Added suspend/resume detection to the GpuWatchdogThread

Added suspend/resume detection to the GpuWatchdogThread

The watchdog will be "paused" while the system is suspending/suspended to
prevent false positives that may result from erratic behavior during hibernate.

BUG=258617
Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=215961

[Ken Russell (switch to Gerrit), 2013-08-05 22:08:01]

This LGTM as far as I can tell -- the threading and state machinery looks
correct -- but please get reviews from apatrick and jbauman.

https://codereview.chromium.org/22289002/diff/1/content/gpu/gpu_watchdog_thre...
File content/gpu/gpu_watchdog_thread.cc (right):

https://codereview.chromium.org/22289002/diff/1/content/gpu/gpu_watchdog_thre...
content/gpu/gpu_watchdog_thread.cc:179: // Should not get here while system is
suspended
Please complete the sentence: "while the system is suspended.".

https://codereview.chromium.org/22289002/diff/1/content/gpu/gpu_watchdog_thre...
content/gpu/gpu_watchdog_thread.cc:245: OnAcknowledge();
Is this correct and sufficient? The situation I'm concerned about is:

 - OnSuspend is called
 - It calls OnAcknowledge
 - That posts a delayed OnCheck message to its message loop (with
was_suspended=false)
 - The machine actually suspends
 - The machine wakes up
 - OnResume is called; suspended_=false
 - The previously posted delayed task is run

Will this work correctly?

Also, a DCHECK should be added to OnAcknowledge that it's being called on the
expected thread.

https://codereview.chromium.org/22289002/diff/1/content/gpu/gpu_watchdog_thre...
content/gpu/gpu_watchdog_thread.cc:253: OnCheck(true);
Similarly for OnCheck. Per discussion with apatrick, perhaps make it a CHECK
rather than a DCHECK since the watchdog's disabled in debug builds.

[bajones, 2013-08-05 22:37:05]

Updated to address Ken's feedback. Not sure if this is the best way to do the
thread CHECKs. Borrowed it from Time::EnableHighResolutionTimer
(https://code.google.com/p/chromium/codesearch#chromium/src/base/time/time_win...)

[apatrick_chromium, 2013-08-05 23:08:34]

Do you want to disable the watchdog for a period of time after OnResume gets
called?

[bajones, 2013-08-05 23:18:46]

On 2013/08/05 23:08:34, apatrick_chromium wrote:
> Do you want to disable the watchdog for a period of time after OnResume gets
> called?

We're calling OnCheck with after_suspend == true, which was a pre-existing
parameter but seemed nicely appropriate in this case. When using that the next
terminate will be scheduled 3 times further out than normal, which should give
us some buffer room if the system needs a little more time. All indications on
OSX show that by the time OnResume is called the system is pretty much done
resuming anyway, though.

[apatrick_chromium, 2013-08-06 00:30:31]

lgtm

[jbauman, 2013-08-06 00:33:06]

lgtm

[commit-bot: I haz the power, 2013-08-06 00:52:02]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bajones@chromium.org/22289002/6001

[commit-bot: I haz the power, 2013-08-06 01:52:35]

Retried try job too often on mac_rel for step(s) content_browsertests
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=mac_rel&nu...

[commit-bot: I haz the power, 2013-08-06 17:41:32]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/bajones@chromium.org/22289002/33001

[commit-bot: I haz the power, 2013-08-06 20:23:54]

Change committed as 215961