Proliant ML 350 Linux sporadic freezes and questions

[Erhard Schwenk]

Hello,

We have a Compaq Proliant ML350 set up last week (delivered two weeks ago) and want to use the Machine with GNU/Debian Linux (Woody 3.0r2, samba 3.0.4 from backports.org).

Linux Kernel is self-compiled 2.4.26 (kernel-sources from woody-proposed-updates). Kernel Configuration is identical with 2.4.26-686-smp besides the thing that SmartArray Support is enabled for booting.

On first setup everything seems to run fine except three things:

- After several hours of perfect operation the whole system suddenly freezes competely. It does continue to answer ICMP Ping requests, but no login via ssh, console keeps blank, no entry in any logfile (including hpimlogview). Reboot using Magic SysRq (Alt-B) is possible, but sync fails on Device 68:03.

Googling around I found some information that this may be caused by tg3.o (the network driver). So I downloaded bcm5700 sources from the HP/Compaq Pages and changed from tg3.o to bcm5700.o. Since then (two days ago), the System has hung again two times.

- The aic7xxx Kernel-Module for the adaptec 3960D complains about an NMI at booting. Using aic7xxx_old everything seems running fine. Curiously, after rmmod aic7xxx_old it is also possible to use modprobe aic7xxx without any complaints.

- in /proc/interrupts, there are correctly shown two CPUs (single Xeon Hyperthreading System), but no Interrupts being routet to CPU1.

Does anybody know any hints where to continue looking for fixes?