Hi!

I experience cold freezes of my new system reproducably (can still
toggle numlock, but on alt+sysrq+t it freezes completely, watchdog_nmi=1
does not help so no printout over serial console) when I enable any 
udma mode >0 using the old or new pdc202xx driver.

However the system seems to work stably (as far I can tell) when using
the mdma0 or pio modes.

The setup is asus a7v8x with sound/ide/firewire onboard. Two 180GB WDC
WD1800JB-00DUA0 on the primary and secondary internal VIA controller and
3 drives on two pdc20268, where the third hard disk is on the secondary
controller (I explain later why!). All harddisk are jumpered to be
master's.
The harddisks on the pdc are Maxtor 4R080J0 (this one is the lonely
secondary one) Maxtor 4D080K4 (primary) Maxtor 4D080H4 (secondary)

Cables are all 80pin, the drives all udma5.

This system has 1.5G ddr333 memory (checked for 2days ~ 17passes)
harddisks were tested with the manufactures tools -> all ok. However
cannot say anything about the cables. Not sure whether they can cause a
system freeze though.

The five disks span a RAID5 (5* 80G partition).

The freeze happens immediately after the raid starts reconstruction when
udma4/5 is used. On udma2 the system freezes immediately when using
bonnie to benchmark the raid (the reconstruction of the raid is
successful though).

This all happens when booting with init=/bin/sh (so no X no nothing).

The pdc's are in slot 3 and 6 of the mainboard, so the first pdc shares
irq's with the agp slot (nvidia card/ but no nvidia module loaded) and
the second shares with Onboard SATA/1394 (no drivers loaded). All other
cards are empty.

I have no ideas how to further narrow down this problem. I am grateful
for any ideas on how to solve this.
 
If you need further infos just tell me.

Now the explanation why one harddisk is used as secondary and not
primary: If I attach the disk to the primary, then it gets correctly
detected as drive x. However when doing fdisk /dev/hdx one of the WDC
drives is shown instead of the maxtor drive.

Both problems occur in kernel version 2.4.19/20/21-pre2/21-pre3.  

Thanks for any suggestions,
Soeren.