On Sun, Nov 06, 2005 at 11:03:25AM -0800, John Klos wrote:
> Hi,
> 
> I don't have more information at this point, but hardware which has been 
> running NetBSD 2.0 and 2.1 for many weeks crashes (locks hard, can't get 
> into debugger) with NetBSD 3 and current.
> 
> I will try to test this with another machine so I can isolate the hardware 
> which is causing the problem, but this is happening on a 100 MHz AGP G4 
> with a 2 GHz 7447a, a 64 bit PCI gigabit wm* card, and a 64 bit PCI ATTO 
> U3LD Ultra-160 esiop* card. When I boot either NetBSD 3 or current, it'll 
> lock within minutes if I do any disk activity (even if I just let it try to 
> finish the fsck). I unplugged the SCSI bus, and it seemed to last longer, 
> but I couldn't test it with no SCSI for very long, so that's definitely not 
> conclusive.
> 
> Does anyone have any clue as to why 2.0 and 2.1 will work for weeks, and 
> even complete a pkgsrc bulk build, but NetBSD 3 and current can't run for 
> minutes on the same hardware? dmesg from NetBSD 2.1 follows.

I note that the esiop is the only piece of hardware using a shared IRQ.
Maybe shared interrupts cause problems ? I guess you can test this
by replacing esiop* with esiop0 in your kernel config file

-- 
Manuel Bouyer <bouyer@antioche.eu.org>
     NetBSD: 26 ans d'experience feront toujours la difference
--