3 Commits

Author SHA1 Message Date
pbos@webrtc.org
b9906743da Split suppressons of thread.cc and messagequeue.cc.
Most calls have either of these in the stack, meaning that pretty much
all races are suppressed.

BUG=
R=kjellander@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/17349004

git-svn-id: http://webrtc.googlecode.com/svn/trunk@7154 4adac7df-926f-26a2-2b94-8c16560cd09d
2014-09-11 14:59:06 +00:00
pbos@webrtc.org
c9b3f77e65 Fix data races in VideoAdapterTest.
Adressing clear races between the test thread and capturer thread shown
as heap-use-after-free in vpx_codec_destroy in
WebRtcVideoMediaChannelTest.SetSend (way later in the rest run).

When capturing a frame the test copied it to a separate frame that would
then be read by the test without synchronization, if the test didn't
manage to examine the frame in between captures the adapted frame would
be overwritten by the following frame during accesses to it.

The actual races are suppressed by race:webrtc/base/messagequeue.cc and
race:webrtc/base/thread.cc. These fixes reduce the suppression count
locally from around 3000 to 30 for VideoAdapterTest.*.

Also removing tsan suppressions for talk/base as it's been moved to
webrtc/base.

R=tommi@webrtc.org
BUG=3671

Review URL: https://webrtc-codereview.appspot.com/22169004

git-svn-id: http://webrtc.googlecode.com/svn/trunk@6979 4adac7df-926f-26a2-2b94-8c16560cd09d
2014-08-26 12:33:18 +00:00
kjellander@webrtc.org
c23923447c Roll chromium_revision 289723:291647
To pick up recent fixes after the Chromium Git switch.

Relevant changes pulled in by this roll:
* r291168 refactor sanitizer_options (we can now remove some hacks)
* r291647 roll of nss.gyp (its paths work with how we build for iOS).

BUG=2863,3731
R=iannucci@chromium.org, tommi@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/22489004

git-svn-id: http://webrtc.googlecode.com/svn/trunk@6967 4adac7df-926f-26a2-2b94-8c16560cd09d
2014-08-25 14:16:32 +00:00