ASYNC mode...

ASYNC mode...

[Peter Bessman]

...what's up with it?  If you pass SND_PCM_ASYNC to snd_pcm_open(), it
doesn't work, which strikes me as counter-intuitive.  Actually, it
seems to be really finicky, and after playing around with pcm.c for
something on the order of 8 hours, I still don't "get it" well enough
to confidently make use of it.  Are there better examples out there,
or is this mode dysfunctional?  FWIW, I'm running the latest
PlanetCCRMA, which puts my alsa-lib at 0.9.6 (and a cvs checkout, at
that).

[pb]


-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/

Re: ASYNC mode...

[Jean-Marc Desprez]

Peter Bessman wrote:

>...what's up with it?  If you pass SND_PCM_ASYNC to snd_pcm_open(), it
>doesn't work, which strikes me as counter-intuitive.  Actually, it
>seems to be really finicky, and after playing around with pcm.c for
>something on the order of 8 hours, I still don't "get it" well enough
>to confidently make use of it.  Are there better examples out there,
>or is this mode dysfunctional?  FWIW, I'm running the latest
>PlanetCCRMA, which puts my alsa-lib at 0.9.6 (and a cvs checkout, at
>that).
>
>[pb]
>  
>
hi

I did not checked recently, but on an alsa 0.9.7 lib:
- using snd_pcm_open with SND_PCM_ASYNC opens the pcm fd with the 
O_ASYNC flag (that seems correct), and then
- setting up the callback with snd_async_add_pcm_handler performs a 
fcntl to set O_ASYNC flag in the fd (kernel) context, and supply the pid 
ot the calling process that will receive the SIGIO

the problem (bug/feature ?) within the linux kernel (at least until 
2.4.20), is that when O_ASYNC is already set in the fd kernel context, 
subsequent fcntl dealing with O_ASYNC are silently ignored, whereas 
fcntl is supposed to perform more init than open (read linux/fs/open.c 
and fcntl.c for more)

so as you say, it is conter-intuitive, but asserting the SND_PCM_ASYNC 
flag in snd_pcm_open forbids snd_async_add_pcm_handler to work!

I already suggested on this list that SND_PCM_ASYNC be only kept to 
authorise the use of snd_async_add_pcm_handler, but as far as quite few 
people are interested in async mode, I believe that priorities are 
elsewhere and it had no effect. Beside, the async mode works fine once 
this init problem is solved.

Regards


>
>-------------------------------------------------------
>This SF.net email is sponsored by: SF.net Giveback Program.
>Does SourceForge.net help you be more productive?  Does it
>help you create better code?  SHARE THE LOVE, and help us help
>YOU!  Click Here: http://sourceforge.net/donate/
>_______________________________________________
>Alsa-devel mailing list
>Alsa-devel@lists.sourceforge.net
>https://lists.sourceforge.net/lists/listinfo/alsa-devel
>
>  
>




-------------------------------------------------------
This SF.net email is sponsored by: SF.net Giveback Program.
Does SourceForge.net help you be more productive?  Does it
help you create better code?  SHARE THE LOVE, and help us help
YOU!  Click Here: http://sourceforge.net/donate/