Merge tag 'amd-drm-next-6.1-2022-09-23' of https://gitlab.freedesktop.org/agd5f/linux into drm-next

amd-drm-next-6.1-2022-09-23:

amdgpu:
- SDMA fix
- Add new firmware types to debugfs/IOCTL version queries
- Misc spelling and grammar fixes
- Misc code cleanups
- DCN 3.2.x fixes
- DCN 3.1.x fixes
- CS cleanup
- Gang submit support
- Clang fixes
- Non-DC audio fix
- GPUVM locking fixes
- Vega10 PWN fan speed fix

amdkgd:
- MQD manager cleanup
- Misc spelling and grammar fixes

UAPI:
- Add new firmware types to the FW version query IOCTL

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220923215729.6061-1-alexander.deucher@amd.com

27 files changed, 864 insertions, 409 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 79bb6fd83094..ae9371b172e3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -885,6 +885,7 @@ struct amdgpu_device {
 	u64				fence_context;
 	unsigned			num_rings;
 	struct amdgpu_ring		*rings[AMDGPU_MAX_RINGS];
+	struct dma_fence __rcu		*gang_submit;
 	bool				ib_pool_ready;
 	struct amdgpu_sa_manager	ib_pools[AMDGPU_IB_POOL_MAX];
 	struct amdgpu_sched		gpu_sched[AMDGPU_HW_IP_NUM][AMDGPU_RING_PRIO_MAX];
@@ -1294,6 +1295,8 @@ u32 amdgpu_device_pcie_port_rreg(struct amdgpu_device *adev,
 				u32 reg);
 void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 				u32 reg, u32 v);
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang);
 
 /* atpx handler */
 #if defined(CONFIG_VGA_SWITCHEROO)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index 4f5bd96000ec..c7b1a2dfde13 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -686,6 +686,7 @@ int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
 	ib->length_dw = ib_len;
 	/* This works for NO_HWS. TODO: need to handle without knowing VMID */
 	job->vmid = vmid;
+	job->num_ibs = 1;
 
 	ret = amdgpu_ib_schedule(ring, 1, ib, job, &f);
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
index b7933c2ce765..491d4846fc02 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_connectors.c
@@ -1674,10 +1674,12 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 						   adev->mode_info.dither_property,
 						   AMDGPU_FMT_DITHER_DISABLE);
 
-			if (amdgpu_audio != 0)
+			if (amdgpu_audio != 0) {
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
+			}
 
 			subpixel_order = SubPixelHorizontalRGB;
 			connector->interlace_allowed = true;
@@ -1799,6 +1801,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
@@ -1852,6 +1855,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
@@ -1902,6 +1906,7 @@ amdgpu_connector_add(struct amdgpu_device *adev,
 				drm_object_attach_property(&amdgpu_connector->base.base,
 							   adev->mode_info.audio_property,
 							   AMDGPU_AUDIO_AUTO);
+				amdgpu_connector->audio = AMDGPU_AUDIO_AUTO;
 			}
 			drm_object_attach_property(&amdgpu_connector->base.base,
 						   adev->mode_info.dither_property,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index e104d7ef3c3d..1bbd39b3b0fc 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -64,11 +64,51 @@ static int amdgpu_cs_parser_init(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
+static int amdgpu_cs_job_idx(struct amdgpu_cs_parser *p,
+			     struct drm_amdgpu_cs_chunk_ib *chunk_ib)
+{
+	struct drm_sched_entity *entity;
+	unsigned int i;
+	int r;
+
+	r = amdgpu_ctx_get_entity(p->ctx, chunk_ib->ip_type,
+				  chunk_ib->ip_instance,
+				  chunk_ib->ring, &entity);
+	if (r)
+		return r;
+
+	/*
+	 * Abort if there is no run queue associated with this entity.
+	 * Possibly because of disabled HW IP.
+	 */
+	if (entity->rq == NULL)
+		return -EINVAL;
+
+	/* Check if we can add this IB to some existing job */
+	for (i = 0; i < p->gang_size; ++i)
+		if (p->entities[i] == entity)
+			return i;
+
+	/* If not increase the gang size if possible */
+	if (i == AMDGPU_CS_GANG_SIZE)
+		return -EINVAL;
+
+	p->entities[i] = entity;
+	p->gang_size = i + 1;
+	return i;
+}
+
 static int amdgpu_cs_p1_ib(struct amdgpu_cs_parser *p,
 			   struct drm_amdgpu_cs_chunk_ib *chunk_ib,
 			   unsigned int *num_ibs)
 {
-	++(*num_ibs);
+	int r;
+
+	r = amdgpu_cs_job_idx(p, chunk_ib);
+	if (r < 0)
+		return r;
+
+	++(num_ibs[r]);
 	return 0;
 }
 
@@ -142,11 +182,12 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 			   union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	unsigned int num_ibs[AMDGPU_CS_GANG_SIZE] = { };
 	struct amdgpu_vm *vm = &fpriv->vm;
 	uint64_t *chunk_array_user;
 	uint64_t *chunk_array;
-	unsigned size, num_ibs = 0;
 	uint32_t uf_offset = 0;
+	unsigned int size;
 	int ret;
 	int i;
 
@@ -209,7 +250,7 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 			if (size < sizeof(struct drm_amdgpu_cs_chunk_ib))
 				goto free_partial_kdata;
 
-			ret = amdgpu_cs_p1_ib(p, p->chunks[i].kdata, &num_ibs);
+			ret = amdgpu_cs_p1_ib(p, p->chunks[i].kdata, num_ibs);
 			if (ret)
 				goto free_partial_kdata;
 			break;
@@ -246,17 +287,28 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 		}
 	}
 
-	ret = amdgpu_job_alloc(p->adev, num_ibs, &p->job, vm);
-	if (ret)
-		goto free_all_kdata;
+	if (!p->gang_size)
+		return -EINVAL;
+
+	for (i = 0; i < p->gang_size; ++i) {
+		ret = amdgpu_job_alloc(p->adev, num_ibs[i], &p->jobs[i], vm);
+		if (ret)
+			goto free_all_kdata;
+
+		ret = drm_sched_job_init(&p->jobs[i]->base, p->entities[i],
+					 &fpriv->vm);
+		if (ret)
+			goto free_all_kdata;
+	}
+	p->gang_leader = p->jobs[p->gang_size - 1];
 
-	if (p->ctx->vram_lost_counter != p->job->vram_lost_counter) {
+	if (p->ctx->vram_lost_counter != p->gang_leader->vram_lost_counter) {
 		ret = -ECANCELED;
 		goto free_all_kdata;
 	}
 
 	if (p->uf_entry.tv.bo)
-		p->job->uf_addr = uf_offset;
+		p->gang_leader->uf_addr = uf_offset;
 	kvfree(chunk_array);
 
 	/* Use this opportunity to fill in task info for the vm */
@@ -278,93 +330,69 @@ free_chunk:
 	return ret;
 }
 
-static int amdgpu_cs_ib_fill(struct amdgpu_device *adev,
-			     struct amdgpu_cs_parser *parser)
+static int amdgpu_cs_p2_ib(struct amdgpu_cs_parser *p,
+			   struct amdgpu_cs_chunk *chunk,
+			   unsigned int *ce_preempt,
+			   unsigned int *de_preempt)
 {
-	struct amdgpu_fpriv *fpriv = parser->filp->driver_priv;
+	struct drm_amdgpu_cs_chunk_ib *chunk_ib = chunk->kdata;
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	struct amdgpu_vm *vm = &fpriv->vm;
-	int r, ce_preempt = 0, de_preempt = 0;
 	struct amdgpu_ring *ring;
-	int i, j;
-
-	for (i = 0, j = 0; i < parser->nchunks && j < parser->job->num_ibs; i++) {
-		struct amdgpu_cs_chunk *chunk;
-		struct amdgpu_ib *ib;
-		struct drm_amdgpu_cs_chunk_ib *chunk_ib;
-		struct drm_sched_entity *entity;
-
-		chunk = &parser->chunks[i];
-		ib = &parser->job->ibs[j];
-		chunk_ib = (struct drm_amdgpu_cs_chunk_ib *)chunk->kdata;
-
-		if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
-			continue;
-
-		if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX &&
-		    chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
-			if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
-				ce_preempt++;
-			else
-				de_preempt++;
+	struct amdgpu_job *job;
+	struct amdgpu_ib *ib;
+	int r;
 
-			/* each GFX command submit allows 0 or 1 IB preemptible for CE & DE */
-			if (ce_preempt > 1 || de_preempt > 1)
-				return -EINVAL;
-		}
+	r = amdgpu_cs_job_idx(p, chunk_ib);
+	if (r < 0)
+		return r;
 
-		r = amdgpu_ctx_get_entity(parser->ctx, chunk_ib->ip_type,
-					  chunk_ib->ip_instance, chunk_ib->ring,
-					  &entity);
-		if (r)
-			return r;
+	job = p->jobs[r];
+	ring = amdgpu_job_ring(job);
+	ib = &job->ibs[job->num_ibs++];
 
-		if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
-			parser->job->preamble_status |=
-				AMDGPU_PREAMBLE_IB_PRESENT;
+	/* MM engine doesn't support user fences */
+	if (p->uf_entry.tv.bo && ring->funcs->no_user_fence)
+		return -EINVAL;
 
-		if (parser->entity && parser->entity != entity)
-			return -EINVAL;
+	if (chunk_ib->ip_type == AMDGPU_HW_IP_GFX &&
+	    chunk_ib->flags & AMDGPU_IB_FLAG_PREEMPT) {
+		if (chunk_ib->flags & AMDGPU_IB_FLAG_CE)
+			(*ce_preempt)++;
+		else
+			(*de_preempt)++;
 
-		/* Return if there is no run queue associated with this entity.
-		 * Possibly because of disabled HW IP*/
-		if (entity->rq == NULL)
+		/* Each GFX command submit allows only 1 IB max
+		 * preemptible for CE & DE */
+		if (*ce_preempt > 1 || *de_preempt > 1)
 			return -EINVAL;
+	}
 
-		parser->entity = entity;
-
-		ring = to_amdgpu_ring(entity->rq->sched);
-		r =  amdgpu_ib_get(adev, vm, ring->funcs->parse_cs ?
-				   chunk_ib->ib_bytes : 0,
-				   AMDGPU_IB_POOL_DELAYED, ib);
-		if (r) {
-			DRM_ERROR("Failed to get ib !\n");
-			return r;
-		}
-
-		ib->gpu_addr = chunk_ib->va_start;
-		ib->length_dw = chunk_ib->ib_bytes / 4;
-		ib->flags = chunk_ib->flags;
+	if (chunk_ib->flags & AMDGPU_IB_FLAG_PREAMBLE)
+		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT;
 
-		j++;
+	r =  amdgpu_ib_get(p->adev, vm, ring->funcs->parse_cs ?
+			   chunk_ib->ib_bytes : 0,
+			   AMDGPU_IB_POOL_DELAYED, ib);
+	if (r) {
+		DRM_ERROR("Failed to get ib !\n");
+		return r;
 	}
 
-	/* MM engine doesn't support user fences */
-	ring = to_amdgpu_ring(parser->entity->rq->sched);
-	if (parser->job->uf_addr && ring->funcs->no_user_fence)
-		return -EINVAL;
-
+	ib->gpu_addr = chunk_ib->va_start;
+	ib->length_dw = chunk_ib->ib_bytes / 4;
+	ib->flags = chunk_ib->flags;
 	return 0;
 }
 
-static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
-				       struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_p2_dependencies(struct amdgpu_cs_parser *p,
+				     struct amdgpu_cs_chunk *chunk)
 {
+	struct drm_amdgpu_cs_chunk_dep *deps = chunk->kdata;
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
 	unsigned num_deps;
 	int i, r;
-	struct drm_amdgpu_cs_chunk_dep *deps;
 
-	deps = (struct drm_amdgpu_cs_chunk_dep *)chunk->kdata;
 	num_deps = chunk->length_dw * 4 /
 		sizeof(struct drm_amdgpu_cs_chunk_dep);
 
