[ACPI-sppt] nforce2-400 motherboard crashes with apic

Hello,

My nforce2 motherboard has what seems like a problem with APIC that I have been
unable to figure out so far.  I keep having intermittent crashes/freezes
(system just stops responding, no kernel panic, no error messages in logs). 
This is a problem that many other people seem to have: see
http://www.nvnews.net/vbulletin/showthread.php?threadid=16264l ,
http://www.nforcershq.com/forum/viewtopic.php?t=27003 , and even on this list,
http://sourceforge.net/mailarchive/forum.php?thread_id=2941359&forum_id=7803 .

I finally found a 100% reproducible way to cause a crash on my machine:
transferring data over ftp (at 6-7mb/sec) to another machine on the local net
(using a tulip pci card, not the onboard 3com) causes a crash after
transferring a few hundred megabytes to a gigabyte.  I've tested a number of
kernels and just about any possible combo of bios settings and kernel command
line, and nothing works.

Here is my hardware setup:
motherboard: FIC AU13 Pro (nforce2 ultra400 chipset, chipset revision 162,
motherboard hardware release 2, BIOS TFA42)
cpu: Athlon XP 2600+
memory: 256MB Kingston 2700 DDR in slot 1
pci cards:  ATI Rage IIC, Promise Ultra100TX2 (PDC 20268 chip), Lite-On
LNE100TX (tulip chip), SoundBlaster Live, WinTV GO (Bt878 chip)

Here is what I have tested so far:

kernels 2.4.19, 2.4.21 with bios apic/acpi turned on or off => crash
 same with any combination of apic=off, noapic, noacpi on kernel command line
 if apic is on, any access to drives on the promise controller (irq=17) times
out with "kernel: hdg: lost interrupt", and the pci ethernet card (irq=18)
times out as well

kernel 2.4.22-rc2 (includes apic-20030730-pre8 patch):
- bios apic=on acpi=on, nothing on kernel command line => crash; has millions
of spurious interrupts for irqs >=16 in /proc/interrupts
- bios apic=off acpi=off, nothing on kernel command line => crash; no spurious
interrupts (and no allocated irqs >=16)
- bios apic=off acpi=off, kernel command line nolapic => crash; no spurious
interrupts
- bios apic=on acpi=on, kernel command line nolapic => slightly more stable,
but still crashes eventually (around 10Gb transferred over network before
crash); has spurious interrupts; syslog has hundreds of copies of the following
error message: "kernel: APIC error on CPU0: 02(02)"; disk and network are about
1/2 the speed they are normally
- bios apic=off acpi=on, kernel command line nolapic => very early crash; no
spurious interrupts
- bios apic=on acpi=on, kernel command line noacpi noapic nolapic acpi=off
apic=off => crash; no spurious interrupts

For testing, I have disabled just about every onboard device (USB, IEE1394,
parallel/serial ports, floppy, onboard sound, onboard ethernet) in the BIOS. 
Btw, I have ran both memtest and cpuburn, which are completely stable.

Any suggestions on how I can get my system to run stably?  How can I help debug
this?  I'll be happy to try any kernel patch, or send you any logs or debugging
info that might help figure this out (dmesg, lspci, etc...).

Thank you,
Alex




-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0