Move gamepad_service out of content/ and into device/

Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330

Committed: https://crrev.com/bee6a64bd524847e541f6ef11c73c0328094b9d3
Cr-Commit-Position: refs/heads/master@{#437716}

[ke.he, 2016-12-09 06:39:51]

Patchset #1 (id:1) has been deleted

[ke.he, 2016-12-09 06:40:13]

The CQ bit was checked by ke.he@intel.com to run a CQ dry run

[commit-bot: I haz the power, 2016-12-09 06:40:31]

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

[ke.he, 2016-12-09 06:41:08]

Description was changed from

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

BUG = 612330

Convert Gamepad IPC messages into mojo interface.

This patch is a subtask of "Decoupling Gamepad from //content".

The Gamepad IPC messages are converted into mojo interfaces, so the
GamepadBrowserMessageFilter doesn't need to extend BrowserMessageFilter anymore
so we renamed it as GamepadMonitor.
GamepadMonitor still extend GamepadConsumer interface. Each RenderProcessHost
owns one GamepadMonitor. GamepadMonitor instance is created by "StrongBinding"
and will be registered into the "ConsumerSet" of GamepadService.
The base::SharedMemory handle are wrapped into mojo::SharedMemoryHandle and
passed as parameter of mojo interface from Browser to Renderer, Renderer Process
unwrap it to get the real shared memory contents.
Note that to obtain a new SharedMemoryHandle to wrap in a Mojo handle we simply
clone the existing one. While it's not obvious, this is functionally equivalent
to the old approach of passing the peer handle: when the renderer unwraps the
Mojo handle that it obtained into a base::SharedMemoryHandle, it fills in that
handle with its own process ID.
==========

to

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

BUG = 612330
==========

[commit-bot: I haz the power, 2016-12-09 07:40:37]

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

[commit-bot: I haz the power, 2016-12-09 07:40:38]

Dry run: This issue passed the CQ dry run.

[ke.he, 2016-12-09 07:44:54]

ke.he@intel.com changed reviewers:
+ bajones@chromium.org, blundell@chromium.org

[ke.he, 2016-12-09 07:44:55]

Hi, blundell@, This CL is almost a "copy" from CL
(https://codereview.chromium.org/2466073002/), the owner is bajones@.  ---Hi,
bajones@, thanks:)

Based on it, I fixed the unittest error, completely removed the
"content/common/gamepad_hardware_buffer.h", and also move the gamepad_monitor
into device/gamepad.

So, PTAL, thanks very much!

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_monitor.cc (left):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_monitor.cc:18: DCHECK_CURRENTLY_ON(BrowserThread::IO);
GamepadMonitor is created in "StrongBinding" way, RenderProcessHost register it
in IO thread.
So we don't need to this DCHECK anymore.

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_provider_unittest.cc (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_provider_unittest.cc:103: do {
Before this patch, it is the "MockGamepadSharedBuffer" that the provider holds,
so here we can get shared memory contents directly.
In this patch, the provider creates a "real" GamepadSharedBuffer by default in
its constructor, so we have to deal with the seqlock read operations.

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_shared_buffer.h (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_shared_buffer.h:17: GamepadHardwareBuffer is stored in
shared memory that's shared between the
copied from /content/common/gamepad_hardware_buffer.h, which has been deleted by
this patch.

[blundell, 2016-12-09 13:35:22]

LGTM, thanks!

Please add to the bottom of the CL description something like:

"Based on a patch by bajones@chromium.org."

It's always nice to explicitly give credit in these kinds of cases :).

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_provider_unittest.cc (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_provider_unittest.cc:103: do {
On 2016/12/09 07:44:55, ke.he wrote:
> Before this patch, it is the "MockGamepadSharedBuffer" that the provider
holds,
> so here we can get shared memory contents directly.
> In this patch, the provider creates a "real" GamepadSharedBuffer by default in
> its constructor, so we have to deal with the seqlock read operations.

Seems like it would be worthwhile to make a helper function for this given that
it's repeated several times.

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_service_unittest.cc (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_service_unittest.cc:68: void TearDown() override {
base::RunLoop().RunUntilIdle(); }
Why is this necessary?

[ke.he, 2016-12-09 14:57:03]

Description was changed from

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

BUG = 612330
==========

to

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330
==========

[ke.he, 2016-12-09 15:12:23]

Patchset #2 (id:40001) has been deleted

[ke.he, 2016-12-09 15:15:22]

ke.he@intel.com changed reviewers:
+ jam@chromium.org, tsepez@chromium.org

[ke.he, 2016-12-09 15:15:23]

blundell@, thanks very much. CL updated.

jam@, Could PTAL on those files in content/ and
device/BUILD.gn and
ppapi/shared_impl/ppb_gamepad_shared.h"  Thanks!

tsepez@, I make trivial changes in
device/gamepad/public/interfaces/gamepad.mojom, PTAL. thanks!

