includes/CryptRand.php

   1 <?php
   2 /**
   3  * A cryptographic random generator class used for generating secret keys
   4  *
   5  * This is based in part on Drupal code as well as what we used in our own code
   6  * prior to introduction of this class.
   7  *
   8  * @author Daniel Friesen
   9  * @file
  10  */
  11
  12 class MWCryptRand {
  13
  14         /**
  15          * Minimum number of iterations we want to make in our drift calculations.
  16          */
  17         const MIN_ITERATIONS = 1000;
  18
  19         /**
  20          * Number of milliseconds we want to spend generating each separate byte
  21          * of the final generated bytes.
  22          * This is used in combination with the hash length to determine the duration
  23          * we should spend doing drift calculations.
  24          */
  25         const MSEC_PER_BYTE = 0.5;
  26
  27         /**
  28          * Singleton instance for public use
  29          */
  30         protected static $singleton = null;
  31
  32         /**
  33          * The hash algorithm being used
  34          */
  35         protected $algo = null;
  36
  37         /**
  38          * The number of bytes outputted by the hash algorithm
  39          */
  40         protected $hashLength = null;
  41
  42         /**
  43          * A boolean indicating whether the previous random generation was done using
  44          * cryptographically strong random number generator or not.
  45          */
  46         protected $strong = null;
  47
  48         /**
  49          * Initialize an initial random state based off of whatever we can find
  50          */
  51         protected function initialRandomState() {
  52                 // $_SERVER contains a variety of unstable user and system specific information
  53                 // It'll vary a little with each page, and vary even more with separate users
  54                 // It'll also vary slightly across different machines
  55                 $state = serialize( $_SERVER );
  56
  57                 // To try vary the system information of the state a bit more
  58                 // by including the system's hostname into the state
  59                 $state .= wfHostname();
  60
  61                 // Try to gather a little entropy from the different php rand sources
  62                 $state .= rand() . uniqid( mt_rand(), true );
  63
  64                 // Include some information about the filesystem's current state in the random state
  65                 $files = array();
  66
  67                 // We know this file is here so grab some info about ourself
  68                 $files[] = __FILE__;
  69
  70                 // We must also have a parent folder, and with the usual file structure, a grandparent
  71                 $files[] = dirname( __FILE__ );
  72                 $files[] = dirname( dirname( __FILE__ ) );
  73
  74                 // The config file is likely the most often edited file we know should be around
  75                 // so include its stat info into the state.
  76                 // The constant with its location will almost always be defined, as WebStart.php defines
  77                 // MW_CONFIG_FILE to $IP/LocalSettings.php unless being configured with MW_CONFIG_CALLBACK (eg. the installer)
  78                 if ( defined( 'MW_CONFIG_FILE' ) ) {
  79                         $files[] = MW_CONFIG_FILE;
  80                 }
  81
  82                 foreach ( $files as $file ) {
  83                         wfSuppressWarnings();
  84                         $stat = stat( $file );
  85                         wfRestoreWarnings();
  86                         if ( $stat ) {
  87                                 // stat() duplicates data into numeric and string keys so kill off all the numeric ones
  88                                 foreach ( $stat as $k => $v ) {
  89                                         if ( is_numeric( $k ) ) {
  90                                                 unset( $k );
  91                                         }
  92                                 }
  93                                 // The absolute filename itself will differ from install to install so don't leave it out
  94                                 $state .= realpath( $file );
  95                                 $state .= implode( '', $stat );
  96                         } else {
  97                                 // The fact that the file isn't there is worth at least a
  98                                 // minuscule amount of entropy.
  99                                 $state .= '0';
 100                         }
 101                 }
 102
 103                 // Try and make this a little more unstable by including the varying process
 104                 // id of the php process we are running inside of if we are able to access it
 105                 if ( function_exists( 'getmypid' ) ) {
 106                         $state .= getmypid();
 107                 }
 108
 109                 // If available try to increase the instability of the data by throwing in
 110                 // the precise amount of memory that we happen to be using at the moment.
