CryptRand.php

Go to the documentation of this file.

<?php
class MWCryptRand {

        const MIN_ITERATIONS = 1000;

        const MSEC_PER_BYTE = 0.5;

        protected static $singleton = null;

        protected $algo = null;

        protected $hashLength = null;

        protected $strong = null;

        protected function initialRandomState() {
                // $_SERVER contains a variety of unstable user and system specific information
                // It'll vary a little with each page, and vary even more with separate users
                // It'll also vary slightly across different machines
                $state = serialize( $_SERVER );

                // To try and vary the system information of the state a bit more
                // by including the system's hostname into the state
                $state .= wfHostname();

                // Try to gather a little entropy from the different php rand sources
                $state .= rand() . uniqid( mt_rand(), true );

                // Include some information about the filesystem's current state in the random state
                $files = array();
                // We know this file is here so grab some info about ourself
                $files[] = __FILE__;
                // The config file is likely the most often edited file we know should be around
                // so if the constant with it's location is defined include it's stat info into the state
                if ( defined( 'MW_CONFIG_FILE' ) ) {
                        $files[] = MW_CONFIG_FILE;
                }
                foreach ( $files as $file ) {
                        wfSuppressWarnings();
                        $stat = stat( $file );
                        wfRestoreWarnings();
                        if ( $stat ) {
                                // stat() duplicates data into numeric and string keys so kill off all the numeric ones
                                foreach ( $stat as $k => $v ) {
                                        if ( is_numeric( $k ) ) {
                                                unset( $k );
                                        }
                                }
                                // The absolute filename itself will differ from install to install so don't leave it out
                                $state .= realpath( $file );
                                $state .= implode( '', $stat );
                        } else {
                                // The fact that the file isn't there is worth at least a
                                // minuscule amount of entropy.
                                $state .= '0';
                        }
                }

                // Try and make this a little more unstable by including the varying process
                // id of the php process we are running inside of if we are able to access it
                if ( function_exists( 'getmypid' ) ) {
                        $state .= getmypid();
                }

                // If available try to increase the instability of the data by throwing in
                // the precise amount of memory that we happen to be using at the moment.
                if ( function_exists( 'memory_get_usage' ) ) {
                        $state .= memory_get_usage( true );
                }

                // It's mostly worthless but throw the wiki's id into the data for a little more variance
                $state .= wfWikiID();

                // If we have a secret key or proxy key set then throw it into the state as well
                global $wgSecretKey, $wgProxyKey;
                if ( $wgSecretKey ) {
                        $state .= $wgSecretKey;
                } elseif ( $wgProxyKey ) {
                        $state .= $wgProxyKey;
                }

                return $state;
        }

        protected function driftHash( $data ) {
                // Minimum number of iterations (to avoid slow operations causing the loop to gather little entropy)
                $minIterations = self::MIN_ITERATIONS;
                // Duration of time to spend doing calculations (in seconds)
                $duration = ( self::MSEC_PER_BYTE / 1000 ) * $this->hashLength();
                // Create a buffer to use to trigger memory operations
                $bufLength = 10000000;
                $buffer = str_repeat( ' ', $bufLength );
                $bufPos = 0;

                // Iterate for $duration seconds or at least $minIerations number of iterations
                $iterations = 0;
                $startTime = microtime( true );
                $currentTime = $startTime;
                while ( $iterations < $minIterations || $currentTime - $startTime < $duration ) {
                        // Trigger some memory writing to trigger some bus activity
                        // This may create variance in the time between iterations
                        $bufPos = ( $bufPos + 13 ) % $bufLength;
                        $buffer[$bufPos] = ' ';
                        // Add the drift between this iteration and the last in as entropy
                        $nextTime = microtime( true );
                        $delta = (int)( ( $nextTime - $currentTime ) * 1000000 );
                        $data .= $delta;
                        // Every 100 iterations hash the data and entropy
                        if ( $iterations % 100 === 0 ) {
                                $data = sha1( $data );
                        }
                        $currentTime = $nextTime;
                        $iterations++;
                }
                $timeTaken = $currentTime - $startTime;
                $data = $this->hash( $data );

                wfDebug( __METHOD__ . ": Clock drift calculation " .
                        "(time-taken=" . ( $timeTaken * 1000 ) . "ms, " .
                        "iterations=$iterations, " .
                        "time-per-iteration=" . ( $timeTaken / $iterations * 1e6 ) . "us)\n" );
                return $data;
        }

        protected function randomState() {
                static $state = null;
                if ( is_null( $state ) ) {
                        // Initialize the state with whatever unstable data we can find
                        // It's important that this data is hashed right afterwards to prevent
                        // it from being leaked into the output stream
                        $state = $this->hash( $this->initialRandomState() );
                }
                // Generate a new random state based on the initial random state or previous
                // random state by combining it with clock drift
                $state = $this->driftHash( $state );
                return $state;
        }

        protected function hashAlgo() {
                if ( !is_null( $this->algo ) ) {
                        return $this->algo;
                }

