Add more trace events to "benchmark" category

Add more trace events to "benchmark" category

This is to make the trace events captured during smoothness test more
useful in debugging smoothness/latency issues.

These added trace events are supposed to be giving most useful information
about the input/scheduling/rendering pipeline and have low occurrence in
each frame so as to minimize the performance overhead.

Following are the list for the added trace events roughly in the order
of the input->rendering pipeline:

LatencyInfo.Flow                                 // so we can follow the input latency flow
RenderWidgetHostViewAndroid::OnVSync             // when frame is started
InputHandlerProxy::HandleInputEvent              // when the input event is handled on impl
RenderWidget::OnHandleInputEvent                 // when the input event is handled on main
Scheduler::BeginFrame                            // cc decides to schedule a frame
Scheduler::BeginRetroFrame                       // cc starts a retro frame
Scheduler::BeginImplFrame                        // impl frame is scheduled
Scheduler::OnBeginImplFrameDeadline              // impl frame actually starts
OnSwapCompositorFrame                            // browser gets notified that renderer frame is ready
CompositorImpl::PostComposit                     // browser schedules a frame
SingleThreadProxy::CompositeImmediately          // browser starts a frame
SingleThreadProxy::DidSwapBuffersCompleteOnImplThread  // browser frame is done

BUG=449258
Committed: https://crrev.com/0190b029a83fad58e31a039bd8bba62d9d12ee5a
Cr-Commit-Position: refs/heads/master@{#312664}

[Yufeng Shen (Slow to review), 2015-01-15 20:46:35]

miletus@chromium.org changed reviewers:
+ brianderson@chromium.org, sievers@chromium.org, skyostil@chromium.org

[Yufeng Shen (Slow to review), 2015-01-15 20:48:40]

Daniel, Sami and Brian,

Do you guys feel these added traces are useful in debugging
jank/latency issues ? Any other trace events you guys want 
to see ?

[Sami, 2015-01-19 11:32:44]

These look like a reasonable set of events to add without introducing too much
overhead. lgtm for the ones in cc/.

I was wondering if we should have at least one event in the gpu process, but I
guess the swap complete covers that indirectly. WDYT?

[Yufeng Shen (Slow to review), 2015-01-19 18:31:04]

On 2015/01/19 11:32:44, Sami wrote:
> These look like a reasonable set of events to add without introducing too much
> overhead. lgtm for the ones in cc/.
> 
> I was wondering if we should have at least one event in the gpu process, but I
> guess the swap complete covers that indirectly. WDYT?

Yeah, we are tracking the final gpu swap buffer already.
But if you feel there is more steps in the gpu side worth tracking please let me
know.

[brianderson, 2015-01-21 10:26:22]

lgtm

https://codereview.chromium.org/855553002/diff/40001/cc/trees/single_thread_p...
File cc/trees/single_thread_proxy.cc (right):

https://codereview.chromium.org/855553002/diff/40001/cc/trees/single_thread_p...
cc/trees/single_thread_proxy.cc:484:
"SingleThreadProxy::DidSwapBuffersCompleteOnImplThread");
Can you also add this for the ThreadProxy?

[Yufeng Shen (Slow to review), 2015-01-21 16:31:13]

On 2015/01/21 10:26:22, brianderson wrote:
> lgtm
> 
>
https://codereview.chromium.org/855553002/diff/40001/cc/trees/single_thread_p...
> File cc/trees/single_thread_proxy.cc (right):
> 
>
https://codereview.chromium.org/855553002/diff/40001/cc/trees/single_thread_p...
> cc/trees/single_thread_proxy.cc:484:
> "SingleThreadProxy::DidSwapBuffersCompleteOnImplThread");
> Can you also add this for the ThreadProxy?

done.

[no sievers, 2015-01-22 19:36:50]

lgtm

[Yufeng Shen (Slow to review), 2015-01-22 19:45:24]

The CQ bit was checked by miletus@chromium.org

[commit-bot: I haz the power, 2015-01-22 19:46:36]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/855553002/50001

[commit-bot: I haz the power, 2015-01-22 20:18:51]

Committed patchset #4 (id:50001)

[commit-bot: I haz the power, 2015-01-22 20:19:41]

Patchset 4 (id:??) landed as
https://crrev.com/0190b029a83fad58e31a039bd8bba62d9d12ee5a
Cr-Commit-Position: refs/heads/master@{#312664}

[Yufeng Shen (Slow to review), 2015-01-22 20:24:00]

A revert of this CL (patchset #4 id:50001) has been created in
https://codereview.chromium.org/867773002/ by miletus@chromium.org.

The reason for reverting is: This CL has perf implication and currently all the
android perf bots are red.

So lets delay the landing until we get perf bots back to normal.