| commit | ecc4f04f7ba31e68a2fb066605657bce2ab530dd | |
| author | perkj@chromium.org <perkj@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | |
| Fri, 9 May 2014 13:04:41 +0000 (9 13:04 +0000) | ||
| committer | perkj@chromium.org <perkj@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | |
| Fri, 9 May 2014 13:04:41 +0000 (9 13:04 +0000) | ||
| tree | 34f30cad9eaeb8a98738503b8fa8190045120c8c | treesnapshot (tar.gz zip) |
| parent | aecd897efa16cad0d3e7bb21334139e5cddfab4b | commitdiff |
Revert 269222 "Make sure webrtc::VideoSourceInterface is release..."
Due to memory leak in:
http://build.chromium.org/p/chromium.memory/builders/Linux%20ASan%20LSan%20Tests%20%282%29/builds/2453/steps/content_unittests/logs/addStreamWithStoppedAudioAndVideoTrack
> Make sure webrtc::VideoSourceInterface is released on the main render thread.
> webrtc::VideoSourceInterface must be released on the main render thread since it needs a libjingle thread wrapper. Fix problem with where a const scoped_refptr<webrtc::VideoSourceInterface>& source must be released on the main render thread. The problem was introduced in https://codereview.chromium.org/252393006/ where the video frames are now delivered on the IO-thread that may result in that that WebRtcVideoTrackAdapter::WebRtcVideoSourceAdapter that holds the last reference to it is destroyed on the IO-thread.
>
> TEST= http://googlechrome.github.io/webrtc/samples/web/content/constraints/ click Ge media and click Connect. When video is flowing, refresh the page.
> R=tommi@chromium.org
>
> Review URL: https://codereview.chromium.org/279533003
TBR=perkj@chromium.org
Review URL: https://codereview.chromium.org/277933002
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@269252 0039d316-1c4b-4281-b951-d872f2087c98
Due to memory leak in:
http://build.chromium.org/p/chromium.memory/builders/Linux%20ASan%20LSan%20Tests%20%282%29/builds/2453/steps/content_unittests/logs/addStreamWithStoppedAudioAndVideoTrack
> Make sure webrtc::VideoSourceInterface is released on the main render thread.
> webrtc::VideoSourceInterface must be released on the main render thread since it needs a libjingle thread wrapper. Fix problem with where a const scoped_refptr<webrtc::VideoSourceInterface>& source must be released on the main render thread. The problem was introduced in https://codereview.chromium.org/252393006/ where the video frames are now delivered on the IO-thread that may result in that that WebRtcVideoTrackAdapter::WebRtcVideoSourceAdapter that holds the last reference to it is destroyed on the IO-thread.
>
> TEST= http://googlechrome.github.io/webrtc/samples/web/content/constraints/ click Ge media and click Connect. When video is flowing, refresh the page.
> R=tommi@chromium.org
>
> Review URL: https://codereview.chromium.org/279533003
TBR=perkj@chromium.org
Review URL: https://codereview.chromium.org/277933002
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@269252 0039d316-1c4b-4281-b951-d872f2087c98