@@ -402,7 +430,7 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
 			dma_fence_put(old);
 		}
 
-		r = amdgpu_sync_fence(&p->job->sync, fence);
+		r = amdgpu_sync_fence(&p->gang_leader->sync, fence);
 		dma_fence_put(fence);
 		if (r)
 			return r;
@@ -410,9 +438,9 @@ static int amdgpu_cs_process_fence_dep(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
-static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
-						 uint32_t handle, u64 point,
-						 u64 flags)
+static int amdgpu_syncobj_lookup_and_add(struct amdgpu_cs_parser *p,
+					 uint32_t handle, u64 point,
+					 u64 flags)
 {
 	struct dma_fence *fence;
 	int r;
@@ -424,25 +452,23 @@ static int amdgpu_syncobj_lookup_and_add_to_sync(struct amdgpu_cs_parser *p,
 		return r;
 	}
 
-	r = amdgpu_sync_fence(&p->job->sync, fence);
+	r = amdgpu_sync_fence(&p->gang_leader->sync, fence);
 	dma_fence_put(fence);
 
 	return r;
 }
 
-static int amdgpu_cs_process_syncobj_in_dep(struct amdgpu_cs_parser *p,
-					    struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_p2_syncobj_in(struct amdgpu_cs_parser *p,
+				   struct amdgpu_cs_chunk *chunk)
 {
-	struct drm_amdgpu_cs_chunk_sem *deps;
+	struct drm_amdgpu_cs_chunk_sem *deps = chunk->kdata;
 	unsigned num_deps;
 	int i, r;
 
-	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
 	num_deps = chunk->length_dw * 4 /
 		sizeof(struct drm_amdgpu_cs_chunk_sem);
 	for (i = 0; i < num_deps; ++i) {
-		r = amdgpu_syncobj_lookup_and_add_to_sync(p, deps[i].handle,
-							  0, 0);
+		r = amdgpu_syncobj_lookup_and_add(p, deps[i].handle, 0, 0);
 		if (r)
 			return r;
 	}
@@ -450,21 +476,19 @@ static int amdgpu_cs_process_syncobj_in_dep(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
-static int amdgpu_cs_process_syncobj_timeline_in_dep(struct amdgpu_cs_parser *p,
-						     struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_p2_syncobj_timeline_wait(struct amdgpu_cs_parser *p,
+					      struct amdgpu_cs_chunk *chunk)
 {
-	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps;
+	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps = chunk->kdata;
 	unsigned num_deps;
 	int i, r;
 
-	syncobj_deps = (struct drm_amdgpu_cs_chunk_syncobj *)chunk->kdata;
 	num_deps = chunk->length_dw * 4 /
 		sizeof(struct drm_amdgpu_cs_chunk_syncobj);
 	for (i = 0; i < num_deps; ++i) {
-		r = amdgpu_syncobj_lookup_and_add_to_sync(p,
-							  syncobj_deps[i].handle,
-							  syncobj_deps[i].point,
-							  syncobj_deps[i].flags);
+		r = amdgpu_syncobj_lookup_and_add(p, syncobj_deps[i].handle,
+						  syncobj_deps[i].point,
+						  syncobj_deps[i].flags);
 		if (r)
 			return r;
 	}
@@ -472,14 +496,13 @@ static int amdgpu_cs_process_syncobj_timeline_in_dep(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
-static int amdgpu_cs_process_syncobj_out_dep(struct amdgpu_cs_parser *p,
-					     struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_p2_syncobj_out(struct amdgpu_cs_parser *p,
+				    struct amdgpu_cs_chunk *chunk)
 {
-	struct drm_amdgpu_cs_chunk_sem *deps;
+	struct drm_amdgpu_cs_chunk_sem *deps = chunk->kdata;
 	unsigned num_deps;
 	int i;
 
-	deps = (struct drm_amdgpu_cs_chunk_sem *)chunk->kdata;
 	num_deps = chunk->length_dw * 4 /
 		sizeof(struct drm_amdgpu_cs_chunk_sem);
 
@@ -507,15 +530,13 @@ static int amdgpu_cs_process_syncobj_out_dep(struct amdgpu_cs_parser *p,
 	return 0;
 }
 
-
-static int amdgpu_cs_process_syncobj_timeline_out_dep(struct amdgpu_cs_parser *p,
-						      struct amdgpu_cs_chunk *chunk)
+static int amdgpu_cs_p2_syncobj_timeline_signal(struct amdgpu_cs_parser *p,
+						struct amdgpu_cs_chunk *chunk)
 {
-	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps;
+	struct drm_amdgpu_cs_chunk_syncobj *syncobj_deps = chunk->kdata;
 	unsigned num_deps;
 	int i;
 
-	syncobj_deps = (struct drm_amdgpu_cs_chunk_syncobj *)chunk->kdata;
 	num_deps = chunk->length_dw * 4 /
 		sizeof(struct drm_amdgpu_cs_chunk_syncobj);
 
@@ -552,9 +573,9 @@ static int amdgpu_cs_process_syncobj_timeline_out_dep(struct amdgpu_cs_parser *p
 	return 0;
 }
 
-static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
-				  struct amdgpu_cs_parser *p)
+static int amdgpu_cs_pass2(struct amdgpu_cs_parser *p)
 {
+	unsigned int ce_preempt = 0, de_preempt = 0;
 	int i, r;
 
 	for (i = 0; i < p->nchunks; ++i) {
@@ -563,29 +584,34 @@ static int amdgpu_cs_dependencies(struct amdgpu_device *adev,
 		chunk = &p->chunks[i];
 
 		switch (chunk->chunk_id) {
+		case AMDGPU_CHUNK_ID_IB:
+			r = amdgpu_cs_p2_ib(p, chunk, &ce_preempt, &de_preempt);
+			if (r)
+				return r;
+			break;
 		case AMDGPU_CHUNK_ID_DEPENDENCIES:
 		case AMDGPU_CHUNK_ID_SCHEDULED_DEPENDENCIES:
-			r = amdgpu_cs_process_fence_dep(p, chunk);
+			r = amdgpu_cs_p2_dependencies(p, chunk);
 			if (r)
 				return r;
 			break;
 		case AMDGPU_CHUNK_ID_SYNCOBJ_IN:
-			r = amdgpu_cs_process_syncobj_in_dep(p, chunk);
+			r = amdgpu_cs_p2_syncobj_in(p, chunk);
 			if (r)
 				return r;
 			break;
 		case AMDGPU_CHUNK_ID_SYNCOBJ_OUT:
-			r = amdgpu_cs_process_syncobj_out_dep(p, chunk);
+			r = amdgpu_cs_p2_syncobj_out(p, chunk);
 			if (r)
 				return r;
 			break;
 		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_WAIT:
-			r = amdgpu_cs_process_syncobj_timeline_in_dep(p, chunk);
+			r = amdgpu_cs_p2_syncobj_timeline_wait(p, chunk);
 			if (r)
 				return r;
 			break;
 		case AMDGPU_CHUNK_ID_SYNCOBJ_TIMELINE_SIGNAL:
-			r = amdgpu_cs_process_syncobj_timeline_out_dep(p, chunk);
+			r = amdgpu_cs_p2_syncobj_timeline_signal(p, chunk);
 			if (r)
 				return r;
 			break;
@@ -830,6 +856,7 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	struct amdgpu_vm *vm = &fpriv->vm;
 	struct amdgpu_bo_list_entry *e;
 	struct list_head duplicates;
+	unsigned int i;
 	int r;
 
 	INIT_LIST_HEAD(&p->validated);
@@ -913,16 +940,6 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 		e->bo_va = amdgpu_vm_bo_find(vm, bo);
 	}
 
-	/* Move fence waiting after getting reservation lock of
-	 * PD root. Then there is no need on a ctx mutex lock.
-	 */
-	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entity);
-	if (unlikely(r != 0)) {
-		if (r != -ERESTARTSYS)
-			DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n");
-		goto error_validate;
-	}
-
 	amdgpu_cs_get_threshold_for_moves(p->adev, &p->bytes_moved_threshold,
 					  &p->bytes_moved_vis_threshold);
 	p->bytes_moved = 0;
@@ -943,124 +960,138 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
 	if (r)
 		goto error_validate;
 
-	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
-				     p->bytes_moved_vis);
-
-	amdgpu_job_set_resources(p->job, p->bo_list->gds_obj,
-				 p->bo_list->gws_obj, p->bo_list->oa_obj);
-
-	if (!r && p->uf_entry.tv.bo) {
+	if (p->uf_entry.tv.bo) {
 		struct amdgpu_bo *uf = ttm_to_amdgpu_bo(p->uf_entry.tv.bo);
 
 		r = amdgpu_ttm_alloc_gart(&uf->tbo);
-		p->job->uf_addr += amdgpu_bo_gpu_offset(uf);
+		if (r)
+			goto error_validate;
+
+		p->gang_leader->uf_addr += amdgpu_bo_gpu_offset(uf);
 	}
 
+	amdgpu_cs_report_moved_bytes(p->adev, p->bytes_moved,
+				     p->bytes_moved_vis);
+
+	for (i = 0; i < p->gang_size; ++i)
+		amdgpu_job_set_resources(p->jobs[i], p->bo_list->gds_obj,
+					 p->bo_list->gws_obj,
+					 p->bo_list->oa_obj);
+	return 0;
+
 error_validate:
-	if (r)
-		ttm_eu_backoff_reservation(&p->ticket, &p->validated);
+	ttm_eu_backoff_reservation(&p->ticket, &p->validated);
 
 out_free_user_pages:
-	if (r) {
-		amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
-			struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
+		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
 
-			if (!e->user_pages)
-				continue;
-			amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
-			kvfree(e->user_pages);
-			e->user_pages = NULL;
-		}
-		mutex_unlock(&p->bo_list->bo_list_mutex);
+		if (!e->user_pages)
+			continue;
+		amdgpu_ttm_tt_get_user_pages_done(bo->tbo.ttm);
+		kvfree(e->user_pages);
+		e->user_pages = NULL;
 	}
 	return r;
 }
 
-static void trace_amdgpu_cs_ibs(struct amdgpu_cs_parser *parser)
+static void trace_amdgpu_cs_ibs(struct amdgpu_cs_parser *p)
 {
-	int i;
+	int i, j;
 
 	if (!trace_amdgpu_cs_enabled())
 		return;
 
-	for (i = 0; i < parser->job->num_ibs; i++)
-		trace_amdgpu_cs(parser, i);
+	for (i = 0; i < p->gang_size; ++i) {
+		struct amdgpu_job *job = p->jobs[i];
+
+		for (j = 0; j < job->num_ibs; ++j)
+			trace_amdgpu_cs(p, job, &job->ibs[j]);
+	}
 }
 
-static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
+static int amdgpu_cs_patch_ibs(struct amdgpu_cs_parser *p,
+			       struct amdgpu_job *job)
 {
-	struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched);
-	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-	struct amdgpu_device *adev = p->adev;
-	struct amdgpu_vm *vm = &fpriv->vm;
-	struct amdgpu_bo_list_entry *e;
-	struct amdgpu_bo_va *bo_va;
-	struct amdgpu_bo *bo;
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	unsigned int i;
 	int r;
 
 	/* Only for UVD/VCE VM emulation */
-	if (ring->funcs->parse_cs || ring->funcs->patch_cs_in_place) {
-		unsigned i, j;
-
-		for (i = 0, j = 0; i < p->nchunks && j < p->job->num_ibs; i++) {
-			struct drm_amdgpu_cs_chunk_ib *chunk_ib;
-			struct amdgpu_bo_va_mapping *m;
-			struct amdgpu_bo *aobj = NULL;
-			struct amdgpu_cs_chunk *chunk;
-			uint64_t offset, va_start;
-			struct amdgpu_ib *ib;
-			uint8_t *kptr;
-
-			chunk = &p->chunks[i];
-			ib = &p->job->ibs[j];
-			chunk_ib = chunk->kdata;
-
-			if (chunk->chunk_id != AMDGPU_CHUNK_ID_IB)
-				continue;
+	if (!ring->funcs->parse_cs && !ring->funcs->patch_cs_in_place)
+		return 0;
 
