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