Re: Kernel panic : VFS : unable to mount root fr on 03:05

Snow Cat (snowcat@netgate.net)
Fri, 18 Oct 1996 22:50:43 -0700 (PDT)


Theodore Y. Ts'o once wrote:
>
> Stephane, the problem is that your hard disk has developed a bad spot in
> a very critical place --- in the partition table of your main disk, to
> be specific. There's little you can do to recover from this situation,
> except to get the disk replaced.
>
> There's one very last ditch thing you can try doing, which is to remap
> the bad block --- *IF* your disk's hardware supports that. Boot a Linux
> repair disk, and then try running the command:
>
> dd if=/dev/zero of=/dev/hda bs=1k count=1
>
Why not bs=512?

> This will *wipe out* your partition table, but it's already been wiped
> out because of a disk failure on this block. If you're lucky though,
> your hard disk *may* remap the block. If so, you will be able to
> recreate your partition table using fdisk (again, use the one off of
> your Linux hard disk). This assumes that you know the original
> partition layout; if you do not, you may have to reinstall from
> scratch. (Which you would have had to do anyway; this procedure is
> basically a last ditch attempt to try to save you a little money by not
> requiring you to replace the disk.)
>

I had a similar situation recently, where my 2.5G HD had one
unreadable sector and kept crashing my system every several
days. Eventually I wrote a program to read every sector on Linux
partition and write it back. No problems or suspicious noises since
then.

This makes me wonder, why this functionality is not part of e2fsck and
fdisk? It seams that e2fsck can do a much better repair job if it
knows that some sector is bad instead of just reading some garbage after
a program simular to mine ran.

--
_. _ .
(_ ,_ _ , . / ` _ _L | Email: Oleg Kibirev <snowcat@netgate.net>
._)| U(_)\/\/ \_,(_L/L | Visit http://math.math.CSUFresno.EDU/~oleg/math.html
------------------------' to get my programs + PGP public key