How to interperet snd_pcm_htimestamp available frames?

How to interperet snd_pcm_htimestamp available frames?

[Eric Buddington]

I am recording from multiple (cheap USB) mics, and trying to
synchronize them. I am assuming that there will be a common minimum
latency (reflecting an uncontested USB bus), with occasional spikes up
- but not down.

In my loop, I am calling snd_pcm_mmap_begin(), snd_pcm_htimestamp(),
and snd__mmap_commit() in that order.

My question is about the 'avail' value returned from htimestamp. It is
documented as "Number of available frames when timestamp was grabbed".
Does this mean frames since the last frame consumed with
_mmap_commit()?

Is there a possibility of there being no timestamp, or one where
'avail' refers to some older state?

Thanks for any insight.

-Eric

Re: How to interperet snd_pcm_htimestamp available frames?

[Takashi Iwai]

On Sun, 22 Nov 2020 17:03:35 +0100,
Eric Buddington wrote:
> 
> I am recording from multiple (cheap USB) mics, and trying to
> synchronize them. I am assuming that there will be a common minimum
> latency (reflecting an uncontested USB bus), with occasional spikes up
> - but not down.
> 
> In my loop, I am calling snd_pcm_mmap_begin(), snd_pcm_htimestamp(),
> and snd__mmap_commit() in that order.
> 
> My question is about the 'avail' value returned from htimestamp. It is
> documented as "Number of available frames when timestamp was grabbed".
> Does this mean frames since the last frame consumed with
> _mmap_commit()?

It's the value that is equivalent with the one in snd_pcm_status and
co: the available frames in the ring buffer.

> Is there a possibility of there being no timestamp, or one where
> 'avail' refers to some older state?

It was a quite long time ago that I tested it, but I don't think there
has been any significant changes in this regard.
I guess Pierre can have better insight.


Takashi