mac: Add a metric for the time blocked by WaitForSurface().

mac: Add a metric for the time blocked by WaitForSurface().

The method RenderWidgetHostImpl::WaitForSurface() synchronously blocks the
browser's main thread, waiting for a frame of the right size to arrive from the
GPU process. It has a timeout of 50 milliseconds. This CL adds a metric
recording the time blocked by this method.

BUG=446369
Committed: https://crrev.com/edac18f894587d3461655bdade8210114d935c5e
Cr-Commit-Position: refs/heads/master@{#310124}

[erikchen, 2015-01-06 02:03:06]

Patchset #1 (id:1) has been deleted

[erikchen, 2015-01-06 02:04:09]

Patchset #1 (id:20001) has been deleted

[erikchen, 2015-01-06 02:06:14]

Patchset #1 (id:40001) has been deleted

[erikchen, 2015-01-06 02:07:09]

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

[erikchen, 2015-01-06 02:07:09]

ccameron: Please review.

[ccameron, 2015-01-06 02:14:21]

lgtm (with the one nit, up to you if you want to change it).

https://codereview.chromium.org/841443002/diff/60001/content/browser/renderer...
File content/browser/renderer_host/render_widget_host_impl.cc (right):

https://codereview.chromium.org/841443002/diff/60001/content/browser/renderer...
content/browser/renderer_host/render_widget_host_impl.cc:839: 
It should be possible to combine the base::TimeTicks::Now() from lines 837 and
840 (maybe have start_time, current_time, end_time). Not a huge difference,
though.

[erikchen, 2015-01-06 03:23:16]

https://codereview.chromium.org/841443002/diff/60001/content/browser/renderer...
File content/browser/renderer_host/render_widget_host_impl.cc (right):

https://codereview.chromium.org/841443002/diff/60001/content/browser/renderer...
content/browser/renderer_host/render_widget_host_impl.cc:839: 
On 2015/01/06 02:14:21, ccameron1 wrote:
> It should be possible to combine the base::TimeTicks::Now() from lines 837 and
> 840 (maybe have start_time, current_time, end_time). Not a huge difference,
> though.

It saves a call to TimeTicks::Now(), plus the original code was structured oddly
anyways. Done.

[erikchen, 2015-01-06 03:24:21]

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

[erikchen, 2015-01-06 03:24:22]

isherman: Please review.

The metric is being emitted very often. Is there a preferred method for
throttling emission rate?

[Ilya Sherman, 2015-01-06 04:33:42]

Histograms LGTM.

> The metric is being emitted very often. Is there a preferred method for
> throttling emission rate?

Typically, histograms don't need to be throttled -- they were designed to be
super low-overhead, to support analysis of networking code.  So, no throttling
is probably the best option, unless you need throttling for other reasons.

[erikchen, 2015-01-06 18:07:50]

The CQ bit was checked by erikchen@chromium.org

[commit-bot: I haz the power, 2015-01-06 18:08:13]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/841443002/80001

[commit-bot: I haz the power, 2015-01-06 20:10:09]

Committed patchset #2 (id:80001)

[commit-bot: I haz the power, 2015-01-06 20:11:50]

Patchset 2 (id:??) landed as
https://crrev.com/edac18f894587d3461655bdade8210114d935c5e
Cr-Commit-Position: refs/heads/master@{#310124}