Merge tag 'amd-drm-next-6.5-2023-06-16' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.5-2023-06-16: amdgpu: - Misc display fixes - W=1 fixes - Improve scheduler naming - DCN 3.1.4 fixes - kdoc fixes - Enable W=1 - VCN 4.0 fix - xgmi fixes - TOPDOWN fix for large BAR systems - eDP fix - PSR fixes - SubVP fixes - Freesync fix - DPIA fix - SMU 13.0.5 fixes - vblflash fix - RAS fixes - SDMA 4 fix - BO locking fix - BO backing store fix - NBIO 7.9 fixes - GC 9.4.3 fixes - GPU reset recovery fixes - HMM fix amdkfd: - Fix NULL check - Trap fixes - Queue count fix - Add event age tracking radeon: - fbdev client fix scheduler: - Avoid an infinite loop UAPI: - Add KFD event age tracking: Proposed ROCT-Thunk-Interface: efdbf6cfbc 1820ae0a2d Proposed ROCR-Runtime: https://github.com/RadeonOpenCompute/ROCR-Runtime/compare/master...zhums:ROCR-Runtime:new_event_wait_review e1f5bdb88e 7d26afd141 drm: - DP MST fix Signed-off-by: Dave Airlie <airlied@redhat.com> From: Alex Deucher <alexander.deucher@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/20230616163548.7706-1-alexander.deucher@amd.com
2025-01-07 21:53:44 +00:00 · 2023-06-19 11:57:25 +10:00 · 2023-06-19 11:57:25 +10:00 · bcbede6fbe
commit bcbede6fbe
parent 4e237d84ee 72f1de49ff
93 changed files with 1221 additions and 430 deletions
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@ -40,7 +40,18 @@ ccflags-y := -I$(FULL_AMD_PATH)/include/asic_reg \
 	-I$(FULL_AMD_PATH)/amdkfd

 subdir-ccflags-y := -Wextra
-subdir-ccflags-y += $(call cc-option, -Wunused-but-set-variable)
+subdir-ccflags-y += -Wunused
+subdir-ccflags-y += -Wmissing-prototypes
+subdir-ccflags-y += -Wmissing-declarations
+subdir-ccflags-y += -Wmissing-include-dirs
+subdir-ccflags-y += -Wold-style-definition
+subdir-ccflags-y += -Wmissing-format-attribute
+# Need this to avoid recursive variable evaluation issues
+cond-flags := $(call cc-option, -Wunused-but-set-variable) \
+	$(call cc-option, -Wunused-const-variable) \
+	$(call cc-option, -Wstringop-truncation) \
+	$(call cc-option, -Wpacked-not-aligned)
+subdir-ccflags-y += $(cond-flags)
 subdir-ccflags-y += -Wno-unused-parameter
 subdir-ccflags-y += -Wno-type-limits
 subdir-ccflags-y += -Wno-sign-compare
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@ -309,7 +309,7 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 	}
 	p->gang_leader = p->jobs[p->gang_leader_idx];

-	if (p->ctx->vram_lost_counter != p->gang_leader->vram_lost_counter) {
+	if (p->ctx->generation != p->gang_leader->generation) {
 		ret = -ECANCELED;
 		goto free_all_kdata;
 	}
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@ -333,7 +333,7 @@ static int amdgpu_ctx_init(struct amdgpu_ctx_mgr *mgr, int32_t priority,

 	ctx->reset_counter = atomic_read(&mgr->adev->gpu_reset_counter);
 	ctx->reset_counter_query = ctx->reset_counter;
-	ctx->vram_lost_counter = atomic_read(&mgr->adev->vram_lost_counter);
+	ctx->generation = amdgpu_vm_generation(mgr->adev, &fpriv->vm);
 	ctx->init_priority = priority;
 	ctx->override_priority = AMDGPU_CTX_PRIORITY_UNSET;

@ -432,6 +432,7 @@ int amdgpu_ctx_get_entity(struct amdgpu_ctx *ctx, u32 hw_ip, u32 instance,
 			  u32 ring, struct drm_sched_entity **entity)
 {
 	int r;
+	struct drm_sched_entity *ctx_entity;

 	if (hw_ip >= AMDGPU_HW_IP_NUM) {
 		DRM_ERROR("unknown HW IP type: %d\n", hw_ip);
@ -455,7 +456,14 @@ int amdgpu_ctx_get_entity(struct amdgpu_ctx *ctx, u32 hw_ip, u32 instance,
 			return r;
 	}

-	*entity = &ctx->entities[hw_ip][ring]->entity;
+	ctx_entity = &ctx->entities[hw_ip][ring]->entity;
+	r = drm_sched_entity_error(ctx_entity);
+	if (r) {
+		DRM_DEBUG("error entity %p\n", ctx_entity);
+		return r;
+	}
+
+	*entity = ctx_entity;
 	return 0;
 }

@ -586,7 +594,7 @@ static int amdgpu_ctx_query2(struct amdgpu_device *adev,
 	if (ctx->reset_counter != atomic_read(&adev->gpu_reset_counter))
 		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_RESET;

-	if (ctx->vram_lost_counter != atomic_read(&adev->vram_lost_counter))
+	if (ctx->generation != amdgpu_vm_generation(adev, &fpriv->vm))
 		out->state.flags |= AMDGPU_CTX_QUERY2_FLAGS_VRAMLOST;

 	if (atomic_read(&ctx->guilty))
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.h
@ -47,7 +47,7 @@ struct amdgpu_ctx {
 	struct amdgpu_ctx_mgr		*mgr;
 	unsigned			reset_counter;
 	unsigned			reset_counter_query;
-	uint32_t			vram_lost_counter;
+	uint64_t			generation;
 	spinlock_t			ring_lock;
 	struct amdgpu_ctx_entity	*entities[AMDGPU_HW_IP_NUM][AMDGPU_MAX_ENTITY_NUM];
 	bool				preamble_presented;
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_discovery.c
@ -1515,6 +1515,7 @@ static int amdgpu_discovery_get_mall_info(struct amdgpu_device *adev)
 				mall_size += mall_size_per_umc;
 		}
 		adev->gmc.mall_size = mall_size;
+		adev->gmc.m_half_use = half_use;
 		break;
 	default:
 		dev_err(adev->dev,
@ -1896,6 +1897,8 @@ static int amdgpu_discovery_set_gc_ip_blocks(struct amdgpu_device *adev)
 		amdgpu_device_ip_block_add(adev, &gfx_v9_0_ip_block);
 		break;
 	case IP_VERSION(9, 4, 3):
+		if (!amdgpu_exp_hw_support)
+			return -EINVAL;
 		amdgpu_device_ip_block_add(adev, &gfx_v9_4_3_ip_block);
 		break;
 	case IP_VERSION(10, 1, 10):
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_fence.c
@ -691,6 +691,30 @@ void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring)
 	}
 }

+/**
+ * amdgpu_fence_driver_set_error - set error code on fences
+ * @ring: the ring which contains the fences
+ * @error: the error code to set
+ *
+ * Set an error code to all the fences pending on the ring.
+ */
+void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error)
+{
+	struct amdgpu_fence_driver *drv = &ring->fence_drv;
+	unsigned long flags;
+
+	spin_lock_irqsave(&drv->lock, flags);
+	for (unsigned int i = 0; i <= drv->num_fences_mask; ++i) {
+		struct dma_fence *fence;
+
+		fence = rcu_dereference_protected(drv->fences[i],
+						  lockdep_is_held(&drv->lock));
+		if (fence && !dma_fence_is_signaled_locked(fence))
+			dma_fence_set_error(fence, error);
+	}
+	spin_unlock_irqrestore(&drv->lock, flags);
+}
+
 /**
 * amdgpu_fence_driver_force_completion - force signal latest fence of ring
 *
@ -699,6 +723,7 @@ void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring)
 */
 void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring)
 {
+	amdgpu_fence_driver_set_error(ring, -ECANCELED);
 	amdgpu_fence_write(ring, ring->fence_drv.sync_seq);
 	amdgpu_fence_process(ring);
 }
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@ -301,6 +301,8 @@ struct amdgpu_gmc {

 	/* MALL size */
 	u64 mall_size;
+	uint32_t m_half_use;
+
 	/* number of UMC instances */
 	int num_umc;
 	/* mode2 save restore */
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_hmm.c
@ -190,8 +190,8 @@ int amdgpu_hmm_range_get_pages(struct mmu_interval_notifier *notifier,
 		pr_debug("hmm range: start = 0x%lx, end = 0x%lx",
 			hmm_range->start, hmm_range->end);

-		/* Assuming 512MB takes maxmium 1 second to fault page address */
-		timeout = max((hmm_range->end - hmm_range->start) >> 29, 1UL);
+		/* Assuming 128MB takes maximum 1 second to fault page address */
+		timeout = max((hmm_range->end - hmm_range->start) >> 27, 1UL);
 		timeout *= HMM_RANGE_DEFAULT_TIMEOUT;
 		timeout = jiffies + msecs_to_jiffies(timeout);

--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@ -467,7 +467,8 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
 			handled = true;

 	} else {
-		DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
+		DRM_DEBUG("Unregistered interrupt src_id: %d of client_id:%d\n",
+			src_id, client_id);
 	}

 	/* Send it to amdkfd as well if it isn't already handled */
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@ -109,7 +109,7 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	(*job)->vm = vm;

 	amdgpu_sync_create(&(*job)->explicit_sync);
-	(*job)->vram_lost_counter = atomic_read(&adev->vram_lost_counter);
+	(*job)->generation = amdgpu_vm_generation(adev, vm);
 	(*job)->vm_pd_addr = AMDGPU_BO_INVALID_OFFSET;

 	if (!entity)
@ -258,16 +258,27 @@ amdgpu_job_prepare_job(struct drm_sched_job *sched_job,
 	struct dma_fence *fence = NULL;
 	int r;

+	/* Ignore soft recovered fences here */
+	r = drm_sched_entity_error(s_entity);
+	if (r && r != -ENODATA)
+		goto error;
+
 	if (!fence && job->gang_submit)
 		fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit);

 	while (!fence && job->vm && !job->vmid) {
 		r = amdgpu_vmid_grab(job->vm, ring, job, &fence);
-		if (r)
+		if (r) {
 			DRM_ERROR("Error getting VM ID (%d)\n", r);
+			goto error;
+		}
 	}

 	return fence;
+
+error:
+	dma_fence_set_error(&job->base.s_fence->finished, r);
+	return NULL;
 }

 static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
@ -284,7 +295,7 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 	trace_amdgpu_sched_run_job(job);

 	/* Skip job if VRAM is lost and never resubmit gangs */
-	if (job->vram_lost_counter != atomic_read(&adev->vram_lost_counter) ||
+	if (job->generation != amdgpu_vm_generation(adev, job->vm) ||
 	    (job->job_run_counter && job->gang_submit))
 		dma_fence_set_error(finished, -ECANCELED);

--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@ -61,7 +61,7 @@ struct amdgpu_job {
 	uint32_t		gds_base, gds_size;
 	uint32_t		gws_base, gws_size;
 	uint32_t		oa_base, oa_size;
-	uint32_t		vram_lost_counter;
+	uint64_t		generation;

 	/* user fence handling */
 	uint64_t		uf_addr;
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_nbio.h
@ -99,8 +99,6 @@ struct amdgpu_nbio_funcs {
 	int (*get_compute_partition_mode)(struct amdgpu_device *adev);
 	u32 (*get_memory_partition_mode)(struct amdgpu_device *adev,
 					 u32 *supp_modes);
-	void (*set_compute_partition_mode)(struct amdgpu_device *adev,
-					   enum amdgpu_gfx_partition mode);
 };

 struct amdgpu_nbio {
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@ -150,7 +150,7 @@ void amdgpu_bo_placement_from_domain(struct amdgpu_bo *abo, u32 domain)

 		if (flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
 			places[c].lpfn = min_not_zero(places[c].lpfn, visible_pfn);
-		else if (adev->gmc.real_vram_size != adev->gmc.visible_vram_size)
+		else
 			places[c].flags |= TTM_PL_FLAG_TOPDOWN;

 		if (flags & AMDGPU_GEM_CREATE_VRAM_CONTIGUOUS)
@ -1282,8 +1282,12 @@ void amdgpu_bo_move_notify(struct ttm_buffer_object *bo,
 void amdgpu_bo_get_memory(struct amdgpu_bo *bo,
 			  struct amdgpu_mem_stats *stats)
 {
-	unsigned int domain;
 	uint64_t size = amdgpu_bo_size(bo);
+	unsigned int domain;
+
+	/* Abort if the BO doesn't currently have a backing store */
+	if (!bo->tbo.resource)
+		return;

 	domain = amdgpu_mem_type_to_domain(bo->tbo.resource->mem_type);
 	switch (domain) {
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@ -569,6 +569,26 @@ int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
 	return -ETIME;
 }

+int psp_wait_for_spirom_update(struct psp_context *psp, uint32_t reg_index,
+			       uint32_t reg_val, uint32_t mask, uint32_t msec_timeout)
+{
+	uint32_t val;
+	int i;
+	struct amdgpu_device *adev = psp->adev;
+
+	if (psp->adev->no_hw_access)
+		return 0;
+
+	for (i = 0; i < msec_timeout; i++) {
+		val = RREG32(reg_index);
+		if ((val & mask) == reg_val)
+			return 0;
+		msleep(1);
+	}
+
+	return -ETIME;
+}
+
 static const char *psp_gfx_cmd_name(enum psp_gfx_cmd_id cmd_id)
 {
 	switch (cmd_id) {
@ -1653,10 +1673,11 @@ int psp_ras_initialize(struct psp_context *psp)

 	if (amdgpu_ras_is_poison_mode_supported(adev))
 		ras_cmd->ras_in_message.init_flags.poison_mode_en = 1;
-	if (!adev->gmc.xgmi.connected_to_cpu)
+	if (!adev->gmc.xgmi.connected_to_cpu && !adev->gmc.is_app_apu)
 		ras_cmd->ras_in_message.init_flags.dgpu_mode = 1;
 	ras_cmd->ras_in_message.init_flags.xcc_mask =
 		adev->gfx.xcc_mask;
+	ras_cmd->ras_in_message.init_flags.channel_dis_num = hweight32(adev->gmc.m_half_use) * 2;

 	ret = psp_ta_load(psp, &psp->ras_context.context);

--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
@ -455,6 +455,8 @@ extern const struct amdgpu_ip_block_version psp_v13_0_4_ip_block;

 extern int psp_wait_for(struct psp_context *psp, uint32_t reg_index,
 			uint32_t field_val, uint32_t mask, bool check_changed);
+extern int psp_wait_for_spirom_update(struct psp_context *psp, uint32_t reg_index,
+			uint32_t field_val, uint32_t mask, uint32_t msec_timeout);

 int psp_gpu_reset(struct amdgpu_device *adev);
 int psp_update_vcn_sram(struct amdgpu_device *adev, int inst_idx,
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@ -171,8 +171,7 @@ static int amdgpu_reserve_page_direct(struct amdgpu_device *adev, uint64_t addre

 	memset(&err_rec, 0x0, sizeof(struct eeprom_table_record));
 	err_data.err_addr = &err_rec;
-	amdgpu_umc_fill_error_record(&err_data, address,
-			(address >> AMDGPU_GPU_PAGE_SHIFT), 0, 0);
+	amdgpu_umc_fill_error_record(&err_data, address, address, 0, 0);

 	if (amdgpu_bad_page_threshold != 0) {
 		amdgpu_ras_add_bad_pages(adev, err_data.err_addr,
@ -490,7 +489,8 @@ static ssize_t amdgpu_ras_debugfs_ctrl_write(struct file *f,
 		ret = amdgpu_ras_feature_enable(adev, &data.head, 1);
 		break;
 	case 2:
-		if ((data.inject.address >= adev->gmc.mc_vram_size) ||
+		if ((data.inject.address >= adev->gmc.mc_vram_size &&
+		    adev->gmc.mc_vram_size) ||
 		    (data.inject.address >= RAS_UMC_INJECT_ADDR_LIMIT)) {
 			dev_warn(adev->dev, "RAS WARN: input address "
 					"0x%llx is invalid.",
@ -1491,6 +1491,7 @@ static int amdgpu_ras_sysfs_remove_all(struct amdgpu_device *adev)
 static struct dentry *amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *adev)
 {
 	struct amdgpu_ras *con = amdgpu_ras_get_context(adev);
+	struct amdgpu_ras_eeprom_control *eeprom = &con->eeprom_control;
 	struct drm_minor  *minor = adev_to_drm(adev)->primary;
 	struct dentry     *dir;

@ -1501,6 +1502,7 @@ static struct dentry *amdgpu_ras_debugfs_create_ctrl_node(struct amdgpu_device *
 			    &amdgpu_ras_debugfs_eeprom_ops);
 	debugfs_create_u32("bad_page_cnt_threshold", 0444, dir,
 			   &con->bad_page_cnt_threshold);
+	debugfs_create_u32("ras_num_recs", 0444, dir, &eeprom->ras_num_recs);
 	debugfs_create_x32("ras_hw_enabled", 0444, dir, &adev->ras_hw_enabled);
 	debugfs_create_x32("ras_enabled", 0444, dir, &adev->ras_enabled);
 	debugfs_create_file("ras_eeprom_size", S_IRUGO, dir, adev,
@ -1685,8 +1687,7 @@ static void amdgpu_ras_interrupt_poison_consumption_handler(struct ras_manager *
 		}
 	}

-	if (!adev->gmc.xgmi.connected_to_cpu)
-		amdgpu_umc_poison_handler(adev, false);
+	amdgpu_umc_poison_handler(adev, false);

 	if (block_obj->hw_ops && block_obj->hw_ops->handle_poison_consumption)
 		poison_stat = block_obj->hw_ops->handle_poison_consumption(adev);
@ -2451,11 +2452,10 @@ static void amdgpu_ras_check_supported(struct amdgpu_device *adev)
 {
 	adev->ras_hw_enabled = adev->ras_enabled = 0;

-	if (!adev->is_atom_fw ||
-	    !amdgpu_ras_asic_supported(adev))
+	if (!amdgpu_ras_asic_supported(adev))
 		return;

-	if (!adev->gmc.xgmi.connected_to_cpu) {
+	if (!adev->gmc.xgmi.connected_to_cpu &&	!adev->gmc.is_app_apu) {
 		if (amdgpu_atomfirmware_mem_ecc_supported(adev)) {
 			dev_info(adev->dev, "MEM ECC is active.\n");
 			adev->ras_hw_enabled |= (1 << AMDGPU_RAS_BLOCK__UMC |
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras_eeprom.c
@ -72,7 +72,7 @@
 /* Bad GPU tag ‘BADG’ */
 #define RAS_TABLE_HDR_BAD       0x42414447

