 |
LXX. Mcrypt Encryption Functions
This is an interface to the mcrypt library, which supports a wide
variety of block algorithms such as DES, TripleDES, Blowfish
(default), 3-WAY, SAFER-SK64, SAFER-SK128, TWOFISH, TEA, RC2 and
GOST in CBC, OFB, CFB and ECB cipher modes. Additionally, it
supports RC6 and IDEA which are considered "non-free".
These functions work using mcrypt.
To use it, download libmcrypt-x.x.tar.gz from http://mcrypt.sourceforge.net/ and follow the included
installation instructions. Windows users will find all the
needed compiled mcrypt binaries at
http://ftp.emini.dk/pub/php/win32/mcrypt/.
As of PHP 5.0.0 you will need libmcrypt Version 2.5.6 or greater.
If you linked against libmcrypt 2.4.x or higher, the following additional
block algorithms are supported: CAST, LOKI97, RIJNDAEL, SAFERPLUS,
SERPENT and the following stream ciphers: ENIGMA (crypt), PANAMA, RC4 and
WAKE. With libmcrypt 2.4.x or higher another cipher mode is also
available; nOFB.
You need to compile PHP with the --with-mcrypt[=DIR] parameter to enable this
extension. DIR is the mcrypt install directory. Make sure you compile
libmcrypt with the option
--disable-posix-threads.
这些函数的行为受 php.ini 的影响。
表格 1. Mcrypt configuration options Name | Default | Changeable | Changelog |
---|
mcrypt.algorithms_dir | NULL | PHP_INI_ALL | Available since PHP 4.0.2. | mcrypt.modes_dir | NULL | PHP_INI_ALL | Available since PHP 4.0.2. |
有关 PHP_INI_* 常量进一步的细节与定义参见 附录 G。
以下常量由本扩展模块定义,因此只有在本扩展模块被编译到
PHP 中,或者在运行时被动态加载后才有效。
Mcrypt can operate in four block cipher modes (CBC, OFB, CFB, and
ECB). If linked against libmcrypt-2.4.x or higher the functions can also operate
in the block cipher mode nOFB and in STREAM mode. Below you find a list
with all supported encryption modes together with the constants that are
defines for the encryption mode. For a more complete reference and
discussion see Applied Cryptography by Schneier (ISBN 0-471-11709-9).
MCRYPT_MODE_ECB (electronic codebook) is suitable for random data,
such as encrypting other keys. Since data there is short and random,
the disadvantages of ECB have a favorable negative effect.
MCRYPT_MODE_CBC (cipher block chaining) is especially suitable for
encrypting files where the security is increased over ECB
significantly.
MCRYPT_MODE_CFB (cipher feedback) is the best mode for encrypting byte
streams where single bytes must be encrypted.
MCRYPT_MODE_OFB (output feedback, in 8bit) is comparable to CFB, but
can be used in applications where error propagation cannot
be tolerated. It's insecure (because it operates in 8bit
mode) so it is not recommended to use it.
MCRYPT_MODE_NOFB (output feedback, in nbit) is comparable to OFB, but
more secure because it operates on the block size of the algorithm.
MCRYPT_MODE_STREAM is an extra mode to include some stream algorithms
like WAKE or RC4.
Some other mode and random device constants:
Here is a list of ciphers which are currently supported by the mcrypt
extension. For a complete list of supported ciphers, see the defines at
the end of mcrypt.h. The general rule with the
mcrypt-2.2.x API is that you can access the cipher from PHP with
MCRYPT_ciphername. With the libmcrypt-2.4.x and libmcrypt-2.5.x API these constants also work,
but it is possible to specify the name of the cipher as a string with a
call to mcrypt_module_open().
