On Tue, 21 Sep 2010 11:23:13 +0200, Paolo Giangrandi <paolo@xxxxxxxxxxx> wrote: > Hi, > I'm using NILFS2 as root filesystem (it's on /dev/sda2, only /boot was > apart on an EXT2 /dev/sda1, if I recall correctly) on my Archlinux > (with Linux 2.6.35.4 and nilfs-utils 2.0.19). > I got to manually run nilfs_cleanerd anytime I boot (read somewhere > it's because my /usr is on NILFS2 as well). > Yesterday, by mistake, I ran nilfs_cleanerd twice. Shut the system as > soon as I realized but now it's not booting anymore: a kernel oops > gets printed... > > Waiting 10 seconds for device /dev/sda2 ... > attempt to access beyond end of device > sda2: rw=16, want=4283850835945523848, limit=15695505 > Pid :1, comm: init Not tainted 2.6.35-ARCH #1 > Call Trace: > *snip* > NILFS: btree level mismatch: 68 != 1 > NILFS error (device sda2): nilfs_ifile_get_inode_block: ifile is broken > ERROR: Root device mounted successfully, but /sbin/init does not exist. > Bailing out, you are on your own. Good luck. > > /dev/sda2 gets mounted anyway on ramfs' /new_root, but if I try to > access most of its subdirectories more kernel oops appear, and only > very few files are shown. My home dir is completely gone. > > So, uh, other than pointing out this bug (I guess a second > nilfs_cleanerd should just quit, not shred your system...) I'd like to > know if there's any way to get my data back... I'll backup (through > dd) my whole /dev/sda (it's an 8GB SD) hoping to be able to recover it > soon... > > Thanks, > Paolo. Sorry for my late reply. If you haven't yet reformatted the partition, try the following manual recovery: $ cd your-work-directory $ git clone http://git.nilfs.org/nilfs2-utils.git $ cd nilfs2-utils $ git fetch http://git.nilfs.org/nilfs2-utils-devel.git fsck0:fsck0 -- This will download fsck0.nilfs2 from nilfs2-utils-devel git repo. $ git checkout fsck0 $ aclocal && autoheader && libtoolize -c --force && automake -a -c && autoconf $ ./configure $ make -- These steps will build fsck0.nilfs2. You may need to install libtool, autoconf (-2.6 or later), and automake packages. $ cd sbin/fsck $ su # ./fsck0.nilfs2 /dev/sda2 Super-block: revision = 2.0 blocksize = 4096 write time = 2010-09-25 22:19:18 indicated log: blocknr = 116170 segnum = 56, seq = 56, cno=1134 ^^ Clean FS. A valid log is pointed to by superblock (No change needed): blocknr = 116170 segnum = 56, seq = 56, cno=1134 creation time = 2010-09-25 21:15:04 # dd if=/dev/zero of=/dev/sda2 bs=8M count=1 seek=56 ^^ -- Get the number of latest segment and destroy it. This is destructive, so you are strongly recommended to backup the partition with dd or something before trying this. # ./fsck0.nilfs2 /dev/sda2 ... Do you wish to overwrite super block (y/N)? - Run fsck0.nilfs2 again, and press "y" if the above prompt appears. # mount -t nilfs2 -o ro,norecovery /dev/sda2 /nilfs - Mount the partition with "norecovery" mount option, and salvage data if succeeded to mount it. You may need to destroy more segments (i.e. 55,54,.. in the above example) to find a mountable checkpoint. Regards, Ryusuke Konishi -- To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
