cc: Trigger deadline immediately when commit is aborted.

cc: Trigger deadline immediately when commit is aborted.

The scheduler should trigger it's deadline immediately if there's no
more work to be done in a given frame. This happens for instance when a
commit is aborted and there's nothing to draw.

BUG=432981
Committed: https://crrev.com/7f6eb54c666afea307ad3035c404e00692cacbf1
Cr-Commit-Position: refs/heads/master@{#311404}

[sunnyps, 2015-01-09 23:06:57]

sunnyps@chromium.org changed reviewers:
+ brianderson@chromium.org

[sunnyps, 2015-01-09 23:06:58]

PTAL

[brianderson, 2015-01-09 23:54:22]

Thanks for picking this up.

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_state...
File cc/scheduler/scheduler_state_machine.cc (right):

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_state...
cc/scheduler/scheduler_state_machine.cc:858: if (pending_swaps_ >=
max_pending_swaps_)
Putting this inside the if(needs_redraw) I think will make us not wait until the
swap ack to send the BeginMainFrame when needs_commit_ is true and we are
otherwise idle. This will delay the BeginMainFrame until the next BeginFrame,
which will hurt throughput.

Can you put this above the if(needs_redraw) and add a test for the above case?

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_unitt...
File cc/scheduler/scheduler_unittest.cc (right):

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_unitt...
cc/scheduler/scheduler_unittest.cc:2203: TEST(SchedulerTest,
BeginMainFrameAbortedTriggersImmediateDeadline) {
Does this test fail before this patch?

[sunnyps, 2015-01-13 19:36:01]

PTAL. It will be helpful to look at the overall diff rather than diff from the
first patch set.

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_state...
File cc/scheduler/scheduler_state_machine.cc (right):

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_state...
cc/scheduler/scheduler_state_machine.cc:858: if (pending_swaps_ >=
max_pending_swaps_)
On 2015/01/09 23:54:22, brianderson wrote:
> Putting this inside the if(needs_redraw) I think will make us not wait until
the
> swap ack to send the BeginMainFrame when needs_commit_ is true and we are
> otherwise idle. This will delay the BeginMainFrame until the next BeginFrame,
> which will hurt throughput.
> 
> Can you put this above the if(needs_redraw) and add a test for the above case?

Done. The test I had to change in scheduler_state_machine_unittests kinda
already tests this so I reverted my change there.

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_unitt...
File cc/scheduler/scheduler_unittest.cc (right):

https://codereview.chromium.org/822933004/diff/1/cc/scheduler/scheduler_unitt...
cc/scheduler/scheduler_unittest.cc:2203: TEST(SchedulerTest,
BeginMainFrameAbortedTriggersImmediateDeadline) {
On 2015/01/09 23:54:22, brianderson wrote:
> Does this test fail before this patch?

Yes, it does.

[brianderson, 2015-01-14 01:34:39]

lgtm

[sunnyps, 2015-01-14 01:34:55]

The CQ bit was checked by sunnyps@chromium.org

[commit-bot: I haz the power, 2015-01-14 01:35:43]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/822933004/40001

[commit-bot: I haz the power, 2015-01-14 03:47:59]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-01-14 03:49:04]

Patchset 3 (id:??) landed as
https://crrev.com/7f6eb54c666afea307ad3035c404e00692cacbf1
Cr-Commit-Position: refs/heads/master@{#311404}

[sunnyps, 2015-01-22 20:43:39]

A revert of this CL (patchset #3 id:40001) has been created in
https://codereview.chromium.org/871613002/ by sunnyps@chromium.org.

The reason for reverting is: caused major regression in tough_compositor_cases
crbug.com/450897.