[PATCH] drm/amdgpu:fix waiting on dirty fence

[PATCH] drm/amdgpu:fix waiting on dirty fence

[Monk Liu]

if bo->shadow is NULL (race issue:BO shadow was just released
and gpu-reset kick in but BO hasn't yet) recover_vram_from_shadow
won't set @next, so the following "fence=next"
will wrongly use a fence pointer which may already dirty.
fixing it by set next to NULL prior to recover_vram_from_shadow

Change-Id: Iba595789f8ce095c0c023e01193715212691239d
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 4dbd121..815e44a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2601,6 +2601,7 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, struct amdgpu_job *job, b
 	ring = adev->mman.buffer_funcs_ring;
 	mutex_lock(&adev->shadow_list_lock);
 	list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
+		next = NULL;
 		amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
 		if (fence) {
 			r = fence_wait(fence, false);
@@ -2766,6 +2767,7 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 			DRM_INFO("recover vram bo from shadow\n");
 			mutex_lock(&adev->shadow_list_lock);
 			list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
+				next = NULL;
 				amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
 				if (fence) {
 					r = fence_wait(fence, false);
-- 
2.7.4

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

RE: [PATCH] drm/amdgpu:fix waiting on dirty fence

[Zhou, David(ChunMing)]

Reviewed-by: Chunming Zhou<david1.zhou@amd.com>

-----Original Message-----
From: amd-gfx [mailto:amd-gfx-bounces@lists.freedesktop.org] On Behalf Of Monk Liu
Sent: Monday, May 01, 2017 4:18 PM
To: amd-gfx@lists.freedesktop.org
Cc: Liu, Monk <Monk.Liu@amd.com>
Subject: [PATCH] drm/amdgpu:fix waiting on dirty fence

if bo->shadow is NULL (race issue:BO shadow was just released and gpu-reset kick in but BO hasn't yet) recover_vram_from_shadow won't set @next, so the following "fence=next"
will wrongly use a fence pointer which may already dirty.
fixing it by set next to NULL prior to recover_vram_from_shadow

Change-Id: Iba595789f8ce095c0c023e01193715212691239d
Signed-off-by: Monk Liu <Monk.Liu@amd.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index 4dbd121..815e44a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -2601,6 +2601,7 @@ int amdgpu_sriov_gpu_reset(struct amdgpu_device *adev, struct amdgpu_job *job, b
 	ring = adev->mman.buffer_funcs_ring;
 	mutex_lock(&adev->shadow_list_lock);
 	list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
+		next = NULL;
 		amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
 		if (fence) {
 			r = fence_wait(fence, false);
@@ -2766,6 +2767,7 @@ int amdgpu_gpu_reset(struct amdgpu_device *adev)
 			DRM_INFO("recover vram bo from shadow\n");
 			mutex_lock(&adev->shadow_list_lock);
 			list_for_each_entry_safe(bo, tmp, &adev->shadow_list, shadow_list) {
+				next = NULL;
 				amdgpu_recover_vram_from_shadow(adev, ring, bo, &next);
 				if (fence) {
 					r = fence_wait(fence, false);
--
2.7.4

_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx