Could https://codereview.chromium.org/2712003002 be related?

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/76e94d0aa96fe99356b5009716739027fa4d0159

commit 76e94d0aa96fe99356b5009716739027fa4d0159
Author: liaoyuke <liaoyuke@chromium.org>
Date: Thu Mar 02 22:36:24 2017

Disabling NumTasksBeforeDetach in base_unittests due to flakiness. This
test failed on iOS-simulator twice during the past 30 runs.

https://build.chromium.org/p/chromium.mac/builders/ios-simulator/builds/15082
https://build.chromium.org/p/chromium.mac/builders/ios-simulator/builds/15100

BUG= 698046 
TBR=fdoray@chromium.org

Review-Url: https://codereview.chromium.org/2728833003
Cr-Commit-Position: refs/heads/master@{#454410}

[modify] https://crrev.com/76e94d0aa96fe99356b5009716739027fa4d0159/base/task_scheduler/scheduler_worker_pool_impl_unittest.cc

Given that this has been disabled, I'm going to go ahead and remove the RB-Beta Flag.

Rob, RBB was here to signify that the bug needs to be reenabled before we release a beta. At least it is our process in iOS.
If you don't want to block the beta release, we should then reenable the test although it is flaky. The thing we should avoid is a disabled test that goes unnoticed.

+cmasso regarding RBB process and tracking this.

I don't know if we can impose our iOS process on this bug, but we should at least make sure we do anything else that needs to happen on disabled tests.

This bug will need to be re-enabled before M59 branch point since we are branching M58 today. As described by lpromero@, this is how things work on iOS. Thanks!

This code resides in base and not iOS specific code. Does the iOS process still make sense for this? For what it's worth, the fix is currently in flight.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/404de72480129c6549041267a171a9b0f1f8189f

commit 404de72480129c6549041267a171a9b0f1f8189f
Author: robliao <robliao@chromium.org>
Date: Tue Mar 14 00:30:45 2017

Fix TaskSchedulerWorkerPoolHistogramTest.NumTasksBeforeDetach

This fix passes the pointer for the PlatformThreadRef to the task for
comparisons and actually gets the PlatformThreadRef from
PlatformThread::CurrentRef().

Previously, the test would race and pass by value. This meant that
if the previous task did not complete in time, it would pass the empty
reference. Since we were accidentally comparing against that, then the
test succeeded. When code ran slowly enough to set the value before
the next PostTask, then we would have the correct PlatformThreadRef,
but incorrectly compare it to the empty PlatformThreadRef, correctly
failing the test.

It's amazing that the tests passed in the first place!

BUG= 697697 

Review-Url: https://codereview.chromium.org/2729573004
Cr-Commit-Position: refs/heads/master@{#456558}

[modify] https://crrev.com/404de72480129c6549041267a171a9b0f1f8189f/base/task_scheduler/scheduler_worker_pool_impl_unittest.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/fbad1aa8a39929e5b58cb07dcdd09d62e88dc679

commit fbad1aa8a39929e5b58cb07dcdd09d62e88dc679
Author: robliao <robliao@chromium.org>
Date: Wed Mar 15 20:05:29 2017

Remove Obsolete Comment from https://codereview.chromium.org/2729573004/

BUG= 698046 

TBR=gab@chromium.org
NOTRY=true
Comment only change.

Review-Url: https://codereview.chromium.org/2750103004
Cr-Commit-Position: refs/heads/master@{#457180}

[modify] https://crrev.com/fbad1aa8a39929e5b58cb07dcdd09d62e88dc679/base/task_scheduler/scheduler_worker_pool_impl_unittest.cc