Re: ext2_free_block already cleared for block xxxx
Dan Hollis (goemon@sasami.anime.net)
Tue, 8 Jul 1997 01:09:28 -0700 (PDT)
On Mon, 7 Jul 1997, Jon Lewis wrote:
> On Mon, 7 Jul 1997, Theodore Y. Ts'o wrote:
> > I believe that most of these cases (where the kernel reports ext2
> > problems, but rebooting fixings it and e2fsck doesn't detect any errors)
> > are caused by hardware problems --- most likely bits getting flipped as
> > they get transfered to memory, although possibly they could be a rare
> > memory problem. I'd take a very careful look at the controllers and the
> > disk cabling.
> I'm not sure how much I believe in that theory. My news server (where I
> see this the most) uses granite digital internal ribbon cables and active
> termination on the drive ends of the cables. Next time I have the system
> down, I'm going to pull the passive terminators from my NCR 810 cards and
> put on external active terminators.
If you're using a recent ncr driver, try passing "ncr53c8xx=burst:0" to
the kernel at boot time, and see if that doesn't get rid of your errors.
At least, that helped for me. My kernel line reads:
ncr53c8xx=sync:10,tags:4,burst:0,verb:2
-Dan
[If it does fix your problems, then I've got a pretty good argument for
Gerard that something is up with PCI timing.]