Re: Help needed - NetBSD 5.0 - generic Apr-26 fails in fsck ..

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Fri, Jun 05, 2009 at 11:17:30PM +1000, Thilo Jeremias wrote:
> Good try (new bios) but didn't help, after N+1 fsck's on a 1tb raid, I
> decided to downgrade.... (sigh)
> 
> but now I have an empty playbox which still crashes!!!!
> 
> I am able to isolate the problem only non-raw accesses of >1m blocksize
> causes the system to crash:
> 
> dd if=/dev/wd0d of=/dev/null bs=1040k count=100    --- DOES NOT CORE
> dd if=/dev/wd0d of=/dev/null bs=1200k  count=100  --- DOES CORE !!!!
> dd if=/dev/rwd0d of=/dev/null bs=1200k  count=100  --- is fine too.   (
> Raw device )
> dd if=/dev/wd1d of=/dev/null bs=2000k count=100 --- is fine ( P-ATA drive )
> wd0 is a 1tb sata drive (ffs - no raid)
> 
> Any expert got any ideas?
> (I used the 5_0 iso from ftp.netbsd)
> I have a core dump of the crashed system (ask and I put it somewhere) -
> below is a dmesg before the crash.
> it very reliable produces "ohci4: 8 scheduling overruns" messages before
> crashing.
> 
> So I disagree with the flaky memory timing hypothesis and I think it is
> a bug in the SATA/kernel  area .

I would more think about a bug in the ohci driver, or the chipset.
Does the system still crash if you disable USB in BIOS, or the
ohci/ehci drivers in the kernel ?

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--