-/**
+/*
 * EEPROM Table structure v1
 * ---------------------------------
 * |                               |
@ -94,7 +94,7 @@
 #define RAS_MAX_RECORD_COUNT    ((RAS_TBL_SIZE_BYTES - RAS_TABLE_HEADER_SIZE) \
 				 / RAS_TABLE_RECORD_SIZE)

-/**
+/*
 * EEPROM Table structrue v2.1
 * ---------------------------------
 * |                               |
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@ -433,11 +433,18 @@ void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
 bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
 			       struct dma_fence *fence)
 {
+	unsigned long flags;
+
 	ktime_t deadline = ktime_add_us(ktime_get(), 10000);

 	if (amdgpu_sriov_vf(ring->adev) || !ring->funcs->soft_recovery || !fence)
 		return false;

+	spin_lock_irqsave(fence->lock, flags);
+	if (!dma_fence_is_signaled_locked(fence))
+		dma_fence_set_error(fence, -ENODATA);
+	spin_unlock_irqrestore(fence->lock, flags);
+
 	atomic_inc(&ring->adev->gpu_reset_counter);
 	while (!dma_fence_is_signaled(fence) &&
 	       ktime_to_ns(ktime_sub(deadline, ktime_get())) > 0)
@ -561,6 +568,17 @@ static const struct file_operations amdgpu_debugfs_mqd_fops = {
 	.llseek = default_llseek
 };

+static int amdgpu_debugfs_ring_error(void *data, u64 val)
+{
+	struct amdgpu_ring *ring = data;
+
+	amdgpu_fence_driver_set_error(ring, val);
+	return 0;
+}
+
+DEFINE_DEBUGFS_ATTRIBUTE_SIGNED(amdgpu_debugfs_error_fops, NULL,
+				amdgpu_debugfs_ring_error, "%lld\n");
+
 #endif

 void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
@ -582,6 +600,11 @@ void amdgpu_debugfs_ring_init(struct amdgpu_device *adev,
 					 &amdgpu_debugfs_mqd_fops,
 					 ring->mqd_size);
 	}
+
+	sprintf(name, "amdgpu_error_%s", ring->name);
+	debugfs_create_file(name, 0200, root, ring,
+			    &amdgpu_debugfs_error_fops);
+
 #endif
 }

--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
@ -126,6 +126,7 @@ struct amdgpu_fence_driver {
 extern const struct drm_sched_backend_ops amdgpu_sched_ops;

 void amdgpu_fence_driver_clear_job_fences(struct amdgpu_ring *ring);
+void amdgpu_fence_driver_set_error(struct amdgpu_ring *ring, int error);
 void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);

 int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring);
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@ -228,7 +228,7 @@ static int amdgpu_ttm_map_buffer(struct ttm_buffer_object *bo,
 	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
 	num_bytes = num_pages * 8 * AMDGPU_GPU_PAGES_IN_CPU_PAGE;

-	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.entity,
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4 + num_bytes,
 				     AMDGPU_IB_POOL_DELAYED, &job);
@ -1456,7 +1456,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 		memcpy(adev->mman.sdma_access_ptr, buf, len);

 	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
-	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.entity,
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4, AMDGPU_IB_POOL_DELAYED,
 				     &job);
@ -2032,7 +2032,7 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)

 		ring = adev->mman.buffer_funcs_ring;
 		sched = &ring->sched;
-		r = drm_sched_entity_init(&adev->mman.entity,
+		r = drm_sched_entity_init(&adev->mman.high_pr,
 					  DRM_SCHED_PRIORITY_KERNEL, &sched,
 					  1, NULL);
 		if (r) {
@ -2041,7 +2041,7 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 			return;
 		}

-		r = drm_sched_entity_init(&adev->mman.delayed,
+		r = drm_sched_entity_init(&adev->mman.low_pr,
 					  DRM_SCHED_PRIORITY_NORMAL, &sched,
 					  1, NULL);
 		if (r) {
@ -2050,8 +2050,8 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 			goto error_free_entity;
 		}
 	} else {
-		drm_sched_entity_destroy(&adev->mman.entity);
-		drm_sched_entity_destroy(&adev->mman.delayed);
+		drm_sched_entity_destroy(&adev->mman.high_pr);
+		drm_sched_entity_destroy(&adev->mman.low_pr);
 		dma_fence_put(man->move);
 		man->move = NULL;
 	}
@ -2067,7 +2067,7 @@ void amdgpu_ttm_set_buffer_funcs_status(struct amdgpu_device *adev, bool enable)
 	return;

 error_free_entity:
-	drm_sched_entity_destroy(&adev->mman.entity);
+	drm_sched_entity_destroy(&adev->mman.high_pr);
 }

 static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
@ -2082,8 +2082,8 @@ static int amdgpu_ttm_prepare_job(struct amdgpu_device *adev,
 		AMDGPU_IB_POOL_DIRECT :
 		AMDGPU_IB_POOL_DELAYED;
 	int r;
-	struct drm_sched_entity *entity = delayed ? &adev->mman.delayed :
-						    &adev->mman.entity;
+	struct drm_sched_entity *entity = delayed ? &adev->mman.low_pr :
+						    &adev->mman.high_pr;
 	r = amdgpu_job_alloc_with_ib(adev, entity,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4, pool, job);
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.h
@ -59,10 +59,10 @@ struct amdgpu_mman {
 	bool					buffer_funcs_enabled;

 	struct mutex				gtt_window_lock;
-	/* Scheduler entity for buffer moves */
-	struct drm_sched_entity			entity;
-	/* Scheduler entity for VRAM clearing */
-	struct drm_sched_entity			delayed;
+	/* High priority scheduler entity for buffer moves */
+	struct drm_sched_entity			high_pr;
+	/* Low priority scheduler entity for VRAM clearing */
+	struct drm_sched_entity			low_pr;

 	struct amdgpu_vram_mgr vram_mgr;
 	struct amdgpu_gtt_mgr gtt_mgr;
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_umc.c
@ -169,27 +169,31 @@ int amdgpu_umc_poison_handler(struct amdgpu_device *adev, bool reset)
 {
 	int ret = AMDGPU_RAS_SUCCESS;

-	if (!amdgpu_sriov_vf(adev)) {
-		if (!adev->gmc.xgmi.connected_to_cpu) {
-			struct ras_err_data err_data = {0, 0, 0, NULL};
-			struct ras_common_if head = {
-				.block = AMDGPU_RAS_BLOCK__UMC,
-			};
-			struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
-
-			ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
-
-			if (ret == AMDGPU_RAS_SUCCESS && obj) {
-				obj->err_data.ue_count += err_data.ue_count;
-				obj->err_data.ce_count += err_data.ce_count;
-			}
-		} else if (reset) {
+	if (adev->gmc.xgmi.connected_to_cpu ||
+		adev->gmc.is_app_apu) {
+		if (reset) {
 			/* MCA poison handler is only responsible for GPU reset,
 			 * let MCA notifier do page retirement.
 			 */
 			kgd2kfd_set_sram_ecc_flag(adev->kfd.dev);
 			amdgpu_ras_reset_gpu(adev);
 		}
+		return ret;
+	}
+
+	if (!amdgpu_sriov_vf(adev)) {
+		struct ras_err_data err_data = {0, 0, 0, NULL};
+		struct ras_common_if head = {
+			.block = AMDGPU_RAS_BLOCK__UMC,
+		};
+		struct ras_manager *obj = amdgpu_ras_find_obj(adev, &head);
+
+		ret = amdgpu_umc_do_page_retirement(adev, &err_data, NULL, reset);
+
+		if (ret == AMDGPU_RAS_SUCCESS && obj) {
+			obj->err_data.ue_count += err_data.ue_count;
+			obj->err_data.ce_count += err_data.ce_count;
+		}
 	} else {
 		if (adev->virt.ops && adev->virt.ops->ras_poison_handler)
 			adev->virt.ops->ras_poison_handler(adev);
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@ -266,6 +266,32 @@ static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
 	spin_unlock(&vm_bo->vm->status_lock);
 }

+/**
+ * amdgpu_vm_bo_reset_state_machine - reset the vm_bo state machine
+ * @vm: the VM which state machine to reset
+ *
+ * Move all vm_bo object in the VM into a state where they will be updated
+ * again during validation.
+ */
+static void amdgpu_vm_bo_reset_state_machine(struct amdgpu_vm *vm)
+{
+	struct amdgpu_vm_bo_base *vm_bo, *tmp;
+
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->done, &vm->invalidated);
+	list_for_each_entry(vm_bo, &vm->invalidated, vm_status)
+		vm_bo->moved = true;
+	list_for_each_entry_safe(vm_bo, tmp, &vm->idle, vm_status) {
+		struct amdgpu_bo *bo = vm_bo->bo;
+
+		if (!bo || bo->tbo.type != ttm_bo_type_kernel)
+			list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
+		else if (bo->parent)
+			list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
+	}
+	spin_unlock(&vm->status_lock);
+}
+
 /**
 * amdgpu_vm_bo_base_init - Adds bo to the list of bos associated with the vm
 *
@ -351,6 +377,58 @@ void amdgpu_vm_move_to_lru_tail(struct amdgpu_device *adev,
 	spin_unlock(&adev->mman.bdev.lru_lock);
 }

+/* Create scheduler entities for page table updates */
+static int amdgpu_vm_init_entities(struct amdgpu_device *adev,
+				   struct amdgpu_vm *vm)
+{
+	int r;
+
+	r = drm_sched_entity_init(&vm->immediate, DRM_SCHED_PRIORITY_NORMAL,
+				  adev->vm_manager.vm_pte_scheds,
+				  adev->vm_manager.vm_pte_num_scheds, NULL);
+	if (r)
+		goto error;
+
+	return drm_sched_entity_init(&vm->delayed, DRM_SCHED_PRIORITY_NORMAL,
+				     adev->vm_manager.vm_pte_scheds,
+				     adev->vm_manager.vm_pte_num_scheds, NULL);
+
+error:
+	drm_sched_entity_destroy(&vm->immediate);
+	return r;
+}
+
+/* Destroy the entities for page table updates again */
+static void amdgpu_vm_fini_entities(struct amdgpu_vm *vm)
+{
+	drm_sched_entity_destroy(&vm->immediate);
+	drm_sched_entity_destroy(&vm->delayed);
+}
+
+/**
+ * amdgpu_vm_generation - return the page table re-generation counter
+ * @adev: the amdgpu_device
+ * @vm: optional VM to check, might be NULL
+ *
+ * Returns a page table re-generation token to allow checking if submissions
+ * are still valid to use this VM. The VM parameter might be NULL in which case
+ * just the VRAM lost counter will be used.
+ */
+uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm)
+{
+	uint64_t result = (u64)atomic_read(&adev->vram_lost_counter) << 32;
+
+	if (!vm)
+		return result;
+
+	result += vm->generation;
+	/* Add one if the page tables will be re-generated on next CS */
+	if (drm_sched_entity_error(&vm->delayed))
+		++result;
+
+	return result;
+}
+
 /**
 * amdgpu_vm_validate_pt_bos - validate the page table BOs
 *
@ -373,6 +451,15 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	struct amdgpu_bo *bo;
 	int r;

+	if (drm_sched_entity_error(&vm->delayed)) {
+		++vm->generation;
+		amdgpu_vm_bo_reset_state_machine(vm);
+		amdgpu_vm_fini_entities(vm);
+		r = amdgpu_vm_init_entities(adev, vm);
+		if (r)
+			return r;
+	}
+
 	spin_lock(&vm->status_lock);
 	while (!list_empty(&vm->evicted)) {
 		bo_base = list_first_entry(&vm->evicted,
@ -920,42 +1007,51 @@ int amdgpu_vm_update_range(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 	return r;
 }

+static void amdgpu_vm_bo_get_memory(struct amdgpu_bo_va *bo_va,
+				    struct amdgpu_mem_stats *stats)
+{
+	struct amdgpu_vm *vm = bo_va->base.vm;
+	struct amdgpu_bo *bo = bo_va->base.bo;
+
+	if (!bo)
+		return;
+
+	/*
+	 * For now ignore BOs which are currently locked and potentially
+	 * changing their location.
+	 */
+	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv &&
+	    !dma_resv_trylock(bo->tbo.base.resv))
+		return;
+
+	amdgpu_bo_get_memory(bo, stats);
+	if (bo->tbo.base.resv != vm->root.bo->tbo.base.resv)
+	    dma_resv_unlock(bo->tbo.base.resv);
+}
+
 void amdgpu_vm_get_memory(struct amdgpu_vm *vm,
 			  struct amdgpu_mem_stats *stats)
 {
 	struct amdgpu_bo_va *bo_va, *tmp;

 	spin_lock(&vm->status_lock);
-	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
-	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status) {
-		if (!bo_va->base.bo)
-			continue;
-		amdgpu_bo_get_memory(bo_va->base.bo, stats);
-	}
+	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->evicted, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->relocated, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
+
+	list_for_each_entry_safe(bo_va, tmp, &vm->done, base.vm_status)
+		amdgpu_vm_bo_get_memory(bo_va, stats);
 	spin_unlock(&vm->status_lock);
 }

@ -2039,19 +2135,10 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	INIT_LIST_HEAD(&vm->pt_freed);
 	INIT_WORK(&vm->pt_free_work, amdgpu_vm_pt_free_work);

-	/* create scheduler entities for page table updates */
-	r = drm_sched_entity_init(&vm->immediate, DRM_SCHED_PRIORITY_NORMAL,
-				  adev->vm_manager.vm_pte_scheds,
-				  adev->vm_manager.vm_pte_num_scheds, NULL);
+	r = amdgpu_vm_init_entities(adev, vm);
 	if (r)
 		return r;

-	r = drm_sched_entity_init(&vm->delayed, DRM_SCHED_PRIORITY_NORMAL,
-				  adev->vm_manager.vm_pte_scheds,
-				  adev->vm_manager.vm_pte_num_scheds, NULL);
-	if (r)
-		goto error_free_immediate;
-
 	vm->pte_support_ats = false;
 	vm->is_compute_context = false;

@ -2072,6 +2159,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	vm->last_update = dma_fence_get_stub();
 	vm->last_unlocked = dma_fence_get_stub();
 	vm->last_tlb_flush = dma_fence_get_stub();
+	vm->generation = 0;

 	mutex_init(&vm->eviction_lock);
 	vm->evicting = false;
@ -2112,10 +2200,7 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 error_free_delayed:
 	dma_fence_put(vm->last_tlb_flush);
 	dma_fence_put(vm->last_unlocked);
-	drm_sched_entity_destroy(&vm->delayed);
-
-error_free_immediate:
-	drm_sched_entity_destroy(&vm->immediate);
+	amdgpu_vm_fini_entities(vm);

 	return r;
 }
@ -2268,8 +2353,7 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	amdgpu_bo_unref(&root);
 	WARN_ON(vm->root.bo);

-	drm_sched_entity_destroy(&vm->immediate);
-	drm_sched_entity_destroy(&vm->delayed);
+	amdgpu_vm_fini_entities(vm);

 	if (!RB_EMPTY_ROOT(&vm->va.rb_root)) {
 		dev_err(adev->dev, "still active bo inside vm\n");
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@ -295,6 +295,9 @@ struct amdgpu_vm {
 	atomic64_t		tlb_seq;
 	struct dma_fence	*last_tlb_flush;

+	/* How many times we had to re-generate the page tables */
+	uint64_t		generation;
+
 	/* Last unlocked submission to the scheduler entities */
 	struct dma_fence	*last_unlocked;

@ -397,6 +400,7 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
 			 struct list_head *validated,
 			 struct amdgpu_bo_list_entry *entry);
 bool amdgpu_vm_ready(struct amdgpu_vm *vm);
+uint64_t amdgpu_vm_generation(struct amdgpu_device *adev, struct amdgpu_vm *vm);
 int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			      int (*callback)(void *p, struct amdgpu_bo *bo),
 			      void *param);
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@ -325,6 +325,36 @@ static ssize_t amdgpu_xgmi_show_device_id(struct device *dev,

 }

+static ssize_t amdgpu_xgmi_show_num_hops(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i;
+
+	for (i = 0; i < top->num_nodes; i++)
+		sprintf(buf + 3 * i, "%02x ", top->nodes[i].num_hops);
+
+	return sysfs_emit(buf, "%s\n", buf);
+}
+
+static ssize_t amdgpu_xgmi_show_num_links(struct device *dev,
+					struct device_attribute *attr,
+					char *buf)
+{
+	struct drm_device *ddev = dev_get_drvdata(dev);
+	struct amdgpu_device *adev = drm_to_adev(ddev);
+	struct psp_xgmi_topology_info *top = &adev->psp.xgmi_context.top_info;
+	int i;
+
+	for (i = 0; i < top->num_nodes; i++)
+		sprintf(buf + 3 * i, "%02x ", top->nodes[i].num_links);
+
+	return sysfs_emit(buf, "%s\n", buf);
+}
+
 #define AMDGPU_XGMI_SET_FICAA(o)	((o) | 0x456801)
 static ssize_t amdgpu_xgmi_show_error(struct device *dev,
 				      struct device_attribute *attr,
@ -361,6 +391,8 @@ static ssize_t amdgpu_xgmi_show_error(struct device *dev,

 static DEVICE_ATTR(xgmi_device_id, S_IRUGO, amdgpu_xgmi_show_device_id, NULL);
 static DEVICE_ATTR(xgmi_error, S_IRUGO, amdgpu_xgmi_show_error, NULL);
+static DEVICE_ATTR(xgmi_num_hops, S_IRUGO, amdgpu_xgmi_show_num_hops, NULL);
+static DEVICE_ATTR(xgmi_num_links, S_IRUGO, amdgpu_xgmi_show_num_links, NULL);

 static int amdgpu_xgmi_sysfs_add_dev_info(struct amdgpu_device *adev,
 					 struct amdgpu_hive_info *hive)
@ -380,6 +412,15 @@ static int amdgpu_xgmi_sysfs_add_dev_info(struct amdgpu_device *adev,
 	if (ret)
 		pr_err("failed to create xgmi_error\n");

+	/* Create xgmi num hops file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_num_hops);
+	if (ret)
+		pr_err("failed to create xgmi_num_hops\n");
+
+	/* Create xgmi num links file */
+	ret = device_create_file(adev->dev, &dev_attr_xgmi_num_links);
+	if (ret)
+		pr_err("failed to create xgmi_num_links\n");

