I have seen many of the problems that you have seen. I ran into a similar
problem with my Promise controllers, but I suspect that my problem is
different than yours. In the middle of crash testing, I swapped m/b's in
two of my computers went from having an Ultra100 and Ultra66 Promise
controllers to an onboard Ultra100 and regular Ultra100. What screwed
things up is that with default CMOS options it would try to boot off of the
onboard controller which had a hard drive with an OS on it (Linux of
course). Then, once the kernel was loaded, it would switch the order of the
controllers so that hde would suddenly be hdi and get all confused. I had
to tweak the CMOS options to not try to boot off of the onboard controller
so that it would try the off board controller instead.
One other thing that I did a couple months back was to reflash the off
board Promise controllers. Prior to V ~1.4 I read somewhere that there are
known problems. When I went to V 2.x I found that I could get higher
performance out of my drives. (Went from 35.56 MB/s to 35.75 MB/s on my
Ultra66 controller using Kernel 2.4.1 and a 45 GB IBM drive.)
>
> [] loading module /lib/reiserfs.o
> VFS: cannot open root device "2105" or 21:05
> please append a correct "root=" boot option
> kernel panic: VFS unable to mount root fs on 21:05
>
> It seems coincidental that I also have a Promise Ultra UDMA 66 card with
> 2 hard drives (Maxtor - master, Quantum Fireball - slave). Although,
> all these problems still occur with only the Maxtor attached. Could the
> support for Promise Ultra cards be broken?
