Added Destroy() functions for GPUTimer and GPUTracer/GPUTrace.

Added Destroy() functions for GPUTimer and GPUTracer/GPUTrace.

BUG=467202
TEST=Shut down during gpu.device trace has no warnings.

Committed: https://crrev.com/dbfb90d4dac51818e040c01ee2986bf64a18dfe3
Cr-Commit-Position: refs/heads/master@{#322304}

[David Yen, 2015-03-17 22:39:35]

dyen@chromium.org changed reviewers:
+ dcastagna@chromium.org, piman@chromium.org, vmiura@chromium.org

[David Yen, 2015-03-17 22:39:35]

+vmiura for Gles2CmdDecoder and GPUTracer changes
+dcastagna for GPUTiming changes
+piman for perf changes

I've added a destroy() function to properly handle deletions of the GL Query
objects. PTAL.

[piman, 2015-03-18 01:53:56]

lgtm

[Daniele Castagna, 2015-03-18 17:02:33]

lgtm.

You could consider avoiding calling destroy on timers (and probably on GPUTrace)
if there is no context, instead of propagating have_context all the way down.

[David Yen, 2015-03-18 23:39:42]

On 2015/03/18 17:02:33, Daniele Castagna wrote:
> lgtm.
> 
> You could consider avoiding calling destroy on timers (and probably on
GPUTrace)
> if there is no context, instead of propagating have_context all the way down.

I think it's good to call destroy() even if we do not have the context.

I am also using destroy() to memset the queries array so I can DCHECK to be sure
destroy() is called in the destructor.

[vmiura, 2015-03-19 22:56:36]

LGTM

https://codereview.chromium.org/1018773002/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/gles2_cmd_decoder.cc (right):

https://codereview.chromium.org/1018773002/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/gles2_cmd_decoder.cc:3754:
gpu_tracer_->Destroy(have_context);
Should we also delete gpu_tracer_here? gpu_tracer_ = NULL?

[David Yen, 2015-03-25 23:33:58]

https://codereview.chromium.org/1018773002/diff/20001/gpu/command_buffer/serv...
File gpu/command_buffer/service/gles2_cmd_decoder.cc (right):

https://codereview.chromium.org/1018773002/diff/20001/gpu/command_buffer/serv...
gpu/command_buffer/service/gles2_cmd_decoder.cc:3754:
gpu_tracer_->Destroy(have_context);
On 2015/03/19 22:56:36, vmiura wrote:
> Should we also delete gpu_tracer_here? gpu_tracer_ = NULL?

Done.

[David Yen, 2015-03-25 23:34:16]

The CQ bit was checked by dyen@chromium.org

[David Yen, 2015-03-25 23:34:16]

The patchset sent to the CQ was uploaded after l-g-t-m from
dcastagna@chromium.org, vmiura@chromium.org, piman@chromium.org
Link to the patchset: https://codereview.chromium.org/1018773002/#ps40001
(title: "Delete GPU Tracer after destroying it in decoder")

[commit-bot: I haz the power, 2015-03-25 23:35:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1018773002/40001

[commit-bot: I haz the power, 2015-03-26 02:41:57]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-03-26 02:42:45]

Patchset 3 (id:??) landed as
https://crrev.com/dbfb90d4dac51818e040c01ee2986bf64a18dfe3
Cr-Commit-Position: refs/heads/master@{#322304}