MCRYPT_3DES MCRYPT_ARCFOUR_IV (libmcrypt > 2.4.x only) MCRYPT_ARCFOUR (libmcrypt > 2.4.x only) MCRYPT_BLOWFISH MCRYPT_CAST_128 MCRYPT_CAST_256 MCRYPT_CRYPT MCRYPT_DES MCRYPT_DES_COMPAT (libmcrypt 2.2.x only) MCRYPT_ENIGMA (libmcrypt > 2.4.x only, alias for MCRYPT_CRYPT) MCRYPT_GOST MCRYPT_IDEA (non-free) MCRYPT_LOKI97 (libmcrypt > 2.4.x only) MCRYPT_MARS (libmcrypt > 2.4.x only, non-free) MCRYPT_PANAMA (libmcrypt > 2.4.x only) MCRYPT_RIJNDAEL_128 (libmcrypt > 2.4.x only) MCRYPT_RIJNDAEL_192 (libmcrypt > 2.4.x only) MCRYPT_RIJNDAEL_256 (libmcrypt > 2.4.x only) MCRYPT_RC2 MCRYPT_RC4 (libmcrypt 2.2.x only) MCRYPT_RC6 (libmcrypt > 2.4.x only) MCRYPT_RC6_128 (libmcrypt 2.2.x only) MCRYPT_RC6_192 (libmcrypt 2.2.x only) MCRYPT_RC6_256 (libmcrypt 2.2.x only) MCRYPT_SAFER64 MCRYPT_SAFER128 MCRYPT_SAFERPLUS (libmcrypt > 2.4.x only) MCRYPT_SERPENT(libmcrypt > 2.4.x only) MCRYPT_SERPENT_128 (libmcrypt 2.2.x only) MCRYPT_SERPENT_192 (libmcrypt 2.2.x only) MCRYPT_SERPENT_256 (libmcrypt 2.2.x only) MCRYPT_SKIPJACK (libmcrypt > 2.4.x only) MCRYPT_TEAN (libmcrypt 2.2.x only) MCRYPT_THREEWAY MCRYPT_TRIPLEDES (libmcrypt > 2.4.x only) MCRYPT_TWOFISH (for older mcrypt 2.x versions, or mcrypt > 2.4.x ) MCRYPT_TWOFISH128 (TWOFISHxxx are available in newer 2.x versions, but not in the 2.4.x versions) MCRYPT_TWOFISH192 MCRYPT_TWOFISH256 MCRYPT_WAKE (libmcrypt > 2.4.x only) MCRYPT_XTEA (libmcrypt > 2.4.x only)
You must (in CFB and OFB mode) or can (in CBC mode) supply an
initialization vector (IV) to the respective cipher function. The
IV must be unique and must be the same when
decrypting/encrypting. With data which is stored encrypted, you
can take the output of a function of the index under which the
data is stored (e.g. the MD5 key of the filename).
Alternatively, you can transmit the IV together with the encrypted
data (see chapter 9.3 of Applied Cryptography by Schneier (ISBN 0-471-11709-9) for a
discussion of this topic).
Mcrypt can be used to encrypt and decrypt using the above
mentioned ciphers. If you linked against libmcrypt-2.2.x, the
four important mcrypt commands (mcrypt_cfb(),
mcrypt_cbc(), mcrypt_ecb(),
and mcrypt_ofb()) can operate in both modes
which are named MCRYPT_ENCRYPT and MCRYPT_DECRYPT, respectively.
例子 1. Encrypt an input value with TripleDES under 2.2.x in ECB mode
<?php $key = "this is a secret key"; $input = "Let us meet at 9 o'clock at the secret place.";
$encrypted_data = mcrypt_ecb (MCRYPT_3DES, $key, $input, MCRYPT_ENCRYPT); ?>
|
|
This example will give you the encrypted data as a string in
$encrypted_data.
If you linked against libmcrypt 2.4.x or 2.5.x, these functions are still
available, but it is recommended that you use the advanced functions.
例子 2. Encrypt an input value with TripleDES under 2.4.x and higher in ECB mode
<?php $key = "this is a secret key"; $input = "Let us meet at 9 o'clock at the secret place.";
$td = mcrypt_module_open('tripledes', '', 'ecb', ''); $iv = mcrypt_create_iv (mcrypt_enc_get_iv_size($td), MCRYPT_RAND); mcrypt_generic_init($td, $key, $iv); $encrypted_data = mcrypt_generic($td, $input); mcrypt_generic_deinit($td); mcrypt_module_close($td); ?>
|
|
This example will give you the encrypted data as a string in
$encrypted_data. For a full example see
mcrypt_module_open().
rolf at winmutt dot com
11-Mar-2006 04:19
mysql AES_ENCRYPT() compatibly function for PHP :
function mysql_aes_encrypt($val,$ky) {
$mode=MCRYPT_MODE_ECB;
$enc=MCRYPT_RIJNDAEL_128;
$val=str_pad($val, (16*(floor(strlen($val) / 16)+(strlen($val) % 16==0?2:1))), chr(16-(strlen($val) % 16)));
return mcrypt_encrypt($enc, $ky, $val, $mode, mcrypt_create_iv( mcrypt_get_iv_size($enc, $mode), MCRYPT_DEV_URANDOM));
}
Please note that if the strlen($ky)>16 then this function will not be compatible.
vincent at verbrugh dot nl
02-Mar-2006 05:55
If you plan to use Mcrypt Encryption to store encrypted data (e.g. passwords) in a (MySQL) database make sure to set the column to BLOB rather than VARCHAR. Otherwise the data may change which can give unexpected results if you decrypt the value.
