So recently we added support for Lotus Notes/Domino 8 hashes (-m 9100).
Since we know that many of you are interested in the details about this hash algorithm, I will try to explain the algo here.
It seems that for the same reason as for -m 8700 = Lotus Notes/Domino 6, i.e. mainly backwards-compatibility, the algo somehow still depends on (both of) the older hash types:
-m 8600 = Lotus Notes/Domino 5 aka SEC_pwddigest_V1 , start ([0-F]
-m 8700 = Lotus Notes/Domino 6 aka SEC_pwddigest_V2 , start (G
Basicallly you need to generate those digests/hashes first and continue w/ the new algorithm starting from the -m 8700 encoded hash.
The encoding too is the same as for the other 2 algos, but the new hashes (following SEC_pwddigest_V3) are longer and starts w/ '(H' (an example is on the wiki page: http://hashcat.net/wiki/doku.php?id=example_hashes ).
Furthermore, the hashes themself encode the following information:
- 16 byte salt (first 5 needed for -m 8700, remember -m 8600 is unsalted)
- round number (length 10, in ascii)
- 2 additional chars
- 8 bytes (a part of the) digest
So we start to generate the -m 8700 hash with the first 5 bytes of the salt.
After that PBKDF2-HMACSHA1 is used w/ the now generated "(G[known "small" salt]...)" hash as password, the full 16 bytes salt and the round number found in the encoded hash.
From the full digest (output of PBKDF2), only the first 8 bytes of the digest are then used in the final encoding step.
Attached you will find POC code I wrote down very quickly (I can't promise that it has no bugs and is easily readable![Tongue]( "Tongue")
).
Algo design problems one should think about:
As hash crackers and password experts we of course need/should also analyze how good/bad such an algo is...
and we identified several possible problems that could/should be considered when designing such an algorithm.
1. The password for the PBKDF2 round is partially known, we know quite a bit of that "password", we know the "small" salt, we know that it starts w/ "(G" and ends w/ ")",
furthermore we of course know salt and rounds from the encoded hash.
So, while it is still quite difficult to brute-force the last step (i.e. only the PBKDF2 part, remember we know 8 byte digest and "only" part of "(G[salt]...,)" is known), the actual keyspace is reduced heavily (total keyspace would be 64^22 w/ pass length 22 - with base64 restriction - , but it is reduced further to 64^(22-9))
2. The other problem we immediately need to think when seeing this reduced digest is, that collissions become much more likely than w/ the full digest in the hash (of course we have salts etc, but it may be considered uneccessary to cut the digest here).
3. This may be considered as a minor problem, but some other (more secure algos?) mix for instance part of the user defined input again, when it comes to e.g. the last step (e.g. the PBKDF2 part), s.t. it is much more difficult to "reverse" the hash step by step.
Consider for a moment that you have cracked the last part (very unlikely, right?) and know the correct and full "(G...)" hash (the "second password"), now you can forget about PBKDF2 and crack the -m 8700 hash on its own).
Said this, of course we know that it is still too difficult to reverse/bf the single steps (and therefore oclHashcat also needs to take all 3 steps anyway), but this may be some of the problems (please complete/correct the list with a reply if you know of any other possible problem) of this -m 9100 algo that we and hopefully other security experts may discuss about.
Special thanks go as always to atom for helping me with this discoveries etc.
CODE (don't blame me for this quickly hacked down code):
P.S. I've forgot a very important detail, as far as we know, oclHashcat is the first cracker that supports this algo, see also https://twitter.com/hashcat/status/490231216885735424
Since we know that many of you are interested in the details about this hash algorithm, I will try to explain the algo here.
It seems that for the same reason as for -m 8700 = Lotus Notes/Domino 6, i.e. mainly backwards-compatibility, the algo somehow still depends on (both of) the older hash types:
-m 8600 = Lotus Notes/Domino 5 aka SEC_pwddigest_V1 , start ([0-F]
-m 8700 = Lotus Notes/Domino 6 aka SEC_pwddigest_V2 , start (G
Basicallly you need to generate those digests/hashes first and continue w/ the new algorithm starting from the -m 8700 encoded hash.
The encoding too is the same as for the other 2 algos, but the new hashes (following SEC_pwddigest_V3) are longer and starts w/ '(H' (an example is on the wiki page: http://hashcat.net/wiki/doku.php?id=example_hashes ).
Furthermore, the hashes themself encode the following information:
- 16 byte salt (first 5 needed for -m 8700, remember -m 8600 is unsalted)
- round number (length 10, in ascii)
- 2 additional chars
- 8 bytes (a part of the) digest
So we start to generate the -m 8700 hash with the first 5 bytes of the salt.