-			va_start = chunk_ib->va_start & AMDGPU_GMC_HOLE_MASK;
-			r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
-			if (r) {
-				DRM_ERROR("IB va_start is invalid\n");
-				return r;
-			}
+	for (i = 0; i < job->num_ibs; ++i) {
+		struct amdgpu_ib *ib = &job->ibs[i];
+		struct amdgpu_bo_va_mapping *m;
+		struct amdgpu_bo *aobj;
+		uint64_t va_start;
+		uint8_t *kptr;
 
-			if ((va_start + chunk_ib->ib_bytes) >
-			    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
-				DRM_ERROR("IB va_start+ib_bytes is invalid\n");
-				return -EINVAL;
-			}
+		va_start = ib->gpu_addr & AMDGPU_GMC_HOLE_MASK;
+		r = amdgpu_cs_find_mapping(p, va_start, &aobj, &m);
+		if (r) {
+			DRM_ERROR("IB va_start is invalid\n");
+			return r;
+		}
 
-			/* the IB should be reserved at this point */
-			r = amdgpu_bo_kmap(aobj, (void **)&kptr);
-			if (r) {
-				return r;
-			}
+		if ((va_start + ib->length_dw * 4) >
+		    (m->last + 1) * AMDGPU_GPU_PAGE_SIZE) {
+			DRM_ERROR("IB va_start+ib_bytes is invalid\n");
+			return -EINVAL;
+		}
 
-			offset = m->start * AMDGPU_GPU_PAGE_SIZE;
-			kptr += va_start - offset;
-
-			if (ring->funcs->parse_cs) {
-				memcpy(ib->ptr, kptr, chunk_ib->ib_bytes);
-				amdgpu_bo_kunmap(aobj);
-
-				r = amdgpu_ring_parse_cs(ring, p, p->job, ib);
-				if (r)
-					return r;
-			} else {
-				ib->ptr = (uint32_t *)kptr;
-				r = amdgpu_ring_patch_cs_in_place(ring, p, p->job, ib);
-				amdgpu_bo_kunmap(aobj);
-				if (r)
-					return r;
-			}
+		/* the IB should be reserved at this point */
+		r = amdgpu_bo_kmap(aobj, (void **)&kptr);
+		if (r) {
+			return r;
+		}
+
+		kptr += va_start - (m->start * AMDGPU_GPU_PAGE_SIZE);
 
-			j++;
+		if (ring->funcs->parse_cs) {
+			memcpy(ib->ptr, kptr, ib->length_dw * 4);
+			amdgpu_bo_kunmap(aobj);
+
+			r = amdgpu_ring_parse_cs(ring, p, job, ib);
+			if (r)
+				return r;
+		} else {
+			ib->ptr = (uint32_t *)kptr;
+			r = amdgpu_ring_patch_cs_in_place(ring, p, job, ib);
+			amdgpu_bo_kunmap(aobj);
+			if (r)
+				return r;
 		}
 	}
 
-	if (!p->job->vm)
-		return 0;
+	return 0;
+}
+
+static int amdgpu_cs_patch_jobs(struct amdgpu_cs_parser *p)
+{
+	unsigned int i;
+	int r;
+
+	for (i = 0; i < p->gang_size; ++i) {
+		r = amdgpu_cs_patch_ibs(p, p->jobs[i]);
+		if (r)
+			return r;
+	}
+	return 0;
+}
+
+static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
+{
+	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_job *job = p->gang_leader;
+	struct amdgpu_device *adev = p->adev;
+	struct amdgpu_vm *vm = &fpriv->vm;
+	struct amdgpu_bo_list_entry *e;
+	struct amdgpu_bo_va *bo_va;
+	struct amdgpu_bo *bo;
+	unsigned int i;
+	int r;
 
 	r = amdgpu_vm_clear_freed(adev, vm, NULL);
 	if (r)
@@ -1070,7 +1101,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_sync_fence(&p->job->sync, fpriv->prt_va->last_pt_update);
+	r = amdgpu_sync_fence(&job->sync, fpriv->prt_va->last_pt_update);
 	if (r)
 		return r;
 
@@ -1081,7 +1112,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (r)
 			return r;
 
-		r = amdgpu_sync_fence(&p->job->sync, bo_va->last_pt_update);
+		r = amdgpu_sync_fence(&job->sync, bo_va->last_pt_update);
 		if (r)
 			return r;
 	}
@@ -1100,7 +1131,7 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 		if (r)
 			return r;
 
-		r = amdgpu_sync_fence(&p->job->sync, bo_va->last_pt_update);
+		r = amdgpu_sync_fence(&job->sync, bo_va->last_pt_update);
 		if (r)
 			return r;
 	}
@@ -1113,11 +1144,18 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 	if (r)
 		return r;
 
-	r = amdgpu_sync_fence(&p->job->sync, vm->last_update);
+	r = amdgpu_sync_fence(&job->sync, vm->last_update);
 	if (r)
 		return r;
 
-	p->job->vm_pd_addr = amdgpu_gmc_pd_addr(vm->root.bo);
+	for (i = 0; i < p->gang_size; ++i) {
+		job = p->jobs[i];
+
+		if (!job->vm)
+			continue;
+
+		job->vm_pd_addr = amdgpu_gmc_pd_addr(vm->root.bo);
+	}
 
 	if (amdgpu_vm_debug) {
 		/* Invalidate all BOs to test for userspace bugs */
@@ -1138,7 +1176,9 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
 static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
+	struct amdgpu_job *leader = p->gang_leader;
 	struct amdgpu_bo_list_entry *e;
+	unsigned int i;
 	int r;
 
 	list_for_each_entry(e, &p->validated, tv.head) {
@@ -1148,12 +1188,23 @@ static int amdgpu_cs_sync_rings(struct amdgpu_cs_parser *p)
 
 		sync_mode = amdgpu_bo_explicit_sync(bo) ?
 			AMDGPU_SYNC_EXPLICIT : AMDGPU_SYNC_NE_OWNER;
-		r = amdgpu_sync_resv(p->adev, &p->job->sync, resv, sync_mode,
+		r = amdgpu_sync_resv(p->adev, &leader->sync, resv, sync_mode,
 				     &fpriv->vm);
 		if (r)
 			return r;
 	}
-	return 0;
+
+	for (i = 0; i < p->gang_size - 1; ++i) {
+		r = amdgpu_sync_clone(&leader->sync, &p->jobs[i]->sync);
+		if (r)
+			return r;
+	}
+
+	r = amdgpu_ctx_wait_prev_fence(p->ctx, p->entities[p->gang_size - 1]);
+	if (r && r != -ERESTARTSYS)
+		DRM_ERROR("amdgpu_ctx_wait_prev_fence failed.\n");
+
+	return r;
 }
 
 static void amdgpu_cs_post_dependencies(struct amdgpu_cs_parser *p)
@@ -1177,20 +1228,28 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 			    union drm_amdgpu_cs *cs)
 {
 	struct amdgpu_fpriv *fpriv = p->filp->driver_priv;
-	struct drm_sched_entity *entity = p->entity;
+	struct amdgpu_job *leader = p->gang_leader;
 	struct amdgpu_bo_list_entry *e;
-	struct amdgpu_job *job;
+	unsigned int i;
 	uint64_t seq;
 	int r;
 
-	job = p->job;
-	p->job = NULL;
+	for (i = 0; i < p->gang_size; ++i)
+		drm_sched_job_arm(&p->jobs[i]->base);
 
-	r = drm_sched_job_init(&job->base, entity, &fpriv->vm);
-	if (r)
-		goto error_unlock;
+	for (i = 0; i < (p->gang_size - 1); ++i) {
+		struct dma_fence *fence;
 
-	drm_sched_job_arm(&job->base);
+		fence = &p->jobs[i]->base.s_fence->scheduled;
+		r = amdgpu_sync_fence(&leader->sync, fence);
+		if (r)
+			goto error_cleanup;
+	}
+
+	if (p->gang_size > 1) {
+		for (i = 0; i < p->gang_size; ++i)
+			amdgpu_job_set_gang_leader(p->jobs[i], leader);
+	}
 
 	/* No memory allocation is allowed while holding the notifier lock.
 	 * The lock is held until amdgpu_cs_submit is finished and fence is
@@ -1201,6 +1260,7 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	/* If userptr are invalidated after amdgpu_cs_parser_bos(), return
 	 * -EAGAIN, drmIoctl in libdrm will restart the amdgpu_cs_ioctl.
 	 */
+	r = 0;
 	amdgpu_bo_list_for_each_userptr_entry(e, p->bo_list) {
 		struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
 
@@ -1208,62 +1268,65 @@ static int amdgpu_cs_submit(struct amdgpu_cs_parser *p,
 	}
 	if (r) {
 		r = -EAGAIN;
-		goto error_abort;
+		goto error_unlock;
 	}
 
-	p->fence = dma_fence_get(&job->base.s_fence->finished);
+	p->fence = dma_fence_get(&leader->base.s_fence->finished);
+	list_for_each_entry(e, &p->validated, tv.head) {
 
-	seq = amdgpu_ctx_add_fence(p->ctx, entity, p->fence);
+		/* Everybody except for the gang leader uses READ */
+		for (i = 0; i < (p->gang_size - 1); ++i) {
+			dma_resv_add_fence(e->tv.bo->base.resv,
+					   &p->jobs[i]->base.s_fence->finished,
+					   DMA_RESV_USAGE_READ);
+		}
+
+		/* The gang leader is remembered as writer */
+		e->tv.num_shared = 0;
+	}
+
+	seq = amdgpu_ctx_add_fence(p->ctx, p->entities[p->gang_size - 1],
+				   p->fence);
 	amdgpu_cs_post_dependencies(p);
 
-	if ((job->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
+	if ((leader->preamble_status & AMDGPU_PREAMBLE_IB_PRESENT) &&
 	    !p->ctx->preamble_presented) {
-		job->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
+		leader->preamble_status |= AMDGPU_PREAMBLE_IB_PRESENT_FIRST;
 		p->ctx->preamble_presented = true;
 	}
 
 	cs->out.handle = seq;
-	job->uf_sequence = seq;
-
-	amdgpu_job_free_resources(job);
+	leader->uf_sequence = seq;
 
-	trace_amdgpu_cs_ioctl(job);
 	amdgpu_vm_bo_trace_cs(&fpriv->vm, &p->ticket);
-	drm_sched_entity_push_job(&job->base);
+	for (i = 0; i < p->gang_size; ++i) {
+		amdgpu_job_free_resources(p->jobs[i]);
+		trace_amdgpu_cs_ioctl(p->jobs[i]);
+		drm_sched_entity_push_job(&p->jobs[i]->base);
+		p->jobs[i] = NULL;
+	}
 
 	amdgpu_vm_move_to_lru_tail(p->adev, &fpriv->vm);
-
-	/* Make sure all BOs are remembered as writers */
-	amdgpu_bo_list_for_each_entry(e, p->bo_list)
-		e->tv.num_shared = 0;
-
 	ttm_eu_fence_buffer_objects(&p->ticket, &p->validated, p->fence);
+
 	mutex_unlock(&p->adev->notifier_lock);
 	mutex_unlock(&p->bo_list->bo_list_mutex);
-
 	return 0;
 
-error_abort:
-	drm_sched_job_cleanup(&job->base);
+error_unlock:
 	mutex_unlock(&p->adev->notifier_lock);
 
-error_unlock:
-	amdgpu_job_free(job);
+error_cleanup:
+	for (i = 0; i < p->gang_size; ++i)
+		drm_sched_job_cleanup(&p->jobs[i]->base);
 	return r;
 }
 
 /* Cleanup the parser structure */
-static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
-				  bool backoff)
+static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser)
 {
 	unsigned i;
 
