Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/b6acafa1d1e2107f66deeed51865fb233b03641b
Cr-Commit-Position: refs/heads/master@{#422688}

[Fady Samuel, 2016-10-01 17:40:42]

Description was changed from

==========
Tie SurfaceFactory to a single FrameSinkId

BUG=
==========

to

==========
Tie SurfaceFactory to a single FrameSinkId

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

[Fady Samuel, 2016-10-01 17:40:43]

Description was changed from

==========
Tie SurfaceFactory to a single FrameSinkId

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.chromium.linux:linux_site_isolation
==========

to

==========
Tie SurfaceFactory to a single FrameSinkId

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[Fady Samuel, 2016-10-01 20:51:38]

Description was changed from

==========
Tie SurfaceFactory to a single FrameSinkId

BUG=
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId anymore, we can
simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[Fady Samuel, 2016-10-01 20:52:10]

fsamuel@chromium.org changed reviewers:
+ boliu@chromium.org, sky@chromium.org

[Fady Samuel, 2016-10-01 20:53:08]

fsamuel@chromium.org changed reviewers:
+ dtrainor@chromium.org, enne@chromium.org, jam@chromium.org

[Fady Samuel, 2016-10-01 20:53:09]

+boliu: android things
+sky: ui and services/ui things
+enne: cc things
+dtrainor: blimp things
+jam: content things

[Fady Samuel, 2016-10-01 20:55:02]

The CQ bit was checked by fsamuel@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-01 20:55:11]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[Fady Samuel, 2016-10-01 20:56:17]

Description was changed from

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId anymore, we can
simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId 
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[commit-bot: I haz the power, 2016-10-01 21:08:19]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-01 21:08:20]

Dry run: Exceeded global retry quota

[Fady Samuel, 2016-10-01 22:11:56]

The CQ bit was checked by fsamuel@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-01 22:12:02]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-01 22:25:11]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-01 22:25:12]

Dry run: Try jobs failed on following builders:
  cast_shell_linux on master.tryserver.chromium.linux (JOB_FAILED,
http://build.chromium.org/p/tryserver.chromium.linux/builders/cast_shell_linu...)

[Fady Samuel, 2016-10-01 22:35:44]

The CQ bit was checked by fsamuel@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-01 22:35:52]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-01 23:17:42]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-01 23:17:43]

Dry run: This issue passed the CQ dry run.

[Fady Samuel, 2016-10-01 23:38:54]

fsamuel@chromium.org changed reviewers:
+ reveman@chromium.org

[Fady Samuel, 2016-10-01 23:38:55]

+reveman@ for exo

[reveman, 2016-10-02 12:58:27]

components/exo lgtm

[boliu, 2016-10-03 05:47:04]

lgtm

[jam, 2016-10-03 17:40:36]

On 2016/10/01 20:53:09, Fady Samuel wrote:
> +boliu: android things
> +sky: ui and services/ui things
> +enne: cc things
> +dtrainor: blimp things
> +jam: content things

I'm not familiar with this code, please have someone else review it. if you
can't find a content/owner or content/browser owner, then have someone else
familiar with it review it and i can rs

[enne (OOO), 2016-10-03 18:06:38]

lgtm

[enne (OOO), 2016-10-03 18:14:24]

I'm not a content/ owner, but I took a look at the whole patch and it looks good
to me.  I've made similar changes in the past to SurfaceFactory and its clients,
so am probably a good choice to give an overall review here.

[jam, 2016-10-03 18:19:41]

On 2016/10/03 18:14:24, enne wrote:
> I'm not a content/ owner, but I took a look at the whole patch and it looks
good
> to me.  I've made similar changes in the past to SurfaceFactory and its
clients,
> so am probably a good choice to give an overall review here.

lgtm

[sky, 2016-10-03 19:18:26]

LGTM

https://codereview.chromium.org/2383373002/diff/80001/services/ui/surfaces/di...
File services/ui/surfaces/display_compositor.h (right):

https://codereview.chromium.org/2383373002/diff/80001/services/ui/surfaces/di...
services/ui/surfaces/display_compositor.h:71: cc::FrameSinkId frame_sink_id_;
If possible make this const.

https://codereview.chromium.org/2383373002/diff/80001/services/ui/ws/server_w...
File services/ui/ws/server_window_surface.h (right):

https://codereview.chromium.org/2383373002/diff/80001/services/ui/ws/server_w...
services/ui/ws/server_window_surface.h:64: cc::FrameSinkId frame_sink_id_;
Same comment here about const.

https://codereview.chromium.org/2383373002/diff/80001/ui/android/delegated_fr...
File ui/android/delegated_frame_host_android.h (right):

https://codereview.chromium.org/2383373002/diff/80001/ui/android/delegated_fr...
ui/android/delegated_frame_host_android.h:79: cc::FrameSinkId frame_sink_id_;
And here.

https://codereview.chromium.org/2383373002/diff/80001/ui/compositor/compositor.h
File ui/compositor/compositor.h (right):

https://codereview.chromium.org/2383373002/diff/80001/ui/compositor/composito...
ui/compositor/compositor.h:417: cc::FrameSinkId frame_sink_id_;
And here.

