I suspect this may be hardware, but somebody else may be able to
confirm or deny this.  I've just changed a motherboard (to KR7A) and
one of the disks no longer works right (apparently dies on write
operations).  The disk used to work fine, but the previous motherboard
fried in the middle of the night, so it may have taken the disk with
it...

What I get (when fsck tries to write) is:

 pciide0:1:0: lost interrupt
         type: ata tc_bcount: 8192 tc_skip: 0
 pciide0:1:0: bus-master DMA error: missing interrupr, status=0x20

Then the following error sequence, repeated continuously:

 pciide0 channel 1: reset failed for drive 0
 wd1a: DMA error writing fsbn 16 of 16-31 (wd1 bn 79; cn 0 tn 1 sn 16),
         retrying
 pciide:0:1:0: not ready, st=0xd0, err=0x00
 pciide0 channel 1: reset failed for drive 0

Once or twice I've seen:

 wd1: flush cache command timeout

It seems to do it on both IDE channels, with and without a slave
device to share with, and regardless of what I do with the BIOS.

It's a Fujitsu drive, 8G, two or three years old (wd tyoe disk).
Any suggestions of possible software or BIOS fixes for this welcome;
otherwise, I suppose I'll consign it to the junk heap.

Mark <><