lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Sat, 11 Jan 2014 09:09:11 -0600 (CST)
From: Steve Thomas <steve@...tu.com>
To: discussions@...sword-hashing.net
Subject: Scripting memory (not so) high vs Catena in PHP (with
 optimizations)

I wanted to get benchmarks to compare a hashing algorithm optimized for a
scripting language vs another algorithm that did not consider this. Since I'm
familiar with Catena I decided to implement Catena-3-SHA512 in PHP.

Current scripting memory (not so) high vs current Catena:
2 MiB: 544 ms vs 2030 ms (3.73x)
1 MiB: 249 ms vs 1040 ms (4.17x)
512 KiB: 124 ms vs 514 ms (4.13x)
256 KiB: 60.5 ms vs 240 ms (3.96x)
128 KiB: 30.0 ms vs 118 ms (3.94x)
Basically scripting memory (not so) high can use 4x more ram than Catena and run
in about the same time.

Optimized scripting memory (not so) high vs optimized Catena:
2 MiB: 431 ms vs 995 ms (2.31x)
1 MiB: 195 ms vs 499 ms (2.56x)
512 KiB: 97.0 ms vs 249 ms (2.57x)
256 KiB: 48.5 ms vs 121 ms (2.49x)
128 KiB: 24.0 ms vs 59.7 ms (2.48x)

Catena takes about 2.5 times longer for the same amount of memory, but Catena is
much better at defending against attackers that use less memory. Pretty sure
it's O(N^2) vs O(N).

Test machine: Core2 Quad (Q9300 2.5GHz), PHP 5.3.4 (64 bit), Windows 7


*** Optimizations ***

Scripting memory (not so) high:
        $ja = unpack('V', $x);
vs
        $ja = unpack('V', substr($x, -4));
The optimized version's look ups are aligned to 384 bytes with a $blocksize of
384 bytes vs byte aligned with a $blocksize of 431 bytes. Note both require four
SHA512 blocks.

Catena:
    for ($c = $garlic; $c <= $garlic; $c++)
vs
    for ($c = 1; $c <= $garlic; $c++)
2x increase in speed by not calculating the smaller sizes. Note that you could
have "minGarlic" and "maxGarlic". To save on calculations and still be
upgradeable.


*** Code ***

Scripting memory (not so) high:
function smhkdf($in, $salt, $m_cost)
{
    $k = 4;
    $blocksize = 128 * $k - 16 - 1 - 64;

    if ($m_cost * 64 < $blocksize)
        $m_cost = ceil($blocksize / 64);

    $v = '';
    $x = hash('sha512', $in, TRUE) . $salt;
    $count = $m_cost;
    do {
        $v .= $x = hash('sha512', $x, TRUE);
    } while (--$count);

    $mod = ($m_cost * 64) - $blocksize + 1;
    $count = $m_cost;
    do {
        $ja = unpack('V', substr($x, -4));
# XXX: optionally align $j to 64-bit boundary
        $x = hash('sha512', $x . substr($v, ($ja[1] & 0x7fffffff) % $mod,
$blocksize), TRUE);
    } while (--$count);

# XXX: use $outlen
# XXX: use $t_cost (as parallelism?)
    $x = substr($x, -32);

    return $x;
}

function smhkdfOptimized($in, $salt, $m_cost)
{
    $k = 4; // Can't change, this was unrolled
    $blocks = 2 * $k - 2;
    $m_cost = ceil($m_cost / $blocks);

    $v = array();
    $x = hash('sha512', $in, TRUE) . $salt;
    $count = $m_cost;
    do {
        // $blocks == 6
        $x0 = hash('sha512', $x,  TRUE);
        $x1 = hash('sha512', $x0, TRUE);
        $x2 = hash('sha512', $x1, TRUE);
        $x3 = hash('sha512', $x2, TRUE);
        $x4 = hash('sha512', $x3, TRUE);
        $x  = hash('sha512', $x4, TRUE);
        $v[] = $x0 . $x1 . $x2 . $x3 . $x4 . $x;
    } while (--$count);
    $count = $blocks * $m_cost;
    do {
        $ja = unpack('V', $x);
        $x = hash('sha512', $x . $v[($ja[1] & 0x7fffffff) % $m_cost], TRUE);
    } while (--$count);

# XXX: use $outlen
# XXX: use $t_cost (as parallelism?)
    $x = substr($x, -32);

    return $x;
}

Catena-3-SHA512:
define('H_LEN',  64);
define('LAMBDA',  3);
define('REGULAR', 0);
define('CLIENT',  1);
define('PASSWORD_HASHING_MODE', 0);
define('KEY_DERIVATION_MODE',   1);

function LBRH($x, $garlic)
{
    $c = 1 << $garlic;
    $r = array();