 	/* Create sysfs link to hive info folder on the first device */
 	if (hive->kobj.parent != (&adev->dev->kobj)) {
@ -407,6 +448,9 @@ static int amdgpu_xgmi_sysfs_add_dev_info(struct amdgpu_device *adev,

 remove_file:
 	device_remove_file(adev->dev, &dev_attr_xgmi_device_id);
+	device_remove_file(adev->dev, &dev_attr_xgmi_error);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_hops);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_links);

 success:
 	return ret;
@ -420,6 +464,8 @@ static void amdgpu_xgmi_sysfs_rem_dev_info(struct amdgpu_device *adev,

 	device_remove_file(adev->dev, &dev_attr_xgmi_device_id);
 	device_remove_file(adev->dev, &dev_attr_xgmi_error);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_hops);
+	device_remove_file(adev->dev, &dev_attr_xgmi_num_links);

 	if (hive->kobj.parent != (&adev->dev->kobj))
 		sysfs_remove_link(&adev->dev->kobj,"xgmi_hive_info");
--- a/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram_reg_init.c
+++ b/drivers/gpu/drm/amd/amdgpu/aqua_vanjaram_reg_init.c
@ -518,9 +518,6 @@ static int aqua_vanjaram_switch_partition_mode(struct amdgpu_xcp_mgr *xcp_mgr,
 		adev->gfx.funcs->switch_partition_mode(xcp_mgr->adev,
 						       num_xcc_per_xcp);

-	if (adev->nbio.funcs->set_compute_partition_mode)
-		adev->nbio.funcs->set_compute_partition_mode(adev, mode);
-
 	/* Init info about new xcps */
 	*num_xcps = num_xcc / num_xcc_per_xcp;
 	amdgpu_xcp_init(xcp_mgr, *num_xcps, mode);
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c
@ -623,22 +623,16 @@ static void gfx_v9_4_3_select_me_pipe_q(struct amdgpu_device *adev,
 static int gfx_v9_4_3_switch_compute_partition(struct amdgpu_device *adev,
 						int num_xccs_per_xcp)
 {
-	int i, num_xcc;
-	u32 tmp = 0;
+	int ret;

-	num_xcc = NUM_XCC(adev->gfx.xcc_mask);
-
-	for (i = 0; i < num_xcc; i++) {
-		tmp = REG_SET_FIELD(tmp, CP_HYP_XCP_CTL, NUM_XCC_IN_XCP,
-				    num_xccs_per_xcp);
-		tmp = REG_SET_FIELD(tmp, CP_HYP_XCP_CTL, VIRTUAL_XCC_ID,
-				    i % num_xccs_per_xcp);
-		WREG32_SOC15(GC, GET_INST(GC, i), regCP_HYP_XCP_CTL, tmp);
-	}
+	ret = psp_spatial_partition(&adev->psp, NUM_XCC(adev->gfx.xcc_mask) /
+							num_xccs_per_xcp);
+	if (ret)
+		return ret;

 	adev->gfx.num_xcc_per_xcp = num_xccs_per_xcp;

-	return 0;
+	return ret;
 }

 static int gfx_v9_4_3_ih_to_xcc_inst(struct amdgpu_device *adev, int ih_node)
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c
@ -382,7 +382,7 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	 * translation. Avoid this by doing the invalidation from the SDMA
 	 * itself.
 	 */
-	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->mman.entity,
+	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     16 * 4, AMDGPU_IB_POOL_IMMEDIATE,
 				     &job);
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@ -2107,6 +2107,7 @@ static int gmc_v9_0_sw_init(void *handle)
 		bitmap_or(adev->vmhubs_mask, adev->vmhubs_mask, &inst_mask, 32);

 		amdgpu_vm_adjust_size(adev, 256 * 1024, 9, 3, 48);
+		adev->gmc.translate_further = adev->vm_manager.num_level > 1;
 		break;
 	default:
 		break;
--- a/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
+++ b/drivers/gpu/drm/amd/amdgpu/nbio_v7_9.c
@ -390,22 +390,7 @@ static int nbio_v7_9_get_compute_partition_mode(struct amdgpu_device *adev)
 	px = REG_GET_FIELD(tmp, BIF_BX_PF0_PARTITION_COMPUTE_STATUS,
 			   PARTITION_MODE);

-	return ffs(px);
-}
-
-static void nbio_v7_9_set_compute_partition_mode(struct amdgpu_device *adev,
-					enum amdgpu_gfx_partition mode)
-{
-	u32 tmp;
-
-	/* Each bit represents DPX,TPX,QPX,CPX mode. No bit set means default
-	 * SPX mode.
-	 */
-	tmp = RREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_COMPUTE_STATUS);
-	tmp = REG_SET_FIELD(tmp, BIF_BX_PF0_PARTITION_COMPUTE_STATUS,
-			    PARTITION_MODE, mode ? BIT(mode - 1) : mode);
-
-	WREG32_SOC15(NBIO, 0, regBIF_BX_PF0_PARTITION_COMPUTE_STATUS, tmp);
+	return px;
 }

 static u32 nbio_v7_9_get_memory_partition_mode(struct amdgpu_device *adev,
@ -463,7 +448,6 @@ const struct amdgpu_nbio_funcs nbio_v7_9_funcs = {
 	.ih_control = nbio_v7_9_ih_control,
 	.remap_hdp_registers = nbio_v7_9_remap_hdp_registers,
 	.get_compute_partition_mode = nbio_v7_9_get_compute_partition_mode,
-	.set_compute_partition_mode = nbio_v7_9_set_compute_partition_mode,
 	.get_memory_partition_mode = nbio_v7_9_get_memory_partition_mode,
 	.init_registers = nbio_v7_9_init_registers,
 };
--- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.c
@ -624,10 +624,11 @@ static int psp_v13_0_exec_spi_cmd(struct psp_context *psp, int cmd)
 	WREG32_SOC15(MP0, 0, regMP0_SMN_C2PMSG_73, 1);

 	if (cmd == C2PMSG_CMD_SPI_UPDATE_FLASH_IMAGE)
-		return 0;
-
-	ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
-				MBOX_READY_FLAG, MBOX_READY_MASK, false);
+		ret = psp_wait_for_spirom_update(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+						 MBOX_READY_FLAG, MBOX_READY_MASK, PSP_SPIROM_UPDATE_TIMEOUT);
+	else
+		ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, regMP0_SMN_C2PMSG_115),
+				   MBOX_READY_FLAG, MBOX_READY_MASK, false);
 	if (ret) {
 		dev_err(adev->dev, "SPI cmd %x timed out, ret = %d", cmd, ret);
 		return ret;
--- a/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v13_0.h
@ -25,6 +25,8 @@

 #include "amdgpu_psp.h"

+#define PSP_SPIROM_UPDATE_TIMEOUT   60000   /* 60s */
+
 void psp_v13_0_set_psp_funcs(struct psp_context *psp);

 #endif
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c
@ -2312,7 +2312,7 @@ const struct amd_ip_funcs sdma_v4_0_ip_funcs = {

 static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
@ -2344,7 +2344,7 @@ static const struct amdgpu_ring_funcs sdma_v4_0_ring_funcs = {

 static const struct amdgpu_ring_funcs sdma_v4_0_page_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.secure_submission_supported = true,
--- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
+++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c
@ -1434,9 +1434,11 @@ static int sdma_v4_4_2_hw_fini(void *handle)
 		return 0;

 	inst_mask = GENMASK(adev->sdma.num_instances - 1, 0);
-	for (i = 0; i < adev->sdma.num_instances; i++) {
-		amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
-			       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for (i = 0; i < adev->sdma.num_instances; i++) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
 	}

 	sdma_v4_4_2_inst_ctx_switch_enable(adev, false, inst_mask);
@ -1823,7 +1825,7 @@ const struct amd_ip_funcs sdma_v4_4_2_ip_funcs = {

 static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
@ -1854,7 +1856,7 @@ static const struct amdgpu_ring_funcs sdma_v4_4_2_ring_funcs = {

 static const struct amdgpu_ring_funcs sdma_v4_4_2_page_ring_funcs = {
 	.type = AMDGPU_RING_TYPE_SDMA,
-	.align_mask = 0xf,
+	.align_mask = 0xff,
 	.nop = SDMA_PKT_NOP_HEADER_OP(SDMA_OP_NOP),
 	.support_64bit_ptrs = true,
 	.get_rptr = sdma_v4_4_2_ring_get_rptr,
@ -2073,9 +2075,11 @@ static int sdma_v4_4_2_xcp_suspend(void *handle, uint32_t inst_mask)
 	uint32_t tmp_mask = inst_mask;
 	int i;

-	for_each_inst(i, tmp_mask) {
-		amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
-			       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+	if (amdgpu_ras_is_supported(adev, AMDGPU_RAS_BLOCK__SDMA)) {
+		for_each_inst(i, tmp_mask) {
+			amdgpu_irq_put(adev, &adev->sdma.ecc_irq,
+				       AMDGPU_SDMA_IRQ_INSTANCE0 + i);
+		}
 	}

 	sdma_v4_4_2_inst_ctx_switch_enable(adev, false, inst_mask);
--- a/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/ta_ras_if.h
@ -130,6 +130,7 @@ struct ta_ras_init_flags {
 	uint8_t poison_mode_en;
 	uint8_t dgpu_mode;
 	uint16_t xcc_mask;
+	uint8_t channel_dis_num;
 };

 struct ta_ras_output_flags {
--- a/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h
+++ b/drivers/gpu/drm/amd/amdgpu/umc_v8_10.h
@ -33,7 +33,8 @@

 /* Total channel instances for all available umc nodes */
 #define UMC_V8_10_TOTAL_CHANNEL_NUM(adev) \
-	(UMC_V8_10_CHANNEL_INSTANCE_NUM * UMC_V8_10_UMC_INSTANCE_NUM * (adev)->gmc.num_umc)
+	(UMC_V8_10_CHANNEL_INSTANCE_NUM * UMC_V8_10_UMC_INSTANCE_NUM * \
+	(adev)->gmc.num_umc - hweight32((adev)->gmc.m_half_use) * 2)

 /* UMC regiser per channel offset */
 #define UMC_V8_10_PER_CHANNEL_OFFSET	0x400
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
@ -129,7 +129,11 @@ static int vcn_v4_0_sw_init(void *handle)
 		if (adev->vcn.harvest_config & (1 << i))
 			continue;

-		atomic_set(&adev->vcn.inst[i].sched_score, 0);
+		/* Init instance 0 sched_score to 1, so it's scheduled after other instances */
+		if (i == 0)
+			atomic_set(&adev->vcn.inst[i].sched_score, 1);
+		else
+			atomic_set(&adev->vcn.inst[i].sched_score, 0);

 		/* VCN UNIFIED TRAP */
 		r = amdgpu_irq_add_id(adev, amdgpu_ih_clientid_vcns[i],
--- a/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_crat.c
@ -30,9 +30,6 @@
 #include "amdgpu.h"
 #include "amdgpu_amdkfd.h"

-/* Fixme: Fake 32GB for 1PNPS1 mode bringup */
-#define DUMMY_VRAM_SIZE 31138512896
-
 /* GPU Processor ID base for dGPUs for which VCRAT needs to be created.
 * GPU processor ID are expressed with Bit[31]=1.
 * The base is set to 0x8000_0000 + 0x1000 to avoid collision with GPU IDs
@ -1056,8 +1053,6 @@ static int kfd_parse_subtype_mem(struct crat_subtype_memory *mem,

 			props->heap_type = heap_type;
 			props->flags = flags;
-			if (size_in_bytes == 0)
-				size_in_bytes = DUMMY_VRAM_SIZE; /* Fixme: TBD */
 			props->size_in_bytes = size_in_bytes;
 			props->width = width;

--- a/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_debug.c
@ -1097,7 +1097,7 @@ void kfd_dbg_set_enabled_debug_exception_mask(struct kfd_process *target,

 	pqm = &target->pqm;
 	list_for_each_entry(pqn, &pqm->queues, process_queue_list) {
-		if (!pqn)
+		if (!pqn->q)
 			continue;

 		found_mask |= pqn->q->properties.exception_status;
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@ -227,7 +227,7 @@ static int add_queue_mes(struct device_queue_manager *dqm, struct queue *q,
 	queue_input.tba_addr = qpd->tba_addr;
 	queue_input.tma_addr = qpd->tma_addr;
 	queue_input.trap_en = KFD_GC_VERSION(q->device) < IP_VERSION(11, 0, 0) ||
-			      KFD_GC_VERSION(q->device) >= IP_VERSION(12, 0, 0);
+			      KFD_GC_VERSION(q->device) > IP_VERSION(11, 0, 3);
 	queue_input.skip_process_ctx_clear = qpd->pqm->process->debug_trap_enabled;

 	queue_type = convert_to_mes_queue_type(q->properties.type);
@ -1807,7 +1807,7 @@ static int create_queue_cpsch(struct device_queue_manager *dqm, struct queue *q,
 	q->properties.is_evicted = !!qpd->evicted;
 	q->properties.is_dbg_wa = qpd->pqm->process->debug_trap_enabled &&
 			KFD_GC_VERSION(q->device) >= IP_VERSION(11, 0, 0) &&
-			KFD_GC_VERSION(q->device) < IP_VERSION(12, 0, 0);
+			KFD_GC_VERSION(q->device) <= IP_VERSION(11, 0, 3);

 	if (qd)
 		mqd_mgr->restore_mqd(mqd_mgr, &q->mqd, q->mqd_mem_obj, &q->gart_mqd_addr,
@ -2089,8 +2089,8 @@ static int destroy_queue_cpsch(struct device_queue_manager *dqm,
 	list_del(&q->list);
 	qpd->queue_count--;
 	if (q->properties.is_active) {
+		decrement_queue_count(dqm, qpd, q);
 		if (!dqm->dev->kfd->shared_resources.enable_mes) {
-			decrement_queue_count(dqm, qpd, q);
 			retval = execute_queues_cpsch(dqm,
 						      KFD_UNMAP_QUEUES_FILTER_DYNAMIC_QUEUES, 0,
 						      USE_DEFAULT_GRACE_PERIOD);
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.c
@ -41,6 +41,7 @@ struct kfd_event_waiter {
 	wait_queue_entry_t wait;
 	struct kfd_event *event; /* Event to wait for */
 	bool activated;		 /* Becomes true when event is signaled */
+	bool event_age_enabled;  /* set to true when last_event_age is non-zero */
 };

 /*
@ -431,6 +432,7 @@ int kfd_event_create(struct file *devkfd, struct kfd_process *p,
 	if (!ret) {
 		*event_id = ev->event_id;
 		*event_trigger_data = ev->event_id;
+		ev->event_age = 1;
 	} else {
 		kfree(ev);
 	}
@ -629,6 +631,11 @@ static void set_event(struct kfd_event *ev)
 	 * updating the wait queues in kfd_wait_on_events.
 	 */
 	ev->signaled = !ev->auto_reset || !waitqueue_active(&ev->wq);
+	if (!(++ev->event_age)) {
+		/* Never wrap back to reserved/default event age 0/1 */
+		ev->event_age = 2;
+		WARN_ONCE(1, "event_age wrap back!");
+	}

 	list_for_each_entry(waiter, &ev->wq.head, wait.entry)
 		WRITE_ONCE(waiter->activated, true);
@ -791,9 +798,9 @@ static struct kfd_event_waiter *alloc_event_waiters(uint32_t num_events)

 static int init_event_waiter(struct kfd_process *p,
 		struct kfd_event_waiter *waiter,
-		uint32_t event_id)
+		struct kfd_event_data *event_data)
 {
-	struct kfd_event *ev = lookup_event_by_id(p, event_id);
+	struct kfd_event *ev = lookup_event_by_id(p, event_data->event_id);

 	if (!ev)
 		return -EINVAL;
@ -802,6 +809,15 @@ static int init_event_waiter(struct kfd_process *p,
 	waiter->event = ev;
 	waiter->activated = ev->signaled;
 	ev->signaled = ev->signaled && !ev->auto_reset;
+
+	/* last_event_age = 0 reserved for backward compatible */
+	if (waiter->event->type == KFD_EVENT_TYPE_SIGNAL &&
+		event_data->signal_event_data.last_event_age) {
+		waiter->event_age_enabled = true;
+		if (ev->event_age != event_data->signal_event_data.last_event_age)
+			waiter->activated = true;
+	}
+
 	if (!waiter->activated)
 		add_wait_queue(&ev->wq, &waiter->wait);
 	spin_unlock(&ev->lock);
@ -849,22 +865,29 @@ static int copy_signaled_event_data(uint32_t num_events,
 		struct kfd_event_waiter *event_waiters,
 		struct kfd_event_data __user *data)
 {
-	struct kfd_hsa_memory_exception_data *src;
-	struct kfd_hsa_memory_exception_data __user *dst;
+	void *src;
+	void __user *dst;
 	struct kfd_event_waiter *waiter;
 	struct kfd_event *event;
-	uint32_t i;
+	uint32_t i, size = 0;

 	for (i = 0; i < num_events; i++) {
 		waiter = &event_waiters[i];
 		event = waiter->event;
 		if (!event)
 			return -EINVAL; /* event was destroyed */
-		if (waiter->activated && event->type == KFD_EVENT_TYPE_MEMORY) {
-			dst = &data[i].memory_exception_data;
-			src = &event->memory_exception_data;
-			if (copy_to_user(dst, src,
-				sizeof(struct kfd_hsa_memory_exception_data)))
+		if (waiter->activated) {
+			if (event->type == KFD_EVENT_TYPE_MEMORY) {
+				dst = &data[i].memory_exception_data;
+				src = &event->memory_exception_data;
+				size = sizeof(struct kfd_hsa_memory_exception_data);
+			} else if (event->type == KFD_EVENT_TYPE_SIGNAL &&
+				waiter->event_age_enabled) {
+				dst = &data[i].signal_event_data.last_event_age;
+				src = &event->event_age;
+				size = sizeof(u64);
+			}
+			if (size && copy_to_user(dst, src, size))
 				return -EFAULT;
 		}
 	}
@ -942,8 +965,7 @@ int kfd_wait_on_events(struct kfd_process *p,
 			goto out_unlock;
 		}

