Make IPC channels track their peer process ID

Make IPC channels track their peer process ID

BrokerDuplicateHandle() requires the target process ID on Windows. Rather than plumb this into every channel class independently, I'm adding it at the top.

TEST=IPCSyncChannelTest.Verified

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=130817

[jschuh, 2012-04-04 18:41:15]

I may need to do follow-ups to remove the (probably) unnecessary peer_pid use in
the OnChannelConnected message, but that seems better to do as a separate patch
(because I'll need to merge this to stable to fix an open bug).

[jam, 2012-04-04 19:05:00]

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:389: base::ProcessId ChannelProxy::peer_pid() const {
ChannelProxy gets called on the listener thread, but the IPC::Channel lives on
the IO thread, so it's not correct to call its member functions on the
listener's thread..

[Tom Sepez, 2012-04-04 19:07:07]

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pos...
File ipc/ipc_channel_posix.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pos...
ipc/ipc_channel_posix.h:68: base::ProcessId peer_pid() const { return peer_pid_;
}
Include what you use.

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:392: return 0;
return base::kNullProcessId constant?  Any reason to believe base::ProcessID
will be both an int/ptr as opposed to a struct, and/or 0.

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.h:171: base::ProcessId peer_pid() const;
include what you use?

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_win.h
File ipc/ipc_channel_win.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_win...
ipc/ipc_channel_win.h:36: base::ProcessId peer_pid() const { return peer_pid_; }
IWYU?

[jschuh, 2012-04-04 19:18:07]

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:389: base::ProcessId ChannelProxy::peer_pid() const {
On 2012/04/04 19:05:00, John Abd-El-Malek wrote:
> ChannelProxy gets called on the listener thread, but the IPC::Channel lives on
> the IO thread, so it's not correct to call its member functions on the
> listener's thread..

That is a subtlety I definitely did not appreciate. So, I'll just keep the
ChannelProxy::peer_pid_ as it was before and update it when the channel
connects.

[jschuh, 2012-04-04 21:05:13]

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pos...
File ipc/ipc_channel_posix.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pos...
ipc/ipc_channel_posix.h:68: base::ProcessId peer_pid() const { return peer_pid_;
}
On 2012/04/04 19:07:07, Tom Sepez wrote:
> Include what you use.

Done.

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:392: return 0;
On 2012/04/04 19:07:07, Tom Sepez wrote:
> return base::kNullProcessId constant?  Any reason to believe base::ProcessID
> will be both an int/ptr as opposed to a struct, and/or 0.

Done.

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.h:171: base::ProcessId peer_pid() const;
On 2012/04/04 19:07:07, Tom Sepez wrote:
> include what you use?

Done.

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_win.h
File ipc/ipc_channel_win.h (right):

https://chromiumcodereview.appspot.com/9968111/diff/15001/ipc/ipc_channel_win...
ipc/ipc_channel_win.h:36: base::ProcessId peer_pid() const { return peer_pid_; }
On 2012/04/04 19:07:07, Tom Sepez wrote:
> IWYU?

Done.

[jam, 2012-04-05 00:11:41]

lgtm with 1 question

https://chromiumcodereview.appspot.com/9968111/diff/17002/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/17002/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:391: return context_.get() ? context_->peer_pid_ :
base::kNullProcessId;
are you sure you need to null check context_?

[jschuh, 2012-04-05 00:23:06]

https://chromiumcodereview.appspot.com/9968111/diff/17002/ipc/ipc_channel_pro...
File ipc/ipc_channel_proxy.cc (right):

https://chromiumcodereview.appspot.com/9968111/diff/17002/ipc/ipc_channel_pro...
ipc/ipc_channel_proxy.cc:391: return context_.get() ? context_->peer_pid_ :
base::kNullProcessId;
On 2012/04/05 00:11:42, John Abd-El-Malek wrote:
> are you sure you need to null check context_?

You're right. It's always passed or created in the constructor. Since it's a
simple getter I'll just move it to an inline in the header.

[commit-bot: I haz the power, 2012-04-05 00:27:55]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/jschuh@chromium.org/9968111/22001

[commit-bot: I haz the power, 2012-04-05 02:38:35]

Change committed as 130817