Jerry Hathaway
17-Feb-2006 02:23
After benchmarking AES in 256-bit operation, I've concluded that CBC is far faster than OFB. Using a 14.9 MiB file, on average...
Encrypt in CBC: 1.9 seconds
Encrypt in OFB: 45.7 seconds (same as CFB)
Just reading the file: ~.53 seconds
After some research, I've concluded that OFB and CFB are slightly more secure than CBC, however I believe the performance difference to be due to an implementation issue.
As a side note on ECB: As stated in the wiki linked to below, ECB is wholly inadequate. It not use an IV (whether it was supplied ot MCrypt or not), meaning the same key and plaintext always produce the same ciphertext and it doesn't hide patterns. The site shows an excellent example of this.
Very useful info:
http://en.wikipedia.org/wiki/Block_cipher_mode_of_operation
triptripon at gmail dot com
05-Feb-2006 07:28
Regarding storing the result on a postgres DB that uses Unicode (follow up to a post below).
You don't need to change the DB's encoding to ASCII.
Simply use BASE64 to encode the result, it's perfectly safe.
use base64_decode before you decrypt.
-- Tomer Levinboim
coz AT metamule D0T com
04-Nov-2005 09:15
If you're going to encrypt data with something like this and store it in postgres.
function encrypt($string, $key){
$result = '';
for($i=1; $i<=strlen($string); $i++){
$char = substr($string, $i-1, 1);
$keychar = substr($key, ($i % strlen($key))-1, 1);
$char = chr(ord($char)+ord($keychar));
$result.=$char;
}
return $result;
}
Make sure when you create your database to set the encoding to SQL_ASCII because you won't be able to store this data in a database that uses UNICODE
joseph-at-digiweb-dot-net-dot-nz
31-Oct-2005 09:22
A further note for those doing interop between PHP and Java:
If you're using the BouncyCastle library on the Java side, then you can use the ZeroBytePadding mode now available in it. Mcrypt pads the data with Nulls rather than spaces....
I've sucessfully done interop using Blowfish/CBC/ZeroBytePadding between PHP and Java this way.
amiller of connext point net
27-Oct-2005 01:55
As regards the whole one-time-pad discussion below, I would also add that any one-time pad scheme which involves the key passing through a computer in any fashion is blown from the very beginning. At best, you have to rely on a less-secure cipher for key transmission; at (far-from-uncommon) worst, your key has been written in the clear to some sectors of the disk (for example, by being swapped out of memory), and awaits only someone with a sector-scanning tool to reveal it.
If you're thinking about Internet security, considering a one-time pad at all, unless you're planning to use one to protect another cipher system's keys for transmission (and probably even then!), is a waste of your time.
torcuato
16-Sep-2005 05:28
There seems to be an error on the list of ciphers supported by the mcrypt extension.
MCRYPT_3DES should be MCRYPT_TRIPLEDES
paul dot lewis at absolutegenius dot co dot uk
13-Jul-2005 11:33
I've spent the majority of the day attempting to get mcrypt to work under IIS6 with Windows Server 2003 (Web Edition) and PHP 5.0.4
There seems to be some incompatability with enabling certain extensions (mcrypt being one) when you are running PHP as ISAPI in this environment.
The way to solve the problem (the error will be that it cannot load php_mcrypt.dll - access is denied) is to run in CGI. While this isn't supposed to be as good performance wise, if you need the mcrypt support (or Oracle support, too, I believe) then this is the only way I've found to do it.
rg at rg dot mejoramos dot com
24-Jun-2005 06:59
Master [thilo-at-hardtware.de], (First Post)
Much thanks by you script.
Now i happy. :-)
Maybe - lines:
<?php
function you($string,$key,$a){
if(empty($a))$string=base64_decode($string);
$salida='';
for($i=0;$i<strlen($string);$i++){
$char=substr($string,$i,1);
$keychar=substr($key,($i%strlen($key))-1,1);
if($a)$char=chr(ord($char)+ord($keychar));
else $char=chr(ord($char)-ord($keychar));
$salida.=$char;
}if($a)$salida=base64_encode($salida);
return $salida;}
$a='thilo-at-hardtware.de';
echo you($a,'xs:a/55p;',1);
$b='r+DcptBclqmdo9nlntWmlqfVadzY';
echo '<hr>';
echo you($b,'xs:a/55p;',0);
exit;
?>Thanks newly.
Regards from CO
MGTech
21-May-2005 03:02
Please, mind that a XOR Encryption can't compete with a block cipher like AES or IDEA, but if you really want to use a stream cipher the more secure RC4 is the right alternative.
