Mojo: Properly cancel further read on all read failures in RawChannelPosix.

Mojo: Properly cancel further read on all read failures in RawChannelPosix.

(We weren't do so properly in the "too many handles" case.)

R=yzshen@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=289310

[yzshen1, 2014-08-12 21:21:17]

LGTM

https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
File mojo/system/raw_channel_posix.cc (right):

https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
mojo/system/raw_channel_posix.cc:179: IOResult rv = ReadImpl(bytes_read);
optional: I think for windows code, sometimes we use
do {
  ...
  break;
  ...
  return;  // On success
} while (false);
//error handling

[viettrungluu, 2014-08-12 21:40:45]

On 2014/08/12 21:21:17, yzshen1 wrote:
> LGTM
> 
>
https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
> File mojo/system/raw_channel_posix.cc (right):
> 
>
https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
> mojo/system/raw_channel_posix.cc:179: IOResult rv = ReadImpl(bytes_read);
> optional: I think for windows code, sometimes we use
> do {
>   ...
>   break;
>   ...
>   return;  // On success
> } while (false);
> //error handling

Yeah, I thought about doing that. It has the advantage of keeping all the code
in one place, but the disadvantage of being an odd construct.

[yzshen1, 2014-08-12 22:15:49]

On 2014/08/12 21:40:45, viettrungluu wrote:
> On 2014/08/12 21:21:17, yzshen1 wrote:
> > LGTM
> > 
> >
>
https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
> > File mojo/system/raw_channel_posix.cc (right):
> > 
> >
>
https://codereview.chromium.org/463883004/diff/1/mojo/system/raw_channel_posi...
> > mojo/system/raw_channel_posix.cc:179: IOResult rv = ReadImpl(bytes_read);
> > optional: I think for windows code, sometimes we use
> > do {
> >   ...
> >   break;
> >   ...
> >   return;  // On success
> > } while (false);
> > //error handling
> 
> Yeah, I thought about doing that. It has the advantage of keeping all the code
> in one place, but the disadvantage of being an odd construct.

Agreed.
Still LGTM

[viettrungluu, 2014-08-13 16:22:36]

Committed patchset #1 manually as 289310 (presubmit successful).