It seems that the problem is 100% reproducible if one unpacks gcc instead of the linux kernel. Steps to reproduce:
amd:/tmp# wget ftp.gnu.org:/gnu/gcc/gcc-4.4.0/gcc-4.4.0.tar.bz2 amd:~# lvcreate -L 30G -n test amd amd:~# mkfs.ext4 /dev/mapper/amd-test amd:~# mount -o noatime /dev/mapper/amd-test /mnt amd:~# cd /mnt amd:/mnt# for i in `seq 1 10`; do mkdir $i; (cd $i && tar xjf /tmp/gcc-4.4.0.tar.bz2); done amd:~# umount /mnt amd:~# e2fsck -f -D /dev/mapper/amd-test e2fsck 1.41.8 (11-July-2009) Pass 1: Checking inodes, blocks, and sizes Pass 2: Checking directory structure Pass 3: Checking directory connectivity Pass 3A: Optimizing directories Failed to optimize directory /1/gcc-4.4.0/gcc/ada (28832): No 'next' extent Failed to optimize directory /2/gcc-4.4.0/gcc/ada (94773): No 'next' extent Failed to optimize directory /3/gcc-4.4.0/gcc/ada (161120): No 'next' extent Failed to optimize directory /4/gcc-4.4.0/gcc/ada (227008): No 'next' extent Failed to optimize directory /5/gcc-4.4.0/gcc/ada (292863): No 'next' extent Failed to optimize directory /6/gcc-4.4.0/gcc/ada (358748): No 'next' extent Failed to optimize directory /7/gcc-4.4.0/gcc/ada (424608): No 'next' extent Failed to optimize directory /8/gcc-4.4.0/gcc/ada (490488): No 'next' extent Failed to optimize directory /9/gcc-4.4.0/gcc/ada (556342): No 'next' extent Failed to optimize directory /10/gcc-4.4.0/gcc/ada (622228): No 'next' extent Pass 4: Checking reference counts Pass 5: Checking group summary information /dev/mapper/amd-test: ***** FILE SYSTEM WAS MODIFIED ***** /dev/mapper/amd-test: 658711/1966080 files (0.0% non-contiguous), 1552469/7864320 blocks amd:~# e2fsck -f -y /dev/mapper/amd-test e2fsck 1.41.8 (11-July-2009) Pass 1: Checking inodes, blocks, and sizes Inode 28832, i_size is 57344, should be 69632. Fix? yes Inode 28832, i_blocks is 136, should be 128. Fix? yes Inode 94773, i_size is 57344, should be 69632. Fix? yes Inode 94773, i_blocks is 136, should be 128. Fix? yes Inode 161120, i_size is 57344, should be 69632. Fix? yes Inode 161120, i_blocks is 136, should be 128. Fix? yes Inode 227008, i_size is 57344, should be 69632. Fix? yes Inode 227008, i_blocks is 136, should be 128. Fix? yes Inode 292863, i_size is 57344, should be 69632. Fix? yes Inode 292863, i_blocks is 136, should be 128. Fix? yes Inode 358748, i_size is 57344, should be 69632. Fix? yes Inode 358748, i_blocks is 136, should be 128. Fix? yes Inode 424608, i_size is 57344, should be 69632. Fix? yes Inode 424608, i_blocks is 136, should be 128. Fix? yes [..] After this the file system is corrupt, and e2fsck cannot repair it any longer. -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
