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

Issue 698441 link

Starred by 1 user

Issue metadata

Status: Fixed
Owner:
Closed: Mar 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Windows
Pri: 2
Type: Bug



Sign in to add a comment

Flaky content_browsertests failures on WebRTC Win 8 Tester

Project Member Reported by emir...@chromium.org, Mar 4 2017

Issue description

I am not sure what CL triggered this flakiness because I cannot find anything suspicious within the range. I will look a bit more.

First failure: https://build.chromium.org/p/chromium.webrtc/builders/Win8%20Tester/builds/32806

Output:
[2360:2176:0301/211408.344:1231953:ERROR:media_foundation_video_encode_accelerator_win.cc(477)] Couldn't set output media type, HRESULT: 0xc00d36b4
[2360:2176:0301/211408.344:1231953:ERROR:mf_helpers.cc(12)] Error in dxva_video_decode_accelerator_win.cc on line 477
[2360:2176:0301/211408.353:1231968:ERROR:media_foundation_video_encode_accelerator_win.cc(477)] Couldn't set output media type, HRESULT: 0xc00d36b4
[2360:2176:0301/211408.353:1231968:ERROR:mf_helpers.cc(12)] Error in dxva_video_decode_accelerator_win.cc on line 477
[1220:904:0301/211408.558:1232171:INFO:CONSOLE(90)] "Recorder object created, mimeType = video/x-matroska;codecs=AVC1", source: http://127.0.0.1:50135/media/mediarecorder_test.html (90)
[1220:904:0301/211408.558:1232171:INFO:CONSOLE(22)] "Waiting for", source: http://127.0.0.1:50135/media/mediarecorder_test_utils.js (22)
Backtrace:
	media::GpuVideoEncodeAcceleratorHost::OnRequireBitstreamBuffers [0x015A7293+339]
	IPC::DispatchToMethod<media::GpuVideoEncodeAcceleratorHost,void (__thiscall media::GpuVideoEncodeAcceleratorHost::*)(unsigned int,gfx::Size const &,unsigned int),void,std::tuple<unsigned int,gfx::Size,unsigned int> > [0x015A60C0+24]
	IPC::MessageT<AcceleratedVideoEncoderHostMsg_RequireBitstreamBuffers_Meta,std::tuple<unsigned int,gfx::Size,unsigned int>,void>::Dispatch<media::GpuVideoEncodeAcceleratorHost,media::GpuVideoEncodeAcceleratorHost,void,void (__thiscall media::GpuVideoEncode [0x015A5FE6+141]
	media::GpuVideoEncodeAcceleratorHost::OnMessageReceived [0x015A6E86+470]
	base::internal::InvokeHelper<1,void>::MakeItSo<void (__thiscall base::CancelableCallback<void __cdecl(IPC::Message const &)>::*const &)(IPC::Message const &)const ,base::WeakPtr<base::CancelableCallback<void __cdecl(IPC::Message const &)> > const &,IPC::M [0x02CB8098+36]
	base::internal::Invoker<base::internal::BindState<base::internal::IgnoreResultHelper<bool (__thiscall IPC::Listener::*)(IPC::Message const &)>,base::WeakPtr<IPC::Listener>,IPC::Message>,void __cdecl(void)>::Run [0x014FDF47+23]
	base::debug::TaskAnnotator::RunTask [0x0241BD97+407]
	base::MessageLoop::RunTask [0x023B1DF1+1233]
	base::MessageLoop::DoWork [0x023B0E25+741]
	base::MessagePumpDefault::Run [0x0241CBFB+219]
	base::MessageLoop::RunHandler [0x023B1917+247]
	base::RunLoop::Run [0x023A8AB4+132]
	base::Thread::Run [0x023E8FED+173]
	base::Thread::ThreadMain [0x023E9AEE+622]
	base::PlatformThread::Sleep [0x023AEFD2+290]
	BaseThreadInitThunk [0x74CC7C04+36]
	RtlInitializeExceptionChain [0x7718AB8F+143]
	RtlInitializeExceptionChain [0x7718AB5A+90]
[  FAILED  ] WebRtcMediaRecorderTest.StartAndDataAvailable/6, where GetParam() = 28-byte object <00-69 64-65 38-7F F1-00 62-6D 3B-63 6F-64 65-63 73-3D 56-50 1C-00 00-00 1F-00 00-00> (930 ms)
[133/133] WebRtcMediaRecorderTest.StartAndDataAvailable/6 (1060 ms)
1 test failed:
    WebRtcMediaRecorderTest.StartAndDataAvailable/6 (e:\b\c\b\win_builder\src\content\browser\webrtc\webrtc_media_recorder_browsertest.cc:76)
 
