How to get panic trace

[gdub]

I just performed a new Gentoo installation on
an Intel MB. The only HW I haven't used before
is the Promise SATA150 TX4 4-port SATA (no
raid) controller.

Anyway, during a prolonged data transfer over
Samba (a Windows PC backup operation), I con-
sistently get a hard panic with only the delcarative
message:

"....not syncing Aee!; killing interrupt handler!"

but no stack trace or any part of a stack trace.
There isn't one in the logs, either.

My question is: How can I get a stack trace in
this situation?

Sorry if it's a dumb question. I operate lots
of Gentoo and other Linux systems but have never
gotten a panic without a trace I could work from.

This seems like a basic skill I should have. :)

TIA, and happy new year,

-dub

[gdub]

Storytime...

This was a fun one. I was able to
get a trace by doing some heavy disk work on
the box itself. So, of course, I assumed I had
a fix on the problem (I was wrong).

I used the traces to track the problem to memory
paging. Turns out that the system didn't like the
disk segments I had chosen for swap space. I fixed
it, ran my stress test on the box several times with-
out failure. So, I can go home, right?

Wrong. I ran the Windows backup over Samba and,
boom, original kernel panic, no trace.

So, I built a 2.6.9-r13 kernel and reverted to it from
the 2.6.10 I was using. Now the long file transfer over
Samba works like a charm.

For my purpose, I am happy now, but in case someone
wants to track this down:

MB/Proc/Mem: Intel D845GEBV2/P4 2.4 GHz/1GB RAM
SATA: Promise SATA150 TX4, 3 WD 80GB SATA HDD
PATA: built-in adapter, WD 250GB HDD, WD 40GB HDD
eth0: built-in Intel, e100 driver
video: built-in Intel

devfs, no udev

EVMS vol mgmt :
md raid5 on 3 SATA 80GB drives, lvm2
boot dev and volume snapshots on 250GB PATA HDD

non-EVMS:
swap on 40GB PATA HDD

-dub