8 @@ static int kfd_ioctl_update_queue(struct file *filp, struct kfd_process *p, properties.pm4_target_xcc = (args->queue_percentage >> 8) & 0xFF; properties.priority = args->queue_priority; - pr_debug("Updating queue id %d for pasid 0x%x\n", - args->queue_id, p->pasid); + pr_debug("Updating queue id %d for process pid %d\n", + args->queue_id, p->lead_thread->pid); mutex_lock(&p->mutex); @@ -705,7 +705,7 @@ static int kfd_ioctl_get_process_apertures(struct file *filp, struct kfd_process_device_apertures *pAperture; int i; - dev_dbg(kfd_device, "get apertures for PASID 0x%x", p->pasid); + dev_dbg(kfd_device, "get apertures for process pid %d", p->lead_thread->pid); args->num_of_nodes = 0; @@ -757,7 +757,8 @@ static int kfd_ioctl_get_process_apertures_new(struct file *filp, int ret; int i; - dev_dbg(kfd_device, "get apertures for PASID 0x%x", p->pasid); + dev_dbg(kfd_device, "get apertures for process pid %d", + p->lead_thread->pid); if (args->num_of_nodes == 0) { /* Return number of nodes, so that user space can alloacate @@ -3375,12 +3376,12 @@ static int kfd_mmio_mmap(struct kfd_node *dev, struct kfd_process *process, vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot); - pr_debug("pasid 0x%x mapping mmio page\n" + pr_debug("process pid %d mapping mmio page\n" " target user address == 0x%08llX\n" " physical address == 0x%08llX\n" " vm_flags == 0x%04lX\n" " size == 0x%04lX\n", - process->pasid, (unsigned long long) vma->vm_start, + process->lead_thread->pid, (unsigned long long) vma->vm_start, address, vma->vm_flags, PAGE_SIZE); return io_remap_pfn_range(vma, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c index a8abc3091801..12456c61ffa5 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c @@ -204,11 +204,12 @@ bool kfd_set_dbg_ev_from_interrupt(struct kfd_node *dev, size_t exception_data_size) { struct kfd_process *p; + struct kfd_process_device *pdd = NULL; bool signaled_to_debugger_or_runtime = false; - p = kfd_lookup_process_by_pasid(pasid); + p = kfd_lookup_process_by_pasid(pasid, &pdd); - if (!p) + if (!pdd) return false; if (!kfd_dbg_ev_raise(trap_mask, p, dev, doorbell_id, true, @@ -238,9 +239,8 @@ bool kfd_set_dbg_ev_from_interrupt(struct kfd_node *dev, mutex_unlock(&p->mutex); } else if (trap_mask & KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION)) { - kfd_dqm_evict_pasid(dev->dqm, p->pasid); - kfd_signal_vm_fault_event(dev, p->pasid, NULL, - exception_data); + kfd_evict_process_device(pdd); + kfd_signal_vm_fault_event(pdd, NULL, exception_data); signaled_to_debugger_or_runtime = true; } @@ -276,8 +276,8 @@ int kfd_dbg_send_exception_to_runtime(struct kfd_process *p, data = (struct kfd_hsa_memory_exception_data *) pdd->vm_fault_exc_data; - kfd_dqm_evict_pasid(pdd->dev->dqm, p->pasid); - kfd_signal_vm_fault_event(pdd->dev, p->pasid, NULL, data); + kfd_evict_process_device(pdd); + kfd_signal_vm_fault_event(pdd, NULL, data); error_reason &= ~KFD_EC_MASK(EC_DEVICE_MEMORY_VIOLATION); } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device.c b/drivers/gpu/drm/amd/amdkfd/kfd_device.c index 6cefd338f23d..bf978b368f6a 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device.c @@ -1558,7 +1558,7 @@ bool kgd2kfd_vmfault_fast_path(struct amdgpu_device *adev, struct amdgpu_iv_entr u32 cam_index; if (entry->ih == &adev->irq.ih_soft || entry->ih == &adev->irq.ih1) { - p = kfd_lookup_process_by_pasid(entry->pasid); + p = kfd_lookup_process_by_pasid(entry->pasid, NULL); if (!p) return true; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c index ad9cb50a9fa3..35ae3c55a97f 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c @@ -208,7 +208,7 @@ static int add_queue_mes(struct device_queue_manager *dqm, struct queue *q, return -EIO; memset(&queue_input, 0x0, sizeof(struct mes_add_queue_input)); - queue_input.process_id = qpd->pqm->process->pasid; + queue_input.process_id = pdd->pasid; queue_input.page_table_base_addr = qpd->page_table_base; queue_input.process_va_start = 0; queue_input.process_va_end = adev->vm_manager.max_pfn - 1; @@ -527,6 +527,7 @@ static int allocate_vmid(struct device_queue_manager *dqm, struct qcm_process_device *qpd, struct queue *q) { + struct kfd_process_device *pdd = qpd_to_pdd(qpd); struct device *dev = dqm->dev->adev->dev; int allocated_vmid = -1, i; @@ -545,9 +546,9 @@ static int allocate_vmid(struct device_queue_manager *dqm, pr_debug("vmid allocated: %d\n", allocated_vmid); - dqm->vmid_pasid[allocated_vmid] = q->process->pasid; + dqm->vmid_pasid[allocated_vmid] = pdd->pasid; - set_pasid_vmid_mapping(dqm, q->process->pasid, allocated_vmid); + set_pasid_vmid_mapping(dqm, pdd->pasid, allocated_vmid); qpd->vmid = allocated_vmid; q->properties.vmid = allocated_vmid; @@ -799,6 +800,11 @@ static int dbgdev_wave_reset_wavefronts(struct kfd_node *dev, struct kfd_process return -EOPNOTSUPP; } + /* taking the VMID for that process on the safe way using PDD */ + pdd = kfd_get_process_device_data(dev, p); + if (!pdd) + return -EFAULT; + /* Scan all registers in the range ATC_VMID8_PASID_MAPPING .. * ATC_VMID15_PASID_MAPPING * to check which VMID the current process is mapped to. @@ -808,23 +814,19 @@ static int dbgdev_wave_reset_wavefronts(struct kfd_node *dev, struct kfd_process status = dev->kfd2kgd->get_atc_vmid_pasid_mapping_info (dev->adev, vmid, &queried_pasid); - if (status && queried_pasid == p->pasid) { - pr_debug("Killing wave fronts of vmid %d and pasid 0x%x\n", - vmid, p->pasid); + if (status && queried_pasid == pdd->pasid) { + pr_debug("Killing wave fronts of vmid %d and process pid %d\n", + vmid, p->lead_thread->pid); break; } } if (vmid > last_vmid_to_scan) { - dev_err(dev->adev->dev, "Didn't find vmid for pasid 0x%x\n", p->pasid); + dev_err(dev->adev->dev, "Didn't find vmid for process pid %d\n", + p->lead_thread->pid); return -EFAULT; } - /* taking the VMID for that process on the safe way using PDD */ - pdd = kfd_get_process_device_data(dev, p); - if (!pdd) - return -EFAULT; - reg_gfx_index.bits.sh_broadcast_writes = 1; reg_gfx_index.bits.se_broadcast_writes = 1; reg_gfx_index.bits.instance_broadcast_writes = 1; @@ -1060,8 +1062,8 @@ static int suspend_single_queue(struct device_queue_manager *dqm, if (q->properties.is_suspended) return 0; - pr_debug("Suspending PASID %u queue [%i]\n", - pdd->process->pasid, + pr_debug("Suspending process pid %d queue [%i]\n", + pdd->process->lead_thread->pid, q->properties.queue_id); is_new = q->properties.exception_status & KFD_EC_MASK(EC_QUEUE_NEW); @@ -1108,8 +1110,8 @@ static int resume_single_queue(struct device_queue_manager *dqm, pdd = qpd_to_pdd(qpd); - pr_debug("Restoring from suspend PASID %u queue [%i]\n", - pdd->process->pasid, + pr_debug("Restoring from suspend process pid %d queue [%i]\n", + pdd->process->lead_thread->pid, q->properties.queue_id); q->properties.is_suspended = false; @@ -1142,8 +1144,8 @@ static int evict_process_queues_nocpsch(struct device_queue_manager *dqm, goto out; pdd = qpd_to_pdd(qpd); - pr_debug_ratelimited("Evicting PASID 0x%x queues\n", - pdd->process->pasid); + pr_debug_ratelimited("Evicting process pid %d queues\n", + pdd->process->lead_thread->pid); pdd->last_evict_timestamp = get_jiffies_64(); /* Mark all queues as evicted. Deactivate all active queues on @@ -1200,8 +1202,8 @@ static int evict_process_queues_cpsch(struct device_queue_manager *dqm, if (!pdd->drm_priv) goto out; - pr_debug_ratelimited("Evicting PASID 0x%x queues\n", - pdd->process->pasid); + pr_debug_ratelimited("Evicting process pid %d queues\n", + pdd->process->lead_thread->pid); /* Mark all queues as evicted. Deactivate all active queues on * the qpd. @@ -1261,8 +1263,8 @@ static int restore_process_queues_nocpsch(struct device_queue_manager *dqm, goto out; } - pr_debug_ratelimited("Restoring PASID 0x%x queues\n", - pdd->process->pasid); + pr_debug_ratelimited("Restoring process pid %d queues\n", + pdd->process->lead_thread->pid); /* Update PD Base in QPD */ qpd->page_table_base = pd_base; @@ -1345,8 +1347,8 @@ static int restore_process_queues_cpsch(struct device_queue_manager *dqm, if (!pdd->drm_priv) goto vm_not_acquired; - pr_debug_ratelimited("Restoring PASID 0x%x queues\n", - pdd->process->pasid); + pr_debug_ratelimited("Restoring process pid %d queues\n", + pdd->process->lead_thread->pid); /* Update PD Base in QPD */ qpd->page_table_base = amdgpu_amdkfd_gpuvm_get_process_page_dir(pdd->drm_priv); @@ -2137,8 +2139,8 @@ static void set_queue_as_reset(struct device_queue_manager *dqm, struct queue *q { struct kfd_process_device *pdd = qpd_to_pdd(qpd); - dev_err(dqm->dev->adev->dev, "queue id 0x%0x at pasid 0x%0x is reset\n", - q->properties.queue_id, q->process->pasid); + dev_err(dqm->dev->adev->dev, "queue id 0x%0x at pasid %d is reset\n", + q->properties.queue_id, pdd->process->lead_thread->pid); pdd->has_reset_queue = true; if (q->properties.is_active) { @@ -2491,14 +2493,6 @@ failed_try_destroy_debugged_queue: return retval; } -/* - * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to - * stay in user mode. - */ -#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL -/* APE1 limit is inclusive and 64K aligned. */ -#define APE1_LIMIT_ALIGNMENT 0xFFFF - static bool set_cache_memory_policy(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -2513,34 +2507,6 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, dqm_lock(dqm); - if (alternate_aperture_size == 0) { - /* base > limit disables APE1 */ - qpd->sh_mem_ape1_base = 1; - qpd->sh_mem_ape1_limit = 0; - } else { - /* - * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, - * SH_MEM_APE1_BASE[31:0], 0x0000 } - * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, - * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } - * Verify that the base and size parameters can be - * represented in this format and convert them. - * Additionally restrict APE1 to user-mode addresses. - */ - - uint64_t base = (uintptr_t)alternate_aperture_base; - uint64_t limit = base + alternate_aperture_size - 1; - - if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || - (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { - retval = false; - goto out; - } - - qpd->sh_mem_ape1_base = base >> 16; - qpd->sh_mem_ape1_limit = limit >> 16; - } - retval = dqm->asic_ops.set_cache_memory_policy( dqm, qpd, @@ -2549,6 +2515,9 @@ static bool set_cache_memory_policy(struct device_queue_manager *dqm, alternate_aperture_base, alternate_aperture_size); + if (retval) + goto out; + if ((dqm->sched_policy == KFD_SCHED_POLICY_NO_HWS) && (qpd->vmid != 0)) program_sh_mem_settings(dqm, qpd); @@ -2974,20 +2943,19 @@ void device_queue_manager_uninit(struct device_queue_manager *dqm) int kfd_dqm_suspend_bad_queue_mes(struct kfd_node *knode, u32 pasid, u32 doorbell_id) { - struct kfd_process_device *pdd; - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process_device *pdd = NULL; + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, &pdd); struct device_queue_manager *dqm = knode->dqm; struct device *dev = dqm->dev->adev->dev; struct qcm_process_device *qpd; struct queue *q = NULL; int ret = 0; - if (!p) + if (!pdd) return -EINVAL; dqm_lock(dqm); - pdd = kfd_get_process_device_data(dqm->dev, p); if (pdd) { qpd = &pdd->qpd; @@ -3020,6 +2988,7 @@ int kfd_dqm_suspend_bad_queue_mes(struct kfd_node *knode, u32 pasid, u32 doorbel out: dqm_unlock(dqm); + kfd_unref_process(p); return ret; } @@ -3061,24 +3030,21 @@ out: return ret; } -int kfd_dqm_evict_pasid(struct device_queue_manager *dqm, u32 pasid) +int kfd_evict_process_device(struct kfd_process_device *pdd) { - struct kfd_process_device *pdd; - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct device_queue_manager *dqm; + struct kfd_process *p; int ret = 0; - if (!p) - return -EINVAL; + p = pdd->process; + dqm = pdd->dev->dqm; + WARN(debug_evictions, "Evicting pid %d", p->lead_thread->pid); - pdd = kfd_get_process_device_data(dqm->dev, p); - if (pdd) { - if (dqm->dev->kfd->shared_resources.enable_mes) - ret = kfd_dqm_evict_pasid_mes(dqm, &pdd->qpd); - else - ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd); - } - kfd_unref_process(p); + if (dqm->dev->kfd->shared_resources.enable_mes) + ret = kfd_dqm_evict_pasid_mes(dqm, &pdd->qpd); + else + ret = dqm->ops.evict_process_queues(dqm, &pdd->qpd); return ret; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c index d4d95c7f2e5d..32bedef912b3 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_cik.c @@ -27,6 +27,14 @@ #include "oss/oss_2_4_sh_mask.h" #include "gca/gfx_7_2_sh_mask.h" +/* + * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to + * stay in user mode. + */ +#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL +/* APE1 limit is inclusive and 64K aligned. */ +#define APE1_LIMIT_ALIGNMENT 0xFFFF + static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -84,6 +92,36 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, { uint32_t default_mtype; uint32_t ape1_mtype; + unsigned int temp; + bool retval = true; + + if (alternate_aperture_size == 0) { + /* base > limit disables APE1 */ + qpd->sh_mem_ape1_base = 1; + qpd->sh_mem_ape1_limit = 0; + } else { + /* + * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, + * SH_MEM_APE1_BASE[31:0], 0x0000 } + * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, + * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } + * Verify that the base and size parameters can be + * represented in this format and convert them. + * Additionally restrict APE1 to user-mode addresses. + */ + + uint64_t base = (uintptr_t)alternate_aperture_base; + uint64_t limit = base + alternate_aperture_size - 1; + + if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || + (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { + retval = false; + goto out; + } + + qpd->sh_mem_ape1_base = base >> 16; + qpd->sh_mem_ape1_limit = limit >> 16; + } default_mtype = (default_policy == cache_policy_coherent) ? MTYPE_NONCACHED : @@ -97,37 +135,22 @@ static bool set_cache_memory_policy_cik(struct device_queue_manager *dqm, | ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED) | DEFAULT_MTYPE(default_mtype) | APE1_MTYPE(ape1_mtype); - - return true; -} - -static int update_qpd_cik(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) -{ - struct kfd_process_device *pdd; - unsigned int temp; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - ALIGNMENT_MODE(SH_MEM_ALIGNMENT_MODE_UNALIGNED) | - DEFAULT_MTYPE(MTYPE_NONCACHED) | - APE1_MTYPE(MTYPE_NONCACHED); - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - /* On dGPU we're always in GPUVM64 addressing mode with 64-bit * aperture addresses. */ - temp = get_sh_mem_bases_nybble_64(pdd); + temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd)); qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp); pr_debug("is32bit process: %d sh_mem_bases nybble: 0x%X and register 0x%X\n", qpd->pqm->process->is_32bit_user_mode, temp, qpd->sh_mem_bases); +out: + return retval; +} + +static int update_qpd_cik(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c index 245a90dfc2f6..b5f5f141353b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v10.c @@ -31,10 +31,17 @@ static int update_qpd_v10(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v10(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v10( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v10; asic_ops->update_qpd = update_qpd_v10; asic_ops->init_sdma_vm = init_sdma_vm_v10; asic_ops->mqd_manager_init = mqd_manager_init_v10; @@ -49,27 +56,27 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v10(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v10(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c index 2e129da7acb4..f436878d0d62 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v11.c @@ -30,10 +30,17 @@ static int update_qpd_v11(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v11(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v11( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v11; asic_ops->update_qpd = update_qpd_v11; asic_ops->init_sdma_vm = init_sdma_vm_v11; asic_ops->mqd_manager_init = mqd_manager_init_v11; @@ -48,28 +55,28 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v11(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v11(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c index 4f3295b29dfb..62ca1c8fcbaf 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v12.c @@ -30,10 +30,17 @@ static int update_qpd_v12(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v12(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v12( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v12; asic_ops->update_qpd = update_qpd_v12; asic_ops->init_sdma_vm = init_sdma_vm_v12; asic_ops->mqd_manager_init = mqd_manager_init_v12; @@ -48,28 +55,28 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } -static int update_qpd_v12(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) +static bool set_cache_memory_policy_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) { - struct kfd_process_device *pdd; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - (SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | - (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } + qpd->sh_mem_config = (SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | + (3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT); - qpd->sh_mem_bases = compute_sh_mem_bases_64bit(pdd); + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); pr_debug("sh_mem_bases 0x%X\n", qpd->sh_mem_bases); + return true; +} +static int update_qpd_v12(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c index 67137e674f1d..d85eadaa1e11 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_v9.c @@ -30,10 +30,17 @@ static int update_qpd_v9(struct device_queue_manager *dqm, struct qcm_process_device *qpd); static void init_sdma_vm_v9(struct device_queue_manager *dqm, struct queue *q, struct qcm_process_device *qpd); +static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size); void device_queue_manager_init_v9( struct device_queue_manager_asic_ops *asic_ops) { + asic_ops->set_cache_memory_policy = set_cache_memory_policy_v9; asic_ops->update_qpd = update_qpd_v9; asic_ops->init_sdma_vm = init_sdma_vm_v9; asic_ops->mqd_manager_init = mqd_manager_init_v9; @@ -48,10 +55,36 @@ static uint32_t compute_sh_mem_bases_64bit(struct kfd_process_device *pdd) private_base; } +static bool set_cache_memory_policy_v9(struct device_queue_manager *dqm, + struct qcm_process_device *qpd, + enum cache_policy default_policy, + enum cache_policy alternate_policy, + void __user *alternate_aperture_base, + uint64_t alternate_aperture_size) +{ + qpd->sh_mem_config = SH_MEM_ALIGNMENT_MODE_UNALIGNED << + SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT; + + if (dqm->dev->kfd->noretry) + qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; + + if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) || + KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4)) + qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT); + + qpd->sh_mem_ape1_limit = 0; + qpd->sh_mem_ape1_base = 0; + qpd->sh_mem_bases = compute_sh_mem_bases_64bit(qpd_to_pdd(qpd)); + + pr_debug("sh_mem_bases 0x%X sh_mem_config 0x%X\n", qpd->sh_mem_bases, + qpd->sh_mem_config); + return true; +} + static int update_qpd_v9(struct device_queue_manager *dqm, struct qcm_process_device *qpd) { - struct kfd_process_device *pdd; + struct kfd_process_device *pdd = qpd_to_pdd(qpd); pdd = qpd_to_pdd(qpd); @@ -64,8 +97,7 @@ static int update_qpd_v9(struct device_queue_manager *dqm, qpd->sh_mem_config |= 1 << SH_MEM_CONFIG__RETRY_DISABLE__SHIFT; if (KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 3) || - KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4) || - KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 5, 0)) + KFD_GC_VERSION(dqm->dev->kfd) == IP_VERSION(9, 4, 4)) qpd->sh_mem_config |= (1 << SH_MEM_CONFIG__F8_MODE__SHIFT); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c index b291ee0fab94..320518f41890 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager_vi.c @@ -27,6 +27,14 @@ #include "gca/gfx_8_0_sh_mask.h" #include "oss/oss_3_0_sh_mask.h" +/* + * Low bits must be 0000/FFFF as required by HW, high bits must be 0 to + * stay in user mode. + */ +#define APE1_FIXED_BITS_MASK 0xFFFF80000000FFFFULL +/* APE1 limit is inclusive and 64K aligned. */ +#define APE1_LIMIT_ALIGNMENT 0xFFFF + static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, struct qcm_process_device *qpd, enum cache_policy default_policy, @@ -85,6 +93,36 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, { uint32_t default_mtype; uint32_t ape1_mtype; + unsigned int temp; + bool retval = true; + + if (alternate_aperture_size == 0) { + /* base > limit disables APE1 */ + qpd->sh_mem_ape1_base = 1; + qpd->sh_mem_ape1_limit = 0; + } else { + /* + * In FSA64, APE1_Base[63:0] = { 16{SH_MEM_APE1_BASE[31]}, + * SH_MEM_APE1_BASE[31:0], 0x0000 } + * APE1_Limit[63:0] = { 16{SH_MEM_APE1_LIMIT[31]}, + * SH_MEM_APE1_LIMIT[31:0], 0xFFFF } + * Verify that the base and size parameters can be + * represented in this format and convert them. + * Additionally restrict APE1 to user-mode addresses. + */ + + uint64_t base = (uintptr_t)alternate_aperture_base; + uint64_t limit = base + alternate_aperture_size - 1; + + if (limit <= base || (base & APE1_FIXED_BITS_MASK) != 0 || + (limit & APE1_FIXED_BITS_MASK) != APE1_LIMIT_ALIGNMENT) { + retval = false; + goto out; + } + + qpd->sh_mem_ape1_base = base >> 16; + qpd->sh_mem_ape1_limit = limit >> 16; + } default_mtype = (default_policy == cache_policy_coherent) ? MTYPE_UC : @@ -100,40 +138,21 @@ static bool set_cache_memory_policy_vi(struct device_queue_manager *dqm, default_mtype << SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT | ape1_mtype << SH_MEM_CONFIG__APE1_MTYPE__SHIFT; - return true; -} - -static int update_qpd_vi(struct device_queue_manager *dqm, - struct qcm_process_device *qpd) -{ - struct kfd_process_device *pdd; - unsigned int temp; - - pdd = qpd_to_pdd(qpd); - - /* check if sh_mem_config register already configured */ - if (qpd->sh_mem_config == 0) { - qpd->sh_mem_config = - SH_MEM_ALIGNMENT_MODE_UNALIGNED << - SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT | - MTYPE_UC << - SH_MEM_CONFIG__DEFAULT_MTYPE__SHIFT | - MTYPE_UC << - SH_MEM_CONFIG__APE1_MTYPE__SHIFT; - - qpd->sh_mem_ape1_limit = 0; - qpd->sh_mem_ape1_base = 0; - } - /* On dGPU we're always in GPUVM64 addressing mode with 64-bit * aperture addresses. */ - temp = get_sh_mem_bases_nybble_64(pdd); + temp = get_sh_mem_bases_nybble_64(qpd_to_pdd(qpd)); qpd->sh_mem_bases = compute_sh_mem_bases_64bit(temp); pr_debug("sh_mem_bases nybble: 0x%X and register 0x%X\n", temp, qpd->sh_mem_bases); +out: + return retval; +} +static int update_qpd_vi(struct device_queue_manager *dqm, + struct qcm_process_device *qpd) +{ return 0; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_events.c b/drivers/gpu/drm/amd/amdkfd/kfd_events.c index d075f24e5f9f..fecdb6794075 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c @@ -727,7 +727,7 @@ void kfd_signal_event_interrupt(u32 pasid, uint32_t partial_id, * to process context, kfd_process could attempt to exit while we are * running so the lookup function increments the process ref count. */ - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL); if (!p) return; /* Presumably process exited. */ @@ -1139,8 +1139,8 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p, if (type == KFD_EVENT_TYPE_MEMORY) { dev_warn(kfd_device, - "Sending SIGSEGV to process %d (pasid 0x%x)", - p->lead_thread->pid, p->pasid); + "Sending SIGSEGV to process pid %d", + p->lead_thread->pid); send_sig(SIGSEGV, p->lead_thread, 0); } @@ -1148,13 +1148,13 @@ static void lookup_events_by_type_and_signal(struct kfd_process *p, if (send_signal) { if (send_sigterm) { dev_warn(kfd_device, - "Sending SIGTERM to process %d (pasid 0x%x)", - p->lead_thread->pid, p->pasid); + "Sending SIGTERM to process pid %d", + p->lead_thread->pid); send_sig(SIGTERM, p->lead_thread, 0); } else { dev_err(kfd_device, - "Process %d (pasid 0x%x) got unhandled exception", - p->lead_thread->pid, p->pasid); + "Process pid %d got unhandled exception", + p->lead_thread->pid); } } @@ -1168,7 +1168,7 @@ void kfd_signal_hw_exception_event(u32 pasid) * to process context, kfd_process could attempt to exit while we are * running so the lookup function increments the process ref count. */ - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL); if (!p) return; /* Presumably process exited. */ @@ -1177,22 +1177,20 @@ void kfd_signal_hw_exception_event(u32 pasid) kfd_unref_process(p); } -void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid, +void kfd_signal_vm_fault_event(struct kfd_process_device *pdd, struct kfd_vm_fault_info *info, struct kfd_hsa_memory_exception_data *data) { struct kfd_event *ev; uint32_t id; - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = pdd->process; struct kfd_hsa_memory_exception_data memory_exception_data; int user_gpu_id; - if (!p) - return; /* Presumably process exited. */ - - user_gpu_id = kfd_process_get_user_gpu_id(p, dev->id); + user_gpu_id = kfd_process_get_user_gpu_id(p, pdd->dev->id); if (unlikely(user_gpu_id == -EINVAL)) { - WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", dev->id); + WARN_ONCE(1, "Could not get user_gpu_id from dev->id:%x\n", + pdd->dev->id); return; } @@ -1229,7 +1227,6 @@ void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid, } rcu_read_unlock(); - kfd_unref_process(p); } void kfd_signal_reset_event(struct kfd_node *dev) @@ -1264,7 +1261,8 @@ void kfd_signal_reset_event(struct kfd_node *dev) } if (unlikely(!pdd)) { - WARN_ONCE(1, "Could not get device data from pasid:0x%x\n", p->pasid); + WARN_ONCE(1, "Could not get device data from process pid:%d\n", + p->lead_thread->pid); continue; } @@ -1273,8 +1271,15 @@ void kfd_signal_reset_event(struct kfd_node *dev) if (dev->dqm->detect_hang_count) { struct amdgpu_task_info *ti; + struct amdgpu_fpriv *drv_priv; + + if (unlikely(amdgpu_file_to_fpriv(pdd->drm_file, &drv_priv))) { + WARN_ONCE(1, "Could not get vm for device %x from pid:%d\n", + dev->id, p->lead_thread->pid); + continue; + } - ti = amdgpu_vm_get_task_info_pasid(dev->adev, p->pasid); + ti = amdgpu_vm_get_task_info_vm(&drv_priv->vm); if (ti) { dev_err(dev->adev->dev, "Queues reset on process %s tid %d thread %s pid %d\n", @@ -1311,7 +1316,7 @@ void kfd_signal_reset_event(struct kfd_node *dev) void kfd_signal_poison_consumed_event(struct kfd_node *dev, u32 pasid) { - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL); struct kfd_hsa_memory_exception_data memory_exception_data; struct kfd_hsa_hw_exception_data hw_exception_data; struct kfd_event *ev; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c index b3f988b275a8..c5f97e6e36ff 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v11.c @@ -194,7 +194,7 @@ static void event_interrupt_poison_consumption_v11(struct kfd_node *dev, enum amdgpu_ras_block block = 0; int ret = -EINVAL; uint32_t reset = 0; - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL); if (!p) return; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c index 0cb5c582ce7d..b8a91bf4ef30 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_int_process_v9.c @@ -146,7 +146,7 @@ static void event_interrupt_poison_consumption_v9(struct kfd_node *dev, { enum amdgpu_ras_block block = 0; uint32_t reset = 0; - struct kfd_process *p = kfd_lookup_process_by_pasid(pasid); + struct kfd_process *p = kfd_lookup_process_by_pasid(pasid, NULL); enum ras_event_type type = RAS_EVENT_TYPE_POISON_CONSUMPTION; u64 event_id; int old_poison, ret; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c index 1f9f5bfeaf86..d56525201155 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_v9.c @@ -47,7 +47,7 @@ static int pm_map_process_v9(struct packet_manager *pm, packet->bitfields2.exec_cleaner_shader = 1; packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0; packet->bitfields2.process_quantum = 10; - packet->bitfields2.pasid = qpd->pqm->process->pasid; + packet->bitfields2.pasid = pdd->pasid; packet->bitfields14.gds_size = qpd->gds_size & 0x3F; packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF; packet->bitfields14.num_gws = (qpd->mapped_gws_queue) ? qpd->num_gws : 0; @@ -106,7 +106,7 @@ static int pm_map_process_aldebaran(struct packet_manager *pm, packet->bitfields2.exec_cleaner_shader = 1; packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0; packet->bitfields2.process_quantum = 10; - packet->bitfields2.pasid = qpd->pqm->process->pasid; + packet->bitfields2.pasid = pdd->pasid; packet->bitfields14.gds_size = qpd->gds_size & 0x3F; packet->bitfields14.gds_size_hi = (qpd->gds_size >> 6) & 0xF; packet->bitfields14.num_gws = (qpd->mapped_gws_queue) ? qpd->num_gws : 0; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c index c1199d06d131..347c86e1c378 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager_vi.c @@ -42,6 +42,7 @@ unsigned int pm_build_pm4_header(unsigned int opcode, size_t packet_size) static int pm_map_process_vi(struct packet_manager *pm, uint32_t *buffer, struct qcm_process_device *qpd) { + struct kfd_process_device *pdd = qpd_to_pdd(qpd); struct pm4_mes_map_process *packet; packet = (struct pm4_mes_map_process *)buffer; @@ -52,7 +53,7 @@ static int pm_map_process_vi(struct packet_manager *pm, uint32_t *buffer, sizeof(struct pm4_mes_map_process)); packet->bitfields2.diq_enable = (qpd->is_debug) ? 1 : 0; packet->bitfields2.process_quantum = 10; - packet->bitfields2.pasid = qpd->pqm->process->pasid; + packet->bitfields2.pasid = pdd->pasid; packet->bitfields3.page_table_base = qpd->page_table_base; packet->bitfields10.gds_size = qpd->gds_size; packet->bitfields10.num_gws = qpd->num_gws; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index d8cd913aa772..0a99c5c9cadc 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -851,6 +851,8 @@ struct kfd_process_device { /* Tracks queue reset status */ bool has_reset_queue; + + u32 pasid; }; #define qpd_to_pdd(x) container_of(x, struct kfd_process_device, qpd) @@ -910,8 +912,6 @@ struct kfd_process { /* We want to receive a notification when the mm_struct is destroyed */ struct mmu_notifier mmu_notifier; - u32 pasid; - /* * Array of kfd_process_device pointers, * one for each device the process is using. @@ -1039,7 +1039,8 @@ void kfd_process_destroy_wq(void); void kfd_cleanup_processes(void); struct kfd_process *kfd_create_process(struct task_struct *thread); struct kfd_process *kfd_get_process(const struct task_struct *task); -struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid); +struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid, + struct kfd_process_device **pdd); struct kfd_process *kfd_lookup_process_by_mm(const struct mm_struct *mm); int kfd_process_gpuidx_from_gpuid(struct kfd_process *p, uint32_t gpu_id); @@ -1337,7 +1338,7 @@ void device_queue_manager_uninit(struct device_queue_manager *dqm); struct kernel_queue *kernel_queue_init(struct kfd_node *dev, enum kfd_queue_type type); void kernel_queue_uninit(struct kernel_queue *kq); -int kfd_dqm_evict_pasid(struct device_queue_manager *dqm, u32 pasid); +int kfd_evict_process_device(struct kfd_process_device *pdd); int kfd_dqm_suspend_bad_queue_mes(struct kfd_node *knode, u32 pasid, u32 doorbell_id); /* Process Queue Manager */ @@ -1492,7 +1493,7 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p, int kfd_get_num_events(struct kfd_process *p); int kfd_event_destroy(struct kfd_process *p, uint32_t event_id); -void kfd_signal_vm_fault_event(struct kfd_node *dev, u32 pasid, +void kfd_signal_vm_fault_event(struct kfd_process_device *pdd, struct kfd_vm_fault_info *info, struct kfd_hsa_memory_exception_data *data); diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process.c b/drivers/gpu/drm/amd/amdkfd/kfd_process.c index c3f2c0428e01..7c0c24732481 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process.c @@ -283,8 +283,8 @@ static int kfd_get_cu_occupancy(struct attribute *attr, char *buffer) cu_cnt = 0; proc = pdd->process; if (pdd->qpd.queue_count == 0) { - pr_debug("Gpu-Id: %d has no active queues for process %d\n", - dev->id, proc->pasid); + pr_debug("Gpu-Id: %d has no active queues for process pid %d\n", + dev->id, (int)proc->lead_thread->pid); return snprintf(buffer, PAGE_SIZE, "%d\n", cu_cnt); } @@ -328,12 +328,9 @@ static int kfd_get_cu_occupancy(struct attribute *attr, char *buffer) static ssize_t kfd_procfs_show(struct kobject *kobj, struct attribute *attr, char *buffer) { - if (strcmp(attr->name, "pasid") == 0) { - struct kfd_process *p = container_of(attr, struct kfd_process, - attr_pasid); - - return snprintf(buffer, PAGE_SIZE, "%d\n", p->pasid); - } else if (strncmp(attr->name, "vram_", 5) == 0) { + if (strcmp(attr->name, "pasid") == 0) + return snprintf(buffer, PAGE_SIZE, "%d\n", 0); + else if (strncmp(attr->name, "vram_", 5) == 0) { struct kfd_process_device *pdd = container_of(attr, struct kfd_process_device, attr_vram); return snprintf(buffer, PAGE_SIZE, "%llu\n", atomic64_read(&pdd->vram_usage)); @@ -842,6 +839,14 @@ struct kfd_process *kfd_create_process(struct task_struct *thread) return ERR_PTR(-EINVAL); } + /* If the process just called exec(3), it is possible that the + * cleanup of the kfd_process (following the release of the mm + * of the old process image) is still in the cleanup work queue. + * Make sure to drain any job before trying to recreate any + * resource for this process. + */ + flush_workqueue(kfd_process_wq); + /* * take kfd processes mutex before starting of process creation * so there won't be a case where two threads of the same process @@ -862,14 +867,6 @@ struct kfd_process *kfd_create_process(struct task_struct *thread) if (process) { pr_debug("Process already found\n"); } else { - /* If the process just called exec(3), it is possible that the - * cleanup of the kfd_process (following the release of the mm - * of the old process image) is still in the cleanup work queue. - * Make sure to drain any job before trying to recreate any - * resource for this process. - */ - flush_workqueue(kfd_process_wq); - process = create_process(thread); if (IS_ERR(process)) goto out; @@ -1057,17 +1054,13 @@ static void kfd_process_destroy_pdds(struct kfd_process *p) for (i = 0; i < p->n_pdds; i++) { struct kfd_process_device *pdd = p->pdds[i]; - pr_debug("Releasing pdd (topology id %d) for process (pasid 0x%x)\n", - pdd->dev->id, p->pasid); - + pr_debug("Releasing pdd (topology id %d, for pid %d)\n", + pdd->dev->id, p->lead_thread->pid); kfd_process_device_destroy_cwsr_dgpu(pdd); kfd_process_device_destroy_ib_mem(pdd); - if (pdd->drm_file) { - amdgpu_amdkfd_gpuvm_release_process_vm( - pdd->dev->adev, pdd->drm_priv); + if (pdd->drm_file) fput(pdd->drm_file); - } if (pdd->qpd.cwsr_kaddr && !pdd->qpd.cwsr_base) free_pages((unsigned long)pdd->qpd.cwsr_kaddr, @@ -1191,7 +1184,6 @@ static void kfd_process_wq_release(struct work_struct *work) kfd_event_free_process(p); - kfd_pasid_free(p->pasid); mutex_destroy(&p->mutex); put_task_struct(p->lead_thread); @@ -1542,12 +1534,6 @@ static struct kfd_process *create_process(const struct task_struct *thread) atomic_set(&process->debugged_process_count, 0); sema_init(&process->runtime_enable_sema, 0); - process->pasid = kfd_pasid_alloc(); - if (process->pasid == 0) { - err = -ENOSPC; - goto err_alloc_pasid; - } - err = pqm_init(&process->pqm, process); if (err != 0) goto err_process_pqm_init; @@ -1601,8 +1587,6 @@ err_init_svm_range_list: err_init_apertures: pqm_uninit(&process->pqm); err_process_pqm_init: - kfd_pasid_free(process->pasid); -err_alloc_pasid: kfd_event_free_process(process); err_event_init: mutex_destroy(&process->mutex); @@ -1721,15 +1705,19 @@ int kfd_process_device_init_vm(struct kfd_process_device *pdd, if (ret) goto err_init_cwsr; - ret = amdgpu_amdkfd_gpuvm_set_vm_pasid(dev->adev, avm, p->pasid); - if (ret) - goto err_set_pasid; + if (unlikely(!avm->pasid)) { + dev_warn(pdd->dev->adev->dev, "WARN: vm %p has no pasid associated", + avm); + ret = -EINVAL; + goto err_get_pasid; + } + pdd->pasid = avm->pasid; pdd->drm_file = drm_file; return 0; -err_set_pasid: +err_get_pasid: kfd_process_device_destroy_cwsr_dgpu(pdd); err_init_cwsr: kfd_process_device_destroy_ib_mem(pdd); @@ -1815,25 +1803,50 @@ void kfd_process_device_remove_obj_handle(struct kfd_process_device *pdd, idr_remove(&pdd->alloc_idr, handle); } -/* This increments the process->ref counter. */ -struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid) +static struct kfd_process_device *kfd_lookup_process_device_by_pasid(u32 pasid) { - struct kfd_process *p, *ret_p = NULL; + struct kfd_process_device *ret_p = NULL; + struct kfd_process *p; unsigned int temp; - - int idx = srcu_read_lock(&kfd_processes_srcu); + int i; hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { - if (p->pasid == pasid) { - kref_get(&p->ref); - ret_p = p; - break; + for (i = 0; i < p->n_pdds; i++) { + if (p->pdds[i]->pasid == pasid) { + ret_p = p->pdds[i]; + break; + } } + if (ret_p) + break; + } + return ret_p; +} + +/* This increments the process->ref counter. */ +struct kfd_process *kfd_lookup_process_by_pasid(u32 pasid, + struct kfd_process_device **pdd) +{ + struct kfd_process_device *ret_p; + + int idx = srcu_read_lock(&kfd_processes_srcu); + + ret_p = kfd_lookup_process_device_by_pasid(pasid); + if (ret_p) { + if (pdd) + *pdd = ret_p; + kref_get(&ret_p->process->ref); + + srcu_read_unlock(&kfd_processes_srcu, idx); + return ret_p->process; } srcu_read_unlock(&kfd_processes_srcu, idx); - return ret_p; + if (pdd) + *pdd = NULL; + + return NULL; } /* This increments the process->ref counter. */ @@ -1989,7 +2002,7 @@ static void evict_process_worker(struct work_struct *work) */ p = container_of(dwork, struct kfd_process, eviction_work); - pr_debug("Started evicting pasid 0x%x\n", p->pasid); + pr_debug("Started evicting process pid %d\n", p->lead_thread->pid); ret = kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_TTM); if (!ret) { /* If another thread already signaled the eviction fence, @@ -2001,9 +2014,9 @@ static void evict_process_worker(struct work_struct *work) msecs_to_jiffies(PROCESS_RESTORE_TIME_MS))) kfd_process_restore_queues(p); - pr_debug("Finished evicting pasid 0x%x\n", p->pasid); + pr_debug("Finished evicting process pid %d\n", p->lead_thread->pid); } else - pr_err("Failed to evict queues of pasid 0x%x\n", p->pasid); + pr_err("Failed to evict queues of process pid %d\n", p->lead_thread->pid); } static int restore_process_helper(struct kfd_process *p) @@ -2020,9 +2033,11 @@ static int restore_process_helper(struct kfd_process *p) ret = kfd_process_restore_queues(p); if (!ret) - pr_debug("Finished restoring pasid 0x%x\n", p->pasid); + pr_debug("Finished restoring process pid %d\n", + p->lead_thread->pid); else - pr_err("Failed to restore queues of pasid 0x%x\n", p->pasid); + pr_err("Failed to restore queues of process pid %d\n", + p->lead_thread->pid); return ret; } @@ -2039,7 +2054,7 @@ static void restore_process_worker(struct work_struct *work) * lifetime of this thread, kfd_process p will be valid */ p = container_of(dwork, struct kfd_process, restore_work); - pr_debug("Started restoring pasid 0x%x\n", p->pasid); + pr_debug("Started restoring process pasid %d\n", (int)p->lead_thread->pid); /* Setting last_restore_timestamp before successful restoration. * Otherwise this would have to be set by KGD (restore_process_bos) @@ -2055,8 +2070,8 @@ static void restore_process_worker(struct work_struct *work) ret = restore_process_helper(p); if (ret) { - pr_debug("Failed to restore BOs of pasid 0x%x, retry after %d ms\n", - p->pasid, PROCESS_BACK_OFF_TIME_MS); + pr_debug("Failed to restore BOs of process pid %d, retry after %d ms\n", + p->lead_thread->pid, PROCESS_BACK_OFF_TIME_MS); if (mod_delayed_work(kfd_restore_wq, &p->restore_work, msecs_to_jiffies(PROCESS_RESTORE_TIME_MS))) kfd_process_restore_queues(p); @@ -2072,7 +2087,7 @@ void kfd_suspend_all_processes(void) WARN(debug_evictions, "Evicting all processes"); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { if (kfd_process_evict_queues(p, KFD_QUEUE_EVICTION_TRIGGER_SUSPEND)) - pr_err("Failed to suspend process 0x%x\n", p->pasid); + pr_err("Failed to suspend process pid %d\n", p->lead_thread->pid); signal_eviction_fence(p); } srcu_read_unlock(&kfd_processes_srcu, idx); @@ -2086,8 +2101,8 @@ int kfd_resume_all_processes(void) hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { if (restore_process_helper(p)) { - pr_err("Restore process %d failed during resume\n", - p->pasid); + pr_err("Restore process pid %d failed during resume\n", + p->lead_thread->pid); ret = -EFAULT; } } @@ -2142,7 +2157,7 @@ int kfd_process_drain_interrupts(struct kfd_process_device *pdd) memset(irq_drain_fence, 0, sizeof(irq_drain_fence)); irq_drain_fence[0] = (KFD_IRQ_FENCE_SOURCEID << 8) | KFD_IRQ_FENCE_CLIENTID; - irq_drain_fence[3] = pdd->process->pasid; + irq_drain_fence[3] = pdd->pasid; /* * For GFX 9.4.3/9.5.0, send the NodeId also in IH cookie DW[3] @@ -2173,7 +2188,7 @@ void kfd_process_close_interrupt_drain(unsigned int pasid) { struct kfd_process *p; - p = kfd_lookup_process_by_pasid(pasid); + p = kfd_lookup_process_by_pasid(pasid, NULL); if (!p) return; @@ -2294,8 +2309,8 @@ int kfd_debugfs_mqds_by_process(struct seq_file *m, void *data) int idx = srcu_read_lock(&kfd_processes_srcu); hash_for_each_rcu(kfd_processes_table, temp, p, kfd_processes) { - seq_printf(m, "Process %d PASID 0x%x:\n", - p->lead_thread->tgid, p->pasid); + seq_printf(m, "Process %d PASID %d:\n", + p->lead_thread->tgid, p->lead_thread->pid); mutex_lock(&p->mutex); r = pqm_debugfs_mqds(m, &p->pqm); 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 d79caa1a6867..662c595ce783 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c @@ -69,8 +69,8 @@ static int find_available_queue_slot(struct process_queue_manager *pqm, pr_debug("The new slot id %lu\n", found); if (found >= KFD_MAX_NUM_OF_QUEUES_PER_PROCESS) { - pr_info("Cannot open more queues for process with pasid 0x%x\n", - pqm->process->pasid); + pr_info("Cannot open more queues for process with pid %d\n", + pqm->process->lead_thread->pid); return -ENOMEM; } @@ -451,8 +451,8 @@ int pqm_create_queue(struct process_queue_manager *pqm, } if (retval != 0) { - pr_err("Pasid 0x%x DQM create queue type %d failed. ret %d\n", - pqm->process->pasid, type, retval); + pr_err("process pid %d DQM create queue type %d failed. ret %d\n", + pqm->process->lead_thread->pid, type, retval); goto err_create_queue; } @@ -546,7 +546,7 @@ int pqm_destroy_queue(struct process_queue_manager *pqm, unsigned int qid) retval = dqm->ops.destroy_queue(dqm, &pdd->qpd, pqn->q); if (retval) { pr_err("Pasid 0x%x destroy queue %d failed, ret %d\n", - pqm->process->pasid, + pdd->pasid, pqn->q->properties.queue_id, retval); if (retval != -ETIME && retval != -EIO) goto err_destroy_queue; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c index d1cf9dd35290..116116a9f578 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c @@ -563,7 +563,8 @@ svm_range_vram_node_new(struct kfd_node *node, struct svm_range *prange, int r; p = container_of(prange->svms, struct kfd_process, svms); - pr_debug("pasid: %x svms 0x%p [0x%lx 0x%lx]\n", p->pasid, prange->svms, + pr_debug("process pid: %d svms 0x%p [0x%lx 0x%lx]\n", + p->lead_thread->pid, prange->svms, prange->start, prange->last); if (svm_range_validate_svm_bo(node, prange)) @@ -2973,7 +2974,7 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid, return -EFAULT; } - p = kfd_lookup_process_by_pasid(pasid); + p = kfd_lookup_process_by_pasid(pasid, NULL); if (!p) { pr_debug("kfd process not founded pasid 0x%x\n", pasid); return 0; @@ -3024,7 +3025,7 @@ retry_write_locked: /* check if this page fault time stamp is before svms->checkpoint_ts */ if (svms->checkpoint_ts[gpuidx] != 0) { - if (amdgpu_ih_ts_after(ts, svms->checkpoint_ts[gpuidx])) { + if (amdgpu_ih_ts_after_or_equal(ts, svms->checkpoint_ts[gpuidx])) { pr_debug("draining retry fault, drop fault 0x%llx\n", addr); r = -EAGAIN; goto out_unlock_svms; @@ -3239,7 +3240,8 @@ void svm_range_list_fini(struct kfd_process *p) struct svm_range *prange; struct svm_range *next; - pr_debug("pasid 0x%x svms 0x%p\n", p->pasid, &p->svms); + pr_debug("process pid %d svms 0x%p\n", p->lead_thread->pid, + &p->svms); cancel_delayed_work_sync(&p->svms.restore_work); @@ -3262,7 +3264,8 @@ void svm_range_list_fini(struct kfd_process *p) mutex_destroy(&p->svms.lock); - pr_debug("pasid 0x%x svms 0x%p done\n", p->pasid, &p->svms); + pr_debug("process pid %d svms 0x%p done\n", + p->lead_thread->pid, &p->svms); } int svm_range_list_init(struct kfd_process *p) @@ -3625,8 +3628,8 @@ svm_range_set_attr(struct kfd_process *p, struct mm_struct *mm, bool flush_tlb; int r, ret = 0; - pr_debug("pasid 0x%x svms 0x%p [0x%llx 0x%llx] pages 0x%llx\n", - p->pasid, &p->svms, start, start + size - 1, size); + pr_debug("process pid %d svms 0x%p [0x%llx 0x%llx] pages 0x%llx\n", + p->lead_thread->pid, &p->svms, start, start + size - 1, size); r = svm_range_check_attr(p, nattr, attrs); if (r) @@ -3734,8 +3737,8 @@ out_unlock_range: out: mutex_unlock(&process_info->lock); - pr_debug("pasid 0x%x svms 0x%p [0x%llx 0x%llx] done, r=%d\n", p->pasid, - &p->svms, start, start + size - 1, r); + pr_debug("process pid %d svms 0x%p [0x%llx 0x%llx] done, r=%d\n", + p->lead_thread->pid, &p->svms, start, start + size - 1, r); return ret ? ret : r; } diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c index ceb9fb475ef1..98317eda2cdb 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_topology.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_topology.c @@ -1683,17 +1683,32 @@ static int fill_in_l2_l3_pcache(struct kfd_cache_properties **props_ext, int cache_type, unsigned int cu_processor_id, struct kfd_node *knode) { - unsigned int cu_sibling_map_mask; + unsigned int cu_sibling_map_mask = 0; int first_active_cu; int i, j, k, xcc, start, end; int num_xcc = NUM_XCC(knode->xcc_mask); struct kfd_cache_properties *pcache = NULL; enum amdgpu_memory_partition mode; struct amdgpu_device *adev = knode->adev; + bool found = false; start = ffs(knode->xcc_mask) - 1; end = start + num_xcc; - cu_sibling_map_mask = cu_info->bitmap[start][0][0]; + + /* To find the bitmap in the first active cu in the first + * xcc, it is based on the assumption that evrey xcc must + * have at least one active cu. + */ + for (i = 0; i < gfx_info->max_shader_engines && !found; i++) { + for (j = 0; j < gfx_info->max_sh_per_se && !found; j++) { + if (cu_info->bitmap[start][i % 4][j % 4]) { + cu_sibling_map_mask = + cu_info->bitmap[start][i % 4][j % 4]; + found = true; + } + } + } + cu_sibling_map_mask &= ((1 << pcache_info[cache_type].num_cu_shared) - 1); first_active_cu = ffs(cu_sibling_map_mask); @@ -2000,15 +2015,12 @@ static void kfd_topology_set_capabilities(struct kfd_topology_device *dev) dev->node_props.capability |= HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED; - dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED; + if (!amdgpu_sriov_vf(dev->gpu->adev)) + dev->node_props.capability |= HSA_CAP_PER_QUEUE_RESET_SUPPORTED; } else { dev->node_props.debug_prop |= HSA_DBG_WATCH_ADDR_MASK_LO_BIT_GFX10 | HSA_DBG_WATCH_ADDR_MASK_HI_BIT; - if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(11, 0, 0)) - dev->node_props.capability |= - HSA_CAP_TRAP_DEBUG_PRECISE_MEMORY_OPERATIONS_SUPPORTED; - if (KFD_GC_VERSION(dev->gpu) >= IP_VERSION(12, 0, 0)) dev->node_props.capability |= HSA_CAP_TRAP_DEBUG_PRECISE_ALU_OPERATIONS_SUPPORTED; 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 80a3cbd2cbe5..4801dcde2cb3 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -367,6 +367,8 @@ get_crtc_by_otg_inst(struct amdgpu_device *adev, static inline bool is_dc_timing_adjust_needed(struct dm_crtc_state *old_state, struct dm_crtc_state *new_state) { + if (new_state->stream->adjust.timing_adjust_pending) + return true; if (new_state->freesync_config.state == VRR_STATE_ACTIVE_FIXED) return true; else if (amdgpu_dm_crtc_vrr_active(old_state) != amdgpu_dm_crtc_vrr_active(new_state)) @@ -1912,26 +1914,6 @@ static enum dmub_ips_disable_type dm_get_default_ips_mode( switch (amdgpu_ip_version(adev, DCE_HWIP, 0)) { case IP_VERSION(3, 5, 0): - /* - * On DCN35 systems with Z8 enabled, it's possible for IPS2 + Z8 to - * cause a hard hang. A fix exists for newer PMFW. - * - * As a workaround, for non-fixed PMFW, force IPS1+RCG as the deepest - * IPS state in all cases, except for s0ix and all displays off (DPMS), - * where IPS2 is allowed. - * - * When checking pmfw version, use the major and minor only. - */ - if ((adev->pm.fw_version & 0x00FFFF00) < 0x005D6300) - ret = DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF; - else if (amdgpu_ip_version(adev, GC_HWIP, 0) > IP_VERSION(11, 5, 0)) - /* - * Other ASICs with DCN35 that have residency issues with - * IPS2 in idle. - * We want them to use IPS2 only in display off cases. - */ - ret = DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF; - break; case IP_VERSION(3, 5, 1): ret = DMUB_IPS_RCG_IN_ACTIVE_IPS2_IN_OFF; break; @@ -3293,16 +3275,16 @@ static void dm_gpureset_commit_state(struct dc_state *dc_state, for (k = 0; k < dc_state->stream_count; k++) { bundle->stream_update.stream = dc_state->streams[k]; - for (m = 0; m < dc_state->stream_status->plane_count; m++) { + for (m = 0; m < dc_state->stream_status[k].plane_count; m++) { bundle->surface_updates[m].surface = - dc_state->stream_status->plane_states[m]; + dc_state->stream_status[k].plane_states[m]; bundle->surface_updates[m].surface->force_full_update = true; } update_planes_and_stream_adapter(dm->dc, UPDATE_TYPE_FULL, - dc_state->stream_status->plane_count, + dc_state->stream_status[k].plane_count, dc_state->streams[k], &bundle->stream_update, bundle->surface_updates); @@ -3404,11 +3386,6 @@ static int dm_resume(struct amdgpu_ip_block *ip_block) return 0; } - - /* leave display off for S4 sequence */ - if (adev->in_s4) - return 0; - /* Recreate dc_state - DC invalidates it when setting power state to S3. */ dc_state_release(dm_state->context); dm_state->context = dc_state_create(dm->dc, NULL); @@ -5662,9 +5639,9 @@ fill_plane_color_attributes(const struct drm_plane_state *plane_state, case DRM_COLOR_YCBCR_BT2020: if (full_range) - *color_space = COLOR_SPACE_2020_YCBCR; + *color_space = COLOR_SPACE_2020_YCBCR_FULL; else - return -EINVAL; + *color_space = COLOR_SPACE_2020_YCBCR_LIMITED; break; default: @@ -6160,7 +6137,7 @@ get_output_color_space(const struct dc_crtc_timing *dc_crtc_timing, if (dc_crtc_timing->pixel_encoding == PIXEL_ENCODING_RGB) color_space = COLOR_SPACE_2020_RGB_FULLRANGE; else - color_space = COLOR_SPACE_2020_YCBCR; + color_space = COLOR_SPACE_2020_YCBCR_LIMITED; break; case DRM_MODE_COLORIMETRY_DEFAULT: // ITU601 default: @@ -7505,12 +7482,12 @@ cleanup: } struct dc_stream_state * -create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, +create_validate_stream_for_sink(struct drm_connector *connector, const struct drm_display_mode *drm_mode, const struct dm_connector_state *dm_state, const struct dc_stream_state *old_stream) { - struct drm_connector *connector = &aconnector->base; + struct amdgpu_dm_connector *aconnector = NULL; struct amdgpu_device *adev = drm_to_adev(connector->dev); struct dc_stream_state *stream; const struct drm_connector_state *drm_state = dm_state ? &dm_state->base : NULL; @@ -7521,8 +7498,12 @@ create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, if (!dm_state) return NULL; - if (aconnector->dc_link->connector_signal == SIGNAL_TYPE_HDMI_TYPE_A || - aconnector->dc_link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER) + if (connector->connector_type != DRM_MODE_CONNECTOR_WRITEBACK) + aconnector = to_amdgpu_dm_connector(connector); + + if (aconnector && + (aconnector->dc_link->connector_signal == SIGNAL_TYPE_HDMI_TYPE_A || + aconnector->dc_link->dpcd_caps.dongle_type == DISPLAY_DONGLE_DP_HDMI_CONVERTER)) bpc_limit = 8; do { @@ -7534,10 +7515,11 @@ create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, break; } - if (aconnector->base.connector_type == DRM_MODE_CONNECTOR_WRITEBACK) + dc_result = dc_validate_stream(adev->dm.dc, stream); + + if (!aconnector) /* writeback connector */ return stream; - dc_result = dc_validate_stream(adev->dm.dc, stream); if (dc_result == DC_OK && stream->signal == SIGNAL_TYPE_DISPLAY_PORT_MST) dc_result = dm_dp_mst_is_port_support_mode(aconnector, stream); @@ -7567,7 +7549,7 @@ create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, __func__, __LINE__); aconnector->force_yuv420_output = true; - stream = create_validate_stream_for_sink(aconnector, drm_mode, + stream = create_validate_stream_for_sink(connector, drm_mode, dm_state, old_stream); aconnector->force_yuv420_output = false; } @@ -7582,6 +7564,9 @@ enum drm_mode_status amdgpu_dm_connector_mode_valid(struct drm_connector *connec struct dc_sink *dc_sink; /* TODO: Unhardcode stream count */ struct dc_stream_state *stream; + /* we always have an amdgpu_dm_connector here since we got + * here via the amdgpu_dm_connector_helper_funcs + */ struct amdgpu_dm_connector *aconnector = to_amdgpu_dm_connector(connector); if ((mode->flags & DRM_MODE_FLAG_INTERLACE) || @@ -7606,7 +7591,7 @@ enum drm_mode_status amdgpu_dm_connector_mode_valid(struct drm_connector *connec drm_mode_set_crtcinfo(mode, 0); - stream = create_validate_stream_for_sink(aconnector, mode, + stream = create_validate_stream_for_sink(connector, mode, to_dm_connector_state(connector->state), NULL); if (stream) { @@ -8386,7 +8371,7 @@ static int amdgpu_dm_i2c_xfer(struct i2c_adapter *i2c_adap, int i; int result = -EIO; - if (!ddc_service->ddc_pin || !ddc_service->ddc_pin->hw_info.hw_supported) + if (!ddc_service->ddc_pin) return result; cmd.payloads = kcalloc(num, sizeof(struct i2c_payload), GFP_KERNEL); @@ -10658,7 +10643,7 @@ static int dm_update_crtc_state(struct amdgpu_display_manager *dm, if (!drm_atomic_crtc_needs_modeset(new_crtc_state)) goto skip_modeset; - new_stream = create_validate_stream_for_sink(aconnector, + new_stream = create_validate_stream_for_sink(connector, &new_crtc_state->mode, dm_new_conn_state, dm_old_crtc_state->stream); @@ -10901,6 +10886,9 @@ static bool should_reset_plane(struct drm_atomic_state *state, state->allow_modeset) return true; + if (amdgpu_in_reset(adev) && state->allow_modeset) + return true; + /* Exit early if we know that we're adding or removing the plane. */ if (old_plane_state->crtc != new_plane_state->crtc) return true; @@ -12618,7 +12606,7 @@ int amdgpu_dm_process_dmub_aux_transfer_sync( * Transient states before tunneling is enabled could * lead to this error. We can ignore this for now. */ - if (p_notify->result != AUX_RET_ERROR_PROTOCOL_ERROR) { + if (p_notify->result == AUX_RET_ERROR_PROTOCOL_ERROR) { DRM_WARN("DPIA AUX failed on 0x%x(%d), error %d\n", payload->address, payload->length, p_notify->result); @@ -12627,22 +12615,15 @@ int amdgpu_dm_process_dmub_aux_transfer_sync( goto out; } + payload->reply[0] = adev->dm.dmub_notify->aux_reply.command & 0xF; + if (adev->dm.dmub_notify->aux_reply.command & 0xF0) + /* The reply is stored in the top nibble of the command. */ + payload->reply[0] = (adev->dm.dmub_notify->aux_reply.command >> 4) & 0xF; - payload->reply[0] = adev->dm.dmub_notify->aux_reply.command; - if (!payload->write && p_notify->aux_reply.length && - (payload->reply[0] == AUX_TRANSACTION_REPLY_AUX_ACK)) { - - if (payload->length != p_notify->aux_reply.length) { - DRM_WARN("invalid read length %d from DPIA AUX 0x%x(%d)!\n", - p_notify->aux_reply.length, - payload->address, payload->length); - *operation_result = AUX_RET_ERROR_INVALID_REPLY; - goto out; - } - + /*write req may receive a byte indicating partially written number as well*/ + if (p_notify->aux_reply.length) memcpy(payload->data, p_notify->aux_reply.data, p_notify->aux_reply.length); - } /* success */ ret = p_notify->aux_reply.length; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h index d2703ca7dff3..195fec9048df 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h @@ -989,7 +989,7 @@ int amdgpu_dm_process_dmub_set_config_sync(struct dc_context *ctx, unsigned int struct set_config_cmd_payload *payload, enum set_config_status *operation_result); struct dc_stream_state * - create_validate_stream_for_sink(struct amdgpu_dm_connector *aconnector, + create_validate_stream_for_sink(struct drm_connector *connector, const struct drm_display_mode *drm_mode, const struct dm_connector_state *dm_state, const struct dc_stream_state *old_stream); diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c index 049046c60462..c7d13e743e6c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c @@ -1169,7 +1169,7 @@ static int amdgpu_current_colorspace_show(struct seq_file *m, void *data) case COLOR_SPACE_2020_RGB_FULLRANGE: seq_puts(m, "BT2020_RGB"); break; - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: seq_puts(m, "BT2020_YCC"); break; default: diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c index c0dc23244049..10ba4d7bf632 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_hdcp.c @@ -172,7 +172,10 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, struct mod_hdcp_display_adjustment display_adjust; unsigned int conn_index = aconnector->base.index; - mutex_lock(&hdcp_w->mutex); + guard(mutex)(&hdcp_w->mutex); + drm_connector_get(&aconnector->base); + if (hdcp_w->aconnector[conn_index]) + drm_connector_put(&hdcp_w->aconnector[conn_index]->base); hdcp_w->aconnector[conn_index] = aconnector; memset(&link_adjust, 0, sizeof(link_adjust)); @@ -209,7 +212,6 @@ void hdcp_update_display(struct hdcp_workqueue *hdcp_work, mod_hdcp_update_display(&hdcp_w->hdcp, conn_index, &link_adjust, &display_adjust, &hdcp_w->output); process_output(hdcp_w); - mutex_unlock(&hdcp_w->mutex); } static void hdcp_remove_display(struct hdcp_workqueue *hdcp_work, @@ -220,8 +222,7 @@ static void hdcp_remove_display(struct hdcp_workqueue *hdcp_work, struct drm_connector_state *conn_state = aconnector->base.state; unsigned int conn_index = aconnector->base.index; - mutex_lock(&hdcp_w->mutex); - hdcp_w->aconnector[conn_index] = aconnector; + guard(mutex)(&hdcp_w->mutex); /* the removal of display will invoke auth reset -> hdcp destroy and * we'd expect the Content Protection (CP) property changed back to @@ -237,9 +238,11 @@ static void hdcp_remove_display(struct hdcp_workqueue *hdcp_work, } mod_hdcp_remove_display(&hdcp_w->hdcp, aconnector->base.index, &hdcp_w->output); - + if (hdcp_w->aconnector[conn_index]) { + drm_connector_put(&hdcp_w->aconnector[conn_index]->base); + hdcp_w->aconnector[conn_index] = NULL; + } process_output(hdcp_w); - mutex_unlock(&hdcp_w->mutex); } void hdcp_reset_display(struct hdcp_workqueue *hdcp_work, unsigned int link_index) @@ -247,7 +250,7 @@ void hdcp_reset_display(struct hdcp_workqueue *hdcp_work, unsigned int link_inde struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index]; unsigned int conn_index; - mutex_lock(&hdcp_w->mutex); + guard(mutex)(&hdcp_w->mutex); mod_hdcp_reset_connection(&hdcp_w->hdcp, &hdcp_w->output); @@ -256,11 +259,13 @@ void hdcp_reset_display(struct hdcp_workqueue *hdcp_work, unsigned int link_inde for (conn_index = 0; conn_index < AMDGPU_DM_MAX_DISPLAY_INDEX; conn_index++) { hdcp_w->encryption_status[conn_index] = MOD_HDCP_ENCRYPTION_STATUS_HDCP_OFF; + if (hdcp_w->aconnector[conn_index]) { + drm_connector_put(&hdcp_w->aconnector[conn_index]->base); + hdcp_w->aconnector[conn_index] = NULL; + } } process_output(hdcp_w); - - mutex_unlock(&hdcp_w->mutex); } void hdcp_handle_cpirq(struct hdcp_workqueue *hdcp_work, unsigned int link_index) @@ -277,7 +282,7 @@ static void event_callback(struct work_struct *work) hdcp_work = container_of(to_delayed_work(work), struct hdcp_workqueue, callback_dwork); - mutex_lock(&hdcp_work->mutex); + guard(mutex)(&hdcp_work->mutex); cancel_delayed_work(&hdcp_work->callback_dwork); @@ -285,8 +290,6 @@ static void event_callback(struct work_struct *work) &hdcp_work->output); process_output(hdcp_work); - - mutex_unlock(&hdcp_work->mutex); } static void event_property_update(struct work_struct *work) @@ -323,7 +326,7 @@ static void event_property_update(struct work_struct *work) continue; drm_modeset_lock(&dev->mode_config.connection_mutex, NULL); - mutex_lock(&hdcp_work->mutex); + guard(mutex)(&hdcp_work->mutex); if (conn_state->commit) { ret = wait_for_completion_interruptible_timeout(&conn_state->commit->hw_done, @@ -355,7 +358,6 @@ static void event_property_update(struct work_struct *work) drm_hdcp_update_content_protection(connector, DRM_MODE_CONTENT_PROTECTION_DESIRED); } - mutex_unlock(&hdcp_work->mutex); drm_modeset_unlock(&dev->mode_config.connection_mutex); } } @@ -368,7 +370,7 @@ static void event_property_validate(struct work_struct *work) struct amdgpu_dm_connector *aconnector; unsigned int conn_index; - mutex_lock(&hdcp_work->mutex); + guard(mutex)(&hdcp_work->mutex); for (conn_index = 0; conn_index < AMDGPU_DM_MAX_DISPLAY_INDEX; conn_index++) { @@ -408,8 +410,6 @@ static void event_property_validate(struct work_struct *work) schedule_work(&hdcp_work->property_update_work); } } - - mutex_unlock(&hdcp_work->mutex); } static void event_watchdog_timer(struct work_struct *work) @@ -420,7 +420,7 @@ static void event_watchdog_timer(struct work_struct *work) struct hdcp_workqueue, watchdog_timer_dwork); - mutex_lock(&hdcp_work->mutex); + guard(mutex)(&hdcp_work->mutex); cancel_delayed_work(&hdcp_work->watchdog_timer_dwork); @@ -429,8 +429,6 @@ static void event_watchdog_timer(struct work_struct *work) &hdcp_work->output); process_output(hdcp_work); - - mutex_unlock(&hdcp_work->mutex); } static void event_cpirq(struct work_struct *work) @@ -439,13 +437,11 @@ static void event_cpirq(struct work_struct *work) hdcp_work = container_of(work, struct hdcp_workqueue, cpirq_work); - mutex_lock(&hdcp_work->mutex); + guard(mutex)(&hdcp_work->mutex); mod_hdcp_process_event(&hdcp_work->hdcp, MOD_HDCP_EVENT_CPIRQ, &hdcp_work->output); process_output(hdcp_work); - - mutex_unlock(&hdcp_work->mutex); } void hdcp_destroy(struct kobject *kobj, struct hdcp_workqueue *hdcp_work) @@ -479,7 +475,7 @@ static bool enable_assr(void *handle, struct dc_link *link) dtm_cmd = (struct ta_dtm_shared_memory *)psp->dtm_context.context.mem_context.shared_buf; - mutex_lock(&psp->dtm_context.mutex); + guard(mutex)(&psp->dtm_context.mutex); memset(dtm_cmd, 0, sizeof(struct ta_dtm_shared_memory)); dtm_cmd->cmd_id = TA_DTM_COMMAND__TOPOLOGY_ASSR_ENABLE; @@ -494,8 +490,6 @@ static bool enable_assr(void *handle, struct dc_link *link) res = false; } - mutex_unlock(&psp->dtm_context.mutex); - return res; } @@ -504,6 +498,7 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) struct hdcp_workqueue *hdcp_work = handle; struct amdgpu_dm_connector *aconnector = config->dm_stream_ctx; int link_index = aconnector->dc_link->link_index; + unsigned int conn_index = aconnector->base.index; struct mod_hdcp_display *display = &hdcp_work[link_index].display; struct mod_hdcp_link *link = &hdcp_work[link_index].link; struct hdcp_workqueue *hdcp_w = &hdcp_work[link_index]; @@ -557,13 +552,14 @@ static void update_config(void *handle, struct cp_psp_stream_config *config) (!!aconnector->base.state) ? aconnector->base.state->hdcp_content_type : -1); - mutex_lock(&hdcp_w->mutex); + guard(mutex)(&hdcp_w->mutex); mod_hdcp_add_display(&hdcp_w->hdcp, link, display, &hdcp_w->output); - + drm_connector_get(&aconnector->base); + if (hdcp_w->aconnector[conn_index]) + drm_connector_put(&hdcp_w->aconnector[conn_index]->base); + hdcp_w->aconnector[conn_index] = aconnector; process_output(hdcp_w); - mutex_unlock(&hdcp_w->mutex); - } /** diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c index fbd80d8545a8..a2532907c7be 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c @@ -912,7 +912,7 @@ dm_helpers_probe_acpi_edid(void *data, u8 *buf, unsigned int block, size_t len) { struct drm_connector *connector = data; struct acpi_device *acpidev = ACPI_COMPANION(connector->dev->dev); - unsigned char start = block * EDID_LENGTH; + unsigned short start = block * EDID_LENGTH; struct edid *edid; int r; diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 07e744da7bf4..c3759a1c32ce 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -51,6 +51,9 @@ #define PEAK_FACTOR_X1000 1006 +/* + * This function handles both native AUX and I2C-Over-AUX transactions. + */ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, struct drm_dp_aux_msg *msg) { @@ -59,6 +62,7 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, enum aux_return_code_type operation_result; struct amdgpu_device *adev; struct ddc_service *ddc; + uint8_t copy[16]; if (WARN_ON(msg->size > 16)) return -E2BIG; @@ -74,6 +78,11 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, (msg->request & DP_AUX_I2C_WRITE_STATUS_UPDATE) != 0; payload.defer_delay = 0; + if (payload.write) { + memcpy(copy, msg->buffer, msg->size); + payload.data = copy; + } + result = dc_link_aux_transfer_raw(TO_DM_AUX(aux)->ddc_service, &payload, &operation_result); @@ -87,15 +96,25 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, if (adev->dm.aux_hpd_discon_quirk) { if (msg->address == DP_SIDEBAND_MSG_DOWN_REQ_BASE && operation_result == AUX_RET_ERROR_HPD_DISCON) { - result = 0; + result = msg->size; operation_result = AUX_RET_SUCCESS; } } - if (payload.write && result >= 0) - result = msg->size; + /* + * result equals to 0 includes the cases of AUX_DEFER/I2C_DEFER + */ + if (payload.write && result >= 0) { + if (result) { + /*one byte indicating partially written bytes*/ + drm_dbg_dp(adev_to_drm(adev), "amdgpu: AUX partially written\n"); + result = payload.data[0]; + } else if (!payload.reply[0]) + /*I2C_ACK|AUX_ACK*/ + result = msg->size; + } - if (result < 0) + if (result < 0) { switch (operation_result) { case AUX_RET_SUCCESS: break; @@ -114,6 +133,13 @@ static ssize_t dm_dp_aux_transfer(struct drm_dp_aux *aux, break; } + drm_dbg_dp(adev_to_drm(adev), "amdgpu: DP AUX transfer fail:%d\n", operation_result); + } + + if (payload.reply[0]) + drm_dbg_dp(adev_to_drm(adev), "amdgpu: AUX reply command not ACK: 0x%02x.", + payload.reply[0]); + return result; } @@ -1625,7 +1651,6 @@ int pre_validate_dsc(struct drm_atomic_state *state, if (ind >= 0) { struct drm_connector *connector; - struct amdgpu_dm_connector *aconnector; struct drm_connector_state *drm_new_conn_state; struct dm_connector_state *dm_new_conn_state; struct dm_crtc_state *dm_old_crtc_state; @@ -1633,15 +1658,14 @@ int pre_validate_dsc(struct drm_atomic_state *state, connector = amdgpu_dm_find_first_crtc_matching_connector(state, state->crtcs[ind].ptr); - aconnector = to_amdgpu_dm_connector(connector); drm_new_conn_state = drm_atomic_get_new_connector_state(state, - &aconnector->base); + connector); dm_new_conn_state = to_dm_connector_state(drm_new_conn_state); dm_old_crtc_state = to_dm_crtc_state(state->crtcs[ind].old_state); local_dc_state->streams[i] = - create_validate_stream_for_sink(aconnector, + create_validate_stream_for_sink(connector, &state->crtcs[ind].new_state->mode, dm_new_conn_state, dm_old_crtc_state->stream); diff --git a/drivers/gpu/drm/amd/display/dc/basics/dc_common.c b/drivers/gpu/drm/amd/display/dc/basics/dc_common.c index b2fc4f8e6482..a51c2701da24 100644 --- a/drivers/gpu/drm/amd/display/dc/basics/dc_common.c +++ b/drivers/gpu/drm/amd/display/dc/basics/dc_common.c @@ -40,7 +40,8 @@ bool is_rgb_cspace(enum dc_color_space output_color_space) case COLOR_SPACE_YCBCR709: case COLOR_SPACE_YCBCR601_LIMITED: case COLOR_SPACE_YCBCR709_LIMITED: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: + case COLOR_SPACE_2020_YCBCR_FULL: return false; default: /* Add a case to switch */ diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c index 7d18f372ce7a..6bc59b7ef007 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table2.c @@ -101,7 +101,6 @@ static void init_dig_encoder_control(struct bios_parser *bp) bp->cmd_tbl.dig_encoder_control = encoder_control_digx_v1_5; break; default: - dm_output_to_console("Don't have dig_encoder_control for v%d\n", version); bp->cmd_tbl.dig_encoder_control = encoder_control_fallback; break; } @@ -238,7 +237,6 @@ static void init_transmitter_control(struct bios_parser *bp) bp->cmd_tbl.transmitter_control = transmitter_control_v1_7; break; default: - dm_output_to_console("Don't have transmitter_control for v%d\n", crev); bp->cmd_tbl.transmitter_control = transmitter_control_fallback; break; } @@ -408,8 +406,6 @@ static void init_set_pixel_clock(struct bios_parser *bp) bp->cmd_tbl.set_pixel_clock = set_pixel_clock_v7; break; default: - dm_output_to_console("Don't have set_pixel_clock for v%d\n", - BIOS_CMD_TABLE_PARA_REVISION(setpixelclock)); bp->cmd_tbl.set_pixel_clock = set_pixel_clock_fallback; break; } @@ -554,7 +550,6 @@ static void init_set_crtc_timing(struct bios_parser *bp) set_crtc_using_dtd_timing_v3; break; default: - dm_output_to_console("Don't have set_crtc_timing for v%d\n", dtd_version); bp->cmd_tbl.set_crtc_timing = NULL; break; } @@ -671,8 +666,6 @@ static void init_enable_crtc(struct bios_parser *bp) bp->cmd_tbl.enable_crtc = enable_crtc_v1; break; default: - dm_output_to_console("Don't have enable_crtc for v%d\n", - BIOS_CMD_TABLE_PARA_REVISION(enablecrtc)); bp->cmd_tbl.enable_crtc = NULL; break; } @@ -864,8 +857,6 @@ static void init_set_dce_clock(struct bios_parser *bp) bp->cmd_tbl.set_dce_clock = set_dce_clock_v2_1; break; default: - dm_output_to_console("Don't have set_dce_clock for v%d\n", - BIOS_CMD_TABLE_PARA_REVISION(setdceclock)); bp->cmd_tbl.set_dce_clock = NULL; break; } diff --git a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c index 73458e295103..df8139bda142 100644 --- a/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c +++ b/drivers/gpu/drm/amd/display/dc/bios/command_table_helper2.c @@ -87,8 +87,7 @@ bool dal_bios_parser_init_cmd_tbl_helper2( return true; default: - /* Unsupported DCE */ - BREAK_TO_DEBUGGER(); + *h = dal_cmd_tbl_helper_dce112_get_table2(); return false; } } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c index a0fb4481d2f1..e4d22f74f986 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn315/dcn315_clk_mgr.c @@ -130,7 +130,7 @@ static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base, struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk; struct dc *dc = clk_mgr_base->ctx->dc; - int display_count; + int display_count = 0; bool update_dppclk = false; bool update_dispclk = false; bool dpp_clock_lowered = false; @@ -194,8 +194,6 @@ static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base, // workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow. if (new_clocks->dppclk_khz < MIN_DPP_DISP_CLK) new_clocks->dppclk_khz = MIN_DPP_DISP_CLK; - if (new_clocks->dispclk_khz < MIN_DPP_DISP_CLK) - new_clocks->dispclk_khz = MIN_DPP_DISP_CLK; if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) { if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz) @@ -204,15 +202,19 @@ static void dcn315_update_clocks(struct clk_mgr *clk_mgr_base, update_dppclk = true; } - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) { - /* No