On Fri, 1 Apr 2005, Ludwig Nussel wrote:

Mike Rose wrote:

...

I guess this is a security matter in a way if you consider that installing the 2.6.5-7.151-default kernel on a computer using software raid for / means it can no longer boot, therefore it is very secure...

anyway this kernel rpm removed the raid1 module from /etc/modules.conf so that a computer booting from a linux raid partition could no longer boot. restoring /etc/modules.conf fixed it. no warnings during rpm install (-ivh used).

I suppose you are talking about SUSE LINUX 9.1 and modprobe.conf? Which line exactly got removed?

Thanks for responding so quickly (admittedly I expect someone would, as they usually do, on this list which is why I posted here so that others might avoid this problem). My apologies for not posting enough information. Yes, Suse Linux 9.1 My mistake, also, about modprobe.conf. The 2.6.5-7.151-default rpm in fact seems to have changed /etc/sysconfig/kernel. from: INITRD_MODULES="ata_piix jbd ext3 raid1" to: INITRD_MODULES="ata_piix jbd ext3" it also had a go at re-creating all of the initrd files in /boot: -rw-r--r-- 1 root root 1255496 Apr 1 13:27 initrd-2.6.4-52-smp -rw-r--r-- 1 root root 1251364 Apr 1 13:27 initrd-2.6.5-7.145-default -rw-r--r-- 1 root root 1292274 Apr 1 13:27 initrd-2.6.5-7.151-default -rw-r--r-- 1 root root 1252259 Apr 1 15:03 initrd-2.6.5-7.151-default.suse This means that the computer could not boot from any previous kernel because no initrd had the modules in it for mounting a linux software raid partition. Interestingly the first time I broke a computer (booting from linux raid partition) with the 2.6.5-7.151-default rpm the Suse Linux 9.1 repair process (from install CD) worked. The second time the repair process ran and found that the raidtools rpm was missing (so no /sbin/raidautorun during boot) which was a bit surprising (I do not know why that was missing as the computer had been rebooted fairly recently), it repaired /etc/sysconfig/kernel and then had a go at running mk_initrd, but failed. Quickest solution was to grab a working initrd from another computer, as you can see from the file list above. Dear "person that writes insulting emails to an entire list based upon their interpretation of an email", I hope tomorrow is a better day for you. regards, M.

cu Ludwig

-- (o_ Ludwig Nussel //\ SUSE LINUX Products GmbH, Development V_/_ http://www.suse.de/

-- Check the headers for your unsubscription address For additional commands, e-mail: suse-security-help@suse.com Security-related bug reports go to security@suse.de, not here

The Friday 2005-04-01 at 17:29 +0200, Ludwig Nussel wrote:

Now you are talking about yet another file. Anyways, I just tried to reproduce on a 9.1 installation. No such problem happened. Neither /etc/modprobe.conf nor /etc/sysconfig/kernel got changed, the raid1 module was still in the initrd after installing new kernels. So it must be a local issue on your side.

I had no such problem, but another one, related: YOU did not run mkinitrd at all. Nothing in the log about that, however, but next boot I noticed problems. 2005-03-27 14:06:27 <1> nimrodel(8042) [packagemanager] InstYou.cc(log):1446 Installing kernel: "The Linux Kernel on x86 architecture." 2005-03-27 14:06:27 <1> nimrodel(8042) [rpmdb] RpmDb.cc(installPackage):1545 RpmDb::installPackage(/var/lib/YaST2/you/mnt/i386/update/9.1/rpm/i586/kernel-default-2.6.5-7.151.i586.rpm,156) -- Cheers, Carlos Robinson