On Monday 16 May 2005 07:04, Peter Jones wrote:
> Are you really sure you're using a proper test 3 tree, rather than
> test
>
> 2? AFAIK, this:
> > <3>audit(1116033649.581:0): avc: denied { transition } for
> > path=/usr/sbin/libgcc_post_upgrade dev=dm-0 ino=17072409
> > scontext=system_u:system_r:kernel_t
> > tcontext=system_u:system_r:unconfined_t tclass=process
> > <6>libgcc_post_upg[747]: segfault at ffffffffffffffd0 rip
> > 00000000004004d3 rsp 00007fffffddbd18 error 6
>
> Was fixed before test3 went out.
>
> After that, all bets are off. Though frankly, I'd run memtest86 on
> the box, and make sure your boot.iso CD is good as well.
Thanks for the reply, Peter. :) After getting no replies for a while, I
gave up on this thread and filed a bug (the right thing to do in any
case). Probably all future discussion should happen there.
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=157729
Here are a few summary tidbits from the bug report:
Yup, this is FC4t3, and yup, I did 18 minutes of memtest86 with no
errors. The NFS install always crashed for me, while a DVD install
burned from the same DVD image succeeded.
FC3 failed in nearly the same way on this machine, with this FC2 NFS
server.
The failure appears to involve cramfs's inability to decompress blocks
from stage2.img *after* all the rpms have been installed on the target
machine. It worked with stage2.img just fine at the beginning of the
install. The eventual crash is a segfault of the child anaconda
process and an illegal instruction by the parent anaconda. loader (or
init?) then gives an "Install exited abnormally" error right before
shutting down.
I did forget to point out the avc denied's in the bug report -- should I
submit another report for that?
David