-	if (error && backoff) {
-		ttm_eu_backoff_reservation(&parser->ticket,
-					   &parser->validated);
-		mutex_unlock(&parser->bo_list->bo_list_mutex);
-	}
-
 	for (i = 0; i < parser->num_post_deps; i++) {
 		drm_syncobj_put(parser->post_deps[i].syncobj);
 		kfree(parser->post_deps[i].chain);
@@ -1280,8 +1343,10 @@ static void amdgpu_cs_parser_fini(struct amdgpu_cs_parser *parser, int error,
 	for (i = 0; i < parser->nchunks; i++)
 		kvfree(parser->chunks[i].kdata);
 	kvfree(parser->chunks);
-	if (parser->job)
-		amdgpu_job_free(parser->job);
+	for (i = 0; i < parser->gang_size; ++i) {
+		if (parser->jobs[i])
+			amdgpu_job_free(parser->jobs[i]);
+	}
 	if (parser->uf_entry.tv.bo) {
 		struct amdgpu_bo *uf = ttm_to_amdgpu_bo(parser->uf_entry.tv.bo);
 
@@ -1293,7 +1358,6 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 {
 	struct amdgpu_device *adev = drm_to_adev(dev);
 	struct amdgpu_cs_parser parser;
-	bool reserved_buffers = false;
 	int r;
 
 	if (amdgpu_ras_intr_triggered())
@@ -1306,22 +1370,16 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 	if (r) {
 		if (printk_ratelimit())
 			DRM_ERROR("Failed to initialize parser %d!\n", r);
-		goto out;
+		return r;
 	}
 
 	r = amdgpu_cs_pass1(&parser, data);
 	if (r)
-		goto out;
+		goto error_fini;
 
-	r = amdgpu_cs_ib_fill(adev, &parser);
+	r = amdgpu_cs_pass2(&parser);
 	if (r)
-		goto out;
-
-	r = amdgpu_cs_dependencies(adev, &parser);
-	if (r) {
-		DRM_ERROR("Failed in the dependencies handling %d!\n", r);
-		goto out;
-	}
+		goto error_fini;
 
 	r = amdgpu_cs_parser_bos(&parser, data);
 	if (r) {
@@ -1329,25 +1387,36 @@ int amdgpu_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 			DRM_ERROR("Not enough memory for command submission!\n");
 		else if (r != -ERESTARTSYS && r != -EAGAIN)
 			DRM_ERROR("Failed to process the buffer list %d!\n", r);
-		goto out;
+		goto error_fini;
 	}
 
-	reserved_buffers = true;
-
-	trace_amdgpu_cs_ibs(&parser);
+	r = amdgpu_cs_patch_jobs(&parser);
+	if (r)
+		goto error_backoff;
 
 	r = amdgpu_cs_vm_handling(&parser);
 	if (r)
-		goto out;
+		goto error_backoff;
 
 	r = amdgpu_cs_sync_rings(&parser);
 	if (r)
-		goto out;
+		goto error_backoff;
+
+	trace_amdgpu_cs_ibs(&parser);
 
 	r = amdgpu_cs_submit(&parser, data);
-out:
-	amdgpu_cs_parser_fini(&parser, r, reserved_buffers);
+	if (r)
+		goto error_backoff;
+
+	amdgpu_cs_parser_fini(&parser);
+	return 0;
+
+error_backoff:
+	ttm_eu_backoff_reservation(&parser.ticket, &parser.validated);
+	mutex_unlock(&parser.bo_list->bo_list_mutex);
 
+error_fini:
+	amdgpu_cs_parser_fini(&parser);
 	return r;
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h
index 30ecc4917f81..cbaa19b2b8a3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.h
@@ -27,6 +27,8 @@
 #include "amdgpu_bo_list.h"
 #include "amdgpu_ring.h"
 
+#define AMDGPU_CS_GANG_SIZE	4
+
 struct amdgpu_bo_va_mapping;
 
 struct amdgpu_cs_chunk {
@@ -50,9 +52,11 @@ struct amdgpu_cs_parser {
 	unsigned		nchunks;
 	struct amdgpu_cs_chunk	*chunks;
 
-	/* scheduler job object */
-	struct amdgpu_job	*job;
-	struct drm_sched_entity	*entity;
+	/* scheduler job objects */
+	unsigned int		gang_size;
+	struct drm_sched_entity	*entities[AMDGPU_CS_GANG_SIZE];
+	struct amdgpu_job	*jobs[AMDGPU_CS_GANG_SIZE];
+	struct amdgpu_job	*gang_leader;
 
 	/* buffer objects */
 	struct ww_acquire_ctx		ticket;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 47aac179951a..c04ea7f1e819 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -3511,6 +3511,7 @@ int amdgpu_device_init(struct amdgpu_device *adev,
 	adev->gmc.gart_size = 512 * 1024 * 1024;
 	adev->accel_working = false;
 	adev->num_rings = 0;
+	RCU_INIT_POINTER(adev->gang_submit, dma_fence_get_stub());
 	adev->mman.buffer_funcs = NULL;
 	adev->mman.buffer_funcs_ring = NULL;
 	adev->vm_manager.vm_pte_funcs = NULL;
@@ -3992,6 +3993,7 @@ void amdgpu_device_fini_sw(struct amdgpu_device *adev)
 	release_firmware(adev->firmware.gpu_info_fw);
 	adev->firmware.gpu_info_fw = NULL;
 	adev->accel_working = false;
+	dma_fence_put(rcu_dereference_protected(adev->gang_submit, true));
 
 	amdgpu_reset_fini(adev);
 
@@ -4749,6 +4751,7 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 	struct amdgpu_device *tmp_adev = NULL;
 	bool need_full_reset, skip_hw_reset, vram_lost = false;
 	int r = 0;
+	bool gpu_reset_for_dev_remove = 0;
 
 	/* Try reset handler method first */
 	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
@@ -4768,6 +4771,10 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 		test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
 	skip_hw_reset = test_bit(AMDGPU_SKIP_HW_RESET, &reset_context->flags);
 
+	gpu_reset_for_dev_remove =
+		test_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context->flags) &&
+			test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
+
 	/*
 	 * ASIC reset has to be done on all XGMI hive nodes ASAP
 	 * to allow proper links negotiation in FW (within 1 sec)
@@ -4812,6 +4819,18 @@ int amdgpu_do_asic_reset(struct list_head *device_list_handle,
 		amdgpu_ras_intr_cleared();
 	}
 
+	/* Since the mode1 reset affects base ip blocks, the
+	 * phase1 ip blocks need to be resumed. Otherwise there
+	 * will be a BIOS signature error and the psp bootloader
+	 * can't load kdb on the next amdgpu install.
+	 */
+	if (gpu_reset_for_dev_remove) {
+		list_for_each_entry(tmp_adev, device_list_handle, reset_list)
+			amdgpu_device_ip_resume_phase1(tmp_adev);
+
+		goto end;
+	}
+
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
 		if (need_full_reset) {
 			/* post card */
@@ -5134,6 +5153,11 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	bool need_emergency_restart = false;
 	bool audio_suspended = false;
 	int tmp_vram_lost_counter;
+	bool gpu_reset_for_dev_remove = false;
+
+	gpu_reset_for_dev_remove =
+			test_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context->flags) &&
+				test_bit(AMDGPU_NEED_FULL_RESET, &reset_context->flags);
 
 	/*
 	 * Special case: RAS triggered and full reset isn't supported
@@ -5169,8 +5193,11 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 	 */
 	INIT_LIST_HEAD(&device_list);
 	if (!amdgpu_sriov_vf(adev) && (adev->gmc.xgmi.num_physical_nodes > 1)) {
-		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head)
+		list_for_each_entry(tmp_adev, &hive->device_list, gmc.xgmi.head) {
 			list_add_tail(&tmp_adev->reset_list, &device_list);
+			if (gpu_reset_for_dev_remove && adev->shutdown)
+				tmp_adev->shutdown = true;
+		}
 		if (!list_is_first(&adev->reset_list, &device_list))
 			list_rotate_to_front(&adev->reset_list, &device_list);
 		device_list_handle = &device_list;
@@ -5253,6 +5280,10 @@ int amdgpu_device_gpu_recover(struct amdgpu_device *adev,
 
 retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 	list_for_each_entry(tmp_adev, device_list_handle, reset_list) {
+		if (gpu_reset_for_dev_remove) {
+			/* Workaroud for ASICs need to disable SMC first */
+			amdgpu_device_smu_fini_early(tmp_adev);
+		}
 		r = amdgpu_device_pre_asic_reset(tmp_adev, reset_context);
 		/*TODO Should we stop ?*/
 		if (r) {
@@ -5286,6 +5317,9 @@ retry:	/* Rest of adevs pre asic reset from XGMI hive. */
 			adev->asic_reset_res = 0;
 			goto retry;
 		}
+
+		if (!r && gpu_reset_for_dev_remove)
+			goto recover_end;
 	}
 
 skip_hw_reset:
@@ -5359,6 +5393,7 @@ skip_sched_resume:
 		amdgpu_device_unset_mp1_state(tmp_adev);
 	}
 
+recover_end:
 	tmp_adev = list_first_entry(device_list_handle, struct amdgpu_device,
 					    reset_list);
 	amdgpu_device_unlock_reset_domain(tmp_adev->reset_domain);
@@ -5927,3 +5962,36 @@ void amdgpu_device_pcie_port_wreg(struct amdgpu_device *adev,
 	(void)RREG32(data);
 	spin_unlock_irqrestore(&adev->pcie_idx_lock, flags);
 }
+
+/**
+ * amdgpu_device_switch_gang - switch to a new gang
+ * @adev: amdgpu_device pointer
+ * @gang: the gang to switch to
+ *
+ * Try to switch to a new gang.
+ * Returns: NULL if we switched to the new gang or a reference to the current
+ * gang leader.
+ */
+struct dma_fence *amdgpu_device_switch_gang(struct amdgpu_device *adev,
+					    struct dma_fence *gang)
+{
+	struct dma_fence *old = NULL;
+
+	do {
+		dma_fence_put(old);
+		rcu_read_lock();
+		old = dma_fence_get_rcu_safe(&adev->gang_submit);
+		rcu_read_unlock();
+
+		if (old == gang)
+			break;
+
+		if (!dma_fence_is_signaled(old))
+			return old;
+
+	} while (cmpxchg((struct dma_fence __force **)&adev->gang_submit,
+			 old, gang) != old);
+
+	dma_fence_put(old);
+	return NULL;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
index 5b09c8f4fe95..23998f727c7f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_display.c
@@ -39,6 +39,7 @@
 #include <linux/pm_runtime.h>
 #include <drm/drm_crtc_helper.h>
 #include <drm/drm_damage_helper.h>
+#include <drm/drm_drv.h>
 #include <drm/drm_edid.h>
 #include <drm/drm_gem_framebuffer_helper.h>
 #include <drm/drm_fb_helper.h>
@@ -497,6 +498,11 @@ bool amdgpu_display_ddc_probe(struct amdgpu_connector *amdgpu_connector,
 static const struct drm_framebuffer_funcs amdgpu_fb_funcs = {
 	.destroy = drm_gem_fb_destroy,
 	.create_handle = drm_gem_fb_create_handle,
+};
+
+static const struct drm_framebuffer_funcs amdgpu_fb_funcs_atomic = {
+	.destroy = drm_gem_fb_destroy,
+	.create_handle = drm_gem_fb_create_handle,
 	.dirty = drm_atomic_helper_dirtyfb,
 };
 
@@ -1102,7 +1108,10 @@ static int amdgpu_display_gem_fb_verify_and_init(struct drm_device *dev,
 	if (ret)
 		goto err;
 
-	ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
+	if (drm_drv_uses_atomic_modeset(dev))
+		ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs_atomic);
+	else
+		ret = drm_framebuffer_init(dev, &rfb->base, &amdgpu_fb_funcs);
 	if (ret)
 		goto err;
 
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 728a0933ea6f..16f6a313335e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -102,9 +102,10 @@
  * - 3.46.0 - To enable hot plug amdgpu tests in libdrm
  * - 3.47.0 - Add AMDGPU_GEM_CREATE_DISCARDABLE and AMDGPU_VM_NOALLOC flags
  * - 3.48.0 - Add IP discovery version info to HW INFO
+ *   3.49.0 - Add gang submit into CS IOCTL
  */
 #define KMS_DRIVER_MAJOR	3
-#define KMS_DRIVER_MINOR	48
+#define KMS_DRIVER_MINOR	49
 #define KMS_DRIVER_PATCHLEVEL	0
 
 int amdgpu_vram_limit;
@@ -2186,6 +2187,37 @@ amdgpu_pci_remove(struct pci_dev *pdev)
 		pm_runtime_forbid(dev->dev);
 	}
 
