Re: Backported sbxfi driver, possible fix

[Takashi Iwai <tiwai@suse.de>]

At Sat, 25 Oct 2008 21:40:40 +0100,
Jason Harvey wrote:
> 
> Takashi Iwai wrote:
> > At Sat, 25 Oct 2008 21:42:55 +0200,
> > Thomas Scheunemann wrote:
> >   
> >>> My solution is to use roundup_pow_of_two() instead of the own
> >>> funciton.  This should work better in general.
> >>>       
> >> I certainly agree that it is better to use an already defined function
> >> instead of reinventing the wheel.
> >>
> >>     
> >>> Anyway, I updated the repo (and rebased, sorry), updated the snapshot,
> >>> too.
> >>>       
> >> But this new version causes an immediate reboot on my machine even with
> >> the previously working speaker-test. If I interpret linux/log2.h correctly
> >> the desired function should be order_base_2 instead of roundup_pow_of_two.
> >> At least it works for me after that exchange.
> >>     
> >
> > Oh my.  You're right, it must be order_base_2(), of course.
> > I was too hurry to fix a bug after the server crash :)
> >
> > Now fixed and updated.  Thanks!
> >   
> Latest unstable (25Oct 19:57) works from command line with mplayer, 
> without the proc oss fix.
> aplay still works but only with dmix.

Do you have no sound quality problem with dmix like other people?

> Now also almost working in gnome with pulse, sound is recognisable but 
> with lots of interference/corruption.

What about base_rate=48000 option?

> Machine did not crash at any point!

Heh, finally.

> This was running on 2.6.26... is this a problem? SUPPORTED_KERNELS says 
> 2.6.25 or earlier?

You can ignore what stated in SUPPORTED_KERNELS.
It's just an excuse not to support vendor kernels.  Distro vendor
kernels are often way too modified to support properly by the
out-of-tree driver.

FWIW, the current alsa-driver (and unstable) snapshot should work fine
with up to 2.6.28-rc1 kernel.  sbxfi is marked to be available on
2.6.25 or later kernels.  It won't be built for earlier kernels.


thanks,

Takashi