[alsa-devel] Question: alsa-lib snd_pcm_htimestamp inner working and precision

[gianluca a]

Hi all,

I am trying to understand the precision of htimestamp for sample captured and I am not sure if I understood well, because I have found old informations togheter with new informations and I've the doubt that something changed. I am interested in this because of a project for my degree thesis.

snd_pcm_hw_params_supports_audio_wallclock_ts returns 0 on capture, so my hardware doesn't support wallclock timestamps on capture.

I've understood that SND_PCM_TSTAMP_ENABLE is equivalent to SND_PCM_TSTAMP_MMAP, so it updates timestamp at every position update on DMA from the driver. I've read elsewhere that this timestamp is updated every period-size ( equal interrupt), is it correct?

I've digged in the code to understand well inner working of htimestamp, but I've found that snd_pcm_htimestamp in pcm.c calls:

err = pcm->fast_ops->htimestamp(pcm->fast_op_arg, avail, tstamp);

And I have not found where fast_ops is initialized, maybe in the driver? Anyway I thougth that maybe in my case the called function is snd_pcm_hw_htimestamp in pcm_hw.c :

"while (1) {
		avail1 = snd_pcm_hw_avail_update(pcm);
		if (avail1 < 0)
			return avail1;
		if (ok && (snd_pcm_uframes_t)avail1 == *avail)
			break;
		*avail = avail1;
		*tstamp = snd_pcm_hw_fast_tstamp(pcm);
		ok = 1;
	}"

that calls snd_pcm_hw_fast_tstamp, that calls finally:

"#define FAST_PCM_TSTAMP(hw) \
	((hw)->mmap_status->tstamp)
"
And finally is this the timestamp that is updated every period-size? Am I correct that in the case that hardware doesn't support wallclock timestamp is snd_pcm_hw_htimestamp in pcm_hw.c the called function?

Thanks in advance