 111                 if ( function_exists( 'memory_get_usage' ) ) {
 112                         $state .= memory_get_usage( true );
 113                 }
 114
 115                 // It's mostly worthless but throw the wiki's id into the data for a little more variance
 116                 $state .= wfWikiID();
 117
 118                 // If we have a secret key or proxy key set then throw it into the state as well
 119                 global $wgSecretKey, $wgProxyKey;
 120                 if ( $wgSecretKey ) {
 121                         $state .= $wgSecretKey;
 122                 } elseif ( $wgProxyKey ) {
 123                         $state .= $wgProxyKey;
 124                 }
 125
 126                 return $state;
 127         }
 128
 129         /**
 130          * Randomly hash data while mixing in clock drift data for randomness
 131          *
 132          * @param $data string The data to randomly hash.
 133          * @return String The hashed bytes
 134          * @author Tim Starling
 135          */
 136         protected function driftHash( $data ) {
 137                 // Minimum number of iterations (to avoid slow operations causing the loop to gather little entropy)
 138                 $minIterations = self::MIN_ITERATIONS;
 139                 // Duration of time to spend doing calculations (in seconds)
 140                 $duration = ( self::MSEC_PER_BYTE / 1000 ) * $this->hashLength();
 141                 // Create a buffer to use to trigger memory operations
 142                 $bufLength = 10000000;
 143                 $buffer = str_repeat( ' ', $bufLength );
 144                 $bufPos = 0;
 145
 146                 // Iterate for $duration seconds or at least $minIerations number of iterations
 147                 $iterations = 0;
 148                 $startTime = microtime( true );
 149                 $currentTime = $startTime;
 150                 while ( $iterations < $minIterations || $currentTime - $startTime < $duration ) {
 151                         // Trigger some memory writing to trigger some bus activity
 152                         // This may create variance in the time between iterations
 153                         $bufPos = ( $bufPos + 13 ) % $bufLength;
 154                         $buffer[$bufPos] = ' ';
 155                         // Add the drift between this iteration and the last in as entropy
 156                         $nextTime = microtime( true );
 157                         $delta = (int)( ( $nextTime - $currentTime ) * 1000000 );
 158                         $data .= $delta;
 159                         // Every 100 iterations hash the data and entropy
 160                         if ( $iterations % 100 === 0 ) {
 161                                 $data = sha1( $data );
 162                         }
 163                         $currentTime = $nextTime;
 164                         $iterations++;
 165                 }
 166                 $timeTaken = $currentTime - $startTime;
 167                 $data = $this->hash( $data );
 168
 169                 wfDebug( __METHOD__ . ": Clock drift calculation " .
 170                         "(time-taken=" . ( $timeTaken * 1000 ) . "ms, " .
 171                         "iterations=$iterations, " .
 172                         "time-per-iteration=" . ( $timeTaken / $iterations * 1e6 ) . "us)\n" );
 173                 return $data;
 174         }
 175
 176         /**
 177          * Return a rolling random state initially build using data from unstable sources
 178          * @return string A new weak random state
 179          */
 180         protected function randomState() {
 181                 static $state = null;
 182                 if ( is_null( $state ) ) {
 183                         // Initialize the state with whatever unstable data we can find
 184                         // It's important that this data is hashed right afterwards to prevent
 185                         // it from being leaked into the output stream
 186                         $state = $this->hash( $this->initialRandomState() );
 187                 }
 188                 // Generate a new random state based on the initial random state or previous
 189                 // random state by combining it with clock drift
 190                 $state = $this->driftHash( $state );
 191                 return $state;
 192         }
 193
 194         /**
 195          * Decide on the best acceptable hash algorithm we have available for hash()
 196          * @throws MWException
 197          * @return String A hash algorithm
 198          */
 199         protected function hashAlgo() {
 200                 if ( !is_null( $this->algo ) ) {
 201                         return $this->algo;
 202                 }
 203
 204                 $algos = hash_algos();
 205                 $preference = array( 'whirlpool', 'sha256', 'sha1', 'md5' );
 206
 207                 foreach ( $preference as $algorithm ) {
 208                         if ( in_array( $algorithm, $algos ) ) {
 209                                 $this->algo = $algorithm;
 210                                 wfDebug( __METHOD__ . ": Using the {$this->algo} hash algorithm.\n" );
 211                                 return $this->algo;
 212                         }
 213                 }
 214
 215                 // We only reach here if no acceptable hash is found in the list, this should
 216                 // be a technical impossibility since most of php's hash list is fixed and
 217                 // some of the ones we list are available as their own native functions
 218                 // But since we already require at least 5.2 and hash() was default in
 219                 // 5.1.2 we don't bother falling back to methods like sha1 and md5.
