Skip to contentSkip to conversation

FAILED: patch "[PATCH] drm/amdgpu: add missing lock to amdgpu_ttm_access_memory_sdma" failed to apply to 6.1-stable tree

PATCH New [email protected] 2 participants 3 messages Jan 5 – Jan 7

Subsystems (3)
Stable branch DRM drivers Radeon and amdgpu DRM drivers
lore
Share & tools
lore

[email protected]

author maintainer
#0 lore raw

The patch below does not apply to the 6.1-stable tree.
If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <[email protected]>.

To reproduce the conflict and resubmit, you may use the following commands:

git fetch https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/ linux-6.1.y
git checkout FETCH_HEAD
git cherry-pick -x 4fa944255be521b1bbd9780383f77206303a3a5c
# <resolve conflicts, build, test, etc.>
git commit -s
git send-email --to '<[email protected]>' --in-reply-to '2026010504-compacter-plow-0408@gregkh' --subject-prefix 'PATCH 6.1.y' HEAD^..

Possible dependencies:

thanks,

greg k-h

------------------ original commit in Linus's tree ------------------

From 4fa944255be521b1bbd9780383f77206303a3a5c Mon Sep 17 00:00:00 2001
From: Pierre-Eric Pelloux-Prayer <[email protected]>
Date: Tue, 25 Nov 2025 10:48:39 +0100
Subject: [PATCH] drm/amdgpu: add missing lock to amdgpu_ttm_access_memory_sdma
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Users of ttm entities need to hold the gtt_window_lock before using them to guarantee proper ordering of jobs.

Cc [email protected]
Fixes cb5cc4f573e1 ("drm/amdgpu: improve debug VRAM access performance using sdma")
Signed-off-by Pierre-Eric Pelloux-Prayer <[email protected]>
Reviewed-by Christian König <[email protected]>
Signed-off-by Alex Deucher <[email protected]>
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +2 -0
index 5475f7117f10..1b799f895dbf 100644
@@ -1486,6 +1486,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	if (r)
 		goto out;
 
+	mutex_lock(&adev->mman.gtt_window_lock);
 	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);
 	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +
 		src_mm.start;

@@ -1500,6 +1501,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	WARN_ON(job->ibs[0].length_dw > num_dw);
 
 	fence = amdgpu_job_submit(job);
+	mutex_unlock(&adev->mman.gtt_window_lock);
 
 	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))
 		r = -ETIMEDOUT;
maintainer
#1 lore raw

PATCH 1/2 drm/amdgpu: cleanup scheduler job initialization v2

From: Christian König <[email protected]>

[ Upstream commit f7d66fb2ea43a3016e78a700a2ca6c77a74579f9 ]

Init the DRM scheduler base class while allocating the job.

This makes the whole handling much more cleaner.

v2: fix coding style

Signed-off-by Christian König <[email protected]>
Reviewed-by Luben Tuikov <[email protected]>
Link https://patchwork.freedesktop.org/patch/msgid/[email protected]

Stable-dep-of: 4fa944255be5 ("drm/amdgpu: add missing lock to amdgpu_ttm_access_memory_sdma")

Signed-off-by Sasha Levin <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c  |  2 +-
 drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c      |  8 +--
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.c     | 44 ++++++++-------
 drivers/gpu/drm/amd/amdgpu/amdgpu_job.h     | 14 +++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c    |  7 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c     | 56 +++++++------------
 drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c     |  9 ++-
 drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c     | 13 +++--
 drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c     | 22 +++++---
 drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c | 61 +++++++++++----------
 drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c      | 12 ++--
 drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c       |  8 +--
 drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c       | 12 ++--
 drivers/gpu/drm/amd/amdkfd/kfd_migrate.c    | 17 ++----
 14 files changed, 135 insertions(+), 150 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c +1 -1
index c31e5f9d63da..330d1404988c 100644
@@ -673,7 +673,7 @@ int amdgpu_amdkfd_submit_ib(struct amdgpu_device *adev,
 		goto err;
 	}
 
-	ret = amdgpu_job_alloc(adev, 1, &job, NULL);
+	ret = amdgpu_job_alloc(adev, NULL, NULL, NULL, 1, &job);
 	if (ret)
 		goto err;
drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c +2 -6
index f83eff3c62e1..faeabe197dc6 100644
@@ -285,12 +285,8 @@ static int amdgpu_cs_pass1(struct amdgpu_cs_parser *p,
 	}
 