After that PBKDF2-HMACSHA1 is used w/ the now generated "(G[known "small" salt]...)" hash as password, the full 16 bytes salt and the round number found in the encoded hash.
From the full digest (output of PBKDF2), only the first 8 bytes of the digest are then used in the final encoding step.
Attached you will find POC code I wrote down very quickly (I can't promise that it has no bugs and is easily readable
).Algo design problems one should think about:
As hash crackers and password experts we of course need/should also analyze how good/bad such an algo is...
and we identified several possible problems that could/should be considered when designing such an algorithm.
1. The password for the PBKDF2 round is partially known, we know quite a bit of that "password", we know the "small" salt, we know that it starts w/ "(G" and ends w/ ")",
furthermore we of course know salt and rounds from the encoded hash.
So, while it is still quite difficult to brute-force the last step (i.e. only the PBKDF2 part, remember we know 8 byte digest and "only" part of "(G[salt]...,)" is known), the actual keyspace is reduced heavily (total keyspace would be 64^22 w/ pass length 22 - with base64 restriction - , but it is reduced further to 64^(22-9))
2. The other problem we immediately need to think when seeing this reduced digest is, that collissions become much more likely than w/ the full digest in the hash (of course we have salts etc, but it may be considered uneccessary to cut the digest here).
3. This may be considered as a minor problem, but some other (more secure algos?) mix for instance part of the user defined input again, when it comes to e.g. the last step (e.g. the PBKDF2 part), s.t. it is much more difficult to "reverse" the hash step by step.
Consider for a moment that you have cracked the last part (very unlikely, right?) and know the correct and full "(G...)" hash (the "second password"), now you can forget about PBKDF2 and crack the -m 8700 hash on its own).
Said this, of course we know that it is still too difficult to reverse/bf the single steps (and therefore oclHashcat also needs to take all 3 steps anyway), but this may be some of the problems (please complete/correct the list with a reply if you know of any other possible problem) of this -m 9100 algo that we and hopefully other security experts may discuss about.
Special thanks go as always to atom for helping me with this discoveries etc.
CODE (don't blame me for this quickly hacked down code):
Code:
#!/usr/bin/env perl
#
# Author: philsmd
# Released to the public domain
# Special thx to atom for his help
#
use strict;
use warnings;
use Crypt::PBKDF2;
#
# Example:
# pass: "hashcat" (=> -m 8600 = 3DD2E1E5AC03E230243D58B8C5ADA076 - remember NO SALT, therefore always the same)
# random bytes: DA D3 E8 93 40 14 AC 9B 91 1E 42 A3 8A 61 EE 88 (len () = 16)
# -m 8700: (GsjFebq3Wtc0qppQH+aT)
# iterations: 5000
# new hash: (HsjFebq0Kh9kH7aAZYc7kY30mC30mC3KmC30mCluagXrvWKj1)
#
my $word_buf;
my $hash_target;
# example 1
$word_buf = "hashcat";
$hash_target = "(HsjFebq0Kh9kH7aAZYc7kY30mC30mC3KmC30mCluagXrvWKj1)";
# example 2
#$word_buf = "pleaseletmein";
#$hash_target = "(HosOQowHtnaYQqFo/XlScup0mC30mC3KmC30mCeACAxpjQN2u)";
# example 3
#$word_buf = "hashcatrocks!";
#$hash_target = "(HoDO+09St57fgxQ0xBE/1+p0mC30mC3KmC30mCYAgjrdL28sv)";
# example 4
#$word_buf ="hashcat";
#$hash_target = "(H7b36MY73EvYe15ugumPD2p0mC30mCJ0mC30mChXS0YQeAAmy)";