-		ret = init_event_waiter(p, &event_waiters[i],
-					event_data.event_id);
+		ret = init_event_waiter(p, &event_waiters[i], &event_data);
 		if (ret)
 			goto out_unlock;
 	}
--- a/drivers/gpu/drm/amd/amdkfd/kfd_events.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_events.h
@ -53,6 +53,7 @@ struct signal_page;

 struct kfd_event {
 	u32 event_id;
+	u64 event_age;

 	bool signaled;
 	bool auto_reset;
--- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
@ -64,7 +64,7 @@ svm_migrate_gart_map(struct amdgpu_ring *ring, uint64_t npages,
 	num_dw = ALIGN(adev->mman.buffer_funcs->copy_num_dw, 8);
 	num_bytes = npages * 8;

-	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.entity,
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.high_pr,
 				     AMDGPU_FENCE_OWNER_UNDEFINED,
 				     num_dw * 4 + num_bytes,
 				     AMDGPU_IB_POOL_DELAYED,
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@ -3267,6 +3267,7 @@ static void dm_handle_mst_sideband_msg(struct amdgpu_dm_connector *aconnector)

 	while (dret == dpcd_bytes_to_read &&
 		process_count < max_process_count) {
+		u8 ack[DP_PSR_ERROR_STATUS - DP_SINK_COUNT_ESI] = {};
 		u8 retry;
 		dret = 0;

@ -3275,28 +3276,29 @@ static void dm_handle_mst_sideband_msg(struct amdgpu_dm_connector *aconnector)
 		DRM_DEBUG_DRIVER("ESI %02x %02x %02x\n", esi[0], esi[1], esi[2]);
 		/* handle HPD short pulse irq */
 		if (aconnector->mst_mgr.mst_state)
-			drm_dp_mst_hpd_irq(
-				&aconnector->mst_mgr,
-				esi,
-				&new_irq_handled);
+			drm_dp_mst_hpd_irq_handle_event(&aconnector->mst_mgr,
+							esi,
+							ack,
+							&new_irq_handled);

 		if (new_irq_handled) {
 			/* ACK at DPCD to notify down stream */
-			const int ack_dpcd_bytes_to_write =
-				dpcd_bytes_to_read - 1;
-
 			for (retry = 0; retry < 3; retry++) {
-				u8 wret;
+				ssize_t wret;

-				wret = drm_dp_dpcd_write(
-					&aconnector->dm_dp_aux.aux,
-					dpcd_addr + 1,
-					&esi[1],
-					ack_dpcd_bytes_to_write);
-				if (wret == ack_dpcd_bytes_to_write)
+				wret = drm_dp_dpcd_writeb(&aconnector->dm_dp_aux.aux,
+							  dpcd_addr + 1,
+							  ack[1]);
+				if (wret == 1)
 					break;
 			}

+			if (retry == 3) {
+				DRM_ERROR("Failed to ack MST event.\n");
+				return;
+			}
+
+			drm_dp_mst_hpd_irq_send_new_request(&aconnector->mst_mgr);
 			/* check if there is new irq to be handled */
 			dret = drm_dp_dpcd_read(
 				&aconnector->dm_dp_aux.aux,
@ -7256,7 +7258,13 @@ static int amdgpu_dm_connector_get_modes(struct drm_connector *connector)
 				drm_add_modes_noedid(connector, 1920, 1080);
 	} else {
 		amdgpu_dm_connector_ddc_get_modes(connector, edid);
-		amdgpu_dm_connector_add_common_modes(encoder, connector);
+		/* most eDP supports only timings from its edid,
+		 * usually only detailed timings are available
+		 * from eDP edid. timings which are not from edid
+		 * may damage eDP
+		 */
+		if (connector->connector_type != DRM_MODE_CONNECTOR_eDP)
+			amdgpu_dm_connector_add_common_modes(encoder, connector);
 		amdgpu_dm_connector_add_freesync_modes(connector, edid);
 	}
 	amdgpu_dm_fbc_init(connector);
@ -8273,6 +8281,12 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
 		if (acrtc_state->abm_level != dm_old_crtc_state->abm_level)
 			bundle->stream_update.abm_level = &acrtc_state->abm_level;

+		mutex_lock(&dm->dc_lock);
+		if ((acrtc_state->update_type > UPDATE_TYPE_FAST) &&
+				acrtc_state->stream->link->psr_settings.psr_allow_active)
+			amdgpu_dm_psr_disable(acrtc_state->stream);
+		mutex_unlock(&dm->dc_lock);
+
 		/*
 		 * If FreeSync state on the stream has changed then we need to
 		 * re-adjust the min/max bounds now that DC doesn't handle this
@ -8286,10 +8300,6 @@ static void amdgpu_dm_commit_planes(struct drm_atomic_state *state,
 			spin_unlock_irqrestore(&pcrtc->dev->event_lock, flags);
 		}
 		mutex_lock(&dm->dc_lock);
-		if ((acrtc_state->update_type > UPDATE_TYPE_FAST) &&
-				acrtc_state->stream->link->psr_settings.psr_allow_active)
-			amdgpu_dm_psr_disable(acrtc_state->stream);
-
 		update_planes_and_stream_adapter(dm->dc,
 					 acrtc_state->update_type,
 					 planes_count,
@ -8951,10 +8961,17 @@ static void amdgpu_dm_atomic_commit_tail(struct drm_atomic_state *state)

 	drm_atomic_helper_cleanup_planes(dev, state);

-	/* return the stolen vga memory back to VRAM */
-	if (!adev->mman.keep_stolen_vga_memory)
-		amdgpu_bo_free_kernel(&adev->mman.stolen_vga_memory, NULL, NULL);
-	amdgpu_bo_free_kernel(&adev->mman.stolen_extended_memory, NULL, NULL);
+	/* Don't free the memory if we are hitting this as part of suspend.
+	 * This way we don't free any memory during suspend; see
+	 * amdgpu_bo_free_kernel().  The memory will be freed in the first
+	 * non-suspend modeset or when the driver is torn down.
+	 */
+	if (!adev->in_suspend) {
+		/* return the stolen vga memory back to VRAM */
+		if (!adev->mman.keep_stolen_vga_memory)
+			amdgpu_bo_free_kernel(&adev->mman.stolen_vga_memory, NULL, NULL);
+		amdgpu_bo_free_kernel(&adev->mman.stolen_extended_memory, NULL, NULL);
+	}

 	/*
 	 * Finally, drop a runtime PM reference for each newly disabled CRTC,
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_debugfs.c
@ -2760,6 +2760,32 @@ static int psr_read_residency(void *data, u64 *val)
 	return 0;
 }

+/* read allow_edp_hotplug_detection */
+static int allow_edp_hotplug_detection_get(void *data, u64 *val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	*val = adev->dm.dc->config.allow_edp_hotplug_detection;
+
+	return 0;
+}
+
+/* set allow_edp_hotplug_detection */
+static int allow_edp_hotplug_detection_set(void *data, u64 val)
+{
+	struct amdgpu_dm_connector *aconnector = data;
+	struct drm_connector *connector = &aconnector->base;
+	struct drm_device *dev = connector->dev;
+	struct amdgpu_device *adev = drm_to_adev(dev);
+
+	adev->dm.dc->config.allow_edp_hotplug_detection = (uint32_t) val;
+
+	return 0;
+}
+
 /*
 * Set dmcub trace event IRQ enable or disable.
 * Usage to enable dmcub trace event IRQ: echo 1 > /sys/kernel/debug/dri/0/amdgpu_dm_dmcub_trace_event_en
@ -2798,6 +2824,10 @@ DEFINE_DEBUGFS_ATTRIBUTE(psr_fops, psr_get, NULL, "%llu\n");
 DEFINE_DEBUGFS_ATTRIBUTE(psr_residency_fops, psr_read_residency, NULL,
 			 "%llu\n");

+DEFINE_DEBUGFS_ATTRIBUTE(allow_edp_hotplug_detection_fops,
+			allow_edp_hotplug_detection_get,
+			allow_edp_hotplug_detection_set, "%llu\n");
+
 DEFINE_SHOW_ATTRIBUTE(current_backlight);
 DEFINE_SHOW_ATTRIBUTE(target_backlight);

@ -2968,6 +2998,8 @@ void connector_debugfs_init(struct amdgpu_dm_connector *connector)
 				    &target_backlight_fops);
 		debugfs_create_file("ilr_setting", 0644, dir, connector,
 					&edp_ilr_debugfs_fops);
+		debugfs_create_file("allow_edp_hotplug_detection", 0644, dir, connector,
+					&allow_edp_hotplug_detection_fops);
 	}

 	for (i = 0; i < ARRAY_SIZE(connector_debugfs_entries); i++) {
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_helpers.c
@ -44,9 +44,6 @@
 #include "dm_helpers.h"
 #include "ddc_service_types.h"

-/* MST Dock */
-static const uint8_t SYNAPTICS_DEVICE_ID[] = "SYNA";
-
 /* dm_helpers_parse_edid_caps
 *
 * Parse edid caps
@ -702,6 +699,9 @@ static void apply_synaptics_fifo_reset_wa(struct drm_dp_aux *aux)
 	DC_LOG_DC("Done apply_synaptics_fifo_reset_wa\n");
 }

+/* MST Dock */
+static const uint8_t SYNAPTICS_DEVICE_ID[] = "SYNA";
+
 static uint8_t write_dsc_enable_synaptics_non_virtual_dpcd_mst(
 		struct drm_dp_aux *aux,
 		const struct dc_stream_state *stream,
--- a/drivers/gpu/drm/amd/display/dc/core/dc.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc.c
@ -1296,7 +1296,7 @@ static void disable_vbios_mode_if_required(

 					if (pix_clk_100hz != requested_pix_clk_100hz) {
 						dc->link_srv->set_dpms_off(pipe);
-						pipe->stream->dpms_off = true;
+						pipe->stream->dpms_off = false;
 					}
 				}
 			}
@ -3480,23 +3480,21 @@ static void build_dmub_update_dirty_rect(


 /**
- * ************************************************************************************************
- * build_dmub_cmd_list: Build an array of DMCUB commands to be sent to DMCUB
+ * build_dmub_cmd_list() - Build an array of DMCUB commands to be sent to DMCUB
 *
- * @param [in]: dc: Current DC state
- * @param [in]: srf_updates: Array of surface updates
- * @param [in]: surface_count: Number of surfaces that have an updated
- * @param [in]: stream: Correponding stream to be updated in the current flip
- * @param [in]: context: New DC state to be programmed
+ * @dc: Current DC state
+ * @srf_updates: Array of surface updates
+ * @surface_count: Number of surfaces that have an updated
+ * @stream: Corresponding stream to be updated in the current flip
+ * @context: New DC state to be programmed
 *
- * @param [out]: dc_dmub_cmd: Array of DMCUB commands to be sent to DMCUB
- * @param [out]: dmub_cmd_count: Count indicating the number of DMCUB commands in dc_dmub_cmd array
+ * @dc_dmub_cmd: Array of DMCUB commands to be sent to DMCUB
+ * @dmub_cmd_count: Count indicating the number of DMCUB commands in dc_dmub_cmd array
 *
 * This function builds an array of DMCUB commands to be sent to DMCUB. This function is required
 * to build an array of commands and have them sent while the OTG lock is acquired.
 *
- * @return: void
- * ************************************************************************************************
+ * Return: void
 */
 static void build_dmub_cmd_list(struct dc *dc,
 		struct dc_surface_update *srf_updates,
@ -4201,20 +4199,18 @@ static bool commit_minimal_transition_state(struct dc *dc,
 }

 /**
- * *******************************************************************************
- * update_seamless_boot_flags: Helper function for updating seamless boot flags
+ * update_seamless_boot_flags() - Helper function for updating seamless boot flags
 *
- * @param [in]: dc: Current DC state
- * @param [in]: context: New DC state to be programmed
- * @param [in]: surface_count: Number of surfaces that have an updated
- * @param [in]: stream: Correponding stream to be updated in the current flip
+ * @dc: Current DC state
+ * @context: New DC state to be programmed
+ * @surface_count: Number of surfaces that have an updated
+ * @stream: Corresponding stream to be updated in the current flip
 *
 * Updating seamless boot flags do not need to be part of the commit sequence. This
 * helper function will update the seamless boot flags on each flip (if required)
 * outside of the HW commit sequence (fast or slow).
 *
- * @return: void
- * *******************************************************************************
+ * Return: void
 */
 static void update_seamless_boot_flags(struct dc *dc,
 		struct dc_state *context,
--- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c
+++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c
@ -562,6 +562,29 @@ void hwss_build_fast_sequence(struct dc *dc,
 				(*num_steps)++;
 			}

+			if (current_mpc_pipe->stream->update_flags.bits.out_csc) {
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.mpc = dc->res_pool->mpc;
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.mpcc_id = current_mpc_pipe->plane_res.hubp->inst;
+				block_sequence[*num_steps].params.power_on_mpc_mem_pwr_params.power_on = true;
+				block_sequence[*num_steps].func = MPC_POWER_ON_MPC_MEM_PWR;
+				(*num_steps)++;
+
+				if (current_mpc_pipe->stream->csc_color_matrix.enable_adjustment == true) {
+					block_sequence[*num_steps].params.set_output_csc_params.mpc = dc->res_pool->mpc;
+					block_sequence[*num_steps].params.set_output_csc_params.opp_id = current_mpc_pipe->stream_res.opp->inst;
+					block_sequence[*num_steps].params.set_output_csc_params.regval = current_mpc_pipe->stream->csc_color_matrix.matrix;
+					block_sequence[*num_steps].params.set_output_csc_params.ocsc_mode = MPC_OUTPUT_CSC_COEF_A;
+					block_sequence[*num_steps].func = MPC_SET_OUTPUT_CSC;
+					(*num_steps)++;
+				} else {
+					block_sequence[*num_steps].params.set_ocsc_default_params.mpc = dc->res_pool->mpc;
+					block_sequence[*num_steps].params.set_ocsc_default_params.opp_id = current_mpc_pipe->stream_res.opp->inst;
+					block_sequence[*num_steps].params.set_ocsc_default_params.color_space = current_mpc_pipe->stream->output_color_space;
+					block_sequence[*num_steps].params.set_ocsc_default_params.ocsc_mode = MPC_OUTPUT_CSC_COEF_A;
+					block_sequence[*num_steps].func = MPC_SET_OCSC_DEFAULT;
+					(*num_steps)++;
+				}
+			}
 			current_mpc_pipe = current_mpc_pipe->bottom_pipe;
 		}
 		current_pipe = current_pipe->next_odm_pipe;
@ -661,6 +684,15 @@ void hwss_execute_sequence(struct dc *dc,
 					params->update_visual_confirm_params.pipe_ctx,
 					params->update_visual_confirm_params.mpcc_id);
 			break;
+		case MPC_POWER_ON_MPC_MEM_PWR:
+			hwss_power_on_mpc_mem_pwr(params);
+			break;
+		case MPC_SET_OUTPUT_CSC:
+			hwss_set_output_csc(params);
+			break;
+		case MPC_SET_OCSC_DEFAULT:
+			hwss_set_ocsc_default(params);
+			break;
 		case DMUB_SEND_DMCUB_CMD:
 			hwss_send_dmcub_cmd(params);
 			break;
@ -718,6 +750,44 @@ void hwss_program_bias_and_scale(union block_sequence_params *params)
 		dpp->funcs->dpp_program_bias_and_scale(dpp, &bns_params);
 }

+void hwss_power_on_mpc_mem_pwr(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->power_on_mpc_mem_pwr_params.mpc;
+	int mpcc_id = params->power_on_mpc_mem_pwr_params.mpcc_id;
+	bool power_on = params->power_on_mpc_mem_pwr_params.power_on;
+
+	if (mpc->funcs->power_on_mpc_mem_pwr)
+		mpc->funcs->power_on_mpc_mem_pwr(mpc, mpcc_id, power_on);
+}
+
+void hwss_set_output_csc(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->set_output_csc_params.mpc;
+	int opp_id = params->set_output_csc_params.opp_id;
+	const uint16_t *matrix = params->set_output_csc_params.regval;
+	enum mpc_output_csc_mode ocsc_mode = params->set_output_csc_params.ocsc_mode;
+
+	if (mpc->funcs->set_output_csc != NULL)
+		mpc->funcs->set_output_csc(mpc,
+				opp_id,
+				matrix,
+				ocsc_mode);
+}
+
+void hwss_set_ocsc_default(union block_sequence_params *params)
+{
+	struct mpc *mpc = params->set_ocsc_default_params.mpc;
+	int opp_id = params->set_ocsc_default_params.opp_id;
+	enum dc_color_space colorspace = params->set_ocsc_default_params.color_space;
+	enum mpc_output_csc_mode ocsc_mode = params->set_ocsc_default_params.ocsc_mode;
+
+	if (mpc->funcs->set_ocsc_default != NULL)
+		mpc->funcs->set_ocsc_default(mpc,
+				opp_id,
+				colorspace,
+				ocsc_mode);
+}
+
 void get_mclk_switch_visual_confirm_color(
 		struct dc *dc,
 		struct dc_state *context,
--- a/drivers/gpu/drm/amd/display/dc/dc.h
+++ b/drivers/gpu/drm/amd/display/dc/dc.h
@ -1,5 +1,5 @@
 /*
- * Copyright 2012-14 Advanced Micro Devices, Inc.
+ * Copyright 2012-2023 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
@ -45,7 +45,7 @@ struct aux_payload;
 struct set_config_cmd_payload;
 struct dmub_notification;

-#define DC_VER "3.2.237"
+#define DC_VER "3.2.239"

 #define MAX_SURFACES 3
 #define MAX_PLANES 6
@ -60,7 +60,9 @@ struct dc_versions {
 };

 enum dp_protocol_version {
-	DP_VERSION_1_4,
+	DP_VERSION_1_4 = 0,
+	DP_VERSION_2_1,
+	DP_VERSION_UNKNOWN,
 };

 enum dc_plane_type {
@ -264,6 +266,7 @@ struct dc_caps {
 	uint16_t subvp_pstate_allow_width_us;
 	uint16_t subvp_vertical_int_margin_us;
 	bool seamless_odm;
+	uint32_t max_v_total;
 	uint8_t subvp_drr_vblank_start_margin_us;
 };