 	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);
+		ret = amdgpu_job_alloc(p->adev, vm, p->entities[i], vm,
+				       num_ibs[i], &p->jobs[i]);
 		if (ret)
 			goto free_all_kdata;
 	}
drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +23 -21

85 lines not on this page · raw message

drivers/gpu/drm/amd/amdgpu/amdgpu_job.h +8 -6
index ab7b150e5d50..e2ecab977d0d 100644
@@ -78,18 +78,20 @@ 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,
-		enum amdgpu_ib_pool_type pool, struct amdgpu_job **job);
+int amdgpu_job_alloc(struct amdgpu_device *adev, struct amdgpu_vm *vm,
+		     struct drm_sched_entity *entity, void *owner,
+		     unsigned int num_ibs, struct amdgpu_job **job);
+int amdgpu_job_alloc_with_ib(struct amdgpu_device *adev,
+			     struct drm_sched_entity *entity, void *owner,
+			     size_t size, enum amdgpu_ib_pool_type pool_type,
+			     struct amdgpu_job **job);
 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);
+struct dma_fence *amdgpu_job_submit(struct amdgpu_job *job);
 int amdgpu_job_submit_direct(struct amdgpu_job *job, struct amdgpu_ring *ring,
 			     struct dma_fence **fence);
drivers/gpu/drm/amd/amdgpu/amdgpu_jpeg.c +4 -3
index d5ae89876a06..b6d52c98e52d 100644
@@ -152,14 +152,15 @@ static int amdgpu_jpeg_dec_set_reg(struct amdgpu_ring *ring, uint32_t handle,
 	const unsigned ib_size_dw = 16;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job);
 	if (r)
 		return r;
 
 	ib = &job->ibs[0];
 
-	ib->ptr[0] = PACKETJ(adev->jpeg.internal.jpeg_pitch, 0, 0, PACKETJ_TYPE0);
+	ib->ptr[0] = PACKETJ(adev->jpeg.internal.jpeg_pitch, 0, 0,
+			     PACKETJ_TYPE0);
 	ib->ptr[1] = 0xDEADBEEF;
 	for (i = 2; i < 16; i += 2) {
 		ib->ptr[i] = PACKETJ(0, 0, 0, PACKETJ_TYPE6);
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +21 -35

117 lines not on this page · raw message

drivers/gpu/drm/amd/amdgpu/amdgpu_uvd.c +4 -5
index 6eac649499d3..8baddf79635b 100644
@@ -1132,7 +1132,9 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 	unsigned offset_idx = 0;
 	unsigned offset[3] = { UVD_BASE_SI, 0, 0 };
 
-	r = amdgpu_job_alloc_with_ib(adev, 64, direct ? AMDGPU_IB_POOL_DIRECT :
+	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->uvd.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     64, direct ? AMDGPU_IB_POOL_DIRECT :
 				     AMDGPU_IB_POOL_DELAYED, &job);
 	if (r)
 		return r;

@@ -1181,10 +1183,7 @@ static int amdgpu_uvd_send_msg(struct amdgpu_ring *ring, struct amdgpu_bo *bo,
 		if (r)
 			goto err_free;
 
-		r = amdgpu_job_submit(job, &adev->uvd.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
-		if (r)
-			goto err_free;
+		f = amdgpu_job_submit(job);
 	}
 
 	amdgpu_bo_reserve(bo, true);
drivers/gpu/drm/amd/amdgpu/amdgpu_vce.c +8 -5
index 251416ad4652..1030d15ba4c8 100644
@@ -450,8 +450,10 @@ static int amdgpu_vce_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
 	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-				     AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, &ring->adev->vce.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job);
 	if (r)
 		return r;
 

@@ -538,7 +540,9 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	struct dma_fence *f = NULL;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
+	r = amdgpu_job_alloc_with_ib(ring->adev, &ring->adev->vce.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     ib_size_dw * 4,
 				     direct ? AMDGPU_IB_POOL_DIRECT :
 				     AMDGPU_IB_POOL_DELAYED, &job);
 	if (r)

