Make the lazy TLS pointer to the current MessageLoop into a Leaky LazyInstance.

Make the lazy TLS pointer to the current MessageLoop into a Leaky LazyInstance.

If a joinable thread is not joined before AtExit (e.g. if it is owned by a leaky
singleton), the AtExitManager will destroy its MessageLoop pointer while the
thread is still running. Make the pointer Leaky to prevent that from happening.

BUG=270110
R=darin, jar

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=219953

[earthdok, 2013-08-20 19:32:13]

please take a look

[Nico, 2013-08-21 19:31:19]

(i think alexeypa has touched this recently?)

[earthdok, 2013-08-21 19:46:25]

On 2013/08/21 19:31:19, Nico (vacation Aug 2 - Aug 22) wrote:
> (i think alexeypa has touched this recently?)

I think the pointer has been this way forever. The reason we haven't seen this
bug before is because there's a rather unlikely race condition involved.

[earthdok, 2013-08-27 13:03:54]

Adding another reviewer. Jim, please take a look.

[jar (doing other things), 2013-08-27 17:54:49]

This seems super safe.  We don't need to recycle these pointers (we really just
use 'em up...) and leaking just ensures the thread initializer and teardown code
(for a given TLS slot) continues to be called at thread creation/destruction
time (I think access to data in the slots from existing use of the pointer will
*probably* keep working in either case).   Such teardown code should be very
safe to call (I know it is for users such as TCMalloc).  There is of course a
tiny chance of a slower shutdown, since we'll call more thread-cleanup code as
we teardown... but it seems much safer to have the slots cleanly (officially?)
accessible till "the end of time."


Bottom line:  lets do it.

LGTM

[commit-bot: I haz the power, 2013-08-27 18:50:53]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-27 18:58:47]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-27 19:14:29]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-27 19:38:47]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-27 20:07:57]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-28 00:38:08]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/earthdok@chromium.org/22816018/1

[commit-bot: I haz the power, 2013-08-28 07:51:23]

Change committed as 219953