Temporarily change ThreadPosix to CHECK (crash) if we ever spend more than 30 seconds waiting for thread shutdown. There are cases on build bots where it looks like we're hitting this problem, but reproducing locally has been a struggle.

TBR=pbos@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/41799004

Cr-Commit-Position: refs/heads/master@{#8273}
git-svn-id: http://webrtc.googlecode.com/svn/trunk@8273 4adac7df-926f-26a2-2b94-8c16560cd09d
This commit is contained in:
tommi@webrtc.org 2015-02-06 19:26:42 +00:00
parent 13a0e184ee
commit ccd7e99f0a

View File

@ -110,7 +110,10 @@ bool ThreadPosix::Start(unsigned int& thread_id) {
if (result != 0)
return false;
CHECK_EQ(kEventSignaled, params.started->Wait(WEBRTC_EVENT_INFINITE));
// TODO(tommi): Change the timeout back to WEBRTC_EVENT_INFINITE when we've
// figured out where we're seeing tests hang.
auto state = params.started->Wait(30 * 1000);
CHECK_EQ(kEventSignaled, state);
DCHECK_NE(thread_id_, 0);
thread_id = thread_id_;