while trying to build some software (OpenJade) to be able to run gnome 2.0 ... I had the message about an impossible mkdir command (I was logged as superuser !) and after that the partition which was holding the /usr tree was impossible to be mounted.

The superblock get corrupted with an inode count of 0 (nul) ???
I try desesperatly to run e2fsck with the argument -b 32769 to get a copy of the superblock, it didn't work !!! I tried the arguments -b 65537 , -b 98305 . all of these returned an error about a bad magic number ...

the command dumpe2fs returns some information as a blocksize of 4096, a blockgroup of 32768, a blockcount of 2097152 ...

Is there any way to recover a correct superblock ?

I had that kind of problems after a live partition resive on a server.
I actually used an option (s or b) of mkfs. Do go through the manpages though to make sure that's what you need...

Best of luck,

hi everybody,
hello ecb,
in fact I was able to run e2fsck on the corrupted partition, BUT instead of specifying an offset of 32769, which made e2fsck to answer about a bad magic number, I used the offset 32768.
At this location I found a "valid" superblock. e2fsck worked quite a long time but recovered part of my configuration.
I still had to reinstall a small part of the system software, and was able to recover quite a lot of my lost files.

anyway I am about to burn several backup CD's in case
long live to linux
c:.