Re: [BUG] + [PATCH]: handling bad inodes in 2.4.x kernels

[Daniel Mack]

On Thu, Feb 14, 2002 at 07:33:19PM -0600, Ken Brownfield wrote:
> One could argue that a corrupt filesystem is a corrupt filesystem, but

but a corrupted filesystem should not change the behaviour of the entire
kernel. write-opening a bad inode should not end up with the effect that 
the rename() syscall does not work anymore, i wasn't even able to securely
shut down by box everytime that happend.

> I've seen this behavior first hand (without using debugfs,
> unfortunately). 

thats how i found that bug - it exists in real life.

> I think it's worth someone with filesystem fu taking a
> look at this patch.  Or "seconded", whatever. :)

there's another possibilty for fixing this in my latest posting.

daniel