From: Matthias Urlichs (urlichs@smurf.sub.org)
Date: 07/01/93


From: urlichs@smurf.sub.org (Matthias Urlichs)
Subject: Re: fun with e2fsck
Date: 1 Jul 1993 12:56:04 +0200

In comp.os.linux, article <5680@uswnvg.uswnvg.com>,
  sstarck@uswnvg.com (Steve Starck) writes:
> Short version:
> Are there alternate superblocks in e2fsck?
>
Won't help...

> Long version:
>
> After booting from the floppy, I tried to use e2fsck to fix whatever
> went wrong, but it complains about some sort of "bad block" or other
> (I didn't write this down and the computer's at home) But suffice it
> to say that I've given every option I can to e2fsck and it still
> pukes.
>
The ext2 file system maintains, for every inode, a list of blocks in the
inode. Unfortunately, when a block number on this number is bad (points
to a block which already is in use by somebody else, has an invalid
block number or one which points into the middle of some other data
structure), e2fs says it's zeroing said number. Surprise -- it isn't,
thus the error won't go away.

e2fsck also sorely needs an option to report which "other file" the
offending block belongs to, if any. (This should be an option because
it eats memory and thus may not work without a swap file.)

e2fsck also can't handle corrupt directories, i.e. whose "." and ".." entries
are missing. Worse, if you mount the corrupt file system and then try to
remove that directory, the system hangs.

-- 
I know nothing except the fact of my ignorance.
                                        -- Socrates
-- 
Matthias Urlichs  --  urlichs@smurf.sub.org -- urlichs@smurf.ira.uka.de   /(o\
Humboldtstrasse 7 -- 7500 Karlsruhe 1 -- Germany  --  +49-721-9612521     \o)/