Re:8tracks Password Security Alert
by scovetta on Saturday July 15, 2017 @05:13PM
(#14778981)
SHA-1 and MD5 are susceptible to collisions, not pre-image attacks, so replacing either with SHA-256 wouldn't really impact this scenario. The real question is whether they used iterated hashing (and to what degree), peppering, a keyed hash, or something else that would slow down a dictionary attack.
[ Parent
| on HN ]
Re:8tracks Password Security Alert
by JosephRedfern on Saturday July 15, 2017 @06:11PM
(#14779291)
I don't think OP suggested otherwise?
While "inadequate" and "completely broken" are totally valid qualifiers for those hashing algorithms, the way in which they are broken doesn't affect sha1crypt or md5crypt (only a pre-image attack would break them, collisions are bad but not enough). That doesn't mean you should use them though (you should use bcrypt or similar).
I think it's relevant to GGP's use of "completely broken" - collisions are far less useful than actually extracting the password string, since one is transferable and has long-term potential, and the other will probably lose all its value pretty quickly, especially with a change of salt.