@@ -570,8 +574,7 @@ static int amdgpu_vce_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
 	if (direct)
 		r = amdgpu_job_submit_direct(job, ring, &f);
 	else
-		r = amdgpu_job_submit(job, &ring->adev->vce.entity,
-				      AMDGPU_FENCE_OWNER_UNDEFINED, &f);
+		f = amdgpu_job_submit(job);
 	if (r)
 		goto err;
drivers/gpu/drm/amd/amdgpu/amdgpu_vcn.c +13 -9
index e2475f656ff2..015f4f18f5f9 100644
@@ -627,15 +627,16 @@ static int amdgpu_vcn_dec_send_msg(struct amdgpu_ring *ring,
 				   struct amdgpu_ib *ib_msg,
 				   struct dma_fence **fence)
 {
+	u64 addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
 	struct amdgpu_device *adev = ring->adev;
 	struct dma_fence *f = NULL;
 	struct amdgpu_job *job;
 	struct amdgpu_ib *ib;
-	uint64_t addr = AMDGPU_GPU_PAGE_ALIGN(ib_msg->gpu_addr);
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(adev, 64,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     64, AMDGPU_IB_POOL_DIRECT,
+				     &job);
 	if (r)
 		goto err;
 

@@ -813,8 +814,9 @@ static int amdgpu_vcn_dec_sw_send_msg(struct amdgpu_ring *ring,
 	if (adev->vcn.using_unified_queue)
 		ib_size_dw += 8;
 
-	r = amdgpu_job_alloc_with_ib(adev, ib_size_dw * 4,
-				AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job);
 	if (r)
 		goto err;
 

@@ -942,8 +944,9 @@ static int amdgpu_vcn_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t hand
 	if (adev->vcn.using_unified_queue)
 		ib_size_dw += 8;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job);
 	if (r)
 		return r;
 

@@ -1008,8 +1011,9 @@ static int amdgpu_vcn_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t han
 	if (adev->vcn.using_unified_queue)
 		ib_size_dw += 8;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL,
+				     ib_size_dw * 4, AMDGPU_IB_POOL_DIRECT,
+				     &job);
 	if (r)
 		return r;
drivers/gpu/drm/amd/amdgpu/amdgpu_vm_sdma.c +31 -30

111 lines not on this page · raw message

drivers/gpu/drm/amd/amdgpu/gmc_v10_0.c +4 -8
index 35921b41fc27..918cf169cb95 100644
@@ -371,7 +371,9 @@ 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(adev, 16 * 4, AMDGPU_IB_POOL_IMMEDIATE,
+	r = amdgpu_job_alloc_with_ib(ring->adev, &adev->mman.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     16 * 4, AMDGPU_IB_POOL_IMMEDIATE,
 				     &job);
 	if (r)
 		goto error_alloc;

@@ -380,10 +382,7 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 	job->vm_needs_flush = true;
 	job->ibs->ptr[job->ibs->length_dw++] = ring->funcs->nop;
 	amdgpu_ring_pad_ib(ring, &job->ibs[0]);
-	r = amdgpu_job_submit(job, &adev->mman.entity,
-			      AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
-	if (r)
-		goto error_submit;
+	fence = amdgpu_job_submit(job);
 
 	mutex_unlock(&adev->mman.gtt_window_lock);
 

@@ -392,9 +391,6 @@ static void gmc_v10_0_flush_gpu_tlb(struct amdgpu_device *adev, uint32_t vmid,
 
 	return;
 
-error_submit:
-	amdgpu_job_free(job);
-
 error_alloc:
 	mutex_unlock(&adev->mman.gtt_window_lock);
 	DRM_ERROR("Error flushing GPU TLB using the SDMA (%d)!\n", r);
drivers/gpu/drm/amd/amdgpu/uvd_v6_0.c +4 -4
index 375c440957dc..5fe872f4bea7 100644
@@ -216,8 +216,8 @@ static int uvd_v6_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle
 	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job);
 	if (r)
 		return r;
 

@@ -280,8 +280,8 @@ static int uvd_v6_0_enc_get_destroy_msg(struct amdgpu_ring *ring,
 	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job);
 	if (r)
 		return r;
drivers/gpu/drm/amd/amdgpu/uvd_v7_0.c +6 -6
index 1c5d79528ca7..91ef914b4cdc 100644
@@ -213,7 +213,7 @@ static int uvd_v7_0_enc_ring_test_ring(struct amdgpu_ring *ring)
  *
  * Open up a stream for HW test
  */
-static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle,
+static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, u32 handle,
 				       struct amdgpu_bo *bo,
 				       struct dma_fence **fence)
 {