A XOR Encryption is only a bit useful if you encrypt data for private use, otherwise it is a frankly a security risk.
thilo-at-hardtware.de
29-Apr-2005 03:37
I have modified the xor-encryption of Anonymous. It is now returning and eating base64-encoded strings. That's much better for saving and transporting (e.g. saving in a database) the encrypted string.
<?php
function encrypt($string, $key) {
$result = '';
for($i=0; $i<strlen($string); $i++) {
$char = substr($string, $i, 1);
$keychar = substr($key, ($i % strlen($key))-1, 1);
$char = chr(ord($char)+ord($keychar));
$result.=$char;
}
return base64_encode($result);
}
function decrypt($string, $key) {
$result = '';
$string = base64_decode($string);
for($i=0; $i<strlen($string); $i++) {
$char = substr($string, $i, 1);
$keychar = substr($key, ($i % strlen($key))-1, 1);
$char = chr(ord($char)-ord($keychar));
$result.=$char;
}
return $result;
}
?>
IRLCoder
24-Mar-2005 06:13
Re: Just a followup note about mcrypt<-->java interop (have a php app encrypting a string and a java app decrypting it - and vice-versa!)
Sorry, the PHP had a bug and should have been the following:
$dlen = strlen($data);
$pad = 16 - fmod($dlen, 16); //change here
if ($pad > 0) {
$i = (int)$pad;
while ($i > 0) {
$data.=" ";
$i--;
}
}
IRLCoder
24-Mar-2005 05:12
Just a followup note about mcrypt<-->java interop (have a php app encrypting a string and a java app decrypting it - and vice-versa!)
it seems that mcrypt pads with nulls (0x00) instead of spaces so to allow the interop to work properly (where both java and php/mcrypt encode strings to the same value) - I padded the PHP strings with spaces first before encrypting and did the same on the Java side - thus the padding was the same on each side and the interop is complete. Here's sample PHP for padding:
$dlen = strlen($data);
$pad = fmod($dlen, 16);
if ($pad > 0) {
$i = (int)$pad;
while ($i > 0) {
$data.=" ";
$i--;
}
}
and here's the Java padding:
public static String PadString(String in) {
int slen = (in.length() % 16);
int i = (16 - slen);
if ((i > 0) && (i < 16)){
StringBuffer buf = new StringBuffer(in.length() + i);
buf.insert(0, in);
for (i = (16 - slen); i > 0; i--) {
buf.append(" ");
}
return buf.toString();
}
else {
return in;
}
}
With trimming on each side for decryption ... all works well.
davisd50 at yahoo dot com
19-Mar-2005 12:37
I was creating a cross-platform (unix + windows) application and ran into some issues with different versions of mcrypt (2.4.x vs 2.5.x) not working with Windows. After further investigation, I found the following:
Libmcrypt 2.5.7 is not usable with PHP 4.x.
The reason is a very Windows-specific issue. DLLs are Windows' way of providing shared code - things get loaded once into RAM and are guaranteed identical across all applications that use them. When an application references a routine/function in a DLL, it may do so either by function name or by reference number. Referring to a routine by name is considered slow by some people and does slow down the startup of applications that use large numbers of routines in DLLs (applications such as PHP). Referring to a routine by reference number is very quick but breaks if a new DLL version changes the reference number assignments for its routines.
Libmcrypt 2.5.6 & 7's DLLs do not have the same reference number assignments as old Libmcrypt.dll files did. Specifically, function 148 (mdecrypt_generic) has been moved in 2.5.6/7 to number 149 due to the addition of the mcrypt_mutex_register function which moved everything above it alphebetically up a reference number. PHP's php_mcrypt.dll, which provides the PHP language bindings to the low-level Libmcrypt.dll, refers to Libmcrypt.dll's functions by number instead of name. PHP 4.x's php_mcrypt.dll refers to function 148 while PHP 5.0's php_mcrypt.dll refers to function 149 and notes that PHP 5.0 under Windows can only work with Libmcrypt 2.5.6 and higher (because function 149 in old Libmcrypt is actually the Panama encryption algorithm, not the expected mcrypt_generic routine).
So, unless you can find a PHP 4.x php_mcrypt.dll that has been compiled for Libmcrypt 2.5.6 and higher, PHP 4.x won't be able to decrypt data with Libmcrypt 2.5.7 under Windows.
GMScribe
05-Mar-2005 07:58
TO:
16-Feb-2005 01:53
No key is ever truely random, thus it is breakable.
php at stock-consulting dot com
16-Feb-2005 09:53
amiller of connext point net is correct. The 'one-time pad' is unbreakable, if used correctly. Correct use requires that:
1. the key length equals or exceeds the data length
2. the key must not be used more than once to encrypt data
3. the key must be truly random.
