1331371 - Intermittent file:///C:/slave/test/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/variables/variable-supports-13.html | application terminated with exit code 1 after Assertion failure: [GFX1]: Failed to create DrawTarget

Status: RESOLVED INCOMPLETE

(Core :: Graphics, defect, P3)

Description

[Treeherder Bug Filer]

Filed by: cbook [at] mozilla.com

https://treeherder.mozilla.org/logviewer.html#?job_id=69264469&repo=autoland

https://archive.mozilla.org/pub/firefox/tinderbox-builds/autoland-win32-debug/1484560342/autoland_win7_vm_gfx-debug_test-reftest-1-bm139-tests1-windows-build486.txt.gz

https://hg.mozilla.org/mozilla-central/raw-file/tip/layout/tools/reftest/reftest-analyzer.xhtml#logurl=https://archive.mozilla.org/pub/firefox/tinderbox-builds/autoland-win32-debug/1484560342/autoland_win7_vm_gfx-debug_test-reftest-1-bm139-tests1-windows-build486.txt.gz&only_show_unexpected=1

Comment 1

[Vincent Liu[:vliu]]

The assertion listed here


The "Type: 3" shows the backend type was set to SKIA.

03:00:37     INFO - Assertion failure: [GFX1]: Failed to create DrawTarget, Type: 3 Size: Size(800,1000), at c:\builds\moz2_slave\autoland-w32-d-000000000000000\build\src\obj-firefox\dist\include\mozilla/gfx/Logging.h:515

03:00:48     INFO - #01: mozilla::gfx::Log<1,mozilla::gfx::CriticalLogger>::Flush() [obj-firefox/dist/include/mozilla/gfx/Logging.h:282]

Comment 2

[Vincent Liu[:vliu]]

In local win7-32bit, I can't reproduce this issue. The requested Surface Size(800,1000) didn't exceed the maximum value that Skia was limited in [1]. 

[1]:https://dxr.mozilla.org/mozilla-central/source/gfx/2d/DrawTargetSkia.h#153

I will try to reproduce it on try with some log message to see if it still reproducible.

Comment 3

[Vincent Liu[:vliu]]

(In reply to Vincent Liu[:vliu] from comment #2)
> In local win7-32bit, I can't reproduce this issue. The requested Surface
> Size(800,1000) didn't exceed the maximum value that Skia was limited in [1]. 
> 
> [1]:https://dxr.mozilla.org/mozilla-central/source/gfx/2d/DrawTargetSkia.
> h#153
> 
> I will try to reproduce it on try with some log message to see if it still
> reproducible.

It seems that this issue can reproduce on try easily. Please see [1]

[1]: https://treeherder.mozilla.org/#/jobs?repo=try&revision=a88f476b3a818638ce5c6c7012590990ccc1f4e7&selectedJob=69871142

Comment 5

[Vincent Liu[:vliu]]

From looked into it more details. I have some founds in current stage.

1. Printing message showed that the reason why crash happens was Skia can't alloc fail in [1]. The size is 3200000. 

  [1]: https://dxr.mozilla.org/mozilla-central/source/gfx/skia/skia/src/core/SkMallocPixelRef.cpp#82

2. The reason why Skia alloc fails seems system has run out of memory because many tests didn't release their resource by description flow in time.
   [2] should be called to release the resource.  

  [2]: https://dxr.mozilla.org/mozilla-central/source/gfx/skia/skia/src/core/SkMallocPixelRef.cpp#14

3. From the try link in [3], Win7 debug / Win7-VM debug can reproduce this issue. Other platforms like Mac or Linux doesn't have. 

   [3]: https://treeherder.mozilla.org/#/jobs?repo=try&revision=a44c0f7bd59057bde8346594a58e02e22ed205fd

Comment 6

[Vincent Liu[:vliu]]

(In reply to Vincent Liu[:vliu] from comment #5)
> From looked into it more details. I have some founds in current stage.
> 
> 1. Printing message showed that the reason why crash happens was Skia can't
> alloc fail in [1]. The size is 3200000. 
> 
>   [1]:
> https://dxr.mozilla.org/mozilla-central/source/gfx/skia/skia/src/core/
> SkMallocPixelRef.cpp#82
> 
> 2. The reason why Skia alloc fails seems system has run out of memory
> because many tests didn't release their resource by description flow in time.
>    [2] should be called to release the resource.  
> 
>   [2]:
> https://dxr.mozilla.org/mozilla-central/source/gfx/skia/skia/src/core/
> SkMallocPixelRef.cpp#14
> 
> 3. From the try link in [3], Win7 debug / Win7-VM debug can reproduce this
> issue. Other platforms like Mac or Linux doesn't have. 
> 
>    [3]:
> https://treeherder.mozilla.org/#/
> jobs?repo=try&revision=a44c0f7bd59057bde8346594a58e02e22ed205fd

After discuss internally, it seems that Win 7 has limited memory size and OOM happens. One possible way to fix this is increasing the number of thunk. ihsiao, is any way I can send a try with configurable parameter to increase the number of thunk to see if we can fix this? Thanks

Comment 7

[Vincent Liu[:vliu]]

The current status is 

1. It is hard to reproduce the issue in recent try I sent. Some reftests have been disabled by Bug 1316482 and Bug 1311244 might be a reason. I would try to revert them and try again. 

2. I'd tried to build up Win7 virtual machine to reproduce this issue. But currently there is no any update from this.

Comment 8

[Vincent Liu[:vliu]]

(In reply to Vincent Liu[:vliu] from comment #7)
> The current status is 
> 

> 2. I'd tried to build up Win7 virtual machine to reproduce this issue. But
> currently there is no any update from this.

  Here is an update that I'd run reftest repeatedly under layout/reftest/ for the past couple days but still can't reproduce this issue.

From bug 1300355 Comment 18, it also pointed out the system memory almost exhausted when issue happens. Based on this, I would like to discuss with people in ateam if I can try the comment I left in bug 1300355 Comment 34.

Comment 9

[Vincent Liu[:vliu]]

I will set the owner to Nobody because it depends on Bug 1300355.