cc: Unflake and re-enable the new pinch zoom unit tests.

cc: Unflake and re-enable the new pinch zoom unit tests.

This fixes the LayerTreeHostTestCrispUpAfterPinchEnds and
LayerTreeHostTestContinuousDrawWhenCreatingVisibleTiles unit tests to
not be flaky, by figuring out all the race conditions taking place and
ensuring each one passes before continuing.

R=enne, vmpstr
BUG=433208
Committed: https://crrev.com/c209e26d344bd88a276d1dc656b945e0b3bcd2c4
Cr-Commit-Position: refs/heads/master@{#304450}

[danakj, 2014-11-14 21:17:32]

https://codereview.chromium.org/732503005/diff/1/cc/trees/layer_tree_host_uni...
File cc/trees/layer_tree_host_unittest.cc (left):

https://codereview.chromium.org/732503005/diff/1/cc/trees/layer_tree_host_uni...
cc/trees/layer_tree_host_unittest.cc:5483: // This frame will not have any
damage, since it's actually the same as
oops this comment should stay now. fixed.

[vmpstr, 2014-11-14 21:29:49]

https://codereview.chromium.org/732503005/diff/20001/cc/trees/layer_tree_host...
File cc/trees/layer_tree_host_unittest.cc (right):

https://codereview.chromium.org/732503005/diff/20001/cc/trees/layer_tree_host...
cc/trees/layer_tree_host_unittest.cc:5551: // On frame 3, we will have a low res
tile complete for the pinch-out
Is doing a ASSERT_EQ(LOW_RESOLUTION,
tile->{priority(ACTIVE_TREE),priority(PENDING_TREE),combined_priority()}.resolution);
a safe thing here to verify that the comment isn't lying?

Where one of active, pending, combined can verify this?

[danakj, 2014-11-14 21:36:20]

PTAL

[danakj, 2014-11-17 17:06:01]

Addressed some more flaky that I found. PTAL

[vmpstr, 2014-11-17 18:40:07]

lgtm

[danakj, 2014-11-17 18:40:18]

The CQ bit was checked by danakj@chromium.org

[commit-bot: I haz the power, 2014-11-17 18:41:15]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/732503005/120001

[commit-bot: I haz the power, 2014-11-17 19:21:20]

Committed patchset #7 (id:120001)

[commit-bot: I haz the power, 2014-11-17 19:21:56]

Patchset 7 (id:??) landed as
https://crrev.com/c209e26d344bd88a276d1dc656b945e0b3bcd2c4
Cr-Commit-Position: refs/heads/master@{#304450}

[enne (OOO), 2014-11-17 19:25:13]

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
File cc/trees/layer_tree_host_unittest.cc (right):

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
cc/trees/layer_tree_host_unittest.cc:5693: // We may get another draw if we
activate due to the pinch (eg LCD text
Weird.  Could you just start without LCD text enabled and have no extra draws? 
It seems like we should be able to control this.

[danakj, 2014-11-17 19:28:36]

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
File cc/trees/layer_tree_host_unittest.cc (right):

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
cc/trees/layer_tree_host_unittest.cc:5693: // We may get another draw if we
activate due to the pinch (eg LCD text
On 2014/11/17 19:25:13, enne wrote:
> Weird.  Could you just start without LCD text enabled and have no extra draws?

> It seems like we should be able to control this.

In the LCD patch, the layer gets init with LCD true, and then goes to false when
a scale is used. I could force LCD off at the start of this test, either by
putting a scale != 1, or by adding something to FakePictureLayer. Think it is
worth doing?

[enne (OOO), 2014-11-17 20:37:51]

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
File cc/trees/layer_tree_host_unittest.cc (right):

https://codereview.chromium.org/732503005/diff/120001/cc/trees/layer_tree_hos...
cc/trees/layer_tree_host_unittest.cc:5693: // We may get another draw if we
activate due to the pinch (eg LCD text
On 2014/11/17 19:28:36, danakj wrote:
> On 2014/11/17 19:25:13, enne wrote:
> > Weird.  Could you just start without LCD text enabled and have no extra
draws?
> 
> > It seems like we should be able to control this.
> 
> In the LCD patch, the layer gets init with LCD true, and then goes to false
when
> a scale is used. I could force LCD off at the start of this test, either by
> putting a scale != 1, or by adding something to FakePictureLayer. Think it is
> worth doing?

I think it'd be worth doing (maybe with opacity != 1?) to make the test more
predictable.  Then if anybody accidentally adds an additional draw after pinch,
this test would catch it, rather than silently being robust to arbitrary
flakiness?