Stop unnecessary flushing in Canvas2DLayerBridge

Stop unnecessary flushing in Canvas2DLayerBridge

Experiments with chrome://tracing revealed non-negligible
CPU overhead from calling SkCanvas::flush() and gl->Flush()
unnecessarily when nothing has been drawn to the canvas since the
previous flush. This change adds a simple early exit.

Performance gain will be visible through existing telemetry benchmarks.

BUG=681200
Review-Url: https://codereview.chromium.org/2637553002
Cr-Commit-Position: refs/heads/master@{#443917}
Committed: https://chromium.googlesource.com/chromium/src/+/11dadab7208d79cf95542cd9ac8c285844141a6a

[Justin Novosad, 2017-01-13 23:44:34]

The CQ bit was checked by junov@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2017-01-13 23:45:15]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Justin Novosad, 2017-01-13 23:46:06]

junov@chromium.org changed reviewers:
+ xlai@chromium.org

[Justin Novosad, 2017-01-13 23:46:07]

PTAL

[commit-bot: I haz the power, 2017-01-14 01:23:22]

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

[commit-bot: I haz the power, 2017-01-14 01:23:24]

Dry run: This issue passed the CQ dry run.

[xlai (Olivia), 2017-01-16 15:57:28]

lgtm

[Justin Novosad, 2017-01-16 16:30:30]

The CQ bit was checked by junov@chromium.org

[commit-bot: I haz the power, 2017-01-16 16:30:58]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[xlai (Olivia), 2017-01-16 16:58:19]

https://codereview.chromium.org/2637553002/diff/1/third_party/WebKit/Source/p...
File third_party/WebKit/Source/platform/graphics/Canvas2DLayerBridge.cpp
(right):

https://codereview.chromium.org/2637553002/diff/1/third_party/WebKit/Source/p...
third_party/WebKit/Source/platform/graphics/Canvas2DLayerBridge.cpp:788: if
(!m_didDrawSinceLastFlush)
I have a second thought about this CL: how about we add a small unit test to it
to verifies it works and prevent future code change from modifying the behavior?
There is an extra code branch introduced here and a unit test seems to be
necessary in this case.

[commit-bot: I haz the power, 2017-01-16 17:47:20]

CQ is committing da patch.

Bot data: {"patchset_id": 1, "attempt_start_ts": 1484584230363100, "parent_rev":
"c0b65a8ae7ba10b3fca69c19237f3b0f91ceed5c", "commit_rev":
"11dadab7208d79cf95542cd9ac8c285844141a6a"}

[commit-bot: I haz the power, 2017-01-16 17:47:49]

Description was changed from

==========
Stop unnecessary flushing in Canvas2DLayerBridge

Experiments with chrome://tracing revealed non-negligible
CPU overhead from calling SkCanvas::flush() and gl->Flush()
unnecessarily when nothing has been drawn to the canvas since the
previous flush. This change adds a simple early exit.

Performance gain will be visible through existing telemetry benchmarks.

BUG=681200
==========

to

==========
Stop unnecessary flushing in Canvas2DLayerBridge

Experiments with chrome://tracing revealed non-negligible
CPU overhead from calling SkCanvas::flush() and gl->Flush()
unnecessarily when nothing has been drawn to the canvas since the
previous flush. This change adds a simple early exit.

Performance gain will be visible through existing telemetry benchmarks.

BUG=681200

Review-Url: https://codereview.chromium.org/2637553002
Cr-Commit-Position: refs/heads/master@{#443917}
Committed:
https://chromium.googlesource.com/chromium/src/+/11dadab7208d79cf95542cd9ac8c...
==========

[commit-bot: I haz the power, 2017-01-16 17:47:50]

Committed patchset #1 (id:1) as
https://chromium.googlesource.com/chromium/src/+/11dadab7208d79cf95542cd9ac8c...

[Justin Novosad, 2017-01-16 18:25:01]

On 2017/01/16 16:58:19, xlai (Olivia) wrote:
>
https://codereview.chromium.org/2637553002/diff/1/third_party/WebKit/Source/p...
> File third_party/WebKit/Source/platform/graphics/Canvas2DLayerBridge.cpp
> (right):
> 
>
https://codereview.chromium.org/2637553002/diff/1/third_party/WebKit/Source/p...
> third_party/WebKit/Source/platform/graphics/Canvas2DLayerBridge.cpp:788: if
> (!m_didDrawSinceLastFlush)
> I have a second thought about this CL: how about we add a small unit test to
it
> to verifies it works and prevent future code change from modifying the
behavior?
> There is an extra code branch introduced here and a unit test seems to be
> necessary in this case.

If we loose this behavior, performance tests will fire alerts.
But it is still a valid point. I will follow-up...