cc: Add MainAndImplFrameTimeDelta UMA.

cc/scheduler/scheduler_state_machine.cc

Index: cc/scheduler/scheduler_state_machine.cc
diff --git a/cc/scheduler/scheduler_state_machine.cc b/cc/scheduler/scheduler_state_machine.cc
index b168034c3bb632006ca90dd6c572cb80ca2f67bf..23715ae6dd38b3a5629fe52881a9fa0a6c6747f5 100644
--- a/cc/scheduler/scheduler_state_machine.cc
+++ b/cc/scheduler/scheduler_state_machine.cc
@@ -613,7 +613,6 @@ void SchedulerStateMachine::WillCommit(bool commit_has_no_updates) {
       forced_redraw_state_ == FORCED_REDRAW_STATE_WAITING_FOR_DRAW) {
     DCHECK(!has_pending_tree_);
     needs_redraw_ = true;
-    active_tree_needs_first_draw_ = true;
   }
 
   // This post-commit work is common to both completed and aborted commits.
@@ -643,6 +642,13 @@ void SchedulerStateMachine::WillActivate() {
 }
 
 void SchedulerStateMachine::WillDrawInternal() {
+  // If a new active tree is pending after the one we are about to draw,
+  // the main thread is in a high latency mode.
+  // main_thread_missed_last_deadline_ is here in addition to
+  // OnBeginImplFrameIdle for cases where the scheduler aborts draws outside
+  // of the deadline.
+  main_thread_missed_last_deadline_ = CommitPending() || has_pending_tree_;

[brianderson, 2016/02/19 01:52:17]

Need to add a unit test for this, but probably won't split it out to a separate
patch this time.

I think this could cause us to unintentionally recover main latency when
can_draw_ goes from false to true.

Without this line, the HudWithRootLayerChange was failing the following DCHECK
in CompositorTimingHistory:

DCHECK_EQ(main_thread_missed_last_deadline,
                !main_and_impl_delta.is_zero());

+
   // We need to reset needs_redraw_ before we draw since the
   // draw itself might request another draw.
   needs_redraw_ = false;