IDE problems

[Justin Moore]

> > > We are currently running 2.2.19-6.2.1 kernel as it came from Red Hat.
> >
> >    There are some known problems with the ServerWorks LE chipset that
> > cause filesystem corruption on some recent kernels.  The bottom line is
> > that it appears to be a BIOS bug, and that disabling DMA on your IDE
> > chains will prevent that.  A chain of BIOS updates may or may not solve
> > your problem, but I haven't seen anything too encouraging yet.  A detailed
> > bug report can be found at
> >
> > http://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=38429
> >
> > Good luck.
> > -jdm
> >
>
> The bug report in question is describing bugs in the 2.4 kernel.

   The next-to-last comment describes your problem with a Tyan board that
uses the same chipset, running the 2.2.19 kernel from RedHat.

> Has anyone seen the problems like that in 2.2 kernels besides me?--
> Also, turning DMA off is a non-starter for us...we depend on DMA I/O
> for the application we are running.
>
> The vendor has tried three types of hard disk on the nodes so far...
> Maxtor, Seagate (which didn't show hda errors but did
> show massive filesystem corruption as described in the bug report above)
> and now Western Digital, where there are filesystem
> errors that hang the machine but the system usually comes back fine
> after reboot and fsck.
> They are now trying to go to IBM drives (which are working fine
> as data drives on these systems, albeit on the secondary IDE bus.)

   Short of getting PCI IDE controllers I'm not quite sure what to
suggest.  I'm almost positive[1] that it really is a BIOS problem you're
running into, and that's why your problem is hard-drive independent.

-jdm

[1] Positive (adj): Mistaken at the top of one's voice.
    - "The Devil's Dictionary", Ambrose Bierce

Department of Computer Science, Duke University, Durham, NC 27708-0129
Email:  justin at cs.duke.edu