                $algos = hash_algos();
                $preference = array( 'whirlpool', 'sha256', 'sha1', 'md5' );

                foreach ( $preference as $algorithm ) {
                        if ( in_array( $algorithm, $algos ) ) {
                                $this->algo = $algorithm;
                                wfDebug( __METHOD__ . ": Using the {$this->algo} hash algorithm.\n" );
                                return $this->algo;
                        }
                }

                // We only reach here if no acceptable hash is found in the list, this should
                // be a technical impossibility since most of php's hash list is fixed and
                // some of the ones we list are available as their own native functions
                // But since we already require at least 5.2 and hash() was default in
                // 5.1.2 we don't bother falling back to methods like sha1 and md5.
                throw new MWException( "Could not find an acceptable hashing function in hash_algos()" );
        }

        protected function hashLength() {
                if ( is_null( $this->hashLength ) ) {
                        $this->hashLength = strlen( $this->hash( '' ) );
                }
                return $this->hashLength;
        }

        protected function hash( $data ) {
                return hash( $this->hashAlgo(), $data, true );
        }

        protected function hmac( $data, $key ) {
                return hash_hmac( $this->hashAlgo(), $data, $key, true );
        }

        public function realWasStrong() {
                if ( is_null( $this->strong ) ) {
                        throw new MWException( __METHOD__ . ' called before generation of random data' );
                }
                return $this->strong;
        }

        public function realGenerate( $bytes, $forceStrong = false ) {
                wfProfileIn( __METHOD__ );

                wfDebug( __METHOD__ . ": Generating cryptographic random bytes for " . wfGetAllCallers( 5 ) . "\n" );

                $bytes = floor( $bytes );
                static $buffer = '';
                if ( is_null( $this->strong ) ) {
                        // Set strength to false initially until we know what source data is coming from
                        $this->strong = true;
                }

                if ( strlen( $buffer ) < $bytes ) {
                        // If available make use of mcrypt_create_iv URANDOM source to generate randomness
                        // On unix-like systems this reads from /dev/urandom but does it without any buffering
                        // and bypasses openbasdir restrictions so it's preferable to reading directly
                        // On Windows starting in PHP 5.3.0 Windows' native CryptGenRandom is used to generate
                        // entropy so this is also preferable to just trying to read urandom because it may work
                        // on Windows systems as well.
                        if ( function_exists( 'mcrypt_create_iv' ) ) {
                                wfProfileIn( __METHOD__ . '-mcrypt' );
                                $rem = $bytes - strlen( $buffer );
                                $iv = mcrypt_create_iv( $rem, MCRYPT_DEV_URANDOM );
                                if ( $iv === false ) {
                                        wfDebug( __METHOD__ . ": mcrypt_create_iv returned false.\n" );
                                } else {
                                        $buffer .= $iv;
                                        wfDebug( __METHOD__ . ": mcrypt_create_iv generated " . strlen( $iv ) . " bytes of randomness.\n" );
                                }
                                wfProfileOut( __METHOD__ . '-mcrypt' );
                        }
                }

                if ( strlen( $buffer ) < $bytes ) {
                        // If available make use of openssl's random_pesudo_bytes method to attempt to generate randomness.
                        // However don't do this on Windows with PHP < 5.3.4 due to a bug:
                        // http://stackoverflow.com/questions/1940168/openssl-random-pseudo-bytes-is-slow-php
                        if ( function_exists( 'openssl_random_pseudo_bytes' )
                                && ( !wfIsWindows() || version_compare( PHP_VERSION, '5.3.4', '>=' ) )
                        ) {
                                wfProfileIn( __METHOD__ . '-openssl' );
                                $rem = $bytes - strlen( $buffer );
                                $openssl_bytes = openssl_random_pseudo_bytes( $rem, $openssl_strong );
                                if ( $openssl_bytes === false ) {
                                        wfDebug( __METHOD__ . ": openssl_random_pseudo_bytes returned false.\n" );
                                } else {
                                        $buffer .= $openssl_bytes;
                                        wfDebug( __METHOD__ . ": openssl_random_pseudo_bytes generated " . strlen( $openssl_bytes ) . " bytes of " . ( $openssl_strong ? "strong" : "weak" ) . " randomness.\n" );
                                }
                                if ( strlen( $buffer ) >= $bytes ) {
                                        // openssl tells us if the random source was strong, if some of our data was generated
                                        // using it use it's say on whether the randomness is strong
                                        $this->strong = !!$openssl_strong;
                                }
                                wfProfileOut( __METHOD__ . '-openssl' );
                        }
                }