The last point is a bit of a problem. "Random Number Generator" algorithms are only pseudo-random. Once the attacker finds out which PRNG was used, breaking the encryption becomes easy. Hardware solutions for the generation of random data (like, sampling noise via the sound card input and taking the LSB) have also proven to be of dubious quality.
Violation of point 2 makes the encryption worthless. A known-plaintext attack will reveal the key at once.
amiller of connext point net
21-Dec-2004 02:03
"Anonymous'" comment of Nov 19th refers to an encryption scheme known as a 'one-time pad'. Anonymous forgot to mention that, while it's true the one-time pad is secure as long as the key's kept secret, it's also necessary to avoid reusing keys -- multiple plaintexts encrypted using a single key are insecure, too, because of collisions, and that's why it's called a 'one-time pad'. Hope this helps --
Jurgen Schwietering
23-Nov-2004 05:20
Attention when using keys longer than the actual key size (i.e. 160 bit instead of 128 bit).
It will work inbetween PHP scripts, but might cause problems when using openssl or other packages with this integration of mcrypt. Cut them always to the supported size (mcrypt_enc_get_key_size) to avoid sleepless hours.
Anonymous
20-Nov-2004 12:18
Or, if you don't have the mcrypt library, you can just use these functions:
<?php
function Encrypt($string, $key)
{
$result = '';
for($i=1; $i<=strlen($string); $i++)
{
$char = substr($string, $i-1, 1);
$keychar = substr($key, ($i % strlen($key))-1, 1);
$char = chr(ord($char)+ord($keychar));
$result.=$char;
}
return $result;
}
function Decrypt($string, $key)
{
$result = '';
for($i=1; $i<=strlen($string); $i++)
{
$char = substr($string, $i-1, 1);
$keychar = substr($key, ($i % strlen($key))-1, 1);
$char = chr(ord($char)-ord($keychar));
$result.=$char;
}
return $result;
}
?>
It's very simple encryption, but as long as the key stays secret, very powerful.
emaher at newbay dot com
06-Nov-2004 12:37
Just a note about mcrypt<-->java interop (have a php app encrypting a string and a java app decrypting it)
php mcrypt pads the plaintext with spaces until the plaintext is a multiple of the block size (e.g 16 for most symmetric ciphers). These spaces have to be stripped on when decrypting on the java side. Seems to be no other way of using a sensible padding (e.g. with PKCS #5) on the mcrypt side.
The following php and java will interop
$cipher = "rijndael-128";
$mode = "cbc";
echo "CIPHER: $cipher | MODE: $mode\n";
// data,key, iv
$data = "blah";
$key = "01234567890abcdef";
$iv = "fedcba9876543210";
// set up and encyrpt
$td = mcrypt_module_open($cipher, "", $mode, $iv);
mcrypt_generic_init($td, $key, $iv);
$encrypted_data = mcrypt_generic($td, $data);
echo bin2hex($encrypted_data);
// tear down
mcrypt_generic_deinit($td);
mcrypt_module_close($td);
Java side
Cipher cipher = Cipher.getInstance("AES/CBC/NoPadding");
SecretKeySpec keySpec = new SecretKeySpec("01234567890abcdef".getBytes(), "AES");
IvParameterSpec ivSpec = new IvParameterSpec("fedcba9876543210".getBytes());
cipher.init(Cipher.DECRYPT_MODE, keySpec, ivSpec);
byte[] outText = cipher.doFinal(fromHexString(input));
System.out.println(new String(outText).trim());
Jan Drda <yan at yan dot cz>
16-Oct-2004 09:24
Sometimes mcrypt library is not aviable e.g. commercial hosting, customer restrictions, etc. Because of these conditions I wrote alternative free synchronous encryption library in pure PHP. It is not so complex as mcrypt, but it is sufficient. Everything including detailed documentation and API description is avaiable at http://www.yan.cz/brutuslib/. Hoping it helps somebody.
Moises Deniz Aleman
24-Aug-2004 07:48
This is a modified version of a previous script that test the algorithms and modes of mcrypt dll. This new script prints out the result in a table and avoid warnings to be printed.
