On Fri, 20 Dec 1996 09:34:46 -0800 (4 @ ) 
 Christopher Wiles <wileyc@fsr.net> wrote:

 > With NetBSD-1.2 (stock) installed on a Sun 3/160 with a Xylogics 451
 > controller, I see the following error logged:
 > 
 > Nov 24 20:43:10 mikkabi /netbsd: xyc0: watchdog timeout
 > 
 > ... over and over again.  Happens during _any_ disk access, though it's
 > more frequent with heavy access.

Chuck Cranor (on the CC list) would be the one to ask, but...

...I seem to remember his mentioning that there was a slight bug
in the xyc/xy code that caused the controller to wedge.  He added the
timeout code to work around it, since he never found the bug.  IIRC,
it's actually not getting an interrupt that it expects.

 > Usually it claims that a soft reset cleared up the problem, but if the
 > system was in the middle of a write it gives up -- with resultant data
 > lossage.

Ick.  It should probably at _least_ retry the request... I haven't looked
at that code in a long time... I don't know how hard that would be.

 > What's really nasty is when this happens while it's writing to swap.  The
 > machine became so unstable that I've been resorting to netbooting it ...
 > which isn't a good thing from my employer's point of view.
 > 
 > Ideas?

I don't really have any... but Chuck might :-)

Jason R. Thorpe                                       thorpej@nas.nasa.gov
NASA Ames Research Center                               Home: 408.866.1912
NAS: M/S 258-6                                          Work: 415.604.0935
Moffett Field, CA 94035                                Pager: 415.428.6939