+	if (adev->ip_versions[MP1_HWIP][0] == IP_VERSION(13, 0, 2)) {
+		bool need_to_reset_gpu = false;
+
+		if (adev->gmc.xgmi.num_physical_nodes > 1) {
+			struct amdgpu_hive_info *hive;
+
+			hive = amdgpu_get_xgmi_hive(adev);
+			if (hive->device_remove_count == 0)
+				need_to_reset_gpu = true;
+			hive->device_remove_count++;
+			amdgpu_put_xgmi_hive(hive);
+		} else {
+			need_to_reset_gpu = true;
+		}
+
+		/* Workaround for ASICs need to reset SMU.
+		 * Called only when the first device is removed.
+		 */
+		if (need_to_reset_gpu) {
+			struct amdgpu_reset_context reset_context;
+
+			adev->shutdown = true;
+			memset(&reset_context, 0, sizeof(reset_context));
+			reset_context.method = AMD_RESET_METHOD_NONE;
+			reset_context.reset_req_dev = adev;
+			set_bit(AMDGPU_NEED_FULL_RESET, &reset_context.flags);
+			set_bit(AMDGPU_RESET_FOR_DEVICE_REMOVE, &reset_context.flags);
+			amdgpu_device_gpu_recover(adev, NULL, &reset_context);
+		}
+	}
+
 	amdgpu_driver_unload_kms(dev);
 
 	drm_dev_unplug(dev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
index 8abdf41d0f83..b91ab919ee70 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h
@@ -304,6 +304,10 @@ struct amdgpu_gfx {
 	uint32_t			rlc_srlg_feature_version;
 	uint32_t			rlc_srls_fw_version;
 	uint32_t			rlc_srls_feature_version;
+	uint32_t			rlcp_ucode_version;
+	uint32_t			rlcp_ucode_feature_version;
+	uint32_t			rlcv_ucode_version;
+	uint32_t			rlcv_ucode_feature_version;
 	uint32_t			mec_feature_version;
 	uint32_t			mec2_feature_version;
 	bool				mec_fw_write_wait;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
index 877b3c22a8a8..46c99331d7f1 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c
@@ -105,7 +105,6 @@ int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
 	 */
 	(*job)->base.sched = &adev->rings[0]->sched;
 	(*job)->vm = vm;
-	(*job)->num_ibs = num_ibs;
 
 	amdgpu_sync_create(&(*job)->sync);
 	amdgpu_sync_create(&(*job)->sched_sync);
@@ -125,6 +124,7 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
 	if (r)
 		return r;
 
+	(*job)->num_ibs = 1;
 	r = amdgpu_ib_get(adev, NULL, size, pool_type, &(*job)->ibs[0]);
 	if (r)
 		kfree(*job);
@@ -173,11 +173,29 @@ static void amdgpu_job_free_cb(struct drm_sched_job *s_job)
 	dma_fence_put(&job->hw_fence);
 }
 
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader)
+{
+	struct dma_fence *fence = &leader->base.s_fence->scheduled;
+
+	WARN_ON(job->gang_submit);
+
+	/*
+	 * Don't add a reference when we are the gang leader to avoid circle
+	 * dependency.
+	 */
+	if (job != leader)
+		dma_fence_get(fence);
+	job->gang_submit = fence;
+}
+
 void amdgpu_job_free(struct amdgpu_job *job)
 {
 	amdgpu_job_free_resources(job);
 	amdgpu_sync_free(&job->sync);
 	amdgpu_sync_free(&job->sched_sync);
+	if (job->gang_submit != &job->base.s_fence->scheduled)
+		dma_fence_put(job->gang_submit);
 
 	if (!job->hw_fence.ops)
 		kfree(job);
@@ -247,12 +265,16 @@ static struct dma_fence *amdgpu_job_dependency(struct drm_sched_job *sched_job,
 		fence = amdgpu_sync_get_fence(&job->sync);
 	}
 
+	if (!fence && job->gang_submit)
+		fence = amdgpu_device_switch_gang(ring->adev, job->gang_submit);
+
 	return fence;
 }
 
 static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 {
 	struct amdgpu_ring *ring = to_amdgpu_ring(sched_job->sched);
+	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *fence = NULL, *finished;
 	struct amdgpu_job *job;
 	int r = 0;
@@ -264,8 +286,10 @@ static struct dma_fence *amdgpu_job_run(struct drm_sched_job *sched_job)
 
 	trace_amdgpu_sched_run_job(job);
 
-	if (job->vram_lost_counter != atomic_read(&ring->adev->vram_lost_counter))
-		dma_fence_set_error(finished, -ECANCELED);/* skip IB as well if VRAM lost */
+	/* Skip job if VRAM is lost and never resubmit gangs */
+	if (job->vram_lost_counter != atomic_read(&adev->vram_lost_counter) ||
+	    (job->job_run_counter && job->gang_submit))
+		dma_fence_set_error(finished, -ECANCELED);
 
 	if (finished->error < 0) {
 		DRM_INFO("Skip scheduling IBs!\n");
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
index 2a1961bf1194..ab7b150e5d50 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.h
@@ -50,6 +50,7 @@ struct amdgpu_job {
 	struct amdgpu_sync	sync;
 	struct amdgpu_sync	sched_sync;
 	struct dma_fence	hw_fence;
+	struct dma_fence	*gang_submit;
 	uint32_t		preamble_status;
 	uint32_t                preemption_status;
 	bool                    vm_needs_flush;
@@ -72,6 +73,11 @@ struct amdgpu_job {
 	struct amdgpu_ib	ibs[];
 };
 
+static inline struct amdgpu_ring *amdgpu_job_ring(struct amdgpu_job *job)
+{
+	return to_amdgpu_ring(job->base.entity->rq->sched);
+}
+
 int amdgpu_job_alloc(struct amdgpu_device *adev, unsigned num_ibs,
 		     struct amdgpu_job **job, struct amdgpu_vm *vm);
 int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
@@ -79,6 +85,8 @@ int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev, unsigned size,
 void amdgpu_job_set_resources(struct amdgpu_job *job, struct amdgpu_bo *gds,
 			      struct amdgpu_bo *gws, struct amdgpu_bo *oa);
 void amdgpu_job_free_resources(struct amdgpu_job *job);
+void amdgpu_job_set_gang_leader(struct amdgpu_job *job,
+				struct amdgpu_job *leader);
 void amdgpu_job_free(struct amdgpu_job *job);
 int amdgpu_job_submit(struct amdgpu_job *job, struct drm_sched_entity *entity,
 		      void *owner, struct dma_fence **f);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 77668c3dae5b..fe23e09eec98 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -247,6 +247,14 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->ver = adev->gfx.rlc_srls_fw_version;
 		fw_info->feature = adev->gfx.rlc_srls_feature_version;
 		break;
+	case AMDGPU_INFO_FW_GFX_RLCP:
+		fw_info->ver = adev->gfx.rlcp_ucode_version;
+		fw_info->feature = adev->gfx.rlcp_ucode_feature_version;
+		break;
+	case AMDGPU_INFO_FW_GFX_RLCV:
+		fw_info->ver = adev->gfx.rlcv_ucode_version;
+		fw_info->feature = adev->gfx.rlcv_ucode_feature_version;
+		break;
 	case AMDGPU_INFO_FW_GFX_MEC:
 		if (query_fw->index == 0) {
 			fw_info->ver = adev->gfx.mec_fw_version;
@@ -328,6 +336,14 @@ static int amdgpu_firmware_info(struct drm_amdgpu_info_firmware *fw_info,
 		fw_info->ver = adev->psp.cap_fw_version;
 		fw_info->feature = adev->psp.cap_feature_version;
 		break;
+	case AMDGPU_INFO_FW_MES_KIQ:
+		fw_info->ver = adev->mes.ucode_fw_version[0];
+		fw_info->feature = 0;
+		break;
+	case AMDGPU_INFO_FW_MES:
+		fw_info->ver = adev->mes.ucode_fw_version[1];
+		fw_info->feature = 0;
+		break;
 	default:
 		return -EINVAL;
 	}
@@ -1469,6 +1485,22 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused)
 	seq_printf(m, "RLC SRLS feature version: %u, firmware version: 0x%08x\n",
 		   fw_info.feature, fw_info.ver);
 
+	/* RLCP */
+	query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLCP;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLCP feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* RLCV */
+        query_fw.fw_type = AMDGPU_INFO_FW_GFX_RLCV;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "RLCV feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
 	/* MEC */
 	query_fw.fw_type = AMDGPU_INFO_FW_GFX_MEC;
 	query_fw.index = 0;
@@ -1581,6 +1613,22 @@ static int amdgpu_debugfs_firmware_info_show(struct seq_file *m, void *unused)
 				fw_info.feature, fw_info.ver);
 	}
 
+	/* MES_KIQ */
+	query_fw.fw_type = AMDGPU_INFO_FW_MES_KIQ;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MES_KIQ feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
+	/* MES */
+	query_fw.fw_type = AMDGPU_INFO_FW_MES;
+	ret = amdgpu_firmware_info(&fw_info, &query_fw, adev);
+	if (ret)
+		return ret;
+	seq_printf(m, "MES feature version: %u, firmware version: 0x%08x\n",
+		   fw_info.feature, fw_info.ver);
+
 	seq_printf(m, "VBIOS version: %s\n", ctx->vbios_version);
 
 	return 0;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index c4848522be16..effa7df3ddbf 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -511,6 +511,11 @@ static int psp_sw_fini(void *handle)
 	kfree(cmd);
 	cmd = NULL;
 
+	if (psp->km_ring.ring_mem)
+		amdgpu_bo_free_kernel(&adev->firmware.rbuf,
+				      &psp->km_ring.ring_mem_mc_addr,
+				      (void **)&psp->km_ring.ring_mem);
+
 	amdgpu_bo_free_kernel(&psp->fw_pri_bo,
 			      &psp->fw_pri_mc_addr, &psp->fw_pri_buf);
 	amdgpu_bo_free_kernel(&psp->fence_buf_bo,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
index f71b83c42590..dc43fcb93eac 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_reset.h
@@ -31,6 +31,7 @@ enum AMDGPU_RESET_FLAGS {
 	AMDGPU_NEED_FULL_RESET = 0,
 	AMDGPU_SKIP_HW_RESET = 1,
 	AMDGPU_SKIP_MODE2_RESET = 2,
+	AMDGPU_RESET_FOR_DEVICE_REMOVE = 3,
 };
 
 struct amdgpu_reset_context {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
index 06dfcf297a8d..5e6ddc7e101c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_trace.h
@@ -140,8 +140,10 @@ TRACE_EVENT(amdgpu_bo_create,
 );
 
 TRACE_EVENT(amdgpu_cs,
-	    TP_PROTO(struct amdgpu_cs_parser *p, int i),
-	    TP_ARGS(p, i),
+	    TP_PROTO(struct amdgpu_cs_parser *p,
+		     struct amdgpu_job *job,
+		     struct amdgpu_ib *ib),
+	    TP_ARGS(p, job, ib),
 	    TP_STRUCT__entry(
 			     __field(struct amdgpu_bo_list *, bo_list)
 			     __field(u32, ring)
@@ -151,10 +153,10 @@ TRACE_EVENT(amdgpu_cs,
 