[Fady Samuel, 2016-10-03 19:45:29]

Addressed comments.

https://codereview.chromium.org/2383373002/diff/80001/services/ui/surfaces/di...
File services/ui/surfaces/display_compositor.h (right):

https://codereview.chromium.org/2383373002/diff/80001/services/ui/surfaces/di...
services/ui/surfaces/display_compositor.h:71: cc::FrameSinkId frame_sink_id_;
On 2016/10/03 19:18:25, sky wrote:
> If possible make this const.

Done.

https://codereview.chromium.org/2383373002/diff/80001/services/ui/ws/server_w...
File services/ui/ws/server_window_surface.h (right):

https://codereview.chromium.org/2383373002/diff/80001/services/ui/ws/server_w...
services/ui/ws/server_window_surface.h:64: cc::FrameSinkId frame_sink_id_;
On 2016/10/03 19:18:25, sky wrote:
> Same comment here about const.

Done.

https://codereview.chromium.org/2383373002/diff/80001/ui/android/delegated_fr...
File ui/android/delegated_frame_host_android.h (right):

https://codereview.chromium.org/2383373002/diff/80001/ui/android/delegated_fr...
ui/android/delegated_frame_host_android.h:79: cc::FrameSinkId frame_sink_id_;
On 2016/10/03 19:18:25, sky wrote:
> And here.

Done.

https://codereview.chromium.org/2383373002/diff/80001/ui/compositor/compositor.h
File ui/compositor/compositor.h (right):

https://codereview.chromium.org/2383373002/diff/80001/ui/compositor/composito...
ui/compositor/compositor.h:417: cc::FrameSinkId frame_sink_id_;
On 2016/10/03 19:18:25, sky wrote:
> And here.

Done.

[Fady Samuel, 2016-10-03 19:45:41]

The CQ bit was checked by fsamuel@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-10-03 19:46:05]

Dry run: CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-03 20:43:13]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-03 20:43:15]

Dry run: This issue passed the CQ dry run.

[Fady Samuel, 2016-10-03 22:10:48]

The CQ bit was checked by fsamuel@chromium.org

[Fady Samuel, 2016-10-03 22:10:48]

The patchset sent to the CQ was uploaded after l-g-t-m from
reveman@chromium.org, sky@chromium.org, jam@chromium.org, boliu@chromium.org,
enne@chromium.org
Link to the patchset: https://codereview.chromium.org/2383373002/#ps120001
(title: "Rebased")

[commit-bot: I haz the power, 2016-10-03 22:12:08]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-03 23:35:14]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-10-03 23:35:22]

Failed to apply patch for cc/test/test_compositor_frame_sink.cc:
While running git apply --index -3 -p1;
  error: patch failed: cc/test/test_compositor_frame_sink.cc:31
  Falling back to three-way merge...
  Applied patch to 'cc/test/test_compositor_frame_sink.cc' with conflicts.
  U cc/test/test_compositor_frame_sink.cc

Patch:       cc/test/test_compositor_frame_sink.cc
Index: cc/test/test_compositor_frame_sink.cc
diff --git a/cc/test/test_compositor_frame_sink.cc
b/cc/test/test_compositor_frame_sink.cc
index
c06eac6b38a026e5644602c9d5032d685383d592..31d80d4f5ec63c680f113d6ad4325e12ae173987
100644
--- a/cc/test/test_compositor_frame_sink.cc
+++ b/cc/test/test_compositor_frame_sink.cc
@@ -14,8 +14,7 @@
 #include "cc/output/output_surface.h"
 #include "cc/output/texture_mailbox_deleter.h"
 
-static constexpr uint32_t kCompositorClientId = 1;
-static constexpr uint32_t kCompositorSinkId = 1;
+static constexpr cc::FrameSinkId kCompositorFrameSinkId(1, 1);
 
 namespace cc {
 
@@ -31,10 +30,11 @@ TestCompositorFrameSink::TestCompositorFrameSink(
     bool force_disable_reclaim_resources)
     : CompositorFrameSink(std::move(compositor_context_provider),
                           std::move(worker_context_provider)),
+      frame_sink_id_(kCompositorFrameSinkId),
       surface_manager_(new SurfaceManager),
-      surface_id_allocator_(new SurfaceIdAllocator(
-          FrameSinkId(kCompositorClientId, kCompositorSinkId))),
-      surface_factory_(new SurfaceFactory(surface_manager_.get(), this)) {
+      surface_id_allocator_(new SurfaceIdAllocator(frame_sink_id_)),
+      surface_factory_(
+          new SurfaceFactory(frame_sink_id_, surface_manager_.get(), this)) {
   std::unique_ptr<SyntheticBeginFrameSource> begin_frame_source;
   std::unique_ptr<DisplayScheduler> scheduler;
   if (!synchronous_composite) {
@@ -89,11 +89,9 @@ bool
TestCompositorFrameSink::BindToClient(CompositorFrameSinkClient* client) {
   if (!capabilities_.delegated_sync_points_required && context_provider())
     context_provider()->SetLostContextCallback(base::Closure());
 