 220                 throw new MWException( "Could not find an acceptable hashing function in hash_algos()" );
 221         }
 222
 223         /**
 224          * Return the byte-length output of the hash algorithm we are
 225          * using in self::hash and self::hmac.
 226          *
 227          * @return int Number of bytes the hash outputs
 228          */
 229         protected function hashLength() {
 230                 if ( is_null( $this->hashLength ) ) {
 231                         $this->hashLength = strlen( $this->hash( '' ) );
 232                 }
 233                 return $this->hashLength;
 234         }
 235
 236         /**
 237          * Generate an acceptably unstable one-way-hash of some text
 238          * making use of the best hash algorithm that we have available.
 239          *
 240          * @param $data string
 241          * @return String A raw hash of the data
 242          */
 243         protected function hash( $data ) {
 244                 return hash( $this->hashAlgo(), $data, true );
 245         }
 246
 247         /**
 248          * Generate an acceptably unstable one-way-hmac of some text
 249          * making use of the best hash algorithm that we have available.
 250          *
 251          * @param $data string
 252          * @param $key string
 253          * @return String A raw hash of the data
 254          */
 255         protected function hmac( $data, $key ) {
 256                 return hash_hmac( $this->hashAlgo(), $data, $key, true );
 257         }
 258
 259         /**
 260          * @see self::wasStrong()
 261          */
 262         public function realWasStrong() {
 263                 if ( is_null( $this->strong ) ) {
 264                         throw new MWException( __METHOD__ . ' called before generation of random data' );
 265                 }
 266                 return $this->strong;
 267         }
 268
 269         /**
 270          * @see self::generate()
 271          */
 272         public function realGenerate( $bytes, $forceStrong = false ) {
 273                 wfProfileIn( __METHOD__ );
 274
 275                 wfDebug( __METHOD__ . ": Generating cryptographic random bytes for " . wfGetAllCallers( 5 ) . "\n" );
 276
 277                 $bytes = floor( $bytes );
 278                 static $buffer = '';
 279                 if ( is_null( $this->strong ) ) {
 280                         // Set strength to false initially until we know what source data is coming from
 281                         $this->strong = true;
 282                 }
 283
 284                 if ( strlen( $buffer ) < $bytes ) {
 285                         // If available make use of mcrypt_create_iv URANDOM source to generate randomness
 286                         // On unix-like systems this reads from /dev/urandom but does it without any buffering
 287                         // and bypasses openbasedir restrictions, so it's preferable to reading directly
 288                         // On Windows starting in PHP 5.3.0 Windows' native CryptGenRandom is used to generate
 289                         // entropy so this is also preferable to just trying to read urandom because it may work
 290                         // on Windows systems as well.
 291                         if ( function_exists( 'mcrypt_create_iv' ) ) {
 292                                 wfProfileIn( __METHOD__ . '-mcrypt' );
 293                                 $rem = $bytes - strlen( $buffer );
 294                                 $iv = mcrypt_create_iv( $rem, MCRYPT_DEV_URANDOM );
 295                                 if ( $iv === false ) {
 296                                         wfDebug( __METHOD__ . ": mcrypt_create_iv returned false.\n" );
 297                                 } else {
 298                                         $buffer .= $iv;
 299                                         wfDebug( __METHOD__ . ": mcrypt_create_iv generated " . strlen( $iv ) . " bytes of randomness.\n" );
 300                                 }
 301                                 wfProfileOut( __METHOD__ . '-mcrypt' );
 302                         }
 303                 }
 304
 305                 if ( strlen( $buffer ) < $bytes ) {
 306                         // If available make use of openssl's random_pseudo_bytes method to attempt to generate randomness.