 	    TP_fast_assign(
 			   __entry->bo_list = p->bo_list;
-			   __entry->ring = to_amdgpu_ring(p->entity->rq->sched)->idx;
-			   __entry->dw = p->job->ibs[i].length_dw;
+			   __entry->ring = to_amdgpu_ring(job->base.sched)->idx;
+			   __entry->dw = ib->length_dw;
 			   __entry->fences = amdgpu_fence_count_emitted(
-				to_amdgpu_ring(p->entity->rq->sched));
+				to_amdgpu_ring(job->base.sched));
 			   ),
 	    TP_printk("bo_list=%p, ring=%u, dw=%u, fences=%u",
 		      __entry->bo_list, __entry->ring, __entry->dw,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
index 939c8614f0e3..dd0bc649a57d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.c
@@ -164,70 +164,138 @@ void amdgpu_ucode_print_rlc_hdr(const struct common_firmware_header *hdr)
 	} else if (version_major == 2) {
 		const struct rlc_firmware_header_v2_0 *rlc_hdr =
 			container_of(hdr, struct rlc_firmware_header_v2_0, header);
+		const struct rlc_firmware_header_v2_1 *rlc_hdr_v2_1 =
+			container_of(rlc_hdr, struct rlc_firmware_header_v2_1, v2_0);
+		const struct rlc_firmware_header_v2_2 *rlc_hdr_v2_2 =
+			container_of(rlc_hdr_v2_1, struct rlc_firmware_header_v2_2, v2_1);
+		const struct rlc_firmware_header_v2_3 *rlc_hdr_v2_3 =
+			container_of(rlc_hdr_v2_2, struct rlc_firmware_header_v2_3, v2_2);
+		const struct rlc_firmware_header_v2_4 *rlc_hdr_v2_4 =
+			container_of(rlc_hdr_v2_3, struct rlc_firmware_header_v2_4, v2_3);
 
-		DRM_DEBUG("ucode_feature_version: %u\n",
-			  le32_to_cpu(rlc_hdr->ucode_feature_version));
-		DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
-		DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
-		DRM_DEBUG("save_and_restore_offset: %u\n",
-			  le32_to_cpu(rlc_hdr->save_and_restore_offset));
-		DRM_DEBUG("clear_state_descriptor_offset: %u\n",
-			  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
-		DRM_DEBUG("avail_scratch_ram_locations: %u\n",
-			  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
-		DRM_DEBUG("reg_restore_list_size: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_restore_list_size));
-		DRM_DEBUG("reg_list_format_start: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_start));
-		DRM_DEBUG("reg_list_format_separate_start: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
-		DRM_DEBUG("starting_offsets_start: %u\n",
-			  le32_to_cpu(rlc_hdr->starting_offsets_start));
-		DRM_DEBUG("reg_list_format_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
-		DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
-		DRM_DEBUG("reg_list_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
-		DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
-		DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
-		DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
-		DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
-		DRM_DEBUG("reg_list_separate_array_offset_bytes: %u\n",
-			  le32_to_cpu(rlc_hdr->reg_list_separate_array_offset_bytes));
-		if (version_minor == 1) {
-			const struct rlc_firmware_header_v2_1 *v2_1 =
-				container_of(rlc_hdr, struct rlc_firmware_header_v2_1, v2_0);
+		switch (version_minor) {
+		case 0:
+			/* rlc_hdr v2_0 */
+			DRM_DEBUG("ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr->ucode_feature_version));
+			DRM_DEBUG("jt_offset: %u\n", le32_to_cpu(rlc_hdr->jt_offset));
+			DRM_DEBUG("jt_size: %u\n", le32_to_cpu(rlc_hdr->jt_size));
+			DRM_DEBUG("save_and_restore_offset: %u\n",
+				  le32_to_cpu(rlc_hdr->save_and_restore_offset));
+			DRM_DEBUG("clear_state_descriptor_offset: %u\n",
+				  le32_to_cpu(rlc_hdr->clear_state_descriptor_offset));
+			DRM_DEBUG("avail_scratch_ram_locations: %u\n",
+				  le32_to_cpu(rlc_hdr->avail_scratch_ram_locations));
+			DRM_DEBUG("reg_restore_list_size: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_restore_list_size));
+			DRM_DEBUG("reg_list_format_start: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_start));
+			DRM_DEBUG("reg_list_format_separate_start: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_start));
+			DRM_DEBUG("starting_offsets_start: %u\n",
+				  le32_to_cpu(rlc_hdr->starting_offsets_start));
+			DRM_DEBUG("reg_list_format_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_size_bytes));
+			DRM_DEBUG("reg_list_format_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_array_offset_bytes));
+			DRM_DEBUG("reg_list_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_size_bytes));
+			DRM_DEBUG("reg_list_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_array_offset_bytes));
+			DRM_DEBUG("reg_list_format_separate_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_size_bytes));
+			DRM_DEBUG("reg_list_format_separate_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_format_separate_array_offset_bytes));
+			DRM_DEBUG("reg_list_separate_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_separate_size_bytes));
+			DRM_DEBUG("reg_list_separate_array_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr->reg_list_separate_array_offset_bytes));
+			break;
+		case 1:
+			/* rlc_hdr v2_1 */
 			DRM_DEBUG("reg_list_format_direct_reg_list_length: %u\n",
-				  le32_to_cpu(v2_1->reg_list_format_direct_reg_list_length));
+				  le32_to_cpu(rlc_hdr_v2_1->reg_list_format_direct_reg_list_length));
 			DRM_DEBUG("save_restore_list_cntl_ucode_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_cntl_ucode_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_ucode_ver));
 			DRM_DEBUG("save_restore_list_cntl_feature_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_cntl_feature_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_feature_ver));
 			DRM_DEBUG("save_restore_list_cntl_size_bytes %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_cntl_size_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_size_bytes));
 			DRM_DEBUG("save_restore_list_cntl_offset_bytes: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_cntl_offset_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_cntl_offset_bytes));
 			DRM_DEBUG("save_restore_list_gpm_ucode_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_gpm_ucode_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_ucode_ver));
 			DRM_DEBUG("save_restore_list_gpm_feature_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_gpm_feature_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_feature_ver));
 			DRM_DEBUG("save_restore_list_gpm_size_bytes %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_gpm_size_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_size_bytes));
 			DRM_DEBUG("save_restore_list_gpm_offset_bytes: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_gpm_offset_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_gpm_offset_bytes));
 			DRM_DEBUG("save_restore_list_srm_ucode_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_srm_ucode_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_ucode_ver));
 			DRM_DEBUG("save_restore_list_srm_feature_ver: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_srm_feature_ver));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_feature_ver));
 			DRM_DEBUG("save_restore_list_srm_size_bytes %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_srm_size_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_size_bytes));
 			DRM_DEBUG("save_restore_list_srm_offset_bytes: %u\n",
-				  le32_to_cpu(v2_1->save_restore_list_srm_offset_bytes));
+				  le32_to_cpu(rlc_hdr_v2_1->save_restore_list_srm_offset_bytes));
+			break;
+		case 2:
+			/* rlc_hdr v2_2 */
+			DRM_DEBUG("rlc_iram_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_iram_ucode_size_bytes));
+			DRM_DEBUG("rlc_iram_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_iram_ucode_offset_bytes));
+			DRM_DEBUG("rlc_dram_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_dram_ucode_size_bytes));
+			DRM_DEBUG("rlc_dram_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_2->rlc_dram_ucode_offset_bytes));
+			break;
+		case 3:
+			/* rlc_hdr v2_3 */
+			DRM_DEBUG("rlcp_ucode_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_version));
+			DRM_DEBUG("rlcp_ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_feature_version));
+			DRM_DEBUG("rlcp_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_size_bytes));
+			DRM_DEBUG("rlcp_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcp_ucode_offset_bytes));
+			DRM_DEBUG("rlcv_ucode_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_version));
+			DRM_DEBUG("rlcv_ucode_feature_version: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_feature_version));
+			DRM_DEBUG("rlcv_ucode_size_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_size_bytes));
+			DRM_DEBUG("rlcv_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_3->rlcv_ucode_offset_bytes));
+			break;
+		case 4:
+			/* rlc_hdr v2_4 */
+			DRM_DEBUG("global_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->global_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("global_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->global_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se0_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se0_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se0_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se0_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se1_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se1_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se1_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se1_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se2_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se2_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se2_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se2_tap_delays_ucode_offset_bytes));
+			DRM_DEBUG("se3_tap_delays_ucode_size_bytes :%u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se3_tap_delays_ucode_size_bytes));
+			DRM_DEBUG("se3_tap_delays_ucode_offset_bytes: %u\n",
+				  le32_to_cpu(rlc_hdr_v2_4->se3_tap_delays_ucode_offset_bytes));
+			break;
+		default:
+			DRM_ERROR("Unknown RLC v2 ucode: v2.%u\n", version_minor);
+			break;
 		}
 	} else {
 		DRM_ERROR("Unknown RLC ucode version: %u.%u\n", version_major, version_minor);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
index 3975bcaa2c89..1c36235b4539 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ucode.h
@@ -261,8 +261,12 @@ struct rlc_firmware_header_v2_2 {
 /* version_major=2, version_minor=3 */
 struct rlc_firmware_header_v2_3 {
     struct rlc_firmware_header_v2_2 v2_2;
+    uint32_t rlcp_ucode_version;
+    uint32_t rlcp_ucode_feature_version;
     uint32_t rlcp_ucode_size_bytes;
     uint32_t rlcp_ucode_offset_bytes;
+    uint32_t rlcv_ucode_version;
+    uint32_t rlcv_ucode_feature_version;
     uint32_t rlcv_ucode_size_bytes;
     uint32_t rlcv_ucode_offset_bytes;
 };
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 59cac347baa3..83b0c5d86e48 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -183,10 +183,12 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
 	struct amdgpu_bo *bo = vm_bo->bo;
 
 	vm_bo->moved = true;
+	spin_lock(&vm_bo->vm->status_lock);
 	if (bo->tbo.type == ttm_bo_type_kernel)
 		list_move(&vm_bo->vm_status, &vm->evicted);
 	else
 		list_move_tail(&vm_bo->vm_status, &vm->evicted);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 /**
  * amdgpu_vm_bo_moved - vm_bo is moved
@@ -198,7 +200,9 @@ static void amdgpu_vm_bo_evicted(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
 {
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->moved);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -211,7 +215,9 @@ static void amdgpu_vm_bo_moved(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
 {
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->idle);
+	spin_unlock(&vm_bo->vm->status_lock);
 	vm_bo->moved = false;
 }
 
@@ -225,9 +231,9 @@ static void amdgpu_vm_bo_idle(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
 {
-	spin_lock(&vm_bo->vm->invalidated_lock);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->invalidated);
-	spin_unlock(&vm_bo->vm->invalidated_lock);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -240,10 +246,13 @@ static void amdgpu_vm_bo_invalidated(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
 {
-	if (vm_bo->bo->parent)
+	if (vm_bo->bo->parent) {
+		spin_lock(&vm_bo->vm->status_lock);
 		list_move(&vm_bo->vm_status, &vm_bo->vm->relocated);
-	else
+		spin_unlock(&vm_bo->vm->status_lock);
+	} else {
 		amdgpu_vm_bo_idle(vm_bo);
+	}
 }
 
 /**
@@ -256,9 +265,9 @@ static void amdgpu_vm_bo_relocated(struct amdgpu_vm_bo_base *vm_bo)
  */
 static void amdgpu_vm_bo_done(struct amdgpu_vm_bo_base *vm_bo)
 {
-	spin_lock(&vm_bo->vm->invalidated_lock);
+	spin_lock(&vm_bo->vm->status_lock);
 	list_move(&vm_bo->vm_status, &vm_bo->vm->done);
-	spin_unlock(&vm_bo->vm->invalidated_lock);
+	spin_unlock(&vm_bo->vm->status_lock);
 }
 
 /**
@@ -363,12 +372,20 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			      int (*validate)(void *p, struct amdgpu_bo *bo),
 			      void *param)
 {
-	struct amdgpu_vm_bo_base *bo_base, *tmp;
+	struct amdgpu_vm_bo_base *bo_base;
+	struct amdgpu_bo *shadow;
+	struct amdgpu_bo *bo;
 	int r;
 
