On Tue, Apr 04, 2000 at 05:40:46PM +0200, Manuel Bouyer wrote:
> On Tue, Apr 04, 2000 at 03:09:17PM +0200, Manuel Bouyer wrote:
> > Hi,
> > Ok, I finaly got my AudioPCI 97 working (it seems it doen't like to share IRQ
> > with my SMC ethernet board). However I noticed short when changing the
> > volume. A good way for me to make it show up is to start xmix and play with
> > the volumes bars - occasionally the mouse will hang.
> > Did someone else notice this (also note that xmix complains about
> > 'Invalid recording source'. I'm not sure if this is related.
> > 
> > Can someone else reproduce the problem or is it just me (I'm running current
> > cvs up'ed today) ?
> 
> Also, I just noticed (running systat vmstat & realplayer at the same time)
> That the kernel cpu usage grows when the hang occurs. Really looks like a
> loop somewhere ...

Bill and I saw this when we were triggering system hangs using IDE I/O.  But
I think one *other* thing we discovered is that something's wrong with
kernel profiling -- we couldn't find any routine with enough "self" time to
exdplain this, and further investigation made me suspect that the times
reported by gprof were in at least some cases bogus.  Also, since the ELF
switch for x86 kernels, gprof seems unable to exclude routines...

-- 
Thor Lancelot Simon	                                      tls@rek.tjls.com
	"And where do all these highways go, now that we are free?"