gpu: Fix crash in gpu tracer

gpu: Fix crash in gpu tracer

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Committed: https://crrev.com/b063270934133a2e13bbae92803eb4708090da70
Cr-Commit-Position: refs/heads/master@{#407023}

[boliu, 2016-07-21 18:55:46]

Description was changed from

==========
gpu: Fix crash in gpu tracer

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
==========

to

==========
gpu: Fix crash in gpu tracer

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

[boliu, 2016-07-21 18:58:11]

boliu@chromium.org changed reviewers:
+ piman@chromium.org

[boliu, 2016-07-21 18:58:12]

ptal

[piman, 2016-07-21 20:18:56]

LGTM, sorry about that. How does that happen? Is it that IsTracing can change
value from another thread, and is fundamentally racy?

[boliu, 2016-07-21 20:24:29]

On 2016/07/21 20:18:56, piman wrote:
> LGTM, sorry about that. How does that happen? Is it that IsTracing can change
> value from another thread, and is fundamentally racy?

I'm not entirely sure. Those pointers are pointing memory controlled by the
tracing system, and it can change when user starts tracing. I'm not sure about
any thread safety concerns..

[boliu, 2016-07-21 21:56:07]

On 2016/07/21 20:24:29, boliu wrote:
> On 2016/07/21 20:18:56, piman wrote:
> > LGTM, sorry about that. How does that happen? Is it that IsTracing can
change
> > value from another thread, and is fundamentally racy?
> 
> I'm not entirely sure. Those pointers are pointing memory controlled by the
> tracing system, and it can change when user starts tracing. I'm not sure about
> any thread safety concerns..

Looks like it is racy. It's marked as a benign race, with a comment saying it
doesn't matter if threads see the enabled a bit later.  
https://cs.chromium.org/chromium/src/base/trace_event/trace_log.cc?rcl=146911...

[boliu, 2016-07-21 21:56:16]

The CQ bit was checked by boliu@chromium.org

[commit-bot: I haz the power, 2016-07-21 21:56:42]

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, 2016-07-21 23:51:01]

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

[commit-bot: I haz the power, 2016-07-21 23:51:02]

Try jobs failed on following builders:
  win_chromium_rel_ng on master.tryserver.chromium.win (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.win/builders/win_chromium_rel_...)

[boliu, 2016-07-21 23:54:33]

The CQ bit was checked by boliu@chromium.org

[commit-bot: I haz the power, 2016-07-21 23:55:13]

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, 2016-07-22 03:27:57]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2016-07-22 03:29:39]

Description was changed from

==========
gpu: Fix crash in gpu tracer

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel
==========

to

==========
gpu: Fix crash in gpu tracer

It is possible to start tracing, ie GPUTracer::IsTracing goes from false
to true, between GPUTracer::BeginDecoding and GPUTracer::Begin. If that
happens, outputter_ is null, which calls a NPE. Fix that.
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_optional_gpu_tests_rel;master.tryserver.chromium.mac:mac_optional_gpu_tests_rel;master.tryserver.chromium.win:win_optional_gpu_tests_rel

Committed: https://crrev.com/b063270934133a2e13bbae92803eb4708090da70
Cr-Commit-Position: refs/heads/master@{#407023}
==========

[commit-bot: I haz the power, 2016-07-22 03:29:40]

Patchset 1 (id:??) landed as
https://crrev.com/b063270934133a2e13bbae92803eb4708090da70
Cr-Commit-Position: refs/heads/master@{#407023}