Insert CHECK to look for abnormal flushes.

Insert CHECK to look for abnormal flushes/fences.

R=piman@chromium.org
BUG=606112
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel

[David Yen, 2016-05-11 01:09:32]

Description was changed from

==========
Insert DCHECK to look for abnormal flushes.

R=piman@chromium.org
BUG=606112
==========

to

==========
Insert DCHECK to look for abnormal flushes.

R=piman@chromium.org
BUG=606112
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[David Yen, 2016-05-11 01:10:39]

On 2016/05/11 01:09:32, David Yen wrote:

I will run this on mac_chromium_rel_ng here a couple times, if nothing
interesting happens I will try to temporarily land it to see if it catches any
interesting callstacks in the CQ wild.

[piman, 2016-05-11 01:24:11]

https://codereview.chromium.org/1972443002/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/sync_point_manager.cc (right):

https://codereview.chromium.org/1972443002/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/sync_point_manager.cc:261: DCHECK(release ==
fence_sync_release_ + 1 || release == UINT64_MAX);
nit: make it a CHECK if you want to catch release bots too. It's ok if it's
temporary.

[David Yen, 2016-05-11 17:27:27]

I changed it to a CHECK, although it didn't trigger any failures in the 5 times
I ran it. Since the release count is quite off (1 vs 12) I was thinking this
check would trigger much more frequently... really unsure how this could happen.

https://codereview.chromium.org/1972443002/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/sync_point_manager.cc (right):

https://codereview.chromium.org/1972443002/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/sync_point_manager.cc:261: DCHECK(release ==
fence_sync_release_ + 1 || release == UINT64_MAX);
On 2016/05/11 01:24:11, piman wrote:
> nit: make it a CHECK if you want to catch release bots too. It's ok if it's
> temporary.

Done.

[David Yen, 2016-05-11 17:27:47]

Description was changed from

==========
Insert DCHECK to look for abnormal flushes.

R=piman@chromium.org
BUG=606112
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
Insert CHECK to look for abnormal flushes/fences.

R=piman@chromium.org
BUG=606112
CQ_INCLUDE_TRYBOTS=tryserver.chromium.linux:linux_optional_gpu_tests_rel;tryserver.chromium.mac:mac_optional_gpu_tests_rel;tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[piman, 2016-05-11 17:40:40]

LGTM - we can land it temporarily on trunk, and see if this helps the
investigation.

[David Yen, 2016-05-11 18:28:39]

The CQ bit was checked by dyen@chromium.org

[commit-bot: I haz the power, 2016-05-11 18:29:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1972443002/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1972443002/40001

[David Yen, 2016-05-11 18:30:58]

The CQ bit was unchecked by dyen@chromium.org

[David Yen, 2016-05-11 18:34:27]

I actually got a callstack running this locally 50 times, I'll hold off on
landing it since it looks like I can reproduce something.

[David Yen, 2016-05-12 22:03:59]

closed since we solved the issue it looks no longer relevant.