--- a/drivers/gpu/drm/amd/display/dc/dc_dp_types.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_dp_types.h
@ -61,7 +61,7 @@ enum dc_link_rate {
 	 */
 	LINK_RATE_UHBR10 = 1000,	// UHBR10 - 10.0 Gbps/Lane
 	LINK_RATE_UHBR13_5 = 1350,	// UHBR13.5 - 13.5 Gbps/Lane
-	LINK_RATE_UHBR20 = 2000,	// UHBR10 - 20.0 Gbps/Lane
+	LINK_RATE_UHBR20 = 2000,	// UHBR20 - 20.0 Gbps/Lane
 };

 enum dc_link_spread {
--- a/drivers/gpu/drm/amd/display/dc/dc_types.h
+++ b/drivers/gpu/drm/amd/display/dc/dc_types.h
@ -597,6 +597,7 @@ enum dc_psr_state {
 	PSR_STATE4b_FULL_FRAME,
 	PSR_STATE4c_FULL_FRAME,
 	PSR_STATE4_FULL_FRAME_POWERUP,
+	PSR_STATE4_FULL_FRAME_HW_LOCK,
 	PSR_STATE5,
 	PSR_STATE5a,
 	PSR_STATE5b,
--- a/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
+++ b/drivers/gpu/drm/amd/display/dc/dce/dmub_psr.c
@ -87,6 +87,8 @@ static enum dc_psr_state convert_psr_state(uint32_t raw_state)
 		state = PSR_STATE4c_FULL_FRAME;
 	else if (raw_state == 0x4E)
 		state = PSR_STATE4_FULL_FRAME_POWERUP;
+	else if (raw_state == 0x4F)
+		state = PSR_STATE4_FULL_FRAME_HW_LOCK;
 	else if (raw_state == 0x60)
 		state = PSR_STATE_HWLOCK_MGR;
 	else if (raw_state == 0x61)
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_opp.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_opp.c
@ -37,14 +37,14 @@
 #define CTX \
 	oppn10->base.ctx

-
-/************* FORMATTER ************/
-
 /**
- *	set_truncation
+ * opp1_set_truncation():
 *	1) set truncation depth: 0 for 18 bpp or 1 for 24 bpp
 *	2) enable truncation
 *	3) HW remove 12bit FMT support for DCE11 power saving reason.
+ *
+ * @oppn10: output_pixel_processor struct instance for dcn10.
+ * @params: pointer to bit_depth_reduction_params.
 */
 static void opp1_set_truncation(
 		struct dcn10_opp *oppn10,
@ -149,11 +149,12 @@ void opp1_program_bit_depth_reduction(
 }

 /**
- *	set_pixel_encoding
- *
- *	Set Pixel Encoding
+ * opp1_set_pixel_encoding():
 *		0: RGB 4:4:4 or YCbCr 4:4:4 or YOnly
 *		1: YCbCr 4:2:2
+ *
+ * @oppn10: output_pixel_processor struct instance for dcn10.
+ * @params: pointer to clamping_and_pixel_encoding_params.
 */
 static void opp1_set_pixel_encoding(
 	struct dcn10_opp *oppn10,
@ -180,13 +181,16 @@ static void opp1_set_pixel_encoding(
 }

 /**
- *	Set Clamping
+ * opp1_set_clamping():
 *	1) Set clamping format based on bpc - 0 for 6bpc (No clamping)
 *		1 for 8 bpc
 *		2 for 10 bpc
 *		3 for 12 bpc
 *		7 for programable
 *	2) Enable clamp if Limited range requested
+ *
+ * @oppn10: output_pixel_processor struct instance for dcn10.
+ * @params: pointer to clamping_and_pixel_encoding_params.
 */
 static void opp1_set_clamping(
 	struct dcn10_opp *oppn10,
--- a/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn10/dcn10_optc.c
@ -42,11 +42,13 @@
 #define STATIC_SCREEN_EVENT_MASK_RANGETIMING_DOUBLE_BUFFER_UPDATE_EN 0x100

 /**
-* apply_front_porch_workaround  TODO FPGA still need?
-*
-* This is a workaround for a bug that has existed since R5xx and has not been
-* fixed keep Front porch at minimum 2 for Interlaced mode or 1 for progressive.
-*/
+ * apply_front_porch_workaround() - This is a workaround for a bug that has
+ *                                  existed since R5xx and has not been fixed
+ *                                  keep Front porch at minimum 2 for Interlaced
+ *                                  mode or 1 for progressive.
+ *
+ * @timing: Timing parameters used to configure DCN blocks.
+ */
 static void apply_front_porch_workaround(struct dc_crtc_timing *timing)
 {
 	if (timing->flags.INTERLACE == 1) {
@ -133,9 +135,20 @@ void optc1_setup_vertical_interrupt2(
 }

 /**
- * program_timing_generator   used by mode timing set
- * Program CRTC Timing Registers - OTG_H_*, OTG_V_*, Pixel repetition.
- * Including SYNC. Call BIOS command table to program Timings.
+ * optc1_program_timing() - used by mode timing set Program
+ *                          CRTC Timing Registers - OTG_H_*,
+ *                          OTG_V_*, Pixel repetition.
+ *                          Including SYNC. Call BIOS command table to program Timings.
+ *
+ * @optc: timing_generator instance.
+ * @dc_crtc_timing: Timing parameters used to configure DCN blocks.
+ * @vready_offset: Vready's starting position.
+ * @vstartup_start: Vstartup period.
+ * @vupdate_offset: Vupdate starting position.
+ * @vupdate_width: Vupdate duration.
+ * @signal: DC signal types.
+ * @use_vbios: to program timings from BIOS command table.
+ *
 */
 void optc1_program_timing(
 	struct timing_generator *optc,
@ -385,6 +398,9 @@ void optc1_set_blank_data_double_buffer(struct timing_generator *optc, bool enab
 * Sets double buffer point for V_TOTAL, H_TOTAL, VTOTAL_MIN,
 * VTOTAL_MAX, VTOTAL_MIN_SEL and VTOTAL_MAX_SEL registers.
 *
+ * @optc: timing_generator instance.
+ * @enable: Enable DRR double buffering control if true, disable otherwise.
+ *
 * Options: any time,  start of frame, dp start of frame (range timing)
 */
 void optc1_set_timing_double_buffer(struct timing_generator *optc, bool enable)
@ -397,8 +413,9 @@ void optc1_set_timing_double_buffer(struct timing_generator *optc, bool enable)
 }

 /**
- * unblank_crtc
- * Call ASIC Control Object to UnBlank CRTC.
+ * optc1_unblank_crtc() - Call ASIC Control Object to UnBlank CRTC.
+ *
+ * @optc: timing_generator instance.
 */
 static void optc1_unblank_crtc(struct timing_generator *optc)
 {
@ -419,8 +436,9 @@ static void optc1_unblank_crtc(struct timing_generator *optc)
 }

 /**
- * blank_crtc
- * Call ASIC Control Object to Blank CRTC.
+ * optc1_blank_crtc() - Call ASIC Control Object to Blank CRTC.
+ *
+ * @optc: timing_generator instance.
 */

 static void optc1_blank_crtc(struct timing_generator *optc)
@ -493,8 +511,9 @@ void optc1_enable_optc_clock(struct timing_generator *optc, bool enable)
 }

 /**
- * Enable CRTC
- * Enable CRTC - call ASIC Control Object to enable Timing generator.
+ * optc1_enable_crtc() - Enable CRTC - call ASIC Control Object to enable Timing generator.
+ *
+ * @optc: timing_generator instance.
 */
 static bool optc1_enable_crtc(struct timing_generator *optc)
 {
@ -890,15 +909,11 @@ static void optc1_program_manual_trigger(struct timing_generator *optc)
 			MANUAL_FLOW_CONTROL, 0);
 }

-
 /**
- *****************************************************************************
- *  Function: set_drr
+ * optc1_set_drr() - Program dynamic refresh rate registers m_OTGx_OTG_V_TOTAL_*.
 *
- *  @brief
- *     Program dynamic refresh rate registers m_OTGx_OTG_V_TOTAL_*.
- *
- *****************************************************************************
+ * @optc: timing_generator instance.
+ * @params: parameters used for Dynamic Refresh Rate.
 */
 void optc1_set_drr(
 	struct timing_generator *optc,
--- a/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn20/dcn20_hwseq.c
@ -1741,17 +1741,6 @@ static void dcn20_program_pipe(

 		if (hws->funcs.setup_vupdate_interrupt)
 			hws->funcs.setup_vupdate_interrupt(dc, pipe_ctx);
-
-		if (hws->funcs.calculate_dccg_k1_k2_values && dc->res_pool->dccg->funcs->set_pixel_rate_div) {
-			unsigned int k1_div, k2_div;
-
-			hws->funcs.calculate_dccg_k1_k2_values(pipe_ctx, &k1_div, &k2_div);
-
-			dc->res_pool->dccg->funcs->set_pixel_rate_div(
-				dc->res_pool->dccg,
-				pipe_ctx->stream_res.tg->inst,
-				k1_div, k2_div);
-		}
 	}

 	if (pipe_ctx->update_flags.bits.odm)
--- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_optc.c
@ -279,6 +279,9 @@ static void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, in
 * Sets double buffer point for V_TOTAL, H_TOTAL, VTOTAL_MIN,
 * VTOTAL_MAX, VTOTAL_MIN_SEL and VTOTAL_MAX_SEL registers.
 *
+ * @optc: timing_generator instance.
+ * @enable: Enable DRR double buffering control if true, disable otherwise.
+ *
 * Options: any time,  start of frame, dp start of frame (range timing)
 */
 static void optc3_set_timing_double_buffer(struct timing_generator *optc, bool enable)
--- a/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn30/dcn30_resource.c
@ -2328,6 +2328,7 @@ static bool dcn30_resource_construct(
 	dc->caps.color.mpc.ocsc = 1;

 	dc->caps.dp_hdmi21_pcon_support = true;
+	dc->caps.max_v_total = (1 << 15) - 1;

 	/* read VBIOS LTTPR caps */
 	{
--- a/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn302/dcn302_resource.c
@ -1227,6 +1227,7 @@ static bool dcn302_resource_construct(
 	dc->caps.force_dp_tps4_for_cp2520 = true;
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
+	dc->caps.max_v_total = (1 << 15) - 1;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn303/dcn303_resource.c
@ -1152,6 +1152,7 @@ static bool dcn303_resource_construct(
 	dc->caps.force_dp_tps4_for_cp2520 = true;
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
+	dc->caps.max_v_total = (1 << 15) - 1;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn31/dcn31_dccg.c
@ -47,6 +47,14 @@ void dccg31_update_dpp_dto(struct dccg *dccg, int dpp_inst, int req_dppclk)
 {
 	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);

+	if (dccg->dpp_clock_gated[dpp_inst]) {
+		/*
+		 * Do not update the DPPCLK DTO if the clock is stopped.
+		 * It is treated the same as if the pipe itself were in PG.
+		 */
+		return;
+	}
+
 	if (dccg->ref_dppclk && req_dppclk) {
 		int ref_dppclk = dccg->ref_dppclk;
 		int modulo, phase;
--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_dccg.c
@ -332,6 +332,9 @@ static void dccg314_dpp_root_clock_control(
 {
 	struct dcn_dccg *dccg_dcn = TO_DCN_DCCG(dccg);

+	if (dccg->dpp_clock_gated[dpp_inst] == clock_on)
+		return;
+
 	if (clock_on) {
 		/* turn off the DTO and leave phase/modulo at max */
 		REG_UPDATE(DPPCLK_DTO_CTRL, DPPCLK_DTO_ENABLE[dpp_inst], 0);
@ -345,6 +348,8 @@ static void dccg314_dpp_root_clock_control(
 			  DPPCLK0_DTO_PHASE, 0,
 			  DPPCLK0_DTO_MODULO, 1);
 	}
+
+	dccg->dpp_clock_gated[dpp_inst] = !clock_on;
 }

 static const struct dccg_funcs dccg314_funcs = {
--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_hwseq.c
@ -337,14 +337,13 @@ void dcn314_enable_power_gating_plane(struct dce_hwseq *hws, bool enable)
 		REG_SET(DC_IP_REQUEST_CNTL, 0, IP_REQUEST_EN, 0);
 }

-unsigned int dcn314_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div)
+void dcn314_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div)
 {
 	struct dc_stream_state *stream = pipe_ctx->stream;
-	unsigned int odm_combine_factor = 0;
 	bool two_pix_per_container = false;

 	two_pix_per_container = optc2_is_two_pixels_per_containter(&stream->timing);
-	odm_combine_factor = get_odm_config(pipe_ctx, NULL);
+	get_odm_config(pipe_ctx, NULL);

 	if (stream->ctx->dc->link_srv->dp_is_128b_132b_signal(pipe_ctx)) {
 		*k1_div = PIXEL_RATE_DIV_BY_1;
@ -362,15 +361,11 @@ unsigned int dcn314_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsig
 		} else {
 			*k1_div = PIXEL_RATE_DIV_BY_1;
 			*k2_div = PIXEL_RATE_DIV_BY_4;
-			if (odm_combine_factor == 2)
-				*k2_div = PIXEL_RATE_DIV_BY_2;
 		}
 	}

 	if ((*k1_div == PIXEL_RATE_DIV_NA) && (*k2_div == PIXEL_RATE_DIV_NA))
 		ASSERT(false);
-
-	return odm_combine_factor;
 }

 void dcn314_set_pixels_per_cycle(struct pipe_ctx *pipe_ctx)
--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_hwseq.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_hwseq.h
@ -37,7 +37,7 @@ void dcn314_dsc_pg_control(struct dce_hwseq *hws, unsigned int dsc_inst, bool po

 void dcn314_enable_power_gating_plane(struct dce_hwseq *hws, bool enable);

-unsigned int dcn314_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div);
+void dcn314_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div);

 void dcn314_set_pixels_per_cycle(struct pipe_ctx *pipe_ctx);

--- a/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn314/dcn314_resource.c
@ -874,8 +874,8 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.force_abm_enable = false,
 	.timing_trace = false,
 	.clock_trace = true,
-	.disable_dpp_power_gate = true,
-	.disable_hubp_power_gate = true,
+	.disable_dpp_power_gate = false,
+	.disable_hubp_power_gate = false,
 	.disable_pplib_clock_request = false,
 	.pipe_split_policy = MPC_SPLIT_DYNAMIC,
 	.force_single_disp_pipe_split = false,
@ -904,6 +904,22 @@ static const struct dc_debug_options debug_defaults_drv = {
 			.afmt = true,
 		}
 	},
+
+	.root_clock_optimization = {
+			.bits = {
+					.dpp = true,
+					.dsc = false,
+					.hdmistream = false,
+					.hdmichar = false,
+					.dpstream = false,
+					.symclk32_se = false,
+					.symclk32_le = false,
+					.symclk_fe = false,
+					.physymclk = false,
+					.dpiasymclk = false,
+			}
+	},
+
 	.seamless_boot_odm_combine = true
 };

@ -1867,6 +1883,13 @@ static bool dcn314_resource_construct(
 	/* Use pipe context based otg sync logic */
 	dc->config.use_pipe_ctx_sync_logic = true;

+	/* Disable pipe power gating when unsupported */
+	if (ctx->asic_id.hw_internal_rev == 0x01 ||
+			ctx->asic_id.hw_internal_rev == 0x80) {
+		dc->debug.disable_dpp_power_gate = true;
+		dc->debug.disable_hubp_power_gate = true;
+	}
+
 	/* read VBIOS LTTPR caps */
 	{
 		if (ctx->dc_bios->funcs->get_lttpr_caps) {
--- a/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn315/dcn315_resource.c
@ -1742,6 +1742,7 @@ static int dcn315_populate_dml_pipes_from_context(
 			/* Do not use asymetric crb if not enough for pstate support */
 			if (remaining_det_segs < 0) {
 				pipes[pipe_cnt].pipe.src.det_size_override = 0;
+				pipe_cnt++;
 				continue;
 			}

--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.c
@ -1141,10 +1141,9 @@ void dcn32_update_odm(struct dc *dc, struct dc_state *context, struct pipe_ctx *
 	}
 }

-unsigned int dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div)
+void dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div)
 {
 	struct dc_stream_state *stream = pipe_ctx->stream;
-	unsigned int odm_combine_factor = 0;
 	bool two_pix_per_container = false;

 	// For phantom pipes, use the same programming as the main pipes
@ -1152,7 +1151,6 @@ unsigned int dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsign
 		stream = pipe_ctx->stream->mall_stream_config.paired_stream;
 	}
 	two_pix_per_container = optc2_is_two_pixels_per_containter(&stream->timing);
-	odm_combine_factor = get_odm_config(pipe_ctx, NULL);

 	if (stream->ctx->dc->link_srv->dp_is_128b_132b_signal(pipe_ctx)) {
 		*k1_div = PIXEL_RATE_DIV_BY_1;
@ -1170,15 +1168,13 @@ unsigned int dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsign
 		} else {
 			*k1_div = PIXEL_RATE_DIV_BY_1;
 			*k2_div = PIXEL_RATE_DIV_BY_4;
-			if ((odm_combine_factor == 2) || dcn32_is_dp_dig_pixel_rate_div_policy(pipe_ctx))
+			if (dcn32_is_dp_dig_pixel_rate_div_policy(pipe_ctx))
 				*k2_div = PIXEL_RATE_DIV_BY_2;
 		}
 	}

 	if ((*k1_div == PIXEL_RATE_DIV_NA) && (*k2_div == PIXEL_RATE_DIV_NA))
 		ASSERT(false);
-
-	return odm_combine_factor;
 }

 void dcn32_set_pixels_per_cycle(struct pipe_ctx *pipe_ctx)
@ -1211,7 +1207,8 @@ void dcn32_resync_fifo_dccg_dio(struct dce_hwseq *hws, struct dc *dc, struct dc_
 		if (pipe->top_pipe || pipe->prev_odm_pipe)
 			continue;

-		if (pipe->stream && (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal))) {
+		if (pipe->stream && (pipe->stream->dpms_off || dc_is_virtual_signal(pipe->stream->signal))
+			&& pipe->stream->mall_stream_config.type != SUBVP_PHANTOM) {
 			pipe->stream_res.tg->funcs->disable_crtc(pipe->stream_res.tg);
 			reset_sync_context_for_pipe(dc, context, i);
 			otg_disabled[i] = true;
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_hwseq.h
@ -71,7 +71,7 @@ void dcn32_update_force_pstate(struct dc *dc, struct dc_state *context);

 void dcn32_update_odm(struct dc *dc, struct dc_state *context, struct pipe_ctx *pipe_ctx);

-unsigned int dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div);
+void dcn32_calculate_dccg_k1_k2_values(struct pipe_ctx *pipe_ctx, unsigned int *k1_div, unsigned int *k2_div);

 void dcn32_set_pixels_per_cycle(struct pipe_ctx *pipe_ctx);

