ChannelMojo: Make MOJO_RESULT_FAILED_PRECONDITION an error.

ChannelMojo: Make MOJO_RESULT_FAILED_PRECONDITION an error.

ChannelMojo used to ignore this error, which signals a closed peer,
and silently closes the channel. This CL let it resuls an error
because the old behavior seems incompatible to traditional
IPC::Channel implementations.

To sheriffs: This is likely to be what you are looking for
if you're chasing mysterious hangs or crashes with no stack traces.
The change is subtle, but possibly causes such kind of errors.

R=viettrungluu@chromium.org,agl@chromium.org
BUG=488291
Committed: https://crrev.com/0b35325d7b5e25f4942f79e3f4d6269c36f1c2e0
Cr-Commit-Position: refs/heads/master@{#333995}

[Hajime Morrita, 2015-06-10 23:09:02]

PTAL? Subtle change, bots say this is OK.

[Hajime Morrita, 2015-06-10 23:25:12]

On 2015/06/10 23:09:02, Hajime Morrita wrote:
> PTAL? Subtle change, bots say this is OK.

I didn't remember why I added that guard.
It's pretty early on the development, so it's likely that
I just didn't understand the semantics at that time.

[agl, 2015-06-11 01:51:06]

LGTM (mostly I'm just trusting that you know this code better than anyone at
this point.)

[Hajime Morrita, 2015-06-11 17:43:05]

On 2015/06/11 01:51:06, agl wrote:
> LGTM (mostly I'm just trusting that you know this code better than anyone at
> this point.)
Thanks! Landing...

[Hajime Morrita, 2015-06-11 17:43:11]

The CQ bit was checked by morrita@chromium.org

[commit-bot: I haz the power, 2015-06-11 17:43:39]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1175083002/1

[commit-bot: I haz the power, 2015-06-11 18:39:08]

Committed patchset #1 (id:1)

[commit-bot: I haz the power, 2015-06-11 18:41:23]

Patchset 1 (id:??) landed as
https://crrev.com/0b35325d7b5e25f4942f79e3f4d6269c36f1c2e0
Cr-Commit-Position: refs/heads/master@{#333995}