Xenix Syscrash on Syscall Trace

[Tanner Andrews]

> ... tracing syscall causes syscrash ...

Experience here shows that the problem is particularly bad on 8086-based
systems.  The poor little 8086 can't protect itself.  In order to make
a multi-user 8086 system, other hardware is required to protect the
CPU.  Syscall mechanism is to stash a funny address in a funny addr and
call indirectly to that.  Play around in that funny address setting
breakpoints to trace things, mash reset when you're done.

With the `286 but version 1 xenix (you want to talk about bugs?  let
me tell you about bugs...) the hardware was different, but the thing
was happy to crash for this sort of thing anyway.  Also crashed for
full moon, half-moon, wind_out_of_east, and running "vi" as I recall.

We now have version 3.0b xenix (no, not 3.1 -- that'll be out a couple
of months ago).  It does not crash on misguided attempts to trace a
syscall.  I assume that by now IBM and ALTOS are shipping xenix version
3.xxx -- at least for `286 systems.  If this is not the case, would
somebody warn me before the boss lays the money down to buy a couple
of these things?

Of course most of the words used above are trademarks of someone.  And
of course, none of the opinions { any version of xenix (most esp
vers 3.xxx) might possibly have more bugs than a NYC slum or that any
chips are not good for anything but controlling elevators } belong to
anyone but myself.

-- 
<std dsclm, copies upon request>	   Tanner Andrews, KI4PV
uucp:					...!decvax!ucf-cs!ki4pv!tanner