[pulseaudio-discuss] alsa sink latency - how to account for startup delay

[Georg Chini]

On 23.03.2016 11:04, Tanu Kaskinen wrote:
> On Tue, 2016-03-22 at 12:57 +0100, Georg Chini wrote:
>> On 22.03.2016 12:51, Tanu Kaskinen wrote:
>>> On Tue, 2016-03-22 at 12:33 +0100, Georg Chini wrote:
>>>> On 22.03.2016 12:20, Tanu Kaskinen wrote:
>>>>> On Tue, 2016-03-22 at 10:11 +0100, Georg Chini wrote:
>>>>>> Hi,
>>>>>>
>>>>>> when a sink is started, there is some delay before the first sample is
>>>>>> really played.
>>>>>> This delay is a constant part of the sink latency that will be always
>>>>>> present, so the
>>>>>> minimum sink latency cannot go below that start delay.
>>>>>> Would it be acceptable to adjust the latency range for the device after
>>>>>> each unsuspend
>>>>>> to reflect that?
>>>>>> USB devices (those I have access to) for example have a startup delay in
>>>>>> the range of
>>>>>> 10ms, but have a latency range that starts at 0.5ms which does not make
>>>>>> a lot of sense
>>>>>> in my opinion.
>>>>> I don't understand why the startup delay would limit the minimum
>>>>> latency once the stream is flowing. Imagine a sound card that is
>>>>> powered by a nuclear power plant. I don't know how long it takes to
>>>>> start a nuclear power plant, but let's say it takes a couple of days.
>>>>> Now the sound card startup delay is a couple of days, but there's no
>>>>> reason that the audio latency has to be a couple of days once the power
>>>>> plant is running. Where would all that audio be buffered anyway?
>>>>>
>>>> Hi Tanu,
>>>>
>>>> you are wrong.
>>> I don't believe you :)
>> Look at the code of alsa-sink. It never drops samples. The only way to
>> compensate
>> for the startup delay would be to drop audio as long as the sink is not
>> yet playing,
>> but that is not done. I could try to implement that however and then you
>> would be
>> right, but with the current code at least for the alsa-sink the startup
>> delay will persist.
> The sink isn't responsible for dropping samples in any case. The
> connection between the start delay and the runtime latency just doesn't
> exist at the sink level.
Again, sorry, but you are wrong. The startup delay does not vanish
magically. I can only point you to the code. I have been working with it
for a couple of month now and I know what I am saying. What happens
for USB devices with timer based scheduling is the following:

1) First audio data is written to the card
2) snd_pcm_start() is called
3) More data is written to the card
4) The reported delay of the card goes up exactly by the amount that was 
written
5) This repeats a couple of times
6) The card starts playing after 10 -15ms
7) Now there is some difference between the delay and the amount of data 
written

This means that the card is forced to buffer audio until it has started 
up. This also
means that module-loopback can't do anything about it - see below.
>   It's module-loopback that introduces that
> connection, and if samples are to be dropped, module-loopback is where
> that should happen. And indeed, module-loopback should drop samples
> when waiting for the sink to start - I've seen a bluetooth situation
> where starting the sink took something like 8 seconds to start, and
> module-loopback kept buffering all that time, which did not result in a
> pleasant phone call experience :)
Well, the version of module-loopback I am using does exactly that, but
the startup delay is not reflected there for alsa-devices. The sink-input
calls pop immediately because the alsa sink code assumes that the sink
is running after calling snd_pcm_start(), but that is just not true. So
module-loopback cannot see the start delay in that case, this is why
the resulting latency is offset by that amount.

>
> By the way, your mails have bad word wrapping. See how it looks:
> https://lists.freedesktop.org/archives/pulseaudio-discuss/2016-March/025830.html
>
Mh, don't know why this happens. Maybe I should not use CR.