 307                         // However don't do this on Windows with PHP < 5.3.4 due to a bug:
 308                         // http://stackoverflow.com/questions/1940168/openssl-random-pseudo-bytes-is-slow-php
 309                         // http://git.php.net/?p=php-src.git;a=commitdiff;h=cd62a70863c261b07f6dadedad9464f7e213cad5
 310                         if ( function_exists( 'openssl_random_pseudo_bytes' )
 311                                 && ( !wfIsWindows() || version_compare( PHP_VERSION, '5.3.4', '>=' ) )
 312                         ) {
 313                                 wfProfileIn( __METHOD__ . '-openssl' );
 314                                 $rem = $bytes - strlen( $buffer );
 315                                 $openssl_bytes = openssl_random_pseudo_bytes( $rem, $openssl_strong );
 316                                 if ( $openssl_bytes === false ) {
 317                                         wfDebug( __METHOD__ . ": openssl_random_pseudo_bytes returned false.\n" );
 318                                 } else {
 319                                         $buffer .= $openssl_bytes;
 320                                         wfDebug( __METHOD__ . ": openssl_random_pseudo_bytes generated " . strlen( $openssl_bytes ) . " bytes of " . ( $openssl_strong ? "strong" : "weak" ) . " randomness.\n" );
 321                                 }
 322                                 if ( strlen( $buffer ) >= $bytes ) {
 323                                         // openssl tells us if the random source was strong, if some of our data was generated
 324                                         // using it use it's say on whether the randomness is strong
 325                                         $this->strong = !!$openssl_strong;
 326                                 }
 327                                 wfProfileOut( __METHOD__ . '-openssl' );
 328                         }
 329                 }
 330
 331                 // Only read from urandom if we can control the buffer size or were passed forceStrong
 332                 if ( strlen( $buffer ) < $bytes && ( function_exists( 'stream_set_read_buffer' ) || $forceStrong ) ) {
 333                         wfProfileIn( __METHOD__ . '-fopen-urandom' );
 334                         $rem = $bytes - strlen( $buffer );
 335                         if ( !function_exists( 'stream_set_read_buffer' ) && $forceStrong ) {
 336                                 wfDebug( __METHOD__ . ": Was forced to read from /dev/urandom without control over the buffer size.\n" );
 337                         }
 338                         // /dev/urandom is generally considered the best possible commonly
 339                         // available random source, and is available on most *nix systems.
 340                         wfSuppressWarnings();
 341                         $urandom = fopen( "/dev/urandom", "rb" );
 342                         wfRestoreWarnings();
 343
 344                         // Attempt to read all our random data from urandom
 345                         // php's fread always does buffered reads based on the stream's chunk_size
 346                         // so in reality it will usually read more than the amount of data we're
 347                         // asked for and not storing that risks depleting the system's random pool.
 348                         // If stream_set_read_buffer is available set the chunk_size to the amount
 349                         // of data we need. Otherwise read 8k, php's default chunk_size.
 350                         if ( $urandom ) {
 351                                 // php's default chunk_size is 8k
 352                                 $chunk_size = 1024 * 8;
 353                                 if ( function_exists( 'stream_set_read_buffer' ) ) {
 354                                         // If possible set the chunk_size to the amount of data we need
 355                                         stream_set_read_buffer( $urandom, $rem );
 356                                         $chunk_size = $rem;
 357                                 }
 358                                 $random_bytes = fread( $urandom, max( $chunk_size, $rem ) );
 359                                 $buffer .= $random_bytes;
 360                                 fclose( $urandom );
 361                                 wfDebug( __METHOD__ . ": /dev/urandom generated " . strlen( $random_bytes ) . " bytes of randomness.\n" );
 362                                 if ( strlen( $buffer ) >= $bytes ) {
 363                                         // urandom is always strong, set to true if all our data was generated using it
 364                                         $this->strong = true;
 365                                 }
 366                         } else {
 367                                 wfDebug( __METHOD__ . ": /dev/urandom could not be opened.\n" );
 368                         }
 369                         wfProfileOut( __METHOD__ . '-fopen-urandom' );
 370                 }
 371
 372                 // If we cannot use or generate enough data from a secure source
 373                 // use this loop to generate a good set of pseudo random data.
 374                 // This works by initializing a random state using a pile of unstable data
 375                 // and continually shoving it through a hash along with a variable salt.
 376                 // We hash the random state with more salt to avoid the state from leaking
 377                 // out and being used to predict the /randomness/ that follows.
