strange e2fs behavior ...

[Sven LUTHER <luther@maxime.u-strasbg.fr>]

Hello, ...

i was compiling a second X tree with debugging info on a partition i didn't use
for a long time (800MB at the end of a 2.7GB old 5.25" double height seagate
SCSI drive). so i did a mke2fs on my old debian still glibc1.99 package
archive, unpacked X on it, and compiled it. But after a time i got lots of :


EXT2-fs error (device 08:13): ext2_readdir: bad entry in directory #131138:
rec_len % 4 != 0 - offset=0, inode=1249750673, rec_len=44015, name_len=122
/usr/local/SRC$ ls
EXT2-fs error (device 08:13): ext2_readdir: bad entry in directory #131138:
rec_len % 4 != 0 - offset=0, inode=1249750673, rec_len=44015, name_len=122
/usr/local/SRC$ ls
EXT2-fs error (device 08:13): ext2_readdir: bad entry in directory #131138:
rec_len % 4 != 0 - offset=0, inode=1249750673, rec_len=44015, name_len=122
/usr/local/SRC$ ls
EXT2-fs error (device 08:13): ext2_readdir: bad entry in directory #131138:
rec_len % 4 != 0 - offset=0, inode=1249750673, rec_len=44015, name_len=122
/usr/local/SRC$ EXT2-fs error (device 08:13): ext2_readdir: bad entry in
directory #137516: rec_len % 4 != 0 - offset=0, inode=1383296101,
rec_len=25701, name_len=58
EXT2-fs error (device 08:13): ext2_readdir: bad entry in directory #112865:
directory entry across blocks - offset=0, inode=906445859, rec_len=62532,
name_len=97

and the compiling stopped after not finding stuff. 

Since then i made a fsck on it and all blocks where bad, or some other kind of
message, and later this same problem appeared on the partition just before that
i was using previously. and so i lost my second X tree. 

Does anyone know what is the problem here ? Is it a kernel problem (was using
2.2.8, but reverting to 2.2.1prex didn't solve the thing, or could it be maybe
a mount or fsck or mke2fs problem ? i upgraded to latest debian/potato before
this problem appeared.

Any help would be very welcome.

Friendly,

Sven LUTHER