                // Only read from urandom if we can control the buffer size or were passed forceStrong
                if ( strlen( $buffer ) < $bytes && ( function_exists( 'stream_set_read_buffer' ) || $forceStrong ) ) {
                        wfProfileIn( __METHOD__ . '-fopen-urandom' );
                        $rem = $bytes - strlen( $buffer );
                        if ( !function_exists( 'stream_set_read_buffer' ) && $forceStrong ) {
                                wfDebug( __METHOD__ . ": Was forced to read from /dev/urandom without control over the buffer size.\n" );
                        }
                        // /dev/urandom is generally considered the best possible commonly
                        // available random source, and is available on most *nix systems.
                        wfSuppressWarnings();
                        $urandom = fopen( "/dev/urandom", "rb" );
                        wfRestoreWarnings();

                        // Attempt to read all our random data from urandom
                        // php's fread always does buffered reads based on the stream's chunk_size
                        // so in reality it will usually read more than the amount of data we're
                        // asked for and not storing that risks depleting the system's random pool.
                        // If stream_set_read_buffer is available set the chunk_size to the amount
                        // of data we need. Otherwise read 8k, php's default chunk_size.
                        if ( $urandom ) {
                                // php's default chunk_size is 8k
                                $chunk_size = 1024 * 8;
                                if ( function_exists( 'stream_set_read_buffer' ) ) {
                                        // If possible set the chunk_size to the amount of data we need
                                        stream_set_read_buffer( $urandom, $rem );
                                        $chunk_size = $rem;
                                }
                                $random_bytes = fread( $urandom, max( $chunk_size, $rem ) );
                                $buffer .= $random_bytes;
                                fclose( $urandom );
                                wfDebug( __METHOD__ . ": /dev/urandom generated " . strlen( $random_bytes ) . " bytes of randomness.\n" );
                                if ( strlen( $buffer ) >= $bytes ) {
                                        // urandom is always strong, set to true if all our data was generated using it
                                        $this->strong = true;
                                }
                        } else {
                                wfDebug( __METHOD__ . ": /dev/urandom could not be opened.\n" );
                        }
                        wfProfileOut( __METHOD__ . '-fopen-urandom' );
                }

                // If we cannot use or generate enough data from a secure source
                // use this loop to generate a good set of pseudo random data.
                // This works by initializing a random state using a pile of unstable data
                // and continually shoving it through a hash along with a variable salt.
                // We hash the random state with more salt to avoid the state from leaking
                // out and being used to predict the /randomness/ that follows.
                if ( strlen( $buffer ) < $bytes ) {
                        wfDebug( __METHOD__ . ": Falling back to using a pseudo random state to generate randomness.\n" ); 
                }
                while ( strlen( $buffer ) < $bytes ) {
                        wfProfileIn( __METHOD__ . '-fallback' );
                        $buffer .= $this->hmac( $this->randomState(), mt_rand() );
                        // This code is never really cryptographically strong, if we use it
                        // at all, then set strong to false.
                        $this->strong = false;
                        wfProfileOut( __METHOD__ . '-fallback' );
                }

                // Once the buffer has been filled up with enough random data to fulfill
                // the request shift off enough data to handle the request and leave the
                // unused portion left inside the buffer for the next request for random data
                $generated = substr( $buffer, 0, $bytes );
                $buffer = substr( $buffer, $bytes );

                wfDebug( __METHOD__ . ": " . strlen( $buffer ) . " bytes of randomness leftover in the buffer.\n" );

                wfProfileOut( __METHOD__ );
                return $generated;
        }

        public function realGenerateHex( $chars, $forceStrong = false ) {
                // hex strings are 2x the length of raw binary so we divide the length in half
                // odd numbers will result in a .5 that leads the generate() being 1 character
                // short, so we use ceil() to ensure that we always have enough bytes
                $bytes = ceil( $chars / 2 );
                // Generate the data and then convert it to a hex string
                $hex = bin2hex( $this->generate( $bytes, $forceStrong ) );
                // A bit of paranoia here, the caller asked for a specific length of string
                // here, and it's possible (eg when given an odd number) that we may actually
                // have at least 1 char more than they asked for. Just in case they made this
                // call intending to insert it into a database that does truncation we don't
                // want to give them too much and end up with their database and their live
                // code having two different values because part of what we gave them is truncated
                // hence, we strip out any run of characters longer than what we were asked for.
                return substr( $hex, 0, $chars );
        }

        protected static function singleton() {
                if ( is_null( self::$singleton ) ) {
                        self::$singleton = new self;
                }
                return self::$singleton;
        }

        public static function wasStrong() {
                return self::singleton()->realWasStrong();
        }

        public static function generate( $bytes, $forceStrong = false ) {
                return self::singleton()->realGenerate( $bytes, $forceStrong );
        }

        public static function generateHex( $chars, $forceStrong = false ) {
                return self::singleton()->realGenerateHex( $chars, $forceStrong );
        }

}