lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  PHC 
Open Source and information security mailing list archives
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 16 Feb 2013 15:26:43 -0800
From: Jeremi Gosney <>
To: Marsh Ray <>
CC: Jens Christian Hillerup <>, 
 Jens Steube <>,
 "" <>
Subject: Re: [PHC] Different cost settings and optional input

On 2/16/2013 2:51 PM, Marsh Ray wrote:
>> -----Original Message-----
>> From: [] On Behalf
>> However, this attack is not that important I'd argue, because if two equal
>> passwords *exist* it is likely that that password is in some dictionary anyway,
>> or that they originate from the same person.
> A grand total of 38 of the top 100 passwords did not appear in the Twitter blacklist.
> Solar would probably have a better sense of things, but my feeling is that probably of all the passwords that have been chosen by exactly two users independently, most of them are not in a dictionary.

There was at least a 21% overlap between LinkedIn passwords, and
passwords leaked on other sites. And an additional 31%+ were simple
permutations of passwords used on other sites (e.g., + 1.)  Just
something to think about.

> I've tried to convince people of the merits of the following scheme:
> Use a fixed salt 

Fixed salts would give the attacker an advantage, and ultimately defeat
the purpose of salting.

> or otherwise allow the comparison of hashes.
> Maintain a dictionary or other database of 'seen' hashes across all users. 

This could easily backfire on you. Any mechanism you employ that enables
you to know if two users have the same password, will also enable an
attacker to know if two users have the same password. Building a
dictionary for the defender also builds a dictionary for the attacker.

> Incorporate the latest breaches into your dictionary.

We recently helped a customer implement this functionality, except the
check is done against the blacklist before hashing, not after. If the
user selects a password that has been compromised on another website,
the password is simply rejected and the user is told to select another

This check is also performed upon authentication as well (again,
pre-hashing), so that the next time the user authenticates, they are
notified that their password has been compromised on another site, and
are prompted to reset their password.

Thankfully we are not on the hook for maintaining this blacklist though.

> If new user B selects a password (or it appears in a new breach) that is the same as existing user A, force B to select another one and immediately *disable* user A until A successfully performs a password reset operation.
> No one seems to care much for this proposal though :-)

Yes, I can see how that proposal would not be very popular :)


Powered by blists - more mailing lists