Skip to contentSkip to conversation

drm/amdgpu: fix kernel hang when starting VNC server

PATCH In review Xiangliang.Yu 2 participants 2 messages Oct 20, 2017 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 drm/amdgpu: fix kernel hang when starting VNC server

After starting VNC server or running CTS test, kernel will hang and can see below call trace:

[961816] INFO: task khugepaged:42 blocked for more than 120 seconds.
[968581] Tainted: G OE 4.13.0 #1
[973495] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables
this message.
[980962] khugepaged D 0 42 2 0x00000000
[980967] Call Trace:

[980977]  __schedule+0x28d/0x890
[980982]  schedule+0x36/0x80
[980986]  rwsem_down_read_failed+0x139/0x1c0
[980991]  ? update_curr+0x100/0x1c0
[981004]  call_rwsem_down_read_failed+0x18/0x30
[981007]  down_read+0x20/0x40
[981012]  khugepaged_scan_mm_slot+0x78/0x1ac0
[981018]  ? __switch_to+0x23e/0x4a0
[981022]  ? finish_task_switch+0x79/0x240
[981026]  khugepaged+0x146/0x480
[981031]  ? remove_wait_queue+0x60/0x60
[981035]  kthread+0x109/0x140
[981037]  ? khugepaged_scan_mm_slot+0x1ac0/0x1ac0
[981039]  ? kthread_park+0x60/0x60
[981044]  ret_from_fork+0x25/0x30

After checking code and found 'commit b72cf4fca2bb7 ("drm/amdgpu: move taking mmap_sem into get_user_pages v2")' forget to drop one of up_read usage.

---
 drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 5 +----
 1 file changed, 1 insertion(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +1 -4
index fb72edc..08b37c2 100644
@@ -323,7 +323,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
 						 bo->tbo.ttm->pages);
 		if (r)
-			goto unlock_mmap_sem;
+			goto release_object;
 
 		r = amdgpu_bo_reserve(bo, true);
 		if (r)
@@ -348,9 +348,6 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
 free_pages:
 	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages, false);
 
-unlock_mmap_sem:
-	up_read(&current->mm->mmap_sem);
-
 release_object:
 	drm_gem_object_put_unlocked(gobj);
 
Signature · 6 lines
2.7.4

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

A message that is not in the archive (sent off-list or not yet received): 1508492712-28961-1-git-send-email-Xiangliang.Yu-5C7GfCeVMHo@public.gmane.org

Reviewed-by
#2 lore raw

Am 20.10.2017 um 11:45 schrieb Xiangliang.Yu:

Quoted 25 lines

After starting VNC server or running CTS test, kernel will hang and can see below call trace:

[961816] INFO: task khugepaged:42 blocked for more than 120 seconds.
[968581] Tainted: G OE 4.13.0 #1
[973495] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables
this message.
[980962] khugepaged D 0 42 2 0x00000000
[980967] Call Trace:

[980977]  __schedule+0x28d/0x890
[980982]  schedule+0x36/0x80
[980986]  rwsem_down_read_failed+0x139/0x1c0
[980991]  ? update_curr+0x100/0x1c0
[981004]  call_rwsem_down_read_failed+0x18/0x30
[981007]  down_read+0x20/0x40
[981012]  khugepaged_scan_mm_slot+0x78/0x1ac0
[981018]  ? __switch_to+0x23e/0x4a0
[981022]  ? finish_task_switch+0x79/0x240
[981026]  khugepaged+0x146/0x480
[981031]  ? remove_wait_queue+0x60/0x60
[981035]  kthread+0x109/0x140
[981037]  ? khugepaged_scan_mm_slot+0x1ac0/0x1ac0
[981039]  ? kthread_park+0x60/0x60
[981044]  ret_from_fork+0x25/0x30
Quoted 5 lines

After checking code and found 'commit b72cf4fca2bb7 ("drm/amdgpu: move taking mmap_sem into get_user_pages v2")' forget to drop one of up_read usage.

Quoted 18 lines
---
  drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c | 5 +----
  1 file changed, 1 insertion(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +1 -1
index fb72edc..08b37c2 100644
@@ -323,7 +323,7 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
  		r = amdgpu_ttm_tt_get_user_pages(bo->tbo.ttm,
  						 bo->tbo.ttm->pages);
  		if (r)
-			goto unlock_mmap_sem;
+			goto release_object;
 
  		r = amdgpu_bo_reserve(bo, true);
  		if (r)

drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data, · line 353

Quoted 9 lines
drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c +0 -3
@@ -348,9 +348,6 @@ int amdgpu_gem_userptr_ioctl(struct drm_device *dev, void *data,
  free_pages:
  	release_pages(bo->tbo.ttm->pages, bo->tbo.ttm->num_pages, false);
 
-unlock_mmap_sem:
-	up_read(&current->mm->mmap_sem);
-
  release_object:
  	drm_gem_object_put_unlocked(gobj);
 

_______________________________________________
amd-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/amd-gfx