New issue
Advanced search Search tips

Issue 818041 link

Starred by 1 user

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 3
Type: Bug



Sign in to add a comment

Flaky test failure WebglConformance_conformance_ogles_GL_dot_dot_001_to_006 on android_n5x_swarming_rel

Project Member Reported by jdarpinian@chromium.org, Mar 2 2018

Issue description

Just got a flaky test failure on my tryjob:

https://ci.chromium.org/buildbot/tryserver.chromium.android/android_n5x_swarming_rel/371908

Looks like a renderer crash but I don't know how to get the stack trace.
 

Comment 1 by kbr@chromium.org, Mar 6 2018

Cc: ynovikov@chromium.org
Components: Blink>WebGL
The failure I see in that tryjob is of:
WebglConformance_conformance_textures_image_bitmap_from_video_tex_2d_luminance_alpha_luminance_alpha_unsigned_byte

It looks to me like maybe the GPU process crashed. In the test, all of the programs failed to compile and link, and all of the ReadPixels results came back (0, 0, 0). That is consistent with the WebGL context being lost.

Not sure how to make progress on this given that there was no symbolized minidump. Yuly, have you seen flakiness like this before?

Oops, must have pasted the wrong link. The failure I meant to report is here: 

https://ci.chromium.org/buildbot/tryserver.chromium.android/android_n5x_swarming_rel/372106

I suppose both are worth investigating. Is there really no way to get a stack trace out of these failures? Can we fix that?
I do see stack trace in that log:
  	signal 11 (SIGSEGV), code 1, fault addr 0x10e5b81218 in tid 10139 (TaskSchedulerFo)
  	pid: 10082, tid: 10139, name: TaskSchedulerFo  >>> org.chromium.chrome:sandboxed_process0 <<<
  	signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x10e5b81218
  	Stack Trace:
  	  RELADDR   FUNCTION                                        FILE:LINE
  	  0000000001d05a8c  SkTextToPathIter::next(SkPath const**, float*)  ??:0:0
  	  0000000001d06fbc  SkPaint::unflatten(SkReadBuffer&)               ??:0:0

Cool, I didn't know that it was implemented.
I think I did see both of these flakes before, they were just not very frequent.
Given that 3 conformance/ogles/GL tests and 2 conformance/textures/image_bitmap_from_video tests are already marked Flaky, looks like we need to also mark these two.
I'll take care of that.
Project Member

Comment 5 by bugdroid1@chromium.org, Mar 7 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/72b01b4324d5f4894960d76d5ed82584d1e36986

commit 72b01b4324d5f4894960d76d5ed82584d1e36986
Author: Yuly Novikov <ynovikov@chromium.org>
Date: Wed Mar 07 04:40:37 2018

Suppress couple WebGL CTS flakes on Nexus 5X

conformance/ogles/GL/dot/dot_001_to_006.html
conformance/textures/image_bitmap_from_video/tex-2d-luminance_alpha-luminance_alpha-unsigned_byte.html

BUG=818041
TBR=kbr@chromium.org

Cq-Include-Trybots: master.tryserver.chromium.android:android_optional_gpu_tests_rel;master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
Change-Id: I9c8095be1d9f35cd375a1f2bb021ef4e1e3eb70e
Reviewed-on: https://chromium-review.googlesource.com/952300
Commit-Queue: Yuly Novikov <ynovikov@chromium.org>
Reviewed-by: Yuly Novikov <ynovikov@chromium.org>
Cr-Commit-Position: refs/heads/master@{#541332}
[modify] https://crrev.com/72b01b4324d5f4894960d76d5ed82584d1e36986/content/test/gpu/gpu_tests/webgl_conformance_expectations.py

Sign in to add a comment