need to apply the w/a if we haven't taken over from bios yet */ - if (clk_mgr_base->clks.dispclk_khz) - dcn315_disable_otg_wa(clk_mgr_base, context, true); + if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) && + (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) { + int requested_dispclk_khz = new_clocks->dispclk_khz; + dcn315_disable_otg_wa(clk_mgr_base, context, true); + + /* Clamp the requested clock to PMFW based on their limit. */ + if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz) + requested_dispclk_khz = dc->debug.min_disp_clk_khz; + + dcn315_smu_set_dispclk(clk_mgr, requested_dispclk_khz); clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz; - dcn315_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz); - if (clk_mgr_base->clks.dispclk_khz) - dcn315_disable_otg_wa(clk_mgr_base, context, false); + dcn315_disable_otg_wa(clk_mgr_base, context, false); update_dispclk = true; } diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c index c3e50c3aaa60..49efea0c8fcf 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn316/dcn316_clk_mgr.c @@ -140,7 +140,7 @@ static void dcn316_update_clocks(struct clk_mgr *clk_mgr_base, struct clk_mgr_internal *clk_mgr = TO_CLK_MGR_INTERNAL(clk_mgr_base); struct dc_clocks *new_clocks = &context->bw_ctx.bw.dcn.clk; struct dc *dc = clk_mgr_base->ctx->dc; - int display_count; + int display_count = 0; bool update_dppclk = false; bool update_dispclk = false; bool dpp_clock_lowered = false; @@ -201,8 +201,6 @@ static void dcn316_update_clocks(struct clk_mgr *clk_mgr_base, // workaround: Limit dppclk to 100Mhz to avoid lower eDP panel switch to plus 4K monitor underflow. if (new_clocks->dppclk_khz < 100000) new_clocks->dppclk_khz = 100000; - if (new_clocks->dispclk_khz < 100000) - new_clocks->dispclk_khz = 100000; if (should_set_clock(safe_to_lower, new_clocks->dppclk_khz, clk_mgr->base.clks.dppclk_khz)) { if (clk_mgr->base.clks.dppclk_khz > new_clocks->dppclk_khz) @@ -211,11 +209,18 @@ static void dcn316_update_clocks(struct clk_mgr *clk_mgr_base, update_dppclk = true; } - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) { + if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) && + (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) { + int requested_dispclk_khz = new_clocks->dispclk_khz; + dcn316_disable_otg_wa(clk_mgr_base, context, safe_to_lower, true); + /* Clamp the requested clock to PMFW based on their limit. */ + if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz) + requested_dispclk_khz = dc->debug.min_disp_clk_khz; + + dcn316_smu_set_dispclk(clk_mgr, requested_dispclk_khz); clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz; - dcn316_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz); dcn316_disable_otg_wa(clk_mgr_base, context, safe_to_lower, false); update_dispclk = true; diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c index 1648226586e2..1f47931c2daf 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn35/dcn35_clk_mgr.c @@ -467,14 +467,19 @@ void dcn35_update_clocks(struct clk_mgr *clk_mgr_base, update_dppclk = true; } - if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz)) { + if (should_set_clock(safe_to_lower, new_clocks->dispclk_khz, clk_mgr_base->clks.dispclk_khz) && + (new_clocks->dispclk_khz > 0 || (safe_to_lower && display_count == 0))) { + int requested_dispclk_khz = new_clocks->dispclk_khz; + dcn35_disable_otg_wa(clk_mgr_base, context, safe_to_lower, true); - if (dc->debug.min_disp_clk_khz > 0 && new_clocks->dispclk_khz < dc->debug.min_disp_clk_khz) - new_clocks->dispclk_khz = dc->debug.min_disp_clk_khz; + /* Clamp the requested clock to PMFW based on their limit. */ + if (dc->debug.min_disp_clk_khz > 0 && requested_dispclk_khz < dc->debug.min_disp_clk_khz) + requested_dispclk_khz = dc->debug.min_disp_clk_khz; + dcn35_smu_set_dispclk(clk_mgr, requested_dispclk_khz); clk_mgr_base->clks.dispclk_khz = new_clocks->dispclk_khz; - dcn35_smu_set_dispclk(clk_mgr, clk_mgr_base->clks.dispclk_khz); + dcn35_disable_otg_wa(clk_mgr_base, context, safe_to_lower, false); update_dispclk = true; diff --git a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c index 8082bb877611..a3b8e3d4a429 100644 --- a/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c +++ b/drivers/gpu/drm/amd/display/dc/clk_mgr/dcn401/dcn401_clk_mgr.c @@ -24,6 +24,8 @@ #include "dml/dcn401/dcn401_fpu.h" +#define DCN_BASE__INST0_SEG1 0x000000C0 + #define mmCLK01_CLK0_CLK_PLL_REQ 0x16E37 #define mmCLK01_CLK0_CLK0_DFS_CNTL 0x16E69 #define mmCLK01_CLK0_CLK1_DFS_CNTL 0x16E6C diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 4683c7ef4507..0ce0ad7f9839 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -276,6 +276,7 @@ static bool create_links( link->link_id.type = OBJECT_TYPE_CONNECTOR; link->link_id.id = CONNECTOR_ID_VIRTUAL; link->link_id.enum_id = ENUM_ID_1; + link->psr_settings.psr_version = DC_PSR_VERSION_UNSUPPORTED; link->link_enc = kzalloc(sizeof(*link->link_enc), GFP_KERNEL); if (!link->link_enc) { @@ -438,9 +439,12 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, * Don't adjust DRR while there's bandwidth optimizations pending to * avoid conflicting with firmware updates. */ - if (dc->ctx->dce_version > DCE_VERSION_MAX) - if (dc->optimized_required || dc->wm_optimized_required) + if (dc->ctx->dce_version > DCE_VERSION_MAX) { + if (dc->optimized_required || dc->wm_optimized_required) { + stream->adjust.timing_adjust_pending = true; return false; + } + } dc_exit_ips_for_hw_access(dc); @@ -452,6 +456,7 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, if (dc->caps.max_v_total != 0 && (adjust->v_total_max > dc->caps.max_v_total || adjust->v_total_min > dc->caps.max_v_total)) { + stream->adjust.timing_adjust_pending = false; if (adjust->allow_otg_v_count_halt) return set_long_vtotal(dc, stream, adjust); else @@ -465,7 +470,7 @@ bool dc_stream_adjust_vmin_vmax(struct dc *dc, dc->hwss.set_drr(&pipe, 1, *adjust); - + stream->adjust.timing_adjust_pending = false; return true; } } @@ -3127,8 +3132,14 @@ static void copy_stream_update_to_stream(struct dc *dc, if (update->vrr_active_fixed) stream->vrr_active_fixed = *update->vrr_active_fixed; - if (update->crtc_timing_adjust) + if (update->crtc_timing_adjust) { + if (stream->adjust.v_total_min != update->crtc_timing_adjust->v_total_min || + stream->adjust.v_total_max != update->crtc_timing_adjust->v_total_max || + stream->adjust.timing_adjust_pending) + update->crtc_timing_adjust->timing_adjust_pending = true; stream->adjust = *update->crtc_timing_adjust; + update->crtc_timing_adjust->timing_adjust_pending = false; + } if (update->dpms_off) stream->dpms_off = *update->dpms_off; @@ -4902,7 +4913,8 @@ static bool full_update_required(struct dc *dc, stream_update->lut3d_func || stream_update->pending_test_pattern || stream_update->crtc_timing_adjust || - stream_update->scaler_sharpener_update)) + stream_update->scaler_sharpener_update || + stream_update->hw_cursor_req)) return true; if (stream) { diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index 6eb9bae3af91..4f54e75a8f95 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -176,7 +176,7 @@ static bool is_ycbcr2020_type( { bool ret = false; - if (color_space == COLOR_SPACE_2020_YCBCR) + if (color_space == COLOR_SPACE_2020_YCBCR_LIMITED || color_space == COLOR_SPACE_2020_YCBCR_FULL) ret = true; return ret; } @@ -247,7 +247,8 @@ void color_space_to_black_color( case COLOR_SPACE_YCBCR709_BLACK: case COLOR_SPACE_YCBCR601_LIMITED: case COLOR_SPACE_YCBCR709_LIMITED: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: + case COLOR_SPACE_2020_YCBCR_FULL: *black_color = black_color_format[BLACK_COLOR_FORMAT_YUV_CV]; break; @@ -563,6 +564,7 @@ void set_p_state_switch_method( if (!dc->ctx || !dc->ctx->dmub_srv || !pipe_ctx || !vba) return; + pipe_ctx->p_state_type = P_STATE_UNKNOWN; if (vba->DRAMClockChangeSupport[vba->VoltageLevel][vba->maxMpcComb] != dm_dram_clock_change_unsupported) { /* MCLK switching is supported */ @@ -609,6 +611,21 @@ void set_p_state_switch_method( } } +void set_drr_and_clear_adjust_pending( + struct pipe_ctx *pipe_ctx, + struct dc_stream_state *stream, + struct drr_params *params) +{ + /* params can be null.*/ + if (pipe_ctx && pipe_ctx->stream_res.tg && + pipe_ctx->stream_res.tg->funcs->set_drr) + pipe_ctx->stream_res.tg->funcs->set_drr( + pipe_ctx->stream_res.tg, params); + + if (stream) + stream->adjust.timing_adjust_pending = false; +} + void get_fams2_visual_confirm_color( struct dc *dc, struct dc_state *context, diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c index 298668e9729c..375b3b1d1d18 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_resource.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_resource.c @@ -941,6 +941,17 @@ static void calculate_adjust_recout_for_visual_confirm(struct pipe_ctx *pipe_ctx *base_offset = VISUAL_CONFIRM_BASE_DEFAULT; } +static void reverse_adjust_recout_for_visual_confirm(struct rect *recout, + struct pipe_ctx *pipe_ctx) +{ + int dpp_offset, base_offset; + + calculate_adjust_recout_for_visual_confirm(pipe_ctx, &base_offset, + &dpp_offset); + recout->height += base_offset; + recout->height += dpp_offset; +} + static void adjust_recout_for_visual_confirm(struct rect *recout, struct pipe_ctx *pipe_ctx) { @@ -1642,6 +1653,62 @@ bool resource_build_scaling_params(struct pipe_ctx *pipe_ctx) return res; } +bool resource_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) +{ + struct pipe_ctx *test_pipe, *split_pipe; + struct rect r1 = pipe_ctx->plane_res.scl_data.recout; + int r1_right, r1_bottom; + int cur_layer = pipe_ctx->plane_state->layer_index; + + reverse_adjust_recout_for_visual_confirm(&r1, pipe_ctx); + r1_right = r1.x + r1.width; + r1_bottom = r1.y + r1.height; + + /** + * Disable the cursor if there's another pipe above this with a + * plane that contains this pipe's viewport to prevent double cursor + * and incorrect scaling artifacts. + */ + for (test_pipe = pipe_ctx->top_pipe; test_pipe; + test_pipe = test_pipe->top_pipe) { + struct rect r2; + int r2_right, r2_bottom; + // Skip invisible layer and pipe-split plane on same layer + if (!test_pipe->plane_state || + !test_pipe->plane_state->visible || + test_pipe->plane_state->layer_index == cur_layer) + continue; + + r2 = test_pipe->plane_res.scl_data.recout; + reverse_adjust_recout_for_visual_confirm(&r2, test_pipe); + r2_right = r2.x + r2.width; + r2_bottom = r2.y + r2.height; + + /** + * There is another half plane on same layer because of + * pipe-split, merge together per same height. + */ + for (split_pipe = pipe_ctx->top_pipe; split_pipe; + split_pipe = split_pipe->top_pipe) + if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) { + struct rect r2_half; + + r2_half = split_pipe->plane_res.scl_data.recout; + reverse_adjust_recout_for_visual_confirm(&r2_half, split_pipe); + r2.x = min(r2_half.x, r2.x); + r2.width = r2.width + r2_half.width; + r2_right = r2.x + r2.width; + r2_bottom = min(r2_bottom, r2_half.y + r2_half.height); + break; + } + + if (r1.x >= r2.x && r1.y >= r2.y && r1_right <= r2_right && r1_bottom <= r2_bottom) + return true; + } + + return false; +} + enum dc_status resource_build_scaling_params_for_context( const struct dc *dc, @@ -4247,7 +4314,7 @@ static void set_avi_info_frame( break; case COLOR_SPACE_2020_RGB_FULLRANGE: case COLOR_SPACE_2020_RGB_LIMITEDRANGE: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: hdmi_info.bits.EC0_EC2 = COLORIMETRYEX_BT2020RGBYCBCR; hdmi_info.bits.C0_C1 = COLORIMETRY_EXTENDED; break; @@ -4261,7 +4328,7 @@ static void set_avi_info_frame( break; } - if (pixel_encoding && color_space == COLOR_SPACE_2020_YCBCR && + if (pixel_encoding && color_space == COLOR_SPACE_2020_YCBCR_LIMITED && stream->out_transfer_func.tf == TRANSFER_FUNCTION_GAMMA22) { hdmi_info.bits.EC0_EC2 = 0; hdmi_info.bits.C0_C1 = COLORIMETRY_ITU709; diff --git a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c index 44ff9abe2880..87b4c2793df3 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c +++ b/drivers/gpu/drm/amd/display/dc/dc_dmub_srv.c @@ -991,57 +991,11 @@ void dc_dmub_srv_log_diagnostic_data(struct dc_dmub_srv *dc_dmub_srv) DC_LOG_DEBUG(" is_cw6_en : %d", diag_data.is_cw6_enabled); } -static bool dc_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) -{ - struct pipe_ctx *test_pipe, *split_pipe; - const struct scaler_data *scl_data = &pipe_ctx->plane_res.scl_data; - struct rect r1 = scl_data->recout, r2, r2_half; - int r1_r = r1.x + r1.width, r1_b = r1.y + r1.height, r2_r, r2_b; - int cur_layer = pipe_ctx->plane_state->layer_index; - - /** - * Disable the cursor if there's another pipe above this with a - * plane that contains this pipe's viewport to prevent double cursor - * and incorrect scaling artifacts. - */ - for (test_pipe = pipe_ctx->top_pipe; test_pipe; - test_pipe = test_pipe->top_pipe) { - // Skip invisible layer and pipe-split plane on same layer - if (!test_pipe->plane_state->visible || test_pipe->plane_state->layer_index == cur_layer) - continue; - - r2 = test_pipe->plane_res.scl_data.recout; - r2_r = r2.x + r2.width; - r2_b = r2.y + r2.height; - - /** - * There is another half plane on same layer because of - * pipe-split, merge together per same height. - */ - for (split_pipe = pipe_ctx->top_pipe; split_pipe; - split_pipe = split_pipe->top_pipe) - if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) { - r2_half = split_pipe->plane_res.scl_data.recout; - r2.x = (r2_half.x < r2.x) ? r2_half.x : r2.x; - r2.width = r2.width + r2_half.width; - r2_r = r2.x + r2.width; - break; - } - - if (r1.x >= r2.x && r1.y >= r2.y && r1_r <= r2_r && r1_b <= r2_b) - return true; - } - - return false; -} - static bool dc_dmub_should_update_cursor_data(struct pipe_ctx *pipe_ctx) { if (pipe_ctx->plane_state != NULL) { - if (pipe_ctx->plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE) - return false; - - if (dc_can_pipe_disable_cursor(pipe_ctx)) + if (pipe_ctx->plane_state->address.type == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE || + resource_can_pipe_disable_cursor(pipe_ctx)) return false; } diff --git a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h index cc005da75ce4..8bb628ab7855 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h @@ -959,6 +959,14 @@ union dp_128b_132b_supported_lttpr_link_rates { uint8_t raw; }; +union dp_alpm_lttpr_cap { + struct { + uint8_t AUX_LESS_ALPM_SUPPORTED :1; + uint8_t RESERVED :7; + } bits; + uint8_t raw; +}; + union dp_sink_video_fallback_formats { struct { uint8_t dp_1024x768_60Hz_24bpp_support :1; @@ -1118,6 +1126,7 @@ struct dc_lttpr_caps { uint8_t max_ext_timeout; union dp_main_link_channel_coding_lttpr_cap main_link_channel_coding; union dp_128b_132b_supported_lttpr_link_rates supported_128b_132b_rates; + union dp_alpm_lttpr_cap alpm; uint8_t aux_rd_interval[MAX_REPEATER_CNT - 1]; uint8_t lttpr_ieee_oui[3]; uint8_t lttpr_device_id[6]; @@ -1372,6 +1381,9 @@ struct dp_trace { #ifndef DPCD_MAX_UNCOMPRESSED_PIXEL_RATE_CAP #define DPCD_MAX_UNCOMPRESSED_PIXEL_RATE_CAP 0x221c #endif +#ifndef DP_LTTPR_ALPM_CAPABILITIES +#define DP_LTTPR_ALPM_CAPABILITIES 0xF0009 +#endif #ifndef DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE #define DP_REPEATER_CONFIGURATION_AND_STATUS_SIZE 0x50 #endif diff --git a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h index 5ac55601a6da..d562ddeca512 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_hw_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_hw_types.h @@ -653,7 +653,8 @@ enum dc_color_space { COLOR_SPACE_YCBCR709_LIMITED, COLOR_SPACE_2020_RGB_FULLRANGE, COLOR_SPACE_2020_RGB_LIMITEDRANGE, - COLOR_SPACE_2020_YCBCR, + COLOR_SPACE_2020_YCBCR_LIMITED, + COLOR_SPACE_2020_YCBCR_FULL, COLOR_SPACE_ADOBERGB, COLOR_SPACE_DCIP3, COLOR_SPACE_DISPLAYNATIVE, @@ -661,6 +662,7 @@ enum dc_color_space { COLOR_SPACE_APPCTRL, COLOR_SPACE_CUSTOMPOINTS, COLOR_SPACE_YCBCR709_BLACK, + COLOR_SPACE_2020_YCBCR = COLOR_SPACE_2020_YCBCR_LIMITED, }; enum dc_dither_option { @@ -1015,6 +1017,7 @@ struct dc_crtc_timing_adjust { uint32_t v_total_mid; uint32_t v_total_mid_frame_num; uint32_t allow_otg_v_count_halt; + uint8_t timing_adjust_pending; }; diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index 0c2aa91f0a11..e60898c2df01 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -1033,6 +1033,13 @@ struct psr_settings { unsigned int psr_sdp_transmit_line_num_deadline; uint8_t force_ffu_mode; unsigned int psr_power_opt; + + /** + * Some panels cannot handle idle pattern during PSR entry. + * To power down phy before disable stream to avoid sending + * idle pattern. + */ + uint8_t power_down_phy_before_disable_stream; }; enum replay_coasting_vtotal_type { diff --git a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c index d199e4ed2e59..1130d7619b26 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dce_stream_encoder.c @@ -418,7 +418,7 @@ static void dce110_stream_encoder_dp_set_stream_attribute( dynamic_range_rgb = 1; /*limited range*/ break; case COLOR_SPACE_2020_RGB_FULLRANGE: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: case COLOR_SPACE_XR_RGB: case COLOR_SPACE_MSREF_SCRGB: case COLOR_SPACE_ADOBERGB: @@ -430,6 +430,7 @@ static void dce110_stream_encoder_dp_set_stream_attribute( case COLOR_SPACE_APPCTRL: case COLOR_SPACE_CUSTOMPOINTS: case COLOR_SPACE_UNKNOWN: + default: /* do nothing */ break; } diff --git a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c index 88c75c243bf8..ff3b8244ba3d 100644 --- a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c +++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c @@ -418,6 +418,10 @@ static bool dmub_psr_copy_settings(struct dmub_psr *dmub, copy_settings_data->relock_delay_frame_cnt = 0; if (link->dpcd_caps.sink_dev_id == DP_BRANCH_DEVICE_ID_001CF8) copy_settings_data->relock_delay_frame_cnt = 2; + + copy_settings_data->power_down_phy_before_disable_stream = + link->psr_settings.power_down_phy_before_disable_stream; + copy_settings_data->dsc_slice_height = psr_context->dsc_slice_height; dc_wake_and_execute_dmub_cmd(dc, &cmd, DM_DMUB_WAIT_TYPE_WAIT); diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c index d01a8b8f9595..22e66b375a7f 100644 --- a/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dio/dcn10/dcn10_stream_encoder.c @@ -391,7 +391,7 @@ void enc1_stream_encoder_dp_set_stream_attribute( break; case COLOR_SPACE_2020_RGB_LIMITEDRANGE: case COLOR_SPACE_2020_RGB_FULLRANGE: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: case COLOR_SPACE_XR_RGB: case COLOR_SPACE_MSREF_SCRGB: case COLOR_SPACE_ADOBERGB: @@ -404,6 +404,7 @@ void enc1_stream_encoder_dp_set_stream_attribute( case COLOR_SPACE_CUSTOMPOINTS: case COLOR_SPACE_UNKNOWN: case COLOR_SPACE_YCBCR709_BLACK: + default: /* do nothing */ break; } diff --git a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c index 098c2a01a850..9e5072627ec7 100644 --- a/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/dio/dcn401/dcn401_dio_stream_encoder.c @@ -632,7 +632,7 @@ void enc401_stream_encoder_dp_set_stream_attribute( break; case COLOR_SPACE_2020_RGB_LIMITEDRANGE: case COLOR_SPACE_2020_RGB_FULLRANGE: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: case COLOR_SPACE_XR_RGB: case COLOR_SPACE_MSREF_SCRGB: case COLOR_SPACE_ADOBERGB: @@ -645,6 +645,7 @@ void enc401_stream_encoder_dp_set_stream_attribute( case COLOR_SPACE_CUSTOMPOINTS: case COLOR_SPACE_UNKNOWN: case COLOR_SPACE_YCBCR709_BLACK: + default: /* do nothing */ break; } diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c index 47d785204f29..c90dee4e9116 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn35/dcn35_fpu.c @@ -195,9 +195,9 @@ struct _vcs_dpi_soc_bounding_box_st dcn3_5_soc = { .dcn_downspread_percent = 0.5, .gpuvm_min_page_size_bytes = 4096, .hostvm_min_page_size_bytes = 4096, - .do_urgent_latency_adjustment = 0, + .do_urgent_latency_adjustment = 1, .urgent_latency_adjustment_fabric_clock_component_us = 0, - .urgent_latency_adjustment_fabric_clock_reference_mhz = 0, + .urgent_latency_adjustment_fabric_clock_reference_mhz = 3000, }; void dcn35_build_wm_range_table_fpu(struct clk_mgr *clk_mgr) @@ -367,6 +367,8 @@ void dcn35_update_bw_bounding_box_fpu(struct dc *dc, clock_limits[i].socclk_mhz; dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].memclk_mhz = clk_table->entries[i].memclk_mhz * clk_table->entries[i].wck_ratio; + + dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dram_speed_mts = clock_limits[i].dram_speed_mts; dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz = clock_limits[i].dtbclk_mhz; dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dcfclk_levels = diff --git a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c index d9e63c4fdd95..17d0b4923b0c 100644 --- a/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c +++ b/drivers/gpu/drm/amd/display/dc/dml/dcn351/dcn351_fpu.c @@ -401,6 +401,7 @@ void dcn351_update_bw_bounding_box_fpu(struct dc *dc, clock_limits[i].socclk_mhz; dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].memclk_mhz = clk_table->entries[i].memclk_mhz * clk_table->entries[i].wck_ratio; + dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dram_speed_mts = clock_limits[i].dram_speed_mts; dc->dml2_options.bbox_overrides.clks_table.clk_entries[i].dtbclk_mhz = clock_limits[i].dtbclk_mhz; dc->dml2_options.bbox_overrides.clks_table.num_entries_per_clk.num_dcfclk_levels = diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c index 0c8ec30ea672..731fbd4bc600 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_translation_helper.c @@ -910,7 +910,7 @@ static void populate_dml21_plane_config_from_plane_state(struct dml2_context *dm } //TODO : Could be possibly moved to a common helper layer. -static bool dml21_wrapper_get_plane_id(const struct dc_state *context, const struct dc_plane_state *plane, unsigned int *plane_id) +static bool dml21_wrapper_get_plane_id(const struct dc_state *context, unsigned int stream_id, const struct dc_plane_state *plane, unsigned int *plane_id) { int i, j; @@ -918,10 +918,12 @@ static bool dml21_wrapper_get_plane_id(const struct dc_state *context, const str return false; for (i = 0; i < context->stream_count; i++) { - for (j = 0; j < context->stream_status[i].plane_count; j++) { - if (context->stream_status[i].plane_states[j] == plane) { - *plane_id = (i << 16) | j; - return true; + if (context->streams[i]->stream_id == stream_id) { + for (j = 0; j < context->stream_status[i].plane_count; j++) { + if (context->stream_status[i].plane_states[j] == plane) { + *plane_id = (i << 16) | j; + return true; + } } } } @@ -944,14 +946,14 @@ static unsigned int map_stream_to_dml21_display_cfg(const struct dml2_context *d return location; } -static unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, +static unsigned int map_plane_to_dml21_display_cfg(const struct dml2_context *dml_ctx, unsigned int stream_id, const struct dc_plane_state *plane, const struct dc_state *context) { unsigned int plane_id; int i = 0; int location = -1; - if (!dml21_wrapper_get_plane_id(context, plane, &plane_id)) { + if (!dml21_wrapper_get_plane_id(context, stream_id, plane, &plane_id)) { ASSERT(false); return -1; } @@ -1037,7 +1039,7 @@ bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_s dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; } else { for (plane_index = 0; plane_index < context->stream_status[stream_index].plane_count; plane_index++) { - disp_cfg_plane_location = map_plane_to_dml21_display_cfg(dml_ctx, context->stream_status[stream_index].plane_states[plane_index], context); + disp_cfg_plane_location = map_plane_to_dml21_display_cfg(dml_ctx, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], context); if (disp_cfg_plane_location < 0) disp_cfg_plane_location = dml_dispcfg->num_planes++; @@ -1048,7 +1050,7 @@ bool dml21_map_dc_state_into_dml_display_cfg(const struct dc *in_dc, struct dc_s populate_dml21_plane_config_from_plane_state(dml_ctx, &dml_dispcfg->plane_descriptors[disp_cfg_plane_location], context->stream_status[stream_index].plane_states[plane_index], context, stream_index); dml_dispcfg->plane_descriptors[disp_cfg_plane_location].stream_index = disp_cfg_stream_location; - if (dml21_wrapper_get_plane_id(context, context->stream_status[stream_index].plane_states[plane_index], &dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) + if (dml21_wrapper_get_plane_id(context, context->streams[stream_index]->stream_id, context->stream_status[stream_index].plane_states[plane_index], &dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id[disp_cfg_plane_location])) dml_ctx->v21.dml_to_dc_pipe_mapping.disp_cfg_to_plane_id_valid[disp_cfg_plane_location] = true; /* apply forced pstate policy */ diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c index 1e56d995cd0e..930e86cdb88a 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_utils.c @@ -232,7 +232,6 @@ void dml21_program_dc_pipe(struct dml2_context *dml_ctx, struct dc_state *contex context->bw_ctx.bw.dcn.clk.dppclk_khz = pipe_ctx->plane_res.bw.dppclk_khz; dml21_populate_mall_allocation_size(context, dml_ctx, pln_prog, pipe_ctx); - memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[pipe_ctx->pipe_idx], &pln_prog->mcache_allocation, sizeof(struct dml2_mcache_surface_allocation)); bool sub_vp_enabled = is_sub_vp_enabled(pipe_ctx->stream->ctx->dc, context); diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c index d6fd13f43c08..ed6584535e89 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/dml21_wrapper.c @@ -129,6 +129,7 @@ static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_sta struct pipe_ctx *dc_main_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__]; struct pipe_ctx *dc_phantom_pipes[__DML2_WRAPPER_MAX_STREAMS_PLANES__] = {0}; int num_pipes; + unsigned int dml_phantom_prog_idx; context->bw_ctx.bw.dcn.clk.dppclk_khz = 0; @@ -142,6 +143,9 @@ static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_sta context->bw_ctx.bw.dcn.mall_ss_psr_active_size_bytes = 0; context->bw_ctx.bw.dcn.mall_subvp_size_bytes = 0; + /* phantom's start after main planes */ + dml_phantom_prog_idx = in_ctx->v21.mode_programming.programming->display_config.num_planes; + for (dml_prog_idx = 0; dml_prog_idx < DML2_MAX_PLANES; dml_prog_idx++) { pln_prog = &in_ctx->v21.mode_programming.programming->plane_programming[dml_prog_idx]; @@ -167,6 +171,16 @@ static void dml21_calculate_rq_and_dlg_params(const struct dc *dc, struct dc_sta dml21_program_dc_pipe(in_ctx, context, dc_phantom_pipes[dc_pipe_index], pln_prog, stream_prog); } } + + /* copy per plane mcache allocation */ + memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_prog_idx], &pln_prog->mcache_allocation, sizeof(struct dml2_mcache_surface_allocation)); + if (pln_prog->phantom_plane.valid) { + memcpy(&context->bw_ctx.bw.dcn.mcache_allocations[dml_phantom_prog_idx], + &pln_prog->phantom_plane.mcache_allocation, + sizeof(struct dml2_mcache_surface_allocation)); + + dml_phantom_prog_idx++; + } } /* assign global clocks */ @@ -220,7 +234,9 @@ static bool dml21_mode_check_and_programming(const struct dc *in_dc, struct dc_s if (!result) return false; + DC_FP_START(); result = dml2_build_mode_programming(mode_programming); + DC_FP_END(); if (!result) return false; @@ -263,7 +279,9 @@ static bool dml21_check_mode_support(const struct dc *in_dc, struct dc_state *co mode_support->dml2_instance = dml_init->dml2_instance; dml21_map_dc_state_into_dml_display_cfg(in_dc, context, dml_ctx); dml_ctx->v21.mode_programming.dml2_instance->scratch.build_mode_programming_locals.mode_programming_params.programming = dml_ctx->v21.mode_programming.programming; + DC_FP_START(); is_supported = dml2_check_mode_supported(mode_support); + DC_FP_END(); if (!is_supported) return false; @@ -274,16 +292,12 @@ bool dml21_validate(const struct dc *in_dc, struct dc_state *context, struct dml { bool out = false; - DC_FP_START(); - /* Use dml_validate_only for fast_validate path */ if (fast_validate) out = dml21_check_mode_support(in_dc, context, dml_ctx); else out = dml21_mode_check_and_programming(in_dc, context, dml_ctx); - DC_FP_END(); - return out; } diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h index d2d053f2354d..0ab19cf4d242 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/inc/dml_top_types.h @@ -245,6 +245,7 @@ struct dml2_per_plane_programming { struct { bool valid; struct dml2_plane_parameters descriptor; + struct dml2_mcache_surface_allocation mcache_allocation; struct dml2_dchub_per_pipe_register_set *pipe_regs[DML2_MAX_PLANES]; } phantom_plane; }; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c index 7216d25c783e..44d2969a904e 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4.c @@ -253,7 +253,8 @@ static void expand_implict_subvp(const struct display_configuation_with_meta *di static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_instance *core, const struct display_configuation_with_meta *display_cfg, const struct dml2_display_cfg *svp_expanded_display_cfg, struct dml2_display_cfg_programming *programming, struct dml2_core_scratch *scratch) { - unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index; + unsigned int stream_index, plane_index, pipe_offset, stream_already_populated_mask, main_plane_index, mcache_index; + unsigned int total_main_mcaches_required = 0; int total_pipe_regs_copied = 0; int dml_internal_pipe_index = 0; const struct dml2_plane_parameters *main_plane; @@ -324,6 +325,13 @@ static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_in dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); + memcpy(&programming->plane_programming[plane_index].mcache_allocation, + &display_cfg->stage2.mcache_allocations[plane_index], + sizeof(struct dml2_mcache_surface_allocation)); + total_main_mcaches_required += programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane0 + + programming->plane_programming[plane_index].mcache_allocation.num_mcaches_plane1 - + (programming->plane_programming[plane_index].mcache_allocation.last_slice_sharing.plane0_plane1 ? 1 : 0); + for (pipe_offset = 0; pipe_offset < programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { // Assign storage for this pipe's register values programming->plane_programming[plane_index].pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; @@ -362,6 +370,22 @@ static void pack_mode_programming_params_with_implicit_subvp(struct dml2_core_in memcpy(&programming->plane_programming[main_plane_index].phantom_plane.descriptor, phantom_plane, sizeof(struct dml2_plane_parameters)); dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &programming->plane_programming[main_plane_index].svp_size_mall_bytes, dml_internal_pipe_index); + + /* generate mcache allocation, phantoms use identical mcache configuration, but in the MALL set and unique mcache ID's beginning after all main ID's */ + memcpy(&programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation, + &programming->plane_programming[main_plane_index].mcache_allocation, + sizeof(struct dml2_mcache_surface_allocation)); + for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane0; mcache_index++) { + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index] += total_main_mcaches_required; + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane0[mcache_index] = + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane0[mcache_index]; + } + for (mcache_index = 0; mcache_index < programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.num_mcaches_plane1; mcache_index++) { + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index] += total_main_mcaches_required; + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_mall_plane1[mcache_index] = + programming->plane_programming[main_plane_index].phantom_plane.mcache_allocation.global_mcache_ids_plane1[mcache_index]; + } + for (pipe_offset = 0; pipe_offset < programming->plane_programming[main_plane_index].num_dpps_required; pipe_offset++) { // Assign storage for this pipe's register values programming->plane_programming[main_plane_index].phantom_plane.pipe_regs[pipe_offset] = &programming->pipe_regs[total_pipe_regs_copied]; @@ -571,6 +595,10 @@ bool core_dcn4_mode_programming(struct dml2_core_mode_programming_in_out *in_out dml2_core_calcs_get_mall_allocation(&core->clean_me_up.mode_lib, &in_out->programming->plane_programming[plane_index].surface_size_mall_bytes, dml_internal_pipe_index); + memcpy(&in_out->programming->plane_programming[plane_index].mcache_allocation, + &in_out->display_cfg->stage2.mcache_allocations[plane_index], + sizeof(struct dml2_mcache_surface_allocation)); + for (pipe_offset = 0; pipe_offset < in_out->programming->plane_programming[plane_index].num_dpps_required; pipe_offset++) { in_out->programming->plane_programming[plane_index].plane_descriptor = &in_out->programming->display_config.plane_descriptors[plane_index]; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c index c1ff869512f2..a72b4c05e1fb 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_dcn4_calcs.c @@ -2638,6 +2638,9 @@ static void calculate_mcache_setting( // Luma/Chroma combine in the last mcache // In the case of Luma/Chroma combine-mCache (with lc_comb_mcache==1), all mCaches except the last segment are filled as much as possible, when stay aligned to mvmpg boundary if (*p->lc_comb_mcache && l->is_dual_plane) { + /* if luma and chroma planes share an mcache, increase total chroma mcache count */ + *p->num_mcaches_c = *p->num_mcaches_c + 1; + for (n = 0; n < *p->num_mcaches_l - 1; n++) p->mcache_offsets_l[n] = (n + 1) * l->mvmpg_per_mcache_lb_l * l->mvmpg_access_width_l; p->mcache_offsets_l[*p->num_mcaches_l - 1] = l->full_vp_access_width_l; @@ -3710,13 +3713,12 @@ static unsigned int CalculateMaxVStartup( double line_time_us = (double)timing->h_total / ((double)timing->pixel_clock_khz / 1000); unsigned int vblank_actual = timing->v_total - timing->v_active; unsigned int vblank_nom_default_in_line = (unsigned int)math_floor2((double)vblank_nom_default_us / line_time_us, 1.0); - unsigned int vblank_nom_input = (unsigned int)math_min2(timing->vblank_nom, vblank_nom_default_in_line); - unsigned int vblank_avail = (vblank_nom_input == 0) ? vblank_nom_default_in_line : vblank_nom_input; + unsigned int vblank_avail = (timing->vblank_nom == 0) ? vblank_nom_default_in_line : (unsigned int)timing->vblank_nom; vblank_size = (unsigned int)math_min2(vblank_actual, vblank_avail); if (timing->interlaced && !ptoi_supported) - max_vstartup_lines = (unsigned int)(math_floor2(vblank_size / 2.0, 1.0)); + max_vstartup_lines = (unsigned int)(math_floor2((vblank_size - 1) / 2.0, 1.0)); else max_vstartup_lines = vblank_size - (unsigned int)math_max2(1.0, math_ceil2(write_back_delay_us / line_time_us, 1.0)); #ifdef __DML_VBA_DEBUG__ @@ -4912,6 +4914,7 @@ static double get_urgent_bandwidth_required( double ReadBandwidthChroma[], double PrefetchBandwidthLuma[], double PrefetchBandwidthChroma[], + double PrefetchBandwidthOto[], double excess_vactive_fill_bw_l[], double excess_vactive_fill_bw_c[], double cursor_bw[], @@ -4975,8 +4978,9 @@ static double get_urgent_bandwidth_required( l->vm_row_bw = NumberOfDPP[k] * prefetch_vmrow_bw[k]; l->flip_and_active_bw = l->per_plane_flip_bw[k] + ReadBandwidthLuma[k] * l->adj_factor_p0 + ReadBandwidthChroma[k] * l->adj_factor_p1 + cursor_bw[k] * l->adj_factor_cur; l->flip_and_prefetch_bw = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthLuma[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; + l->flip_and_prefetch_bw_oto = l->per_plane_flip_bw[k] + NumberOfDPP[k] * (PrefetchBandwidthOto[k] * l->adj_factor_p0_pre + PrefetchBandwidthChroma[k] * l->adj_factor_p1_pre) + prefetch_cursor_bw[k] * l->adj_factor_cur_pre; l->active_and_excess_bw = (ReadBandwidthLuma[k] + excess_vactive_fill_bw_l[k]) * l->tmp_nom_adj_factor_p0 + (ReadBandwidthChroma[k] + excess_vactive_fill_bw_c[k]) * l->tmp_nom_adj_factor_p1 + dpte_row_bw[k] + meta_row_bw[k]; - surface_required_bw[k] = math_max4(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw); + surface_required_bw[k] = math_max5(l->vm_row_bw, l->flip_and_active_bw, l->flip_and_prefetch_bw, l->active_and_excess_bw, l->flip_and_prefetch_bw_oto); /* export peak required bandwidth for the surface */ surface_peak_required_bw[k] = math_max2(surface_required_bw[k], surface_peak_required_bw[k]); @@ -5174,6 +5178,7 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->Tsw_est3 = 0.0; s->cursor_prefetch_bytes = 0; *p->prefetch_cursor_bw = 0; + *p->RequiredPrefetchBWOTO = 0.0; dcc_mrq_enable = (p->dcc_enable && p->mrq_present); @@ -5387,6 +5392,9 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch s->prefetch_bw_oto += (p->swath_width_chroma_ub * p->myPipe->BytePerPixelC) / s->LineTime; } + /* oto prefetch bw should be always be less than total vactive bw */ + DML2_ASSERT(s->prefetch_bw_oto < s->per_pipe_vactive_sw_bw * p->myPipe->DPPPerSurface); + s->prefetch_bw_oto = math_max2(s->per_pipe_vactive_sw_bw, s->prefetch_bw_oto) * p->mall_prefetch_sdp_overhead_factor; s->prefetch_bw_oto = math_min2(s->prefetch_bw_oto, *p->prefetch_sw_bytes/(s->min_Lsw_oto*s->LineTime)); @@ -5397,6 +5405,12 @@ static bool CalculatePrefetchSchedule(struct dml2_core_internal_scratch *scratch p->vm_bytes * p->HostVMInefficiencyFactor / (31 * s->LineTime) - *p->Tno_bw, (p->PixelPTEBytesPerRow * p->HostVMInefficiencyFactor + p->meta_row_bytes + tdlut_row_bytes) / (15 * s->LineTime)); + /* oto bw needs to be outputted even if the oto schedule isn't being used to avoid ms/mp mismatch. + * mp will fail if ms decides to use equ schedule and mp decides to use oto schedule + * and the required bandwidth increases when going from ms to mp + */ + *p->RequiredPrefetchBWOTO = s->prefetch_bw_oto; + #ifdef __DML_VBA_DEBUG__ dml2_printf("DML::%s: vactive_sw_bw_l = %f\n", __func__, p->vactive_sw_bw_l); dml2_printf("DML::%s: vactive_sw_bw_c = %f\n", __func__, p->vactive_sw_bw_c); @@ -6157,6 +6171,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, + l->zero_array, //PrefetchBWOTO l->zero_array, l->zero_array, l->zero_array, @@ -6193,6 +6208,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, l->zero_array, //PrefetchBandwidthLuma, l->zero_array, //PrefetchBandwidthChroma, + l->zero_array, //PrefetchBWOTO p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6229,6 +6245,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, + p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6265,6 +6282,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, + p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -6301,6 +6319,7 @@ static void calculate_peak_bandwidth_required( p->surface_read_bandwidth_c, p->prefetch_bandwidth_l, p->prefetch_bandwidth_c, + p->prefetch_bandwidth_oto, // to prevent ms/mp mismatch when oto bw > total vactive bw p->excess_vactive_fill_bw_l, p->excess_vactive_fill_bw_c, p->cursor_bw, @@ -9063,6 +9082,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->ms.VRatioPreC[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->ms.RequiredPrefetchPixelDataBWLuma[k]; // prefetch_sw_bw_l CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->ms.RequiredPrefetchPixelDataBWChroma[k]; // prefetch_sw_bw_c + CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &mode_lib->ms.RequiredPrefetchBWOTO[k]; CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->ms.NoTimeForDynamicMetadata[k]; CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->ms.Tno_bw[k]; CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->ms.Tno_bw_flip[k]; @@ -9207,6 +9227,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; @@ -9373,6 +9394,7 @@ static bool dml_core_mode_support(struct dml2_core_calcs_mode_support_ex *in_out calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->ms.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->ms.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->ms.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_oto = mode_lib->ms.RequiredPrefetchBWOTO; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->ms.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->ms.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->ms.cursor_bw; @@ -11289,6 +11311,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex CalculatePrefetchSchedule_params->VRatioPrefetchC = &mode_lib->mp.VRatioPrefetchC[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWLuma = &mode_lib->mp.RequiredPrefetchPixelDataBWLuma[k]; CalculatePrefetchSchedule_params->RequiredPrefetchPixelDataBWChroma = &mode_lib->mp.RequiredPrefetchPixelDataBWChroma[k]; + CalculatePrefetchSchedule_params->RequiredPrefetchBWOTO = &s->dummy_single_array[0][k]; CalculatePrefetchSchedule_params->NotEnoughTimeForDynamicMetadata = &mode_lib->mp.NotEnoughTimeForDynamicMetadata[k]; CalculatePrefetchSchedule_params->Tno_bw = &mode_lib->mp.Tno_bw[k]; CalculatePrefetchSchedule_params->Tno_bw_flip = &mode_lib->mp.Tno_bw_flip[k]; @@ -11431,6 +11454,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[0]; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; @@ -11563,6 +11587,7 @@ static bool dml_core_mode_programming(struct dml2_core_calcs_mode_programming_ex calculate_peak_bandwidth_params->surface_read_bandwidth_c = mode_lib->mp.vactive_sw_bw_c; calculate_peak_bandwidth_params->prefetch_bandwidth_l = mode_lib->mp.RequiredPrefetchPixelDataBWLuma; calculate_peak_bandwidth_params->prefetch_bandwidth_c = mode_lib->mp.RequiredPrefetchPixelDataBWChroma; + calculate_peak_bandwidth_params->prefetch_bandwidth_oto = s->dummy_single_array[k]; calculate_peak_bandwidth_params->excess_vactive_fill_bw_l = mode_lib->mp.excess_vactive_fill_bw_l; calculate_peak_bandwidth_params->excess_vactive_fill_bw_c = mode_lib->mp.excess_vactive_fill_bw_c; calculate_peak_bandwidth_params->cursor_bw = mode_lib->mp.cursor_bw; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h index 23c0fca5515f..b7cb017b59ba 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_core/dml2_core_shared_types.h @@ -484,6 +484,8 @@ struct dml2_core_internal_mode_support { double WriteBandwidth[DML2_MAX_PLANES][DML2_MAX_WRITEBACK]; double RequiredPrefetchPixelDataBWLuma[DML2_MAX_PLANES]; double RequiredPrefetchPixelDataBWChroma[DML2_MAX_PLANES]; + /* oto bw should also be considered when calculating urgent bw to avoid situations oto/equ mismatches between ms and mp */ + double RequiredPrefetchBWOTO[DML2_MAX_PLANES]; double cursor_bw[DML2_MAX_PLANES]; double prefetch_cursor_bw[DML2_MAX_PLANES]; double prefetch_vmrow_bw[DML2_MAX_PLANES]; @@ -1381,6 +1383,7 @@ struct dml2_core_shared_get_urgent_bandwidth_required_locals { double vm_row_bw; double flip_and_active_bw; double flip_and_prefetch_bw; + double flip_and_prefetch_bw_oto; double active_and_excess_bw; }; @@ -1792,6 +1795,7 @@ struct dml2_core_calcs_CalculatePrefetchSchedule_params { double *VRatioPrefetchC; double *RequiredPrefetchPixelDataBWLuma; double *RequiredPrefetchPixelDataBWChroma; + double *RequiredPrefetchBWOTO; bool *NotEnoughTimeForDynamicMetadata; double *Tno_bw; double *Tno_bw_flip; @@ -2025,6 +2029,7 @@ struct dml2_core_calcs_calculate_peak_bandwidth_required_params { double *surface_read_bandwidth_c; double *prefetch_bandwidth_l; double *prefetch_bandwidth_c; + double *prefetch_bandwidth_oto; double *excess_vactive_fill_bw_l; double *excess_vactive_fill_bw_c; double *cursor_bw; diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c index a3324f7b9ba6..15c906c42ec4 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_pmo/dml2_pmo_dcn4_fams2.c @@ -1082,12 +1082,21 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, const struct dml2_fams2_meta *stream_fams2_meta; unsigned int microschedule_vlines; unsigned int i; + unsigned int mcaches_per_plane; + unsigned int total_mcaches_required = 0; unsigned int num_planes_per_stream[DML2_MAX_PLANES] = { 0 }; /* confirm timing it is not a centered timing */ for (i = 0; i < display_config->display_config.num_planes; i++) { plane_descriptor = &display_config->display_config.plane_descriptors[i]; + mcaches_per_plane = 0; + + if (plane_descriptor->surface.dcc.enable) { + mcaches_per_plane += display_config->stage2.mcache_allocations[i].num_mcaches_plane0 + + display_config->stage2.mcache_allocations[i].num_mcaches_plane1 - + (display_config->stage2.mcache_allocations[i].last_slice_sharing.plane0_plane1 ? 1 : 0); + } if (is_bit_set_in_bitfield(mask, (unsigned char)plane_descriptor->stream_index)) { num_planes_per_stream[plane_descriptor->stream_index]++; @@ -1098,7 +1107,19 @@ static bool all_timings_support_svp(const struct dml2_pmo_instance *pmo, plane_descriptor->composition.rotation_angle != dml2_rotation_0) { return false; } + + /* phantom requires same number of mcaches as main */ + if (plane_descriptor->surface.dcc.enable) { + mcaches_per_plane *= 2; + } } + + total_mcaches_required += mcaches_per_plane; + } + + if (total_mcaches_required > pmo->soc_bb->num_dcc_mcaches) { + /* too many mcaches required */ + return false; } for (i = 0; i < DML2_MAX_PLANES; i++) { diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c index a8f58f8448e4..dc2ce5e77f57 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml21/src/dml2_top/dml2_top_soc15.c @@ -831,7 +831,6 @@ static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_program bool uclk_pstate_success = false; bool vmin_success = false; bool stutter_success = false; - unsigned int i; memset(l, 0, sizeof(struct dml2_build_mode_programming_locals)); memset(in_out->programming, 0, sizeof(struct dml2_display_cfg_programming)); @@ -977,13 +976,6 @@ static bool dml2_top_soc15_build_mode_programming(struct dml2_build_mode_program } /* - * Populate mcache programming - */ - for (i = 0; i < in_out->display_config->num_planes; i++) { - in_out->programming->plane_programming[i].mcache_allocation = l->base_display_config_with_meta.stage2.mcache_allocations[i]; - } - - /* * Call DPMM to map all requirements to minimum clock state */ if (result) { diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c index b8a34abaf519..aeb9fae83cac 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_translation_helper.c @@ -969,7 +969,9 @@ static void populate_dml_surface_cfg_from_plane_state(enum dml_project_id dml2_p } } -static void get_scaler_data_for_plane(const struct dc_plane_state *in, struct dc_state *context, struct scaler_data *out) +static struct scaler_data *get_scaler_data_for_plane( + const struct dc_plane_state *in, + struct dc_state *context) { int i; struct pipe_ctx *temp_pipe = &context->res_ctx.temp_pipe; @@ -990,7 +992,7 @@ static void get_scaler_data_for_plane(const struct dc_plane_state *in, struct dc } ASSERT(i < MAX_PIPES); - memcpy(out, &temp_pipe->plane_res.scl_data, sizeof(*out)); + return &temp_pipe->plane_res.scl_data; } static void populate_dummy_dml_plane_cfg(struct dml_plane_cfg_st *out, unsigned int location, @@ -1053,11 +1055,7 @@ static void populate_dml_plane_cfg_from_plane_state(struct dml_plane_cfg_st *out const struct dc_plane_state *in, struct dc_state *context, const struct soc_bounding_box_st *soc) { - struct scaler_data *scaler_data = kzalloc(sizeof(*scaler_data), GFP_KERNEL); - if (!scaler_data) - return; - - get_scaler_data_for_plane(in, context, scaler_data); + struct scaler_data *scaler_data = get_scaler_data_for_plane(in, context); out->CursorBPP[location] = dml_cur_32bit; out->CursorWidth[location] = 256; @@ -1122,8 +1120,6 @@ static void populate_dml_plane_cfg_from_plane_state(struct dml_plane_cfg_st *out out->DynamicMetadataTransmittedBytes[location] = 0; out->NumberOfCursors[location] = 1; - - kfree(scaler_data); } static unsigned int map_stream_to_dml_display_cfg(const struct dml2_context *dml2, diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h index 0f944fcfd5a5..785226945699 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_wrapper.h @@ -159,6 +159,7 @@ struct dml2_clks_table_entry { unsigned int dtbclk_mhz; unsigned int dispclk_mhz; unsigned int dppclk_mhz; + unsigned int dram_speed_mts; /*which is based on wck_ratio*/ }; struct dml2_clks_num_entries { diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c index 40acebd13e46..abf439e743f2 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn30/dcn30_dpp.c @@ -425,11 +425,6 @@ bool dpp3_get_optimal_number_of_taps( int min_taps_y, min_taps_c; enum lb_memory_config lb_config; - if (scl_data->viewport.width > scl_data->h_active && - dpp->ctx->dc->debug.max_downscale_src_width != 0 && - scl_data->viewport.width > dpp->ctx->dc->debug.max_downscale_src_width) - return false; - /* * Set default taps if none are provided * From programming guide: taps = min{ ceil(2*H_RATIO,1), 8} for downscaling @@ -467,6 +462,12 @@ bool dpp3_get_optimal_number_of_taps( else scl_data->taps.h_taps_c = in_taps->h_taps_c; + // Avoid null data in the scl data with this early return, proceed non-adaptive calcualtion first + if (scl_data->viewport.width > scl_data->h_active && + dpp->ctx->dc->debug.max_downscale_src_width != 0 && + scl_data->viewport.width > dpp->ctx->dc->debug.max_downscale_src_width) + return false; + /*Ensure we can support the requested number of vtaps*/ min_taps_y = dc_fixpt_ceil(scl_data->ratios.vert); min_taps_c = dc_fixpt_ceil(scl_data->ratios.vert_c); diff --git a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c index 1236e0f9a256..712aff7e17f7 100644 --- a/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c +++ b/drivers/gpu/drm/amd/display/dc/dpp/dcn401/dcn401_dpp_cm.c @@ -120,10 +120,11 @@ void dpp401_set_cursor_attributes( enum dc_cursor_color_format color_format = cursor_attributes->color_format; int cur_rom_en = 0; - // DCN4 should always do Cursor degamma for Cursor Color modes if (color_format == CURSOR_MODE_COLOR_PRE_MULTIPLIED_ALPHA || color_format == CURSOR_MODE_COLOR_UN_PRE_MULTIPLIED_ALPHA) { - cur_rom_en = 1; + if (cursor_attributes->attribute_flags.bits.ENABLE_CURSOR_DEGAMMA) { + cur_rom_en = 1; + } } REG_UPDATE_3(CURSOR0_CONTROL, diff --git a/drivers/gpu/drm/amd/display/dc/hpo/dcn31/dcn31_hpo_dp_stream_encoder.c b/drivers/gpu/drm/amd/display/dc/hpo/dcn31/dcn31_hpo_dp_stream_encoder.c index 678db949cfe3..759b453385c4 100644 --- a/drivers/gpu/drm/amd/display/dc/hpo/dcn31/dcn31_hpo_dp_stream_encoder.c +++ b/drivers/gpu/drm/amd/display/dc/hpo/dcn31/dcn31_hpo_dp_stream_encoder.c @@ -323,7 +323,7 @@ static void dcn31_hpo_dp_stream_enc_set_stream_attribute( break; case COLOR_SPACE_2020_RGB_LIMITEDRANGE: case COLOR_SPACE_2020_RGB_FULLRANGE: - case COLOR_SPACE_2020_YCBCR: + case COLOR_SPACE_2020_YCBCR_LIMITED: case COLOR_SPACE_XR_RGB: case COLOR_SPACE_MSREF_SCRGB: case COLOR_SPACE_ADOBERGB: @@ -336,6 +336,7 @@ static void dcn31_hpo_dp_stream_enc_set_stream_attribute( case COLOR_SPACE_CUSTOMPOINTS: case COLOR_SPACE_UNKNOWN: case COLOR_SPACE_YCBCR709_BLACK: + default: /* do nothing */ break; } 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 81f4c386c287..94ceccfc0498 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 @@ -1065,7 +1065,8 @@ void dce110_edp_backlight_control( DC_LOG_DC("edp_receiver_ready_T9 skipped\n"); } - if (!enable && link->dpcd_sink_ext_caps.bits.oled) { + if (!enable) { + /*follow oem panel config's requirement*/ pre_T11_delay += link->panel_config.pps.extra_pre_t11_ms; msleep(pre_T11_delay); } @@ -1654,9 +1655,7 @@ enum dc_status dce110_apply_single_controller_ctx_to_hw( params.vertical_total_min = stream->adjust.v_total_min; params.vertical_total_max = stream->adjust.v_total_max; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); // DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min != 0 && stream->adjust.v_total_max != 0) @@ -2104,8 +2103,7 @@ static void set_drr(struct pipe_ctx **pipe_ctx, struct timing_generator *tg = pipe_ctx[i]->stream_res.tg; if ((tg != NULL) && tg->funcs) { - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c index 13f9e9b439f6..bbeaefe1ef0d 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn10/dcn10_hwseq.c @@ -1112,9 +1112,7 @@ static void dcn10_reset_back_end_for_pipe( pipe_ctx->stream_res.tg->funcs->disable_crtc(pipe_ctx->stream_res.tg); pipe_ctx->stream_res.tg->funcs->enable_optc_clock(pipe_ctx->stream_res.tg, false); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); if (dc_is_hdmi_tmds_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->link->phy_state.symclk_ref_cnts.otg = 0; } @@ -3217,8 +3215,7 @@ void dcn10_set_drr(struct pipe_ctx **pipe_ctx, struct timing_generator *tg = pipe_ctx[i]->stream_res.tg; if ((tg != NULL) && tg->funcs) { - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( @@ -3427,52 +3424,6 @@ void dcn10_update_dchub(struct dce_hwseq *hws, struct dchub_init_data *dh_data) hubbub->funcs->update_dchub(hubbub, dh_data); } -static bool dcn10_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) -{ - struct pipe_ctx *test_pipe, *split_pipe; - const struct scaler_data *scl_data = &pipe_ctx->plane_res.scl_data; - struct rect r1 = scl_data->recout, r2, r2_half; - int r1_r = r1.x + r1.width, r1_b = r1.y + r1.height, r2_r, r2_b; - int cur_layer = pipe_ctx->plane_state->layer_index; - - /** - * Disable the cursor if there's another pipe above this with a - * plane that contains this pipe's viewport to prevent double cursor - * and incorrect scaling artifacts. - */ - for (test_pipe = pipe_ctx->top_pipe; test_pipe; - test_pipe = test_pipe->top_pipe) { - // Skip invisible layer and pipe-split plane on same layer - if (!test_pipe->plane_state || - !test_pipe->plane_state->visible || - test_pipe->plane_state->layer_index == cur_layer) - continue; - - r2 = test_pipe->plane_res.scl_data.recout; - r2_r = r2.x + r2.width; - r2_b = r2.y + r2.height; - - /** - * There is another half plane on same layer because of - * pipe-split, merge together per same height. - */ - for (split_pipe = pipe_ctx->top_pipe; split_pipe; - split_pipe = split_pipe->top_pipe) - if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) { - r2_half = split_pipe->plane_res.scl_data.recout; - r2.x = (r2_half.x < r2.x) ? r2_half.x : r2.x; - r2.width = r2.width + r2_half.width; - r2_r = r2.x + r2.width; - break; - } - - if (r1.x >= r2.x && r1.y >= r2.y && r1_r <= r2_r && r1_b <= r2_b) - return true; - } - - return false; -} - void dcn10_set_cursor_position(struct pipe_ctx *pipe_ctx) { struct dc_cursor_position pos_cpy = pipe_ctx->stream->cursor_position; @@ -3572,7 +3523,7 @@ void dcn10_set_cursor_position(struct pipe_ctx *pipe_ctx) == PLN_ADDR_TYPE_VIDEO_PROGRESSIVE) pos_cpy.enable = false; - if (pos_cpy.enable && dcn10_can_pipe_disable_cursor(pipe_ctx)) + if (pos_cpy.enable && resource_can_pipe_disable_cursor(pipe_ctx)) pos_cpy.enable = 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 b78096a7690e..1a07973ead4f 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 @@ -952,9 +952,7 @@ enum dc_status dcn20_enable_stream_timing( params.vertical_total_max = stream->adjust.v_total_max; params.vertical_total_mid = stream->adjust.v_total_mid; params.vertical_total_mid_frame_num = stream->adjust.v_total_mid_frame_num; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); // DRR should set trigger event to monitor surface update event if (stream->adjust.v_total_min != 0 && stream->adjust.v_total_max != 0) @@ -1266,14 +1264,18 @@ static void dcn20_power_on_plane_resources( struct dce_hwseq *hws, struct pipe_ctx *pipe_ctx) { + uint32_t org_ip_request_cntl = 0; + DC_LOGGER_INIT(hws->ctx->logger); if (hws->funcs.dpp_root_clock_control) hws->funcs.dpp_root_clock_control(hws, pipe_ctx->plane_res.dpp->inst, true); if (REG(DC_IP_REQUEST_CNTL)) { - REG_SET(DC_IP_REQUEST_CNTL, 0, - IP_REQUEST_EN, 1); + REG_GET(DC_IP_REQUEST_CNTL, IP_REQUEST_EN, &org_ip_request_cntl); + if (org_ip_request_cntl == 0) + REG_SET(DC_IP_REQUEST_CNTL, 0, + IP_REQUEST_EN, 1); if (hws->funcs.dpp_pg_control) hws->funcs.dpp_pg_control(hws, pipe_ctx->plane_res.dpp->inst, true); @@ -1281,8 +1283,10 @@ static void dcn20_power_on_plane_resources( if (hws->funcs.hubp_pg_control) hws->funcs.hubp_pg_control(hws, pipe_ctx->plane_res.hubp->inst, true); - REG_SET(DC_IP_REQUEST_CNTL, 0, - IP_REQUEST_EN, 0); + if (org_ip_request_cntl == 0) + REG_SET(DC_IP_REQUEST_CNTL, 0, + IP_REQUEST_EN, 0); + DC_LOG_DEBUG( "Un-gated front end for pipe %d\n", pipe_ctx->plane_res.hubp->inst); } @@ -2849,9 +2853,7 @@ void dcn20_reset_back_end_for_pipe( pipe_ctx->stream_res.tg->funcs->set_odm_bypass( pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); /* TODO - convert symclk_ref_cnts for otg to a bit map to solve * the case where the same symclk is shared across multiple otg * instances 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 03ba01f4ace1..38f889826697 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 @@ -538,9 +538,7 @@ static void dcn31_reset_back_end_for_pipe( if (dc_is_hdmi_tmds_signal(pipe_ctx->stream->signal)) pipe_ctx->stream->link->phy_state.symclk_ref_cnts.otg = 0; - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); /* DPMS may already disable or */ /* dpms_off status is incorrect due to fastboot diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c index b907ad1acedd..922b8d71cf1a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn35/dcn35_hwseq.c @@ -1473,8 +1473,7 @@ void dcn35_set_drr(struct pipe_ctx **pipe_ctx, num_frames = 2 * (frame_rate % 60); } } - if (tg->funcs->set_drr) - tg->funcs->set_drr(tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx[i], pipe_ctx[i]->stream, ¶ms); if (adjust.v_total_max != 0 && adjust.v_total_min != 0) if (tg->funcs->set_static_screen_control) tg->funcs->set_static_screen_control( diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 89af3e4afbc2..da8afb08b920 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -830,10 +830,7 @@ enum dc_status dcn401_enable_stream_timing( } hws->funcs.wait_for_blank_complete(pipe_ctx->stream_res.opp); - - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, ¶ms); + set_drr_and_clear_adjust_pending(pipe_ctx, stream, ¶ms); /* Event triggers and num frames initialized for DRR, but can be * later updated for PSR use. Note DRR trigger events are generated @@ -975,52 +972,6 @@ void dcn401_setup_hpo_hw_control(const struct dce_hwseq *hws, bool enable) REG_UPDATE(HPO_TOP_HW_CONTROL, HPO_IO_EN, enable); } -static bool dcn401_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx) -{ - struct pipe_ctx *test_pipe, *split_pipe; - const struct scaler_data *scl_data = &pipe_ctx->plane_res.scl_data; - struct rect r1 = scl_data->recout, r2, r2_half; - int r1_r = r1.x + r1.width, r1_b = r1.y + r1.height, r2_r, r2_b; - int cur_layer = pipe_ctx->plane_state->layer_index; - - /** - * Disable the cursor if there's another pipe above this with a - * plane that contains this pipe's viewport to prevent double cursor - * and incorrect scaling artifacts. - */ - for (test_pipe = pipe_ctx->top_pipe; test_pipe; - test_pipe = test_pipe->top_pipe) { - // Skip invisible layer and pipe-split plane on same layer - if (!test_pipe->plane_state || - !test_pipe->plane_state->visible || - test_pipe->plane_state->layer_index == cur_layer) - continue; - - r2 = test_pipe->plane_res.scl_data.recout; - r2_r = r2.x + r2.width; - r2_b = r2.y + r2.height; - - /** - * There is another half plane on same layer because of - * pipe-split, merge together per same height. - */ - for (split_pipe = pipe_ctx->top_pipe; split_pipe; - split_pipe = split_pipe->top_pipe) - if (split_pipe->plane_state->layer_index == test_pipe->plane_state->layer_index) { - r2_half = split_pipe->plane_res.scl_data.recout; - r2.x = (r2_half.x < r2.x) ? r2_half.x : r2.x; - r2.width = r2.width + r2_half.width; - r2_r = r2.x + r2.width; - break; - } - - if (r1.x >= r2.x && r1.y >= r2.y && r1_r <= r2_r && r1_b <= r2_b) - return true; - } - - return false; -} - void adjust_hotspot_between_slices_for_2x_magnify(uint32_t cursor_width, struct dc_cursor_position *pos_cpy) { if (cursor_width <= 128) { @@ -1211,7 +1162,7 @@ void dcn401_set_cursor_position(struct pipe_ctx *pipe_ctx) pos_cpy.x = (uint32_t)x_pos; pos_cpy.y = (uint32_t)y_pos; - if (pos_cpy.enable && dcn401_can_pipe_disable_cursor(pipe_ctx)) + if (pos_cpy.enable && resource_can_pipe_disable_cursor(pipe_ctx)) pos_cpy.enable = false; x_pos = pos_cpy.x - param.recout.x; @@ -1866,9 +1817,8 @@ void dcn401_reset_back_end_for_pipe( pipe_ctx->stream_res.tg->funcs->set_odm_bypass( pipe_ctx->stream_res.tg, &pipe_ctx->stream->timing); - if (pipe_ctx->stream_res.tg->funcs->set_drr) - pipe_ctx->stream_res.tg->funcs->set_drr( - pipe_ctx->stream_res.tg, NULL); + set_drr_and_clear_adjust_pending(pipe_ctx, pipe_ctx->stream, NULL); + /* TODO - convert symclk_ref_cnts for otg to a bit map to solve * the case where the same symclk is shared across multiple otg * instances @@ -2027,9 +1977,9 @@ static void dcn401_program_pipe( dc->res_pool->hubbub, pipe_ctx->plane_res.hubp->inst, pipe_ctx->hubp_regs.det_size); } - if (pipe_ctx->update_flags.raw || - (pipe_ctx->plane_state && pipe_ctx->plane_state->update_flags.raw) || - pipe_ctx->stream->update_flags.raw) + if (pipe_ctx->plane_state && (pipe_ctx->update_flags.raw || + pipe_ctx->plane_state->update_flags.raw || + pipe_ctx->stream->update_flags.raw)) dc->hwss.update_dchubp_dpp(dc, pipe_ctx, context); if (pipe_ctx->plane_state && (pipe_ctx->update_flags.bits.enable || @@ -2659,3 +2609,37 @@ void dcn401_detect_pipe_changes(struct dc_state *old_state, new_pipe->update_flags.bits.test_pattern_changed = 1; } } + +void dcn401_plane_atomic_power_down(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp) +{ + struct dce_hwseq *hws = dc->hwseq; + uint32_t org_ip_request_cntl = 0; + + DC_LOGGER_INIT(dc->ctx->logger); + + REG_GET(DC_IP_REQUEST_CNTL, IP_REQUEST_EN, &org_ip_request_cntl); + if (org_ip_request_cntl == 0) + REG_SET(DC_IP_REQUEST_CNTL, 0, + IP_REQUEST_EN, 1); + + if (hws->funcs.dpp_pg_control) + hws->funcs.dpp_pg_control(hws, dpp->inst, false); + + if (hws->funcs.hubp_pg_control) + hws->funcs.hubp_pg_control(hws, hubp->inst, false); + + hubp->funcs->hubp_reset(hubp); + dpp->funcs->dpp_reset(dpp); + + if (org_ip_request_cntl == 0) + REG_SET(DC_IP_REQUEST_CNTL, 0, + IP_REQUEST_EN, 0); + + DC_LOG_DEBUG( + "Power gated front end %d\n", hubp->inst); + + if (hws->funcs.dpp_root_clock_control) + hws->funcs.dpp_root_clock_control(hws, dpp->inst, false); +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h index 17cea748789e..dbd69d215b8b 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.h @@ -102,4 +102,7 @@ void dcn401_detect_pipe_changes( struct dc_state *new_state, struct pipe_ctx *old_pipe, struct pipe_ctx *new_pipe); +void dcn401_plane_atomic_power_down(struct dc *dc, + struct dpp *dpp, + struct hubp *hubp); #endif /* __DC_HWSS_DCN401_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c index 44cb376f97c1..a4e3501fadbb 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c @@ -123,7 +123,7 @@ static const struct hwseq_private_funcs dcn401_private_funcs = { .disable_vga = dcn20_disable_vga, .bios_golden_init = dcn10_bios_golden_init, .plane_atomic_disable = dcn20_plane_atomic_disable, - .plane_atomic_power_down = dcn10_plane_atomic_power_down, + .plane_atomic_power_down = dcn401_plane_atomic_power_down, .enable_power_gating_plane = dcn32_enable_power_gating_plane, .hubp_pg_control = dcn32_hubp_pg_control, .program_all_writeback_pipes_in_tree = dcn30_program_all_writeback_pipes_in_tree, diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index a7d66cfd93c9..16ef5250a02e 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -46,6 +46,7 @@ struct dce_hwseq; struct link_resource; struct dc_dmub_cmd; struct pg_block_update; +struct drr_params; struct subvp_pipe_control_lock_fast_params { struct dc *dc; @@ -521,6 +522,11 @@ void set_p_state_switch_method( struct dc_state *context, struct pipe_ctx *pipe_ctx); +void set_drr_and_clear_adjust_pending( + struct pipe_ctx *pipe_ctx, + struct dc_stream_state *stream, + struct drr_params *params); + void hwss_execute_sequence(struct dc *dc, struct block_sequence block_sequence[], int num_steps); diff --git a/drivers/gpu/drm/amd/display/dc/inc/core_types.h b/drivers/gpu/drm/amd/display/dc/inc/core_types.h index d558efc6e12f..652d52040f4e 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/core_types.h +++ b/drivers/gpu/drm/amd/display/dc/inc/core_types.h @@ -627,7 +627,7 @@ struct dc_state { */ struct bw_context bw_ctx; - struct block_sequence block_sequence[50]; + struct block_sequence block_sequence[100]; unsigned int block_sequence_steps; struct dc_dmub_cmd dc_dmub_cmd[10]; unsigned int dmub_cmd_count; diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr_internal.h b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr_internal.h index 7a1ca1e98059..221645c023b5 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr_internal.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/clk_mgr_internal.h @@ -221,6 +221,7 @@ enum dentist_divider_range { CLK_SF(CLK0_CLK_PLL_REQ, FbMult_frac, mask_sh) #define CLK_REG_LIST_DCN401() \ + SR(DENTIST_DISPCLK_CNTL), \ CLK_SR_DCN401(CLK0_CLK_PLL_REQ, CLK01, 0), \ CLK_SR_DCN401(CLK0_CLK0_DFS_CNTL, CLK01, 0), \ CLK_SR_DCN401(CLK0_CLK1_DFS_CNTL, CLK01, 0), \ diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h index 0150f2581ee4..0c5675d1c593 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dpp.h @@ -119,10 +119,14 @@ static const struct dpp_input_csc_matrix __maybe_unused dpp_input_csc_matrix[] = { 0x39a6, 0x2568, 0, 0xe0d6, 0xeedd, 0x2568, 0xf925, 0x9a8, 0, 0x2568, 0x43ee, 0xdbb2 } }, - { COLOR_SPACE_2020_YCBCR, + { COLOR_SPACE_2020_YCBCR_FULL, { 0x2F30, 0x2000, 0, 0xE869, 0xEDB7, 0x2000, 0xFABC, 0xBC6, 0, 0x2000, 0x3C34, 0xE1E6 } }, + { COLOR_SPACE_2020_YCBCR_LIMITED, + { 0x35B9, 0x2543, 0, 0xE2B2, + 0xEB2F, 0x2543, 0xFA01, 0x0B1F, + 0, 0x2543, 0x4489, 0xDB42 } }, { COLOR_SPACE_2020_RGB_LIMITEDRANGE, { 0x35E0, 0x255F, 0, 0xE2B3, 0xEB20, 0x255F, 0xF9FD, 0xB1E, diff --git a/drivers/gpu/drm/amd/display/dc/inc/resource.h b/drivers/gpu/drm/amd/display/dc/inc/resource.h index cd1157d225ab..b32d07ce0f08 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/resource.h +++ b/drivers/gpu/drm/amd/display/dc/inc/resource.h @@ -152,6 +152,8 @@ bool resource_attach_surfaces_to_context( struct dc_state *context, const struct resource_pool *pool); +bool resource_can_pipe_disable_cursor(struct pipe_ctx *pipe_ctx); + #define FREE_PIPE_INDEX_NOT_FOUND -1 /* diff --git a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c index ec7de9c01fab..e95ec72b4096 100644 --- a/drivers/gpu/drm/amd/display/dc/link/link_dpms.c +++ b/drivers/gpu/drm/amd/display/dc/link/link_dpms.c @@ -148,6 +148,7 @@ void link_blank_dp_stream(struct dc_link *link, bool hw_init) void link_set_all_streams_dpms_off_for_link(struct dc_link *link) { struct pipe_ctx *pipes[MAX_PIPES]; + struct dc_stream_state *streams[MAX_PIPES]; struct dc_state *state = link->dc->current_state; uint8_t count; int i; @@ -160,10 +161,18 @@ void link_set_all_streams_dpms_off_for_link(struct dc_link *link) link_get_master_pipes_with_dpms_on(link, state, &count, pipes); + /* The subsequent call to dc_commit_updates_for_stream for a full update + * will release the current state and swap to a new state. Releasing the + * current state results in the stream pointers in the pipe_ctx structs + * to be zero'd. Hence, cache all streams prior to dc_commit_updates_for_stream. + */ + for (i = 0; i < count; i++) + streams[i] = pipes[i]->stream; + for (i = 0; i < count; i++) { - stream_update.stream = pipes[i]->stream; + stream_update.stream = streams[i]; dc_commit_updates_for_stream(link->ctx->dc, NULL, 0, - pipes[i]->stream, &stream_update, + streams[i], &stream_update, state); } 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 44f33e3bc1c5..64e4ae379e34 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 @@ -250,21 +250,21 @@ static uint32_t intersect_frl_link_bw_support( { uint32_t supported_bw_in_kbps = max_supported_frl_bw_in_kbps; - // HDMI_ENCODED_LINK_BW bits are only valid if HDMI Link Configuration bit is 1 (FRL mode) - if (hdmi_encoded_link_bw.bits.FRL_MODE) { - if (hdmi_encoded_link_bw.bits.BW_48Gbps) - supported_bw_in_kbps = 48000000; - else if (hdmi_encoded_link_bw.bits.BW_40Gbps) - supported_bw_in_kbps = 40000000; - else if (hdmi_encoded_link_bw.bits.BW_32Gbps) - supported_bw_in_kbps = 32000000; - else if (hdmi_encoded_link_bw.bits.BW_24Gbps) - supported_bw_in_kbps = 24000000; - else if (hdmi_encoded_link_bw.bits.BW_18Gbps) - supported_bw_in_kbps = 18000000; - else if (hdmi_encoded_link_bw.bits.BW_9Gbps) - supported_bw_in_kbps = 9000000; - } + /* Skip checking FRL_MODE bit, as certain PCON will clear + * it despite supporting the link BW indicated in the other bits. + */ + if (hdmi_encoded_link_bw.bits.BW_48Gbps) + supported_bw_in_kbps = 48000000; + else if (hdmi_encoded_link_bw.bits.BW_40Gbps) + supported_bw_in_kbps = 40000000; + else if (hdmi_encoded_link_bw.bits.BW_32Gbps) + supported_bw_in_kbps = 32000000; + else if (hdmi_encoded_link_bw.bits.BW_24Gbps) + supported_bw_in_kbps = 24000000; + else if (hdmi_encoded_link_bw.bits.BW_18Gbps) + supported_bw_in_kbps = 18000000; + else if (hdmi_encoded_link_bw.bits.BW_9Gbps) + supported_bw_in_kbps = 9000000; return supported_bw_in_kbps; } @@ -945,6 +945,9 @@ bool link_decide_link_settings(struct dc_stream_state *stream, * TODO: add MST specific link training routine */ decide_mst_link_settings(link, link_setting); + } else if (stream->signal == SIGNAL_TYPE_VIRTUAL) { + link_setting->lane_count = LANE_COUNT_FOUR; + link_setting->link_rate = LINK_RATE_HIGH3; } else if (link->connector_signal == SIGNAL_TYPE_EDP) { /* enable edp link optimization for DSC eDP case */ if (stream->timing.flags.DSC) { @@ -967,9 +970,6 @@ bool link_decide_link_settings(struct dc_stream_state *stream, } else { edp_decide_link_settings(link, link_setting, req_bw); } - } else if (stream->signal == SIGNAL_TYPE_VIRTUAL) { - link_setting->lane_count = LANE_COUNT_FOUR; - link_setting->link_rate = LINK_RATE_HIGH3; } else { decide_dp_link_settings(link, link_setting, req_bw); } @@ -1502,7 +1502,7 @@ static bool dpcd_read_sink_ext_caps(struct dc_link *link) enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) { - uint8_t lttpr_dpcd_data[8] = {0}; + uint8_t lttpr_dpcd_data[10] = {0}; enum dc_status status; bool is_lttpr_present; @@ -1552,6 +1552,10 @@ enum dc_status dp_retrieve_lttpr_cap(struct dc_link *link) lttpr_dpcd_data[DP_PHY_REPEATER_128B132B_RATES - DP_LT_TUNABLE_PHY_REPEATER_FIELD_DATA_STRUCTURE_REV]; + link->dpcd_caps.lttpr_caps.alpm.raw = + lttpr_dpcd_data[DP_LTTPR_ALPM_CAPABILITIES - + DP_LT_TUNABLE_PHY_REPEATER_FIELD_DATA_STRUCTURE_REV]; + /* If this chip cap is set, at least one retimer must exist in the chain * Override count to 1 if we receive a known bad count (0 or an invalid value) */ if (((link->chip_caps & AMD_EXT_DISPLAY_PATH_CAPS__EXT_CHIP_MASK) == AMD_EXT_DISPLAY_PATH_CAPS__DP_FIXED_VS_EN) && diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_phy.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_phy.c index 2c73ac87cd66..c27ffec5d84f 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_phy.