Re: /dev/dsp got more sensitive about writes?

On 09/30/16 01:01, Benjamin Kaduk wrote:
> Hi all,
>
> I've updated from r302211 (June 26) to r306427 (yesterday), and the
> behavior of audio/sox (in its play(1) alias) seems to have changed.  I
> used to be able to suspend play(1) for an arbitrary period of time and
> have the output continue when it was foregrounded again.
>
> Now, it seems that I can only suspend it for 4-5 seconds and retain
> playback; for longer sleeps, I get
>
> [1]+  Stopped                 play 01\ Mary\ Hynes.wav
> [root_at_glossolalia ...]# sleep 5; fg
> play 01\ Mary\ Hynes.wav
> play FAIL sox: `/dev/dsp' error writing output file: Invalid argument
> Done.
>
>
> Wrapping play(1) in truss seems to change things, though; in that case, I
> can sleep for at least two minutes and still recover playback.  So that
> makes it a bit hard to see what's going on.
>
> I will note that I did not upgrade ports/packages after the base upgrade,
> so the sox binary in use is potentially somewhat stale.
>
> Does that ring any bells for anyone?

Hi,

Maybe you could ktrace the program. SOX has a bad behaviour that it 
writes one and one sample through GIO when using /dev/dsp. I once made a 
patch for it, but it was never upstreamed.

--HPS