EISA + Wangtek + ESIX = pffft...

[Evan Leibovitch]

Had all sorts of cute experiences trying to install Esix on a 486 EISA
box (an ALR PowerCache 4e, I think it was called -- impressive little
system).

Most of the installation worked OK, except for two things. One was
resolved -- the ESIX wouldn't recognize more than 16 Meg of RAM until
some of the ALR cute caching mechanisms were turned off. The other is a
bit stranger...

The system was equipped with made-by-Wangtek 120-Meg cartridge tape
subsystems, plugged into EISA slots. Using them would sometimes give
perfect results, but just as often crash the system with a kernel trap
and that horrid "trying to dump oodles and oodles of pages before I die"
message.

The vendor was stymied. The addresses, DMA and interrupts selected on the
board matched exactly what the driver expected. As I said, sometimes it
worked fine. But why the crashes?

In a fit of desperation, I moved the board from the EISA slot to one of
the two older ISA (16-bit) slots on the system. Behold, NO CRASHES!

(I'm not totally out of the woods yet, tho'. The tape has hung once on
an overnight, cron-induced backup run. Drive light stayed on, software
woulnd't read or write the device until rebooted. But it's a hell of a
step up from panic kernel traps!)

So what's the scoop? I thought EISA was supposed to be completely
compatible with the old stuff. Why would a board that works under ISA
crash Unix when put in an EISA slot? Is the problem unique to ESIX or
would it affect other 386 *IX products?
-- 
Evan Leibovitch, Sound Software, located in beautiful Brampton, Ontario
evan at telly.on.ca / uunet!attcan!telly!evan / moderator, rec.arts.erotica
           ...quoth the Raven, "Eat My Shorts!" -- Bart