@@ -224,8 +224,8 @@ static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle
 	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job);
 	if (r)
 		return r;
 

@@ -276,7 +276,7 @@ static int uvd_v7_0_enc_get_create_msg(struct amdgpu_ring *ring, uint32_t handle
  *
  * Close up a stream for HW test or if userspace failed to do so
  */
-static int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handle,
+static int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, u32 handle,
 					struct amdgpu_bo *bo,
 					struct dma_fence **fence)
 {

@@ -287,8 +287,8 @@ static int uvd_v7_0_enc_get_destroy_msg(struct amdgpu_ring *ring, uint32_t handl
 	uint64_t addr;
 	int i, r;
 
-	r = amdgpu_job_alloc_with_ib(ring->adev, ib_size_dw * 4,
-					AMDGPU_IB_POOL_DIRECT, &job);
+	r = amdgpu_job_alloc_with_ib(ring->adev, NULL, NULL, ib_size_dw * 4,
+				     AMDGPU_IB_POOL_DIRECT, &job);
 	if (r)
 		return r;
drivers/gpu/drm/amd/amdkfd/kfd_migrate.c +6 -11
index a7ed47cb5bf6..730746512af8 100644
@@ -65,8 +65,11 @@ 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, num_dw * 4 + num_bytes,
-				     AMDGPU_IB_POOL_DELAYED, &job);
+	r = amdgpu_job_alloc_with_ib(adev, &adev->mman.entity,
+				     AMDGPU_FENCE_OWNER_UNDEFINED,
+				     num_dw * 4 + num_bytes,
+				     AMDGPU_IB_POOL_DELAYED,
+				     &job);
 	if (r)
 		return r;
 

@@ -89,18 +92,10 @@ svm_migrate_gart_map(struct amdgpu_ring *ring, uint64_t npages,
 	cpu_addr = &job->ibs[0].ptr[num_dw];
 
 	amdgpu_gart_map(adev, 0, npages, addr, pte_flags, cpu_addr);
-	r = amdgpu_job_submit(job, &adev->mman.entity,
-			      AMDGPU_FENCE_OWNER_UNDEFINED, &fence);
-	if (r)
-		goto error_free;
-
+	fence = amdgpu_job_submit(job);
 	dma_fence_put(fence);
 
 	return r;
-
-error_free:
-	amdgpu_job_free(job);
-	return r;
 }
 
 /**
Signature · 1 line
2.51.0
maintainer
#2 lore raw

PATCH 2/2 drm/amdgpu: add missing lock to amdgpu_ttm_access_memory_sdma

From: Pierre-Eric Pelloux-Prayer <[email protected]>

[ Upstream commit 4fa944255be521b1bbd9780383f77206303a3a5c ]

Users of ttm entities need to hold the gtt_window_lock before using them to guarantee proper ordering of jobs.

Cc [email protected]
Fixes cb5cc4f573e1 ("drm/amdgpu: improve debug VRAM access performance using sdma")
Signed-off-by Pierre-Eric Pelloux-Prayer <[email protected]>
Reviewed-by Christian König <[email protected]>
Signed-off-by Alex Deucher <[email protected]>
Signed-off-by Sasha Levin <[email protected]>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +2 -0
index 3870179acc58..4133afde22b4 100644
@@ -1425,6 +1425,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	if (r)
 		goto out;
 
+	mutex_lock(&adev->mman.gtt_window_lock);
 	amdgpu_res_first(abo->tbo.resource, offset, len, &src_mm);
 	src_addr = amdgpu_ttm_domain_start(adev, bo->resource->mem_type) +
 		src_mm.start;

@@ -1439,6 +1440,7 @@ static int amdgpu_ttm_access_memory_sdma(struct ttm_buffer_object *bo,
 	WARN_ON(job->ibs[0].length_dw > num_dw);
 
 	fence = amdgpu_job_submit(job);
+	mutex_unlock(&adev->mman.gtt_window_lock);
 
 	if (!dma_fence_wait_timeout(fence, false, adev->sdma_timeout))
 		r = -ETIMEDOUT;
Signature · 1 line
2.51.0