--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_optc.c
@ -106,8 +106,11 @@ void optc32_set_h_timing_div_manual_mode(struct timing_generator *optc, bool man
 			OTG_H_TIMING_DIV_MODE_MANUAL, manual_mode ? 1 : 0);
 }
 /**
- * Enable CRTC
- * Enable CRTC - call ASIC Control Object to enable Timing generator.
+ * optc32_enable_crtc() - Enable CRTC - call ASIC Control Object to enable Timing generator.
+ *
+ * @optc: timing_generator instance.
+ *
+ * Return: If CRTC is enabled, return true.
 */
 static bool optc32_enable_crtc(struct timing_generator *optc)
 {
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.c
@ -728,7 +728,7 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.fpo_vactive_margin_us = 2000, // 2000us
 	.disable_fpo_vactive = false,
 	.disable_boot_optimizations = false,
-	.disable_subvp_high_refresh = true,
+	.disable_subvp_high_refresh = false,
 	.disable_dp_plus_plus_wa = true,
 	.fpo_vactive_min_active_margin_us = 200,
 	.fpo_vactive_max_blank_us = 1000,
@ -2175,6 +2175,7 @@ static bool dcn32_resource_construct(
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
 	dc->caps.seamless_odm = true;
+	dc->caps.max_v_total = (1 << 15) - 1;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource.h
@ -169,6 +169,10 @@ double dcn32_determine_max_vratio_prefetch(struct dc *dc, struct dc_state *conte

 bool dcn32_check_native_scaling_for_res(struct pipe_ctx *pipe, unsigned int width, unsigned int height);

+bool dcn32_subvp_drr_admissable(struct dc *dc, struct dc_state *context);
+
+bool dcn32_subvp_vblank_admissable(struct dc *dc, struct dc_state *context, int vlevel);
+
 /* definitions for run time init of reg offsets */

 /* CLK SRC */
--- a/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn32/dcn32_resource_helpers.c
@ -255,8 +255,6 @@ bool dcn32_is_psr_capable(struct pipe_ctx *pipe)
 	return psr_capable;
 }

-#define DCN3_2_NEW_DET_OVERRIDE_MIN_MULTIPLIER 7
-
 /**
 * dcn32_determine_det_override(): Determine DET allocation for each pipe
 *
@ -267,6 +265,7 @@ bool dcn32_is_psr_capable(struct pipe_ctx *pipe)
 * If there is a plane that's driven by more than 1 pipe (i.e. pipe split), then the
 * number of DET for that given plane will be split among the pipes driving that plane.
 *
+ *
 * High level algorithm:
 * 1. Split total DET among number of streams
 * 2. For each stream, split DET among the planes
@ -274,18 +273,6 @@ bool dcn32_is_psr_capable(struct pipe_ctx *pipe)
 *    among those pipes.
 * 4. Assign the DET override to the DML pipes.
 *
- * Special cases:
- *
- * For two displays that have a large difference in pixel rate, we may experience
- *  underflow on the larger display when we divide the DET equally. For this, we
- *  will implement a modified algorithm to assign more DET to larger display.
- *
- * 1. Calculate difference in pixel rates ( multiplier ) between two displays
- * 2. If the multiplier exceeds DCN3_2_NEW_DET_OVERRIDE_MIN_MULTIPLIER, then
- *    implement the modified DET override algorithm.
- * 3. Assign smaller DET size for lower pixel display and higher DET size for
- *    higher pixel display
- *
 * @dc: Current DC state
 * @context: New DC state to be programmed
 * @pipes: Array of DML pipes
@ -303,31 +290,10 @@ void dcn32_determine_det_override(struct dc *dc,
 	struct dc_plane_state *current_plane = NULL;
 	uint8_t stream_count = 0;

-	int phy_pix_clk_mult, lower_mode_stream_index;
-	int phy_pix_clk[MAX_PIPES] = {0};
-	bool use_new_det_override_algorithm = false;
-
 	for (i = 0; i < context->stream_count; i++) {
 		/* Don't count SubVP streams for DET allocation */
-		if (context->streams[i]->mall_stream_config.type != SUBVP_PHANTOM) {
-			phy_pix_clk[i] = context->streams[i]->phy_pix_clk;
+		if (context->streams[i]->mall_stream_config.type != SUBVP_PHANTOM)
 			stream_count++;
-		}
-	}
-
-	/* Check for special case with two displays, one with much higher pixel rate */
-	if (stream_count == 2) {
-		ASSERT((phy_pix_clk[0] > 0) && (phy_pix_clk[1] > 0));
-		if (phy_pix_clk[0] < phy_pix_clk[1]) {
-			lower_mode_stream_index = 0;
-			phy_pix_clk_mult = phy_pix_clk[1] / phy_pix_clk[0];
-		} else {
-			lower_mode_stream_index = 1;
-			phy_pix_clk_mult = phy_pix_clk[0] / phy_pix_clk[1];
-		}
-
-		if (phy_pix_clk_mult >= DCN3_2_NEW_DET_OVERRIDE_MIN_MULTIPLIER)
-			use_new_det_override_algorithm = true;
 	}

 	if (stream_count > 0) {
@ -336,13 +302,6 @@ void dcn32_determine_det_override(struct dc *dc,
 			if (context->streams[i]->mall_stream_config.type == SUBVP_PHANTOM)
 				continue;

-			if (use_new_det_override_algorithm) {
-				if (i == lower_mode_stream_index)
-					stream_segments = 4;
-				else
-					stream_segments = 14;
-			}
-
 			if (context->stream_status[i].plane_count > 0)
 				plane_segments = stream_segments / context->stream_status[i].plane_count;
 			else
@ -660,3 +619,105 @@ bool dcn32_check_native_scaling_for_res(struct pipe_ctx *pipe, unsigned int widt

 	return is_native_scaling;
 }
+
+/**
+ * dcn32_subvp_drr_admissable() - Determine if SubVP + DRR config is admissible
+ *
+ * @dc: Current DC state
+ * @context: New DC state to be programmed
+ *
+ * SubVP + DRR is admissible under the following conditions:
+ * - Config must have 2 displays (i.e., 2 non-phantom master pipes)
+ * - One display is SubVP
+ * - Other display must have Freesync enabled
+ * - The potential DRR display must not be PSR capable
+ *
+ * Return: True if admissible, false otherwise
+ */
+bool dcn32_subvp_drr_admissable(struct dc *dc, struct dc_state *context)
+{
+	bool result = false;
+	uint32_t i;
+	uint8_t subvp_count = 0;
+	uint8_t non_subvp_pipes = 0;
+	bool drr_pipe_found = false;
+	bool drr_psr_capable = false;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+
+		if (pipe->plane_state && !pipe->top_pipe) {
+			if (pipe->stream->mall_stream_config.type == SUBVP_MAIN)
+				subvp_count++;
+			if (pipe->stream->mall_stream_config.type == SUBVP_NONE) {
+				non_subvp_pipes++;
+				drr_psr_capable = (drr_psr_capable || dcn32_is_psr_capable(pipe));
+				if (pipe->stream->ignore_msa_timing_param &&
+						(pipe->stream->allow_freesync || pipe->stream->vrr_active_variable)) {
+					drr_pipe_found = true;
+				}
+			}
+		}
+	}
+
+	if (subvp_count == 1 && non_subvp_pipes == 1 && drr_pipe_found && !drr_psr_capable)
+		result = true;
+
+	return result;
+}
+
+/**
+ * dcn32_subvp_vblank_admissable() - Determine if SubVP + Vblank config is admissible
+ *
+ * @dc: Current DC state
+ * @context: New DC state to be programmed
+ * @vlevel: Voltage level calculated by DML
+ *
+ * SubVP + Vblank is admissible under the following conditions:
+ * - Config must have 2 displays (i.e., 2 non-phantom master pipes)
+ * - One display is SubVP
+ * - Other display must not have Freesync capability
+ * - DML must have output DRAM clock change support as SubVP + Vblank
+ * - The potential vblank display must not be PSR capable
+ *
+ * Return: True if admissible, false otherwise
+ */
+bool dcn32_subvp_vblank_admissable(struct dc *dc, struct dc_state *context, int vlevel)
+{
+	bool result = false;
+	uint32_t i;
+	uint8_t subvp_count = 0;
+	uint8_t non_subvp_pipes = 0;
+	bool drr_pipe_found = false;
+	struct vba_vars_st *vba = &context->bw_ctx.dml.vba;
+	bool vblank_psr_capable = false;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+
+		if (pipe->plane_state && !pipe->top_pipe) {
+			if (pipe->stream->mall_stream_config.type == SUBVP_MAIN)
+				subvp_count++;
+			if (pipe->stream->mall_stream_config.type == SUBVP_NONE) {
+				non_subvp_pipes++;
+				vblank_psr_capable = (vblank_psr_capable || dcn32_is_psr_capable(pipe));
+				if (pipe->stream->ignore_msa_timing_param &&
+						(pipe->stream->allow_freesync || pipe->stream->vrr_active_variable)) {
+					drr_pipe_found = true;
+				}
+			}
+		}
+	}
+
+	if (subvp_count == 1 && non_subvp_pipes == 1 && !drr_pipe_found && !vblank_psr_capable &&
+			vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_vblank_w_mall_sub_vp)
+		result = true;
+
+	return result;
+}
--- a/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c
+++ b/drivers/gpu/drm/amd/display/dc/dcn321/dcn321_resource.c
@ -727,7 +727,7 @@ static const struct dc_debug_options debug_defaults_drv = {
 	.fpo_vactive_margin_us = 2000, // 2000us
 	.disable_fpo_vactive = false,
 	.disable_boot_optimizations = false,
-	.disable_subvp_high_refresh = true,
+	.disable_subvp_high_refresh = false,
 	.fpo_vactive_min_active_margin_us = 200,
 	.fpo_vactive_max_blank_us = 1000,
 };
@ -1718,6 +1718,7 @@ static bool dcn321_resource_construct(
 	dc->caps.edp_dsc_support = true;
 	dc->caps.extended_aux_timeout_support = true;
 	dc->caps.dmcub_support = true;
+	dc->caps.max_v_total = (1 << 15) - 1;

 	/* Color pipeline capabilities */
 	dc->caps.color.dpp.dcn_arch = 1;
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn314/display_mode_vba_314.c
@ -5557,6 +5557,65 @@ void dml314_ModeSupportAndSystemConfigurationFull(struct display_mode_lib *mode_
 			}
 		}
 	}
+	for (i = v->soc.num_states; i >= 0; i--) {
+		for (j = 0; j < 2; j++) {
+			enum dm_validation_status status = DML_VALIDATION_OK;
+
+			if (!v->ScaleRatioAndTapsSupport) {
+				status = DML_FAIL_SCALE_RATIO_TAP;
+			} else if (!v->SourceFormatPixelAndScanSupport) {
+				status = DML_FAIL_SOURCE_PIXEL_FORMAT;
+			} else if (!v->ViewportSizeSupport[i][j]) {
+				status = DML_FAIL_VIEWPORT_SIZE;
+			} else if (P2IWith420) {
+				status = DML_FAIL_P2I_WITH_420;
+			} else if (DSCOnlyIfNecessaryWithBPP) {
+				status = DML_FAIL_DSC_ONLY_IF_NECESSARY_WITH_BPP;
+			} else if (DSC422NativeNotSupported) {
+				status = DML_FAIL_NOT_DSC422_NATIVE;
+			} else if (!v->ODMCombine4To1SupportCheckOK[i]) {
+				status = DML_FAIL_ODM_COMBINE4TO1;
+			} else if (v->NotEnoughDSCUnits[i]) {
+				status = DML_FAIL_NOT_ENOUGH_DSC;
+			} else if (!v->ROBSupport[i][j]) {
+				status = DML_FAIL_REORDERING_BUFFER;
+			} else if (!v->DISPCLK_DPPCLK_Support[i][j]) {
+				status = DML_FAIL_DISPCLK_DPPCLK;
+			} else if (!v->TotalAvailablePipesSupport[i][j]) {
+				status = DML_FAIL_TOTAL_AVAILABLE_PIPES;
+			} else if (!EnoughWritebackUnits) {
+				status = DML_FAIL_ENOUGH_WRITEBACK_UNITS;
+			} else if (!v->WritebackLatencySupport) {
+				status = DML_FAIL_WRITEBACK_LATENCY;
+			} else if (!v->WritebackScaleRatioAndTapsSupport) {
+				status = DML_FAIL_WRITEBACK_SCALE_RATIO_TAP;
+			} else if (!v->CursorSupport) {
+				status = DML_FAIL_CURSOR_SUPPORT;
+			} else if (!v->PitchSupport) {
+				status = DML_FAIL_PITCH_SUPPORT;
+			} else if (ViewportExceedsSurface) {
+				status = DML_FAIL_VIEWPORT_EXCEEDS_SURFACE;
+			} else if (!v->PrefetchSupported[i][j]) {
+				status = DML_FAIL_PREFETCH_SUPPORT;
+			} else if (!v->DynamicMetadataSupported[i][j]) {
+				status = DML_FAIL_DYNAMIC_METADATA;
+			} else if (!v->TotalVerticalActiveBandwidthSupport[i][j]) {
+				status = DML_FAIL_TOTAL_V_ACTIVE_BW;
+			} else if (!v->VRatioInPrefetchSupported[i][j]) {
+				status = DML_FAIL_V_RATIO_PREFETCH;
+			} else if (!v->PTEBufferSizeNotExceeded[i][j]) {
+				status = DML_FAIL_PTE_BUFFER_SIZE;
+			} else if (v->NonsupportedDSCInputBPC) {
+				status = DML_FAIL_DSC_INPUT_BPC;
+			} else if ((v->HostVMEnable
+					&& !v->ImmediateFlipSupportedForState[i][j])) {
+				status = DML_FAIL_HOST_VM_IMMEDIATE_FLIP;
+			} else if (FMTBufferExceeded) {
+				status = DML_FAIL_FMT_BUFFER_EXCEEDED;
+			}
+			mode_lib->vba.ValidationStatus[i] = status;
+		}
+	}

 	{
 		unsigned int MaximumMPCCombine = 0;
--- a/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
+++ b/drivers/gpu/drm/amd/display/dc/dml/dcn32/dcn32_fpu.c
@ -679,7 +679,6 @@ static bool dcn32_assign_subvp_pipe(struct dc *dc,
 	unsigned int max_frame_time = 0;
 	bool valid_assignment_found = false;
 	unsigned int free_pipes = dcn32_get_num_free_pipes(dc, context);
-	bool current_assignment_freesync = false;
 	struct vba_vars_st *vba = &context->bw_ctx.dml.vba;

 	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
@ -720,19 +719,10 @@ static bool dcn32_assign_subvp_pipe(struct dc *dc,
 				struct dc_stream_state *stream = pipe->stream;
 				unsigned int frame_us = (stream->timing.v_total * stream->timing.h_total /
 						(double)(stream->timing.pix_clk_100hz * 100)) * 1000000;
-				if (frame_us > max_frame_time && !stream->ignore_msa_timing_param) {
+				if (frame_us > max_frame_time) {
 					*index = i;
 					max_frame_time = frame_us;
 					valid_assignment_found = true;
-					current_assignment_freesync = false;
-				/* For the 2-Freesync display case, still choose the one with the
-			     * longest frame time
-			     */
-				} else if (stream->ignore_msa_timing_param && (!valid_assignment_found ||
-						(current_assignment_freesync && frame_us > max_frame_time))) {
-					*index = i;
-					valid_assignment_found = true;
-					current_assignment_freesync = true;
 				}
 			}
 		}
@ -864,10 +854,9 @@ static bool subvp_subvp_schedulable(struct dc *dc, struct dc_state *context)
 }

 /**
- * subvp_drr_schedulable - Determine if SubVP + DRR config is schedulable
+ * subvp_drr_schedulable() - Determine if SubVP + DRR config is schedulable
 * @dc: current dc state
 * @context: new dc state
- * @drr_pipe: DRR pipe_ctx for the SubVP + DRR config
 *
 * High level algorithm:
 * 1. Get timing for SubVP pipe, phantom pipe, and DRR pipe
@ -878,11 +867,12 @@ static bool subvp_subvp_schedulable(struct dc *dc, struct dc_state *context)
 *
 * Return: True if the SubVP + DRR config is schedulable, false otherwise
 */
-static bool subvp_drr_schedulable(struct dc *dc, struct dc_state *context, struct pipe_ctx *drr_pipe)
+static bool subvp_drr_schedulable(struct dc *dc, struct dc_state *context)
 {
 	bool schedulable = false;
 	uint32_t i;
 	struct pipe_ctx *pipe = NULL;
+	struct pipe_ctx *drr_pipe = NULL;
 	struct dc_crtc_timing *main_timing = NULL;
 	struct dc_crtc_timing *phantom_timing = NULL;
 	struct dc_crtc_timing *drr_timing = NULL;
@ -908,6 +898,19 @@ static bool subvp_drr_schedulable(struct dc *dc, struct dc_state *context, struc
 			break;
 	}

+	// Find the DRR pipe
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		drr_pipe = &context->res_ctx.pipe_ctx[i];
+
+		// We check for master pipe only
+		if (!drr_pipe->stream || !drr_pipe->plane_state || drr_pipe->top_pipe || drr_pipe->prev_odm_pipe)
+			continue;
+
+		if (drr_pipe->stream->mall_stream_config.type == SUBVP_NONE && drr_pipe->stream->ignore_msa_timing_param &&
+				(drr_pipe->stream->allow_freesync || drr_pipe->stream->vrr_active_variable))
+			break;
+	}
+
 	main_timing = &pipe->stream->timing;
 	phantom_timing = &pipe->stream->mall_stream_config.paired_stream->timing;
 	drr_timing = &drr_pipe->stream->timing;
@ -993,13 +996,7 @@ static bool subvp_vblank_schedulable(struct dc *dc, struct dc_state *context)
 		if (!subvp_pipe && pipe->stream->mall_stream_config.type == SUBVP_MAIN)
 			subvp_pipe = pipe;
 	}
-	// Use ignore_msa_timing_param and VRR active, or Freesync flag to identify as DRR On
-	if (found && context->res_ctx.pipe_ctx[vblank_index].stream->ignore_msa_timing_param &&
-			(context->res_ctx.pipe_ctx[vblank_index].stream->allow_freesync ||
-			context->res_ctx.pipe_ctx[vblank_index].stream->vrr_active_variable)) {
-		// SUBVP + DRR case -- only allowed if run through DRR validation path
-		schedulable = false;
-	} else if (found) {
+	if (found) {
 		main_timing = &subvp_pipe->stream->timing;
 		phantom_timing = &subvp_pipe->stream->mall_stream_config.paired_stream->timing;
 		vblank_timing = &context->res_ctx.pipe_ctx[vblank_index].stream->timing;
@ -1028,6 +1025,53 @@ static bool subvp_vblank_schedulable(struct dc *dc, struct dc_state *context)
 	return schedulable;
 }

