Re: exclusive sleep mutex pcm0 (sound cdev) r = 0

On Feb 24, Don Lewis wrote:
> On 24 Feb, Jun Kuriyama wrote:
> > 
> > I got this warning on my current box:
> > 
> > malloc() of "256" with the following non-sleepable locks held:
> > exclusive sleep mutex pcm0 (sound cdev) r = 0 (0xc73e63c0) locked _at_ dev/sound/pcm/dsp.c:213
> > 
> > I think there is no need to wait in vchan_create().  If kernel does
> > not have resources to return small 256 bytes memory for sound device,
> > we are not in the situation to care whether sound is played correctly
> > or not...
> 
> There are a bunch more problems with the "sound cdev" mutex, many of
> which can't be fixed by this type of change unless having sporadic sound
> system failures due to temporary malloc() failures is acceptable.
> 
> I currently have some patches for the channel mutexes that are
> undergoing testing.  Once these have been committed, I'll take another
> look at the "sound cdev" mutex.  My current thought is to convert most
> uses of it to a lockmgr lock.

	Don,
	I've only recently come up-to date with the freebsd-life and I
plan to test your outstanding patches in the next day or so.
	I think before you take another you do another iteration of
bug fixing we should re-consider vchans.  (Although I think it may be
wise at looking at fixing the buffer size and not allowing the
re-allocation).  Are you interested at looking at that first?

	--Mat
-- 
	In general, a standard is very useful, whether it's de facto
	or du jour.
	- Microsoft's Greg Sullivan
	as misquoted by News.Com