Temp fix for flaky test here: https://codereview.chromium.org/2727633008/.
Project Member

Comment 2 by bugdroid1@chromium.org, Mar 4 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/26c6a0e07e73b1e9d357edcea55566873a99bb20

commit 26c6a0e07e73b1e9d357edcea55566873a99bb20
Author: emircan <emircan@chromium.org>
Date: Sat Mar 04 04:16:53 2017

Disable VEA usage in MediaRecorder until flaky test is addressed

BUG= 698441 

Review-Url: https://codereview.chromium.org/2727633008
Cr-Commit-Position: refs/heads/master@{#454759}

[modify] https://crrev.com/26c6a0e07e73b1e9d357edcea55566873a99bb20/content/renderer/media_recorder/video_track_recorder.cc

I believe this has the same underlying issue as  issue 701030 . I will revert to see if bots are green now.
Project Member

Comment 4 by bugdroid1@chromium.org, Mar 17 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8086b9ac7fe4598925a84e6169760187022c3eb3

commit 8086b9ac7fe4598925a84e6169760187022c3eb3
Author: emircan <emircan@chromium.org>
Date: Fri Mar 17 00:22:25 2017

Revert of Disable VEA usage in MediaRecorder until flaky test is addressed (patchset #2 id:20001 of https://codereview.chromium.org/2727633008/ )

Reason for revert:
I believe this has the same underlying issue as https://bugs.chromium.org/p/chromium/issues/detail?id=701030, which is fixed. I will re-enable the tests to see the effects.

Original issue's description:
> Disable VEA usage in MediaRecorder until flaky test is addressed
>
> BUG= 698441 
>
> Review-Url: https://codereview.chromium.org/2727633008
> Cr-Commit-Position: refs/heads/master@{#454759}
> Committed: https://chromium.googlesource.com/chromium/src/+/26c6a0e07e73b1e9d357edcea55566873a99bb20

TBR=mcasas@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 698441 

Review-Url: https://codereview.chromium.org/2754473008
Cr-Commit-Position: refs/heads/master@{#457622}

[modify] https://crrev.com/8086b9ac7fe4598925a84e6169760187022c3eb3/content/renderer/media_recorder/video_track_recorder.cc

Project Member

Comment 5 by bugdroid1@chromium.org, Mar 20 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/b0782aed1dbfcbc1258d33961f961cfb81b20c45

commit b0782aed1dbfcbc1258d33961f961cfb81b20c45
Author: grunell <grunell@chromium.org>
Date: Mon Mar 20 10:33:02 2017

Reland of Disable VEA usage in MediaRecorder until flaky test is addressed (patchset #1 id:1 of https://codereview.chromium.org/2754473008/ )

Reason for revert:
Still flaky after the revert. Re-disabling.

Original issue's description:
> Revert of Disable VEA usage in MediaRecorder until flaky test is addressed (patchset #2 id:20001 of https://codereview.chromium.org/2727633008/ )
>
> Reason for revert:
> I believe this has the same underlying issue as https://bugs.chromium.org/p/chromium/issues/detail?id=701030, which is fixed. I will re-enable the tests to see the effects.
>
> Original issue's description:
> > Disable VEA usage in MediaRecorder until flaky test is addressed
> >
> > BUG= 698441 
> >
> > Review-Url: https://codereview.chromium.org/2727633008
> > Cr-Commit-Position: refs/heads/master@{#454759}
> > Committed: https://chromium.googlesource.com/chromium/src/+/26c6a0e07e73b1e9d357edcea55566873a99bb20
>
> TBR=mcasas@chromium.org
> # Not skipping CQ checks because original CL landed more than 1 days ago.
> BUG= 698441 
>
> Review-Url: https://codereview.chromium.org/2754473008
> Cr-Commit-Position: refs/heads/master@{#457622}
> Committed: https://chromium.googlesource.com/chromium/src/+/8086b9ac7fe4598925a84e6169760187022c3eb3