+/**
+ * subvp_subvp_admissable() - Determine if subvp + subvp config is admissible
+ *
+ * @dc: Current DC state
+ * @context: New DC state to be programmed
+ *
+ * SubVP + SubVP is admissible under the following conditions:
+ * - All SubVP pipes are < 120Hz OR
+ * - All SubVP pipes are >= 120hz
+ *
+ * Return: True if admissible, false otherwise
+ */
+static bool subvp_subvp_admissable(struct dc *dc,
+				struct dc_state *context)
+{
+	bool result = false;
+	uint32_t i;
+	uint8_t subvp_count = 0;
+	uint32_t min_refresh = subvp_high_refresh_list.min_refresh, max_refresh = 0;
+	uint32_t refresh_rate = 0;
+
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe->stream)
+			continue;
+
+		if (pipe->plane_state && !pipe->top_pipe &&
+				pipe->stream->mall_stream_config.type == SUBVP_MAIN) {
+			refresh_rate = (pipe->stream->timing.pix_clk_100hz * 100 +
+					pipe->stream->timing.v_total * pipe->stream->timing.h_total - 1)
+					/ (double)(pipe->stream->timing.v_total * pipe->stream->timing.h_total);
+			if (refresh_rate < min_refresh)
+				min_refresh = refresh_rate;
+			if (refresh_rate > max_refresh)
+				max_refresh = refresh_rate;
+			subvp_count++;
+		}
+	}
+
+	if (subvp_count == 2 && ((min_refresh < 120 && max_refresh < 120) ||
+			(min_refresh >= 120 && max_refresh >= 120)))
+		result = true;
+
+	return result;
+}
+
 /**
 * subvp_validate_static_schedulability - Check which SubVP case is calculated
 * and handle static analysis based on the case.
@ -1046,11 +1090,12 @@ static bool subvp_validate_static_schedulability(struct dc *dc,
 				struct dc_state *context,
 				int vlevel)
 {
-	bool schedulable = true;	// true by default for single display case
+	bool schedulable = false;
 	struct vba_vars_st *vba = &context->bw_ctx.dml.vba;
 	uint32_t i, pipe_idx;
 	uint8_t subvp_count = 0;
 	uint8_t vactive_count = 0;
+	uint8_t non_subvp_pipes = 0;

 	for (i = 0, pipe_idx = 0; i < dc->res_pool->pipe_count; i++) {
 		struct pipe_ctx *pipe = &context->res_ctx.pipe_ctx[i];
@ -1058,14 +1103,18 @@ static bool subvp_validate_static_schedulability(struct dc *dc,
 		if (!pipe->stream)
 			continue;

-		if (pipe->plane_state && !pipe->top_pipe &&
-				pipe->stream->mall_stream_config.type == SUBVP_MAIN)
-			subvp_count++;
+		if (pipe->plane_state && !pipe->top_pipe) {
+			if (pipe->stream->mall_stream_config.type == SUBVP_MAIN)
+				subvp_count++;
+			if (pipe->stream->mall_stream_config.type == SUBVP_NONE) {
+				non_subvp_pipes++;
+			}
+		}

 		// Count how many planes that aren't SubVP/phantom are capable of VACTIVE
 		// switching (SubVP + VACTIVE unsupported). In situations where we force
 		// SubVP for a VACTIVE plane, we don't want to increment the vactive_count.
-		if (vba->ActiveDRAMClockChangeLatencyMargin[vba->pipe_plane[pipe_idx]] > 0 &&
+		if (vba->ActiveDRAMClockChangeLatencyMarginPerState[vlevel][vba->maxMpcComb][vba->pipe_plane[pipe_idx]] > 0 &&
 		    pipe->stream->mall_stream_config.type == SUBVP_NONE) {
 			vactive_count++;
 		}
@ -1074,13 +1123,14 @@ static bool subvp_validate_static_schedulability(struct dc *dc,

 	if (subvp_count == 2) {
 		// Static schedulability check for SubVP + SubVP case
-		schedulable = subvp_subvp_schedulable(dc, context);
-	} else if (vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_vblank_w_mall_sub_vp) {
-		// Static schedulability check for SubVP + VBLANK case. Also handle the case where
-		// DML outputs SubVP + VBLANK + VACTIVE (DML will report as SubVP + VBLANK)
-		if (vactive_count > 0)
-			schedulable = false;
-		else
+		schedulable = subvp_subvp_admissable(dc, context) && subvp_subvp_schedulable(dc, context);
+	} else if (subvp_count == 1 && non_subvp_pipes == 0) {
+		// Single SubVP configs will be supported by default as long as it's suppported by DML
+		schedulable = true;
+	} else if (subvp_count == 1 && non_subvp_pipes == 1) {
+		if (dcn32_subvp_drr_admissable(dc, context))
+			schedulable = subvp_drr_schedulable(dc, context);
+		else if (dcn32_subvp_vblank_admissable(dc, context, vlevel))
 			schedulable = subvp_vblank_schedulable(dc, context);
 	} else if (vba->DRAMClockChangeSupport[vlevel][vba->maxMpcComb] == dm_dram_clock_change_vactive_w_mall_sub_vp &&
 			vactive_count > 0) {
@ -1104,10 +1154,6 @@ static void dcn32_full_validate_bw_helper(struct dc *dc,
 	unsigned int dc_pipe_idx = 0;
 	int i = 0;
 	bool found_supported_config = false;
-	struct pipe_ctx *pipe = NULL;
-	uint32_t non_subvp_pipes = 0;
-	bool drr_pipe_found = false;
-	uint32_t drr_pipe_index = 0;

 	dc_assert_fp_enabled();

@ -1197,31 +1243,12 @@ static void dcn32_full_validate_bw_helper(struct dc *dc,
 				}
 			}

-			if (*vlevel < context->bw_ctx.dml.soc.num_states &&
-			    vba->DRAMClockChangeSupport[*vlevel][vba->maxMpcComb] != dm_dram_clock_change_unsupported
-			    && subvp_validate_static_schedulability(dc, context, *vlevel)) {
+			if (*vlevel < context->bw_ctx.dml.soc.num_states
+			    && subvp_validate_static_schedulability(dc, context, *vlevel))
 				found_supported_config = true;
-			} else if (*vlevel < context->bw_ctx.dml.soc.num_states) {
-				/* Case where 1 SubVP is added, and DML reports MCLK unsupported or DRR is allowed.
-				 * This handles the case for SubVP + DRR, where the DRR display does not support MCLK
-				 * switch at it's native refresh rate / timing, or DRR is allowed for the non-subvp
-				 * display.
-				 */
-				for (i = 0; i < dc->res_pool->pipe_count; i++) {
-					pipe = &context->res_ctx.pipe_ctx[i];
-					if (pipe->stream && pipe->plane_state && !pipe->top_pipe &&
-					    pipe->stream->mall_stream_config.type == SUBVP_NONE) {
-						non_subvp_pipes++;
-						// Use ignore_msa_timing_param flag to identify as DRR
-						if (pipe->stream->ignore_msa_timing_param && pipe->stream->allow_freesync) {
-							drr_pipe_found = true;
-							drr_pipe_index = i;
-						}
-					}
-				}
-				// If there is only 1 remaining non SubVP pipe that is DRR, check static
-				// schedulability for SubVP + DRR.
-				if (non_subvp_pipes == 1 && drr_pipe_found) {
+			if (found_supported_config) {
+				// For SubVP + DRR cases, we can force the lowest vlevel that supports the mode
+				if (dcn32_subvp_drr_admissable(dc, context) && subvp_drr_schedulable(dc, context)) {
 					/* find lowest vlevel that supports the config */
 					for (i = *vlevel; i >= 0; i--) {
 						if (vba->ModeSupport[i][vba->maxMpcComb]) {
@ -1230,9 +1257,6 @@ static void dcn32_full_validate_bw_helper(struct dc *dc,
 							break;
 						}
 					}
-
-					found_supported_config = subvp_drr_schedulable(dc, context,
-										       &context->res_ctx.pipe_ctx[drr_pipe_index]);
 				}
 			}
 		}
@ -2882,16 +2906,34 @@ bool dcn32_allow_subvp_high_refresh_rate(struct dc *dc, struct dc_state *context
 {
 	bool allow = false;
 	uint32_t refresh_rate = 0;
-	uint32_t min_refresh = subvp_high_refresh_list.min_refresh;
-	uint32_t max_refresh = subvp_high_refresh_list.max_refresh;
+	uint32_t subvp_min_refresh = subvp_high_refresh_list.min_refresh;
+	uint32_t subvp_max_refresh = subvp_high_refresh_list.max_refresh;
+	uint32_t min_refresh = subvp_max_refresh;
 	uint32_t i;

-	if (!dc->debug.disable_subvp_high_refresh && pipe->stream &&
+	/* Only allow SubVP on high refresh displays if all connected displays
+	 * are considered "high refresh" (i.e. >= 120hz). We do not want to
+	 * allow combinations such as 120hz (SubVP) + 60hz (SubVP).
+	 */
+	for (i = 0; i < dc->res_pool->pipe_count; i++) {
+		struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[i];
+
+		if (!pipe_ctx->stream)
+			continue;
+		refresh_rate = (pipe_ctx->stream->timing.pix_clk_100hz * 100 +
+				pipe_ctx->stream->timing.v_total * pipe_ctx->stream->timing.h_total - 1)
+						/ (double)(pipe_ctx->stream->timing.v_total * pipe_ctx->stream->timing.h_total);
+
+		if (refresh_rate < min_refresh)
+			min_refresh = refresh_rate;
+	}
+
+	if (!dc->debug.disable_subvp_high_refresh && min_refresh >= subvp_min_refresh && pipe->stream &&
 			pipe->plane_state && !(pipe->stream->vrr_active_variable || pipe->stream->vrr_active_fixed)) {
 		refresh_rate = (pipe->stream->timing.pix_clk_100hz * 100 +
 						pipe->stream->timing.v_total * pipe->stream->timing.h_total - 1)
 						/ (double)(pipe->stream->timing.v_total * pipe->stream->timing.h_total);
-		if (refresh_rate >= min_refresh && refresh_rate <= max_refresh) {
+		if (refresh_rate >= subvp_min_refresh && refresh_rate <= subvp_max_refresh) {
 			for (i = 0; i < SUBVP_HIGH_REFRESH_LIST_LEN; i++) {
 				uint32_t width = subvp_high_refresh_list.res[i].width;
 				uint32_t height = subvp_high_refresh_list.res[i].height;
--- a/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h
+++ b/drivers/gpu/drm/amd/display/dc/dml/display_mode_enums.h
@ -190,6 +190,14 @@ enum dm_validation_status {
 	DML_FAIL_DSC_INPUT_BPC,
 	DML_FAIL_PREFETCH_SUPPORT,
 	DML_FAIL_V_RATIO_PREFETCH,
+	DML_FAIL_P2I_WITH_420,
+	DML_FAIL_DSC_ONLY_IF_NECESSARY_WITH_BPP,
+	DML_FAIL_NOT_DSC422_NATIVE,
+	DML_FAIL_ODM_COMBINE4TO1,
+	DML_FAIL_ENOUGH_WRITEBACK_UNITS,
+	DML_FAIL_VIEWPORT_EXCEEDS_SURFACE,
+	DML_FAIL_DYNAMIC_METADATA,
+	DML_FAIL_FMT_BUFFER_EXCEEDED,
 };

 enum writeback_config {
--- a/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw/dccg.h
@ -68,6 +68,7 @@ struct dccg {
 	const struct dccg_funcs *funcs;
 	int pipe_dppclk_khz[MAX_PIPES];
 	int ref_dppclk;
+	bool dpp_clock_gated[MAX_PIPES];
 	//int dtbclk_khz[MAX_PIPES];/* TODO needs to be removed */
 	//int audio_dtbclk_khz;/* TODO needs to be removed */
 	//int ref_dtbclk_khz;/* TODO needs to be removed */
--- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer.h
@ -114,6 +114,26 @@ struct update_visual_confirm_params {
 	int mpcc_id;
 };

+struct power_on_mpc_mem_pwr_params {
+	struct mpc *mpc;
+	int mpcc_id;
+	bool power_on;
+};
+
+struct set_output_csc_params {
+	struct mpc *mpc;
+	int opp_id;
+	const uint16_t *regval;
+	enum mpc_output_csc_mode ocsc_mode;
+};
+
+struct set_ocsc_default_params {
+	struct mpc *mpc;
+	int opp_id;
+	enum dc_color_space color_space;
+	enum mpc_output_csc_mode ocsc_mode;
+};
+
 union block_sequence_params {
 	struct update_plane_addr_params update_plane_addr_params;
 	struct subvp_pipe_control_lock_fast_params subvp_pipe_control_lock_fast_params;
@ -128,6 +148,9 @@ union block_sequence_params {
 	struct program_bias_and_scale_params program_bias_and_scale_params;
 	struct set_output_transfer_func_params set_output_transfer_func_params;
 	struct update_visual_confirm_params update_visual_confirm_params;
+	struct power_on_mpc_mem_pwr_params power_on_mpc_mem_pwr_params;
+	struct set_output_csc_params set_output_csc_params;
+	struct set_ocsc_default_params set_ocsc_default_params;
 };

 enum block_sequence_func {
@ -144,6 +167,9 @@ enum block_sequence_func {
 	DPP_PROGRAM_BIAS_AND_SCALE,
 	DPP_SET_OUTPUT_TRANSFER_FUNC,
 	MPC_UPDATE_VISUAL_CONFIRM,
+	MPC_POWER_ON_MPC_MEM_PWR,
+	MPC_SET_OUTPUT_CSC,
+	MPC_SET_OCSC_DEFAULT,
 };

 struct block_sequence {
@ -439,4 +465,10 @@ void hwss_setup_dpp(union block_sequence_params *params);

 void hwss_program_bias_and_scale(union block_sequence_params *params);

+void hwss_power_on_mpc_mem_pwr(union block_sequence_params *params);
+
+void hwss_set_output_csc(union block_sequence_params *params);
+
+void hwss_set_ocsc_default(union block_sequence_params *params);
+
 #endif /* __DC_HW_SEQUENCER_H__ */
--- a/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer_private.h
+++ b/drivers/gpu/drm/amd/display/dc/inc/hw_sequencer_private.h
@ -156,7 +156,7 @@ struct hwseq_private_funcs {
 	void (*program_mall_pipe_config)(struct dc *dc, struct dc_state *context);
 	void (*update_force_pstate)(struct dc *dc, struct dc_state *context);
 	void (*update_mall_sel)(struct dc *dc, struct dc_state *context);
-	unsigned int (*calculate_dccg_k1_k2_values)(struct pipe_ctx *pipe_ctx,
+	void (*calculate_dccg_k1_k2_values)(struct pipe_ctx *pipe_ctx,
 			unsigned int *k1_div,
 			unsigned int *k2_div);
 	void (*set_pixels_per_cycle)(struct pipe_ctx *pipe_ctx);
--- a/drivers/gpu/drm/amd/display/dc/link/link_detection.c
+++ b/drivers/gpu/drm/amd/display/dc/link/link_detection.c
@ -984,6 +984,11 @@ static bool detect_link_and_local_sink(struct dc_link *link,
 					(link->dpcd_caps.dongle_type !=
 							DISPLAY_DONGLE_DP_HDMI_CONVERTER))
 				converter_disable_audio = true;
+
+			/* limited link rate to HBR3 for DPIA until we implement USB4 V2 */
+			if (link->ep_type == DISPLAY_ENDPOINT_USB4_DPIA &&
+					link->reported_link_cap.link_rate > LINK_RATE_HIGH3)
+				link->reported_link_cap.link_rate = LINK_RATE_HIGH3;
 			break;
 		}

--- a/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
+++ b/drivers/gpu/drm/amd/display/modules/freesync/freesync.c
@ -1,5 +1,5 @@
 /*
- * Copyright 2016 Advanced Micro Devices, Inc.
+ * Copyright 2016-2023 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
@ -989,6 +989,7 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync,
 	unsigned int refresh_range = 0;
 	unsigned long long min_refresh_in_uhz = 0;
 	unsigned long long max_refresh_in_uhz = 0;
+	unsigned long long min_hardware_refresh_in_uhz = 0;

 	if (mod_freesync == NULL)
 		return;
@ -999,7 +1000,13 @@ void mod_freesync_build_vrr_params(struct mod_freesync *mod_freesync,
 	nominal_field_rate_in_uhz =
 			mod_freesync_calc_nominal_field_rate(stream);

-	min_refresh_in_uhz = in_config->min_refresh_in_uhz;
+	if (stream->ctx->dc->caps.max_v_total != 0 && stream->timing.h_total != 0) {
+		min_hardware_refresh_in_uhz = div64_u64((stream->timing.pix_clk_100hz * 100000000ULL),
+			(stream->timing.h_total * stream->ctx->dc->caps.max_v_total));
+	}
+	/* Limit minimum refresh rate to what can be supported by hardware */
+	min_refresh_in_uhz = min_hardware_refresh_in_uhz > in_config->min_refresh_in_uhz ?
+		min_hardware_refresh_in_uhz : in_config->min_refresh_in_uhz;
 	max_refresh_in_uhz = in_config->max_refresh_in_uhz;

 	/* Full range may be larger than current video timing, so cap at nominal */
--- a/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu11/sienna_cichlid_ppt.c
@ -431,7 +431,13 @@ static int sienna_cichlid_append_powerplay_table(struct smu_context *smu)
 {
 	struct atom_smc_dpm_info_v4_9 *smc_dpm_table;
 	int index, ret;
-	I2cControllerConfig_t *table_member;
+	PPTable_beige_goby_t *ppt_beige_goby;
+	PPTable_t *ppt;
+
+	if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13))
+		ppt_beige_goby = smu->smu_table.driver_pptable;
+	else
+		ppt = smu->smu_table.driver_pptable;

 	index = get_index_into_master_table(atom_master_list_of_data_tables_v2_1,
 					    smc_dpm_info);
@ -440,9 +446,13 @@ static int sienna_cichlid_append_powerplay_table(struct smu_context *smu)
 				      (uint8_t **)&smc_dpm_table);
 	if (ret)
 		return ret;
-	GET_PPTABLE_MEMBER(I2cControllers, &table_member);
-	memcpy(table_member, smc_dpm_table->I2cControllers,
-			sizeof(*smc_dpm_table) - sizeof(smc_dpm_table->table_header));
+
+	if (smu->adev->ip_versions[MP1_HWIP][0] == IP_VERSION(11, 0, 13))
+		smu_memcpy_trailing(ppt_beige_goby, I2cControllers, BoardReserved,
+				    smc_dpm_table, I2cControllers);
+	else
+		smu_memcpy_trailing(ppt, I2cControllers, BoardReserved,
+				    smc_dpm_table, I2cControllers);

 	return 0;
 }
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_0_ppt.c
@ -46,7 +46,6 @@
 #include "asic_reg/mp/mp_13_0_0_sh_mask.h"
 #include "smu_cmn.h"
 #include "amdgpu_ras.h"
-#include "umc_v8_10.h"