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_phy.c @@ -75,7 +75,8 @@ void dp_disable_link_phy(struct dc_link *link, struct dc *dc = link->ctx->dc; if (!link->wa_flags.dp_keep_receiver_powered && - !link->skip_implict_edp_power_control) + !link->skip_implict_edp_power_control && + link->type != dc_connection_none) dpcd_write_rx_power_ctrl(link, false); dc->hwss.disable_link_output(link, link_res, signal); @@ -163,8 +164,9 @@ enum dc_status dp_set_fec_ready(struct dc_link *link, const struct link_resource } else { if (link->fec_state == dc_link_fec_ready) { fec_config = 0; - core_link_write_dpcd(link, DP_FEC_CONFIGURATION, - &fec_config, sizeof(fec_config)); + if (link->type != dc_connection_none) + core_link_write_dpcd(link, DP_FEC_CONFIGURATION, + &fec_config, sizeof(fec_config)); link_enc->funcs->fec_set_ready(link_enc, false); link->fec_state = dc_link_fec_not_ready; diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c index 751c18e592ea..7848ddb94456 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training.c @@ -1782,13 +1782,10 @@ bool perform_link_training_with_retries( is_link_bw_min = ((cur_link_settings.link_rate <= LINK_RATE_LOW) && (cur_link_settings.lane_count <= LANE_COUNT_ONE)); - if (is_link_bw_low) { + if (is_link_bw_low) DC_LOG_WARNING( "%s: Link(%d) bandwidth too low after fallback req_bw(%d) > link_bw(%d)\n", __func__, link->link_index, req_bw, link_bw); - - return false; - } } msleep(delay_between_attempts); diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c index 3bdce32a85e3..ae95ec48e572 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_training_8b_10b.c @@ -36,7 +36,8 @@ link->ctx->logger static int32_t get_cr_training_aux_rd_interval(struct dc_link *link, - const struct dc_link_settings *link_settings) + const struct dc_link_settings *link_settings, + enum lttpr_mode lttpr_mode) { union training_aux_rd_interval training_rd_interval; uint32_t wait_in_micro_secs = 100; @@ -49,6 +50,8 @@ static int32_t get_cr_training_aux_rd_interval(struct dc_link *link, DP_TRAINING_AUX_RD_INTERVAL, (uint8_t *)&training_rd_interval, sizeof(training_rd_interval)); + if (lttpr_mode != LTTPR_MODE_NON_TRANSPARENT) + wait_in_micro_secs = 400; if (training_rd_interval.bits.TRAINIG_AUX_RD_INTERVAL) wait_in_micro_secs = training_rd_interval.bits.TRAINIG_AUX_RD_INTERVAL * 4000; } @@ -110,7 +113,6 @@ void decide_8b_10b_training_settings( */ lt_settings->link_settings.link_spread = link->dp_ss_off ? LINK_SPREAD_DISABLED : LINK_SPREAD_05_DOWNSPREAD_30KHZ; - lt_settings->cr_pattern_time = get_cr_training_aux_rd_interval(link, link_setting); lt_settings->eq_pattern_time = get_eq_training_aux_rd_interval(link, link_setting); lt_settings->pattern_for_cr = decide_cr_training_pattern(link_setting); lt_settings->pattern_for_eq = decide_eq_training_pattern(link, link_setting); @@ -119,6 +121,7 @@ void decide_8b_10b_training_settings( lt_settings->disallow_per_lane_settings = true; lt_settings->always_match_dpcd_with_hw_lane_settings = true; lt_settings->lttpr_mode = dp_decide_8b_10b_lttpr_mode(link); + lt_settings->cr_pattern_time = get_cr_training_aux_rd_interval(link, link_setting, lt_settings->lttpr_mode); dp_hw_to_dpcd_lane_settings(lt_settings, lt_settings->hw_lane_settings, lt_settings->dpcd_lane_settings); } diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c index e0e3bb865359..1e4adbc764ea 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_edp_panel_control.c @@ -675,6 +675,18 @@ bool edp_setup_psr(struct dc_link *link, if (!link) return false; + //Clear PSR cfg + memset(&psr_configuration, 0, sizeof(psr_configuration)); + dm_helpers_dp_write_dpcd( + link->ctx, + link, + DP_PSR_EN_CFG, + &psr_configuration.raw, + sizeof(psr_configuration.raw)); + + if (link->psr_settings.psr_version == DC_PSR_VERSION_UNSUPPORTED) + return false; + dc = link->ctx->dc; dmcu = dc->res_pool->dmcu; psr = dc->res_pool->psr; @@ -685,9 +697,6 @@ bool edp_setup_psr(struct dc_link *link, if (!dc_get_edp_link_panel_inst(dc, link, &panel_inst)) return false; - - memset(&psr_configuration, 0, sizeof(psr_configuration)); - psr_configuration.bits.ENABLE = 1; psr_configuration.bits.CRC_VERIFICATION = 1; psr_configuration.bits.FRAME_CAPTURE_INDICATION = @@ -950,6 +959,16 @@ bool edp_setup_replay(struct dc_link *link, const struct dc_stream_state *stream if (!link) return false; + //Clear Replay config + dm_helpers_dp_write_dpcd(link->ctx, link, + DP_SINK_PR_ENABLE_AND_CONFIGURATION, + (uint8_t *)&(replay_config.raw), sizeof(uint8_t)); + + if (!(link->replay_settings.config.replay_supported)) + return false; + + link->replay_settings.config.replay_error_status.raw = 0; + dc = link->ctx->dc; replay = dc->res_pool->replay; diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c index 14acef036b5a..6c2bb3f63be1 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn315/dcn315_resource.c @@ -1698,7 +1698,7 @@ static int dcn315_populate_dml_pipes_from_context( pipes[pipe_cnt].dout.dsc_input_bpc = 0; DC_FP_START(); dcn31_zero_pipe_dcc_fraction(pipes, pipe_cnt); - if (pixel_rate_crb && !pipe->top_pipe && !pipe->prev_odm_pipe) { + if (pixel_rate_crb) { int bpp = source_format_to_bpp(pipes[pipe_cnt].pipe.src.source_format); /* Ceil to crb segment size */ int approx_det_segs_required_for_pstate = dcn_get_approx_det_segs_required_for_pstate( @@ -1755,28 +1755,26 @@ static int dcn315_populate_dml_pipes_from_context( continue; } - if (!pipe->top_pipe && !pipe->prev_odm_pipe) { - bool split_required = pipe->stream->timing.pix_clk_100hz >= dcn_get_max_non_odm_pix_rate_100hz(&dc->dml.soc) - || (pipe->plane_state && pipe->plane_state->src_rect.width > 5120); - - if (remaining_det_segs > MIN_RESERVED_DET_SEGS && crb_pipes != 0) - pipes[pipe_cnt].pipe.src.det_size_override += (remaining_det_segs - MIN_RESERVED_DET_SEGS) / crb_pipes + - (crb_idx < (remaining_det_segs - MIN_RESERVED_DET_SEGS) % crb_pipes ? 1 : 0); - if (pipes[pipe_cnt].pipe.src.det_size_override > 2 * DCN3_15_MAX_DET_SEGS) { - /* Clamp to 2 pipe split max det segments */ - remaining_det_segs += pipes[pipe_cnt].pipe.src.det_size_override - 2 * (DCN3_15_MAX_DET_SEGS); - pipes[pipe_cnt].pipe.src.det_size_override = 2 * DCN3_15_MAX_DET_SEGS; - } - if (pipes[pipe_cnt].pipe.src.det_size_override > DCN3_15_MAX_DET_SEGS || split_required) { - /* If we are splitting we must have an even number of segments */ - remaining_det_segs += pipes[pipe_cnt].pipe.src.det_size_override % 2; - pipes[pipe_cnt].pipe.src.det_size_override -= pipes[pipe_cnt].pipe.src.det_size_override % 2; - } - /* Convert segments into size for DML use */ - pipes[pipe_cnt].pipe.src.det_size_override *= DCN3_15_CRB_SEGMENT_SIZE_KB; - - crb_idx++; + bool split_required = pipe->stream->timing.pix_clk_100hz >= dcn_get_max_non_odm_pix_rate_100hz(&dc->dml.soc) + || (pipe->plane_state && pipe->plane_state->src_rect.width > 5120); + + if (remaining_det_segs > MIN_RESERVED_DET_SEGS && crb_pipes != 0) + pipes[pipe_cnt].pipe.src.det_size_override += (remaining_det_segs - MIN_RESERVED_DET_SEGS) / crb_pipes + + (crb_idx < (remaining_det_segs - MIN_RESERVED_DET_SEGS) % crb_pipes ? 1 : 0); + if (pipes[pipe_cnt].pipe.src.det_size_override > 2 * DCN3_15_MAX_DET_SEGS) { + /* Clamp to 2 pipe split max det segments */ + remaining_det_segs += pipes[pipe_cnt].pipe.src.det_size_override - 2 * (DCN3_15_MAX_DET_SEGS); + pipes[pipe_cnt].pipe.src.det_size_override = 2 * DCN3_15_MAX_DET_SEGS; + } + if (pipes[pipe_cnt].pipe.src.det_size_override > DCN3_15_MAX_DET_SEGS || split_required) { + /* If we are splitting we must have an even number of segments */ + remaining_det_segs += pipes[pipe_cnt].pipe.src.det_size_override % 2; + pipes[pipe_cnt].pipe.src.det_size_override -= pipes[pipe_cnt].pipe.src.det_size_override % 2; } + /* Convert segments into size for DML use */ + pipes[pipe_cnt].pipe.src.det_size_override *= DCN3_15_CRB_SEGMENT_SIZE_KB; + + crb_idx++; pipe_cnt++; } } diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c index 38a9a0d68058..047f05ab0181 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c @@ -8,7 +8,7 @@ #include "dc_spl_isharp_filters.h" #include "spl_debug.h" -#define IDENTITY_RATIO(ratio) (spl_fixpt_u2d19(ratio) == (1 << 19)) +#define IDENTITY_RATIO(ratio) (spl_fixpt_u3d19(ratio) == (1 << 19)) #define MIN_VIEWPORT_SIZE 12 static bool spl_is_yuv420(enum spl_pixel_format format) @@ -76,6 +76,21 @@ static struct spl_rect shift_rec(const struct spl_rect *rec_in, int x, int y) return rec_out; } +static void spl_opp_adjust_rect(struct spl_rect *rec, const struct spl_opp_adjust *adjust) +{ + if ((rec->x + adjust->x) >= 0) + rec->x += adjust->x; + + if ((rec->y + adjust->y) >= 0) + rec->y += adjust->y; + + if ((rec->width + adjust->width) >= 1) + rec->width += adjust->width; + + if ((rec->height + adjust->height) >= 1) + rec->height += adjust->height; +} + static struct spl_rect calculate_plane_rec_in_timing_active( struct spl_in *spl_in, const struct spl_rect *rec_in) @@ -723,13 +738,15 @@ static void spl_handle_3d_recout(struct spl_in *spl_in, struct spl_rect *recout) } } -static void spl_clamp_viewport(struct spl_rect *viewport) +static void spl_clamp_viewport(struct spl_rect *viewport, int min_viewport_size) { + if (min_viewport_size == 0) + min_viewport_size = MIN_VIEWPORT_SIZE; /* Clamp minimum viewport size */ - if (viewport->height < MIN_VIEWPORT_SIZE) - viewport->height = MIN_VIEWPORT_SIZE; - if (viewport->width < MIN_VIEWPORT_SIZE) - viewport->width = MIN_VIEWPORT_SIZE; + if (viewport->height < min_viewport_size) + viewport->height = min_viewport_size; + if (viewport->width < min_viewport_size) + viewport->width = min_viewport_size; } static enum scl_mode spl_get_dscl_mode(const struct spl_in *spl_in, @@ -767,25 +784,13 @@ static enum scl_mode spl_get_dscl_mode(const struct spl_in *spl_in, return SCL_MODE_SCALING_420_YCBCR_ENABLE; } -static bool spl_choose_lls_policy(enum spl_pixel_format format, - enum spl_transfer_func_type tf_type, - enum spl_transfer_func_predefined tf_predefined_type, +static void spl_choose_lls_policy(enum spl_pixel_format format, enum linear_light_scaling *lls_pref) { - if (spl_is_video_format(format)) { + if (spl_is_subsampled_format(format)) *lls_pref = LLS_PREF_NO; - if ((tf_type == SPL_TF_TYPE_PREDEFINED) || - (tf_type == SPL_TF_TYPE_DISTRIBUTED_POINTS)) - return true; - } else { /* RGB or YUV444 */ - if ((tf_type == SPL_TF_TYPE_PREDEFINED) || - (tf_type == SPL_TF_TYPE_BYPASS)) { - *lls_pref = LLS_PREF_YES; - return true; - } - } - *lls_pref = LLS_PREF_NO; - return false; + else /* RGB or YUV444 */ + *lls_pref = LLS_PREF_YES; } /* Enable EASF ?*/ @@ -794,7 +799,6 @@ static bool enable_easf(struct spl_in *spl_in, struct spl_scratch *spl_scratch) int vratio = 0; int hratio = 0; bool skip_easf = false; - bool lls_enable_easf = true; if (spl_in->disable_easf) skip_easf = true; @@ -810,17 +814,13 @@ static bool enable_easf(struct spl_in *spl_in, struct spl_scratch *spl_scratch) skip_easf = true; /* - * If lls_pref is LLS_PREF_DONT_CARE, then use pixel format and transfer - * function to determine whether to use LINEAR or NONLINEAR scaling + * If lls_pref is LLS_PREF_DONT_CARE, then use pixel format + * to determine whether to use LINEAR or NONLINEAR scaling */ if (spl_in->lls_pref == LLS_PREF_DONT_CARE) - lls_enable_easf = spl_choose_lls_policy(spl_in->basic_in.format, - spl_in->basic_in.tf_type, spl_in->basic_in.tf_predefined_type, + spl_choose_lls_policy(spl_in->basic_in.format, &spl_in->lls_pref); - if (!lls_enable_easf) - skip_easf = true; - /* Check for linear scaling or EASF preferred */ if (spl_in->lls_pref != LLS_PREF_YES && !spl_in->prefer_easf) skip_easf = true; @@ -887,6 +887,8 @@ static bool spl_get_isharp_en(struct spl_in *spl_in, static void spl_get_taps_non_adaptive_scaler( struct spl_scratch *spl_scratch, const struct spl_taps *in_taps) { + bool check_max_downscale = false; + if (in_taps->h_taps == 0) { if (spl_fixpt_ceil(spl_scratch->scl_data.ratios.horz) > 1) spl_scratch->scl_data.taps.h_taps = spl_min(2 * spl_fixpt_ceil( @@ -926,6 +928,23 @@ static void spl_get_taps_non_adaptive_scaler( else spl_scratch->scl_data.taps.h_taps_c = in_taps->h_taps_c; + + /* + * Max downscale supported is 6.0x. Add ASSERT to catch if go beyond that + */ + check_max_downscale = spl_fixpt_le(spl_scratch->scl_data.ratios.horz, + spl_fixpt_from_fraction(6, 1)); + SPL_ASSERT(check_max_downscale); + check_max_downscale = spl_fixpt_le(spl_scratch->scl_data.ratios.vert, + spl_fixpt_from_fraction(6, 1)); + SPL_ASSERT(check_max_downscale); + check_max_downscale = spl_fixpt_le(spl_scratch->scl_data.ratios.horz_c, + spl_fixpt_from_fraction(6, 1)); + SPL_ASSERT(check_max_downscale); + check_max_downscale = spl_fixpt_le(spl_scratch->scl_data.ratios.vert_c, + spl_fixpt_from_fraction(6, 1)); + SPL_ASSERT(check_max_downscale); + if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.horz)) spl_scratch->scl_data.taps.h_taps = 1; if (IDENTITY_RATIO(spl_scratch->scl_data.ratios.vert)) @@ -944,8 +963,8 @@ static bool spl_get_optimal_number_of_taps( bool *enable_isharp) { int num_part_y, num_part_c; - int max_taps_y, max_taps_c; - int min_taps_y, min_taps_c; + unsigned int max_taps_y, max_taps_c; + unsigned int min_taps_y, min_taps_c; enum lb_memory_config lb_config; bool skip_easf = false; bool is_subsampled = spl_is_subsampled_format(spl_in->basic_in.format); @@ -1781,6 +1800,8 @@ static bool spl_calculate_number_of_taps(struct spl_in *spl_in, struct spl_scrat spl_calculate_recout(spl_in, spl_scratch, spl_out); /* depends on pixel format */ spl_calculate_scaling_ratios(spl_in, spl_scratch, spl_out); + /* Adjust recout for opp if needed */ + spl_opp_adjust_rect(&spl_scratch->scl_data.recout, &spl_in->basic_in.opp_recout_adjust); /* depends on scaling ratios and recout, does not calculate offset yet */ spl_calculate_viewport_size(spl_in, spl_scratch); @@ -1817,7 +1838,7 @@ bool spl_calculate_scaler_params(struct spl_in *spl_in, struct spl_out *spl_out) // Handle 3d recout spl_handle_3d_recout(spl_in, &spl_scratch.scl_data.recout); // Clamp - spl_clamp_viewport(&spl_scratch.scl_data.viewport); + spl_clamp_viewport(&spl_scratch.scl_data.viewport, spl_in->min_viewport_size); // Save all calculated parameters in dscl_prog_data structure to program hw registers spl_set_dscl_prog_data(spl_in, &spl_scratch, spl_out, enable_easf_v, enable_easf_h, enable_isharp); diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h index 467af9dd90de..1c3949b24611 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl_types.h @@ -427,6 +427,14 @@ struct spl_out { // SPL inputs +// opp extra adjustment for rect +struct spl_opp_adjust { + int x; + int y; + int width; + int height; +}; + // Basic input information struct basic_in { enum spl_pixel_format format; // Pixel Format @@ -444,6 +452,7 @@ struct basic_in { } num_slices_recout_width; } num_h_slices_recout_width_align; int mpc_h_slice_index; // previous mpc_combine_v - split_idx + struct spl_opp_adjust opp_recout_adjust; // Inputs for adaptive scaler - TODO enum spl_transfer_func_type tf_type; /* Transfer function type */ enum spl_transfer_func_predefined tf_predefined_type; /* Transfer function predefined type */ @@ -484,7 +493,7 @@ struct spl_sharpness_range { }; struct adaptive_sharpness { bool enable; - int sharpness_level; + unsigned int sharpness_level; struct spl_sharpness_range sharpness_range; }; enum linear_light_scaling { // convert it in translation logic @@ -535,6 +544,7 @@ struct spl_in { bool is_hdr_on; int h_active; int v_active; + int min_viewport_size; int sdr_white_level_nits; enum sharpen_policy sharpen_policy; }; diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c index 131f1e3949d3..52d97918a3bd 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.c @@ -346,7 +346,7 @@ struct spl_fixed31_32 spl_fixpt_exp(struct spl_fixed31_32 arg) if (m > 0) return spl_fixpt_shl( spl_fixed31_32_exp_from_taylor_series(r), - (unsigned char)m); + (unsigned int)m); else return spl_fixpt_div_int( spl_fixed31_32_exp_from_taylor_series(r), diff --git a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h index ed2647f9a099..9f349ffe9148 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h +++ b/drivers/gpu/drm/amd/display/dc/spl/spl_fixpt31_32.h @@ -189,7 +189,7 @@ static inline struct spl_fixed31_32 spl_fixpt_clamp( * @brief * result = arg << shift */ -static inline struct spl_fixed31_32 spl_fixpt_shl(struct spl_fixed31_32 arg, unsigned char shift) +static inline struct spl_fixed31_32 spl_fixpt_shl(struct spl_fixed31_32 arg, unsigned int shift) { SPL_ASSERT(((arg.value >= 0) && (arg.value <= LLONG_MAX >> shift)) || ((arg.value < 0) && (arg.value >= ~(LLONG_MAX >> shift)))); @@ -203,7 +203,7 @@ static inline struct spl_fixed31_32 spl_fixpt_shl(struct spl_fixed31_32 arg, uns * @brief * result = arg >> shift */ -static inline struct spl_fixed31_32 spl_fixpt_shr(struct spl_fixed31_32 arg, unsigned char shift) +static inline struct spl_fixed31_32 spl_fixpt_shr(struct spl_fixed31_32 arg, unsigned int shift) { bool negative = arg.value < 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 d0fe324cb537..8cf89aed024b 100644 --- a/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h +++ b/drivers/gpu/drm/amd/display/dmub/inc/dmub_cmd.h @@ -3118,6 +3118,12 @@ struct dmub_cmd_psr_copy_settings_data { * Some panels request main link off before xth vertical line */ uint16_t poweroff_before_vertical_line; + /** + * Some panels cannot handle idle pattern during PSR entry. + * To power down phy before disable stream to avoid sending + * idle pattern. + */ + uint8_t power_down_phy_before_disable_stream; }; /** diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c index d9f31b191c69..1a68b5782cac 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn31.c @@ -83,8 +83,8 @@ static inline void dmub_dcn31_translate_addr(const union dmub_addr *addr_in, void dmub_dcn31_reset(struct dmub_srv *dmub) { union dmub_gpint_data_register cmd; - const uint32_t timeout = 100; - uint32_t in_reset, scratch, i, pwait_mode; + const uint32_t timeout = 100000; + uint32_t in_reset, is_enabled, scratch, i, pwait_mode; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); @@ -108,7 +108,7 @@ void dmub_dcn31_reset(struct dmub_srv *dmub) } for (i = 0; i < timeout; ++i) { - scratch = dmub->hw_funcs.get_gpint_response(dmub); + scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; @@ -125,9 +125,14 @@ void dmub_dcn31_reset(struct dmub_srv *dmub) /* Force reset in case we timed out, DMCUB is likely hung. */ } - REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); - REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); + REG_GET(DMCUB_CNTL, DMCUB_ENABLE, &is_enabled); + + if (is_enabled) { + REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); + REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); + } + REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c index e5e77bd3c31e..01d013a12b94 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn35.c @@ -88,7 +88,7 @@ static inline void dmub_dcn35_translate_addr(const union dmub_addr *addr_in, void dmub_dcn35_reset(struct dmub_srv *dmub) { union dmub_gpint_data_register cmd; - const uint32_t timeout = 100; + const uint32_t timeout = 100000; uint32_t in_reset, is_enabled, scratch, i, pwait_mode; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); @@ -113,7 +113,7 @@ void dmub_dcn35_reset(struct dmub_srv *dmub) } for (i = 0; i < timeout; ++i) { - scratch = dmub->hw_funcs.get_gpint_response(dmub); + scratch = REG_READ(DMCUB_SCRATCH7); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c index 39a8cb6d7523..e1c4fe1c6e3e 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.c @@ -63,8 +63,10 @@ static inline void dmub_dcn401_translate_addr(const union dmub_addr *addr_in, void dmub_dcn401_reset(struct dmub_srv *dmub) { union dmub_gpint_data_register cmd; - const uint32_t timeout = 30; - uint32_t in_reset, scratch, i; + const uint32_t timeout_us = 1 * 1000 * 1000; //1s + const uint32_t poll_delay_us = 1; //1us + uint32_t i = 0; + uint32_t in_reset, scratch, pwait_mode; REG_GET(DMCUB_CNTL2, DMCUB_SOFT_RESET, &in_reset); @@ -75,32 +77,35 @@ void dmub_dcn401_reset(struct dmub_srv *dmub) dmub->hw_funcs.set_gpint(dmub, cmd); - /** - * Timeout covers both the ACK and the wait - * for remaining work to finish. - * - * This is mostly bound by the PHY disable sequence. - * Each register check will be greater than 1us, so - * don't bother using udelay. - */ - - for (i = 0; i < timeout; ++i) { + for (i = 0; i < timeout_us; i++) { if (dmub->hw_funcs.is_gpint_acked(dmub, cmd)) break; + + udelay(poll_delay_us); } - for (i = 0; i < timeout; ++i) { + for (; i < timeout_us; i++) { scratch = dmub->hw_funcs.get_gpint_response(dmub); if (scratch == DMUB_GPINT__STOP_FW_RESPONSE) break; + + udelay(poll_delay_us); } - /* Force reset in case we timed out, DMCUB is likely hung. */ + for (; i < timeout_us; i++) { + REG_GET(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS, &pwait_mode); + if (pwait_mode & (1 << 0)) + break; + + udelay(poll_delay_us); + } + } + + if (i >= timeout_us) { + /* timeout should never occur */ + BREAK_TO_DEBUGGER(); } - REG_UPDATE(DMCUB_CNTL2, DMCUB_SOFT_RESET, 1); - REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); - REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); REG_WRITE(DMCUB_INBOX1_RPTR, 0); REG_WRITE(DMCUB_INBOX1_WPTR, 0); REG_WRITE(DMCUB_OUTBOX1_RPTR, 0); @@ -131,7 +136,10 @@ void dmub_dcn401_backdoor_load(struct dmub_srv *dmub, dmub_dcn401_get_fb_base_offset(dmub, &fb_base, &fb_offset); + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); dmub_dcn401_translate_addr(&cw0->offset, fb_base, fb_offset, &offset); @@ -151,6 +159,7 @@ void dmub_dcn401_backdoor_load(struct dmub_srv *dmub, DMCUB_REGION3_CW1_TOP_ADDRESS, cw1->region.top, DMCUB_REGION3_CW1_ENABLE, 1); + /* release DMCUB reset only to prevent premature execution */ REG_UPDATE_2(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 0, DMCUB_MEM_UNIT_ID, 0x20); } @@ -161,7 +170,10 @@ void dmub_dcn401_backdoor_load_zfb_mode(struct dmub_srv *dmub, { union dmub_addr offset; + /* reset and disable DMCUB and MMHUBBUB DMUIF */ REG_UPDATE(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 1); + REG_UPDATE(MMHUBBUB_SOFT_RESET, DMUIF_SOFT_RESET, 1); + REG_UPDATE(DMCUB_CNTL, DMCUB_ENABLE, 0); offset = cw0->offset; @@ -181,6 +193,7 @@ void dmub_dcn401_backdoor_load_zfb_mode(struct dmub_srv *dmub, DMCUB_REGION3_CW1_TOP_ADDRESS, cw1->region.top, DMCUB_REGION3_CW1_ENABLE, 1); + /* release DMCUB reset only to prevent premature execution */ REG_UPDATE_2(DMCUB_SEC_CNTL, DMCUB_SEC_RESET, 0, DMCUB_MEM_UNIT_ID, 0x20); } diff --git a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h index 4c8843b79695..31f95b27e227 100644 --- a/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h +++ b/drivers/gpu/drm/amd/display/dmub/src/dmub_dcn401.h @@ -169,7 +169,8 @@ struct dmub_srv; DMUB_SF(HOST_INTERRUPT_CSR, HOST_REG_INBOX0_RSP_INT_EN) \ DMUB_SF(HOST_INTERRUPT_CSR, HOST_REG_OUTBOX0_RDY_INT_ACK) \ DMUB_SF(HOST_INTERRUPT_CSR, HOST_REG_OUTBOX0_RDY_INT_STAT) \ - DMUB_SF(HOST_INTERRUPT_CSR, HOST_REG_OUTBOX0_RDY_INT_EN) + DMUB_SF(HOST_INTERRUPT_CSR, HOST_REG_OUTBOX0_RDY_INT_EN) \ + DMUB_SF(DMCUB_CNTL, DMCUB_PWAIT_MODE_STATUS) struct dmub_srv_dcn401_reg_offset { #define DMUB_SR(reg) uint32_t reg; diff --git a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c index a344e2e49b0e..b3d55cac3569 100644 --- a/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c +++ b/drivers/gpu/drm/amd/display/modules/info_packet/info_packet.c @@ -383,10 +383,10 @@ void mod_build_vsc_infopacket(const struct dc_stream_state *stream, colorimetryFormat = ColorimetryYCC_DP_ITU709; else if (cs == COLOR_SPACE_ADOBERGB) colorimetryFormat = ColorimetryYCC_DP_AdobeYCC; - else if (cs == COLOR_SPACE_2020_YCBCR) + else if (cs == COLOR_SPACE_2020_YCBCR_LIMITED) colorimetryFormat = ColorimetryYCC_DP_ITU2020YCbCr; - if (cs == COLOR_SPACE_2020_YCBCR && tf == TRANSFER_FUNC_GAMMA_22) + if (cs == COLOR_SPACE_2020_YCBCR_LIMITED && tf == TRANSFER_FUNC_GAMMA_22) colorimetryFormat = ColorimetryYCC_DP_ITU709; break; diff --git a/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_offset.h b/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_offset.h index c488d4a50cf4..b2252deabc17 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_offset.h +++ b/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_offset.h @@ -203,6 +203,10 @@ #define mmDAGB0_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB0_WR_MISC_CREDIT 0x0058 #define mmDAGB0_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE 0x005b +#define mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x005c +#define mmDAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB0_WRCLI_ASK_PENDING 0x005d #define mmDAGB0_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB0_WRCLI_GO_PENDING 0x005e @@ -455,6 +459,10 @@ #define mmDAGB1_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB1_WR_MISC_CREDIT 0x00d8 #define mmDAGB1_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE 0x00db +#define mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x00dc +#define mmDAGB1_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB1_WRCLI_ASK_PENDING 0x00dd #define mmDAGB1_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB1_WRCLI_GO_PENDING 0x00de @@ -707,6 +715,10 @@ #define mmDAGB2_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB2_WR_MISC_CREDIT 0x0158 #define mmDAGB2_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB2_WRCLI_GPU_SNOOP_OVERRIDE 0x015b +#define mmDAGB2_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB2_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x015c +#define mmDAGB2_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB2_WRCLI_ASK_PENDING 0x015d #define mmDAGB2_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB2_WRCLI_GO_PENDING 0x015e @@ -959,6 +971,10 @@ #define mmDAGB3_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB3_WR_MISC_CREDIT 0x01d8 #define mmDAGB3_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB3_WRCLI_GPU_SNOOP_OVERRIDE 0x01db +#define mmDAGB3_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB3_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x01dc +#define mmDAGB3_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB3_WRCLI_ASK_PENDING 0x01dd #define mmDAGB3_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB3_WRCLI_GO_PENDING 0x01de @@ -1211,6 +1227,10 @@ #define mmDAGB4_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB4_WR_MISC_CREDIT 0x0258 #define mmDAGB4_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB4_WRCLI_GPU_SNOOP_OVERRIDE 0x025b +#define mmDAGB4_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB4_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x025c +#define mmDAGB4_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB4_WRCLI_ASK_PENDING 0x025d #define mmDAGB4_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB4_WRCLI_GO_PENDING 0x025e @@ -4793,6 +4813,10 @@ #define mmDAGB5_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB5_WR_MISC_CREDIT 0x3058 #define mmDAGB5_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB5_WRCLI_GPU_SNOOP_OVERRIDE 0x305b +#define mmDAGB5_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB5_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x305c +#define mmDAGB5_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB5_WRCLI_ASK_PENDING 0x305d #define mmDAGB5_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB5_WRCLI_GO_PENDING 0x305e @@ -5045,6 +5069,10 @@ #define mmDAGB6_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB6_WR_MISC_CREDIT 0x30d8 #define mmDAGB6_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB6_WRCLI_GPU_SNOOP_OVERRIDE 0x30db +#define mmDAGB6_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB6_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x30dc +#define mmDAGB6_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB6_WRCLI_ASK_PENDING 0x30dd #define mmDAGB6_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB6_WRCLI_GO_PENDING 0x30de @@ -5297,6 +5325,10 @@ #define mmDAGB7_WR_DATA_CREDIT_BASE_IDX 1 #define mmDAGB7_WR_MISC_CREDIT 0x3158 #define mmDAGB7_WR_MISC_CREDIT_BASE_IDX 1 +#define mmDAGB7_WRCLI_GPU_SNOOP_OVERRIDE 0x315b +#define mmDAGB7_WRCLI_GPU_SNOOP_OVERRIDE_BASE_IDX 1 +#define mmDAGB7_WRCLI_GPU_SNOOP_OVERRIDE_VALUE 0x315c +#define mmDAGB7_WRCLI_GPU_SNOOP_OVERRIDE_VALUE_BASE_IDX 1 #define mmDAGB7_WRCLI_ASK_PENDING 0x315d #define mmDAGB7_WRCLI_ASK_PENDING_BASE_IDX 1 #define mmDAGB7_WRCLI_GO_PENDING 0x315e diff --git a/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_sh_mask.h b/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_sh_mask.h index 2969fbf282b7..5069d2fd467f 100644 --- a/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_sh_mask.h +++ b/drivers/gpu/drm/amd/include/asic_reg/mmhub/mmhub_9_4_1_sh_mask.h @@ -1532,6 +1532,12 @@ //DAGB0_WRCLI_DBUS_GO_PENDING #define DAGB0_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB0_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB0_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB0_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB0_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB0_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB0_DAGB_DLY #define DAGB0_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB0_DAGB_DLY__CLI__SHIFT 0x8 @@ -3207,6 +3213,12 @@ //DAGB1_WRCLI_DBUS_GO_PENDING #define DAGB1_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB1_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB1_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB1_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB1_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB1_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB1_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB1_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB1_DAGB_DLY #define DAGB1_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB1_DAGB_DLY__CLI__SHIFT 0x8 @@ -4882,6 +4894,12 @@ //DAGB2_WRCLI_DBUS_GO_PENDING #define DAGB2_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB2_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB2_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB2_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB2_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB2_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB2_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB2_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB2_DAGB_DLY #define DAGB2_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB2_DAGB_DLY__CLI__SHIFT 0x8 @@ -6557,6 +6575,12 @@ //DAGB3_WRCLI_DBUS_GO_PENDING #define DAGB3_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB3_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB3_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB3_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB3_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB3_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB3_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB3_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB3_DAGB_DLY #define DAGB3_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB3_DAGB_DLY__CLI__SHIFT 0x8 @@ -8232,6 +8256,12 @@ //DAGB4_WRCLI_DBUS_GO_PENDING #define DAGB4_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB4_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB4_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB4_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB4_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB4_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB4_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB4_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB4_DAGB_DLY #define DAGB4_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB4_DAGB_DLY__CLI__SHIFT 0x8 @@ -28737,6 +28767,12 @@ //DAGB5_WRCLI_DBUS_GO_PENDING #define DAGB5_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB5_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB5_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB5_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB5_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB5_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB5_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB5_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB5_DAGB_DLY #define DAGB5_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB5_DAGB_DLY__CLI__SHIFT 0x8 @@ -30412,6 +30448,12 @@ //DAGB6_WRCLI_DBUS_GO_PENDING #define DAGB6_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB6_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB6_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB6_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB6_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB6_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB6_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB6_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB6_DAGB_DLY #define DAGB6_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB6_DAGB_DLY__CLI__SHIFT 0x8 @@ -32087,6 +32129,12 @@ //DAGB7_WRCLI_DBUS_GO_PENDING #define DAGB7_WRCLI_DBUS_GO_PENDING__BUSY__SHIFT 0x0 #define DAGB7_WRCLI_DBUS_GO_PENDING__BUSY_MASK 0xFFFFFFFFL +//DAGB7_WRCLI_GPU_SNOOP_OVERRIDE +#define DAGB7_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE__SHIFT 0x0 +#define DAGB7_WRCLI_GPU_SNOOP_OVERRIDE__ENABLE_MASK 0xFFFFFFFFL +//DAGB7_WRCLI_GPU_SNOOP_OVERRIDE_VALUE +#define DAGB7_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE__SHIFT 0x0 +#define DAGB7_WRCLI_GPU_SNOOP_OVERRIDE_VALUE__ENABLE_MASK 0xFFFFFFFFL //DAGB7_DAGB_DLY #define DAGB7_DAGB_DLY__DLY__SHIFT 0x0 #define DAGB7_DAGB_DLY__CLI__SHIFT 0x8 diff --git a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c index ed9dac00ebfb..f3f5b7dd15cc 100644 --- a/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c +++ b/drivers/gpu/drm/amd/pm/swsmu/amdgpu_smu.c @@ -2802,6 +2802,7 @@ int smu_get_power_limit(void *handle, switch (amdgpu_ip_version(adev, MP1_HWIP, 0)) { case IP_VERSION(13, 0, 2): case IP_VERSION(13, 0, 6): + case IP_VERSION(13, 0, 12): case IP_VERSION(13, 0, 14): case IP_VERSION(11, 0, 7): case IP_VERSION(11, 0, 11): 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 da7bd9227afe..5f2a824918e3 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 @@ -450,8 +450,9 @@ static int smu_v13_0_6_init_microcode(struct smu_context *smu) int var = (adev->pdev->device & 0xF); char ucode_prefix[15]; - /* No need to load P2S tables in IOV mode */ - if (amdgpu_sriov_vf(adev)) + /* No need to load P2S tables in IOV mode or for smu v13.0.12 */ + if (amdgpu_sriov_vf(adev) || + (amdgpu_ip_version(smu->adev, MP1_HWIP, 0) == IP_VERSION(13, 0, 12))) return 0; if (!