[ATrpms-users] Smart reporting segfault during kernel upgrade.

[John Pilkington]

Hi Axel, Matthias

Thanks for the suggestions.  Some updates added below: I need to do
other things, so if this isn't a widespread problem I'll go quiet for a
bit.

John

On Tue, 2006-04-04 at 01:11 +0200, Axel Thimm wrote:
> On Mon, Apr 03, 2006 at 10:08:16PM +0100, John Pilkington wrote:
> > On Mon, 2006-04-03 at 18:04 +0200, Matthias Hensler wrote:
> > > On Sun, Apr 02, 2006 at 12:28:43PM +0100, John Pilkington wrote:
> > > > Output from kernel-suspend2-2.6.16-1.2080_3.rhfc5.cubbi_suspend2 at i686:
> > > > /var/tmp/rpm-tmp.95583: line 6:  5146 Segmentation fault
> > > > /usr/sbin/module_upgrade 2.6.16-1.2080_3.rhfc5.cubbi_suspend2
> > > > 
> > > > I have seen similar messages during earlier FC5 kernel upgrades.

Reinstalled 1.2080 - got
Output from kernel-2.6.16-1.2080_FC5 at i686:
/var/tmp/rpm-tmp.68207: line 6:  3012 Segmentation
fault      /usr/sbin/module_upgrade 2.6.16-1.2080_FC5
 
> > > 
> > > That is the postscript creating the initrd and updating the bootloader.
> > > 
> > > > Do I need to take some action on this?
> > > 
> > > Well, it should not happen. Especially if you see these messages even
> > > for the mainline kernel, there is something broken. Either you have some
> > > hardwareproblem which causes random failure, or something is not
> > > correctly installed/damaged.
> > > 
> > > mkinitrd and new-kernel-pkg are just simple bash-scripts with some calls
> > > to losetup and nash. Do you have any unusual configuration (eg
> > > non-standard devices mounted)? And did you update from an older Core
> > > version, or is this a clean install?
> > > 
> > > Regards,
> > > MatthiasOn Mon, Apr 03, 2006 at 10:08:16PM +0100, John Pilkington wrote:
> > On Mon, 2006-04-03 at 18:04 +0200, Matthias Hensler wrote:
> > > On Sun, Apr 02, 2006 at 12:28:43PM +0100, John Pilkington wrote:
> > > > Output from kernel-suspend2-2.6.16-1.2080_3.rhfc5.cubbi_suspend2 at i686:
> > > > /var/tmp/rpm-
> > 
> > Thanks for your reply.  It wasn't a clean install but, as I reported
> > earlier 
> > 
> > ([ATrpms-users] Upgrade to FC5 - mainly OK but some problems, March 31),
> > 
> > an upgrade from an RH9 installation that wasn't in good health - I
> > couldn't start X after attempting to update RH9 itself after a fairly
> > long period of neglect.
> 
> Please make sure that the software components are OK by using rpm -V
> on the package names. Especially for /usr/sbin/module_upgrade (kudzu)
> and glibc.

kudzu - no problem found
glibc - /etc/ld.so.conf has been modified, looks plausible.

> 
> But I'd recommend backing up all data and having a clean install. rh9
> to fc5 are 5 (!) interim releases, where kernel-2.6, lvm2, udev,
> (nptl?), tls and God knows what else has been upgraded, so it is very
> likely that you have lots of nasty surprises waiting for you. and if
> the system to start with wasn't in good health, then reinstalling from
> scratch is a must :)