Implement permission_broker for serial devices.

Implement permission_broker for serial devices.

This change makes Chrome on Chrome OS ask permission_broker for access to serial devices.

BUG=173548
TEST=Chrome can access serial devices on Linux, chromeos=1, full Chrome OS.

Committed: https://crrev.com/2aa4e40e4bac2c708c0e91c9a7c03fb9539323fd
Cr-Commit-Position: refs/heads/master@{#300364}

[Jorge Lucangeli Obes, 2014-10-10 21:17:26]

jorgelo@chromium.org changed reviewers:
+ reillyg@chromium.org, rockot@chromium.org

[Jorge Lucangeli Obes, 2014-10-10 21:17:27]

Hi guys,

This is a very early draft to discuss where in the serial device stack we should
put the query to permission_broker for access to serial devices. Let me know
what you think, don't be shy.

Thanks!

[Jorge Lucangeli Obes, 2014-10-10 21:17:27]

Hi guys,

This is a very early draft to discuss where in the serial device stack we should
put the query to permission_broker for access to serial devices. Let me know
what you think, don't be shy.

Thanks!

[Reilly Grant (use Gerrit), 2014-10-10 21:31:53]

Given that the open operation is asynchronous anyways this seems like the
perfect place to put it.

https://chromiumcodereview.appspot.com/646053002/diff/1/device/serial/serial_...
File device/serial/serial_io_handler_posix.cc (right):

https://chromiumcodereview.appspot.com/646053002/diff/1/device/serial/serial_...
device/serial/serial_io_handler_posix.cc:148: io_message_loop->PostTask(
chromeos::PermissionBrokerClient must as I understand it be used from the UI
thread only.

[Jorge Lucangeli Obes, 2014-10-13 20:56:43]

PTAL. PermissionBrokerClient is now called on the UI thread. It required some
plumbing but it's basically the same as what's done for the FILE thread.

https://chromiumcodereview.appspot.com/646053002/diff/1/device/serial/serial_...
File device/serial/serial_io_handler_posix.cc (right):

https://chromiumcodereview.appspot.com/646053002/diff/1/device/serial/serial_...
device/serial/serial_io_handler_posix.cc:148: io_message_loop->PostTask(
On 2014/10/10 21:31:53, reillyg wrote:
> chromeos::PermissionBrokerClient must as I understand it be used from the UI
> thread only.

Done.

[Jorge Lucangeli Obes, 2014-10-15 00:23:27]

Friendly ping. Thanks for taking a look!

[Reilly Grant (use Gerrit), 2014-10-15 00:32:55]

lgtm with a nit

[Reilly Grant (use Gerrit), 2014-10-15 00:33:03]

https://chromiumcodereview.appspot.com/646053002/diff/170001/device/serial/se...
File device/serial/serial_io_handler.cc (right):

https://chromiumcodereview.appspot.com/646053002/diff/170001/device/serial/se...
device/serial/serial_io_handler.cc:39: scoped_refptr<base::MessageLoopProxy>
ui_message_loop) {
Why pass these as parameters when they are available as object members in the
function below?

[Jorge Lucangeli Obes, 2014-10-15 00:35:58]

https://chromiumcodereview.appspot.com/646053002/diff/170001/device/serial/se...
File device/serial/serial_io_handler.cc (right):

https://chromiumcodereview.appspot.com/646053002/diff/170001/device/serial/se...
device/serial/serial_io_handler.cc:39: scoped_refptr<base::MessageLoopProxy>
ui_message_loop) {
On 2014/10/15 00:33:02, reillyg wrote:
> Why pass these as parameters when they are available as object members in the
> function below?

Because they're private members and need to be accessed by RequestAccess in
SerialIoHandlerPosix (to post to the UI thread), which is a subclass.

I could also add accessors, like what's done for file().

[Ken Rockot(use gerrit already), 2014-10-15 21:08:49]

lgtm too, fwiw

[Jorge Lucangeli Obes, 2014-10-15 22:23:06]

On 2014/10/15 21:08:49, Ken Rockot wrote:
> lgtm too, fwiw

Thanks guys! I'm fixing one small detail on the Chrome OS side and then I'll
land this.

[Jorge Lucangeli Obes, 2014-10-17 21:01:29]

On 2014/10/15 22:23:06, Jorge Lucangeli Obes wrote:
> On 2014/10/15 21:08:49, Ken Rockot wrote:
> > lgtm too, fwiw
> 
> Thanks guys! I'm fixing one small detail on the Chrome OS side and then I'll
> land this.

Alright, fixed the Chrome OS side, should I remove the arguments and create
accessors for the object memebers?

[Jorge Lucangeli Obes, 2014-10-20 21:57:42]

On 2014/10/17 21:01:29, Jorge Lucangeli Obes wrote:
> On 2014/10/15 22:23:06, Jorge Lucangeli Obes wrote:
> > On 2014/10/15 21:08:49, Ken Rockot wrote:
> > > lgtm too, fwiw
> > 
> > Thanks guys! I'm fixing one small detail on the Chrome OS side and then I'll
> > land this.
> 
> Alright, fixed the Chrome OS side, should I remove the arguments and create
> accessors for the object memebers?

I'm gonna land this and I can address the accessors in a follow-up CL.

[Jorge Lucangeli Obes, 2014-10-20 21:57:47]

The CQ bit was checked by jorgelo@chromium.org

[commit-bot: I haz the power, 2014-10-20 22:00:58]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/646053002/220001

[commit-bot: I haz the power, 2014-10-20 22:43:08]

Committed patchset #5 (id:220001)

[commit-bot: I haz the power, 2014-10-20 22:44:43]

Patchset 5 (id:??) landed as
https://crrev.com/2aa4e40e4bac2c708c0e91c9a7c03fb9539323fd
Cr-Commit-Position: refs/heads/master@{#300364}