Mojo: Implement local passing of MessagePipes.

Mojo: Implement local passing of MessagePipes.

I.e., allow handles to MessagePipes to be transferred across
MessagePipes whose ends are both local (i.e., in-process).

There's one basic test of this functionality in this change, but much
more is needed.

Also, in theory, a handle to a remote (proxied) MessagePipe can be
transferred across a local MessagePipe, but this is not tested.

R=davemoore@chromium.org, davemoore

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=234863

[DaveMoore, 2013-11-13 02:54:02]

lgtm

https://codereview.chromium.org/66963005/diff/1/mojo/system/dispatcher.cc
File mojo/system/dispatcher.cc (right):

https://codereview.chromium.org/66963005/diff/1/mojo/system/dispatcher.cc#new...
mojo/system/dispatcher.cc:39: uint32_t* num_dispatchers,
Nit: Why do you need an output var for num_dispatchers? Isn't dispatchers
enough?

[viettrungluu, 2013-11-13 17:31:57]

On 2013/11/13 02:54:02, DaveMoore wrote:
> lgtm
> 
> https://codereview.chromium.org/66963005/diff/1/mojo/system/dispatcher.cc
> File mojo/system/dispatcher.cc (right):
> 
>
https://codereview.chromium.org/66963005/diff/1/mojo/system/dispatcher.cc#new...
> mojo/system/dispatcher.cc:39: uint32_t* num_dispatchers,
> Nit: Why do you need an output var for num_dispatchers? Isn't dispatchers
> enough?

No, because if the max is too small, it'll need to tell the caller how many are
needed.

[viettrungluu, 2013-11-13 18:53:18]

Committed patchset #3 manually as r234863 (presubmit successful).