On 17 Sep 2002 12:05:01 -0500, Miguel Mendez wrote:
> Hi,
> 
> I've been able to reproduce this problem on two different machines (a
> IBM laptop and a p3 desktop), also, some other people have reported this
> on IRC. The problem is as follows. During the tarball install, sysinst
> segfaults, causing the install to abort. The workaround I've done so far
> is as follows:
> 
> 1) Reboot
> 2) You reboot and drop to single user, since rc_confugred is not set
> 3) mount /usr if necessary
> 4) mount -u /
> 5) for i in /usr/INSTALL/*.tgz; do tar xzpvf $i;done
> 6) use sed to change rc_configured to YES
> 7) reboot
> 8) now you have a usable system, use vi to setup ifconfig.ifname and
> friends
> 
> Question: has any other people found this problem? I think I still have
> the core files somewhere, could we try to find where the problem lies?
> Both boxes installed 1.5.3 flawlessly.

Yes. I reported this in
http://mail-index.netbsd.org/port-i386/2002/09/12/0007.html
last week.

The current status of this is that we removed every i/o card from
the box except video, and even swapped that with another type of
video card. We swapped the 4 sticks of EDO ram around and removed
the different pairs of them one at a time. We swapped cpus (MP
pentiumpro) and even voltage regulators. Still segfaults.

When sysinst segfaults, it does so when installing misc.tgz - every
time.

The interesting thing is that memtest86 was run for two days
(reported no errors), then immediately after shutting it down and
rebooting, NetBSD1.6 and RH7.3 both installed fine. The box ran
with either one for about two hours, then shut down.  Can't even
get the bootloader to run properly now.  Sysinst does leave a
corefile on /dev/wd0a, but we don't have a good way at the moment
to move it to something we can look at it with. We also ran one of
the cdrom-only linuxes (doesn't use hard drive) and it ran fine,
in addition to RH7.1 and NetBSD1.5.3.