TBR=mcasas@chromium.org,emircan@chromium.org
# Not skipping CQ checks because original CL landed more than 1 days ago.
BUG= 698441 

Review-Url: https://codereview.chromium.org/2758243002
Cr-Commit-Position: refs/heads/master@{#458035}

[modify] https://crrev.com/b0782aed1dbfcbc1258d33961f961cfb81b20c45/content/renderer/media_recorder/video_track_recorder.cc

Cc: mcasas@chromium.org
Status: Started (was: Assigned)
I got a repro after many trials. It looks like we are calling InitializeEncoder() too many times.

[ RUN      ] WebRtcMediaRecorderTest.StartAndDataAvailable/0
[17468:25016:0322/165127.926:82330281:INFO:CONSOLE(90)] "Recorder object created, mimeType = video/x-matroska;codecs=AVC1", source: http://127.0.0.1:59131/media/mediarecorder_test.html (90)
[17468:25016:0322/165127.928:82330281:INFO:CONSOLE(22)] "Waiting for", source: http://127.0.0.1:59131/media/mediarecorder_test_utils.js (22)
[25656:26484:0322/165128.415:82330765:ERROR:video_track_recorder.cc(1308)] InitializeEncoder640x480
[25656:26484:0322/165128.416:82330781:ERROR:video_track_recorder.cc(155)] Accelerated codec found: h264 baseline
[25656:22184:0322/165128.416:82330781:ERROR:video_track_recorder.cc(797)] ConfigureEncoderOnEncodingTaskRunner
[25656:26484:0322/165128.419:82330781:ERROR:video_track_recorder.cc(1308)] InitializeEncoder640x480
[25656:26484:0322/165128.421:82330781:ERROR:video_track_recorder.cc(1308)] InitializeEncoder640x480
[25656:22184:0322/165128.422:82330781:ERROR:video_track_recorder.cc(797)] ConfigureEncoderOnEncodingTaskRunner
[17468:25016:0322/165130.930:82333281:INFO:CONSOLE(27)] "Still waiting for satisfaction of function () {
              return videoSize > 0;
            }", source: http://127.0.0.1:59131/media/mediarecorder_test_utils.js (27)
[17468:25016:0322/165133.931:82336281:INFO:CONSOLE(27)] "Still waiting for satisfaction of function () {
              return videoSize > 0;
            }", source: http://127.0.0.1:59131/media/mediarecorder_test_utils.js (27)
[17468:9608:0322/165136.020:82338375:ERROR:video_capture_device_client.cc(412)] error@ OnIncomingCapturedData@c:\workspace\chromium\src\media\capture\video\video_capture_device_client.cc:191, Too many frames dropped, OS message: The operation completed successfully. (0x0)
Project Member

Comment 7 by bugdroid1@chromium.org, Mar 24 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/1ea19d13673e5fbe40925ba23a49f1bfa6877287

commit 1ea19d13673e5fbe40925ba23a49f1bfa6877287
Author: emircan <emircan@chromium.org>
Date: Fri Mar 24 23:35:03 2017

Fix multiple Initialize() calls and reenable VEA usage in MediaRecorder for Win8

This Cl does two things:
- Reverts the CL that disables VEA usage until flaky test is addressed
https://codereview.chromium.org/2727633008
- Fixes the underlying issue of flakiness by not allowing |encoder_| to be
initialized multiple times.

BUG= 698441 
TEST=Local run of tests consistently pass.

Review-Url: https://codereview.chromium.org/2775453003
Cr-Commit-Position: refs/heads/master@{#459598}

[modify] https://crrev.com/1ea19d13673e5fbe40925ba23a49f1bfa6877287/content/renderer/media_recorder/video_track_recorder.cc

Components: -Infra>Client>WebRTC
Status: Fixed (was: Started)

Sign in to add a comment