Base: Fix another deadlock during shutdown of SequencedWorkerPool.

Base: Fix another deadlock during shutdown of SequencedWorkerPool.

This CL is very similar to https://codereview.chromium.org/956583004/. There is
yet another code path that causes SequencedWorkerPool to destroy some tasks
without releasing its global lock.

BUG=465271
Committed: https://crrev.com/19249bcc5ac81a3dfd5e38be10d56b2ea434162c
Cr-Commit-Position: refs/heads/master@{#320600}

[erikchen, 2015-03-09 22:38:04]

erikchen@chromium.org changed reviewers:
+ brettw@chromium.org

[erikchen, 2015-03-09 22:38:05]

brettw: Please review.

[brettw, 2015-03-12 23:12:42]

https://codereview.chromium.org/994483005/diff/20001/base/threading/sequenced...
File base/threading/sequenced_worker_pool.cc (right):

https://codereview.chromium.org/994483005/diff/20001/base/threading/sequenced...
base/threading/sequenced_worker_pool.cc:808: // deletion must happen outside of
the lock. Since the lock has been
I don't follow this, maybe the comment needs to be clarified.

This talks about the lock "has been released" but when this code is called, the
lock is still held.

[erikchen, 2015-03-12 23:24:52]

brettw: PTAL

https://codereview.chromium.org/994483005/diff/20001/base/threading/sequenced...
File base/threading/sequenced_worker_pool.cc (right):

https://codereview.chromium.org/994483005/diff/20001/base/threading/sequenced...
base/threading/sequenced_worker_pool.cc:808: // deletion must happen outside of
the lock. Since the lock has been
On 2015/03/12 23:12:42, brettw wrote:
> I don't follow this, maybe the comment needs to be clarified.
> 
> This talks about the lock "has been released" but when this code is called,
the
> lock is still held.

I broke the comment in two, moved the location of the second comment, and
rewrote it to be more precise. I've reproduced it here for your convenience:

 // Since the lock has been released, |status| may no longer be
 // accurate. It might read GET_WORK_WAIT even if there are tasks
 // ready to perform work. Jump to the top of the loop to recalculate
 // |status|.

[brettw, 2015-03-13 22:03:25]

very clear now, thanks. LGTM

[erikchen, 2015-03-13 22:05:41]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2015-03-13 22:05:59]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/994483005/60001

[commit-bot: I haz the power, 2015-03-13 22:57:44]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-03-13 22:58:37]

Patchset 4 (id:??) landed as
https://crrev.com/19249bcc5ac81a3dfd5e38be10d56b2ea434162c
Cr-Commit-Position: refs/heads/master@{#320600}