# example 5
#$word_buf = "hashcatpass";
#$hash_target = "(Hg5o/maaMbKGmoOvMLA4om30mC30mCJ8pD3KmCYi8pjK0a508)";
# example - to crack
#$word_buf = $ARGV[0];
#$hash_target = $ARGV[1];
exit (1) if (! $word_buf);
#
# lotus magic table
#
my @lotus_magic_table =
(
0xbd, 0x56, 0xea, 0xf2, 0xa2, 0xf1, 0xac, 0x2a,
0xb0, 0x93, 0xd1, 0x9c, 0x1b, 0x33, 0xfd, 0xd0,
0x30, 0x04, 0xb6, 0xdc, 0x7d, 0xdf, 0x32, 0x4b,
0xf7, 0xcb, 0x45, 0x9b, 0x31, 0xbb, 0x21, 0x5a,
0x41, 0x9f, 0xe1, 0xd9, 0x4a, 0x4d, 0x9e, 0xda,
0xa0, 0x68, 0x2c, 0xc3, 0x27, 0x5f, 0x80, 0x36,
0x3e, 0xee, 0xfb, 0x95, 0x1a, 0xfe, 0xce, 0xa8,
0x34, 0xa9, 0x13, 0xf0, 0xa6, 0x3f, 0xd8, 0x0c,
0x78, 0x24, 0xaf, 0x23, 0x52, 0xc1, 0x67, 0x17,
0xf5, 0x66, 0x90, 0xe7, 0xe8, 0x07, 0xb8, 0x60,
0x48, 0xe6, 0x1e, 0x53, 0xf3, 0x92, 0xa4, 0x72,
0x8c, 0x08, 0x15, 0x6e, 0x86, 0x00, 0x84, 0xfa,
0xf4, 0x7f, 0x8a, 0x42, 0x19, 0xf6, 0xdb, 0xcd,
0x14, 0x8d, 0x50, 0x12, 0xba, 0x3c, 0x06, 0x4e,
0xec, 0xb3, 0x35, 0x11, 0xa1, 0x88, 0x8e, 0x2b,
0x94, 0x99, 0xb7, 0x71, 0x74, 0xd3, 0xe4, 0xbf,
0x3a, 0xde, 0x96, 0x0e, 0xbc, 0x0a, 0xed, 0x77,
0xfc, 0x37, 0x6b, 0x03, 0x79, 0x89, 0x62, 0xc6,
0xd7, 0xc0, 0xd2, 0x7c, 0x6a, 0x8b, 0x22, 0xa3,
0x5b, 0x05, 0x5d, 0x02, 0x75, 0xd5, 0x61, 0xe3,
0x18, 0x8f, 0x55, 0x51, 0xad, 0x1f, 0x0b, 0x5e,
0x85, 0xe5, 0xc2, 0x57, 0x63, 0xca, 0x3d, 0x6c,
0xb4, 0xc5, 0xcc, 0x70, 0xb2, 0x91, 0x59, 0x0d,
0x47, 0x20, 0xc8, 0x4f, 0x58, 0xe0, 0x01, 0xe2,
0x16, 0x38, 0xc4, 0x6f, 0x3b, 0x0f, 0x65, 0x46,
0xbe, 0x7e, 0x2d, 0x7b, 0x82, 0xf9, 0x40, 0xb5,
0x1d, 0x73, 0xf8, 0xeb, 0x26, 0xc7, 0x87, 0x97,
0x25, 0x54, 0xb1, 0x28, 0xaa, 0x98, 0x9d, 0xa5,
0x64, 0x6d, 0x7a, 0xd4, 0x10, 0x81, 0x44, 0xef,
0x49, 0xd6, 0xae, 0x2e, 0xdd, 0x76, 0x5c, 0x2f,
0xa7, 0x1c, 0xc9, 0x09, 0x69, 0x9a, 0x83, 0xcf,
0x29, 0x39, 0xb9, 0xe9, 0x4c, 0xff, 0x43, 0xab
);
#
# start:
#
# decode
my $base64_part = substr ($hash_target, 2, -1);
my ($digest_target, $salt_buf, $iterations, $chars) = domino_85x_decode ($base64_part);
exit (1) if ($iterations < 1);
#
# generate the digests (i.e. check/crack it)
#
# domino 5 hash - SEC_pwddigest_V1 - -m 8600
my @saved_key = map { ord $_; } split "", $word_buf;
my $len = scalar @saved_key;
my @state = domino_big_md (\@saved_key, $len);
# domino 6 hash - SEC_pwddigest_V2 - -m 8700
my $domino_char = undef;
my $salt_part = substr ($salt_buf, 0, 5);
my $str = "(" . unpack ("H*", join ("", (map { chr $_; } @state))) . ")";
@saved_key = map { ord $_; } split "", $salt_part . uc $str;
@state = domino_big_md (\@saved_key, 34);
my $hash_buf = join ("", (map { chr $_; } @state));
my $tmp_hash = sprintf ('(G%s)', domino_encode ($salt_part . $hash_buf, $domino_char));
# domino 8(.5.x) hash - SEC_pwddigest_V3 - -m 9100
my $pbkdf2 = Crypt::PBKDF2->new
(
hash_class => 'HMACSHA1',
iterations => $iterations,
output_len => 8,
salt_len => 16,
);
my $digest_new = $pbkdf2->PBKDF2 ($salt_buf, $tmp_hash);
# done
# done
# done
# print only
my $digest_new_hex = unpack ("H*", $digest_new);
my $digest_target_hex = unpack ("H*", $digest_target);
print "Generated digest: " . $digest_new_hex . "\n";
print "Target digest...: " . $digest_target_hex . "\n\n";
if ($digest_new_hex eq $digest_target_hex)
{
print "SUCCESS!\n";
print ".. to exaggerate I will re-generate the SEC_pwddigest_V3 encoded hash starting from the digest:\n\n";
# just padding for the $iteration part - use a build-in perl function if you would like
my $iteration_str = "" . $iterations;
