Re: MD5 for passwords
- From: "Matthew Fanto" <mfanto@xxxxxxxxx>
- Date: 31 Oct 2006 12:17:50 -0800
Paul Rubin wrote:
Unruh <unruh-spam@xxxxxxxxxxxxxx> writes:
No, you cannot. You cannot create one file which has any given hash. What
you would have to do is to know everything on his drive. You could then
create another drive with different information but the same hash as his
drive with a suitable addition you place on his drive. Ie, the addition
must be placed there AFTER the rest of the information on the disk is
fixed and known by you.
Yes, that is part of the attack: the attacker gets to seize the drive
and then use its contents while constructing the altered version.
If the attacker is the one seizing the drive, there are other things
they can do. This was definately an attack I brought up with them. The
way they thwart the attack is strong chain of evidence and having a
number of different people from legal, HR, and an uninterested 3rd
party being present during the seizure as well as the image stage. But
it's certainly a better idea to use SHA256 rather than MD5, as it stops
the attack. Having witnesses though stops the attack of the person
doing the seizing just planting evidence.
[other message:]
You have to alter the data ( yes in one 512 bit chunk) but while you
are generating the hash. Ie, to make the two hashes the same, you
have to dynamically change the contects of BOTH drives.
Hmm, OK. Anyway you can still potentially carry out an attack like
this one:
http://www.cits.rub.de/MD5Collisions/
but you have to plant the modified version in advance, a slightly less
realistic attack.
Which is exactly the attack I mentioned in the beginning :). An
enterprising criminal can plant modified versions in advance.
.
- Follow-Ups:
- Re: MD5 for passwords
- From: Unruh
- Re: MD5 for passwords
- References:
- Re: MD5 for passwords
- From: Matthew Fanto
- Re: MD5 for passwords
- From: David Wagner
- Re: MD5 for passwords
- From: Matthew Fanto
- Re: MD5 for passwords
- From: David Wagner
- Re: MD5 for passwords
- From: Unruh
- Re: MD5 for passwords
- From: Paul Rubin
- Re: MD5 for passwords
- Prev by Date: Re: MD5 for passwords
- Next by Date: Re: MD5 for passwords
- Previous by thread: Re: MD5 for passwords
- Next by thread: Re: MD5 for passwords
- Index(es):
Relevant Pages
|