<?PHP
/* run a self-test through every listed cipher and mode */
function mcrypt_check_sanity() {
$modes = mcrypt_list_modes();
$algorithms = mcrypt_list_algorithms();
echo "<table border=1>";
echo "<tr><td align=center><strong>Algorithm</strong></td align=center><td><strong>Status</strong></td>";
foreach ($modes as $mode) echo "<td align=center><strong>".strtoupper($mode)."</strong></td>";
echo "</tr>";
foreach ($algorithms as $cipher) {
echo "<tr><td bgcolor=f0f0ff align=left>".strtoupper($cipher)."</td>";
if(mcrypt_module_self_test($cipher)) {
print "<td bgcolor=green align=center>OK</td>";
} else {
print "<td bgcolor=red align=center>NOT OK</td>";
}
foreach ($modes as $mode) {
if($mode == 'stream') {
$result = "<td bgcolor=gray align=center>NOT TESTED</td>";
} else if(mcrypt_test_module_mode($cipher,$mode)) {
$result = "<td bgcolor=green align=center><strong>OK</strong></td>";
} else {
$result = "<td bgcolor=red align=center>NOT OK</td>";
}
print $result;
}
echo "</tr>";
}
echo "</table>";
}
// a variant on the example posted in mdecrypt_generic
function mcrypt_test_module_mode($module,$mode) {
/* Data */
$key = 'this is a very long key, even too long for the cipher';
$plain_text = 'very important data';
/* Open module, and create IV */
$td = mcrypt_module_open($module, '',$mode, '');
$key = substr($key, 0, mcrypt_enc_get_key_size($td));
$iv_size = mcrypt_enc_get_iv_size($td);
$iv = mcrypt_create_iv($iv_size, MCRYPT_RAND);
/* Initialize encryption handle */
if (mcrypt_generic_init($td, $key, $iv) != -1) {
/* Encrypt data */
$c_t = mcrypt_generic($td, $plain_text);
mcrypt_generic_deinit($td);
// close the module
mcrypt_module_close($td);
/* Reinitialize buffers for decryption */
/* Open module */
$td = mcrypt_module_open($module, '', $mode, '');
$key = substr($key, 0, mcrypt_enc_get_key_size($td));
mcrypt_generic_init($td, $key, $iv);
$p_t = trim(mdecrypt_generic($td, $c_t)); //trim to remove padding
/* Clean up */
mcrypt_generic_end($td);
mcrypt_module_close($td);
}
if (strncmp($p_t, $plain_text, strlen($plain_text)) == 0) {
return TRUE;
} else {
return FALSE;
}
}
// function call:
@mcrypt_check_sanity();
?>
wagner at cesnet dot cz
17-Aug-2004 12:38
I have another comment to the script submitted by robert at peakepro dot com. He uses deprecated function mcrypt_generic_end which also closes the module. Subsequent mcrypt_module_close then complains about invalid resource. Moreover, not all combinations of algorith and mode are compatible. It is necessary to check the return value from mcrypt_module_open in order to find whether the module can be open for selected combination. My script also displays some useful information. You can find the full source at my site http://icebearsoft.euweb.cz/php/
Wilmo
11-Aug-2004 10:54
Changing the function as such seems to greatly help.
<?php
function encrypt($encrypt) {
global $key;
srand((double) microtime() * 1000000); //for sake of MCRYPT_RAND
$iv = mcrypt_create_iv(mcrypt_get_iv_size(MCRYPT_RIJNDAEL_256, MCRYPT_MODE_ECB), MCRYPT_RAND);
$passcrypt = mcrypt_encrypt(MCRYPT_RIJNDAEL_256, $key, $encrypt, MCRYPT_MODE_ECB, $iv);
$encode = base64_encode($passcrypt);
return $encode;
}
?>
by adding:
<?php
srand((double) microtime() * 1000000); //for sake of MCRYPT_RAND
?>
which I guess seeds the encryption routine differently everytime.
Thanks to robert at peakepro dot com sourced from
http://www.php.net/manual/en/function.mcrypt-create-iv.php
Cheers,
Wil
Wilmo
11-Aug-2004 10:16
The Algorithm posted by:
Mike Zaccari
29-Jun-2004 03:54
"I'm running PHP 4.3.7 on Apache 2.0.49 on an Xp machine, and after many hours of googling over the internet I found that the easiest way to use the mcrypt function was to do this:"
Thanks for posting this Mike but there seems to be a problem
with your implementation.
On my machine and I suspect others, the output is independent
of the key. I can change the key and this has no effect on
the resulting crypted data. So the input is always encrypted the
same way irregardless of the key and therefore decrypted
with any key.
This would only be secure if an attacker knew nothing about the
algorithm which seems unlikely from an experienced attacker.
I am looking into a fix and will post if resolved.
Does anybody else have this problem? I want to make sure my install is good.
I am running on a gentoo linux box.
