Add browser tests for USB device add/remove events.

Add browser tests for USB device add/remove events.

These tests are essentially copies of the HID device add and remove
event tests. In order to reconcile issues with the MockUsbService
lifetime it is now tracked by a MessageLoop::DestructionObserver at the
generic UsbService level. This is similar to the pattern used by the
HidService until it was moved to the UI thread and could use a
LazyInstance.

BUG=411715
Committed: https://crrev.com/45919813ba21243f14a326b593a9fae52ff4a2f8
Cr-Commit-Position: refs/heads/master@{#312538}

[Reilly Grant (use Gerrit), 2015-01-13 02:07:09]

Patchset #1 (id:1) has been deleted

[Reilly Grant (use Gerrit), 2015-01-13 02:28:55]

reillyg@chromium.org changed reviewers:
+ yoz@chromium.org

[Reilly Grant (use Gerrit), 2015-01-13 02:28:56]

Yoyo, please take a look at these tests (and associated lifecycle changes that
enable them to not leak memory).

[Yoyo Zhou, 2015-01-21 06:57:44]

LGTM

https://codereview.chromium.org/800963005/diff/20001/device/usb/usb_service.cc
File device/usb/usb_service.cc (right):

https://codereview.chromium.org/800963005/diff/20001/device/usb/usb_service.c...
device/usb/usb_service.cc:51: g_service =
UsbServiceImpl::Create(ui_task_runner);
It is odd for UsbService to depend on UsbServiceImpl this way. But you might not
be able to do better than this if you don't have better guarantees about where
UsbService might get created.

https://codereview.chromium.org/800963005/diff/20001/extensions/browser/api/u...
File extensions/browser/api/usb/usb_apitest.cc (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/browser/api/u...
extensions/browser/api/usb/usb_apitest.cc:130: // Public wrapper around the
protected base class method.
Why not make the base class friend class UsbApiTest?

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
File extensions/test/data/api_test/usb/add_event/background.js (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
extensions/test/data/api_test/usb/add_event/background.js:5:
chrome.usb.onDeviceAdded.addListener(function (device) {
nit: while you're here, delete the space after function.

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
File extensions/test/data/api_test/usb/remove_event/background.js (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
extensions/test/data/api_test/usb/remove_event/background.js:7:
chrome.usb.onDeviceRemoved.addListener(function (device) {
ditto in this file

[Reilly Grant (use Gerrit), 2015-01-22 00:45:01]

https://codereview.chromium.org/800963005/diff/20001/device/usb/usb_service.cc
File device/usb/usb_service.cc (right):

https://codereview.chromium.org/800963005/diff/20001/device/usb/usb_service.c...
device/usb/usb_service.cc:51: g_service =
UsbServiceImpl::Create(ui_task_runner);
On 2015/01/21 06:57:43, Yoyo Zhou wrote:
> It is odd for UsbService to depend on UsbServiceImpl this way. But you might
not
> be able to do better than this if you don't have better guarantees about where
> UsbService might get created.

There's a similar pattern in HidService::GetInstance where the platform
appropriate subclass is instantiated by a static function in the base class.
Here in USB there is only one subclass but there will eventually be one per
platform. UsbService acts as a light-weight factory.

https://codereview.chromium.org/800963005/diff/20001/extensions/browser/api/u...
File extensions/browser/api/usb/usb_apitest.cc (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/browser/api/u...
extensions/browser/api/usb/usb_apitest.cc:130: // Public wrapper around the
protected base class method.
On 2015/01/21 06:57:43, Yoyo Zhou wrote:
> Why not make the base class friend class UsbApiTest?

I tried this and no one in my office could figure out why it wasn't working. I
also crashed clang a couple times.

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
File extensions/test/data/api_test/usb/add_event/background.js (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
extensions/test/data/api_test/usb/add_event/background.js:5:
chrome.usb.onDeviceAdded.addListener(function (device) {
On 2015/01/21 06:57:43, Yoyo Zhou wrote:
> nit: while you're here, delete the space after function.

I swear some linter complained about me not putting spaces after function but
Google's style guide agrees with you.

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
File extensions/test/data/api_test/usb/remove_event/background.js (right):

https://codereview.chromium.org/800963005/diff/20001/extensions/test/data/api...
extensions/test/data/api_test/usb/remove_event/background.js:7:
chrome.usb.onDeviceRemoved.addListener(function (device) {
On 2015/01/21 06:57:43, Yoyo Zhou wrote:
> ditto in this file

Done.

[Reilly Grant (use Gerrit), 2015-01-22 01:30:27]

The CQ bit was checked by reillyg@chromium.org

[commit-bot: I haz the power, 2015-01-22 01:32:46]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/800963005/40001

[commit-bot: I haz the power, 2015-01-22 02:40:39]

Committed patchset #2 (id:40001)

[commit-bot: I haz the power, 2015-01-22 02:41:37]

Patchset 2 (id:??) landed as
https://crrev.com/45919813ba21243f14a326b593a9fae52ff4a2f8
Cr-Commit-Position: refs/heads/master@{#312538}