-	list_for_each_entry_safe(bo_base, tmp, &vm->evicted, vm_status) {
-		struct amdgpu_bo *bo = bo_base->bo;
-		struct amdgpu_bo *shadow = amdgpu_bo_shadowed(bo);
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->evicted)) {
+		bo_base = list_first_entry(&vm->evicted,
+					   struct amdgpu_vm_bo_base,
+					   vm_status);
+		spin_unlock(&vm->status_lock);
+
+		bo = bo_base->bo;
+		shadow = amdgpu_bo_shadowed(bo);
 
 		r = validate(param, bo);
 		if (r)
@@ -385,7 +402,9 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
 			vm->update_funcs->map_table(to_amdgpu_bo_vm(bo));
 			amdgpu_vm_bo_relocated(bo_base);
 		}
+		spin_lock(&vm->status_lock);
 	}
+	spin_unlock(&vm->status_lock);
 
 	amdgpu_vm_eviction_lock(vm);
 	vm->evicting = false;
@@ -406,13 +425,18 @@ int amdgpu_vm_validate_pt_bos(struct amdgpu_device *adev, struct amdgpu_vm *vm,
  */
 bool amdgpu_vm_ready(struct amdgpu_vm *vm)
 {
+	bool empty;
 	bool ret;
 
 	amdgpu_vm_eviction_lock(vm);
 	ret = !vm->evicting;
 	amdgpu_vm_eviction_unlock(vm);
 
-	return ret && list_empty(&vm->evicted);
+	spin_lock(&vm->status_lock);
+	empty = list_empty(&vm->evicted);
+	spin_unlock(&vm->status_lock);
+
+	return ret && empty;
 }
 
 /**
@@ -680,9 +704,14 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	struct amdgpu_vm_update_params params;
 	struct amdgpu_vm_bo_base *entry;
 	bool flush_tlb_needed = false;
+	LIST_HEAD(relocated);
 	int r, idx;
 
-	if (list_empty(&vm->relocated))
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->relocated, &relocated);
+	spin_unlock(&vm->status_lock);
+
+	if (list_empty(&relocated))
 		return 0;
 
 	if (!drm_dev_enter(adev_to_drm(adev), &idx))
@@ -697,7 +726,7 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	if (r)
 		goto error;
 
-	list_for_each_entry(entry, &vm->relocated, vm_status) {
+	list_for_each_entry(entry, &relocated, vm_status) {
 		/* vm_flush_needed after updating moved PDEs */
 		flush_tlb_needed |= entry->moved;
 
@@ -713,9 +742,8 @@ int amdgpu_vm_update_pdes(struct amdgpu_device *adev,
 	if (flush_tlb_needed)
 		atomic64_inc(&vm->tlb_seq);
 
-	while (!list_empty(&vm->relocated)) {
-		entry = list_first_entry(&vm->relocated,
-					 struct amdgpu_vm_bo_base,
+	while (!list_empty(&relocated)) {
+		entry = list_first_entry(&relocated, struct amdgpu_vm_bo_base,
 					 vm_status);
 		amdgpu_vm_bo_idle(entry);
 	}
@@ -912,6 +940,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
 {
 	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;
@@ -936,7 +965,6 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
 		amdgpu_bo_get_memory(bo_va->base.bo, vram_mem,
 				gtt_mem, cpu_mem);
 	}
-	spin_lock(&vm->invalidated_lock);
 	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
 		if (!bo_va->base.bo)
 			continue;
@@ -949,7 +977,7 @@ void amdgpu_vm_get_memory(struct amdgpu_vm *vm, uint64_t *vram_mem,
 		amdgpu_bo_get_memory(bo_va->base.bo, vram_mem,
 				gtt_mem, cpu_mem);
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 }
 /**
  * amdgpu_vm_bo_update - update all BO mappings in the vm page table
@@ -1278,24 +1306,29 @@ int amdgpu_vm_clear_freed(struct amdgpu_device *adev,
 int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 			   struct amdgpu_vm *vm)
 {
-	struct amdgpu_bo_va *bo_va, *tmp;
+	struct amdgpu_bo_va *bo_va;
 	struct dma_resv *resv;
 	bool clear;
 	int r;
 
-	list_for_each_entry_safe(bo_va, tmp, &vm->moved, base.vm_status) {
+	spin_lock(&vm->status_lock);
+	while (!list_empty(&vm->moved)) {
+		bo_va = list_first_entry(&vm->moved, struct amdgpu_bo_va,
+					 base.vm_status);
+		spin_unlock(&vm->status_lock);
+
 		/* Per VM BOs never need to bo cleared in the page tables */
 		r = amdgpu_vm_bo_update(adev, bo_va, false);
 		if (r)
 			return r;
+		spin_lock(&vm->status_lock);
 	}
 
-	spin_lock(&vm->invalidated_lock);
 	while (!list_empty(&vm->invalidated)) {
 		bo_va = list_first_entry(&vm->invalidated, struct amdgpu_bo_va,
 					 base.vm_status);
 		resv = bo_va->base.bo->tbo.base.resv;
-		spin_unlock(&vm->invalidated_lock);
+		spin_unlock(&vm->status_lock);
 
 		/* Try to reserve the BO to avoid clearing its ptes */
 		if (!amdgpu_vm_debug && dma_resv_trylock(resv))
@@ -1310,9 +1343,9 @@ int amdgpu_vm_handle_moved(struct amdgpu_device *adev,
 
 		if (!clear)
 			dma_resv_unlock(resv);
-		spin_lock(&vm->invalidated_lock);
+		spin_lock(&vm->status_lock);
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	return 0;
 }
@@ -1387,7 +1420,7 @@ static void amdgpu_vm_bo_insert_map(struct amdgpu_device *adev,
 
 	if (bo && bo->tbo.base.resv == vm->root.bo->tbo.base.resv &&
 	    !bo_va->base.moved) {
-		list_move(&bo_va->base.vm_status, &vm->moved);
+		amdgpu_vm_bo_moved(&bo_va->base);
 	}
 	trace_amdgpu_vm_bo_map(bo_va, mapping);
 }
@@ -1763,9 +1796,9 @@ void amdgpu_vm_bo_del(struct amdgpu_device *adev,
 		}
 	}
 
-	spin_lock(&vm->invalidated_lock);
+	spin_lock(&vm->status_lock);
 	list_del(&bo_va->base.vm_status);
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 
 	list_for_each_entry_safe(mapping, next, &bo_va->valids, list) {
 		list_del(&mapping->list);
@@ -2019,9 +2052,11 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 	INIT_LIST_HEAD(&vm->moved);
 	INIT_LIST_HEAD(&vm->idle);
 	INIT_LIST_HEAD(&vm->invalidated);
-	spin_lock_init(&vm->invalidated_lock);
+	spin_lock_init(&vm->status_lock);
 	INIT_LIST_HEAD(&vm->freed);
 	INIT_LIST_HEAD(&vm->done);
+	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,
@@ -2223,6 +2258,8 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 
 	amdgpu_amdkfd_gpuvm_destroy_cb(adev, vm);
 
+	flush_work(&vm->pt_free_work);
+
 	root = amdgpu_bo_ref(vm->root.bo);
 	amdgpu_bo_reserve(root, true);
 	amdgpu_vm_set_pasid(adev, vm, 0);
@@ -2484,8 +2521,7 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, u32 pasid,
 		/* Intentionally setting invalid PTE flag
 		 * combination to force a no-retry-fault
 		 */
-		flags = AMDGPU_PTE_EXECUTABLE | AMDGPU_PDE_PTE |
-			AMDGPU_PTE_TF;
+		flags = AMDGPU_PTE_SNOOPED | AMDGPU_PTE_PRT;
 		value = 0;
 	} else if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_NEVER) {
 		/* Redirect the access to the dummy page */
@@ -2548,6 +2584,7 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 	unsigned int total_done_objs = 0;
 	unsigned int id = 0;
 
+	spin_lock(&vm->status_lock);
 	seq_puts(m, "\tIdle BOs:\n");
 	list_for_each_entry_safe(bo_va, tmp, &vm->idle, base.vm_status) {
 		if (!bo_va->base.bo)
@@ -2585,7 +2622,6 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 	id = 0;
 
 	seq_puts(m, "\tInvalidated BOs:\n");
-	spin_lock(&vm->invalidated_lock);
 	list_for_each_entry_safe(bo_va, tmp, &vm->invalidated, base.vm_status) {
 		if (!bo_va->base.bo)
 			continue;
@@ -2600,7 +2636,7 @@ void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m)
 			continue;
 		total_done += amdgpu_bo_print_info(id++, bo_va->base.bo, m);
 	}
-	spin_unlock(&vm->invalidated_lock);
+	spin_unlock(&vm->status_lock);
 	total_done_objs = id;
 
 	seq_printf(m, "\tTotal idle size:        %12lld\tobjs:\t%d\n", total_idle,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 9ecb7f663e19..83acb7bd80fe 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -254,6 +254,9 @@ struct amdgpu_vm {
 	bool			evicting;
 	unsigned int		saved_flags;
 
+	/* Lock to protect vm_bo add/del/move on all lists of vm */
+	spinlock_t		status_lock;
+
 	/* BOs who needs a validation */
 	struct list_head	evicted;
 
@@ -268,7 +271,6 @@ struct amdgpu_vm {
 
 	/* regular invalidated BOs, but not yet updated in the PT */
 	struct list_head	invalidated;
-	spinlock_t		invalidated_lock;
 
 	/* BO mappings freed, but not yet updated in the PT */
 	struct list_head	freed;
@@ -276,6 +278,10 @@ struct amdgpu_vm {
 	/* BOs which are invalidated, has been updated in the PTs */
 	struct list_head        done;
 
+	/* PT BOs scheduled to free and fill with zero if vm_resv is not hold */
+	struct list_head	pt_freed;
+	struct work_struct	pt_free_work;
+
 	/* contains the page directory */
 	struct amdgpu_vm_bo_base     root;
 	struct dma_fence	*last_update;
@@ -471,6 +477,7 @@ int amdgpu_vm_pde_update(struct amdgpu_vm_update_params *params,
 int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
 			  uint64_t start, uint64_t end,
 			  uint64_t dst, uint64_t flags);
+void amdgpu_vm_pt_free_work(struct work_struct *work);
 
 #if defined(CONFIG_DEBUG_FS)
 void amdgpu_debugfs_vm_bo_info(struct amdgpu_vm *vm, struct seq_file *m);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
index 88de9f0d4728..358b91243e37 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_pt.c
@@ -637,10 +637,34 @@ static void amdgpu_vm_pt_free(struct amdgpu_vm_bo_base *entry)
 	}
 	ttm_bo_set_bulk_move(&entry->bo->tbo, NULL);
 	entry->bo->vm_bo = NULL;
+
+	spin_lock(&entry->vm->status_lock);
 	list_del(&entry->vm_status);
+	spin_unlock(&entry->vm->status_lock);
 	amdgpu_bo_unref(&entry->bo);
 }
 
+void amdgpu_vm_pt_free_work(struct work_struct *work)
+{
+	struct amdgpu_vm_bo_base *entry, *next;
+	struct amdgpu_vm *vm;
+	LIST_HEAD(pt_freed);
+
+	vm = container_of(work, struct amdgpu_vm, pt_free_work);
+
+	spin_lock(&vm->status_lock);
+	list_splice_init(&vm->pt_freed, &pt_freed);
+	spin_unlock(&vm->status_lock);
+
+	/* flush_work in amdgpu_vm_fini ensure vm->root.bo is valid. */
+	amdgpu_bo_reserve(vm->root.bo, true);
+
+	list_for_each_entry_safe(entry, next, &pt_freed, vm_status)
+		amdgpu_vm_pt_free(entry);
+
+	amdgpu_bo_unreserve(vm->root.bo);
+}
+
 /**
  * amdgpu_vm_pt_free_dfs - free PD/PT levels
  *
@@ -652,11 +676,24 @@ static void amdgpu_vm_pt_free(struct amdgpu_vm_bo_base *entry)
  */
 static void amdgpu_vm_pt_free_dfs(struct amdgpu_device *adev,
 				  struct amdgpu_vm *vm,
-				  struct amdgpu_vm_pt_cursor *start)
+				  struct amdgpu_vm_pt_cursor *start,
+				  bool unlocked)
 {
 	struct amdgpu_vm_pt_cursor cursor;
 	struct amdgpu_vm_bo_base *entry;
 
