3 ms·

Yep, password is common enough to unhash pretty easily. People need to learn to salt their hashes for fuck's sake.

by noamsml 14y ago

alexkus 14y ago
In this case, salting would give almost no benefit given the poor password choice. The pull request included the code that does the hashing:- if( empty( $auth_pass ) || ( isset( $_POST['pass'] ) && ( md5($_POST['pass']) == $auth_pass ) ) ) $_SESSION[md5($_SERVER['HTTP_HOST'])] = true; else printLogin(); So any salt [was hash] used would have to be present in the code too. Given that such a simple password (8 digits) could be brute forced in seconds on an average PC, even with a salt, it doesn't really matter whether it was salted or not. A password of "p*l12nJ9£l ~98as2389bvkqsopfq£3oef2[olpe]wog!wei^og(8ni" would take an unrealistic amount of time to brute force, even if unsalted, and it's beyond the scope of any precomputed rainbow tables for similar reasons. Anyway, it's only a concern if he uses the same password elsewhere.
- gknoy 14y ago
  > A password of "p*l12nJ9£l ~98as2389bvkqsopfq£3oef2[olpe]wog!wei^og(8ni" would take an unrealistic amount of time to brute force, even if unsalted, and it's beyond the scope of any precomputed rainbow tables for similar reasons. Since rainbow tables look up a password via its hashed value, I believe you could find something else that has a hash collision -- for all we know, it has the same hash as 'ponies'.
  - alexkus 14y ago
    Sure, but a hash collision for a specified value is still a needle in a haystack. The sheer size of the keyspace (2^128) means that rainbow tables can only ever exist for a subset of the keyspace. 1M machines running for a year each capable of doing 20M MD5 hashes a second will still only cover 1/539514153540300709th of the MD5 keyspace. But, yes, salting would prevent the use of rainbow tables completely.
    - Freaky 14y ago
      20M? Doesn't alter your point, but a pair of 3-gen-old ATI graphics cards manage on the order of 5000-6000M: Hash type: MD5, Hash: 6df23dc03f9b54cc38a0fc1483df6e21 Device #0: [RV870] 850.00 Mhz 1600 SP Device #1: [RV870] 850.00 Mhz 1600 SP CURPWD: @uX5G DONE: 22.30% ETA: 1m 34s CURSPD: 6436.3M=3217.8M+3218.6M http://www.golubev.com/hashgpu.htm http://www.golubev.com/hashgpu.htm
    - duaneb 14y ago
      This doesn't really change your statement, but 20M MD5 hashes/second is pretty damn wimpy these days. IIRC 20 BILLION is more accurate.

Yep, password is common enough to unhash pretty easily. People need to learn to salt their hashes for fuck's sake.

by noamsml 14y ago

Yep, password is common enough to unhash pretty easily. People need to learn to salt their hashes for fuck's sake.

In this case, salting would give almost no benefit given the poor password choice. The pull request included the code that does the hashing:- if( empty( $auth_pass ) || ( isset( $_POST['pass'] ) && ( md5($_POST['pass']) == $auth_pass ) ) ) $_SESSION[md5($_SERVER['HTTP_HOST'])] = true; else printLogin(); So any salt [was hash] used would have to be present in the code too. Given that such a simple password (8 digits) could be brute forced in seconds on an average PC, even with a salt, it doesn't really matter whether it was salted or not. A password of "p*l12nJ9£l ~98as2389bvkqsopfq£3oef2[olpe]wog!wei^og(8ni" would take an unrealistic amount of time to brute force, even if unsalted, and it's beyond the scope of any precomputed rainbow tables for similar reasons. Anyway, it's only a concern if he uses the same password elsewhere.

gknoy 14y ago
> A password of "p*l12nJ9£l ~98as2389bvkqsopfq£3oef2[olpe]wog!wei^og(8ni" would take an unrealistic amount of time to brute force, even if unsalted, and it's beyond the scope of any precomputed rainbow tables for similar reasons. Since rainbow tables look up a password via its hashed value, I believe you could find something else that has a hash collision -- for all we know, it has the same hash as 'ponies'.
- alexkus 14y ago
  Sure, but a hash collision for a specified value is still a needle in a haystack. The sheer size of the keyspace (2^128) means that rainbow tables can only ever exist for a subset of the keyspace. 1M machines running for a year each capable of doing 20M MD5 hashes a second will still only cover 1/539514153540300709th of the MD5 keyspace. But, yes, salting would prevent the use of rainbow tables completely.
  - Freaky 14y ago
    20M? Doesn't alter your point, but a pair of 3-gen-old ATI graphics cards manage on the order of 5000-6000M: Hash type: MD5, Hash: 6df23dc03f9b54cc38a0fc1483df6e21 Device #0: [RV870] 850.00 Mhz 1600 SP Device #1: [RV870] 850.00 Mhz 1600 SP CURPWD: @uX5G DONE: 22.30% ETA: 1m 34s CURSPD: 6436.3M=3217.8M+3218.6M http://www.golubev.com/hashgpu.htm http://www.golubev.com/hashgpu.htm
  - duaneb 14y ago
    This doesn't really change your statement, but 20M MD5 hashes/second is pretty damn wimpy these days. IIRC 20 BILLION is more accurate.