for (my $i = length ($iterations); $i < 10; $i++)
{
$iterations = "0" . $iterations;
}
$tmp_hash = sprintf ('(H%s)', domino_85x_encode ($salt_buf . $iterations . $chars . $digest_new, $domino_char));
print "Generated \"hash\": " . $tmp_hash . "\n";
print "Target \"hash\"...: " . $hash_target . "\n";
exit (0);
}
else
{
print "FAILED!\n";
exit (1);
}
#
# helper functions only
#
sub domino_base64_decode
{
my $v = shift;
my $n = shift;
my $itoa64 = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz+/";
my $ret = 0;
my $i = 1;
while ($i <= $n)
{
my $idx = (index ($itoa64, substr ($v, $n - $i, 1))) & 0x3f;
$ret += ($idx << (6 * ($i - 1)));
$i = $i + 1;
}
return $ret
}
sub domino_85x_decode
{
my $str = shift;
my $decoded = "";
for (my $i = 0; $i < length ($str); $i += 4)
{
my $num = domino_base64_decode (substr ($str, $i, 4), 4);
$decoded .= chr (($num >> 16) & 0xff) . chr (($num >> 8) & 0xff) . chr ($num & 0xff);
}
my $digest;
my $salt;
my $iterations = -1;
my $chars;
$salt = substr ($decoded, 0, 16); # longer than -m 8700 (5 vs 16 <- new)
my $byte10 = (ord (substr ($salt, 3, 1)) - 4);
if ($byte10 < 0)
{
$byte10 = 256 + $byte10;
}
substr ($salt, 3, 1) = chr ($byte10);
if ($iterations =~ /^?d*$/)
{
# continue
$iterations = substr ($decoded, 16, 10);
$iterations = $iterations + 0; # hack: make sure it is an int now (atoi ())
$chars = substr ($decoded, 26, 2); # in my example it is "02"
$digest = substr ($decoded, 28, 8); # only of length of 8 vs 20 SHA1 bytes
}
return ($digest, $salt, $iterations, $chars);
}
sub domino_base64_encode
{
my $v = shift;
my $n = shift;
my $itoa64 = "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz+/";
my $ret = "";
while (($n - 1) >= 0)
{
$n = $n - 1;
$ret = substr ($itoa64, $v & 0x3f, 1) . $ret;
$v = $v >> 6;
}
return $ret
}
sub domino_encode
{
my $final = shift;
my $char = shift;
my $byte10 = (ord (substr ($final, 3, 1)) + 4);
if ($byte10 > 255)
{
$byte10 = $byte10 - 256;
}
substr ($final, 3, 1) = chr ($byte10);
my $passwd = "";
$passwd .= domino_base64_encode ((int (ord (substr ($final, 0, 1))) << 16) | (int (ord (substr ($final, 1, 1))) << 8) | (int (ord (substr ($final, 2, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 3, 1))) << 16) | (int (ord (substr ($final, 4, 1))) << 8) | (int (ord (substr ($final, 5, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 6, 1))) << 16) | (int (ord (substr ($final, 7, 1))) << 8) | (int (ord (substr ($final, 8, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 9, 1))) << 16) | (int (ord (substr ($final, 10, 1))) << 8) | (int (ord (substr ($final, 11, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 12, 1))) << 16) | (int (ord (substr ($final, 13, 1))) << 8) | (int (ord (substr ($final, 14, 1)))), 4);
if (defined ($char))
{
substr ($passwd, 18, 1) = $char;
}
substr ($passwd, 19, 1) = "";
return $passwd;
}
sub domino_85x_encode
{
my $final = shift;
my $char = shift;
my $byte10 = (ord (substr ($final, 3, 1)) + 4);
if ($byte10 > 255)
{
$byte10 = $byte10 - 256;
}
substr ($final, 3, 1) = chr ($byte10);
my $passwd = "";
