gpu: Fix a bug in the generation of the destruction sync token for GMBs.

gpu: Fix a bug in the generation of the destruction sync token for GMBs.

This bug was recently introduced (https://codereview.chromium.org/1499813003).
It created a race condition that would cause GMBs to be used after they were
destroyed, since the destruction sync token wasn't being properly constructed.

BUG=514815
Committed: https://crrev.com/94fe07eb3484daee0f7b091aaee3f8511c7a34fa
Cr-Commit-Position: refs/heads/master@{#365437}

[erikchen, 2015-12-15 22:10:25]

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

[erikchen, 2015-12-15 22:10:25]

kbr: Please review.

[David Yen, 2015-12-15 22:34:00]

dyen@chromium.org changed reviewers:
+ dyen@chromium.org

[David Yen, 2015-12-15 22:34:00]

Good catch, thank you for fixing this. LGTM.

[David Yen, 2015-12-15 22:35:25]

On 2015/12/15 22:34:00, David Yen wrote:
> Good catch, thank you for fixing this. LGTM.

Is there a bug ID for the race condition it was causing? This patch is probably
also needed:

https://codereview.chromium.org/1528163002/

[erikchen, 2015-12-15 22:38:43]

On 2015/12/15 22:35:25, David Yen wrote:
> On 2015/12/15 22:34:00, David Yen wrote:
> > Good catch, thank you for fixing this. LGTM.
> 
> Is there a bug ID for the race condition it was causing? This patch is
probably
> also needed:
> 
> https://codereview.chromium.org/1528163002/

I haven't filed a bug for the race condition.

I was working on https://code.google.com/p/chromium/issues/detail?id=533617.
After switching WebGL to use overlays, pretty much all WebGL content was running
into this race condition.

[Ken Russell (switch to Gerrit), 2015-12-15 23:41:32]

I'm not familiar with the construction of these tokens, so defer to dyen's
review. It would be nice if these arguments were type-safe so that this sort of
bug where the positional arguments were wrong were a compile-time error.

[erikchen, 2015-12-15 23:49:38]

On 2015/12/15 23:41:32, Ken Russell wrote:
> I'm not familiar with the construction of these tokens, so defer to dyen's
> review. It would be nice if these arguments were type-safe so that this sort
of
> bug where the positional arguments were wrong were a compile-time error.

kbr: Can I get an OWNER stamp then? dyen@ already gave an lgtm.

[Ken Russell (switch to Gerrit), 2015-12-16 00:50:34]

Yes. LGTM

[erikchen, 2015-12-16 00:51:08]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2015-12-16 00:52:48]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1525273003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1525273003/1

[commit-bot: I haz the power, 2015-12-16 03:00:49]

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

[commit-bot: I haz the power, 2015-12-16 03:00:50]

Try jobs failed on following builders:
  android_arm64_dbg_recipe on tryserver.chromium.linux (JOB_TIMED_OUT, no build
URL)

[David Yen, 2015-12-16 03:38:20]

The CQ bit was checked by dyen@chromium.org

[commit-bot: I haz the power, 2015-12-16 03:39:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1525273003/1
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1525273003/1

[commit-bot: I haz the power, 2015-12-16 04:18:52]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-12-16 04:19:39]

Description was changed from

==========
gpu: Fix a bug in the generation of the destruction sync token for GMBs.

This bug was recently introduced (https://codereview.chromium.org/1499813003).
It created a race condition that would cause GMBs to be used after they were
destroyed, since the destruction sync token wasn't being properly constructed.

BUG=514815
==========

to

==========
gpu: Fix a bug in the generation of the destruction sync token for GMBs.

This bug was recently introduced (https://codereview.chromium.org/1499813003).
It created a race condition that would cause GMBs to be used after they were
destroyed, since the destruction sync token wasn't being properly constructed.

BUG=514815

Committed: https://crrev.com/94fe07eb3484daee0f7b091aaee3f8511c7a34fa
Cr-Commit-Position: refs/heads/master@{#365437}
==========

[commit-bot: I haz the power, 2015-12-16 04:19:40]

Patchset 1 (id:??) landed as
https://crrev.com/94fe07eb3484daee0f7b091aaee3f8511c7a34fa
Cr-Commit-Position: refs/heads/master@{#365437}