    $header = chr($garlic) . "\0\0\0\0";
    $r[] = hash('sha512', $header . "\0\0\0\0" . $x, true);

    // Top row
    for ($i = 1; $i < $c; $i++)
    {
        $r[] = hash('sha512', $header . pack('V', $i) . $r[$i - 1], true);
    }

    // Mid rows
    $pShift = 24 - $garlic;
    for ($k = 0; $k < LAMBDA; $k++)
    {
        $r[0] = hash('sha512', $header . "\0\0\0\0" . $r[0] . $r[$c - 1], true);

        // Replace $r[reverse($i, $garlic)] with new value
        $previousR = 0;
        for ($i = 1; $i < $c; $i++)
        {
            // $p = reverse($i, $garlic);
            $p = (($i & 0x0000ff) << 16) | (($i & 0xff0000) >> 16) | ($i &
0x00ff00);
            $p = (($p & 0x0f0f0f) <<  4) | (($p & 0xf0f0f0) >>  4);
            $p = (($p & 0x333333) <<  2) | (($p & 0xcccccc) >>  2);
            $p = (($p & 0x555555) <<  1) | (($p & 0xaaaaaa) >>  1);
            $p = $p >> $pShift;
            $r[$p] = hash('sha512', $header . pack('V', $i) . $r[$previousR] .
$r[$p], true);
            $previousR = $p;
        }
        $k++;
        if ($k >= LAMBDA)
        {
            break;
        }

        // This is now sequential because (reverse(reverse($i, $garlic),
$garlic) == $i)
        $r[0] = hash('sha512', $header . "\0\0\0\0" . $r[0] . $r[$c - 1], true);
        for ($i = 1; $i < $c; $i++)
        {
            $r[$i] = hash('sha512', $header . pack('V', $i) . $r[$i - 1] .
$r[$i], true);
        }
    }

    // reverse($c - 1, $garlic) == $c - 1
    return $r[$c - 1];
}

function catenaOptimized($pwd, $salt, $data, $garlic, $client = REGULAR,
$tweak_id = PASSWORD_HASHING_MODE, $hashlen = H_LEN)
{
    if ($hashlen > H_LEN || $garlic > 24)
    {
        return false;
    }

    // Compute H(AD)
    $x = hash('sha512', $data, true);

    // Compute the initial value to hash
    $x = hash('sha512', "\xff" . chr($tweak_id) . chr(LAMBDA) . chr($hashlen) .
chr(strlen($salt)) . $x . $pwd . $salt, true);
    for ($i = $hashlen; $i < H_LEN; $i++)
    {
        $x[$i] = 0;
    }

    for ($c = $garlic; $c <= $garlic; $c++)
    {
        $x = LBRH($x, $c);
        if ($c == $garlic && $client == CLIENT)
        {
            return $x;
        }

        $x = hash('sha512', chr($c) . "\0\0\0\0" . pack('V', (LAMBDA + 1) << $c)
. $x, true);
        for ($i = $hashlen; $i < H_LEN; $i++)
        {
            $x[$i] = 0;
        }
    }
    return substr($x, 0, $hashlen);
}

function catena($pwd, $salt, $data, $garlic, $client = REGULAR, $tweak_id =
PASSWORD_HASHING_MODE, $hashlen = H_LEN)
{
    if ($hashlen > H_LEN || $garlic > 24)
    {
        return false;
    }

    // Compute H(AD)
    $x = hash('sha512', $data, true);

    // Compute the initial value to hash
    $x = hash('sha512', "\xff" . chr($tweak_id) . chr(LAMBDA) . chr($hashlen) .
chr(strlen($salt)) . $x . $pwd . $salt, true);
    for ($i = $hashlen; $i < H_LEN; $i++)
    {
        $x[$i] = 0;
    }

    for ($c = 1; $c <= $garlic; $c++)
    {
        $x = LBRH($x, $c);
        if ($c == $garlic && $client == CLIENT)
        {
            return $x;
        }

        $x = hash('sha512', chr($c) . "\0\0\0\0" . pack('V', (LAMBDA + 1) << $c)
. $x, true);
        for ($i = $hashlen; $i < H_LEN; $i++)
        {
            $x[$i] = 0;
        }
    }
    return substr($x, 0, $hashlen);
}

* Note I may have messed something up. Please check for errors.
Content of type "text/html" skipped

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