- 
surface_manager_->RegisterFrameSinkId(surface_id_allocator_->frame_sink_id());
-  surface_manager_->RegisterSurfaceFactoryClient(
-      surface_id_allocator_->frame_sink_id(), this);
-  display_->Initialize(this, surface_manager_.get(),
-                       surface_id_allocator_->frame_sink_id());
+  surface_manager_->RegisterFrameSinkId(frame_sink_id_);
+  surface_manager_->RegisterSurfaceFactoryClient(frame_sink_id_, this);
+  display_->Initialize(this, surface_manager_.get(), frame_sink_id_);
   display_->renderer_for_testing()->SetEnlargePassTextureAmountForTesting(
       enlarge_pass_texture_amount_);
   display_->SetVisible(true);
@@ -106,10 +104,8 @@ void TestCompositorFrameSink::DetachFromClient() {
   if (bound_) {
     if (!delegated_surface_id_.is_null())
       surface_factory_->Destroy(delegated_surface_id_);
-    surface_manager_->UnregisterSurfaceFactoryClient(
-        surface_id_allocator_->frame_sink_id());
-    surface_manager_->InvalidateFrameSinkId(
-        surface_id_allocator_->frame_sink_id());
+    surface_manager_->UnregisterSurfaceFactoryClient(frame_sink_id_);
+    surface_manager_->InvalidateFrameSinkId(frame_sink_id_);
     bound_ = false;
   }
   display_ = nullptr;

[Fady Samuel, 2016-10-04 01:59:42]

The CQ bit was checked by fsamuel@chromium.org

[Fady Samuel, 2016-10-04 01:59:43]

The patchset sent to the CQ was uploaded after l-g-t-m from
reveman@chromium.org, sky@chromium.org, jam@chromium.org, boliu@chromium.org,
enne@chromium.org
Link to the patchset: https://codereview.chromium.org/2383373002/#ps140001
(title: "Rebased")

[commit-bot: I haz the power, 2016-10-04 02:00:23]

CQ is trying da patch. Follow status at
 
https://chromium-cq-status.appspot.com/v2/patch-status/codereview.chromium.or...

[commit-bot: I haz the power, 2016-10-04 03:22:42]

Description was changed from

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId 
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId 
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

[commit-bot: I haz the power, 2016-10-04 03:22:46]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2016-10-04 03:28:26]

Description was changed from

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of 
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to 
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of 
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of 
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it 
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId 
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel
==========

to

==========
Reduce SurfaceIdAllocator usage and tie SurfaceFactory to a single FrameSinkId

This CL takes us two steps along the long windy road to an out of
process display compositor.

1. In the near future, SurfaceIds will be allocated in clients in order to
avoid sync IPCs for offscreen canvas. Thus, we need to reduce our
reliance on SurfaceIdAllocator. SurfaceIdAllocator is often used simply
to find out the frame_sink_id(). This CL gives owners of
SurfaceIdAllocator a FrameSinkId directly and removes that accessor
from the ID allocator. This significantly reduces usage of
SurfaceIdAllocator.

2. SurfaceFactory's role was not well defined previously. It practice, it
corresponds to a single CompositorFrameSink but in tests, surface IDs
with different FrameSinkIds were submitting CompositorFrames to
the same SurfaceFactory. This CL establishes the invariant that a
SurfaceFactory belongs to a single CompositorFrameSink. In the near
future, we won't need to pass a SurfaceFactory a full SurfaceId
anymore, we can simply give it the "LocalFrameId" part and assume
a constant FrameSinkId. This will reduce the the data needed to send to
FrameSinkObservers from SurfaceFactories and from frame sources to
SurfaceFactories.

BUG=647852
CQ_INCLUDE_TRYBOTS=master.tryserver.blink:linux_precise_blink_rel

Committed: https://crrev.com/b6acafa1d1e2107f66deeed51865fb233b03641b
Cr-Commit-Position: refs/heads/master@{#422688}
==========

[commit-bot: I haz the power, 2016-10-04 03:28:28]

Patchset 8 (id:??) landed as
https://crrev.com/b6acafa1d1e2107f66deeed51865fb233b03641b
Cr-Commit-Position: refs/heads/master@{#422688}