nanook@rwing.UUCP (Robert Dinse) (09/08/90)
Just a follow up. I figured out what was causing the weird things
to happen to drive configuration.
In my init, I neglected to open stdin, stdout, stderr to /dev/null
(or anything else) prior to invoking a shell to execute commands in
/etc/rc.
In my /etc/rc file there is a command, /etc/verify -f y, which turns
on read-after-write verify for floppy disk writes. In order to do this
it accesses /xenix and writes into /dev/kmem.
What I know: Opening stdin/out/err to /dev/null prior to executing
a shell to execute /etc/rc commands fixed the problem.
What I am speculating: That /etc/verify probably first opens /xenix
to read the symbol table, then opens /dev/kmem to make changes as required.
Then writes a confirmation message to stdout. Because stdin/out/err were
not opened first, /xenix became file descriptor 0, /dev/kmem file descriptor
1, (stdout), and the message got written to /dev/kmem right after the place
where the verify status goes. This of coarse is just speculation, but the
problem was fixed by opening stdin/out/err to /dev/null first.
Thanks to all of you who sent me mail with possible causes and gave
me clues as to where to look.
I really appreciate all of the help I have received on various 6000
problems. These machines are getting close to functional now, too bad it
couldn't have happened >before< they were antiques.. But what the hey, if
they do the job they do the job.