(adev->flags & AMD_IS_APU)) { diff --git a/drivers/gpu/drm/ast/ast_main.c b/drivers/gpu/drm/ast/ast_main.c index bc37c65305d4..96470fc8e6e5 100644 --- a/drivers/gpu/drm/ast/ast_main.c +++ b/drivers/gpu/drm/ast/ast_main.c @@ -96,21 +96,21 @@ static void ast_detect_tx_chip(struct ast_device *ast, bool need_post) /* Check 3rd Tx option (digital output afaik) */ ast->tx_chip = AST_TX_NONE; - /* - * VGACRA3 Enhanced Color Mode Register, check if DVO is already - * enabled, in that case, assume we have a SIL164 TMDS transmitter - * - * Don't make that assumption if we the chip wasn't enabled and - * is at power-on reset, otherwise we'll incorrectly "detect" a - * SIL164 when there is none. - */ - if (!need_post) { - jreg = ast_get_index_reg_mask(ast, AST_IO_VGACRI, 0xa3, 0xff); - if (jreg & 0x80) - ast->tx_chip = AST_TX_SIL164; - } - - if (IS_AST_GEN4(ast) || IS_AST_GEN5(ast) || IS_AST_GEN6(ast)) { + if (AST_GEN(ast) <= 3) { + /* + * VGACRA3 Enhanced Color Mode Register, check if DVO is already + * enabled, in that case, assume we have a SIL164 TMDS transmitter + * + * Don't make that assumption if we the chip wasn't enabled and + * is at power-on reset, otherwise we'll incorrectly "detect" a + * SIL164 when there is none. + */ + if (!need_post) { + jreg = ast_get_index_reg_mask(ast, AST_IO_VGACRI, 0xa3, 0xff); + if (jreg & 0x80) + ast->tx_chip = AST_TX_SIL164; + } + } else if (IS_AST_GEN4(ast) || IS_AST_GEN5(ast) || IS_AST_GEN6(ast)) { /* * On AST GEN4+, look the configuration set by the SoC in * the SOC scratch register #1 bits 11:8 (interestingly marked diff --git a/drivers/gpu/drm/ast/ast_mode.c b/drivers/gpu/drm/ast/ast_mode.c index 9d5321c81e68..a29fe1ae803f 100644 --- a/drivers/gpu/drm/ast/ast_mode.c +++ b/drivers/gpu/drm/ast/ast_mode.c @@ -131,7 +131,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, return false; } - switch (mode->crtc_hdisplay) { + switch (mode->hdisplay) { case 640: vbios_mode->enh_table = &res_640x480[refresh_rate_index]; break; @@ -145,7 +145,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, vbios_mode->enh_table = &res_1152x864[refresh_rate_index]; break; case 1280: - if (mode->crtc_vdisplay == 800) + if (mode->vdisplay == 800) vbios_mode->enh_table = &res_1280x800[refresh_rate_index]; else vbios_mode->enh_table = &res_1280x1024[refresh_rate_index]; @@ -157,7 +157,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, vbios_mode->enh_table = &res_1440x900[refresh_rate_index]; break; case 1600: - if (mode->crtc_vdisplay == 900) + if (mode->vdisplay == 900) vbios_mode->enh_table = &res_1600x900[refresh_rate_index]; else vbios_mode->enh_table = &res_1600x1200[refresh_rate_index]; @@ -166,7 +166,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, vbios_mode->enh_table = &res_1680x1050[refresh_rate_index]; break; case 1920: - if (mode->crtc_vdisplay == 1080) + if (mode->vdisplay == 1080) vbios_mode->enh_table = &res_1920x1080[refresh_rate_index]; else vbios_mode->enh_table = &res_1920x1200[refresh_rate_index]; @@ -210,6 +210,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, hborder = (vbios_mode->enh_table->flags & HBorder) ? 8 : 0; vborder = (vbios_mode->enh_table->flags & VBorder) ? 8 : 0; + adjusted_mode->crtc_hdisplay = vbios_mode->enh_table->hde; adjusted_mode->crtc_htotal = vbios_mode->enh_table->ht; adjusted_mode->crtc_hblank_start = vbios_mode->enh_table->hde + hborder; adjusted_mode->crtc_hblank_end = vbios_mode->enh_table->ht - hborder; @@ -219,6 +220,7 @@ static bool ast_get_vbios_mode_info(const struct drm_format_info *format, vbios_mode->enh_table->hfp + vbios_mode->enh_table->hsync); + adjusted_mode->crtc_vdisplay = vbios_mode->enh_table->vde; adjusted_mode->crtc_vtotal = vbios_mode->enh_table->vt; adjusted_mode->crtc_vblank_start = vbios_mode->enh_table->vde + vborder; adjusted_mode->crtc_vblank_end = vbios_mode->enh_table->vt - vborder; diff --git a/drivers/gpu/drm/bridge/adv7511/adv7511_audio.c b/drivers/gpu/drm/bridge/adv7511/adv7511_audio.c index 657bc3dd18df..98030500a978 100644 --- a/drivers/gpu/drm/bridge/adv7511/adv7511_audio.c +++ b/drivers/gpu/drm/bridge/adv7511/adv7511_audio.c @@ -245,7 +245,9 @@ static const struct hdmi_codec_pdata codec_data = { .ops = &adv7511_codec_ops, .max_i2s_channels = 2, .i2s = 1, + .no_i2s_capture = 1, .spdif = 1, + .no_spdif_capture = 1, }; int adv7511_audio_init(struct device *dev, struct adv7511 *adv7511) diff --git a/drivers/gpu/drm/drm_atomic_helper.c b/drivers/gpu/drm/drm_atomic_helper.c index 32902f77f00d..40e4e1b6c911 100644 --- a/drivers/gpu/drm/drm_atomic_helper.c +++ b/drivers/gpu/drm/drm_atomic_helper.c @@ -574,6 +574,30 @@ mode_valid(struct drm_atomic_state *state) return 0; } +static int drm_atomic_check_valid_clones(struct drm_atomic_state *state, + struct drm_crtc *crtc) +{ + struct drm_encoder *drm_enc; + struct drm_crtc_state *crtc_state = drm_atomic_get_new_crtc_state(state, + crtc); + + drm_for_each_encoder_mask(drm_enc, crtc->dev, crtc_state->encoder_mask) { + if (!drm_enc->possible_clones) { + DRM_DEBUG("enc%d possible_clones is 0\n", drm_enc->base.id); + continue; + } + + if ((crtc_state->encoder_mask & drm_enc->possible_clones) != + crtc_state->encoder_mask) { + DRM_DEBUG("crtc%d failed valid clone check for mask 0x%x\n", + crtc->base.id, crtc_state->encoder_mask); + return -EINVAL; + } + } + + return 0; +} + /** * drm_atomic_helper_check_modeset - validate state object for modeset changes * @dev: DRM device @@ -745,6 +769,10 @@ drm_atomic_helper_check_modeset(struct drm_device *dev, ret = drm_atomic_add_affected_planes(state, crtc); if (ret != 0) return ret; + + ret = drm_atomic_check_valid_clones(state, crtc); + if (ret != 0) + return ret; } /* diff --git a/drivers/gpu/drm/drm_buddy.c b/drivers/gpu/drm/drm_buddy.c index 103c185bb1c8..ca42e6081d27 100644 --- a/drivers/gpu/drm/drm_buddy.c +++ b/drivers/gpu/drm/drm_buddy.c @@ -324,7 +324,7 @@ EXPORT_SYMBOL(drm_buddy_init); */ void drm_buddy_fini(struct drm_buddy *mm) { - u64 root_size, size; + u64 root_size, size, start; unsigned int order; int i; @@ -332,7 +332,8 @@ void drm_buddy_fini(struct drm_buddy *mm) for (i = 0; i < mm->n_roots; ++i) { order = ilog2(size) - ilog2(mm->chunk_size); - __force_merge(mm, 0, size, order); + start = drm_buddy_block_offset(mm->roots[i]); + __force_merge(mm, start, start + size, order); WARN_ON(!drm_buddy_block_is_free(mm->roots[i])); drm_block_free(mm, mm->roots[i]); diff --git a/drivers/gpu/drm/drm_edid.c b/drivers/gpu/drm/drm_edid.c index 13bc4c290b17..9edb3247c767 100644 --- a/drivers/gpu/drm/drm_edid.c +++ b/drivers/gpu/drm/drm_edid.c @@ -6596,6 +6596,7 @@ static void drm_reset_display_info(struct drm_connector *connector) info->has_hdmi_infoframe = false; info->rgb_quant_range_selectable = false; memset(&info->hdmi, 0, sizeof(info->hdmi)); + memset(&connector->hdr_sink_metadata, 0, sizeof(connector->hdr_sink_metadata)); info->edid_hdmi_rgb444_dc_modes = 0; info->edid_hdmi_ycbcr444_dc_modes = 0; diff --git a/drivers/gpu/drm/drm_file.c b/drivers/gpu/drm/drm_file.c index c299cd94d3f7..cf2463090d3a 100644 --- a/drivers/gpu/drm/drm_file.c +++ b/drivers/gpu/drm/drm_file.c @@ -964,6 +964,10 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f) struct drm_file *file = f->private_data; struct drm_device *dev = file->minor->dev; struct drm_printer p = drm_seq_file_printer(m); + int idx; + + if (!drm_dev_enter(dev, &idx)) + return; drm_printf(&p, "drm-driver:\t%s\n", dev->driver->name); drm_printf(&p, "drm-client-id:\t%llu\n", file->client_id); @@ -983,6 +987,8 @@ void drm_show_fdinfo(struct seq_file *m, struct file *f) if (dev->driver->show_fdinfo) dev->driver->show_fdinfo(&p, file); + + drm_dev_exit(idx); } EXPORT_SYMBOL(drm_show_fdinfo); diff --git a/drivers/gpu/drm/drm_gem.c b/drivers/gpu/drm/drm_gem.c index ee811764c3df..c6240bab3fa5 100644 --- a/drivers/gpu/drm/drm_gem.c +++ b/drivers/gpu/drm/drm_gem.c @@ -348,7 +348,7 @@ int drm_gem_dumb_map_offset(struct drm_file *file, struct drm_device *dev, return -ENOENT; /* Don't allow imported objects to be mapped */ - if (obj->import_attach) { + if (drm_gem_is_imported(obj)) { ret = -EINVAL; goto out; } @@ -1178,7 +1178,7 @@ void drm_gem_print_info(struct drm_printer *p, unsigned int indent, drm_vma_node_start(&obj->vma_node)); drm_printf_indent(p, indent, "size=%zu\n", obj->size); drm_printf_indent(p, indent, "imported=%s\n", - str_yes_no(obj->import_attach)); + str_yes_no(drm_gem_is_imported(obj))); if (obj->funcs->print_info) obj->funcs->print_info(p, indent, obj); diff --git a/drivers/gpu/drm/drm_mipi_dbi.c b/drivers/gpu/drm/drm_mipi_dbi.c index 34bca7567576..3ea9f23b4f67 100644 --- a/drivers/gpu/drm/drm_mipi_dbi.c +++ b/drivers/gpu/drm/drm_mipi_dbi.c @@ -404,12 +404,16 @@ static void mipi_dbi_blank(struct mipi_dbi_dev *dbidev) u16 height = drm->mode_config.min_height; u16 width = drm->mode_config.min_width; struct mipi_dbi *dbi = &dbidev->dbi; - size_t len = width * height * 2; + const struct drm_format_info *dst_format; + size_t len; int idx; if (!drm_dev_enter(drm, &idx)) return; + dst_format = drm_format_info(dbidev->pixel_format); + len = drm_format_info_min_pitch(dst_format, 0, width) * height; + memset(dbidev->tx_buf, 0, len); mipi_dbi_set_window_address(dbidev, 0, width - 1, 0, height - 1); diff --git a/drivers/gpu/drm/i915/display/intel_dp_mst.c b/drivers/gpu/drm/i915/display/intel_dp_mst.c index 86d6185fda50..8a6135b179d3 100644 --- a/drivers/gpu/drm/i915/display/intel_dp_mst.c +++ b/drivers/gpu/drm/i915/display/intel_dp_mst.c @@ -221,6 +221,7 @@ int intel_dp_mtp_tu_compute_config(struct intel_dp *intel_dp, to_intel_connector(conn_state->connector); const struct drm_display_mode *adjusted_mode = &crtc_state->hw.adjusted_mode; + bool is_mst = intel_crtc_has_type(crtc_state, INTEL_OUTPUT_DP_MST); fixed20_12 pbn_div; int bpp, slots = -EINVAL; int dsc_slice_count = 0; @@ -271,7 +272,7 @@ int intel_dp_mtp_tu_compute_config(struct intel_dp *intel_dp, link_bpp_x16, &crtc_state->dp_m_n); - if (intel_dp->is_mst) { + if (is_mst) { int remote_bw_overhead; int remote_tu; fixed20_12 pbn; diff --git a/drivers/gpu/drm/i915/pxp/intel_pxp_gsccs.h b/drivers/gpu/drm/i915/pxp/intel_pxp_gsccs.h index 9aae779c4da3..4969d3de2bac 100644 --- a/drivers/gpu/drm/i915/pxp/intel_pxp_gsccs.h +++ b/drivers/gpu/drm/i915/pxp/intel_pxp_gsccs.h @@ -23,6 +23,7 @@ int intel_pxp_gsccs_init(struct intel_pxp *pxp); int intel_pxp_gsccs_create_session(struct intel_pxp *pxp, int arb_session_id); void intel_pxp_gsccs_end_arb_fw_session(struct intel_pxp *pxp, u32 arb_session_id); +bool intel_pxp_gsccs_is_ready_for_sessions(struct intel_pxp *pxp); #else static inline void intel_pxp_gsccs_fini(struct intel_pxp *pxp) @@ -34,8 +35,11 @@ static inline int intel_pxp_gsccs_init(struct intel_pxp *pxp) return 0; } -#endif +static inline bool intel_pxp_gsccs_is_ready_for_sessions(struct intel_pxp *pxp) +{ + return false; +} -bool intel_pxp_gsccs_is_ready_for_sessions(struct intel_pxp *pxp); +#endif #endif /*__INTEL_PXP_GSCCS_H__ */ diff --git a/drivers/gpu/drm/mediatek/mtk_dpi.c b/drivers/gpu/drm/mediatek/mtk_dpi.c index a12ef24c7742..b7d90574df9a 100644 --- a/drivers/gpu/drm/mediatek/mtk_dpi.c +++ b/drivers/gpu/drm/mediatek/mtk_dpi.c @@ -410,12 +410,13 @@ static void mtk_dpi_config_swap_input(struct mtk_dpi *dpi, bool enable) static void mtk_dpi_config_2n_h_fre(struct mtk_dpi *dpi) { - mtk_dpi_mask(dpi, dpi->conf->reg_h_fre_con, H_FRE_2N, H_FRE_2N); + if (dpi->conf->reg_h_fre_con) + mtk_dpi_mask(dpi, dpi->conf->reg_h_fre_con, H_FRE_2N, H_FRE_2N); } static void mtk_dpi_config_disable_edge(struct mtk_dpi *dpi) { - if (dpi->conf->edge_sel_en) + if (dpi->conf->edge_sel_en && dpi->conf->reg_h_fre_con) mtk_dpi_mask(dpi, dpi->conf->reg_h_fre_con, 0, EDGE_SEL_EN); } diff --git a/drivers/gpu/drm/meson/meson_drv.c b/drivers/gpu/drm/meson/meson_drv.c index 81d2ee37e773..49ff9f1f16d3 100644 --- a/drivers/gpu/drm/meson/meson_drv.c +++ b/drivers/gpu/drm/meson/meson_drv.c @@ -169,7 +169,7 @@ static const struct meson_drm_soc_attr meson_drm_soc_attrs[] = { /* S805X/S805Y HDMI PLL won't lock for HDMI PHY freq > 1,65GHz */ { .limits = { - .max_hdmi_phy_freq = 1650000, + .max_hdmi_phy_freq = 1650000000, }, .attrs = (const struct soc_device_attribute []) { { .soc_id = "GXL (S805*)", }, diff --git a/drivers/gpu/drm/meson/meson_drv.h b/drivers/gpu/drm/meson/meson_drv.h index 3f9345c14f31..be4b0e4df6e1 100644 --- a/drivers/gpu/drm/meson/meson_drv.h +++ b/drivers/gpu/drm/meson/meson_drv.h @@ -37,7 +37,7 @@ struct meson_drm_match_data { }; struct meson_drm_soc_limits { - unsigned int max_hdmi_phy_freq; + unsigned long long max_hdmi_phy_freq; }; struct meson_drm { diff --git a/drivers/gpu/drm/meson/meson_encoder_hdmi.c b/drivers/gpu/drm/meson/meson_encoder_hdmi.c index 0593a1cde906..c61e6026adfa 100644 --- a/drivers/gpu/drm/meson/meson_encoder_hdmi.c +++ b/drivers/gpu/drm/meson/meson_encoder_hdmi.c @@ -70,12 +70,12 @@ static void meson_encoder_hdmi_set_vclk(struct meson_encoder_hdmi *encoder_hdmi, { struct meson_drm *priv = encoder_hdmi->priv; int vic = drm_match_cea_mode(mode); - unsigned int phy_freq; - unsigned int vclk_freq; - unsigned int venc_freq; - unsigned int hdmi_freq; + unsigned long long phy_freq; + unsigned long long vclk_freq; + unsigned long long venc_freq; + unsigned long long hdmi_freq; - vclk_freq = mode->clock; + vclk_freq = mode->clock * 1000ULL; /* For 420, pixel clock is half unlike venc clock */ if (encoder_hdmi->output_bus_fmt == MEDIA_BUS_FMT_UYYVYY8_0_5X24) @@ -107,7 +107,8 @@ static void meson_encoder_hdmi_set_vclk(struct meson_encoder_hdmi *encoder_hdmi, if (mode->flags & DRM_MODE_FLAG_DBLCLK) venc_freq /= 2; - dev_dbg(priv->dev, "vclk:%d phy=%d venc=%d hdmi=%d enci=%d\n", + dev_dbg(priv->dev, + "vclk:%lluHz phy=%lluHz venc=%lluHz hdmi=%lluHz enci=%d\n", phy_freq, vclk_freq, venc_freq, hdmi_freq, priv->venc.hdmi_use_enci); @@ -122,10 +123,11 @@ static enum drm_mode_status meson_encoder_hdmi_mode_valid(struct drm_bridge *bri struct meson_encoder_hdmi *encoder_hdmi = bridge_to_meson_encoder_hdmi(bridge); struct meson_drm *priv = encoder_hdmi->priv; bool is_hdmi2_sink = display_info->hdmi.scdc.supported; - unsigned int phy_freq; - unsigned int vclk_freq; - unsigned int venc_freq; - unsigned int hdmi_freq; + unsigned long long clock = mode->clock * 1000ULL; + unsigned long long phy_freq; + unsigned long long vclk_freq; + unsigned long long venc_freq; + unsigned long long hdmi_freq; int vic = drm_match_cea_mode(mode); enum drm_mode_status status; @@ -144,12 +146,12 @@ static enum drm_mode_status meson_encoder_hdmi_mode_valid(struct drm_bridge *bri if (status != MODE_OK) return status; - return meson_vclk_dmt_supported_freq(priv, mode->clock); + return meson_vclk_dmt_supported_freq(priv, clock); /* Check against supported VIC modes */ } else if (!meson_venc_hdmi_supported_vic(vic)) return MODE_BAD; - vclk_freq = mode->clock; + vclk_freq = clock; /* For 420, pixel clock is half unlike venc clock */ if (drm_mode_is_420_only(display_info, mode) || @@ -179,7 +181,8 @@ static enum drm_mode_status meson_encoder_hdmi_mode_valid(struct drm_bridge *bri if (mode->flags & DRM_MODE_FLAG_DBLCLK) venc_freq /= 2; - dev_dbg(priv->dev, "%s: vclk:%d phy=%d venc=%d hdmi=%d\n", + dev_dbg(priv->dev, + "%s: vclk:%lluHz phy=%lluHz venc=%lluHz hdmi=%lluHz\n", __func__, phy_freq, vclk_freq, venc_freq, hdmi_freq); return meson_vclk_vic_supported_freq(priv, phy_freq, vclk_freq); diff --git a/drivers/gpu/drm/meson/meson_vclk.c b/drivers/gpu/drm/meson/meson_vclk.c index 2a942dc6a6dc..3325580d885d 100644 --- a/drivers/gpu/drm/meson/meson_vclk.c +++ b/drivers/gpu/drm/meson/meson_vclk.c @@ -110,7 +110,10 @@ #define HDMI_PLL_LOCK BIT(31) #define HDMI_PLL_LOCK_G12A (3 << 30) -#define FREQ_1000_1001(_freq) DIV_ROUND_CLOSEST(_freq * 1000, 1001) +#define PIXEL_FREQ_1000_1001(_freq) \ + DIV_ROUND_CLOSEST_ULL((_freq) * 1000ULL, 1001ULL) +#define PHY_FREQ_1000_1001(_freq) \ + (PIXEL_FREQ_1000_1001(DIV_ROUND_DOWN_ULL(_freq, 10ULL)) * 10) /* VID PLL Dividers */ enum { @@ -360,11 +363,11 @@ enum { }; struct meson_vclk_params { - unsigned int pll_freq; - unsigned int phy_freq; - unsigned int vclk_freq; - unsigned int venc_freq; - unsigned int pixel_freq; + unsigned long long pll_freq; + unsigned long long phy_freq; + unsigned long long vclk_freq; + unsigned long long venc_freq; + unsigned long long pixel_freq; unsigned int pll_od1; unsigned int pll_od2; unsigned int pll_od3; @@ -372,11 +375,11 @@ struct meson_vclk_params { unsigned int vclk_div; } params[] = { [MESON_VCLK_HDMI_ENCI_54000] = { - .pll_freq = 4320000, - .phy_freq = 270000, - .vclk_freq = 54000, - .venc_freq = 54000, - .pixel_freq = 54000, + .pll_freq = 4320000000, + .phy_freq = 270000000, + .vclk_freq = 54000000, + .venc_freq = 54000000, + .pixel_freq = 54000000, .pll_od1 = 4, .pll_od2 = 4, .pll_od3 = 1, @@ -384,11 +387,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_DDR_54000] = { - .pll_freq = 4320000, - .phy_freq = 270000, - .vclk_freq = 54000, - .venc_freq = 54000, - .pixel_freq = 27000, + .pll_freq = 4320000000, + .phy_freq = 270000000, + .vclk_freq = 54000000, + .venc_freq = 54000000, + .pixel_freq = 27000000, .pll_od1 = 4, .pll_od2 = 4, .pll_od3 = 1, @@ -396,11 +399,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_DDR_148500] = { - .pll_freq = 2970000, - .phy_freq = 742500, - .vclk_freq = 148500, - .venc_freq = 148500, - .pixel_freq = 74250, + .pll_freq = 2970000000, + .phy_freq = 742500000, + .vclk_freq = 148500000, + .venc_freq = 148500000, + .pixel_freq = 74250000, .pll_od1 = 4, .pll_od2 = 1, .pll_od3 = 1, @@ -408,11 +411,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_74250] = { - .pll_freq = 2970000, - .phy_freq = 742500, - .vclk_freq = 74250, - .venc_freq = 74250, - .pixel_freq = 74250, + .pll_freq = 2970000000, + .phy_freq = 742500000, + .vclk_freq = 74250000, + .venc_freq = 74250000, + .pixel_freq = 74250000, .pll_od1 = 2, .pll_od2 = 2, .pll_od3 = 2, @@ -420,11 +423,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_148500] = { - .pll_freq = 2970000, - .phy_freq = 1485000, - .vclk_freq = 148500, - .venc_freq = 148500, - .pixel_freq = 148500, + .pll_freq = 2970000000, + .phy_freq = 1485000000, + .vclk_freq = 148500000, + .venc_freq = 148500000, + .pixel_freq = 148500000, .pll_od1 = 1, .pll_od2 = 2, .pll_od3 = 2, @@ -432,11 +435,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_297000] = { - .pll_freq = 5940000, - .phy_freq = 2970000, - .venc_freq = 297000, - .vclk_freq = 297000, - .pixel_freq = 297000, + .pll_freq = 5940000000, + .phy_freq = 2970000000, + .venc_freq = 297000000, + .vclk_freq = 297000000, + .pixel_freq = 297000000, .pll_od1 = 2, .pll_od2 = 1, .pll_od3 = 1, @@ -444,11 +447,11 @@ struct meson_vclk_params { .vclk_div = 2, }, [MESON_VCLK_HDMI_594000] = { - .pll_freq = 5940000, - .phy_freq = 5940000, - .venc_freq = 594000, - .vclk_freq = 594000, - .pixel_freq = 594000, + .pll_freq = 5940000000, + .phy_freq = 5940000000, + .venc_freq = 594000000, + .vclk_freq = 594000000, + .pixel_freq = 594000000, .pll_od1 = 1, .pll_od2 = 1, .pll_od3 = 2, @@ -456,11 +459,11 @@ struct meson_vclk_params { .vclk_div = 1, }, [MESON_VCLK_HDMI_594000_YUV420] = { - .pll_freq = 5940000, - .phy_freq = 2970000, - .venc_freq = 594000, - .vclk_freq = 594000, - .pixel_freq = 297000, + .pll_freq = 5940000000, + .phy_freq = 2970000000, + .venc_freq = 594000000, + .vclk_freq = 594000000, + .pixel_freq = 297000000, .pll_od1 = 2, .pll_od2 = 1, .pll_od3 = 1, @@ -617,16 +620,16 @@ static void meson_hdmi_pll_set_params(struct meson_drm *priv, unsigned int m, 3 << 20, pll_od_to_reg(od3) << 20); } -#define XTAL_FREQ 24000 +#define XTAL_FREQ (24 * 1000 * 1000) static unsigned int meson_hdmi_pll_get_m(struct meson_drm *priv, - unsigned int pll_freq) + unsigned long long pll_freq) { /* The GXBB PLL has a /2 pre-multiplier */ if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXBB)) - pll_freq /= 2; + pll_freq = DIV_ROUND_DOWN_ULL(pll_freq, 2); - return pll_freq / XTAL_FREQ; + return DIV_ROUND_DOWN_ULL(pll_freq, XTAL_FREQ); } #define HDMI_FRAC_MAX_GXBB 4096 @@ -635,12 +638,13 @@ static unsigned int meson_hdmi_pll_get_m(struct meson_drm *priv, static unsigned int meson_hdmi_pll_get_frac(struct meson_drm *priv, unsigned int m, - unsigned int pll_freq) + unsigned long long pll_freq) { - unsigned int parent_freq = XTAL_FREQ; + unsigned long long parent_freq = XTAL_FREQ; unsigned int frac_max = HDMI_FRAC_MAX_GXL; unsigned int frac_m; unsigned int frac; + u32 remainder; /* The GXBB PLL has a /2 pre-multiplier and a larger FRAC width */ if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXBB)) { @@ -652,11 +656,11 @@ static unsigned int meson_hdmi_pll_get_frac(struct meson_drm *priv, frac_max = HDMI_FRAC_MAX_G12A; /* We can have a perfect match !*/ - if (pll_freq / m == parent_freq && - pll_freq % m == 0) + if (div_u64_rem(pll_freq, m, &remainder) == parent_freq && + remainder == 0) return 0; - frac = div_u64((u64)pll_freq * (u64)frac_max, parent_freq); + frac = mul_u64_u64_div_u64(pll_freq, frac_max, parent_freq); frac_m = m * frac_max; if (frac_m > frac) return frac_max; @@ -666,7 +670,7 @@ static unsigned int meson_hdmi_pll_get_frac(struct meson_drm *priv, } static bool meson_hdmi_pll_validate_params(struct meson_drm *priv, - unsigned int m, + unsigned long long m, unsigned int frac) { if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXBB)) { @@ -694,7 +698,7 @@ static bool meson_hdmi_pll_validate_params(struct meson_drm *priv, } static bool meson_hdmi_pll_find_params(struct meson_drm *priv, - unsigned int freq, + unsigned long long freq, unsigned int *m, unsigned int *frac, unsigned int *od) @@ -706,7 +710,7 @@ static bool meson_hdmi_pll_find_params(struct meson_drm *priv, continue; *frac = meson_hdmi_pll_get_frac(priv, *m, freq * *od); - DRM_DEBUG_DRIVER("PLL params for %dkHz: m=%x frac=%x od=%d\n", + DRM_DEBUG_DRIVER("PLL params for %lluHz: m=%x frac=%x od=%d\n", freq, *m, *frac, *od); if (meson_hdmi_pll_validate_params(priv, *m, *frac)) @@ -718,7 +722,7 @@ static bool meson_hdmi_pll_find_params(struct meson_drm *priv, /* pll_freq is the frequency after the OD dividers */ enum drm_mode_status -meson_vclk_dmt_supported_freq(struct meson_drm *priv, unsigned int freq) +meson_vclk_dmt_supported_freq(struct meson_drm *priv, unsigned long long freq) { unsigned int od, m, frac; @@ -741,7 +745,7 @@ EXPORT_SYMBOL_GPL(meson_vclk_dmt_supported_freq); /* pll_freq is the frequency after the OD dividers */ static void meson_hdmi_pll_generic_set(struct meson_drm *priv, - unsigned int pll_freq) + unsigned long long pll_freq) { unsigned int od, m, frac, od1, od2, od3; @@ -756,7 +760,7 @@ static void meson_hdmi_pll_generic_set(struct meson_drm *priv, od1 = od / od2; } - DRM_DEBUG_DRIVER("PLL params for %dkHz: m=%x frac=%x od=%d/%d/%d\n", + DRM_DEBUG_DRIVER("PLL params for %lluHz: m=%x frac=%x od=%d/%d/%d\n", pll_freq, m, frac, od1, od2, od3); meson_hdmi_pll_set_params(priv, m, frac, od1, od2, od3); @@ -764,17 +768,18 @@ static void meson_hdmi_pll_generic_set(struct meson_drm *priv, return; } - DRM_ERROR("Fatal, unable to find parameters for PLL freq %d\n", + DRM_ERROR("Fatal, unable to find parameters for PLL freq %lluHz\n", pll_freq); } enum drm_mode_status -meson_vclk_vic_supported_freq(struct meson_drm *priv, unsigned int phy_freq, - unsigned int vclk_freq) +meson_vclk_vic_supported_freq(struct meson_drm *priv, + unsigned long long phy_freq, + unsigned long long vclk_freq) { int i; - DRM_DEBUG_DRIVER("phy_freq = %d vclk_freq = %d\n", + DRM_DEBUG_DRIVER("phy_freq = %lluHz vclk_freq = %lluHz\n", phy_freq, vclk_freq); /* Check against soc revision/package limits */ @@ -785,19 +790,19 @@ meson_vclk_vic_supported_freq(struct meson_drm *priv, unsigned int phy_freq, } for (i = 0 ; params[i].pixel_freq ; ++i) { - DRM_DEBUG_DRIVER("i = %d pixel_freq = %d alt = %d\n", + DRM_DEBUG_DRIVER("i = %d pixel_freq = %lluHz alt = %lluHz\n", i, params[i].pixel_freq, - FREQ_1000_1001(params[i].pixel_freq)); - DRM_DEBUG_DRIVER("i = %d phy_freq = %d alt = %d\n", + PIXEL_FREQ_1000_1001(params[i].pixel_freq)); + DRM_DEBUG_DRIVER("i = %d phy_freq = %lluHz alt = %lluHz\n", i, params[i].phy_freq, - FREQ_1000_1001(params[i].phy_freq/1000)*1000); + PHY_FREQ_1000_1001(params[i].phy_freq)); /* Match strict frequency */ if (phy_freq == params[i].phy_freq && vclk_freq == params[i].vclk_freq) return MODE_OK; /* Match 1000/1001 variant */ - if (phy_freq == (FREQ_1000_1001(params[i].phy_freq/1000)*1000) && - vclk_freq == FREQ_1000_1001(params[i].vclk_freq)) + if (phy_freq == PHY_FREQ_1000_1001(params[i].phy_freq) && + vclk_freq == PIXEL_FREQ_1000_1001(params[i].vclk_freq)) return MODE_OK; } @@ -805,8 +810,9 @@ meson_vclk_vic_supported_freq(struct meson_drm *priv, unsigned int phy_freq, } EXPORT_SYMBOL_GPL(meson_vclk_vic_supported_freq); -static void meson_vclk_set(struct meson_drm *priv, unsigned int pll_base_freq, - unsigned int od1, unsigned int od2, unsigned int od3, +static void meson_vclk_set(struct meson_drm *priv, + unsigned long long pll_base_freq, unsigned int od1, + unsigned int od2, unsigned int od3, unsigned int vid_pll_div, unsigned int vclk_div, unsigned int hdmi_tx_div, unsigned int venc_div, bool hdmi_use_enci, bool vic_alternate_clock) @@ -826,15 +832,15 @@ static void meson_vclk_set(struct meson_drm *priv, unsigned int pll_base_freq, meson_hdmi_pll_generic_set(priv, pll_base_freq); } else if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXBB)) { switch (pll_base_freq) { - case 2970000: + case 2970000000: m = 0x3d; frac = vic_alternate_clock ? 0xd02 : 0xe00; break; - case 4320000: + case 4320000000: m = vic_alternate_clock ? 0x59 : 0x5a; frac = vic_alternate_clock ? 0xe8f : 0; break; - case 5940000: + case 5940000000: m = 0x7b; frac = vic_alternate_clock ? 0xa05 : 0xc00; break; @@ -844,15 +850,15 @@ static void meson_vclk_set(struct meson_drm *priv, unsigned int pll_base_freq, } else if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXM) || meson_vpu_is_compatible(priv, VPU_COMPATIBLE_GXL)) { switch (pll_base_freq) { - case 2970000: + case 2970000000: m = 0x7b; frac = vic_alternate_clock ? 0x281 : 0x300; break; - case 4320000: + case 4320000000: m = vic_alternate_clock ? 0xb3 : 0xb4; frac = vic_alternate_clock ? 0x347 : 0; break; - case 5940000: + case 5940000000: m = 0xf7; frac = vic_alternate_clock ? 0x102 : 0x200; break; @@ -861,15 +867,15 @@ static void meson_vclk_set(struct meson_drm *priv, unsigned int pll_base_freq, meson_hdmi_pll_set_params(priv, m, frac, od1, od2, od3); } else if (meson_vpu_is_compatible(priv, VPU_COMPATIBLE_G12A)) { switch (pll_base_freq) { - case 2970000: + case 2970000000: m = 0x7b; frac = vic_alternate_clock ? 0x140b4 : 0x18000; break; - case 4320000: + case 4320000000: m = vic_alternate_clock ? 0xb3 : 0xb4; frac = vic_alternate_clock ? 0x1a3ee : 0; break; - case 5940000: + case 5940000000: m = 0xf7; frac = vic_alternate_clock ? 0x8148 : 0x10000; break; @@ -1025,14 +1031,14 @@ static void meson_vclk_set(struct meson_drm *priv, unsigned int pll_base_freq, } void meson_vclk_setup(struct meson_drm *priv, unsigned int target, - unsigned int phy_freq, unsigned int vclk_freq, - unsigned int venc_freq, unsigned int dac_freq, + unsigned long long phy_freq, unsigned long long vclk_freq, + unsigned long long venc_freq, unsigned long long dac_freq, bool hdmi_use_enci) { bool vic_alternate_clock = false; - unsigned int freq; - unsigned int hdmi_tx_div; - unsigned int venc_div; + unsigned long long freq; + unsigned long long hdmi_tx_div; + unsigned long long venc_div; if (target == MESON_VCLK_TARGET_CVBS) { meson_venci_cvbs_clock_config(priv); @@ -1052,27 +1058,27 @@ void meson_vclk_setup(struct meson_drm *priv, unsigned int target, return; } - hdmi_tx_div = vclk_freq / dac_freq; + hdmi_tx_div = DIV_ROUND_DOWN_ULL(vclk_freq, dac_freq); if (hdmi_tx_div == 0) { - pr_err("Fatal Error, invalid HDMI-TX freq %d\n", + pr_err("Fatal Error, invalid HDMI-TX freq %lluHz\n", dac_freq); return; } - venc_div = vclk_freq / venc_freq; + venc_div = DIV_ROUND_DOWN_ULL(vclk_freq, venc_freq); if (venc_div == 0) { - pr_err("Fatal Error, invalid HDMI venc freq %d\n", + pr_err("Fatal Error, invalid HDMI venc freq %lluHz\n", venc_freq); return; } for (freq = 0 ; params[freq].pixel_freq ; ++freq) { if ((phy_freq == params[freq].phy_freq || - phy_freq == FREQ_1000_1001(params[freq].phy_freq/1000)*1000) && + phy_freq == PHY_FREQ_1000_1001(params[freq].phy_freq)) && (vclk_freq == params[freq].vclk_freq || - vclk_freq == FREQ_1000_1001(params[freq].vclk_freq))) { + vclk_freq == PIXEL_FREQ_1000_1001(params[freq].vclk_freq))) { if (vclk_freq != params[freq].vclk_freq) vic_alternate_clock = true; else @@ -1098,7 +1104,8 @@ void meson_vclk_setup(struct meson_drm *priv, unsigned int target, } if (!params[freq].pixel_freq) { - pr_err("Fatal Error, invalid HDMI vclk freq %d\n", vclk_freq); + pr_err("Fatal Error, invalid HDMI vclk freq %lluHz\n", + vclk_freq); return; } diff --git a/drivers/gpu/drm/meson/meson_vclk.h b/drivers/gpu/drm/meson/meson_vclk.h index 60617aaf18dd..7ac55744e574 100644 --- a/drivers/gpu/drm/meson/meson_vclk.h +++ b/drivers/gpu/drm/meson/meson_vclk.h @@ -20,17 +20,18 @@ enum { }; /* 27MHz is the CVBS Pixel Clock */ -#define MESON_VCLK_CVBS 27000 +#define MESON_VCLK_CVBS (27 * 1000 * 1000) enum drm_mode_status -meson_vclk_dmt_supported_freq(struct meson_drm *priv, unsigned int freq); +meson_vclk_dmt_supported_freq(struct meson_drm *priv, unsigned long long freq); enum drm_mode_status -meson_vclk_vic_supported_freq(struct meson_drm *priv, unsigned int phy_freq, - unsigned int vclk_freq); +meson_vclk_vic_supported_freq(struct meson_drm *priv, + unsigned long long phy_freq, + unsigned long long vclk_freq); void meson_vclk_setup(struct meson_drm *priv, unsigned int target, - unsigned int phy_freq, unsigned int vclk_freq, - unsigned int venc_freq, unsigned int dac_freq, + unsigned long long phy_freq, unsigned long long vclk_freq, + unsigned long long venc_freq, unsigned long long dac_freq, bool hdmi_use_enci); #endif /* __MESON_VCLK_H */ diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c index 7b56da24711e..eca9c7d4ec6f 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.c @@ -2539,6 +2539,38 @@ static int dpu_encoder_virt_add_phys_encs( return 0; } +/** + * dpu_encoder_get_clones - Calculate the possible_clones for DPU encoder + * @drm_enc: DRM encoder pointer + * Returns: possible_clones mask + */ +uint32_t dpu_encoder_get_clones(struct drm_encoder *drm_enc) +{ + struct drm_encoder *curr; + int type = drm_enc->encoder_type; + uint32_t clone_mask = drm_encoder_mask(drm_enc); + + /* + * Set writeback as possible clones of real-time DSI encoders and vice + * versa + * + * Writeback encoders can't be clones of each other and DSI + * encoders can't be clones of each other. + * + * TODO: Add DP encoders as valid possible clones for writeback encoders + * (and vice versa) once concurrent writeback has been validated for DP + */ + drm_for_each_encoder(curr, drm_enc->dev) { + if ((type == DRM_MODE_ENCODER_VIRTUAL && + curr->encoder_type == DRM_MODE_ENCODER_DSI) || + (type == DRM_MODE_ENCODER_DSI && + curr->encoder_type == DRM_MODE_ENCODER_VIRTUAL)) + clone_mask |= drm_encoder_mask(curr); + } + + return clone_mask; +} + static int dpu_encoder_setup_display(struct dpu_encoder_virt *dpu_enc, struct dpu_kms *dpu_kms, struct msm_display_info *disp_info) diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.h b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.h index da133ee4701a..751be231ee7b 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.h +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_encoder.h @@ -60,6 +60,8 @@ enum dpu_intf_mode dpu_encoder_get_intf_mode(struct drm_encoder *encoder); void dpu_encoder_virt_runtime_resume(struct drm_encoder *encoder); +uint32_t dpu_encoder_get_clones(struct drm_encoder *drm_enc); + struct drm_encoder *dpu_encoder_init(struct drm_device *dev, int drm_enc_mode, struct msm_display_info *disp_info); diff --git a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c index 8741dc6fc8dd..b8f4ebba8ac2 100644 --- a/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c +++ b/drivers/gpu/drm/msm/disp/dpu1/dpu_kms.c @@ -2,7 +2,7 @@ /* * Copyright (C) 2013 Red Hat * Copyright (c) 2014-2018, The Linux Foundation. All rights reserved. - * Copyright (c) 2022 Qualcomm Innovation Center, Inc. All rights reserved. + * Copyright (c) 2022-2024 Qualcomm Innovation Center, Inc. All rights reserved. * * Author: Rob Clark <robdclark@gmail.com> */ @@ -834,8 +834,11 @@ static int _dpu_kms_drm_obj_init(struct dpu_kms *dpu_kms) return ret; num_encoders = 0; - drm_for_each_encoder(encoder, dev) + drm_for_each_encoder(encoder, dev) { num_encoders++; + if (catalog->cwb_count > 0) + encoder->possible_clones = dpu_encoder_get_clones(encoder); + } max_crtc_count = min(catalog->mixer_count, num_encoders); diff --git a/drivers/gpu/drm/nouveau/nouveau_fence.c b/drivers/gpu/drm/nouveau/nouveau_fence.c index 7cc84472cece..edddfc036c6d 100644 --- a/drivers/gpu/drm/nouveau/nouveau_fence.c +++ b/drivers/gpu/drm/nouveau/nouveau_fence.c @@ -90,7 +90,7 @@ nouveau_fence_context_kill(struct nouveau_fence_chan *fctx, int error) while (!list_empty(&fctx->pending)) { fence = list_entry(fctx->pending.next, typeof(*fence), head); - if (error) + if (error && !dma_fence_is_signaled_locked(&fence->base)) dma_fence_set_error(&fence->base, error); if (nouveau_fence_signal(fence)) diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c index 58502102926b..bb86b6d4ca49 100644 --- a/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/gsp/r535.c @@ -61,7 +61,7 @@ extern struct dentry *nouveau_debugfs_root; #define GSP_MSG_MIN_SIZE GSP_PAGE_SIZE -#define GSP_MSG_MAX_SIZE GSP_PAGE_MIN_SIZE * 16 +#define GSP_MSG_MAX_SIZE (GSP_MSG_MIN_SIZE * 16) struct r535_gsp_msg { u8 auth_tag_buffer[16]; diff --git a/drivers/gpu/drm/panel/panel-edp.c b/drivers/gpu/drm/panel/panel-edp.c index f8511fe5fb0d..b0315d3ba00a 100644 --- a/drivers/gpu/drm/panel/panel-edp.c +++ b/drivers/gpu/drm/panel/panel-edp.c @@ -1993,6 +1993,7 @@ static const struct edp_panel_entry edp_panels[] = { EDP_PANEL_ENTRY('S', 'H', 'P', 0x154c, &delay_200_500_p2e100, "LQ116M1JW10"), EDP_PANEL_ENTRY('S', 'H', 'P', 0x1593, &delay_200_500_p2e100, "LQ134N1"), + EDP_PANEL_ENTRY('S', 'T', 'A', 0x0004, &delay_200_500_e200, "116KHD024006"), EDP_PANEL_ENTRY('S', 'T', 'A', 0x0100, &delay_100_500_e200, "2081116HHD028001-51D"), { /* sentinal */ } diff --git a/drivers/gpu/drm/panel/panel-jadard-jd9365da-h3.c b/drivers/gpu/drm/panel/panel-jadard-jd9365da-h3.c index 7d68a8acfe2e..eb0f8373258c 100644 --- a/drivers/gpu/drm/panel/panel-jadard-jd9365da-h3.c +++ b/drivers/gpu/drm/panel/panel-jadard-jd9365da-h3.c @@ -129,11 +129,11 @@ static int jadard_unprepare(struct drm_panel *panel) { struct jadard *jadard = panel_to_jadard(panel); - gpiod_set_value(jadard->reset, 1); + gpiod_set_value(jadard->reset, 0); msleep(120); if (jadard->desc->reset_before_power_off_vcioo) { - gpiod_set_value(jadard->reset, 0); + gpiod_set_value(jadard->reset, 1); usleep_range(1000, 2000); } diff --git a/drivers/gpu/drm/panel/panel-simple.c b/drivers/gpu/drm/panel/panel-simple.c index 9b2f128fd309..cf9ab2d1f1d2 100644 --- a/drivers/gpu/drm/panel/panel-simple.c +++ b/drivers/gpu/drm/panel/panel-simple.c @@ -1027,27 +1027,28 @@ static const struct panel_desc auo_g070vvn01 = { }, }; -static const struct drm_display_mode auo_g101evn010_mode = { - .clock = 68930, - .hdisplay = 1280, - .hsync_start = 1280 + 82, - .hsync_end = 1280 + 82 + 2, - .htotal = 1280 + 82 + 2 + 84, - .vdisplay = 800, - .vsync_start = 800 + 8, - .vsync_end = 800 + 8 + 2, - .vtotal = 800 + 8 + 2 + 6, +static const struct display_timing auo_g101evn010_timing = { + .pixelclock = { 64000000, 68930000, 85000000 }, + .hactive = { 1280, 1280, 1280 }, + .hfront_porch = { 8, 64, 256 }, + .hback_porch = { 8, 64, 256 }, + .hsync_len = { 40, 168, 767 }, + .vactive = { 800, 800, 800 }, + .vfront_porch = { 4, 8, 100 }, + .vback_porch = { 4, 8, 100 }, + .vsync_len = { 8, 16, 223 }, }; static const struct panel_desc auo_g101evn010 = { - .modes = &auo_g101evn010_mode, - .num_modes = 1, + .timings = &auo_g101evn010_timing, + .num_timings = 1, .bpc = 6, .size = { .width = 216, .height = 135, }, .bus_format = MEDIA_BUS_FMT_RGB666_1X7X3_SPWG, + .bus_flags = DRM_BUS_FLAG_DE_HIGH, .connector_type = DRM_MODE_CONNECTOR_LVDS, }; diff --git a/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c b/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c index 17a98845fd31..bcbd49882392 100644 --- a/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c +++ b/drivers/gpu/drm/rockchip/rockchip_drm_vop2.c @@ -159,6 +159,7 @@ struct vop2_video_port { struct drm_crtc crtc; struct vop2 *vop2; struct clk *dclk; + struct clk *dclk_src; unsigned int id; const struct vop2_video_port_data *data; @@ -214,6 +215,7 @@ struct vop2 { struct clk *hclk; struct clk *aclk; struct clk *pclk; + struct clk *pll_hdmiphy0; /* optional internal rgb encoder */ struct rockchip_rgb *rgb; @@ -222,6 +224,8 @@ struct vop2 { struct vop2_win win[]; }; +#define VOP2_MAX_DCLK_RATE 600000000 + #define vop2_output_if_is_hdmi(x) ((x) == ROCKCHIP_VOP2_EP_HDMI0 || \ (x) == ROCKCHIP_VOP2_EP_HDMI1) @@ -1155,6 +1159,9 @@ static void vop2_crtc_atomic_disable(struct drm_crtc *crtc, vop2_crtc_disable_irq(vp, VP_INT_DSP_HOLD_VALID); + if (vp->dclk_src) + clk_set_parent(vp->dclk, vp->dclk_src); + clk_disable_unprepare(vp->dclk); vop2->enable_count--; @@ -1547,10 +1554,8 @@ static void vop2_plane_atomic_update(struct drm_plane *plane, rb_swap = vop2_win_rb_swap(fb->format->format); vop2_win_write(win, VOP2_WIN_RB_SWAP, rb_swap); - if (!vop2_cluster_window(win)) { - uv_swap = vop2_win_uv_swap(fb->format->format); - vop2_win_write(win, VOP2_WIN_UV_SWAP, uv_swap); - } + uv_swap = vop2_win_uv_swap(fb->format->format); + vop2_win_write(win, VOP2_WIN_UV_SWAP, uv_swap); if (fb->format->is_yuv) { vop2_win_write(win, VOP2_WIN_UV_VIR, DIV_ROUND_UP(fb->pitches[1], 4)); @@ -2259,6 +2264,27 @@ static void vop2_crtc_atomic_enable(struct drm_crtc *crtc, vop2_vp_write(vp, RK3568_VP_MIPI_CTRL, 0); + /* + * Switch to HDMI PHY PLL as DCLK source for display modes up + * to 4K@60Hz, if available, otherwise keep using the system CRU. + */ + if (vop2->pll_hdmiphy0 && clock <= VOP2_MAX_DCLK_RATE) { + drm_for_each_encoder_mask(encoder, crtc->dev, crtc_state->encoder_mask) { + struct rockchip_encoder *rkencoder = to_rockchip_encoder(encoder); + + if (rkencoder->crtc_endpoint_id == ROCKCHIP_VOP2_EP_HDMI0) { + if (!vp->dclk_src) + vp->dclk_src = clk_get_parent(vp->dclk); + + ret = clk_set_parent(vp->dclk, vop2->pll_hdmiphy0); + if (ret < 0) + drm_warn(vop2->drm, + "Could not switch to HDMI0 PHY PLL: %d\n", ret); + break; + } + } + } + clk_set_rate(vp->dclk, clock); vop2_post_config(crtc); @@ -3699,6 +3725,12 @@ static int vop2_bind(struct device *dev, struct device *master, void *data) return PTR_ERR(vop2->pclk); } + vop2->pll_hdmiphy0 = devm_clk_get_optional(vop2->dev, "pll_hdmiphy0"); + if (IS_ERR(vop2->pll_hdmiphy0)) { + drm_err(vop2->drm, "failed to get pll_hdmiphy0\n"); + return PTR_ERR(vop2->pll_hdmiphy0); + } + vop2->irq = platform_get_irq(pdev, 0); if (vop2->irq < 0) { drm_err(vop2->drm, "cannot find irq for vop2\n"); diff --git a/drivers/gpu/drm/tests/drm_gem_shmem_test.c b/drivers/gpu/drm/tests/drm_gem_shmem_test.c index fd4215e2f982..925fbc2cda70 100644 --- a/drivers/gpu/drm/tests/drm_gem_shmem_test.c +++ b/drivers/gpu/drm/tests/drm_gem_shmem_test.c @@ -216,6 +216,9 @@ static void drm_gem_shmem_test_get_pages_sgt(struct kunit *test) KUNIT_ASSERT_NOT_ERR_OR_NULL(test, sgt); KUNIT_EXPECT_NULL(test, shmem->sgt); + ret = kunit_add_action_or_reset(test, kfree_wrapper, sgt); + KUNIT_ASSERT_EQ(test, ret, 0); + ret = kunit_add_action_or_reset(test, sg_free_table_wrapper, sgt); KUNIT_ASSERT_EQ(test, ret, 0); diff --git a/drivers/gpu/drm/tiny/panel-mipi-dbi.c b/drivers/gpu/drm/tiny/panel-mipi-dbi.c index 0460ecaef4bd..23914a9f7fd3 100644 --- a/drivers/gpu/drm/tiny/panel-mipi-dbi.c +++ b/drivers/gpu/drm/tiny/panel-mipi-dbi.c @@ -390,7 +390,10 @@ static int panel_mipi_dbi_spi_probe(struct spi_device *spi) spi_set_drvdata(spi, drm); - drm_client_setup(drm, NULL); + if (bpp == 16) + drm_client_setup_with_fourcc(drm, DRM_FORMAT_RGB565); + else + drm_client_setup_with_fourcc(drm, DRM_FORMAT_RGB888); return 0; } diff --git a/drivers/gpu/drm/ttm/ttm_bo.c b/drivers/gpu/drm/ttm/ttm_bo.c index ea5e49858857..72c675191a02 100644 --- a/drivers/gpu/drm/ttm/ttm_bo.c +++ b/drivers/gpu/drm/ttm/ttm_bo.c @@ -1092,7 +1092,8 @@ struct ttm_bo_swapout_walk { struct ttm_lru_walk walk; /** @gfp_flags: The gfp flags to use for ttm_tt_swapout() */ gfp_t gfp_flags; - + /** @hit_low: Whether we should attempt to swap BO's with low watermark threshold */ + /** @evict_low: If we cannot swap a bo when @try_low is false (first pass) */ bool hit_low, evict_low; }; diff --git a/drivers/gpu/drm/v3d/v3d_drv.c b/drivers/gpu/drm/v3d/v3d_drv.c index 930737a9347b..852015214e97 100644 --- a/drivers/gpu/drm/v3d/v3d_drv.c +++ b/drivers/gpu/drm/v3d/v3d_drv.c @@ -295,11 +295,21 @@ static int v3d_platform_drm_probe(struct platform_device *pdev) if (ret) return ret; + v3d->clk = devm_clk_get_optional(dev, NULL); + if (IS_ERR(v3d->clk)) + return dev_err_probe(dev, PTR_ERR(v3d->clk), "Failed to get V3D clock\n"); + + ret = clk_prepare_enable(v3d->clk); + if (ret) { + dev_err(&pdev->dev, "Couldn't enable the V3D clock\n"); + return ret; + } + mmu_debug = V3D_READ(V3D_MMU_DEBUG_INFO); mask = DMA_BIT_MASK(30 + V3D_GET_FIELD(mmu_debug, V3D_MMU_PA_WIDTH)); ret = dma_set_mask_and_coherent(dev, mask); if (ret) - return ret; + goto clk_disable; v3d->va_width = 30 + V3D_GET_FIELD(mmu_debug, V3D_MMU_VA_WIDTH); @@ -319,28 +329,29 @@ static int v3d_platform_drm_probe(struct platform_device *pdev) ret = PTR_ERR(v3d->reset); if (ret == -EPROBE_DEFER) - return ret; + goto clk_disable; v3d->reset = NULL; ret = map_regs(v3d, &v3d->bridge_regs, "bridge"); if (ret) { dev_err(dev, "Failed to get reset control or bridge regs\n"); - return ret; + goto clk_disable; } } if (v3d->ver < 41) { ret = map_regs(v3d, &v3d->gca_regs, "gca"); if (ret) - return ret; + goto clk_disable; } v3d->mmu_scratch = dma_alloc_wc(dev, 4096, &v3d->mmu_scratch_paddr, GFP_KERNEL | __GFP_NOWARN | __GFP_ZERO); if (!v3d->mmu_scratch) { dev_err(dev, "Failed to allocate MMU scratch page\n"); - return -ENOMEM; + ret = -ENOMEM; + goto clk_disable; } ret = v3d_gem_init(drm); @@ -369,6 +380,8 @@ gem_destroy: v3d_gem_destroy(drm); dma_free: dma_free_wc(dev, 4096, v3d->mmu_scratch, v3d->mmu_scratch_paddr); +clk_disable: + clk_disable_unprepare(v3d->clk); return ret; } @@ -386,6 +399,8 @@ static void v3d_platform_drm_remove(struct platform_device *pdev) dma_free_wc(v3d->drm.dev, 4096, v3d->mmu_scratch, v3d->mmu_scratch_paddr); + + clk_disable_unprepare(v3d->clk); } static struct platform_driver v3d_platform_driver = { diff --git a/drivers/gpu/drm/v3d/v3d_sched.c b/drivers/gpu/drm/v3d/v3d_sched.c index 6db503a56918..78ebc7d54a0c 100644 --- a/drivers/gpu/drm/v3d/v3d_sched.c +++ b/drivers/gpu/drm/v3d/v3d_sched.c @@ -746,11 +746,16 @@ v3d_gpu_reset_for_timeout(struct v3d_dev *v3d, struct drm_sched_job *sched_job) return DRM_GPU_SCHED_STAT_NOMINAL; } -/* If the current address or return address have changed, then the GPU - * has probably made progress and we should delay the reset. This - * could fail if the GPU got in an infinite loop in the CL, but that - * is pretty unlikely outside of an i-g-t testcase. - */ +static void +v3d_sched_skip_reset(struct drm_sched_job *sched_job) +{ + struct drm_gpu_scheduler *sched = sched_job->sched; + + spin_lock(&sched->job_list_lock); + list_add(&sched_job->list, &sched->pending_list); + spin_unlock(&sched->job_list_lock); +} + static enum drm_gpu_sched_stat v3d_cl_job_timedout(struct drm_sched_job *sched_job, enum v3d_queue q, u32 *timedout_ctca, u32 *timedout_ctra) @@ -760,9 +765,16 @@ v3d_cl_job_timedout(struct drm_sched_job *sched_job, enum v3d_queue q, u32 ctca = V3D_CORE_READ(0, V3D_CLE_CTNCA(q)); u32 ctra = V3D_CORE_READ(0, V3D_CLE_CTNRA(q)); + /* If the current address or return address have changed, then the GPU + * has probably made progress and we should delay the reset. This + * could fail if the GPU got in an infinite loop in the CL, but that + * is pretty unlikely outside of an i-g-t testcase. + */ if (*timedout_ctca != ctca || *timedout_ctra != ctra) { *timedout_ctca = ctca; *timedout_ctra = ctra; + + v3d_sched_skip_reset(sched_job); return DRM_GPU_SCHED_STAT_NOMINAL; } @@ -802,11 +814,13 @@ v3d_csd_job_timedout(struct drm_sched_job *sched_job) struct v3d_dev *v3d = job->base.v3d; u32 batches = V3D_CORE_READ(0, V3D_CSD_CURRENT_CFG4(v3d->ver)); - /* If we've made progress, skip reset and let the timer get - * rearmed. + /* If we've made progress, skip reset, add the job to the pending + * list, and let the timer get rearmed. */ if (job->timedout_batches != batches) { job->timedout_batches = batches; + + v3d_sched_skip_reset(sched_job); return DRM_GPU_SCHED_STAT_NOMINAL; } diff --git a/drivers/gpu/drm/virtio/virtgpu_drv.c b/drivers/gpu/drm/virtio/virtgpu_drv.c index 6a67c6297d58..8719b778a1ff 100644 --- a/drivers/gpu/drm/virtio/virtgpu_drv.c +++ b/drivers/gpu/drm/virtio/virtgpu_drv.c @@ -125,6 +125,14 @@ static void virtio_gpu_remove(struct virtio_device *vdev) drm_dev_put(dev); } +static void virtio_gpu_shutdown(struct virtio_device *vdev) +{ + /* + * drm does its own synchronization on shutdown. + * Do nothing here, opt out of device reset. + */ +} + static void virtio_gpu_config_changed(struct virtio_device *vdev) { struct drm_device *dev = vdev->priv; @@ -159,6 +167,7 @@ static struct virtio_driver virtio_gpu_driver = { .id_table = id_table, .probe = virtio_gpu_probe, .remove = virtio_gpu_remove, + .shutdown = virtio_gpu_shutdown, .config_changed = virtio_gpu_config_changed }; diff --git a/drivers/gpu/drm/xe/display/xe_display.c b/drivers/gpu/drm/xe/display/xe_display.c index b3921dbc52ff..b735e30953ce 100644 --- a/drivers/gpu/drm/xe/display/xe_display.c +++ b/drivers/gpu/drm/xe/display/xe_display.c @@ -346,7 +346,8 @@ static void __xe_display_pm_suspend(struct xe_device *xe, bool runtime) xe_display_flush_cleanup_work(xe); - intel_hpd_cancel_work(xe); + if (!runtime) + intel_hpd_cancel_work(xe); if (!runtime && has_display(xe)) { intel_display_driver_suspend_access(display); diff --git a/drivers/gpu/drm/xe/instructions/xe_gpu_commands.h b/drivers/gpu/drm/xe/instructions/xe_gpu_commands.h index a255946b6f77..8cfcd3360896 100644 --- a/drivers/gpu/drm/xe/instructions/xe_gpu_commands.h +++ b/drivers/gpu/drm/xe/instructions/xe_gpu_commands.h @@ -41,6 +41,7 @@ #define GFX_OP_PIPE_CONTROL(len) ((0x3<<29)|(0x3<<27)|(0x2<<24)|((len)-2)) +#define PIPE_CONTROL0_L3_READ_ONLY_CACHE_INVALIDATE BIT(10) /* gen12 */ #define PIPE_CONTROL0_HDC_PIPELINE_FLUSH BIT(9) /* gen12 */ #define PIPE_CONTROL_COMMAND_CACHE_INVALIDATE (1<<29) diff --git a/drivers/gpu/drm/xe/instructions/xe_mi_commands.h b/drivers/gpu/drm/xe/instructions/xe_mi_commands.h index 10ec2920d31b..d4033278be9f 100644 --- a/drivers/gpu/drm/xe/instructions/xe_mi_commands.h +++ b/drivers/gpu/drm/xe/instructions/xe_mi_commands.h @@ -47,6 +47,10 @@ #define MI_LRI_FORCE_POSTED REG_BIT(12) #define MI_LRI_LEN(x) (((x) & 0xff) + 1) +#define MI_STORE_REGISTER_MEM (__MI_INSTR(0x24) | XE_INSTR_NUM_DW(4)) +#define MI_SRM_USE_GGTT REG_BIT(22) +#define MI_SRM_ADD_CS_OFFSET REG_BIT(19) + #define MI_FLUSH_DW __MI_INSTR(0x26) #define MI_FLUSH_DW_STORE_INDEX REG_BIT(21) #define MI_INVALIDATE_TLB REG_BIT(18) diff --git a/drivers/gpu/drm/xe/regs/xe_gt_regs.h b/drivers/gpu/drm/xe/regs/xe_gt_regs.h index 162f18e975da..ab95d3545a72 100644 --- a/drivers/gpu/drm/xe/regs/xe_gt_regs.h +++ b/drivers/gpu/drm/xe/regs/xe_gt_regs.h @@ -157,6 +157,7 @@ #define XEHPG_SC_INSTDONE_EXTRA2 XE_REG_MCR(0x7108) #define COMMON_SLICE_CHICKEN4 XE_REG(0x7300, XE_REG_OPTION_MASKED) +#define SBE_PUSH_CONSTANT_BEHIND_FIX_ENABLE REG_BIT(12) #define DISABLE_TDC_LOAD_BALANCING_CALC REG_BIT(6) #define COMMON_SLICE_CHICKEN3 XE_REG(0x7304, XE_REG_OPTION_MASKED) @@ -475,6 +476,7 @@ #define TDL_TSL_CHICKEN XE_REG_MCR(0xe4c4, XE_REG_OPTION_MASKED) #define STK_ID_RESTRICT REG_BIT(12) #define SLM_WMTP_RESTORE REG_BIT(11) +#define RES_CHK_SPR_DIS REG_BIT(6) #define ROW_CHICKEN XE_REG_MCR(0xe4f0, XE_REG_OPTION_MASKED) #define UGM_BACKUP_MODE REG_BIT(13) @@ -500,6 +502,9 @@ #define LSC_L1_FLUSH_CTL_3D_DATAPORT_FLUSH_EVENTS_MASK REG_GENMASK(13, 11) #define DIS_ATOMIC_CHAINING_TYPED_WRITES REG_BIT(3) +#define TDL_CHICKEN XE_REG_MCR(0xe5f4, XE_REG_OPTION_MASKED) +#define QID_WAIT_FOR_THREAD_NOT_RUN_DISABLE REG_BIT(12) + #define LSC_CHICKEN_BIT_0 XE_REG_MCR(0xe7c8) #define DISABLE_D8_D16_COASLESCE REG_BIT(30) #define WR_REQ_CHAINING_DIS REG_BIT(26) diff --git a/drivers/gpu/drm/xe/tests/xe_mocs.c b/drivers/gpu/drm/xe/tests/xe_mocs.c index ef1e5256c56a..0e502feaca81 100644 --- a/drivers/gpu/drm/xe/tests/xe_mocs.c +++ b/drivers/gpu/drm/xe/tests/xe_mocs.c @@ -46,8 +46,11 @@ static void read_l3cc_table(struct xe_gt *gt, unsigned int fw_ref, i; u32 reg_val; - fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FW_GT); - KUNIT_ASSERT_NE_MSG(test, fw_ref, 0, "Forcewake Failed.\n"); + fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL); + if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL)) { + xe_force_wake_put(gt_to_fw(gt), fw_ref); + KUNIT_ASSERT_TRUE_MSG(test, true, "Forcewake Failed.\n"); + } for (i = 0; i < info->num_mocs_regs; i++) { if (!(i & 1)) { diff --git a/drivers/gpu/drm/xe/tests/xe_rtp_test.c b/drivers/gpu/drm/xe/tests/xe_rtp_test.c index 36a3b5420fef..b0254b014fe4 100644 --- a/drivers/gpu/drm/xe/tests/xe_rtp_test.c +++ b/drivers/gpu/drm/xe/tests/xe_rtp_test.c @@ -320,7 +320,7 @@ static void xe_rtp_process_to_sr_tests(struct kunit *test) count_rtp_entries++; xe_rtp_process_ctx_enable_active_tracking(&ctx, &active, count_rtp_entries); - xe_rtp_process_to_sr(&ctx, param->entries, reg_sr); + xe_rtp_process_to_sr(&ctx, param->entries, count_rtp_entries, reg_sr); xa_for_each(®_sr->xa, idx, sre) { if (idx == param->expected_reg.addr) diff --git a/drivers/gpu/drm/xe/xe_bo.c b/drivers/gpu/drm/xe/xe_bo.c index 3f5391d416d4..2070aa12059c 100644 --- a/drivers/gpu/drm/xe/xe_bo.c +++ b/drivers/gpu/drm/xe/xe_bo.c @@ -713,6 +713,21 @@ 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. @@ -2142,6 +2157,7 @@ int xe_gem_create_ioctl(struct drm_device *dev, void *data, struct xe_file *xef = to_xe_file(file); struct drm_xe_gem_create *args = data; struct xe_vm *vm = NULL; + ktime_t end = 0; struct xe_bo *bo; unsigned int bo_flags; u32 handle; @@ -2214,6 +2230,10 @@ int xe_gem_create_ioctl(struct drm_device *dev, void *data, vm = xe_vm_lookup(xef, args->vm_id); if (XE_IOCTL_DBG(xe, !vm)) return -ENOENT; + } + +retry: + if (vm) { err = xe_vm_lock(vm, true); if (err) goto out_vm; @@ -2227,6 +2247,8 @@ int xe_gem_create_ioctl(struct drm_device *dev, void *data, if (IS_ERR(bo)) { err = PTR_ERR(bo); + if (xe_vm_validate_should_retry(NULL, err, &end)) + goto retry; goto out_vm; } diff --git a/drivers/gpu/drm/xe/xe_debugfs.c b/drivers/gpu/drm/xe/xe_debugfs.c index 492b4877433f..92e6fa8fe3a1 100644 --- a/drivers/gpu/drm/xe/xe_debugfs.c +++ b/drivers/gpu/drm/xe/xe_debugfs.c @@ -166,7 +166,7 @@ static ssize_t wedged_mode_set(struct file *f, const char __user *ubuf, return -EINVAL; if (xe->wedged.mode == wedged_mode) - return 0; + return size; xe->wedged.mode = wedged_mode; @@ -175,6 +175,7 @@ static ssize_t wedged_mode_set(struct file *f, const char __user *ubuf, ret = xe_guc_ads_scheduler_policy_toggle_reset(>->uc.guc.ads); if (ret) { xe_gt_err(gt, "Failed to update GuC ADS scheduler policy. GuC may still cause engine reset even with wedged_mode=2\n"); + xe_pm_runtime_put(xe); return -EIO; } } diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c index 4e1839b483a0..74516e73ba4e 100644 --- a/drivers/gpu/drm/xe/xe_device.c +++ b/drivers/gpu/drm/xe/xe_device.c @@ -722,7 +722,9 @@ int xe_device_probe(struct xe_device *xe) } /* Allocate and map stolen after potential VRAM resize */ - xe_ttm_stolen_mgr_init(xe); + err = xe_ttm_stolen_mgr_init(xe); + if (err) + return err; /* * Now that GT is initialized (TTM in particular), @@ -734,6 +736,12 @@ int xe_device_probe(struct xe_device *xe) if (err) goto err; + for_each_tile(tile, xe, id) { + err = xe_tile_init(tile); + if (err) + goto err; + } + for_each_gt(gt, xe, id) { last_gt = id; diff --git a/drivers/gpu/drm/xe/xe_drm_client.c b/drivers/gpu/drm/xe/xe_drm_client.c index 2d4874d2b922..31f688e953d7 100644 --- a/drivers/gpu/drm/xe/xe_drm_client.c +++ b/drivers/gpu/drm/xe/xe_drm_client.c @@ -325,6 +325,14 @@ static void show_run_ticks(struct drm_printer *p, struct drm_file *file) unsigned int fw_ref; /* + * RING_TIMESTAMP registers are inaccessible in VF mode. + * Without drm-total-cycles-*, other keys provide little value. + * Show all or none of the optional "run_ticks" keys in this case. + */ + if (IS_SRIOV_VF(xe)) + return; + + /* * Wait for any exec queue going away: their cycles will get updated on * context switch out, so wait for that to happen */ diff --git a/drivers/gpu/drm/xe/xe_gen_wa_oob.c b/drivers/gpu/drm/xe/xe_gen_wa_oob.c index 904cf47925aa..ed9183599e31 100644 --- a/drivers/gpu/drm/xe/xe_gen_wa_oob.c +++ b/drivers/gpu/drm/xe/xe_gen_wa_oob.c @@ -28,10 +28,10 @@ "\n" \ "#endif\n" -static void print_usage(FILE *f) +static void print_usage(FILE *f, const char *progname) { fprintf(f, "usage: %s <input-rule-file> <generated-c-source-file> <generated-c-header-file>\n", - program_invocation_short_name); + progname); } static void print_parse_error(const char *err_msg, const char *line, @@ -144,7 +144,7 @@ int main(int argc, const char *argv[]) if (argc < 3) { fprintf(stderr, "ERROR: wrong arguments\n"); - print_usage(stderr); + print_usage(stderr, argv[0]); return 1; } diff --git a/drivers/gpu/drm/xe/xe_gsc.c b/drivers/gpu/drm/xe/xe_gsc.c index 1eb791ddc375..0f11b1c00960 100644 --- a/drivers/gpu/drm/xe/xe_gsc.c +++ b/drivers/gpu/drm/xe/xe_gsc.c @@ -564,6 +564,28 @@ void xe_gsc_remove(struct xe_gsc *gsc) xe_gsc_proxy_remove(gsc); } +void xe_gsc_stop_prepare(struct xe_gsc *gsc) +{ + struct xe_gt *gt = gsc_to_gt(gsc); + int ret; + + if (!xe_uc_fw_is_loadable(&gsc->fw) || xe_uc_fw_is_in_error_state(&gsc->fw)) + return; + + xe_force_wake_assert_held(gt_to_fw(gt), XE_FW_GSC); + + /* + * If the GSC FW load or the proxy init are interrupted, the only way + * to recover it is to do an FLR and reload the GSC from scratch. + * Therefore, let's wait for the init to complete before stopping + * operations. The proxy init is the last step, so we can just wait on + * that + */ + ret = xe_gsc_wait_for_proxy_init_done(gsc); + if (ret) + xe_gt_err(gt, "failed to wait for GSC init completion before uc stop\n"); +} + /* * wa_14015076503: if the GSC FW is loaded, we need to alert it before doing a * GSC engine reset by writing a notification bit in the GS1 register and then diff --git a/drivers/gpu/drm/xe/xe_gsc.h b/drivers/gpu/drm/xe/xe_gsc.h index e282b9ef6ec4..c31fe24c4b66 100644 --- a/drivers/gpu/drm/xe/xe_gsc.h +++ b/drivers/gpu/drm/xe/xe_gsc.h @@ -16,6 +16,7 @@ struct xe_hw_engine; int xe_gsc_init(struct xe_gsc *gsc); int xe_gsc_init_post_hwconfig(struct xe_gsc *gsc); void xe_gsc_wait_for_worker_completion(struct xe_gsc *gsc); +void xe_gsc_stop_prepare(struct xe_gsc *gsc); void xe_gsc_load_start(struct xe_gsc *gsc); void xe_gsc_remove(struct xe_gsc *gsc); void xe_gsc_hwe_irq_handler(struct xe_hw_engine *hwe, u16 intr_vec); diff --git a/drivers/gpu/drm/xe/xe_gsc_proxy.c b/drivers/gpu/drm/xe/xe_gsc_proxy.c index 24cc6a4f9a96..76636da3d06c 100644 --- a/drivers/gpu/drm/xe/xe_gsc_proxy.c +++ b/drivers/gpu/drm/xe/xe_gsc_proxy.c @@ -71,6 +71,17 @@ bool xe_gsc_proxy_init_done(struct xe_gsc *gsc) HECI1_FWSTS1_PROXY_STATE_NORMAL; } +int xe_gsc_wait_for_proxy_init_done(struct xe_gsc *gsc) +{ + struct xe_gt *gt = gsc_to_gt(gsc); + + /* Proxy init can take up to 500ms, so wait double that for safety */ + return xe_mmio_wait32(>->mmio, HECI_FWSTS1(MTL_GSC_HECI1_BASE), + HECI1_FWSTS1_CURRENT_STATE, + HECI1_FWSTS1_PROXY_STATE_NORMAL, + USEC_PER_SEC, NULL, false); +} + static void __gsc_proxy_irq_rmw(struct xe_gsc *gsc, u32 clr, u32 set) { struct xe_gt *gt = gsc_to_gt(gsc); diff --git a/drivers/gpu/drm/xe/xe_gsc_proxy.h b/drivers/gpu/drm/xe/xe_gsc_proxy.h index c511ade6b863..e2498aa6de18 100644 --- a/drivers/gpu/drm/xe/xe_gsc_proxy.h +++ b/drivers/gpu/drm/xe/xe_gsc_proxy.h @@ -13,6 +13,7 @@ struct xe_gsc; int xe_gsc_proxy_init(struct xe_gsc *gsc); bool xe_gsc_proxy_init_done(struct xe_gsc *gsc); void xe_gsc_proxy_remove(struct xe_gsc *gsc); +int xe_gsc_wait_for_proxy_init_done(struct xe_gsc *gsc); int xe_gsc_proxy_start(struct xe_gsc *gsc); int xe_gsc_proxy_request_handler(struct xe_gsc *gsc); diff --git a/drivers/gpu/drm/xe/xe_gt.c b/drivers/gpu/drm/xe/xe_gt.c index 8a20e6744836..6b4b9eca2c38 100644 --- a/drivers/gpu/drm/xe/xe_gt.c +++ b/drivers/gpu/drm/xe/xe_gt.c @@ -381,6 +381,10 @@ int xe_gt_init_early(struct xe_gt *gt) if (err) return err; + err = xe_tuning_init(gt); + if (err) + return err; + xe_wa_process_oob(gt); xe_force_wake_init_gt(gt, gt_to_fw(gt)); @@ -646,6 +650,9 @@ void xe_gt_mmio_init(struct xe_gt *gt) if (gt->info.type == XE_GT_TYPE_MEDIA) { gt->mmio.adj_offset = MEDIA_GT_GSI_OFFSET; gt->mmio.adj_limit = MEDIA_GT_GSI_LENGTH; + } else { + gt->mmio.adj_offset = 0; + gt->mmio.adj_limit = 0; } if (IS_SRIOV_VF(gt_to_xe(gt))) @@ -858,7 +865,7 @@ void xe_gt_suspend_prepare(struct xe_gt *gt) fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL); - xe_uc_stop_prepare(>->uc); + xe_uc_suspend_prepare(>->uc); xe_force_wake_put(gt_to_fw(gt), fw_ref); } diff --git a/drivers/gpu/drm/xe/xe_gt_debugfs.c b/drivers/gpu/drm/xe/xe_gt_debugfs.c index e7792858b1e4..f7005a3643e6 100644 --- a/drivers/gpu/drm/xe/xe_gt_debugfs.c +++ b/drivers/gpu/drm/xe/xe_gt_debugfs.c @@ -30,6 +30,7 @@ #include "xe_reg_sr.h" #include "xe_reg_whitelist.h" #include "xe_sriov.h" +#include "xe_tuning.h" #include "xe_uc_debugfs.h" #include "xe_wa.h" @@ -91,22 +92,23 @@ static int hw_engines(struct xe_gt *gt, struct drm_printer *p) struct xe_hw_engine *hwe; enum xe_hw_engine_id id; unsigned int fw_ref; + int ret = 0; xe_pm_runtime_get(xe); fw_ref = xe_force_wake_get(gt_to_fw(gt), XE_FORCEWAKE_ALL); if (!xe_force_wake_ref_has_domain(fw_ref, XE_FORCEWAKE_ALL)) { - xe_pm_runtime_put(xe); - xe_force_wake_put(gt_to_fw(gt), fw_ref); - return -ETIMEDOUT; + ret = -ETIMEDOUT; + goto fw_put; } for_each_hw_engine(hwe, gt, id) xe_hw_engine_print(hwe, p); +fw_put: xe_force_wake_put(gt_to_fw(gt), fw_ref); xe_pm_runtime_put(xe); - return 0; + return ret; } static int powergate_info(struct xe_gt *gt, struct drm_printer *p) @@ -217,6 +219,15 @@ static int workarounds(struct xe_gt *gt, struct drm_printer *p) return 0; } +static int tunings(struct xe_gt *gt, struct drm_printer *p) +{ + xe_pm_runtime_get(gt_to_xe(gt)); + xe_tuning_dump(gt, p); + xe_pm_runtime_put(gt_to_xe(gt)); + + return 0; +} + static int pat(struct xe_gt *gt, struct drm_printer *p) { xe_pm_runtime_get(gt_to_xe(gt)); @@ -300,6 +311,7 @@ static const struct drm_info_list debugfs_list[] = { {"powergate_info", .show = xe_gt_debugfs_simple_show, .data = powergate_info}, {"register-save-restore", .show = xe_gt_debugfs_simple_show, .data = register_save_restore}, {"workarounds", .show = xe_gt_debugfs_simple_show, .data = workarounds}, + {"tunings", .show = xe_gt_debugfs_simple_show, .data = tunings}, {"pat", .show = xe_gt_debugfs_simple_show, .data = pat}, {"mocs", .show = xe_gt_debugfs_simple_show, .data = mocs}, {"default_lrc_rcs", .show = xe_gt_debugfs_simple_show, .data = rcs_default_lrc}, diff --git a/drivers/gpu/drm/xe/xe_gt_idle.c b/drivers/gpu/drm/xe/xe_gt_idle.c index ffd3ba7f6656..fbbace7b0b12 100644 --- a/drivers/gpu/drm/xe/xe_gt_idle.c +++ b/drivers/gpu/drm/xe/xe_gt_idle.c @@ -69,6 +69,8 @@ static u64 get_residency_ms(struct xe_gt_idle *gtidle, u64 cur_residency) { u64 delta, overflow_residency, prev_residency; + lockdep_assert_held(>idle->lock); + overflow_residency = BIT_ULL(32); /* @@ -275,8 +277,21 @@ static ssize_t idle_status_show(struct device *dev, return sysfs_emit(buff, "%s\n", gt_idle_state_to_string(state)); } -static DEVICE_ATTR_RO(idle_status); +u64 xe_gt_idle_residency_msec(struct xe_gt_idle *gtidle) +{ + struct xe_guc_pc *pc = gtidle_to_pc(gtidle); + u64 residency; + unsigned long flags; + + raw_spin_lock_irqsave(>idle->lock, flags); + residency = get_residency_ms(gtidle, gtidle->idle_residency(pc)); + raw_spin_unlock_irqrestore(>idle->lock, flags); + + return residency; +} + +static DEVICE_ATTR_RO(idle_status); static ssize_t idle_residency_ms_show(struct device *dev, struct device_attribute *attr, char *buff) { @@ -285,10 +300,10 @@ static ssize_t idle_residency_ms_show(struct device *dev, u64 residency; xe_pm_runtime_get(pc_to_xe(pc)); - residency = gtidle->idle_residency(pc); + residency = xe_gt_idle_residency_msec(gtidle); xe_pm_runtime_put(pc_to_xe(pc)); - return sysfs_emit(buff, "%llu\n", get_residency_ms(gtidle, residency)); + return sysfs_emit(buff, "%llu\n", residency); } static DEVICE_ATTR_RO(idle_residency_ms); @@ -331,6 +346,8 @@ int xe_gt_idle_init(struct xe_gt_idle *gtidle) if (!kobj) return -ENOMEM; + raw_spin_lock_init(>idle->lock); + if (xe_gt_is_media_type(gt)) { snprintf(gtidle->name, sizeof(gtidle->name), "gt%d-mc", gt->info.id); gtidle->idle_residency = xe_guc_pc_mc6_residency; diff --git a/drivers/gpu/drm/xe/xe_gt_idle.h b/drivers/gpu/drm/xe/xe_gt_idle.h index 4455a6501cb0..591a01e181bc 100644 --- a/drivers/gpu/drm/xe/xe_gt_idle.h +++ b/drivers/gpu/drm/xe/xe_gt_idle.h @@ -17,5 +17,6 @@ void xe_gt_idle_disable_c6(struct xe_gt *gt); void xe_gt_idle_enable_pg(struct xe_gt *gt); void xe_gt_idle_disable_pg(struct xe_gt *gt); int xe_gt_idle_pg_print(struct xe_gt *gt, struct drm_printer *p); +u64 xe_gt_idle_residency_msec(struct xe_gt_idle *gtidle); #endif /* _XE_GT_IDLE_H_ */ diff --git a/drivers/gpu/drm/xe/xe_gt_idle_types.h b/drivers/gpu/drm/xe/xe_gt_idle_types.h index b8b297a3f884..a3667c567f8a 100644 --- a/drivers/gpu/drm/xe/xe_gt_idle_types.h +++ b/drivers/gpu/drm/xe/xe_gt_idle_types.h @@ -6,6 +6,7 @@ #ifndef _XE_GT_IDLE_SYSFS_TYPES_H_ #define _XE_GT_IDLE_SYSFS_TYPES_H_ +#include <linux/spinlock.h> #include <linux/types.h> struct xe_guc_pc; @@ -31,6 +32,8 @@ struct xe_gt_idle { u64 cur_residency; /** @prev_residency: previous residency counter */ u64 prev_residency; + /** @lock: Lock protecting idle residency counters */ + raw_spinlock_t lock; /** @idle_status: get the current idle state */ enum xe_gt_idle_state (*idle_status)(struct xe_guc_pc *pc); /** @idle_residency: get idle residency counter */ diff --git a/drivers/gpu/drm/xe/xe_gt_pagefault.c b/drivers/gpu/drm/xe/xe_gt_pagefault.c index 2606cd396df5..0d0207be93ed 100644 --- a/drivers/gpu/drm/xe/xe_gt_pagefault.c +++ b/drivers/gpu/drm/xe/xe_gt_pagefault.c @@ -422,9 +422,16 @@ static int xe_alloc_pf_queue(struct xe_gt *gt, struct pf_queue *pf_queue) num_eus = bitmap_weight(gt->fuse_topo.eu_mask_per_dss, XE_MAX_EU_FUSE_BITS) * num_dss; - /* user can issue separate page faults per EU and per CS */ + /* + * user can issue separate page faults per EU and per CS + * + * XXX: Multiplier required as compute UMD are getting PF queue errors + * without it. Follow on why this multiplier is required. + */ +#define PF_MULTIPLIER 8 pf_queue->num_dw = - (num_eus + XE_NUM_HW_ENGINES) * PF_MSG_LEN_DW; + (num_eus + XE_NUM_HW_ENGINES) * PF_MSG_LEN_DW * PF_MULTIPLIER; +#undef PF_MULTIPLIER pf_queue->gt = gt; pf_queue->data = devm_kcalloc(xe->drm.dev, pf_queue->num_dw, diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c index 6f906c8e8108..c08efca6420e 100644 --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf.c +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf.c @@ -15,7 +15,11 @@ #include "xe_gt_sriov_pf_helpers.h" #include "xe_gt_sriov_pf_migration.h" #include "xe_gt_sriov_pf_service.h" +#include "xe_gt_sriov_printk.h" #include "xe_mmio.h" +#include "xe_pm.h" + +static void pf_worker_restart_func(struct work_struct *w); /* * VF's metadata is maintained in the flexible array where: @@ -41,6 +45,11 @@ static int pf_alloc_metadata(struct xe_gt *gt) return 0; } +static void pf_init_workers(struct xe_gt *gt) +{ + INIT_WORK(>->sriov.pf.workers.restart, pf_worker_restart_func); +} + /** * xe_gt_sriov_pf_init_early - Prepare SR-IOV PF data structures on PF. * @gt: the &xe_gt to initialize @@ -65,6 +74,8 @@ int xe_gt_sriov_pf_init_early(struct xe_gt *gt) if (err) return err; + pf_init_workers(gt); + return 0; } @@ -78,6 +89,12 @@ int xe_gt_sriov_pf_init_early(struct xe_gt *gt) */ int xe_gt_sriov_pf_init(struct xe_gt *gt) { + int err; + + err = xe_gt_sriov_pf_config_init(gt); + if (err) + return err; + return xe_gt_sriov_pf_migration_init(gt); } @@ -155,6 +172,35 @@ void xe_gt_sriov_pf_sanitize_hw(struct xe_gt *gt, unsigned int vfid) pf_clear_vf_scratch_regs(gt, vfid); } +static void pf_restart(struct xe_gt *gt) +{ + struct xe_device *xe = gt_to_xe(gt); + + xe_pm_runtime_get(xe); + xe_gt_sriov_pf_config_restart(gt); + xe_gt_sriov_pf_control_restart(gt); + xe_pm_runtime_put(xe); + + xe_gt_sriov_dbg(gt, "restart completed\n"); +} + +static void pf_worker_restart_func(struct work_struct *w) +{ + struct xe_gt *gt = container_of(w, typeof(*gt), sriov.pf.workers.restart); + + pf_restart(gt); +} + +static void pf_queue_restart(struct xe_gt *gt) +{ + struct xe_device *xe = gt_to_xe(gt); + + xe_gt_assert(gt, IS_SRIOV_PF(xe)); + + if (!queue_work(xe->sriov.wq, >->sriov.pf.workers.restart)) + xe_gt_sriov_dbg(gt, "restart already in queue!\n"); +} + /** * xe_gt_sriov_pf_restart - Restart SR-IOV support after a GT reset. * @gt: the &xe_gt @@ -163,6 +209,5 @@ void xe_gt_sriov_pf_sanitize_hw(struct xe_gt *gt, unsigned int vfid) */ void xe_gt_sriov_pf_restart(struct xe_gt *gt) { - xe_gt_sriov_pf_config_restart(gt); - xe_gt_sriov_pf_control_restart(gt); + pf_queue_restart(gt); } diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c index 4bd255adfb40..27f309e3a76e 100644 --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.c @@ -336,6 +336,26 @@ static int pf_push_full_vf_config(struct xe_gt *gt, unsigned int vfid) return err; } +static int pf_push_vf_cfg(struct xe_gt *gt, unsigned int vfid, bool reset) +{ + int err = 0; + + xe_gt_assert(gt, vfid); + lockdep_assert_held(xe_gt_sriov_pf_master_mutex(gt)); + + if (reset) + err = pf_send_vf_cfg_reset(gt, vfid); + if (!err) + err = pf_push_full_vf_config(gt, vfid); + + return err; +} + +static int pf_refresh_vf_cfg(struct xe_gt *gt, unsigned int vfid) +{ + return pf_push_vf_cfg(gt, vfid, true); +} + static u64 pf_get_ggtt_alignment(struct xe_gt *gt) { struct xe_device *xe = gt_to_xe(gt); @@ -432,6 +452,10 @@ static int pf_provision_vf_ggtt(struct xe_gt *gt, unsigned int vfid, u64 size) return err; pf_release_vf_config_ggtt(gt, config); + + err = pf_refresh_vf_cfg(gt, vfid); + if (unlikely(err)) + return err; } xe_gt_assert(gt, !xe_ggtt_node_allocated(config->ggtt_region)); @@ -757,6 +781,10 @@ static int pf_provision_vf_ctxs(struct xe_gt *gt, unsigned int vfid, u32 num_ctx return ret; pf_release_config_ctxs(gt, config); + + ret = pf_refresh_vf_cfg(gt, vfid); + if (unlikely(ret)) + return ret; } if (!num_ctxs) @@ -1054,6 +1082,10 @@ static int pf_provision_vf_dbs(struct xe_gt *gt, unsigned int vfid, u32 num_dbs) return ret; pf_release_config_dbs(gt, config); + + ret = pf_refresh_vf_cfg(gt, vfid); + if (unlikely(ret)) + return ret; } if (!num_dbs) @@ -2085,10 +2117,7 @@ int xe_gt_sriov_pf_config_push(struct xe_gt *gt, unsigned int vfid, bool refresh xe_gt_assert(gt, vfid); mutex_lock(xe_gt_sriov_pf_master_mutex(gt)); - if (refresh) - err = pf_send_vf_cfg_reset(gt, vfid); - if (!err) - err = pf_push_full_vf_config(gt, vfid); + err = pf_push_vf_cfg(gt, vfid, refresh); mutex_unlock(xe_gt_sriov_pf_master_mutex(gt)); if (unlikely(err)) { @@ -2320,6 +2349,35 @@ int xe_gt_sriov_pf_config_restore(struct xe_gt *gt, unsigned int vfid, return err; } +static void fini_config(void *arg) +{ + struct xe_gt *gt = arg; + struct xe_device *xe = gt_to_xe(gt); + unsigned int n, total_vfs = xe_sriov_pf_get_totalvfs(xe); + + mutex_lock(xe_gt_sriov_pf_master_mutex(gt)); + for (n = 1; n <= total_vfs; n++) + pf_release_vf_config(gt, n); + mutex_unlock(xe_gt_sriov_pf_master_mutex(gt)); +} + +/** + * xe_gt_sriov_pf_config_init - Initialize SR-IOV configuration data. + * @gt: the &xe_gt + * + * This function can only be called on PF. + * + * Return: 0 on success or a negative error code on failure. + */ +int xe_gt_sriov_pf_config_init(struct xe_gt *gt) +{ + struct xe_device *xe = gt_to_xe(gt); + + xe_gt_assert(gt, IS_SRIOV_PF(xe)); + + return devm_add_action_or_reset(xe->drm.dev, fini_config, gt); +} + /** * xe_gt_sriov_pf_config_restart - Restart SR-IOV configurations after a GT reset. * @gt: the &xe_gt diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h index f894e9d4abba..513e6512a575 100644 --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_config.h @@ -63,6 +63,7 @@ int xe_gt_sriov_pf_config_restore(struct xe_gt *gt, unsigned int vfid, bool xe_gt_sriov_pf_config_is_empty(struct xe_gt *gt, unsigned int vfid); +int xe_gt_sriov_pf_config_init(struct xe_gt *gt); void xe_gt_sriov_pf_config_restart(struct xe_gt *gt); int xe_gt_sriov_pf_config_print_ggtt(struct xe_gt *gt, struct drm_printer *p); diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h b/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h index 0426b1a77069..a64a6835ad65 100644 --- a/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h +++ b/drivers/gpu/drm/xe/xe_gt_sriov_pf_types.h @@ -36,7 +36,16 @@ struct xe_gt_sriov_metadata { }; /** + * struct xe_gt_sriov_pf_workers - GT level workers used by the PF. + */ +struct xe_gt_sriov_pf_workers { + /** @restart: worker that executes actions post GT reset */ + struct work_struct restart; +}; + +/** * struct xe_gt_sriov_pf - GT level PF virtualization data. + * @workers: workers data. * @service: service data. * @control: control data. * @policy: policy data. @@ -45,6 +54,7 @@ struct xe_gt_sriov_metadata { * @vfs: metadata for all VFs. */ struct xe_gt_sriov_pf { + struct xe_gt_sriov_pf_workers workers; struct xe_gt_sriov_pf_service service; struct xe_gt_sriov_pf_control control; struct xe_gt_sriov_pf_policy policy; diff --git a/drivers/gpu/drm/xe/xe_gt_sriov_vf.c b/drivers/gpu/drm/xe/xe_gt_sriov_vf.c index 9c30cbd9af6e..a439261bf4d7 100644 --- a/drivers/gpu/drm/xe/xe_gt_sriov_vf.c +++ b/drivers/gpu/drm/xe/xe_gt_sriov_vf.c @@ -47,12 +47,19 @@ static int guc_action_vf_reset(struct xe_guc *guc) return ret > 0 ? -EPROTO : ret; } +#define GUC_RESET_VF_STATE_RETRY_MAX 10 static int vf_reset_guc_state(struct xe_gt *gt) { + unsigned int retry = GUC_RESET_VF_STATE_RETRY_MAX; struct xe_guc *guc = >->uc.guc; int err; - err = guc_action_vf_reset(guc); + do { + err = guc_action_vf_reset(guc); + if (!err || err != -ETIMEDOUT) + break; + } while (--retry); + if (unlikely(err)) xe_gt_sriov_err(gt, "Failed to reset GuC state (%pe)\n", ERR_PTR(err)); return err; @@ -229,6 +236,9 @@ int xe_gt_sriov_vf_bootstrap(struct xe_gt *gt) { int err; + if (!xe_device_uc_enabled(gt_to_xe(gt))) + return -ENODEV; + err = vf_reset_guc_state(gt); if (unlikely(err)) return err; diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c index 9405d83d4db2..084cbdeba8ea 100644 --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.c @@ -419,6 +419,28 @@ int xe_gt_tlb_invalidation_range(struct xe_gt *gt, } /** + * xe_gt_tlb_invalidation_vm - Issue a TLB invalidation on this GT for a VM + * @gt: graphics tile + * @vm: VM to invalidate + * + * Invalidate entire VM's address space + */ +void xe_gt_tlb_invalidation_vm(struct xe_gt *gt, struct xe_vm *vm) +{ + struct xe_gt_tlb_invalidation_fence fence; + u64 range = 1ull << vm->xe->info.va_bits; + int ret; + + xe_gt_tlb_invalidation_fence_init(gt, &fence, true); + + ret = xe_gt_tlb_invalidation_range(gt, &fence, 0, range, vm->usm.asid); + if (ret < 0) + return; + + xe_gt_tlb_invalidation_fence_wait(&fence); +} + +/** * xe_gt_tlb_invalidation_vma - Issue a TLB invalidation on this GT for a VMA * @gt: GT structure * @fence: invalidation fence which will be signal on TLB invalidation diff --git a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h index 672acfcdf0d7..abe9b03d543e 100644 --- a/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h +++ b/drivers/gpu/drm/xe/xe_gt_tlb_invalidation.h @@ -12,6 +12,7 @@ struct xe_gt; struct xe_guc; +struct xe_vm; struct xe_vma; int xe_gt_tlb_invalidation_init_early(struct xe_gt *gt); @@ -21,6 +22,7 @@ int xe_gt_tlb_invalidation_ggtt(struct xe_gt *gt); int xe_gt_tlb_invalidation_vma(struct xe_gt *gt, struct xe_gt_tlb_invalidation_fence *fence, struct xe_vma *vma); +void xe_gt_tlb_invalidation_vm(struct xe_gt *gt, struct xe_vm *vm); int xe_gt_tlb_invalidation_range(struct xe_gt *gt, struct xe_gt_tlb_invalidation_fence *fence, u64 start, u64 end, u32 asid); diff --git a/drivers/gpu/drm/xe/xe_gt_types.h b/drivers/gpu/drm/xe/xe_gt_types.h index 6e66bf0e8b3f..dd2969a1846d 100644 --- a/drivers/gpu/drm/xe/xe_gt_types.h +++ b/drivers/gpu/drm/xe/xe_gt_types.h @@ -413,6 +413,16 @@ struct xe_gt { bool oob_initialized; } wa_active; + /** @tuning_active: keep track of active tunings */ + struct { + /** @tuning_active.gt: bitmap with active GT tunings */ + unsigned long *gt; + /** @tuning_active.engine: bitmap with active engine tunings */ + unsigned long *engine; + /** @tuning_active.lrc: bitmap with active LRC tunings */ + unsigned long *lrc; + } tuning_active; + /** @user_engines: engines present in GT and available to userspace */ struct { /** diff --git a/drivers/gpu/drm/xe/xe_guc_capture.c b/drivers/gpu/drm/xe/xe_guc_capture.c index f6d523e4c5fe..9095618648bc 100644 --- a/drivers/gpu/drm/xe/xe_guc_capture.c +++ b/drivers/gpu/drm/xe/xe_guc_capture.c @@ -359,7 +359,7 @@ static void __fill_ext_reg(struct __guc_mmio_reg_descr *ext, ext->reg = XE_REG(extlist->reg.