Subject: my kernel has broke. scripts don't work.
To: None <port-sparc@NetBSD.ORG>
From: matthew green <mrg@eterna.com.au>
List: port-sparc
Date: 06/17/1998 01:23:54
hi folks.

i attempted to update to a 1.3E (just prior to 1.3F) kernel last night,
and it seemed a bit unstable (kernel builds got segv/bus and eventually
a kernel fault), so i went back to the old kernel.  it has since stopped
running scripts.  of any kind.  it appears as if the argument/environment
vectors are being screwed some how.  eg, if i have a script /tmp/blah,
that looks like:

#!/bin/sh
exit 0

mode 555, and i run `/tmp/blah', i was getting things like:

	TERM=screen-w: Can't open TERM=screen-w

if i run `/tmp/blah foo' i was getting:

	STY=436.ttyp1.splode: Can't open STY=436.ttyp1.splode

if i run "/tmp/blah foo baz' i was getting

	baz: Can't open baz

the first two values in my environment are:

	STY=436.ttyp1.splode
	TERM=screen-w

which makes me believe that the argv/envp pointers are being screwed
somehow.  but it doesn't make sense.

a rebuild 1.3E kernel exhibted the same problem.  a -current 1.3F
kernel does too.  my laet april 1.3E kernel has it.  but my 1.3C february
kernel does *not* (this is the kernel currently running).  the reason
it doesn't make sense is that the late april 1.3E kernel had been running
(mostly happily) since late april.  :(


does anyone have any clues?  i have no more.  christos asked me to try
unlimiting everything but that didn't help.  it doesn't seem to depend on
any programs being run.. if i 'boot -s' scripts fail there too.



.mrg.