2 * Copyright (C) 2012, 2013
6 * Permission is hereby granted, free of charge, to any person obtaining a copy of
7 * this software and associated documentation files (the "Software"), to deal in
8 * the Software without restriction, including without limitation the rights to
9 * use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies
10 * of the Software, and to permit persons to whom the Software is furnished to do
11 * so, subject to the following conditions:
13 * The above copyright notice and this permission notice shall be included in all
14 * copies or substantial portions of the Software.
16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
28 /* TODO: remove globals ... */
29 static uint64_t mem_ab = 0;
30 static uint64_t mem_db = 0;
31 static uint64_t mem_at = 0;
32 static uint64_t mem_dt = 0;
33 static uint64_t mem_pk = 0;
34 static uint64_t mem_hw = 0;
40 struct memblock_t *next;
41 struct memblock_t *prev;
46 if (mem_hw > mem_pk) \
50 static struct memblock_t *mem_start = NULL;
52 void *util_memory_a(size_t byte, unsigned int line, const char *file) {
53 struct memblock_t *info = (struct memblock_t*)malloc(sizeof(struct memblock_t) + byte);
54 void *data = (void*)(info+1);
55 if (!info) return NULL;
60 info->next = mem_start;
62 mem_start->prev = info;
74 void util_memory_d(void *ptrn) {
75 struct memblock_t *info = NULL;
78 info = ((struct memblock_t*)ptrn - 1);
85 info->prev->next = info->next;
87 info->next->prev = info->prev;
88 if (info == mem_start)
89 mem_start = info->next;
94 void *util_memory_r(void *ptrn, size_t byte, unsigned int line, const char *file) {
95 struct memblock_t *oldinfo = NULL;
97 struct memblock_t *newinfo;
100 return util_memory_a(byte, line, file);
106 oldinfo = ((struct memblock_t*)ptrn - 1);
107 newinfo = ((struct memblock_t*)malloc(sizeof(struct memblock_t) + byte));
111 util_memory_d(oldinfo+1);
116 memcpy(newinfo+1, oldinfo+1, oldinfo->byte);
120 oldinfo->prev->next = oldinfo->next;
122 oldinfo->next->prev = oldinfo->prev;
123 if (oldinfo == mem_start)
124 mem_start = oldinfo->next;
127 newinfo->line = line;
128 newinfo->byte = byte;
129 newinfo->file = file;
130 newinfo->prev = NULL;
131 newinfo->next = mem_start;
133 mem_start->prev = newinfo;
136 mem_ab -= oldinfo->byte;
137 mem_hw -= oldinfo->byte;
138 mem_ab += newinfo->byte;
139 mem_hw += newinfo->byte;
148 static void util_dumpmem(struct memblock_t *memory, uint16_t cols) {
150 for (i = 0; i < memory->byte + ((memory->byte % cols) ? (cols - memory->byte % cols) : 0); i++) {
151 if (i % cols == 0) con_out(" 0x%06X: ", i);
152 if (i < memory->byte) con_out("%02X " , 0xFF & ((char*)(memory + 1))[i]);
155 if ((uint16_t)(i % cols) == (cols - 1)) {
156 for (j = i - (cols - 1); j <= i; j++) {
160 : (isprint(((char*)(memory + 1))[j]))
161 ? 0xFF & ((char*)(memory + 1)) [j]
169 static uint64_t vectors = 0;
170 void util_meminfo() {
171 struct memblock_t *info;
173 if (OPTS_OPTION_BOOL(OPTION_DEBUG)) {
174 for (info = mem_start; info; info = info->next) {
175 con_out("lost: %u (bytes) at %s:%u\n",
180 util_dumpmem(info, OPTS_OPTION_U16(OPTION_MEMDUMPCOLS));
184 con_out("Additional Statistics:\n Total vectors used: %lu\n",
188 if (OPTS_OPTION_BOOL(OPTION_DEBUG) ||
189 OPTS_OPTION_BOOL(OPTION_MEMCHK)) {
190 con_out("Memory information:\n\
191 Total allocations: %llu\n\
192 Total deallocations: %llu\n\
193 Total allocated: %f (MB)\n\
194 Total deallocated: %f (MB)\n\
195 Total peak memory: %f (MB)\n\
196 Total leaked memory: %f (MB) in %llu allocations\n",
199 (float)(mem_ab) / 1048576.0f,
200 (float)(mem_db) / 1048576.0f,
201 (float)(mem_pk) / 1048576.0f,
202 (float)(mem_ab - mem_db) / 1048576.0f,
204 /* could be more clever */
211 * Some string utility functions, because strdup uses malloc, and we want
212 * to track all memory (without replacing malloc).
