I tried test10 today by uncommenting arjan's repository in the up2date
sources file and running up2date. It couldn't mount the root filesystem
due to a problem with the initrd. What happens is that the adaptec
module loads fine from the initrd and scans the scsi bus, it prints a
message that it is loading the ext3 module, but mounting / fails with an
error code. So as a result pivot_root fails and it can't find init.
When I rebooted back into 2.4.22-1.2115.nptl, I discovered that /home
(which is located on /dev/hda, my first IDE disk) had at least one
corrupted directory, apparently in homedir.
My system has seen several upgrades so it isn't using fs-labels in
fstab. Don't know if that's relevant, but my initrd's work fine under
kernel 2.4. Also, after upgrading this machine to both RH9 and now FC1,
I had to enter grub commands manually to boot because it apparently
couldn't find grub.conf. This is probably because I recently added
/dev/hda and changed /dev/sda to no longer be the boot disk.
Reinstalling grub manually clears this problem up, but it would be nice
to find a way to fix my system so that anaconda won't get confused next
time I do an upgrade. And I'm wondering if this is related to the initrd
confusion, since I had the same problems when building 2.5.70 myself and
had to statically link with the adaptec driver so that I wouldn't need
an initrd.
/etc/fstab:
/dev/sda6 / ext3 defaults 1 1
/dev/hda1 /boot ext3 defaults 1 2
/dev/hda4 /home ext3 defaults 1 2
/dev/sda5 swap swap pri=0 0 0
/dev/sda2 /f vfat gid=17,umask=002,noexec 0 0
/dev/hda2 /c ntfs gid=17,umask=007,noexec,ro 0 0
/dev/fd0 /mnt/floppy ext2 owner,noauto 0 0
/dev/cdrom /mnt/cdrom iso9660 kudzu,owner,noauto,ro 0 0
none /proc proc defaults 0 0
none /dev/shm tmpfs defaults 0 0
none /dev/pts devpts gid=5,mode=620 0 0
/dev/hda3 swap swap pri=0 0 0
/dev/cdrom1 /mnt/cdrom1 udf,iso9660 noauto,owner,kudzu,ro 0 0