Fix forwarding of XPC messages to launchd on 10.10 and higher.

Fix forwarding of XPC messages to launchd on 10.10 and higher.

Message forwarding was broken from sandboxed clients since launchd does not
associate them with a valid domain.

BUG=382931, 506047
R=mark@chromium.org

Committed: https://crrev.com/4a80416d53b00f1f0bbe612f51b60f726129daef
Cr-Commit-Position: refs/heads/master@{#345883}

[Mark Mentovai, 2015-08-27 03:27:01]

https://codereview.chromium.org/1312893005/diff/1/sandbox/mac/launchd_interce...
File sandbox/mac/launchd_interception_server.cc (right):

https://codereview.chromium.org/1312893005/diff/1/sandbox/mac/launchd_interce...
sandbox/mac/launchd_interception_server.cc:172: kern_return_t kr =
mach_port_mod_refs(mach_task_self(), domain_port,
I assume you’re doing this because xpc consumes a send right.

Is there anything set for domain-port to begin with? If so, do you need to
release it?

[Robert Sesek, 2015-08-27 14:59:53]

https://codereview.chromium.org/1312893005/diff/1/sandbox/mac/launchd_interce...
File sandbox/mac/launchd_interception_server.cc (right):

https://codereview.chromium.org/1312893005/diff/1/sandbox/mac/launchd_interce...
sandbox/mac/launchd_interception_server.cc:172: kern_return_t kr =
mach_port_mod_refs(mach_task_self(), domain_port,
On 2015/08/27 03:27:00, Mark Mentovai - August is over wrote:
> I assume you’re doing this because xpc consumes a send right.

I thought so and several examples (in old WebKit…) I found did so. But after
checking the disassembly and it actually does the mod_refs for you. And then I
checked WebKit ToT and found this:
https://github.com/WebKit/webkit/commit/d710fe9a03d386b0abba4b8558d68644cf19701e.
Removed the mod_refs.

> Is there anything set for domain-port to begin with? If so, do you need to
> release it?

There shouldn't be a domain-port set to begin with. If it is, it'll be from a
sandboxed client so we shouldn't trust it. But if one were present, XPC will
release it when replacing the value (so much nicer than raw Mach).

[Mark Mentovai, 2015-08-27 15:07:03]

LGTM

[Robert Sesek, 2015-08-27 15:11:39]

The CQ bit was checked by rsesek@chromium.org

[commit-bot: I haz the power, 2015-08-27 15:11:54]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1312893005/20001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1312893005/20001

[commit-bot: I haz the power, 2015-08-27 15:28:53]

Committed patchset #2 (id:20001)

[commit-bot: I haz the power, 2015-08-27 15:29:23]

Patchset 2 (id:??) landed as
https://crrev.com/4a80416d53b00f1f0bbe612f51b60f726129daef
Cr-Commit-Position: refs/heads/master@{#345883}