On Mon, 25 Apr 2016 16:50:09 +0200, Enrico Mioso wrote:

Thank you a lot Takashi.

there are two problems: 1 - The warning doesn't manifest always: in some cases (instances of mplayer?) this warning manifests itself from time to time, some minutes.

Yes, this is likely some XRUN by either hardware boundary condition or a system load.

2 - I think the audio stops a little bit (less of a second I think), then things go ahead. Still I don't think any message is printed by alsa libraries. I don't know if this per se indicates there is no overrun, but still my sensation is this. But this system isn't exactly "extremely fast": so the audio might stop even for other reasons. I'll recompile the kernel to apply this patch and let you know. The timing will be slow, but I will try.

My patch won't improve in this regard, I'm afraid. The PCM core already corrects to the position 0 if such an invalid position is detected, and the patch would do it just implicitly. But, it's still interesting to see whether it has any impact on such behavior.

Takashi