New issue
Advanced search Search tips
Note: Color blocks (like █ or █) mean that a user may not be available. Tooltip shows the reason.

Issue 791289 link

Starred by 2 users
Status: Available
Owner: ----
Cc:
█ kbr@chromium.org
geoffl...@chromium.org
zmo@chromium.org
kkinnu...@nvidia.com
jmad...@chromium.org
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

webgl2_conformance_gl_tests failure on Win7 NVIDIA: conformance2_rendering_instanced_rendering_bug

Project Member Reported by jmad...@chromium.org, Dec 2 2017 
Often WebGL rolls causes test failures in WebGL configs that ANGLE uses on the CQ and GPU.FYI bots. The latest is suspicious:

https://ci.chromium.org/buildbot/chromium.gpu.fyi/Win7%20Release%20%28NVIDIA%29/32329

In this case, I'm suspecting 

Roll WebGL e4919fa..05591bb

https://chromium.googlesource.com/external/khronosgroup/webgl.git/+log/e4919fa..05591bb

The change log doesn't seem immediately relevant but I don't know of any other changes in the regression log that seem as suspicious.

Ken, Mo, could we please run the ANGLE main trybot configurations against WebGL rolls to prevent ANGLE and GPU FYI breakage, or at least help alleviate suspicion on WebGL rolls? This is a very common source of breakage and I think it would help a lot.

Comment 1 by jmad...@chromium.org, Dec 2 2017 

Error text:

[3260:4464:1202/011757.964:FATAL:gpu_memory_manager.cc(40)] Check failed: !bytes_allocated_current_. 
Backtrace:
	base::debug::StackTrace::StackTrace [0x66E5D620+32]
	base::debug::StackTrace::StackTrace [0x66E547BD+13]
	logging::LogMessage::~LogMessage [0x66DD9CEE+78]
	gpu::GpuMemoryManager::~GpuMemoryManager [0x68ADC400+132]
	gpu::GpuChannelManager::~GpuChannelManager [0x68ADBA7C+228]
	viz::GpuServiceImpl::~GpuServiceImpl [0x67BEA0ED+509]
	viz::GpuServiceImpl::`scalar deleting destructor' [0x67BE587B+11]
	viz::VizMainImpl::~VizMainImpl [0x6630B5C3+307]
	content::GpuChildThread::~GpuChildThread [0x6844F45A+236]
	content::GpuChildThread::`scalar deleting destructor' [0x6844FCA1+11]
	content::ChildProcess::~ChildProcess [0x67DAA466+172]
	content::GpuMain [0x6844E89B+911]
	content::RunNamedProcessTypeMain [0x66DCDA32+266]
	content::ContentMainRunnerImpl::Run [0x66DCDF9A+278]
	service_manager::Main [0x66DD3A03+675]
	content::ContentMain [0x66DCD901+49]
Project Member

Comment 2 by bugdroid1@chromium.org, Dec 2 2017 

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/469ec8f1d3f17eb603c2a0be1fe5ba6947744a54

commit 469ec8f1d3f17eb603c2a0be1fe5ba6947744a54
Author: Jamie Madill <jmadill@chromium.org>
Date: Sat Dec 02 19:40:31 2017

WebGL: Suppress instanced test on NVIDIA/Win/GL.

Bug: 791289
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: I6f7f3da80dcbb6190e8a05b02e3eed535ac9c6fe
Tbr: zmo@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/804701
Reviewed-by: Jamie Madill <jmadill@chromium.org>
Commit-Queue: Jamie Madill <jmadill@chromium.org>
Cr-Commit-Position: refs/heads/master@{#521220}
[modify] https://crrev.com/469ec8f1d3f17eb603c2a0be1fe5ba6947744a54/content/test/gpu/gpu_tests/webgl2_conformance_expectations.py

Comment 3 by kbr@chromium.org, Dec 5 2017

Cc: oetu...@nvidia.com
Running ANGLE's tryservers against WebGL conformance rolls as of https://chromium-review.googlesource.com/808089 ; thanks for the reminder.

Is this an NVIDIA driver bug or a bug in Chrome?

Comment 4 by jmad...@chromium.org, Dec 5 2017 

NVIDIA is the only config that runs WebGL 2 tests with the GL backend on Windows - there's no reason to believe right now it's an NVIDIA bug. Note that the test expectations file is a bit misleading because it specifies NVIDIA for these tests, since we don't know which are NVIDIA bugs or our bugs, if any.

Comment 5 by oetu...@nvidia.com, Oct 9

Cc: -oetu...@nvidia.com kkinnu...@nvidia.com

Comment 6 by kkinnu...@nvidia.com, Jan 2

Labels: GPU-NVidia
►

Sign in to add a comment