Switch SessionManagerClient::RestartJob to use RestartJobWithAuth

Switch SessionManagerClient::RestartJob to use RestartJobWithAuth

The session manager is once again providing a RestartJobWithAuth
method to allow callers to directly authorize their request to
restart the browser by providing a unix domain socket ready to be
SO_PEERCREDified in lieu of a pid.

This has the advantage of working inside a PID namespace, while
normal dbus mechanisms do not.

BUG=448821
TEST=integration tests TBD
R=stevenjb@chromium.org

Committed: https://crrev.com/ec53492f57498b60bbffe2aba1c80df39589925f
Cr-Commit-Position: refs/heads/master@{#327417}

[Chris Masone, 2015-04-28 00:55:38]

Haven't done my integration tests yet, and this relies on
https://chromium-review.googlesource.com/#/c/267634/ going in first. But, do you
have any general objections to this code?

[stevenjb, 2015-04-28 01:25:26]

lgtm, I think :)

https://codereview.chromium.org/1111653002/diff/1/chromeos/dbus/session_manag...
File chromeos/dbus/session_manager_client.cc (right):

https://codereview.chromium.org/1111653002/diff/1/chromeos/dbus/session_manag...
chromeos/dbus/session_manager_client.cc:133: false);
This is all pretty confusing. I'm pretty sure I followed it through and it looks
OK, but a couple of suggestions:

a) I'm not sure this will actually clarify things, but maybe pass local_auth_fd
to CallRestartJobWithValidFd instead of binding it to on_restart_job here? It
means you have to bind and pass local_auth_fd to the callback in the
session_manager_proxy_->CallMethod() call, but it might make ownership of it a
little easier to follow?

b) Either way, add some comments about ownership of the file descriptors, it's
challenging enough to follow when you're used to these patterns.

[Chris Masone, 2015-04-28 15:00:20]

look better?

[Chris Masone, 2015-04-28 22:21:11]

On 2015/04/28 15:00:20, Chris Masone wrote:
> look better?

The CrOS-side CL has landed, so if this LGTY given the changes, I can go ahead
and commit this, yes? I didn't need to roll system_api, because this method name
has been around for a long time, but not implemented :-)

[stevenjb, 2015-04-28 22:29:48]

Much easier to follow now, thanks.
lgtm w/ comment fix.

https://codereview.chromium.org/1111653002/diff/20001/chromeos/dbus/session_m...
File chromeos/dbus/session_manager_client.cc (right):

https://codereview.chromium.org/1111653002/diff/20001/chromeos/dbus/session_m...
chromeos/dbus/session_manager_client.cc:428: // called on completion of this
method call. This keeps the browser end
Comment truncated?

[Chris Masone, 2015-04-28 23:43:25]

https://codereview.chromium.org/1111653002/diff/20001/chromeos/dbus/session_m...
File chromeos/dbus/session_manager_client.cc (right):

https://codereview.chromium.org/1111653002/diff/20001/chromeos/dbus/session_m...
chromeos/dbus/session_manager_client.cc:428: // called on completion of this
method call. This keeps the browser end
On 2015/04/28 22:29:48, stevenjb wrote:
> Comment truncated?

Done.

[Chris Masone, 2015-04-28 23:43:30]

The CQ bit was checked by cmasone@chromium.org

[Chris Masone, 2015-04-28 23:43:30]

The patchset sent to the CQ was uploaded after l-g-t-m from
stevenjb@chromium.org
Link to the patchset: https://codereview.chromium.org/1111653002/#ps40001
(title: "fix comment")

[commit-bot: I haz the power, 2015-04-28 23:44:34]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1111653002/40001

[commit-bot: I haz the power, 2015-04-29 02:08:07]

Committed patchset #3 (id:40001)

[commit-bot: I haz the power, 2015-04-29 02:09:01]

Patchset 3 (id:??) landed as
https://crrev.com/ec53492f57498b60bbffe2aba1c80df39589925f
Cr-Commit-Position: refs/heads/master@{#327417}