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

Issue 712798 link

Starred by 2 users

Issue metadata

Status: Duplicate
Merged: issue 725664
Owner: ----
Closed: May 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 1
Type: Bug-Regression



Sign in to add a comment

Out of memory in V8 during WebglConformance_deqp_functional_gles3_texturefiltering_3d_* tests on Linux Release (Intel HD 530) GPU.FYI

Project Member Reported by ynovikov@chromium.org, Apr 18 2017

Issue description

Started in https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20%28Intel%20HD%20530%29/builds/472

Stack:
[1:341:0415/121354.190778:FATAL:memory_linux.cc(35)] Out of memory.
#0 0x5654acf7bc37 base::debug::StackTrace::StackTrace()
#1 0x5654acf9573b logging::LogMessage::~LogMessage()
#2 0x5654acfbb8ce base::(anonymous namespace)::OnNoMemory()
#3 0x5654ad01def9 GlibcMallocHook
#4 0x5654ac84c5cf v8::internal::AccountingAllocator::GetSegment()
#5 0x5654ac84cb24 v8::internal::Zone::NewExpand()
#6 0x5654ac84ca00 v8::internal::Zone::New()
#7 0x5654ac1f7842 v8::internal::compiler::BranchElimination::ReduceIf()
#8 0x5654ac23f515 v8::internal::compiler::GraphReducer::ReduceTop()
#9 0x5654ac23ef08 v8::internal::compiler::GraphReducer::ReduceNode()
#10 0x5654ac2e52af v8::internal::compiler::LateOptimizationPhase::Run()
#11 0x5654ac2dd3d8 v8::internal::compiler::PipelineImpl::OptimizeGraph()
#12 0x5654ac2dcea4 v8::internal::compiler::PipelineCompilationJob::ExecuteJobImpl()
#13 0x5654ac1d330e v8::internal::CompilationJob::ExecuteJob()
#14 0x5654ac1d1e08 v8::internal::OptimizingCompileDispatcher::CompileNext()
#15 0x5654ac1d2e3a v8::internal::OptimizingCompileDispatcher::CompileTask::Run()
#16 0x5654ab2ba941 _ZNO4base8CallbackIFvvELNS_8internal8CopyModeE1ELNS2_10RepeatModeE1EE3RunEv
#17 0x5654acff2c17 base::(anonymous namespace)::WorkerThread::ThreadMain()
#18 0x5654acfe835c base::(anonymous namespace)::ThreadFunc()
#19 0x7fde128db6ca start_thread
#20 0x7fde0c02d0af clone

Suspect:
https://codereview.chromium.org/2818233004

machenbach@, could you please triage?
 
Labels: Performance-Memory
Status: Untriaged (was: Unconfirmed)
Uh, please put it to memory sheriff triage:
https://github.com/v8/v8/wiki/Triaging-issues
Owner: ----

Comment 3 by habl...@google.com, Apr 19 2017

Cc: hpayer@chromium.org jgruber@chromium.org
Status: Available (was: Untriaged)
Let's revert V8 to 6.0.2 and roll CLs one by one?
Cc: senorblanco@chromium.org
CC'ing current pixel wrangler who saw this as well.
As far as I can see:
* local reproduction won't work without that specific graphics card (Linux NVidia builds do not have the OOM issue)
* and there's no obvious suspect in the V8 CL range from the initial report.

Is there a way to bisect on the bot itself?
Cc: kbr@chromium.org
Yes, you can do it following these instructions:
https://sites.google.com/a/google.com/client3d/documents/chrome-internal-gpu-pixel-wrangling-instructions
Cc: hablich@chromium.org
 Issue 715881  has been merged into this issue.
Cc: neis@chromium.org
+neis, the oom is happening during optimization off the main thread. Any ideas what could be causing an increase in memory consumption in the CL range? 
This is consistantly happening again in recently builds after being green for a few days:
https://build.chromium.org/p/chromium.gpu.fyi/builders/Linux%20Release%20%28Intel%20HD%20530%29/builds/617

Looks at the history seems like the old issue never fully went away.
Mergedinto: 725664
Status: Duplicate (was: Available)

Sign in to add a comment