Thanks,
Wil
Here is the code I am using to test the algorithm:
<head><title>Encryption</title>
</head>
<body>
<?php print_debug_header(1); ?>
<form name=form method=post action='encrypt.php'>
<table align=center>
<TR><TD>Source Text:</TD><TD><input type=text name=input value=<?php echo $_REQUEST['input']; ?>></TD></TR>
<TR><TD>Key:</TD><TD><input type=text name=key value=<?php echo $_REQUEST['key']; ?>></TD></TR>
</table>
<input type=submit>
</form>
<?php
if(!empty($_REQUEST['input'])){
$encrypted=encrypt($_REQUEST['input']);
$decrypted=decrypt($encrypted);
echo "Encrypted : '$encrypted' Decrypted: '$decrypted' <BR>";
}
$key = $_REQUEST['key'];
//Encrypt Function
function encrypt($encrypt) {
global $key;
$iv = mcrypt_create_iv(mcrypt_get_iv_size(MCRYPT_RIJNDAEL_256, MCRYPT_MODE_ECB), MCRYPT_RAND);
$passcrypt = mcrypt_encrypt(MCRYPT_RIJNDAEL_256, $key, $encrypt, MCRYPT_MODE_ECB, $iv);
$encode = base64_encode($passcrypt);
return $encode;
}
//Decrypt Function
function decrypt($decrypt) {
global $key;
$decoded = base64_decode($decrypt);
$iv = mcrypt_create_iv(mcrypt_get_iv_size(MCRYPT_RIJNDAEL_256, MCRYPT_MODE_ECB), MCRYPT_RAND);
$decrypted = mcrypt_decrypt(MCRYPT_RIJNDAEL_256, $key, $decoded, MCRYPT_MODE_ECB, $iv);
return $decrypted;
}
?>
</body>
</html>
dzelko at hotmail dot com
15-May-2004 03:11
Windows IIS Users - Problems installing/using mcrypt and other extensions
::
I've noticed a lot of users complaining in forums that they have difficult time getting mcrypt extension to function/ finding or installing working .dlls when using IIS:
::
Easy solution that works well for me: (IIS 6 on Win 2003 Svr and IIS on XP Pro) #customized install later for increased security#
1. Install current stable php version using windows installer #gets php up and running quickly#
2. Download Windows Binary Package
3. Extract Package Library to PHP folder installer generates - overwrite all
4. Edit php.ini as appropriate. (specifically for mcrypt uncomment mcrypt=php_mcrypt.dll)
::
No mo' problems.
robert at peakepro dot com
03-Mar-2004 06:59
15-Feb-2004 06:17
"I found this nested loop very useful for checking the sanity of my libmcrypt install. It turned out many of the modules weren't working in certain modes. Hopefully this will save someone some frustration:"
I have ammended the mcrypt_check_sanity() function. This is a bug fix: the function was reinitializing the IV, which is unacceptable. Here is the complete, fixed version. Note that 'stream' mode is not tested.
<?PHP
/* run a self-test through every listed cipher and mode */
function mcrypt_check_sanity() {
$modes = mcrypt_list_modes();
$algorithms = mcrypt_list_algorithms();
foreach ($algorithms as $cipher) {
if(mcrypt_module_self_test($cipher)) {
print $cipher." ok.<br />\n";
} else {
print $cipher." not ok.<br />\n";
}
foreach ($modes as $mode) {
if($mode == 'stream') {
$result = "not tested";
} else if(mcrypt_test_module_mode($cipher,$mode)) {
$result = "ok";
} else {
$result = "not ok";
}
print $cipher." in mode ".$mode." ".$result."<br />\n";
mcrypt_module_close($td);
}
}
}
// a variant on the example posted in mdecrypt_generic
function mcrypt_test_module_mode($module,$mode) {
/* Data */
$key = 'this is a very long key, even too long for the cipher';
$plain_text = 'very important data';
/* Open module, and create IV */
$td = mcrypt_module_open($module, '',$mode, '');
$key = substr($key, 0, mcrypt_enc_get_key_size($td));
$iv_size = mcrypt_enc_get_iv_size($td);
$iv = mcrypt_create_iv($iv_size, MCRYPT_RAND);
/* Initialize encryption handle */
if (mcrypt_generic_init($td, $key, $iv) != -1) {
/* Encrypt data */
$c_t = mcrypt_generic($td, $plain_text);
mcrypt_generic_deinit($td);
// close the module
mcrypt_module_close($td);
/* Reinitialize buffers for decryption */
/* Open module */
$td = mcrypt_module_open($module, '', $mode, '');
$key = substr($key, 0, mcrypt_enc_get_key_size($td));
mcrypt_generic_init($td, $key, $iv);
$p_t = trim(mdecrypt_generic($td, $c_t)); //trim to remove padding
/* Clean up */
mcrypt_generic_end($td);
mcrypt_module_close($td);
}
if (strncmp($p_t, $plain_text, strlen($plain_text)) == 0) {
return TRUE;
} else {
return FALSE;
}
}
// function call:
mcrypt_check_sanity();
?>
Apologies for any confusion the previous version may have caused. The only working mode listed was ecb, since it does not use an IV.