__reg.addr); ext->flags = FIELD_PREP(GUC_REGSET_STEERING_NEEDED, 1); - ext->flags = FIELD_PREP(GUC_REGSET_STEERING_GROUP, slice_id); + ext->flags |= FIELD_PREP(GUC_REGSET_STEERING_GROUP, slice_id); ext->flags |= FIELD_PREP(GUC_REGSET_STEERING_INSTANCE, subslice_id); ext->regname = extlist->name; } diff --git a/drivers/gpu/drm/xe/xe_guc_log.c b/drivers/gpu/drm/xe/xe_guc_log.c index 0ca3056d8bd3..80514a446ba2 100644 --- a/drivers/gpu/drm/xe/xe_guc_log.c +++ b/drivers/gpu/drm/xe/xe_guc_log.c @@ -149,16 +149,12 @@ struct xe_guc_log_snapshot *xe_guc_log_snapshot_capture(struct xe_guc_log *log, size_t remain; int i; - if (!log->bo) { - xe_gt_err(gt, "GuC log buffer not allocated\n"); + if (!log->bo) return NULL; - } snapshot = xe_guc_log_snapshot_alloc(log, atomic); - if (!snapshot) { - xe_gt_err(gt, "GuC log snapshot not allocated\n"); + if (!snapshot) return NULL; - } remain = snapshot->size; for (i = 0; i < snapshot->num_chunks; i++) { diff --git a/drivers/gpu/drm/xe/xe_guc_pc.c b/drivers/gpu/drm/xe/xe_guc_pc.c index f382f5d53ca8..2276d85926fc 100644 --- a/drivers/gpu/drm/xe/xe_guc_pc.c +++ b/drivers/gpu/drm/xe/xe_guc_pc.c @@ -371,16 +371,17 @@ static void tgl_update_rpa_value(struct xe_guc_pc *pc) u32 reg; /* - * For PVC we still need to use fused RP1 as the approximation for RPe - * For other platforms than PVC we get the resolved RPe directly from + * For PVC we still need to use fused RP0 as the approximation for RPa + * For other platforms than PVC we get the resolved RPa directly from * PCODE at a different register */ - if (xe->info.platform == XE_PVC) + if (xe->info.platform == XE_PVC) { reg = xe_mmio_read32(>->mmio, PVC_RP_STATE_CAP); - else + pc->rpa_freq = REG_FIELD_GET(RP0_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + } else { reg = xe_mmio_read32(>->mmio, FREQ_INFO_REC); - - pc->rpa_freq = REG_FIELD_GET(RPA_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + pc->rpa_freq = REG_FIELD_GET(RPA_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + } } static void tgl_update_rpe_value(struct xe_guc_pc *pc) @@ -394,12 +395,13 @@ static void tgl_update_rpe_value(struct xe_guc_pc *pc) * For other platforms than PVC we get the resolved RPe directly from * PCODE at a different register */ - if (xe->info.platform == XE_PVC) + if (xe->info.platform == XE_PVC) { reg = xe_mmio_read32(>->mmio, PVC_RP_STATE_CAP); - else + pc->rpe_freq = REG_FIELD_GET(RP1_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + } else { reg = xe_mmio_read32(>->mmio, FREQ_INFO_REC); - - pc->rpe_freq = REG_FIELD_GET(RPE_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + pc->rpe_freq = REG_FIELD_GET(RPE_MASK, reg) * GT_FREQUENCY_MULTIPLIER; + } } static void pc_update_rp_values(struct xe_guc_pc *pc) diff --git a/drivers/gpu/drm/xe/xe_guc_relay.c b/drivers/gpu/drm/xe/xe_guc_relay.c index 8f62de026724..e5dc94f3e618 100644 --- a/drivers/gpu/drm/xe/xe_guc_relay.c +++ b/drivers/gpu/drm/xe/xe_guc_relay.c @@ -225,7 +225,7 @@ __relay_get_transaction(struct xe_guc_relay *relay, bool incoming, u32 remote, u * with CTB lock held which is marked as used in the reclaim path. * Btw, that's one of the reason why we use mempool here! */ - txn = mempool_alloc(&relay->pool, incoming ? GFP_ATOMIC : GFP_KERNEL); + txn = mempool_alloc(&relay->pool, incoming ? GFP_ATOMIC : GFP_NOWAIT); if (!txn) return ERR_PTR(-ENOMEM); diff --git a/drivers/gpu/drm/xe/xe_hw_engine.c b/drivers/gpu/drm/xe/xe_hw_engine.c index fc447751fe78..b26b6fb5cdb5 100644 --- a/drivers/gpu/drm/xe/xe_hw_engine.c +++ b/drivers/gpu/drm/xe/xe_hw_engine.c @@ -386,12 +386,6 @@ xe_hw_engine_setup_default_lrc_state(struct xe_hw_engine *hwe) blit_cctl_val, XE_RTP_ACTION_FLAG(ENGINE_BASE))) }, - /* Use Fixed slice CCS mode */ - { XE_RTP_NAME("RCU_MODE_FIXED_SLICE_CCS_MODE"), - XE_RTP_RULES(FUNC(xe_hw_engine_match_fixed_cslice_mode)), - XE_RTP_ACTIONS(FIELD_SET(RCU_MODE, RCU_MODE_FIXED_SLICE_CCS_MODE, - RCU_MODE_FIXED_SLICE_CCS_MODE)) - }, /* Disable WMTP if HW doesn't support it */ { XE_RTP_NAME("DISABLE_WMTP_ON_UNSUPPORTED_HW"), XE_RTP_RULES(FUNC(xe_rtp_cfeg_wmtp_disabled)), @@ -400,10 +394,9 @@ xe_hw_engine_setup_default_lrc_state(struct xe_hw_engine *hwe) PREEMPT_GPGPU_THREAD_GROUP_LEVEL)), XE_RTP_ENTRY_FLAG(FOREACH_ENGINE) }, - {} }; - xe_rtp_process_to_sr(&ctx, lrc_setup, &hwe->reg_lrc); + xe_rtp_process_to_sr(&ctx, lrc_setup, ARRAY_SIZE(lrc_setup), &hwe->reg_lrc); } static void @@ -459,10 +452,15 @@ hw_engine_setup_default_state(struct xe_hw_engine *hwe) XE_RTP_ACTIONS(SET(CSFE_CHICKEN1(0), CS_PRIORITY_MEM_READ, XE_RTP_ACTION_FLAG(ENGINE_BASE))) }, - {} + /* Use Fixed slice CCS mode */ + { XE_RTP_NAME("RCU_MODE_FIXED_SLICE_CCS_MODE"), + XE_RTP_RULES(FUNC(xe_hw_engine_match_fixed_cslice_mode)), + XE_RTP_ACTIONS(FIELD_SET(RCU_MODE, RCU_MODE_FIXED_SLICE_CCS_MODE, + RCU_MODE_FIXED_SLICE_CCS_MODE)) + }, }; - xe_rtp_process_to_sr(&ctx, engine_entries, &hwe->reg_sr); + xe_rtp_process_to_sr(&ctx, engine_entries, ARRAY_SIZE(engine_entries), &hwe->reg_sr); } static const struct engine_info *find_engine_info(enum xe_engine_class class, int instance) diff --git a/drivers/gpu/drm/xe/xe_lrc.c b/drivers/gpu/drm/xe/xe_lrc.c index bbb9ffbf6367..5d7629bb6b8d 100644 --- a/drivers/gpu/drm/xe/xe_lrc.c +++ b/drivers/gpu/drm/xe/xe_lrc.c @@ -684,7 +684,7 @@ static inline u32 __xe_lrc_start_seqno_offset(struct xe_lrc *lrc) static u32 __xe_lrc_ctx_job_timestamp_offset(struct xe_lrc *lrc) { - /* The start seqno is stored in the driver-defined portion of PPHWSP */ + /* This is stored in the driver-defined portion of PPHWSP */ return xe_lrc_pphwsp_offset(lrc) + LRC_CTX_JOB_TIMESTAMP_OFFSET; } @@ -864,7 +864,7 @@ static void *empty_lrc_data(struct xe_hw_engine *hwe) static void xe_lrc_set_ppgtt(struct xe_lrc *lrc, struct xe_vm *vm) { - u64 desc = xe_vm_pdp4_descriptor(vm, lrc->tile); + u64 desc = xe_vm_pdp4_descriptor(vm, gt_to_tile(lrc->gt)); xe_lrc_write_ctx_reg(lrc, CTX_PDP0_UDW, upper_32_bits(desc)); xe_lrc_write_ctx_reg(lrc, CTX_PDP0_LDW, lower_32_bits(desc)); @@ -895,6 +895,7 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe, int err; kref_init(&lrc->refcount); + lrc->gt = gt; lrc->flags = 0; lrc_size = ring_size + xe_gt_lrc_size(gt, hwe->class); if (xe_gt_has_indirect_ring_state(gt)) @@ -913,7 +914,6 @@ static int xe_lrc_init(struct xe_lrc *lrc, struct xe_hw_engine *hwe, return PTR_ERR(lrc->bo); lrc->size = lrc_size; - lrc->tile = gt_to_tile(hwe->gt); lrc->ring.size = ring_size; lrc->ring.tail = 0; lrc->ctx_timestamp = 0; diff --git a/drivers/gpu/drm/xe/xe_lrc_types.h b/drivers/gpu/drm/xe/xe_lrc_types.h index 71ecb453f811..cd38586ae989 100644 --- a/drivers/gpu/drm/xe/xe_lrc_types.h +++ b/drivers/gpu/drm/xe/xe_lrc_types.h @@ -25,8 +25,8 @@ struct xe_lrc { /** @size: size of lrc including any indirect ring state page */ u32 size; - /** @tile: tile which this LRC belongs to */ - struct xe_tile *tile; + /** @gt: gt which this LRC belongs to */ + struct xe_gt *gt; /** @flags: LRC flags */ #define XE_LRC_FLAG_INDIRECT_RING_STATE 0x1 diff --git a/drivers/gpu/drm/xe/xe_mmio.c b/drivers/gpu/drm/xe/xe_mmio.c index a48f239cad1c..9c2f60ce0c94 100644 --- a/drivers/gpu/drm/xe/xe_mmio.c +++ b/drivers/gpu/drm/xe/xe_mmio.c @@ -76,12 +76,12 @@ static void mmio_multi_tile_setup(struct xe_device *xe, size_t tile_mmio_size) * is fine as it's going to the root tile's mmio, that's * guaranteed to be initialized earlier in xe_mmio_init() */ - mtcfg = xe_mmio_read64_2x32(mmio, XEHP_MTCFG_ADDR); + mtcfg = xe_mmio_read32(mmio, XEHP_MTCFG_ADDR); tile_count = REG_FIELD_GET(TILE_COUNT, mtcfg) + 1; if (tile_count < xe->info.tile_count) { drm_info(&xe->drm, "tile_count: %d, reduced_tile_count %d\n", - xe->info.tile_count, tile_count); + xe->info.tile_count, tile_count); xe->info.tile_count = tile_count; /* @@ -173,7 +173,7 @@ int xe_mmio_init(struct xe_device *xe) */ xe->mmio.size = pci_resource_len(pdev, GTTMMADR_BAR); xe->mmio.regs = pci_iomap(pdev, GTTMMADR_BAR, 0); - if (xe->mmio.regs == NULL) { + if (!xe->mmio.regs) { drm_err(&xe->drm, "failed to map registers\n"); return -EIO; } @@ -338,8 +338,8 @@ u64 xe_mmio_read64_2x32(struct xe_mmio *mmio, struct xe_reg reg) return (u64)udw << 32 | ldw; } -static int __xe_mmio_wait32(struct xe_mmio *mmio, struct xe_reg reg, u32 mask, u32 val, u32 timeout_us, - u32 *out_val, bool atomic, bool expect_match) +static int __xe_mmio_wait32(struct xe_mmio *mmio, struct xe_reg reg, u32 mask, u32 val, + u32 timeout_us, u32 *out_val, bool atomic, bool expect_match) { ktime_t cur = ktime_get_raw(); const ktime_t end = ktime_add_us(cur, timeout_us); diff --git a/drivers/gpu/drm/xe/xe_oa.c b/drivers/gpu/drm/xe/xe_oa.c index eb6cd91e1e22..abf37d9ab221 100644 --- a/drivers/gpu/drm/xe/xe_oa.c +++ b/drivers/gpu/drm/xe/xe_oa.c @@ -548,6 +548,7 @@ static ssize_t xe_oa_read(struct file *file, char __user *buf, mutex_unlock(&stream->stream_lock); } while (!offset && !ret); } else { + xe_oa_buffer_check_unlocked(stream); mutex_lock(&stream->stream_lock); ret = __xe_oa_read(stream, buf, count, &offset); mutex_unlock(&stream->stream_lock); diff --git a/drivers/gpu/drm/xe/xe_pci.c b/drivers/gpu/drm/xe/xe_pci.c index 39be74848e44..d92b2e5885b9 100644 --- a/drivers/gpu/drm/xe/xe_pci.c +++ b/drivers/gpu/drm/xe/xe_pci.c @@ -150,7 +150,6 @@ static const struct xe_graphics_desc graphics_xehpc = { }; static const struct xe_graphics_desc graphics_xelpg = { - .name = "Xe_LPG", .hw_engine_mask = BIT(XE_HW_ENGINE_RCS0) | BIT(XE_HW_ENGINE_BCS0) | BIT(XE_HW_ENGINE_CCS0), @@ -174,8 +173,6 @@ static const struct xe_graphics_desc graphics_xelpg = { GENMASK(XE_HW_ENGINE_CCS3, XE_HW_ENGINE_CCS0) static const struct xe_graphics_desc graphics_xe2 = { - .name = "Xe2_LPG / Xe2_HPG / Xe3_LPG", - XE2_GFX_FEATURES, }; @@ -200,15 +197,6 @@ static const struct xe_media_desc media_xehpm = { }; static const struct xe_media_desc media_xelpmp = { - .name = "Xe_LPM+", - .hw_engine_mask = - GENMASK(XE_HW_ENGINE_VCS7, XE_HW_ENGINE_VCS0) | - GENMASK(XE_HW_ENGINE_VECS3, XE_HW_ENGINE_VECS0) | - BIT(XE_HW_ENGINE_GSCCS0) -}; - -static const struct xe_media_desc media_xe2 = { - .name = "Xe2_LPM / Xe2_HPM / Xe3_LPM", .hw_engine_mask = GENMASK(XE_HW_ENGINE_VCS7, XE_HW_ENGINE_VCS0) | GENMASK(XE_HW_ENGINE_VECS3, XE_HW_ENGINE_VECS0) | @@ -357,21 +345,21 @@ __diag_pop(); /* Map of GMD_ID values to graphics IP */ static const struct gmdid_map graphics_ip_map[] = { - { 1270, &graphics_xelpg }, - { 1271, &graphics_xelpg }, - { 1274, &graphics_xelpg }, /* Xe_LPG+ */ - { 2001, &graphics_xe2 }, - { 2004, &graphics_xe2 }, - { 3000, &graphics_xe2 }, - { 3001, &graphics_xe2 }, + { 1270, "Xe_LPG", &graphics_xelpg }, + { 1271, "Xe_LPG", &graphics_xelpg }, + { 1274, "Xe_LPG+", &graphics_xelpg }, + { 2001, "Xe2_HPG", &graphics_xe2 }, + { 2004, "Xe2_LPG", &graphics_xe2 }, + { 3000, "Xe3_LPG", &graphics_xe2 }, + { 3001, "Xe3_LPG", &graphics_xe2 }, }; /* Map of GMD_ID values to media IP */ static const struct gmdid_map media_ip_map[] = { - { 1300, &media_xelpmp }, - { 1301, &media_xe2 }, - { 2000, &media_xe2 }, - { 3000, &media_xe2 }, + { 1300, "Xe_LPM+", &media_xelpmp }, + { 1301, "Xe2_HPM", &media_xelpmp }, + { 2000, "Xe2_LPM", &media_xelpmp }, + { 3000, "Xe3_LPM", &media_xelpmp }, }; /* @@ -502,6 +490,7 @@ static void read_gmdid(struct xe_device *xe, enum xe_gmdid_type type, u32 *ver, gt->info.type = XE_GT_TYPE_MAIN; } + xe_gt_mmio_init(gt); xe_guc_comm_init_early(>->uc.guc); /* Don't bother with GMDID if failed to negotiate the GuC ABI */ @@ -566,6 +555,7 @@ static void handle_gmdid(struct xe_device *xe, for (int i = 0; i < ARRAY_SIZE(graphics_ip_map); i++) { if (ver == graphics_ip_map[i].ver) { xe->info.graphics_verx100 = ver; + xe->info.graphics_name = graphics_ip_map[i].name; *graphics = graphics_ip_map[i].ip; break; @@ -586,6 +576,7 @@ static void handle_gmdid(struct xe_device *xe, for (int i = 0; i < ARRAY_SIZE(media_ip_map); i++) { if (ver == media_ip_map[i].ver) { xe->info.media_verx100 = ver; + xe->info.media_name = media_ip_map[i].name; *media = media_ip_map[i].ip; break; diff --git a/drivers/gpu/drm/xe/xe_pci_sriov.c b/drivers/gpu/drm/xe/xe_pci_sriov.c index aaceee748287..09ee8a06fe2e 100644 --- a/drivers/gpu/drm/xe/xe_pci_sriov.c +++ b/drivers/gpu/drm/xe/xe_pci_sriov.c @@ -62,6 +62,55 @@ static void pf_reset_vfs(struct xe_device *xe, unsigned int num_vfs) xe_gt_sriov_pf_control_trigger_flr(gt, n); } +static struct pci_dev *xe_pci_pf_get_vf_dev(struct xe_device *xe, unsigned int vf_id) +{ + struct pci_dev *pdev = to_pci_dev(xe->drm.dev); + + xe_assert(xe, IS_SRIOV_PF(xe)); + + /* caller must use pci_dev_put() */ + return pci_get_domain_bus_and_slot(pci_domain_nr(pdev->bus), + pdev->bus->number, + pci_iov_virtfn_devfn(pdev, vf_id)); +} + +static void pf_link_vfs(struct xe_device *xe, int num_vfs) +{ + struct pci_dev *pdev_pf = to_pci_dev(xe->drm.dev); + struct device_link *link; + struct pci_dev *pdev_vf; + unsigned int n; + + /* + * When both PF and VF devices are enabled on the host, during system + * resume they are resuming in parallel. + * + * But PF has to complete the provision of VF first to allow any VFs to + * successfully resume. + * + * Create a parent-child device link between PF and VF devices that will + * enforce correct resume order. + */ + for (n = 1; n <= num_vfs; n++) { + pdev_vf = xe_pci_pf_get_vf_dev(xe, n - 1); + + /* unlikely, something weird is happening, abort */ + if (!pdev_vf) { + xe_sriov_err(xe, "Cannot find VF%u device, aborting link%s creation!\n", + n, str_plural(num_vfs)); + break; + } + + link = device_link_add(&pdev_vf->dev, &pdev_pf->dev, + DL_FLAG_AUTOREMOVE_CONSUMER); + /* unlikely and harmless, continue with other VFs */ + if (!link) + xe_sriov_notice(xe, "Failed linking VF%u\n", n); + + pci_dev_put(pdev_vf); + } +} + static int pf_enable_vfs(struct xe_device *xe, int num_vfs) { struct pci_dev *pdev = to_pci_dev(xe->drm.dev); @@ -92,6 +141,8 @@ static int pf_enable_vfs(struct xe_device *xe, int num_vfs) if (err < 0) goto failed; + pf_link_vfs(xe, num_vfs); + xe_sriov_info(xe, "Enabled %u of %u VF%s\n", num_vfs, total_vfs, str_plural(total_vfs)); return num_vfs; diff --git a/drivers/gpu/drm/xe/xe_pci_types.h b/drivers/gpu/drm/xe/xe_pci_types.h index 79b0f80376a4..665b4447b2eb 100644 --- a/drivers/gpu/drm/xe/xe_pci_types.h +++ b/drivers/gpu/drm/xe/xe_pci_types.h @@ -44,6 +44,7 @@ struct xe_media_desc { struct gmdid_map { unsigned int ver; + const char *name; const void *ip; }; diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c index dc24baa84092..148d90611eeb 100644 --- a/drivers/gpu/drm/xe/xe_pt.c +++ b/drivers/gpu/drm/xe/xe_pt.c @@ -219,6 +219,20 @@ void xe_pt_destroy(struct xe_pt *pt, u32 flags, struct llist_head *deferred) } /** + * xe_pt_clear() - Clear a page-table. + * @xe: xe device. + * @pt: The page-table. + * + * Clears page-table by setting to zero. + */ +void xe_pt_clear(struct xe_device *xe, struct xe_pt *pt) +{ + struct iosys_map *map = &pt->bo->vmap; + + xe_map_memset(xe, map, 0, 0, SZ_4K); +} + +/** * DOC: Pagetable building * * Below we use the term "page-table" for both page-directories, containing diff --git a/drivers/gpu/drm/xe/xe_pt.h b/drivers/gpu/drm/xe/xe_pt.h index 9ab386431cad..8e43912ae8e9 100644 --- a/drivers/gpu/drm/xe/xe_pt.h +++ b/drivers/gpu/drm/xe/xe_pt.h @@ -13,6 +13,7 @@ struct dma_fence; struct xe_bo; struct xe_device; struct xe_exec_queue; +struct xe_svm_range; struct xe_sync_entry; struct xe_tile; struct xe_vm; @@ -35,6 +36,8 @@ void xe_pt_populate_empty(struct xe_tile *tile, struct xe_vm *vm, void xe_pt_destroy(struct xe_pt *pt, u32 flags, struct llist_head *deferred); +void xe_pt_clear(struct xe_device *xe, struct xe_pt *pt); + int xe_pt_update_ops_prepare(struct xe_tile *tile, struct xe_vma_ops *vops); struct dma_fence *xe_pt_update_ops_run(struct xe_tile *tile, struct xe_vma_ops *vops); diff --git a/drivers/gpu/drm/xe/xe_reg_whitelist.c b/drivers/gpu/drm/xe/xe_reg_whitelist.c index edab5d4e3ba5..23f6c81d9994 100644 --- a/drivers/gpu/drm/xe/xe_reg_whitelist.c +++ b/drivers/gpu/drm/xe/xe_reg_whitelist.c @@ -88,7 +88,6 @@ static const struct xe_rtp_entry_sr register_whitelist[] = { RING_FORCE_TO_NONPRIV_ACCESS_RD | RING_FORCE_TO_NONPRIV_RANGE_4)) }, - {} }; static void whitelist_apply_to_hwe(struct xe_hw_engine *hwe) @@ -137,7 +136,8 @@ void xe_reg_whitelist_process_engine(struct xe_hw_engine *hwe) { struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(hwe); - xe_rtp_process_to_sr(&ctx, register_whitelist, &hwe->reg_whitelist); + xe_rtp_process_to_sr(&ctx, register_whitelist, ARRAY_SIZE(register_whitelist), + &hwe->reg_whitelist); whitelist_apply_to_hwe(hwe); } diff --git a/drivers/gpu/drm/xe/xe_ring_ops.c b/drivers/gpu/drm/xe/xe_ring_ops.c index 9f327f27c072..349317794768 100644 --- a/drivers/gpu/drm/xe/xe_ring_ops.c +++ b/drivers/gpu/drm/xe/xe_ring_ops.c @@ -141,7 +141,8 @@ emit_pipe_control(u32 *dw, int i, u32 bit_group_0, u32 bit_group_1, u32 offset, static int emit_pipe_invalidate(u32 mask_flags, bool invalidate_tlb, u32 *dw, int i) { - u32 flags = PIPE_CONTROL_CS_STALL | + u32 flags0 = 0; + u32 flags1 = PIPE_CONTROL_CS_STALL | PIPE_CONTROL_COMMAND_CACHE_INVALIDATE | PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE | PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE | @@ -152,11 +153,15 @@ static int emit_pipe_invalidate(u32 mask_flags, bool invalidate_tlb, u32 *dw, PIPE_CONTROL_STORE_DATA_INDEX; if (invalidate_tlb) - flags |= PIPE_CONTROL_TLB_INVALIDATE; + flags1 |= PIPE_CONTROL_TLB_INVALIDATE; - flags &= ~mask_flags; + flags1 &= ~mask_flags; - return emit_pipe_control(dw, i, 0, flags, LRC_PPHWSP_SCRATCH_ADDR, 0); + if (flags1 & PIPE_CONTROL_VF_CACHE_INVALIDATE) + flags0 |= PIPE_CONTROL0_L3_READ_ONLY_CACHE_INVALIDATE; + + return emit_pipe_control(dw, i, flags0, flags1, + LRC_PPHWSP_SCRATCH_ADDR, 0); } static int emit_store_imm_ppgtt_posted(u64 addr, u64 value, @@ -229,13 +234,10 @@ static u32 get_ppgtt_flag(struct xe_sched_job *job) static int emit_copy_timestamp(struct xe_lrc *lrc, u32 *dw, int i) { - dw[i++] = MI_COPY_MEM_MEM | MI_COPY_MEM_MEM_SRC_GGTT | - MI_COPY_MEM_MEM_DST_GGTT; + dw[i++] = MI_STORE_REGISTER_MEM | MI_SRM_USE_GGTT | MI_SRM_ADD_CS_OFFSET; + dw[i++] = RING_CTX_TIMESTAMP(0).addr; dw[i++] = xe_lrc_ctx_job_timestamp_ggtt_addr(lrc); dw[i++] = 0; - dw[i++] = xe_lrc_ctx_timestamp_ggtt_addr(lrc); - dw[i++] = 0; - dw[i++] = MI_NOOP; return i; } diff --git a/drivers/gpu/drm/xe/xe_rtp.c b/drivers/gpu/drm/xe/xe_rtp.c index 7a1c78fdfc92..13bb62d3e615 100644 --- a/drivers/gpu/drm/xe/xe_rtp.c +++ b/drivers/gpu/drm/xe/xe_rtp.c @@ -237,6 +237,7 @@ static void rtp_mark_active(struct xe_device *xe, * the save-restore argument. * @ctx: The context for processing the table, with one of device, gt or hwe * @entries: Table with RTP definitions + * @n_entries: Number of entries to process, usually ARRAY_SIZE(entries) * @sr: Save-restore struct where matching rules execute the action. This can be * viewed as the "coalesced view" of multiple the tables. The bits for each * register set are expected not to collide with previously added entries @@ -247,6 +248,7 @@ static void rtp_mark_active(struct xe_device *xe, */ void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx, const struct xe_rtp_entry_sr *entries, + size_t n_entries, struct xe_reg_sr *sr) { const struct xe_rtp_entry_sr *entry; @@ -259,7 +261,9 @@ void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx, if (IS_SRIOV_VF(xe)) return; - for (entry = entries; entry && entry->name; entry++) { + xe_assert(xe, entries); + + for (entry = entries; entry - entries < n_entries; entry++) { bool match = false; if (entry->flags & XE_RTP_ENTRY_FLAG_FOREACH_ENGINE) { diff --git a/drivers/gpu/drm/xe/xe_rtp.h b/drivers/gpu/drm/xe/xe_rtp.h index 38b9f13bba5e..4fe736a11c42 100644 --- a/drivers/gpu/drm/xe/xe_rtp.h +++ b/drivers/gpu/drm/xe/xe_rtp.h @@ -430,7 +430,7 @@ void xe_rtp_process_ctx_enable_active_tracking(struct xe_rtp_process_ctx *ctx, void xe_rtp_process_to_sr(struct xe_rtp_process_ctx *ctx, const struct xe_rtp_entry_sr *entries, - struct xe_reg_sr *sr); + size_t n_entries, struct xe_reg_sr *sr); void xe_rtp_process(struct xe_rtp_process_ctx *ctx, const struct xe_rtp_entry *entries); diff --git a/drivers/gpu/drm/xe/xe_sa.c b/drivers/gpu/drm/xe/xe_sa.c index e055bed7ae55..4e7aba445ebc 100644 --- a/drivers/gpu/drm/xe/xe_sa.c +++ b/drivers/gpu/drm/xe/xe_sa.c @@ -57,8 +57,6 @@ struct xe_sa_manager *xe_sa_bo_manager_init(struct xe_tile *tile, u32 size, u32 } sa_manager->bo = bo; sa_manager->is_iomem = bo->vmap.is_iomem; - - drm_suballoc_manager_init(&sa_manager->base, managed_size, align); sa_manager->gpu_addr = xe_bo_ggtt_addr(bo); if (bo->vmap.is_iomem) { @@ -72,6 +70,7 @@ struct xe_sa_manager *xe_sa_bo_manager_init(struct xe_tile *tile, u32 size, u32 memset(sa_manager->cpu_ptr, 0, bo->ttm.base.size); } + drm_suballoc_manager_init(&sa_manager->base, managed_size, align); ret = drmm_add_action_or_reset(&xe->drm, xe_sa_bo_manager_fini, sa_manager); if (ret) diff --git a/drivers/gpu/drm/xe/xe_tile.c b/drivers/gpu/drm/xe/xe_tile.c index 07cf7cfe4abd..377438ea6b83 100644 --- a/drivers/gpu/drm/xe/xe_tile.c +++ b/drivers/gpu/drm/xe/xe_tile.c @@ -170,17 +170,19 @@ int xe_tile_init_noalloc(struct xe_tile *tile) if (err) return err; + xe_wa_apply_tile_workarounds(tile); + + return xe_tile_sysfs_init(tile); +} + +int xe_tile_init(struct xe_tile *tile) +{ tile->mem.kernel_bb_pool = xe_sa_bo_manager_init(tile, SZ_1M, 16); if (IS_ERR(tile->mem.kernel_bb_pool)) return PTR_ERR(tile->mem.kernel_bb_pool); - xe_wa_apply_tile_workarounds(tile); - - err = xe_tile_sysfs_init(tile); - return 0; } - void xe_tile_migrate_wait(struct xe_tile *tile) { xe_migrate_wait(tile->migrate); diff --git a/drivers/gpu/drm/xe/xe_tile.h b/drivers/gpu/drm/xe/xe_tile.h index 1c9e42ade6b0..eb939316d55b 100644 --- a/drivers/gpu/drm/xe/xe_tile.h +++ b/drivers/gpu/drm/xe/xe_tile.h @@ -12,6 +12,7 @@ struct xe_tile; int xe_tile_init_early(struct xe_tile *tile, struct xe_device *xe, u8 id); int xe_tile_init_noalloc(struct xe_tile *tile); +int xe_tile_init(struct xe_tile *tile); void xe_tile_migrate_wait(struct xe_tile *tile); diff --git a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c index d414421f8c13..d9c9d2547aad 100644 --- a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c +++ b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.c @@ -207,17 +207,16 @@ static u64 detect_stolen(struct xe_device *xe, struct xe_ttm_stolen_mgr *mgr) #endif } -void xe_ttm_stolen_mgr_init(struct xe_device *xe) +int xe_ttm_stolen_mgr_init(struct xe_device *xe) { - struct xe_ttm_stolen_mgr *mgr = drmm_kzalloc(&xe->drm, sizeof(*mgr), GFP_KERNEL); struct pci_dev *pdev = to_pci_dev(xe->drm.dev); + struct xe_ttm_stolen_mgr *mgr; u64 stolen_size, io_size; int err; - if (!mgr) { - drm_dbg_kms(&xe->drm, "Stolen mgr init failed\n"); - return; - } + mgr = drmm_kzalloc(&xe->drm, sizeof(*mgr), GFP_KERNEL); + if (!mgr) + return -ENOMEM; if (IS_SRIOV_VF(xe)) stolen_size = 0; @@ -230,7 +229,7 @@ void xe_ttm_stolen_mgr_init(struct xe_device *xe) if (!stolen_size) { drm_dbg_kms(&xe->drm, "No stolen memory support\n"); - return; + return 0; } /* @@ -246,7 +245,7 @@ void xe_ttm_stolen_mgr_init(struct xe_device *xe) io_size, PAGE_SIZE); if (err) { drm_dbg_kms(&xe->drm, "Stolen mgr init failed: %i\n", err); - return; + return err; } drm_dbg_kms(&xe->drm, "Initialized stolen memory support with %llu bytes\n", @@ -254,6 +253,8 @@ void xe_ttm_stolen_mgr_init(struct xe_device *xe) if (io_size) mgr->mapping = devm_ioremap_wc(&pdev->dev, mgr->io_base, io_size); + + return 0; } u64 xe_ttm_stolen_io_offset(struct xe_bo *bo, u32 offset) diff --git a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.h b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.h index 1777245ff810..8e877d1e839b 100644 --- a/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.h +++ b/drivers/gpu/drm/xe/xe_ttm_stolen_mgr.h @@ -12,7 +12,7 @@ struct ttm_resource; struct xe_bo; struct xe_device; -void xe_ttm_stolen_mgr_init(struct xe_device *xe); +int xe_ttm_stolen_mgr_init(struct xe_device *xe); int xe_ttm_stolen_io_mem_reserve(struct xe_device *xe, struct ttm_resource *mem); bool xe_ttm_stolen_cpu_access_needs_ggtt(struct xe_device *xe); u64 xe_ttm_stolen_io_offset(struct xe_bo *bo, u32 offset); diff --git a/drivers/gpu/drm/xe/xe_tuning.c b/drivers/gpu/drm/xe/xe_tuning.c index 3c78f3d71559..a61a2917590f 100644 --- a/drivers/gpu/drm/xe/xe_tuning.c +++ b/drivers/gpu/drm/xe/xe_tuning.c @@ -7,6 +7,8 @@ #include <kunit/visibility.h> +#include <drm/drm_managed.h> + #include "regs/xe_gt_regs.h" #include "xe_gt_types.h" #include "xe_platform_types.h" @@ -83,8 +85,6 @@ static const struct xe_rtp_entry_sr gt_tunings[] = { XE_RTP_RULES(MEDIA_VERSION(2000)), XE_RTP_ACTIONS(SET(XE2LPM_SCRATCH3_LBCF, RWFLUSHALLEN)) }, - - {} }; static const struct xe_rtp_entry_sr engine_tunings[] = { @@ -93,7 +93,6 @@ static const struct xe_rtp_entry_sr engine_tunings[] = { ENGINE_CLASS(RENDER)), XE_RTP_ACTIONS(SET(SAMPLER_MODE, INDIRECT_STATE_BASE_ADDR_OVERRIDE)) }, - {} }; static const struct xe_rtp_entry_sr lrc_tunings[] = { @@ -131,15 +130,47 @@ static const struct xe_rtp_entry_sr lrc_tunings[] = { XE_RTP_ACTIONS(FIELD_SET(FF_MODE, VS_HIT_MAX_VALUE_MASK, REG_FIELD_PREP(VS_HIT_MAX_VALUE_MASK, 0x3f))) }, - - {} }; +/** + * xe_tuning_init - initialize gt with tunings bookkeeping + * @gt: GT instance to initialize + * + * Returns 0 for success, negative error code otherwise. + */ +int xe_tuning_init(struct xe_gt *gt) +{ + struct xe_device *xe = gt_to_xe(gt); + size_t n_lrc, n_engine, n_gt, total; + unsigned long *p; + + n_gt = BITS_TO_LONGS(ARRAY_SIZE(gt_tunings)); + n_engine = BITS_TO_LONGS(ARRAY_SIZE(engine_tunings)); + n_lrc = BITS_TO_LONGS(ARRAY_SIZE(lrc_tunings)); + total = n_gt + n_engine + n_lrc; + + p = drmm_kzalloc(&xe->drm, sizeof(*p) * total, GFP_KERNEL); + if (!p) + return -ENOMEM; + + gt->tuning_active.gt = p; + p += n_gt; + gt->tuning_active.engine = p; + p += n_engine; + gt->tuning_active.lrc = p; + + return 0; +} +ALLOW_ERROR_INJECTION(xe_tuning_init, ERRNO); /* See xe_pci_probe() */ + void xe_tuning_process_gt(struct xe_gt *gt) { struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(gt); - xe_rtp_process_to_sr(&ctx, gt_tunings, >->reg_sr); + xe_rtp_process_ctx_enable_active_tracking(&ctx, + gt->tuning_active.gt, + ARRAY_SIZE(gt_tunings)); + xe_rtp_process_to_sr(&ctx, gt_tunings, ARRAY_SIZE(gt_tunings), >->reg_sr); } EXPORT_SYMBOL_IF_KUNIT(xe_tuning_process_gt); @@ -147,7 +178,11 @@ void xe_tuning_process_engine(struct xe_hw_engine *hwe) { struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(hwe); - xe_rtp_process_to_sr(&ctx, engine_tunings, &hwe->reg_sr); + xe_rtp_process_ctx_enable_active_tracking(&ctx, + hwe->gt->tuning_active.engine, + ARRAY_SIZE(engine_tunings)); + xe_rtp_process_to_sr(&ctx, engine_tunings, ARRAY_SIZE(engine_tunings), + &hwe->reg_sr); } EXPORT_SYMBOL_IF_KUNIT(xe_tuning_process_engine); @@ -163,5 +198,25 @@ void xe_tuning_process_lrc(struct xe_hw_engine *hwe) { struct xe_rtp_process_ctx ctx = XE_RTP_PROCESS_CTX_INITIALIZER(hwe); - xe_rtp_process_to_sr(&ctx, lrc_tunings, &hwe->reg_lrc); + xe_rtp_process_ctx_enable_active_tracking(&ctx, + hwe->gt->tuning_active.lrc, + ARRAY_SIZE(lrc_tunings)); + xe_rtp_process_to_sr(&ctx, lrc_tunings, ARRAY_SIZE(lrc_tunings), &hwe->reg_lrc); +} + +void xe_tuning_dump(struct xe_gt *gt, struct drm_printer *p) +{ + size_t idx; + + drm_printf(p, "GT Tunings\n"); + for_each_set_bit(idx, gt->tuning_active.gt, ARRAY_SIZE(gt_tunings)) + drm_printf_indent(p, 1, "%s\n", gt_tunings[idx].name); + + drm_printf(p, "\nEngine Tunings\n"); + for_each_set_bit(idx, gt->tuning_active.engine, ARRAY_SIZE(engine_tunings)) + drm_printf_indent(p, 1, "%s\n", engine_tunings[idx].name); + + drm_printf(p, "\nLRC Tunings\n"); + for_each_set_bit(idx, gt->tuning_active.lrc, ARRAY_SIZE(lrc_tunings)) + drm_printf_indent(p, 1, "%s\n", lrc_tunings[idx].name); } diff --git a/drivers/gpu/drm/xe/xe_tuning.h b/drivers/gpu/drm/xe/xe_tuning.h index 4f9c3ac3b516..dd0d3ccc9c65 100644 --- a/drivers/gpu/drm/xe/xe_tuning.h +++ b/drivers/gpu/drm/xe/xe_tuning.h @@ -6,11 +6,14 @@ #ifndef _XE_TUNING_ #define _XE_TUNING_ +struct drm_printer; struct xe_gt; struct xe_hw_engine; +int xe_tuning_init(struct xe_gt *gt); void xe_tuning_process_gt(struct xe_gt *gt); void xe_tuning_process_engine(struct xe_hw_engine *hwe); void xe_tuning_process_lrc(struct xe_hw_engine *hwe); +void xe_tuning_dump(struct xe_gt *gt, struct drm_printer *p); #endif diff --git a/drivers/gpu/drm/xe/xe_uc.c b/drivers/gpu/drm/xe/xe_uc.c index 0d073a9987c2..bb03c524613f 100644 --- a/drivers/gpu/drm/xe/xe_uc.c +++ b/drivers/gpu/drm/xe/xe_uc.c @@ -241,7 +241,7 @@ void xe_uc_gucrc_disable(struct xe_uc *uc) void xe_uc_stop_prepare(struct xe_uc *uc) { - xe_gsc_wait_for_worker_completion(&uc->gsc); + xe_gsc_stop_prepare(&uc->gsc); xe_guc_stop_prepare(&uc->guc); } @@ -275,6 +275,12 @@ again: goto again; } +void xe_uc_suspend_prepare(struct xe_uc *uc) +{ + xe_gsc_wait_for_worker_completion(&uc->gsc); + xe_guc_stop_prepare(&uc->guc); +} + int xe_uc_suspend(struct xe_uc *uc) { /* GuC submission not enabled, nothing to do */ diff --git a/drivers/gpu/drm/xe/xe_uc.h b/drivers/gpu/drm/xe/xe_uc.h index 506517c11333..ba2937ab94cf 100644 --- a/drivers/gpu/drm/xe/xe_uc.h +++ b/drivers/gpu/drm/xe/xe_uc.h @@ -18,6 +18,7 @@ int xe_uc_reset_prepare(struct xe_uc *uc); void xe_uc_stop_prepare(struct xe_uc *uc); void xe_uc_stop(struct xe_uc *uc); int xe_uc_start(struct xe_uc *uc); +void xe_uc_suspend_prepare(struct xe_uc *uc); int xe_uc_suspend(struct xe_uc *uc); int xe_uc_sanitize_reset(struct xe_uc *uc); void xe_uc_remove(struct xe_uc *uc); diff --git a/drivers/gpu/drm/xe/xe_vm.c b/drivers/gpu/drm/xe/xe_vm.c index 5956631c0d40..785b8960050b 100644 --- a/drivers/gpu/drm/xe/xe_vm.c +++ b/drivers/gpu/drm/xe/xe_vm.c @@ -8,6 +8,7 @@ #include <linux/dma-fence-array.h> #include <linux/nospec.h> +#include <drm/drm_drv.h> #include <drm/drm_exec.h> #include <drm/drm_print.h> #include <drm/ttm/ttm_tt.h> @@ -1582,9 +1583,40 @@ err_no_resv: static void xe_vm_close(struct xe_vm *vm) { + struct xe_device *xe = vm->xe; + bool bound; + int idx; + + bound = drm_dev_enter(&xe->drm, &idx); + down_write(&vm->lock); + vm->size = 0; + + if (!((vm->flags & XE_VM_FLAG_MIGRATION))) { + struct xe_tile *tile; + struct xe_gt *gt; + u8 id; + + /* Wait for pending binds */ + dma_resv_wait_timeout(xe_vm_resv(vm), + DMA_RESV_USAGE_BOOKKEEP, + false, MAX_SCHEDULE_TIMEOUT); + + if (bound) { + for_each_tile(tile, xe, id) + if (vm->pt_root[id]) + xe_pt_clear(xe, vm->pt_root[id]); + + for_each_gt(gt, xe, id) + xe_gt_tlb_invalidation_vm(gt, vm); + } + } + up_write(&vm->lock); + + if (bound) + drm_dev_exit(idx); } void xe_vm_close_and_put(struct xe_vm *vm) diff --git a/drivers/gpu/drm/xe/xe_wa.c b/drivers/gpu/drm/xe/xe_wa.c index 2553accf8c51..56257430b364 100644 --- a/drivers/gpu/drm/xe/xe_wa.c +++ b/drivers/gpu/drm/xe/xe_wa.c @@ -279,8 +279,6 @@ static const struct xe_rtp_entry_sr gt_was[] = { XE_RTP_ACTIONS(SET(VDBOX_CGCTL3F10(0), RAMDFTUNIT_CLKGATE_DIS)), XE_RTP_ENTRY_FLAG(FOREACH_ENGINE), }, - - {} }; static const struct xe_rtp_entry_sr engine_was[] = { @@ -613,8 +611,16 @@ static const struct xe_rtp_entry_sr engine_was[] = { XE_RTP_ACTIONS(FIELD_SET(SAMPLER_MODE, SMP_WAIT_FETCH_MERGING_COUNTER, SMP_FORCE_128B_OVERFETCH)) }, - - {} + { XE_RTP_NAME("14023061436"), + XE_RTP_RULES(GRAPHICS_VERSION_RANGE(3000, 3001), + FUNC(xe_rtp_match_first_render_or_compute)), + XE_RTP_ACTIONS(SET(TDL_CHICKEN, QID_WAIT_FOR_THREAD_NOT_RUN_DISABLE)) + }, + { XE_RTP_NAME("13012615864"), + XE_RTP_RULES(GRAPHICS_VERSION_RANGE(3000, 3001), + FUNC(xe_rtp_match_first_render_or_compute)), + XE_RTP_ACTIONS(SET(TDL_TSL_CHICKEN, RES_CHK_SPR_DIS)) + }, }; static const struct xe_rtp_entry_sr lrc_was[] = { @@ -795,6 +801,10 @@ static const struct xe_rtp_entry_sr lrc_was[] = { XE_RTP_RULES(GRAPHICS_VERSION(2001), ENGINE_CLASS(RENDER)), XE_RTP_ACTIONS(SET(CHICKEN_RASTER_1, DIS_CLIP_NEGATIVE_BOUNDING_BOX)) }, + { XE_RTP_NAME("22021007897"), + XE_RTP_RULES(GRAPHICS_VERSION(2001), ENGINE_CLASS(RENDER)), + XE_RTP_ACTIONS(SET(COMMON_SLICE_CHICKEN4, SBE_PUSH_CONSTANT_BEHIND_FIX_ENABLE)) + }, /* Xe3_LPG */ { XE_RTP_NAME("14021490052"), @@ -807,8 +817,6 @@ static const struct xe_rtp_entry_sr lrc_was[] = { DIS_PARTIAL_AUTOSTRIP | DIS_AUTOSTRIP)) }, - - {} }; static __maybe_unused const struct xe_rtp_entry oob_was[] = { @@ -850,7 +858,7 @@ void xe_wa_process_gt(struct xe_gt *gt) xe_rtp_process_ctx_enable_active_tracking(&ctx, gt->wa_active.gt, ARRAY_SIZE(gt_was)); - xe_rtp_process_to_sr(&ctx, gt_was, >->reg_sr); + xe_rtp_process_to_sr(&ctx, gt_was, ARRAY_SIZE(gt_was), >->reg_sr); } EXPORT_SYMBOL_IF_KUNIT(xe_wa_process_gt); @@ -868,7 +876,7 @@ void xe_wa_process_engine(struct xe_hw_engine *hwe) xe_rtp_process_ctx_enable_active_tracking(&ctx, hwe->gt->wa_active.engine, ARRAY_SIZE(engine_was)); - xe_rtp_process_to_sr(&ctx, engine_was, &hwe->reg_sr); + xe_rtp_process_to_sr(&ctx, engine_was, ARRAY_SIZE(engine_was), &hwe->reg_sr); } /** @@ -885,7 +893,7 @@ void xe_wa_process_lrc(struct xe_hw_engine *hwe) xe_rtp_process_ctx_enable_active_tracking(&ctx, hwe->gt->wa_active.lrc, ARRAY_SIZE(lrc_was)); - xe_rtp_process_to_sr(&ctx, lrc_was, &hwe->reg_lrc); + xe_rtp_process_to_sr(&ctx, lrc_was, ARRAY_SIZE(lrc_was), &hwe->reg_lrc); } /** diff --git a/drivers/gpu/drm/xe/xe_wa_oob.rules b/drivers/gpu/drm/xe/xe_wa_oob.rules index 40438c3d9b72..32d3853b08ec 100644 --- a/drivers/gpu/drm/xe/xe_wa_oob.rules +++ b/drivers/gpu/drm/xe/xe_wa_oob.rules @@ -30,8 +30,10 @@ 13011645652 GRAPHICS_VERSION(2004) 14022293748 GRAPHICS_VERSION(2001) GRAPHICS_VERSION(2004) + GRAPHICS_VERSION_RANGE(3000, 3001) 22019794406 GRAPHICS_VERSION(2001) GRAPHICS_VERSION(2004) + GRAPHICS_VERSION_RANGE(3000, 3001) 22019338487 MEDIA_VERSION(2000) GRAPHICS_VERSION(2001) MEDIA_VERSION(3000), MEDIA_STEP(A0, B0), FUNC(xe_rtp_match_not_sriov_vf) |