 /*
 * DO NOT use these for err/warn/info/debug messages.
@ -2609,7 +2608,7 @@ static ssize_t smu_v13_0_0_get_ecc_info(struct smu_context *smu,

 	ecc_table = (EccInfoTable_t *)smu_table->ecc_table;

-	for (i = 0; i < UMC_V8_10_TOTAL_CHANNEL_NUM(adev); i++) {
+	for (i = 0; i < ARRAY_SIZE(ecc_table->EccInfo); i++) {
 		ecc_info_per_channel = &(eccinfo->ecc[i]);
 		ecc_info_per_channel->ce_count_lo_chip =
 				ecc_table->EccInfo[i].ce_count_lo_chip;
--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.c
@ -831,6 +831,8 @@ static int smu_v13_0_5_set_soft_freq_limited_range(struct smu_context *smu,
 							uint32_t max)
 {
 	enum smu_message_type msg_set_min, msg_set_max;
+	uint32_t min_clk = min;
+	uint32_t max_clk = max;
 	int ret = 0;

 	if (!smu_v13_0_5_clk_dpm_is_enabled(smu, clk_type))
@ -851,11 +853,16 @@ static int smu_v13_0_5_set_soft_freq_limited_range(struct smu_context *smu,
 		return -EINVAL;
 	}

-	ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_min, min, NULL);
+	if (clk_type == SMU_VCLK) {
+		min_clk = min << SMU_13_VCLK_SHIFT;
+		max_clk = max << SMU_13_VCLK_SHIFT;
+	}
+
+	ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_min, min_clk, NULL);
 	if (ret)
 		goto out;

-	ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_max, max, NULL);
+	ret = smu_cmn_send_smc_msg_with_param(smu, msg_set_max, max_clk, NULL);
 	if (ret)
 		goto out;

@ -971,31 +978,79 @@ static int smu_v13_0_5_force_clk_levels(struct smu_context *smu,
 	return ret;
 }

+static int smu_v13_0_5_get_dpm_profile_freq(struct smu_context *smu,
+					enum amd_dpm_forced_level level,
+					enum smu_clk_type clk_type,
+					uint32_t *min_clk,
+					uint32_t *max_clk)
+{
+	int ret = 0;
+	uint32_t clk_limit = 0;
+
+	switch (clk_type) {
+	case SMU_GFXCLK:
+	case SMU_SCLK:
+		clk_limit = SMU_13_0_5_UMD_PSTATE_GFXCLK;
+		if (level == AMD_DPM_FORCED_LEVEL_PROFILE_PEAK)
+			smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_SCLK, NULL, &clk_limit);
+		else if (level == AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK)
+			smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_SCLK, &clk_limit, NULL);
+		break;
+	case SMU_VCLK:
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_VCLK, NULL, &clk_limit);
+		break;
+	case SMU_DCLK:
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_DCLK, NULL, &clk_limit);
+		break;
+	default:
+		ret = -EINVAL;
+		break;
+	}
+	*min_clk = *max_clk = clk_limit;
+	return ret;
+}
+
 static int smu_v13_0_5_set_performance_level(struct smu_context *smu,
 						enum amd_dpm_forced_level level)
 {
 	struct amdgpu_device *adev = smu->adev;
 	uint32_t sclk_min = 0, sclk_max = 0;
+	uint32_t vclk_min = 0, vclk_max = 0;
+	uint32_t dclk_min = 0, dclk_max = 0;
 	int ret = 0;

 	switch (level) {
 	case AMD_DPM_FORCED_LEVEL_HIGH:
 		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_SCLK, NULL, &sclk_max);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_VCLK, NULL, &vclk_max);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_DCLK, NULL, &dclk_max);
 		sclk_min = sclk_max;
+		vclk_min = vclk_max;
+		dclk_min = dclk_max;
 		break;
 	case AMD_DPM_FORCED_LEVEL_LOW:
 		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_SCLK, &sclk_min, NULL);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_VCLK, &vclk_min, NULL);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_DCLK, &dclk_min, NULL);
 		sclk_max = sclk_min;
+		vclk_max = vclk_min;
+		dclk_max = dclk_min;
 		break;
 	case AMD_DPM_FORCED_LEVEL_AUTO:
 		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_SCLK, &sclk_min, &sclk_max);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_VCLK, &vclk_min, &vclk_max);
+		smu_v13_0_5_get_dpm_ultimate_freq(smu, SMU_DCLK, &dclk_min, &dclk_max);
 		break;
 	case AMD_DPM_FORCED_LEVEL_PROFILE_STANDARD:
 	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_SCLK:
-	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
 	case AMD_DPM_FORCED_LEVEL_PROFILE_PEAK:
-		/* Temporarily do nothing since the optimal clocks haven't been provided yet */
+		smu_v13_0_5_get_dpm_profile_freq(smu, level, SMU_SCLK, &sclk_min, &sclk_max);
+		smu_v13_0_5_get_dpm_profile_freq(smu, level, SMU_VCLK, &vclk_min, &vclk_max);
+		smu_v13_0_5_get_dpm_profile_freq(smu, level, SMU_DCLK, &dclk_min, &dclk_max);
 		break;
+	case AMD_DPM_FORCED_LEVEL_PROFILE_MIN_MCLK:
+		dev_err(adev->dev, "The performance level profile_min_mclk is not supported.");
+		return -EOPNOTSUPP;
 	case AMD_DPM_FORCED_LEVEL_MANUAL:
 	case AMD_DPM_FORCED_LEVEL_PROFILE_EXIT:
 		return 0;
@ -1016,6 +1071,23 @@ static int smu_v13_0_5_set_performance_level(struct smu_context *smu,
 		smu->gfx_actual_soft_max_freq = sclk_max;
 	}

+	if (vclk_min && vclk_max) {
+		ret = smu_v13_0_5_set_soft_freq_limited_range(smu,
+							      SMU_VCLK,
+							      vclk_min,
+							      vclk_max);
+		if (ret)
+			return ret;
+	}
+
+	if (dclk_min && dclk_max) {
+		ret = smu_v13_0_5_set_soft_freq_limited_range(smu,
+							      SMU_DCLK,
+							      dclk_min,
+							      dclk_max);
+		if (ret)
+			return ret;
+	}
 	return ret;
 }

--- a/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.h
+++ b/drivers/gpu/drm/amd/pm/swsmu/smu13/smu_v13_0_5_ppt.h
@ -24,6 +24,6 @@
 #define __SMU_V13_0_5_PPT_H__

 extern void smu_v13_0_5_set_ppt_funcs(struct smu_context *smu);
-#define SMU_13_0_5_UMD_PSTATE_GFXCLK   1100
+#define SMU_13_0_5_UMD_PSTATE_GFXCLK   700

 #endif
--- a/drivers/gpu/drm/display/drm_dp_mst_topology.c
+++ b/drivers/gpu/drm/display/drm_dp_mst_topology.c
@ -4053,17 +4053,28 @@ static int drm_dp_mst_handle_up_req(struct drm_dp_mst_topology_mgr *mgr)
 }

 /**
- * drm_dp_mst_hpd_irq() - MST hotplug IRQ notify
+ * drm_dp_mst_hpd_irq_handle_event() - MST hotplug IRQ handle MST event
 * @mgr: manager to notify irq for.
 * @esi: 4 bytes from SINK_COUNT_ESI
+ * @ack: 4 bytes used to ack events starting from SINK_COUNT_ESI
 * @handled: whether the hpd interrupt was consumed or not
 *
- * This should be called from the driver when it detects a short IRQ,
+ * This should be called from the driver when it detects a HPD IRQ,
 * along with the value of the DEVICE_SERVICE_IRQ_VECTOR_ESI0. The
- * topology manager will process the sideband messages received as a result
- * of this.
+ * topology manager will process the sideband messages received
+ * as indicated in the DEVICE_SERVICE_IRQ_VECTOR_ESI0 and set the
+ * corresponding flags that Driver has to ack the DP receiver later.
+ *
+ * Note that driver shall also call
+ * drm_dp_mst_hpd_irq_send_new_request() if the 'handled' is set
+ * after calling this function, to try to kick off a new request in
+ * the queue if the previous message transaction is completed.
+ *
+ * See also:
+ * drm_dp_mst_hpd_irq_send_new_request()
 */
-int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, bool *handled)
+int drm_dp_mst_hpd_irq_handle_event(struct drm_dp_mst_topology_mgr *mgr, const u8 *esi,
+				    u8 *ack, bool *handled)
 {
 	int ret = 0;
 	int sc;
@ -4078,18 +4089,47 @@ int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, bool *handl
 	if (esi[1] & DP_DOWN_REP_MSG_RDY) {
 		ret = drm_dp_mst_handle_down_rep(mgr);
 		*handled = true;
+		ack[1] |= DP_DOWN_REP_MSG_RDY;
 	}

 	if (esi[1] & DP_UP_REQ_MSG_RDY) {
 		ret |= drm_dp_mst_handle_up_req(mgr);
 		*handled = true;
+		ack[1] |= DP_UP_REQ_MSG_RDY;
 	}

-	drm_dp_mst_kick_tx(mgr);
 	return ret;
 }
-EXPORT_SYMBOL(drm_dp_mst_hpd_irq);
+EXPORT_SYMBOL(drm_dp_mst_hpd_irq_handle_event);

+/**
+ * drm_dp_mst_hpd_irq_send_new_request() - MST hotplug IRQ kick off new request
+ * @mgr: manager to notify irq for.
+ *
+ * This should be called from the driver when mst irq event is handled
+ * and acked. Note that new down request should only be sent when
+ * previous message transaction is completed. Source is not supposed to generate
+ * interleaved message transactions.
+ */
+void drm_dp_mst_hpd_irq_send_new_request(struct drm_dp_mst_topology_mgr *mgr)
+{
+	struct drm_dp_sideband_msg_tx *txmsg;
+	bool kick = true;
+
+	mutex_lock(&mgr->qlock);
+	txmsg = list_first_entry_or_null(&mgr->tx_msg_downq,
+					 struct drm_dp_sideband_msg_tx, next);
+	/* If last transaction is not completed yet*/
+	if (!txmsg ||
+	    txmsg->state == DRM_DP_SIDEBAND_TX_START_SEND ||
+	    txmsg->state == DRM_DP_SIDEBAND_TX_SENT)
+		kick = false;
+	mutex_unlock(&mgr->qlock);
+
+	if (kick)
+		drm_dp_mst_kick_tx(mgr);
+}
+EXPORT_SYMBOL(drm_dp_mst_hpd_irq_send_new_request);
 /**
 * drm_dp_mst_detect_port() - get connection status for an MST port
 * @connector: DRM connector for this port
--- a/drivers/gpu/drm/i915/display/intel_dp.c
+++ b/drivers/gpu/drm/i915/display/intel_dp.c
@ -4069,9 +4069,7 @@ intel_dp_mst_hpd_irq(struct intel_dp *intel_dp, u8 *esi, u8 *ack)
 {
 	bool handled = false;

-	drm_dp_mst_hpd_irq(&intel_dp->mst_mgr, esi, &handled);
-	if (handled)
-		ack[1] |= esi[1] & (DP_DOWN_REP_MSG_RDY | DP_UP_REQ_MSG_RDY);
+	drm_dp_mst_hpd_irq_handle_event(&intel_dp->mst_mgr, esi, ack, &handled);

 	if (esi[1] & DP_CP_IRQ) {
 		intel_hdcp_handle_cp_irq(intel_dp->attached_connector);
@ -4146,6 +4144,9 @@ intel_dp_check_mst_status(struct intel_dp *intel_dp)

 		if (!intel_dp_ack_sink_irq_esi(intel_dp, ack))
 			drm_dbg_kms(&i915->drm, "Failed to ack ESI\n");
+
+		if (ack[1] & (DP_DOWN_REP_MSG_RDY | DP_UP_REQ_MSG_RDY))
+			drm_dp_mst_hpd_irq_send_new_request(&intel_dp->mst_mgr);
 	}

 	return link_ok;
--- a/drivers/gpu/drm/nouveau/dispnv50/disp.c
+++ b/drivers/gpu/drm/nouveau/dispnv50/disp.c
@ -1359,22 +1359,26 @@ nv50_mstm_service(struct nouveau_drm *drm,
 	u8 esi[8] = {};

 	while (handled) {
+		u8 ack[8] = {};
+
 		rc = drm_dp_dpcd_read(aux, DP_SINK_COUNT_ESI, esi, 8);
 		if (rc != 8) {
 			ret = false;
 			break;
 		}

-		drm_dp_mst_hpd_irq(&mstm->mgr, esi, &handled);
+		drm_dp_mst_hpd_irq_handle_event(&mstm->mgr, esi, ack, &handled);
 		if (!handled)
 			break;

-		rc = drm_dp_dpcd_write(aux, DP_SINK_COUNT_ESI + 1, &esi[1],
-				       3);
-		if (rc != 3) {
+		rc = drm_dp_dpcd_writeb(aux, DP_SINK_COUNT_ESI + 1, ack[1]);
+
+		if (rc != 1) {
 			ret = false;
 			break;
 		}
+
+		drm_dp_mst_hpd_irq_send_new_request(&mstm->mgr);
 	}

 	if (!ret)
--- a/drivers/gpu/drm/radeon/radeon_acpi.h
+++ b/drivers/gpu/drm/radeon/radeon_acpi.h
@ -453,4 +453,13 @@ struct acpi_bus_event;
 * BYTE  - number of active lanes
 */

+#if defined(CONFIG_VGA_SWITCHEROO)
+void radeon_register_atpx_handler(void);
+void radeon_unregister_atpx_handler(void);
+bool radeon_has_atpx_dgpu_power_cntl(void);
+bool radeon_is_atpx_hybrid(void);
+bool radeon_has_atpx(void);
+bool radeon_atpx_dgpu_req_power_for_displays(void);
+#endif
+
 #endif
--- a/drivers/gpu/drm/radeon/radeon_atpx_handler.c
+++ b/drivers/gpu/drm/radeon/radeon_atpx_handler.c
@ -147,7 +147,7 @@ static void radeon_atpx_parse_functions(struct radeon_atpx_functions *f, u32 mas
 }

 /**
- * radeon_atpx_validate_functions - validate ATPX functions
+ * radeon_atpx_validate() - validate ATPX functions
 *
 * @atpx: radeon atpx struct
 *
--- a/drivers/gpu/drm/radeon/radeon_fbdev.c
+++ b/drivers/gpu/drm/radeon/radeon_fbdev.c
@ -304,6 +304,7 @@ static void radeon_fbdev_client_unregister(struct drm_client_dev *client)

 	if (fb_helper->info) {
 		vga_switcheroo_client_fb_set(rdev->pdev, NULL);
+		drm_helper_force_disable_all(dev);
 		drm_fb_helper_unregister_info(fb_helper);
 	} else {
 		drm_client_release(&fb_helper->client);
--- a/drivers/gpu/drm/scheduler/sched_entity.c
+++ b/drivers/gpu/drm/scheduler/sched_entity.c
@ -384,7 +384,7 @@ static bool drm_sched_entity_add_dependency_cb(struct drm_sched_entity *entity)
 	}

 	s_fence = to_drm_sched_fence(fence);
-	if (s_fence && s_fence->sched == sched &&
+	if (!fence->error && s_fence && s_fence->sched == sched &&
 	    !test_bit(DRM_SCHED_FENCE_DONT_PIPELINE, &fence->flags)) {

 		/*
--- a/include/drm/display/drm_dp_mst_helper.h
+++ b/include/drm/display/drm_dp_mst_helper.h
@ -810,8 +810,11 @@ void drm_dp_mst_topology_mgr_destroy(struct drm_dp_mst_topology_mgr *mgr);
 bool drm_dp_read_mst_cap(struct drm_dp_aux *aux, const u8 dpcd[DP_RECEIVER_CAP_SIZE]);
 int drm_dp_mst_topology_mgr_set_mst(struct drm_dp_mst_topology_mgr *mgr, bool mst_state);

-int drm_dp_mst_hpd_irq(struct drm_dp_mst_topology_mgr *mgr, u8 *esi, bool *handled);
-
+int drm_dp_mst_hpd_irq_handle_event(struct drm_dp_mst_topology_mgr *mgr,
+				    const u8 *esi,
+				    u8 *ack,
+				    bool *handled);
+void drm_dp_mst_hpd_irq_send_new_request(struct drm_dp_mst_topology_mgr *mgr);

 int
 drm_dp_mst_detect_port(struct drm_connector *connector,
--- a/include/uapi/linux/kfd_ioctl.h
+++ b/include/uapi/linux/kfd_ioctl.h
@ -39,9 +39,10 @@
 * - 1.11 - Add unified memory for ctx save/restore area
 * - 1.12 - Add DMA buf export ioctl
 * - 1.13 - Add debugger API
+ * - 1.14 - Update kfd_event_data
 */
 #define KFD_IOCTL_MAJOR_VERSION 1
-#define KFD_IOCTL_MINOR_VERSION 13
+#define KFD_IOCTL_MINOR_VERSION 14

 struct kfd_ioctl_get_version_args {
 	__u32 major_version;	/* from KFD */
@ -320,12 +321,20 @@ struct kfd_hsa_hw_exception_data {
 	__u32 gpu_id;
 };

+/* hsa signal event data */
+struct kfd_hsa_signal_event_data {
+	__u64 last_event_age;	/* to and from KFD */
+};
+
 /* Event data */
 struct kfd_event_data {
 	union {
+		/* From KFD */
 		struct kfd_hsa_memory_exception_data memory_exception_data;
 		struct kfd_hsa_hw_exception_data hw_exception_data;
-	};				/* From KFD */
+		/* To and From KFD */
+		struct kfd_hsa_signal_event_data signal_event_data;
+	};
 	__u64 kfd_event_data_ext;	/* pointer to an extension structure
 					   for future exception types */
 	__u32 event_id;		/* to KFD */