groundzero at zuavra dot net
09-Jan-2004 09:08
If you've ever compiled PHP from source (any version) you may be familiar with the [in]famous MCRYPT_BLOWFISH_128 compilation error that appears when you attempt to compile --with-mcrypt. It occurs often on Debian but not only there. The problem: during compilation, the PHP configure script always assumes that libmcrypt has been built in conjunction with libltdl. Whenever that is not the case, PHP compilation will fail later saying certain headers (such as the above blowfish example) are missing from mcrypt.h (which is misleading, they're not supposed to be there nor looked after if libltdl was properly involved). Solution: make sure your libmcrypt was linked against libltdl before you even start configuring PHP. You can check by running 'ldd lybmcrypt.so' and verifying that libltdl appears in the output. libltdl can be found in libltld[3][-dev] on Debian or in libtool-libs on Red Hat.
simms
04-Oct-2003 05:03
DEBIAN users: avoid mcrypt installation headaches.
to add mcrypt support to an existing php installation, get root and run
apt-get install php4-mcrypt
restart your webserver, and voil
steve@itemfront dot ltd dot uk
29-Sep-2003 08:17
Just spent a while getting mcrypt support working with php. Used libmcrypt version 2.5.7 with php 4.3.3. Out of the box it just won't work. Configure as follows:
libmcrypt:
./configure --disable-posix-threads --enable-dynamic-loading
php: ( as you can see, it's built for a SunONE server, but that's the easy bit to configure! )
./configure --with-nsapi=/usr/iplanet/servers --enable-sigchld --with-ldap --with-zlib-dir=/usr/lib --with-mcrypt=<srcdir>/libmcrypt-2.5.7
hth, steve
herowekker at hotmail dot com
31-Jul-2003 08:55
mcrypt_ecb with base64_decode gave some problems, i found out that it had to be chopped to work:
chop(mcrypt_ecb(MCRYPT_BLOWFISH,$key,base64_decode
($input),MCRYPT_DECRYPT));
paul at xmill dot com
06-Nov-2002 02:25
The statement that use of the 'generic' functions is 'recommended' in 2.4+ should probably be 'required'. Having like many others struggled with a "mcrypt module initialization failed" error even when phpinfo() showed mcrypt installed and all its algorithms loaded, I found that:
$key = mcrypt_ecb (MCRYPT_3DES, $magic, $input, MCRYPT_ENCRYPT);
fails, but
$td = mcrypt_module_open (MCRYPT_TripleDES, "", MCRYPT_MODE_ECB, "");
$iv = mcrypt_create_iv (mcrypt_enc_get_iv_size ($td), MCRYPT_RAND);
mcrypt_generic_init ($td, $magic, $iv);
$key = mcrypt_generic ($td, $input);
mcrypt_generic_end ($td);
works fine. You can't leave out the creation of the IV to the right length, even though ECB doesn't use it (happily, because I want it repeatable).
scott at boothcreek dot com
28-Mar-2002 05:28
If you are using ECB mode to encrypt it does not seem to use the iv (initialization vector) for much of anything, given the same key it will always decrypt it no matter what the iv is. If you use CBC mode you must decrypt with the same iv that you encrypted with.
If you use a different iv before decrypting, your decrypt will not work. IMHO it seems better to use CBC mode than ECB as ECB will always encrypt to the same cipher text given the same plain text (leaving you open to know plaintext attacks). CBC uses the random iv which means text encrypts to different things. You probably could get the same effect from using random keys in ECB mode.
Read that in the Schneier book - Applied Cryptography (ISBN 0-471-11709-9) This book is a must for anyone seriously using any type of encryption.
pawelNOSPAM at rsc dot pl
03-Aug-2001 09:37
If you compiled mcrypt and php without problem, but phpinfo() shows there are no supported ciphers and modes, try to change mode to 755 on libdirs (/usr/local/libmcrypt, /usr/local/libcrypt).
mountie at paygate dot net
10-Feb-2000 09:21
the encrypted result data maybe binary data and It make errors in sql query.
so use the base64_encode/base64_decode function with mcrypt()
try below
base64_encode(mcrypt_ecb(MCRYPT_BLOWFISH,$key,$input,MCRYPT_ENCRYPT));
mcrypt_ecb(MCRYPT_BLOWFISH,$key,base64_decode($input),MCRYPT_DECRYPT);
|  |