System crash & reboot after 'modprobe cb_pcidas'

Hello fellow Comedians,

I have run into a showstopper that I am not sure how to address, being relatively new to the real-time Linux game.  I began with a vanilla 2.6.7 kernel, upon which I installed rtai-3.1-testr4.  No problem, everything worked smoothly.  Next, I downloaded the Sept 28 CVS of comedi and comedilib.  Neither of these installs worked, until I figured that what I ought to do was download and untar the most recent stable versions (0.7.68 and 0.7.21, respectively), then untar the CVSs into those directories, overwriting the common files.  This worked very well, everything compiled, made, and installed great.  So, I rebooted, and began playing with the new comedi addition.  Right off the bat, cat /proc/comedi gave me a "not a file or directory", and dmesg was replete with messages about unresolved symbols e.g. RT_REQUEST_SRQ.  Among the modules that failed to load were 8255.o, comedi.o, comedilib.o, cb_pcidas.o.  A quick perusal of the comedi mailing list suggested an ad-hoc solution, which I employed: creating a directory /lib/modules/2.6.7-adeos/rtai, and then having that directory point (ln -s) to /usr/realtime/modules/*.  This I did, and then depmod -a.  At this point, things started working correctly - cat /proc/comedi gave me the correct information about the comedi install itself, and said there were no devices present.  modprobe cb_pcidas rectified that situation, and cat /proc/comedi now gave the correct information about the board, and the comedi test suite worked perfectly.  The final test before I went home for the evening was to reboot and make certain that everything still worked.  The system booted fine, but a quick cat /proc/comedi gave the "not a file or directory" again.  depmod -a brought up comedi, but now life took a dark turn.  modprobe cb_pcidas, and no sooner had I hit the return key than the entire system fell over, an instantaneous, complete reboot, which I assure you brought my heart into my throat.  Fortunately, the filesystem was not corrupted.  On the subsequent reboot, I removed the symbolic links, shut the system down correctly, and rebooted again.  I repeated the protocol I had originally performed, and again, everything worked perfectly normally.  This is where things stand.  I would like to know a couple of things, first, what exactly was it about this workaround that caused the system to fall over, and second, is there a procedure for resolving these unknown symbols without this workaround?  I would be very interested in any advice you might have to offer - I have seen other messages with this same problem, but unfortunately they were often met with a hand-waving "oh, just do a modprobe instead of insmod and everything will be fine," which doesn't really tell me anything.

System specs:
AMD XP 3200+
Asus A7N8-X Deluxe 2
WinXP (hde1)
Mandrake 10.0 / 2.6.7+RTAI-3.1testr4 (hde5)
comedi 0.7.68 + 9.28.2004 CVS
comedilib 0.7.22 + 9.28.2004 CVS
Measurement Computing PC-1600 AD board

Thanks in advance!
Mark

--------------------------
Mark Teagarden, Ph.D.
University of Texas at San Antonio
Dept. of Biology
6900 N. Loop 1604 W
San Antonio TX 78249