 378                 if ( strlen( $buffer ) < $bytes ) {
 379                         wfDebug( __METHOD__ . ": Falling back to using a pseudo random state to generate randomness.\n" );
 380                 }
 381                 while ( strlen( $buffer ) < $bytes ) {
 382                         wfProfileIn( __METHOD__ . '-fallback' );
 383                         $buffer .= $this->hmac( $this->randomState(), mt_rand() );
 384                         // This code is never really cryptographically strong, if we use it
 385                         // at all, then set strong to false.
 386                         $this->strong = false;
 387                         wfProfileOut( __METHOD__ . '-fallback' );
 388                 }
 389
 390                 // Once the buffer has been filled up with enough random data to fulfill
 391                 // the request shift off enough data to handle the request and leave the
 392                 // unused portion left inside the buffer for the next request for random data
 393                 $generated = substr( $buffer, 0, $bytes );
 394                 $buffer = substr( $buffer, $bytes );
 395
 396                 wfDebug( __METHOD__ . ": " . strlen( $buffer ) . " bytes of randomness leftover in the buffer.\n" );
 397
 398                 wfProfileOut( __METHOD__ );
 399                 return $generated;
 400         }
 401
 402         /**
 403          * @see self::generateHex()
 404          */
 405         public function realGenerateHex( $chars, $forceStrong = false ) {
 406                 // hex strings are 2x the length of raw binary so we divide the length in half
 407                 // odd numbers will result in a .5 that leads the generate() being 1 character
 408                 // short, so we use ceil() to ensure that we always have enough bytes
 409                 $bytes = ceil( $chars / 2 );
 410                 // Generate the data and then convert it to a hex string
 411                 $hex = bin2hex( $this->generate( $bytes, $forceStrong ) );
 412                 // A bit of paranoia here, the caller asked for a specific length of string
 413                 // here, and it's possible (eg when given an odd number) that we may actually
 414                 // have at least 1 char more than they asked for. Just in case they made this
 415                 // call intending to insert it into a database that does truncation we don't
 416                 // want to give them too much and end up with their database and their live
 417                 // code having two different values because part of what we gave them is truncated
 418                 // hence, we strip out any run of characters longer than what we were asked for.
 419                 return substr( $hex, 0, $chars );
 420         }
 421
 422         /** Publicly exposed static methods **/
 423
 424         /**
 425          * Return a singleton instance of MWCryptRand
 426          * @return MWCryptRand
 427          */
 428         protected static function singleton() {
 429                 if ( is_null( self::$singleton ) ) {
 430                         self::$singleton = new self;
 431                 }
 432                 return self::$singleton;
 433         }
 434
 435         /**
 436          * Return a boolean indicating whether or not the source used for cryptographic
 437          * random bytes generation in the previously run generate* call
 438          * was cryptographically strong.
 439          *
 440          * @return bool Returns true if the source was strong, false if not.
 441          */
 442         public static function wasStrong() {
 443                 return self::singleton()->realWasStrong();
 444         }
 445
 446         /**
 447          * Generate a run of (ideally) cryptographically random data and return
 448          * it in raw binary form.
 449          * You can use MWCryptRand::wasStrong() if you wish to know if the source used
 450          * was cryptographically strong.
 451          *
 452          * @param $bytes int the number of bytes of random data to generate
 453          * @param $forceStrong bool Pass true if you want generate to prefer cryptographically
 454          *                          strong sources of entropy even if reading from them may steal
 455          *                          more entropy from the system than optimal.
 456          * @return String Raw binary random data
 457          */
 458         public static function generate( $bytes, $forceStrong = false ) {
 459                 return self::singleton()->realGenerate( $bytes, $forceStrong );
 460         }
 461
 462         /**
 463          * Generate a run of (ideally) cryptographically random data and return
 464          * it in hexadecimal string format.
 465          * You can use MWCryptRand::wasStrong() if you wish to know if the source used
 466          * was cryptographically strong.
 467          *
 468          * @param $chars int the number of hex chars of random data to generate
 469          * @param $forceStrong bool Pass true if you want generate to prefer cryptographically
 470          *                          strong sources of entropy even if reading from them may steal
 471          *                          more entropy from the system than optimal.
 472          * @return String Hexadecimal random data
 473          */
 474         public static function generateHex( $chars, $forceStrong = false ) {
 475                 return self::singleton()->realGenerateHex( $chars, $forceStrong );
 476         }
 477
 478 }