$passwd .= domino_base64_encode ((int (ord (substr ($final, 0, 1))) << 16) | (int (ord (substr ($final, 1, 1))) << 8) | (int (ord (substr ($final, 2, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 3, 1))) << 16) | (int (ord (substr ($final, 4, 1))) << 8) | (int (ord (substr ($final, 5, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 6, 1))) << 16) | (int (ord (substr ($final, 7, 1))) << 8) | (int (ord (substr ($final, 8, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 9, 1))) << 16) | (int (ord (substr ($final, 10, 1))) << 8) | (int (ord (substr ($final, 11, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 12, 1))) << 16) | (int (ord (substr ($final, 13, 1))) << 8) | (int (ord (substr ($final, 14, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 15, 1))) << 16) | (int (ord (substr ($final, 16, 1))) << 8) | (int (ord (substr ($final, 17, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 18, 1))) << 16) | (int (ord (substr ($final, 19, 1))) << 8) | (int (ord (substr ($final, 20, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 21, 1))) << 16) | (int (ord (substr ($final, 22, 1))) << 8) | (int (ord (substr ($final, 23, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 24, 1))) << 16) | (int (ord (substr ($final, 25, 1))) << 8) | (int (ord (substr ($final, 26, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 27, 1))) << 16) | (int (ord (substr ($final, 28, 1))) << 8) | (int (ord (substr ($final, 29, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 30, 1))) << 16) | (int (ord (substr ($final, 31, 1))) << 8) | (int (ord (substr ($final, 32, 1)))), 4);
$passwd .= domino_base64_encode ((int (ord (substr ($final, 33, 1))) << 16) | (int (ord (substr ($final, 34, 1))) << 8) | (int (ord (substr ($final, 35, 1)))), 4);
if (defined ($char))
{
substr ($passwd, 18, 1) = $char;
}
# substr ($passwd, 19, 1) = "";
return $passwd;
}
sub pad16
{
my $block_ref = shift;
my $offset = shift;
my $value = 16 - $offset;
for (my $i = $offset; $i < 16; $i++)
{
push @{$block_ref}, $value;
}
}
sub lotus_mix
{
my $in_ref = shift;
my $p = 0;
for (my $i = 0; $i < 18; $i++)
{
for (my $j = 0; $j < 48; $j++)
{
$p = ($p + 48 - $j) & 0xff;
my $c = $lotus_magic_table[$p];
$p = $in_ref->[$j] ^ $c;
$in_ref->[$j] = $p;
}
}
}
sub lotus_transform_password
{
my $in_ref = shift;
my $out_ref = shift;
my $t = $out_ref->[15];
for (my $i = 0; $i < 16; $i++)
{
$t ^= $in_ref->[$i];
my $c = $lotus_magic_table[$t];
$out_ref->[$i] ^= $c;
$t = $out_ref->[$i];
}
}
sub mdtransform_norecalc
{
my $state_ref = shift;
my $block_ref = shift;
my @x;
push (@x, @{$state_ref});
push (@x, @{$block_ref});
for (my $i = 0; $i < 16; $i++)
{
push (@x, $x[0 + $i] ^ $x[16 + $i]);
}
lotus_mix (\@x);
for (my $i = 0; $i < 16; $i++)
{
$state_ref->[$i] = $x[$i];
}
}
sub mdtransform
{
my $state_ref = shift;
my $checksum_ref = shift;
my $block_ref = shift;
mdtransform_norecalc ($state_ref, $block_ref);
lotus_transform_password ($block_ref, $checksum_ref);
}
sub domino_big_md
{
my $saved_key_ref = shift;
my $size = shift;
@{$saved_key_ref} = splice (@{$saved_key_ref}, 0, $size);
my @state = (0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0);
my @checksum;
my $curpos;
for ($curpos = 0; $curpos + 16 < $size; $curpos += 16)
{
my $curpos16 = $curpos + 16;
my @block = splice (@{$saved_key_ref}, 0, 16);
mdtransform (\@state, \@checksum, \@block);
}
my $left = $size - $curpos;
my @block = splice (@{$saved_key_ref}, 0, 16);
pad16 (\@block, $left);
mdtransform (\@state, \@checksum, \@block);
mdtransform_norecalc (\@state, \@checksum);
return @state;
}P.S. I've forgot a very important detail, as far as we know, oclHashcat is the first cracker that supports this algo, see also https://twitter.com/hashcat/status/490231216885735424