+	if (unlocked) {
+		spin_lock(&vm->status_lock);
+		for_each_amdgpu_vm_pt_dfs_safe(adev, vm, start, cursor, entry)
+			list_move(&entry->vm_status, &vm->pt_freed);
+
+		if (start)
+			list_move(&start->entry->vm_status, &vm->pt_freed);
+		spin_unlock(&vm->status_lock);
+		schedule_work(&vm->pt_free_work);
+		return;
+	}
+
 	for_each_amdgpu_vm_pt_dfs_safe(adev, vm, start, cursor, entry)
 		amdgpu_vm_pt_free(entry);
 
@@ -673,7 +710,7 @@ static void amdgpu_vm_pt_free_dfs(struct amdgpu_device *adev,
  */
 void amdgpu_vm_pt_free_root(struct amdgpu_device *adev, struct amdgpu_vm *vm)
 {
-	amdgpu_vm_pt_free_dfs(adev, vm, NULL);
+	amdgpu_vm_pt_free_dfs(adev, vm, NULL, false);
 }
 
 /**
@@ -966,7 +1003,8 @@ int amdgpu_vm_ptes_update(struct amdgpu_vm_update_params *params,
 				if (cursor.entry->bo) {
 					params->table_freed = true;
 					amdgpu_vm_pt_free_dfs(adev, params->vm,
-							      &cursor);
+							      &cursor,
+							      params->unlocked);
 				}
 				amdgpu_vm_pt_next(adev, &cursor);
 			}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
index 03ec099d64e0..2b0669c464f6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c
@@ -212,12 +212,15 @@ static int amdgpu_vm_sdma_update(struct amdgpu_vm_update_params *p,
 	int r;
 
 	/* Wait for PD/PT moves to be completed */
-	dma_resv_for_each_fence(&cursor, bo->tbo.base.resv,
-				DMA_RESV_USAGE_KERNEL, fence) {
+	dma_resv_iter_begin(&cursor, bo->tbo.base.resv, DMA_RESV_USAGE_KERNEL);
+	dma_resv_for_each_fence_unlocked(&cursor, fence) {
 		r = amdgpu_sync_fence(&p->job->sync, fence);
-		if (r)
+		if (r) {
+			dma_resv_iter_end(&cursor);
 			return r;
+		}
 	}
+	dma_resv_iter_end(&cursor);
 
 	do {
 		ndw = p->num_dw_left;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index 10477ca52a41..47159e9a0884 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -406,14 +406,14 @@ struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev)
 		if (adev->reset_domain->type != XGMI_HIVE) {
 			hive->reset_domain =
 				amdgpu_reset_create_reset_domain(XGMI_HIVE, "amdgpu-reset-hive");
-				if (!hive->reset_domain) {
-					dev_err(adev->dev, "XGMI: failed initializing reset domain for xgmi hive\n");
-					ret = -ENOMEM;
-					kobject_put(&hive->kobj);
-					kfree(hive);
-					hive = NULL;
-					goto pro_end;
-				}
+			if (!hive->reset_domain) {
+				dev_err(adev->dev, "XGMI: failed initializing reset domain for xgmi hive\n");
+				ret = -ENOMEM;
+				kobject_put(&hive->kobj);
+				kfree(hive);
+				hive = NULL;
+				goto pro_end;
+			}
 		} else {
 			amdgpu_reset_get_reset_domain(adev->reset_domain);
 			hive->reset_domain = adev->reset_domain;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
index 552e6fb55aa8..30dcc1681b4e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -43,6 +43,7 @@ struct amdgpu_hive_info {
 	} pstate;
 
 	struct amdgpu_reset_domain *reset_domain;
+	uint32_t device_remove_count;
 };
 
 struct amdgpu_pcs_ras_field {
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
index fa718318568e..ce8c792cef1a 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c
@@ -475,8 +475,13 @@ static void gfx_v11_0_init_rlcp_rlcv_microcode(struct amdgpu_device *adev)
 	const struct rlc_firmware_header_v2_3 *rlc_hdr;
 
 	rlc_hdr = (const struct rlc_firmware_header_v2_3 *)adev->gfx.rlc_fw->data;
+	adev->gfx.rlcp_ucode_version = le32_to_cpu(rlc_hdr->rlcp_ucode_version);
+	adev->gfx.rlcp_ucode_feature_version = le32_to_cpu(rlc_hdr->rlcp_ucode_feature_version);
 	adev->gfx.rlc.rlcp_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlcp_ucode_size_bytes);
 	adev->gfx.rlc.rlcp_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlcp_ucode_offset_bytes);
+
+	adev->gfx.rlcv_ucode_version = le32_to_cpu(rlc_hdr->rlcv_ucode_version);
+	adev->gfx.rlcv_ucode_feature_version = le32_to_cpu(rlc_hdr->rlcv_ucode_feature_version);
 	adev->gfx.rlc.rlcv_ucode_size_bytes = le32_to_cpu(rlc_hdr->rlcv_ucode_size_bytes);
 	adev->gfx.rlc.rlcv_ucode = (u8 *)rlc_hdr + le32_to_cpu(rlc_hdr->rlcv_ucode_offset_bytes);
 }
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 4603653916f5..67ca16a8027c 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -1103,10 +1103,13 @@ static void gmc_v9_0_get_vm_pde(struct amdgpu_device *adev, int level,
 			*flags |= AMDGPU_PDE_BFS(0x9);
 
 	} else if (level == AMDGPU_VM_PDB0) {
-		if (*flags & AMDGPU_PDE_PTE)
+		if (*flags & AMDGPU_PDE_PTE) {
 			*flags &= ~AMDGPU_PDE_PTE;
-		else
+			if (!(*flags & AMDGPU_PTE_VALID))
+				*addr |= 1 << PAGE_SHIFT;
+		} else {
 			*flags |= AMDGPU_PTE_TF;
+		}
 	}
 }
 
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
index 39405f0db824..9c8b5fd99037 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c
@@ -1761,21 +1761,23 @@ static const struct amdgpu_ring_funcs vcn_v3_0_dec_sw_ring_vm_funcs = {
 	.emit_reg_write_reg_wait = amdgpu_ring_emit_reg_write_reg_wait_helper,
 };
 
-static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p)
+static int vcn_v3_0_limit_sched(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job)
 {
 	struct drm_gpu_scheduler **scheds;
 
 	/* The create msg must be in the first IB submitted */
-	if (atomic_read(&p->entity->fence_seq))
+	if (atomic_read(&job->base.entity->fence_seq))
 		return -EINVAL;
 
 	scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_DEC]
 		[AMDGPU_RING_PRIO_DEFAULT].sched;
-	drm_sched_entity_modify_sched(p->entity, scheds, 1);
+	drm_sched_entity_modify_sched(job->base.entity, scheds, 1);
 	return 0;
 }
 
-static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr)
+static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
+			    uint64_t addr)
 {
 	struct ttm_operation_ctx ctx = { false, false };
 	struct amdgpu_bo_va_mapping *map;
@@ -1846,7 +1848,7 @@ static int vcn_v3_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr)
 		if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11)
 			continue;
 
-		r = vcn_v3_0_limit_sched(p);
+		r = vcn_v3_0_limit_sched(p, job);
 		if (r)
 			goto out;
 	}
@@ -1860,7 +1862,7 @@ static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
 					   struct amdgpu_job *job,
 					   struct amdgpu_ib *ib)
 {
-	struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched);
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
 	uint32_t msg_lo = 0, msg_hi = 0;
 	unsigned i;
 	int r;
@@ -1879,7 +1881,8 @@ static int vcn_v3_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
 			msg_hi = val;
 		} else if (reg == PACKET0(p->adev->vcn.internal.cmd, 0) &&
 			   val == 0) {
-			r = vcn_v3_0_dec_msg(p, ((u64)msg_hi) << 32 | msg_lo);
+			r = vcn_v3_0_dec_msg(p, job,
+					     ((u64)msg_hi) << 32 | msg_lo);
 			if (r)
 				return r;
 		}
diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
index 09c89faa8c27..b6f73b87c47e 100644
--- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c
@@ -1591,21 +1591,23 @@ static void vcn_v4_0_unified_ring_set_wptr(struct amdgpu_ring *ring)
 	}
 }
 
-static int vcn_v4_0_limit_sched(struct amdgpu_cs_parser *p)
+static int vcn_v4_0_limit_sched(struct amdgpu_cs_parser *p,
+				struct amdgpu_job *job)
 {
 	struct drm_gpu_scheduler **scheds;
 
 	/* The create msg must be in the first IB submitted */
-	if (atomic_read(&p->entity->fence_seq))
+	if (atomic_read(&job->base.entity->fence_seq))
 		return -EINVAL;
 
 	scheds = p->adev->gpu_sched[AMDGPU_HW_IP_VCN_ENC]
 		[AMDGPU_RING_PRIO_0].sched;
-	drm_sched_entity_modify_sched(p->entity, scheds, 1);
+	drm_sched_entity_modify_sched(job->base.entity, scheds, 1);
 	return 0;
 }
 
-static int vcn_v4_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr)
+static int vcn_v4_0_dec_msg(struct amdgpu_cs_parser *p, struct amdgpu_job *job,
+			    uint64_t addr)
 {
 	struct ttm_operation_ctx ctx = { false, false };
 	struct amdgpu_bo_va_mapping *map;
@@ -1676,7 +1678,7 @@ static int vcn_v4_0_dec_msg(struct amdgpu_cs_parser *p, uint64_t addr)
 		if (create[0] == 0x7 || create[0] == 0x10 || create[0] == 0x11)
 			continue;
 
-		r = vcn_v4_0_limit_sched(p);
+		r = vcn_v4_0_limit_sched(p, job);
 		if (r)
 			goto out;
 	}
@@ -1689,32 +1691,34 @@ out:
 #define RADEON_VCN_ENGINE_TYPE_DECODE                                 (0x00000003)
 
 static int vcn_v4_0_ring_patch_cs_in_place(struct amdgpu_cs_parser *p,
-				struct amdgpu_job *job,
-				struct amdgpu_ib *ib)
+					   struct amdgpu_job *job,
+					   struct amdgpu_ib *ib)
 {
-	struct amdgpu_ring *ring = to_amdgpu_ring(p->entity->rq->sched);
-	struct amdgpu_vcn_decode_buffer *decode_buffer = NULL;
+	struct amdgpu_ring *ring = amdgpu_job_ring(job);
+	struct amdgpu_vcn_decode_buffer *decode_buffer;
+	uint64_t addr;
 	uint32_t val;
-	int r = 0;
 
 	/* The first instance can decode anything */
 	if (!ring->me)
-		return r;
+		return 0;
 
 	/* unified queue ib header has 8 double words. */
 	if (ib->length_dw < 8)
-		return r;
+		return 0;
 
 	val = amdgpu_ib_get_value(ib, 6); //RADEON_VCN_ENGINE_TYPE
+	if (val != RADEON_VCN_ENGINE_TYPE_DECODE)
+		return 0;
 
-	if (val == RADEON_VCN_ENGINE_TYPE_DECODE) {
-		decode_buffer = (struct amdgpu_vcn_decode_buffer *)&ib->ptr[10];
+	decode_buffer = (struct amdgpu_vcn_decode_buffer *)&ib->ptr[10];
 
-		if (decode_buffer->valid_buf_flag  & 0x1)
-			r = vcn_v4_0_dec_msg(p, ((u64)decode_buffer->msg_buffer_address_hi) << 32 |
-						decode_buffer->msg_buffer_address_lo);
-	}
-	return r;
+	if (!(decode_buffer->valid_buf_flag  & 0x1))
+		return 0;
+
+	addr = ((u64)decode_buffer->msg_buffer_address_hi) << 32 |
+		decode_buffer->msg_buffer_address_lo;
+	return vcn_v4_0_dec_msg(p, job, addr);
 }
 
 static const struct amdgpu_ring_funcs vcn_v4_0_unified_ring_vm_funcs = {