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

Issue 656840 link

Starred by 2 users

Issue metadata

Status: WontFix
Owner:
Closed: Nov 2016
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: All
Pri: 3
Type: Bug



Sign in to add a comment

DCHECK in AudioOutputResampler when running MediaSourceTest using mojo renderer in the GPU process

Project Member Reported by xhw...@chromium.org, Oct 18 2016

Issue description

Version: M56
OS: All

What steps will reproduce the problem?

(1) Build content_browsertests with

enable_nacl = false
proprietary_codecs = true
ffmpeg_branding = "Chrome"
is_debug = true
use_goma = true
is_component_build = true
target_cpu = "x64"
enable_mojo_media = true
mojo_media_services = ["cdm", "renderer"]

(2) Run MediaSourceTest in content_browsertests as follows:

content_browsertests --gtest_filter=MediaSourceTest.Playback_AudioOnly_Opus_WebM --vmodule=mojo*=3

What is the expected output?

Test passes.

What do you see instead?

DCHECK during test teardown.

[39821:39853:1017/174108:256625491564:VERBOSE1:mojo_demuxer_stream_adapter.cc(33)] ~MojoDemuxerStreamAdapter
[39821:39853:1017/174108:256625491698:VERBOSE1:mojo_decoder_buffer_converter.cc(67)] ~MojoDecoderBufferReader
[39821:39853:1017/174108:256625493316:FATAL:audio_output_resampler.cc(399)] Check failed: HasOneRef(). Only the AudioManager should hold a reference
#0 0x7f4e985f8ade base::debug::StackTrace::StackTrace()
#1 0x7f4e9866662f logging::LogMessage::~LogMessage()
#2 0x7f4e9ec44cfe media::AudioOutputResampler::Shutdown()
#3 0x7f4e9ec1ae78 media::AudioManagerBase::Shutdown()
#4 0x7f4e9ec72b3d media::AudioManagerAlsa::~AudioManagerAlsa()
#5 0x7f4e9ec72b89 media::AudioManagerAlsa::~AudioManagerAlsa()
#6 0x7f4e9ec1928e base::DeleteHelper<>::DoDelete()
#7 0x7f4e987370da _ZN4base8internal13FunctorTraitsIPFvPKvEvE6InvokeIJRKS3_EEEvS5_DpOT_
#8 0x7f4e98737058 _ZN4base8internal12InvokeHelperILb0EvE8MakeItSoIRKPFvPKvEJRKS5_EEEvOT_DpOT0_
#9 0x7f4e98737012 _ZN4base8internal7InvokerINS0_9BindStateIPFvPKvEJS4_EEEFvvEE7RunImplIRKS6_RKSt5tupleIJS4_EEJLm0EEEEvOT_OT0_NS_13IndexSequenceIJXspT1_EEEE
#10 0x7f4e98736f5c _ZN4base8internal7InvokerINS0_9BindStateIPFvPKvEJS4_EEEFvvEE3RunEPNS0_13BindStateBaseE
#11 0x7f4e985fe9c1 _ZNO4base8internal8RunMixinINS_8CallbackIFvvELNS0_8CopyModeE0ELNS0_10RepeatModeE0EEEE3RunEv
#12 0x7f4e985fe3c9 base::debug::TaskAnnotator::RunTask()

Please use labels and text to provide additional information.

Note that the test is running in the GPU process.
https://cs.chromium.org/chromium/src/media/media_options.gni?rcl=0&l=118

This is for test only and only happens with a particular test configuration. So mark this as a P3 issue.
 

Comment 1 by xhw...@chromium.org, Oct 18 2016

Cc: dalecur...@chromium.org
alokp: I remember you did some work around AudioManager. Do you have any idea why this DCHECK fires before I dive in? Note that I am running the media app (new name service) in the GPU process.

Comment 2 by alokp@chromium.org, Oct 18 2016

Looks like AudioOutputStreamSink::stream_ is not getting closed before AudioManager is destroyed.

Comment 3 by xhw...@chromium.org, Nov 11 2016

Status: WontFix (was: Available)
In short, AudioManager shutdown isn't hooked up for the GPU process. So this path isn't officially supported.

Mark as WontFix.

Sign in to add a comment