Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!crdgw1!uunet!ncrcom!whq-top!ncrlnk!ncrwic!srwic.UUCP!sralston
From: sralston@srwic.UUCP (Steve Ralston)
Newsgroups: comp.unix.sysv386
Subject: Re: GDB "Can't exec" Problem (+some other quirks)
Message-ID: <1991Apr05.074329.24091@srwic.UUCP>
Date: 5 Apr 91 07:43:29 GMT
References: <3440016@hparc0.HP.COM>
Organization: My_Home  Wichita, KS
Lines: 60

linas@hparc0.HP.COM (Linas Petras) writes:

>I am trying to run GDB-3.5 on SCO unix 3.2.2 and ever time I try and run a
>test program ("hello world" type) I get an error message back from GDB 
>saying that permission is denied to exec /bin/sh.  

>I'm sure I have seen an answer to this somewhere, but I can't find the
>reference anywhere in my archives. Can anyone tell me what the fix is ???? 

Simple...  Just make sure and run gdb from /bin/sh...  *Seriously*
(I was surprised when I found out by accident)  Gdb runs fine as root...
hummmm....  My login shell was /bin/csh at the time...

Here's my /usr/local/bin/gdb hack, er, FIX for gdb on SCO UNIX:
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
:
SHELL=/bin/sh; export SHELL
/usr/local/bin/gdb.real $*
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=

Wish I could tell you the root cause of the problem.  All I can say is that
I've had much shit break because of csh on SCO UNIX (make utility "weirdness"
comes to mind).

Another poster (Sean Eric Fagan) also mentioned that copying a READABLE
copy of /bin/sh to /usr/local/lib/gsh (along with recompiling gdb, I assume)
will solve this particular problem with gdb.

FYI: All this was before I got industrious one nite and put
/bin/ksh from 3.2v2 on a 3.2.1 (ODT1.0) system.  Yeah, no job control in
the 3.2.1 kernel, but job control kinda went out for me when multi-screens
came in.

On a related note, has anyone else seen something like:
	sh: can't exec /usr/bin/uuname

during bootup on SCO UNIX (3.2.1, aka ODT1.0) coming out of /etc/rc?
Drove me nuts for awhile until I changed the line causing this,
  from:
	/usr/lib/sendmail -q5m		# this is really smail 3.1.20
  to:
	/bin/su root -c '/usr/lib/sendmail -q5m'  # this is really smail 3.1.20

Now, I don't get the message any more...  BUT, the old
"40 second pause until I put up the Password: prompt, you Jerk" problem
has migrated from the initial multi-user login phase to a
"40 second (bet you think your system just crashed, heh, heh)" delay during
execution of /etc/rc during multi-user bootup.

The 40 second delay stuff began the first bootup after installing TCP/IP.

Can anybody offer explanations for:
	(1) Failed "exec's" on files that are executable by everyone
	    (even when uid=0(root) gid=0(root)).
  or	(2) The 40 second delay related to SCO TCP/IP

Thanks in advance!
--
Steve Ralston    (sralston@srwic.UUCP || ..!uunet!ncrlnk!ncrwic!srwic!sralston)
235 N Zelta, Wichita, KS 67206