Re: problems with savedefault

[Thierry Laronde]

On Wed, Jan 31, 2001 at 06:57:51PM +0100, Jochen Hoenicke wrote:
> 
> IMHO there should be a sanity check in savedefault_func, comparing
> some magic bytes in the sector that it changes and giving an error
> message if they are not present.

Rawwriting is, by nature, dangerous. So this solution seems sensible. There
are two cases :

1) We are writing well-defined sectors (boot sectors), or an area (where
*stage1_5 is generally put) that is, normally, not used by an OS (and if
some try to hide informations here, this is not really our problem);

2) We are not in case 1, so we have to verify that we are modifying a sector
which belongs to us. And indeed, the signature is the simple way (it already
exists), even if it can be combined with blocklist stage2 (be sure to modify
something that belongs to what we are supposed to deal with (stage2) && a
particular sector in stage2 (the signature can not be found identical
somewhere else or probability is almost 0)).

-- 
Thierry LARONDE, Centre de Ressources Informatiques, Archamps - France
http://www.cri74.org
PingOO, serveur de com sur distribution GNU/Linux: http://www.pingoo.org