[Cryptography] CSPRNG for password salt

[Stephan Neuhaus]

On 2014-08-20, 10:08, John B wrote:
> Say we were using rand() - if the attacker can submit his/her own
> password AND obtain their own hash back, they could then bruteforce
> the seed (say because srand(time_now)) and now would presumably know
> the list of salts used for all of the subsequent password hashes.
> They can then pre-compute the tables necessary for an 'improved'
> offline attack on those passwords. Does this sound plausible?

I don't think so.  In my opinion, the cost in pre-computing the tables
is in their storage, not their computation.  In that case, it doesn't
matter if I compute the table on the fly or offline.

Fun,

Stephan