Reland 7976047 to fix the alsa output scheduling for low latency audio.

Reland 7976047 to fix the alsa output scheduling for low latency audio.
This patch is to allow using 10ms buffersize or above, and it also optimize the WebAudio performance by using 5ms as the minimum interview between two OnMoreData() calls.

Bug=None
Test=media_unittests, manully tests:
WebAudio: http://chromium.googlecode.com/svn/trunk/samples/audio/shiny-drum-machine.html (change audio_util.cc to return 512 as buffersize)
WebRTC: Internal test app
Html5: http://www.youtube.com/watch?v=siOHh0uzcuY
Chromebook: http://www.youtube.com/watch?v=-9AtssRoc2Q&feature=related

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=109852

[no longer working on chromium, 2011-11-10 13:39:06]

Hello,

This CL is to reland CL7976047 to make the alsa output scheduling working for
low latency audio.
It is tested with html5, WebAudio(512 as buffersize), WebRTC(10ms buffersize) in
the browser; and also done quite a lot tests with html5 in chomebooks (Alex and
Acer AC700).

[scherkus (not reviewing), 2011-11-10 23:29:54]

LGTM from an OWNERS perspective but it'd be good to have enal/rtoy inspect +
test as well

[no longer working on chromium, 2011-11-11 08:41:19]

Thanks Andrew,

Eugene and Raymond, any comment or question for the patch?

BR,
/SX

On Fri, Nov 11, 2011 at 12:29 AM, <scherkus@chromium.org> wrote:

> LGTM from an OWNERS perspective but it'd be good to have enal/rtoy inspect
> +
> test as well
>
>
http://codereview.chromium.**org/8465018/<http://codereview.chromium.org/8465...
>

[enal1, 2011-11-11 17:33:08]

On 2011/11/11 08:41:19, xians1 wrote:
> Thanks Andrew,
> 
> Eugene and Raymond, any comment or question for the patch?
> 
> BR,
> /SX
> 
> On Fri, Nov 11, 2011 at 12:29 AM, <mailto:scherkus@chromium.org> wrote:
> 
> > LGTM from an OWNERS perspective but it'd be good to have enal/rtoy inspect
> > +
> > test as well
> >
> >
>
http://codereview.chromium.**org/8465018/%3Chttp://codereview.chromium.org/84...>
> >

lgtm

(Previous one also looked good, though...)

Previous one included changes to unit tests, while this one does not. Why?

[Raymond Toy (Google), 2011-11-11 22:01:25]

Other than the one question, lgtm.

http://codereview.chromium.org/8465018/diff/1/media/audio/linux/alsa_output.cc
File media/audio/linux/alsa_output.cc (right):

http://codereview.chromium.org/8465018/diff/1/media/audio/linux/alsa_output.c...
media/audio/linux/alsa_output.cc:354: alsa_buffer_frames_ = frames_per_packet_ *
2;
Why is this twice now when it used to be the same?  Is it because our time
interval is no 5 ms instead of 10?

[no longer working on chromium, 2011-11-12 18:31:22]

On 2011/11/11 22:01:25, rtoy wrote:
> Other than the one question, lgtm.
> 
> http://codereview.chromium.org/8465018/diff/1/media/audio/linux/alsa_output.cc
> File media/audio/linux/alsa_output.cc (right):
> 
>
http://codereview.chromium.org/8465018/diff/1/media/audio/linux/alsa_output.c...
> media/audio/linux/alsa_output.cc:354: alsa_buffer_frames_ = frames_per_packet_
*
> 2;
> Why is this twice now when it used to be the same?  Is it because our time
> interval is no 5 ms instead of 10?

The previous value is not correct. It is related to how we setup the
configuration for alsa:
OpenPlaybackDevice(wrapper_, device_name_.c_str(),channels_, sample_rate_,
pcm_format_,latency_micros_);
Where latency_micros_ is deciding the buffersize.
And in our code it is
latency_micros_(std::max(AlsaPcmOutputStream::kMinLatencyMicros,
micros_per_packet_ * 2))
So the buffer size should be twice of packet size.

[commit-bot: I haz the power, 2011-11-12 18:32:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/xians@chromium.org/8465018/1

[no longer working on chromium, 2011-11-12 18:43:48]

On 2011/11/11 17:33:08, enal1 wrote:
> On 2011/11/11 08:41:19, xians1 wrote:
> > Thanks Andrew,
> > 
> > Eugene and Raymond, any comment or question for the patch?
> > 
> > BR,
> > /SX
> > 
> > On Fri, Nov 11, 2011 at 12:29 AM, <mailto:scherkus@chromium.org> wrote:
> > 
> > > LGTM from an OWNERS perspective but it'd be good to have enal/rtoy inspect
> > > +
> > > test as well
> > >
> > >
> >
>
http://codereview.chromium.**org/8465018/%253Chttp://codereview.chromium.org/...>
> > >
> 
> lgtm
> 
> (Previous one also looked good, though...)
> 
> Previous one included changes to unit tests, while this one does not. Why?

The previous CL changes the behavior a bit for html5, while this one I try to
keep it untouch and only optimize for low-latency audio.

[commit-bot: I haz the power, 2011-11-12 19:31:49]

Try job failure for 8465018-1 (retry) on linux_clang for step "compile" (clobber
build).
It's a second try, previously, step "compile" failed.
http://build.chromium.org/p/tryserver.chromium/buildstatus?builder=linux_clan...

[commit-bot: I haz the power, 2011-11-14 08:18:11]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/xians@chromium.org/8465018/12001

[commit-bot: I haz the power, 2011-11-14 09:47:46]

Change committed as 109852