diff options
| author | timurrrr@chromium.org <timurrrr@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-09-21 17:03:46 +0000 |
|---|---|---|
| committer | timurrrr@chromium.org <timurrrr@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2009-09-21 17:03:46 +0000 |
| commit | 3c92e2472c00fb74f6bc01896235f1e98e6dfc19 (patch) | |
| tree | a8830342b061047669a41c299b9cd7ba1397f01a /base | |
| parent | cf8bf30122f38a800ddf5ee7806a7cf8d019ed63 (diff) | |
| download | chromium_src-3c92e2472c00fb74f6bc01896235f1e98e6dfc19.zip chromium_src-3c92e2472c00fb74f6bc01896235f1e98e6dfc19.tar.gz chromium_src-3c92e2472c00fb74f6bc01896235f1e98e6dfc19.tar.bz2 | |
Fix a harmless data race in Thread
Currently, if StopSoon/Stop is called twice (like in ThreadTest.StopSoon) message_loop_ can be accessed simultaneously by ThreadMain():162 and in StopSoon():124.
This data race is harmless since stopping_==true, so message_loop_ value is actually un-needed.
I'd like to swap the conditions to avoid reading message_loop_ in case stopping_==true.
This way ThreadTest.StopSoon will have no race reports under ThreadSanitizer. Currently, this is the only report on base_unittests on ThreadSanitizer linux bot.
Review URL: http://codereview.chromium.org/207039
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@26691 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'base')
| -rw-r--r-- | base/thread.cc | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/base/thread.cc b/base/thread.cc index 5fc8678..09f20a7 100644 --- a/base/thread.cc +++ b/base/thread.cc @@ -121,7 +121,7 @@ void Thread::StopSoon() { // We should only be called on the same thread that started us. DCHECK_NE(thread_id_, PlatformThread::CurrentId()); - if (!message_loop_ || stopping_) + if (stopping_ || !message_loop_) return; stopping_ = true; |