Fix AudioPipeReader to use default pipe buffer size.

Fix AudioPipeReader to use default pipe buffer size.

Pulseaudio uses default pipe buffer size to calculate pipe
latency, ignoring the actual buffer size set by the reading
side. AudioPipeReader was setting buffer size that's about 2
times bigger than default. As result it reports incorrect
latency and some clients did not work properly (particularly
AudioContext API implementation in chrome). Now
AudioPipeReader will be using default buffer size determined
by the OS, which matches pulseaudio's behavior.

BUG=442455
R=lambroslambrou@chromium.org

Committed: https://chromium.googlesource.com/chromium/src/+/cbcd2ec76ca61e9f9e3d529f36011346d4e673da

[Sergey Ulanov, 2014-12-23 22:16:16]

Patchset #1 (id:1) has been deleted

[Sergey Ulanov, 2014-12-23 22:22:21]

sergeyu@chromium.org changed reviewers:
+ lambroslambrou@chromium.org

[Lambros, 2014-12-23 23:07:18]

lgtm

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
File remoting/host/linux/audio_pipe_reader.cc (right):

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
remoting/host/linux/audio_pipe_reader.cc:130: pipe_buffer_size_ =
fpathconf(pipe_.GetPlatformFile(), _PC_PIPE_BUF);
I don't think this call can fail, can it?

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
remoting/host/linux/audio_pipe_reader.cc:132: kSampleBytesPerSecond / 2;
I'm curious to know why we divide by 2? Is it simply so that we capture twice as
often as we really need?

[Sergey Ulanov, 2014-12-24 00:01:29]

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
File remoting/host/linux/audio_pipe_reader.cc (right):

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
remoting/host/linux/audio_pipe_reader.cc:130: pipe_buffer_size_ =
fpathconf(pipe_.GetPlatformFile(), _PC_PIPE_BUF);
On 2014/12/23 23:07:18, Lambros wrote:
> I don't think this call can fail, can it?

it shouldn't, but it's still better to check if it does. Done.

https://codereview.chromium.org/824943002/diff/20001/remoting/host/linux/audi...
remoting/host/linux/audio_pipe_reader.cc:132: kSampleBytesPerSecond / 2;
On 2014/12/23 23:07:18, Lambros wrote:
> I'm curious to know why we divide by 2? Is it simply so that we capture twice
as
> often as we really need?

Yes. We need to capture more often than the buffer length. E.g. if the buffer
can fit 20ms of audio samples then we need to capture a more often than every
20ms. If we try scheduling the timer every 20ms, then, because the timer is not
precise, sometimes more than 20ms will pass between between DoCapture() calls
and as result it won't be reading enough audio samples, starving the stream on
the client.

[Sergey Ulanov, 2014-12-24 00:01:34]

The CQ bit was checked by sergeyu@chromium.org

[commit-bot: I haz the power, 2014-12-24 00:02:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/824943002/40001

[commit-bot: I haz the power, 2014-12-24 00:11:55]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-12-24 00:11:56]

Try jobs failed on following builders:
  android_chromium_gn_compile_rel on tryserver.chromium.linux
(http://build.chromium.org/p/tryserver.chromium.linux/builders/android_chromiu...)

[Sergey Ulanov, 2014-12-24 00:18:17]

The CQ bit was checked by sergeyu@chromium.org

[commit-bot: I haz the power, 2014-12-24 00:18:46]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/824943002/40001

[commit-bot: I haz the power, 2014-12-24 00:27:44]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2014-12-24 00:27:46]

Try jobs failed on following builders:
  android_chromium_gn_compile_rel on tryserver.chromium.linux
(http://build.chromium.org/p/tryserver.chromium.linux/builders/android_chromiu...)

[commit-bot: I haz the power, 2014-12-24 00:45:02]

Patchset 2 (id:??) landed as
https://crrev.com/cbcd2ec76ca61e9f9e3d529f36011346d4e673da
Cr-Commit-Position: refs/heads/master@{#309593}

[Sergey Ulanov, 2014-12-24 00:45:07]

Committed patchset #2 (id:40001) manually as
cbcd2ec76ca61e9f9e3d529f36011346d4e673da (presubmit successful).