Req #62372 [Opn->Ver]: crypt() and broken backwards compability in SHA rounds
| From: | yohgaki@php.net | Date: | Tue, 15 Jul 2014 22:47:35 +0000 |
| Subject: | Req #62372 [Opn->Ver]: crypt() and broken backwards compability in SHA rounds | ||
| References: | 1 | Groups: | php.bugs |
| Request: | Send a blank email to php-bugs+get-186650@lists.php.net to get a copy of this message | ||
Edit report at https://bugs.php.net/bug.php?id=62372&edit=1
ID: 62372
Updated by: yohgaki@php.net
Reported by: jani dot ollikainen at mmd dot net
Summary: crypt() and broken backwards compability in SHA
rounds
-Status: Open
+Status: Verified
Type: Feature/Change Request
Package: *Encryption and hash functions
-Operating System:
+Operating System: *
-PHP Version: 5.4.4
+PHP Version: *
Block user comment: N
Private report: N
New Comment:
http://3v4l.org/WJPSl
HASH:
$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
- CRYPT:
$6$rounds=1000$qNElXs2yMnL2.GNS$/q7trYkbKkoJernsumbObt2IysdXGRx/ytFaG0HBC97rHHhYRQvUcyEuRHP6h5yj8V.fH7XKEw5hjofVmYONw1
NO MATCH
rounds became 10 to 1000.
Previous Comments:
------------------------------------------------------------------------
[2013-02-26 08:23:20] m dot staab at complex-it dot de
I have the very same issues. Updating from PHP5.2.4 to PHP5.4.11 and now my passwords which were
hased using PHP5.2.4 do not match any longer on PHP5.4.11.
I found no way how to re-produce passwords on PHP5.4.11 like they were generated on PHP5.2.4 and
therefore I have no way to upgrade users which already have a password to a newly generated with
PHP5.4.11..
The only way I can think of is a password-reset for all of our users?!
------------------------------------------------------------------------
[2012-06-20 20:33:17] jani dot ollikainen at mmd dot net
Oh now I get what I did wrong with PHP 5.1.6. It doesn't understand rounds at
all, it's just looks like it does, but it uses it as salt and new PHP doesn't
allow = in salt, so it doesn't work as they see different salt in it. So the
compatibility issue is with salt string, not in rounds at all.
This can be work around in PHP 5.1.6 not to use salt's which newer versions
don't understand, but if you already use them, then you're in big trouble.
Another one is that if I have password like: http://3v4l.org/amNND
The hash is generated with patched PHP 5.4.4 with ROUNDS_MIN 1, but it could be
from some other system. PHP cannot verify it, as the rounds limit 1000 is
enforced. To me enforcing it seems to limit compability.
------------------------------------------------------------------------
[2012-06-20 19:32:11] arjen at react dot com
Looks like PHP <= 5.1.7 has a limit on the saltstring of 9 chars.
If no $rounds is specified AND a salstring of 9 chars, all versions return the
same hash: http://3v4l.org/nDiFd (2nd line, 1st line is with
long hash).
------------------------------------------------------------------------
[2012-06-20 11:44:29] jani dot ollikainen at mmd dot net
Tested with patched 5.4.4 where:
#define ROUNDS_MIN 1
HASH:
$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
- CRYPT:
$6$rounds=10$qNElXs2yMnL2.GNS$YwaYQmhwsN2RzBImxlEjIL.0/YLlfYCDmyfozkCQWNKdKgZQtTpK3Y/TAw31deCnJrDzgrqpI6ckvJCBsoeNB/
NO MATCH
So problem isn't only in ROUNDS_MIN. Also I know that my hash hasn't $ after salt, but
that's how PHP 5.1.6 creates it. Tested also with adding $ and result is similar:
HASH:
$6$rounds=10$qNElXs2yMnL2.GNS$3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
- CRYPT:
$6$rounds=10$qNElXs2yMnL2.GNS$YwaYQmhwsN2RzBImxlEjIL.0/YLlfYCDmyfozkCQWNKdKgZQtTpK3Y/TAw31deCnJrDzgrqpI6ckvJCBsoeNB/
NO MATCH
So something else is also different...
------------------------------------------------------------------------
[2012-06-20 10:58:38] jani dot ollikainen at mmd dot net
Description:
------------
http://fi2.php.net/manual/en/function.crypt.php
"CRYPT_SHA512 - SHA-512 hash with a sixteen character salt prefixed with $6$. If the salt
string starts with 'rounds=<N>$', the numeric value of N is used to indicate how
many times the hashing loop should be executed, much like the cost parameter on Blowfish. The
default number of rounds is 5000, there is a minimum of 1000 and a maximum of 999,999,999. Any
selection of N outside this range will be truncated to the nearest limit."
Why is that N put to minium of 1000? Now if I've made hash with old PHP (or with anything else)
which has "$6$rounds=10$" I cannot check that, because if there's rounds mentioned
PHP will do minimum of 1000 rounds and I get totally different hash to compare.
Sounds very nice! I can understant that rounds limit if I don't give salt to crypt(), but if I
give a salt which has "$6$rounds=10$" it really should do just those 10 rounds!
Fails with:
PHP 5.4.4 (cli) (built: Jun 20 2012 13:48:48)
PHP 5.3.14 (cli) (built: Jun 20 2012 13:39:44)
PHP 5.3.3 (cli) (built: May 7 2012 19:58:17)
Works with:
PHP 5.1.6 (cli) (built: May 7 2012 15:03:06)
Test script:
---------------
<?php
$h='$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40';
$p='salasana';
$c=crypt($p,$h);
echo "HASH: $h - CRYPT: $c\n";
if ($c == $h)
{
echo "MATCH OK\n";
}
else
{
echo "NO MATCH\n";
}
?>
Expected result:
----------------
HASH:
$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
- CRYPT:
$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
MATCH OK
Actual result:
--------------
HASH:
$6$rounds=10$qNElXs2yMnL2.GNS3kiM7DqmGbFLdQfIwu2691aJgT3xgJazPLtw7RPKz3Dp8RIc4b5fmJ7qvlq/mPN8a.rE40
- CRYPT:
$6$rounds=1000$qNElXs2yMnL2.GNS$/q7trYkbKkoJernsumbObt2IysdXGRx/ytFaG0HBC97rHHhYRQvUcyEuRHP6h5yj8V.fH7XKEw5hjofVmYONw1
NO MATCH
------------------------------------------------------------------------
--
Edit this bug report at https://bugs.php.net/bug.php?id=62372&edit=1