Red Hat Bugzilla – Bug 18660
Kernel panic: VFS: Unable to mount root fs on 03:05
Last modified: 2007-04-18 12:29:05 EDT
... autorun Done Kernel panic: VFS: Unable to mount root fs on 03:05 I get the kernel panic message when it tries to mount / for the first time. It was successful in loading the kernel from the same hard disk partition. I am able to successfully boot the system from the resucue floppy. The system and kernel in use is from the 7.0 release iso images with the released patches as of 14:12pdt 8 oct. I had hoped that the mount-2.10m-6.i386.rpm and friend would resolve this problem but it didn't.. I thought it possible that the large disk was part of the problem but then why does it load the kernel from hardisk and does / mount with the floppy version of the kernel? For reference here is my fstab and fdisk layout.. LABEL=/ / ext2 defaults 1 1 /dev/hda7 /home ext2 defaults 1 2 /dev/cdrom /mnt/cdrom iso9660 noauto,user,ro,unhide 0 0 /dev/cdrom1 /mnt/cdrw iso9660 noauto,user,unhide 0 0 /dev/fd0 /mnt/floppy auto noauto,user 0 0 /dev/jaz /mnt/jaz auto noauto,user 0 0 /dev/zip /mnt/zip auto noauto,user 0 0 none /proc proc defaults 0 0 none /dev/pts devpts gid=5,mode=620 0 0 /dev/hda5 swap swap defaults 0 0 /dev/hdd1 /vast ext2 defaults 1 2 Disk /dev/hda: 255 heads, 63 sectors, 2055 cylinders Units = cylinders of 16065 * 512 bytes Device Boot Start End Blocks Id System /dev/hda1 * 1 765 6144831 b Win95 FAT32 /dev/hda2 766 2055 10361925 5 Extended /dev/hda5 766 798 265041 82 Linux swap /dev/hda6 * 799 1192 3164773+ 83 Linux /dev/hda7 1193 2055 6932016 83 Linux
Your lilo.conf is broken. Device 03:05 is /dev/hda5, which happens to be your swap partition. I'm quite sure you have a "root=/dev/hda5" somewhere in your /etc/lilo.conf, where it should be "root=/dev/hda6". Edit lilo.conf to fix this, then run /sbin/lilo as root.
/dev/hda5 used to be my / partitiion. I found that an upgrade to 7.0 failed because I didn't have enough space in /usr. I used fdisk to merge 4 partitions together to form a single /. Whille doing this I moved swap from /dev/hda6 to /dev/hda5 and made / /dev/hda6 with my old /dev/hda10 moving to /dev/hda7 when things renumbered. Once the repartion was complete I booted from the install CD and performed a custom install. I chose mbr as the location for lilo. The install went flawlessly but when I rebooted I found the problem. Late last night I thought to compare lilo.conf with the version from my previous install (yess I backed up completely) and noted that there was an initrd entry that my olld one didn't have. The root specification correctly pointed to /dev/hda6 Just for fun, I commented out the initrd and ran lilo. I got an error message: Fatal: sector 18126678 too large for linear mode (try 'lba32' instead) The man page for lilo doesn't show lba32 but I tried it anyway. This time lilo ran without error and the system booted correctly from hard disk. My conclusion is that the install script must not have noticed that lilo failed to install a new mbr since it would appear from your analysis that /dev/hda5 was still marked as the boot partition even though the /etc/lilo.conf did not show this. BTW, when the system booted after rerunning lilo, I saw a brief spash screen (1 sec?) that seemed to show a graphical boot partition selection screen. That would be great to have since I really miss the way the classic Amiga handled selection of a boot disk.
Reassigning to anaconda (mount can't do anything about the installer failing to detect lilo errors). Yes, lilo now has a graphical prompt - if you want to see it for longer, modify /etc/lilo.conf and rerun lilo.
This is related to bug 13614 and bug 16639 and bug 14351. Marking as a dupe of bug 13614. *** This bug has been marked as a duplicate of 13614 ***