Bugzilla – Bug 151884
Installation stage 2: Using ext3 for / results in a panic()
Last modified: 2006-02-17 21:13:18 UTC
I tried to install Beta4 using ext3 for the root-filesystem. Besides the fact that the partitioner formatted /boot with the wrong fs, the partion setup was completed successful and correctly (verified with fdisk). Now, in the second stage of the installation, I received a kernel panic with a message of the sort ``ext3 not supported''. The panic resulted from the failed attempt to mount the root filesystem. I consider this a very critical error that should be corrected very soon, obviously the normal kernel does not support ext3 anymore. I verified that ext3-support for the installation kernel works fine. Please note that many systems (specially root servers) are getting shipped with ext3 by default.
Please post the exact kernel error. We need it to determine if the bug is in the kernel code or in mkinitrd or the installer.
Hello Chris, while I checked out the fstab I realized that there were two entries for /, one with reiserfs and a wrong device (completely wrong) and a correct line. Nontheless, commenting out the wrong line did expectedly change nothing here: Here the exact kernel messages (written down from the screen): (...) # if you need more, ask for it Loading processor Loading thermal Loading fan Loading reiserfs Attempting manual resume rootfs: major=3 minor=4 devn=772 Mounting root /dev/hda4 # correct attempt mount: unknown filesystem type `ext3` unmount: /dev: device is busy unmount: /dev: device is busy # yes, is outputted twice Kernel panic - not syncing: Attempted to kill init It's quite possible the ext3 module was not loaded.
:/ sorry for the leading spaces... I will try out something now.
ok, it looks very likely the initrd is missing ext3. It is probably missing ext3 because it was confused about the list of filesystems. Probably a yast bug, definitely not kernel.
OK, here it is: The following two entries for / in fstab were created by YaST: /dev/hda3 / reiserfs defaults 1 1 /dev/hda4 / ext3 acl,user_xattr 1 1 I removed the wrong line from fstab which makes no difference here. I suppose only required modules are added to the initrd, then the second (and correct) line is getting ignored. This would indicate that this is a YaST problem, specially a problem with the partitioner! Not to confuse the maintainer too much, I will redo the complete thing and create a new bugreport for this issue. Suggestion: It would be a nice `fallback' if the initrd would contain all fs-modules mentioned in the fstab and output some warning if entries are doubled. This would at least not lead to a panic() if the affected fs is the rootfs.
All handled in 151947 now...
*** This bug has been marked as a duplicate of 151947 ***