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_provider_unittest.cc (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_provider_unittest.cc:103: do {
On 2016/12/09 13:35:22, blundell wrote:
> On 2016/12/09 07:44:55, ke.he wrote:
> > Before this patch, it is the "MockGamepadSharedBuffer" that the provider
> holds,
> > so here we can get shared memory contents directly.
> > In this patch, the provider creates a "real" GamepadSharedBuffer by default
in
> > its constructor, so we have to deal with the seqlock read operations.
> 
> Seems like it would be worthwhile to make a helper function for this given
that
> it's repeated several times.

Done.

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
File device/gamepad/gamepad_service_unittest.cc (right):

https://codereview.chromium.org/2563483006/diff/20001/device/gamepad/gamepad_...
device/gamepad/gamepad_service_unittest.cc:68: void TearDown() override {
base::RunLoop().RunUntilIdle(); }
On 2016/12/09 13:35:22, blundell wrote:
> Why is this necessary?

not necessary, removed.

[ke.he, 2016-12-09 15:15:35]

The CQ bit was checked by ke.he@intel.com to run a CQ dry run

[commit-bot: I haz the power, 2016-12-09 15:15:49]

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

[blundell, 2016-12-09 15:17:45]

bajones@ should definitely review this as well :).

[ke.he, 2016-12-09 15:21:00]

On 2016/12/09 15:17:45, blundell wrote:
> bajones@ should definitely review this as well :).

Yeah, bajones@, PTAL. thanks again for your patch:)

[commit-bot: I haz the power, 2016-12-09 16:29:44]

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

[commit-bot: I haz the power, 2016-12-09 16:29:45]

Dry run: This issue passed the CQ dry run.

[Tom Sepez, 2016-12-09 17:15:55]

RS LGTM on mojom

[jam, 2016-12-09 17:40:51]

lgtm

[bajones, 2016-12-09 19:23:45]

LGTM, and thanks for doing all the rebase/test fixing work here to move this
forward!

[ke.he, 2016-12-10 00:04:00]

The CQ bit was checked by ke.he@intel.com

[ke.he, 2016-12-10 00:04:01]

The patchset sent to the CQ was uploaded after l-g-t-m from
blundell@chromium.org
Link to the patchset: https://codereview.chromium.org/2563483006/#ps60001
(title: "addressed Blundell's review comments")

[commit-bot: I haz the power, 2016-12-10 00:05:03]

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

[ke.he, 2016-12-10 00:17:56]

The CQ bit was unchecked by ke.he@intel.com

[ke.he, 2016-12-10 00:17:57]

The CQ bit was checked by ke.he@intel.com

[commit-bot: I haz the power, 2016-12-10 00:18:33]

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, 2016-12-10 01:56:24]

CQ is committing da patch.

Bot data: {"patchset_id": 60001, "attempt_start_ts": 1481329077450590,
"parent_rev": "a79c8e05238bcd28a58c596f62292827d73a2b0c", "commit_rev":
"c6433c19e55143b190ec635176dbbcff1f373048"}

[commit-bot: I haz the power, 2016-12-10 01:56:49]

Description was changed from

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330
==========

to

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330

Review-Url: https://codereview.chromium.org/2563483006
==========

[commit-bot: I haz the power, 2016-12-10 01:56:50]

Committed patchset #2 (id:60001)

[commit-bot: I haz the power, 2016-12-12 14:59:25]

Description was changed from

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330

Review-Url: https://codereview.chromium.org/2563483006
==========

to

==========
Move gamepad_service out of content/ and into device/

This patch is a subtask of "Decoupling Gamepad from //content".

In this patch the GamepadService, GamepadServiceUnittest, GamepadMonitor and
GamepadSharedBuffer are moved out of content/browser/gamepad or render_host,
and into device/gamepad/.

The file //content/common/gamepad_hardware_buffer.h is removed, we move the
definition of GamepadHardwareBuffer into gamepad_shared_memory.h, and change
its definition by using generic SharedMemorySeqLockBuffer<blink::WebGamepads>.

In the unittest of GamepadProvider, we don't pass a MockSharedMemoryBuffer
(which has no seqlock controls) anymore. GamepadProvider will create a
SharedMemoryBuffer by default in its constructor, so we add the seqlock-read
accordingly in its unittest.

Based on a patch by bajones@chromium.org.

BUG = 612330

Committed: https://crrev.com/bee6a64bd524847e541f6ef11c73c0328094b9d3
Cr-Commit-Position: refs/heads/master@{#437716}
==========

[commit-bot: I haz the power, 2016-12-12 14:59:27]

Patchset 2 (id:??) landed as
https://crrev.com/bee6a64bd524847e541f6ef11c73c0328094b9d3
Cr-Commit-Position: refs/heads/master@{#437716}