Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 18 Feb 1997 02:17:07 +1100 (EDT)
From:      Darren Reed <avalon@coombs.anu.edu.au>
To:        kuku@gilberto.physik.rwth-aachen.de
Cc:        msmith@atrad.adelaide.edu.au, luigi@labinfo.iet.unipi.it, hackers@freebsd.org
Subject:   Re: File system/disk recovery tools ?
Message-ID:  <199702171523.HAA21329@freefall.freebsd.org>
In-Reply-To: <199702171443.PAA24698@gilberto.physik.rwth-aachen.de> from "Christoph Kukulies" at Feb 17, 97 03:43:04 pm

next in thread | previous in thread | raw e-mail | index | archive | help
In some mail from Christoph Kukulies, sie said:
> 
> > Luigi Rizzo stands accused of saying:
> > > > dd the whole disk off onto a file and mount it on a vnode?  You might
> > > 
> > > unfortunately is 1.6GB ... I need to find a bigger unit!
> > 
> > dd bits of it at a time? 8)
> > 
> > > > have to reconstruct the label and such, and using fsck's '-b' option
> > > 
> > > actually an annoying thing with fsck -b is that I never know what
> > > to use as an alternate superblock number. 32, says the manpage, is
> > > an alternate superblock, but what are others ? I guess they depend
> > > upon the geometry of the disk... is there a "magic" number that I
> > > can look for ?
> > 
> > You mean you didn't write them down when you made your filesystems? 8)
> > 
> > Ok, presumably your disklabel is OK.  Have a look in /sys/ufs/ffs/fs.h
> > at the superblock layout; it's 8K long, has a fixed checksum, and
> > contains various predictable fields (eg. the last mountpoint).  This
> > should let you hunt a copy of it down.
> 
> Some time ago I wrote a (really quick and dirty) little program which scanned
> the disk for once existing file systems.
> 
> It's too dirty to post in public but if anyone wants it I'll
> send it. At least one case beside of my own is known to have been
> successful.

I've written one of these too (SunOS4) and my comments are the same :-)
(btw, it worked too :-)

> Bottomline is that you scan the disk for a FS magic and then with the
> obtained data (offset and size) you do a dd skip= count= to a file which
> then can be used as /dev/vn0 to be mounted and files can be restored
> from there.

Does dumpfs exist on FreeBSD ?

Darren



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199702171523.HAA21329>