Fix shutdown crash by ensuring valid MessageLoop

Fix shutdown crashes by ensuring a valid MessageLoop.

The WindowServer crashes during shutdown when there are open handles
because ThreadTaskRunnerHandle will crash when there isn't a valid
MessageLoop. Don't manually destroy the MessageLoop.

BUG=689197
Review-Url: https://codereview.chromium.org/2683543003
Cr-Commit-Position: refs/heads/master@{#448742}
Committed: https://chromium.googlesource.com/chromium/src/+/a6de7e04fc91c39ee5c85b4fa098ba2f8d26a9d7

[Elliot Glaysher, 2017-02-07 19:22:38]

The CQ bit was checked by erg@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-07 19:23:19]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-07 20:13:12]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2017-02-07 20:13:13]

Dry run: This issue passed the CQ dry run.

[Elliot Glaysher, 2017-02-07 20:22:04]

Description was changed from

==========
[xxx] If we destroy MessageLoop, we destroy the ThreadTaskRunnerHandle.

BUG=
==========

to

==========
Fix shutdown crashes by ensuring a valid MessageLoop.

The WindowServer crashes during shutdown when there are open handles
because ThreadTaskRunnerHandle will crash when there isn't a valid
MessageLoop. Don't manually destroy the MessageLoop.

BUG=689197
==========

[Elliot Glaysher, 2017-02-07 20:22:05]

erg@chromium.org changed reviewers:
+ rockot@chromium.org

[Elliot Glaysher, 2017-02-07 20:22:09]

The CQ bit was checked by erg@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-02-07 20:22:54]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Elliot Glaysher, 2017-02-07 20:23:11]

I don't understand why the code I'm deleting is there.

One possible alternative is to reset message loop, but then install another
message loop immediately for the final context reset.

[Ken Rockot(use gerrit already), 2017-02-07 20:43:31]

On 2017/02/07 at 20:23:11, erg wrote:
> I don't understand why the code I'm deleting is there.
> 
> One possible alternative is to reset message loop, but then install another
message loop immediately for the final context reset.

TL;DR: LGTM

Long version:

The comment is a bit dated. The rationale was that there may be tasks in the
task queue which have bound unretained references to the Service instance, which
is itself owned by the ServiceContext. This is because the service library code
aimed to guarantee that your Service instance essentially lived forever (at
least, for as long as it could possibly matter.)

In practice I do not agree that this is a useful guarantee to make: binding to
the Service itself is a rare enough occurrence that one can use weakptrs where
appropriate. Furthermore, ServiceRunner isn't the only way to run a Service
instance - it's just a helper for convenience which provides a common way of
doing things - so a Service relying on this guarantee is probably unsafe given
that there is no clear specification compelling all service runners to satisfy
the same guarantee.

Given all of this, I say we kill it and resolve any new crashes which arise (if
any) on a case-by-case basis by fixing individual services' logic.

[Elliot Glaysher, 2017-02-07 20:50:28]

The CQ bit was unchecked by erg@chromium.org

[Elliot Glaysher, 2017-02-07 20:50:32]

The CQ bit was checked by erg@chromium.org

[commit-bot: I haz the power, 2017-02-07 20:51:17]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2017-02-07 22:24:47]

CQ is committing da patch.

Bot data: {"patchset_id": 20001, "attempt_start_ts": 1486500632964860,
"parent_rev": "264a4bdbd88094e20d723c493a07c97df2946859", "commit_rev":
"a6de7e04fc91c39ee5c85b4fa098ba2f8d26a9d7"}

[commit-bot: I haz the power, 2017-02-07 22:25:25]

Description was changed from

==========
Fix shutdown crashes by ensuring a valid MessageLoop.

The WindowServer crashes during shutdown when there are open handles
because ThreadTaskRunnerHandle will crash when there isn't a valid
MessageLoop. Don't manually destroy the MessageLoop.

BUG=689197
==========

to

==========
Fix shutdown crashes by ensuring a valid MessageLoop.

The WindowServer crashes during shutdown when there are open handles
because ThreadTaskRunnerHandle will crash when there isn't a valid
MessageLoop. Don't manually destroy the MessageLoop.

BUG=689197

Review-Url: https://codereview.chromium.org/2683543003
Cr-Commit-Position: refs/heads/master@{#448742}
Committed:
https://chromium.googlesource.com/chromium/src/+/a6de7e04fc91c39ee5c85b4fa098...
==========

[commit-bot: I haz the power, 2017-02-07 22:25:26]

Committed patchset #2 (id:20001) as
https://chromium.googlesource.com/chromium/src/+/a6de7e04fc91c39ee5c85b4fa098...