GitLab

This is probably a false positive. The test currently assumes that, in
the period when it's calling SetClearSiteDataHeader, there are no
requests in flight. This appears to be true, however, it depends on
signals across processes, which I suspect TSan is not taking into
account. The race doesn't trigger with --single-process, which supports
this theory.

Still, adding a lock is easy and seems prudent when state is accessed in
this way across threads.

Bug: 965719
Change-Id: I7c5c34a06df4684845f76c0e8730c4bfea7170c4
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/1626297
Auto-Submit: David Benjamin <davidben@chromium.org>
Commit-Queue: Christian Dullweber <dullweber@chromium.org>
Reviewed-by: Christian Dullweber <dullweber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#662539}