Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!usc!wuarchive!emory!att!cbnews!cbnewsl!urban
From: urban@cbnewsl.att.com (john.urban)
Newsgroups: comp.unix.questions
Subject: Re: System Dump Memory Image in Swap Device after Panic on SCO 3.2.2
Message-ID: <1991May7.144351.10296@cbnewsl.att.com>
Date: 7 May 91 14:43:51 GMT
References: <1171@pemcom.pem-stuttgart.de>
Organization: AT&T Bell Laboratories
Lines: 36

In article <1171@pemcom.pem-stuttgart.de> basien@pemcom.pem-stuttgart.de (Tillmann A. Basien) writes:
>
>Hy netlanders,
>	after I tuned my SCO UNIX 3.2.2 System I got a panic. So far so good.
>	I corrected the parameters, but after every booting I get a message
>	out of /etc/dumpsave:
>
>
>echo 'There may be a system dump memory image in the swap device.'
>echo 'Do you want to save it? (y/n)> \c'
>
>	After I saved the image on the tape, I think that I don`t get this
>	massege after reboot. But I get it all the time.
>
>	How can I clear the swap device to stop getting this message?
>
>
>	Tillmann

AT&T UNIX System V/386 Release 3.2 has this same problem and had many complains
about it.  Finally AT&T fixed it in UNIX System V/386 Release 3.2 Version 2.2
Basically the fix was:

Make /etc/dumpsave a little smarter so it compares the first 512 bytes or so
of /dev/mem to /dev/swap.  If they are then same, then assume that a PANIC
occured and the dump (of memory images) was saved on /dev/swap.  If this first
512 don't match, then no dump was taken, so don't ask any questions.

Then change /etc/bcheckrc so that it does a echo "MARK" > /dev/swap (after the
call to dumpsave).  This is so that later reboots will not find the first 512
bytes of /dev/mem and /dev/swap the same.


Sincerely,

John Urban