214 char *_util_Estrdup(const char *s, const char *file, size_t line) {
218 /* in case of -DNOTRACK */
225 if ((len = strlen(s)) && (ptr = (char*)mem_af(len+1, line, file))) {
232 char *_util_Estrdup_empty(const char *s, const char *file, size_t line) {
236 /* in case of -DNOTRACK */
244 if ((ptr = (char*)mem_af(len+1, line, file))) {
251 void util_debug(const char *area, const char *ms, ...) {
253 if (!OPTS_OPTION_BOOL(OPTION_DEBUG))
256 if (!strcmp(area, "MEM") && !OPTS_OPTION_BOOL(OPTION_MEMCHK))
260 con_out ("[%s] ", area);
266 * only required if big endian .. otherwise no need to swap
269 #if PLATFORM_BYTE_ORDER == GMQCC_BYTE_ORDER_BIG
270 static GMQCC_INLINE void util_swap16(uint16_t *d, size_t l) {
272 d[l] = (d[l] << 8) | (d[l] >> 8);
276 static GMQCC_INLINE void util_swap32(uint32_t *d, size_t l) {
279 v = ((d[l] << 8) & 0xFF00FF00) | ((d[l] >> 8) & 0x00FF00FF);
280 d[l] = (v << 16) | (v >> 16);
284 /* Some strange system doesn't like constants that big, AND doesn't recognize an ULL suffix
285 * so let's go the safe way
287 static GMQCC_INLINE void util_swap64(uint32_t *d, size_t l) {
291 v = ((d[l] << 8) & 0xFF00FF00FF00FF00) | ((d[l] >> 8) & 0x00FF00FF00FF00FF);
292 v = ((v << 16) & 0xFFFF0000FFFF0000) | ((v >> 16) & 0x0000FFFF0000FFFF);
293 d[l] = (v << 32) | (v >> 32);
297 for (i = 0; i < l; i += 2) {
306 void util_endianswap(void *_data, size_t length, unsigned int typesize) {
307 # if PLATFORM_BYTE_ORDER == -1 /* runtime check */
308 if (*((char*)&typesize))
311 /* prevent unused warnings */
316 # if PLATFORM_BYTE_ORDER == GMQCC_BYTE_ORDER_LITTLE
322 util_swap16((uint16_t*)_data, length>>1);
325 util_swap32((uint32_t*)_data, length>>2);
328 util_swap64((uint32_t*)_data, length>>3);
331 default: exit(EXIT_FAILURE); /* please blow the fuck up! */
338 * CRC algorithms vary in the width of the polynomial, the value of said polynomial,
339 * the initial value used for the register, weather the bits of each byte are reflected
340 * before being processed, weather the algorithm itself feeds input bytes through the
341 * register or XORs them with a byte from one end and then straight into the table, as
342 * well as (but not limited to the idea of reflected versions) where the final register
343 * value becomes reversed, and finally weather the value itself is used to XOR the final
344 * register value. AS such you can already imagine how painfully annoying CRCs are,
345 * of course we stand to target Quake, which expects it's certian set of rules for proper
346 * calculation of a CRC.
348 * In most traditional CRC algorithms on uses a reflected table driven method where a value
349 * or register is reflected if it's bits are swapped around it's center. For example:
350 * take the bits 0101 is the 4-bit reflection of 1010, and respectfully 0011 would be the
351 * reflection of 1100. Quake however expects a NON-Reflected CRC on the output, but still
352 * requires a final XOR on the values (0xFFFF and 0x0000) this is a standard CCITT CRC-16
353 * which I respectfully as a programmer don't agree with.
355 * So now you know what we target, and why we target it, despite how unsettling it may seem
356 * but those are what Quake seems to request.
359 static const uint16_t util_crc16_table[] = {
360 0x0000, 0x1021, 0x2042, 0x3063, 0x4084, 0x50A5,
361 0x60C6, 0x70E7, 0x8108, 0x9129, 0xA14A, 0xB16B,
362 0xC18C, 0xD1AD, 0xE1CE, 0xF1EF, 0x1231, 0x0210,
363 0x3273, 0x2252, 0x52B5, 0x4294, 0x72F7, 0x62D6,
364 0x9339, 0x8318, 0xB37B, 0xA35A, 0xD3BD, 0xC39C,
365 0xF3FF, 0xE3DE, 0x2462, 0x3443, 0x0420, 0x1401,
366 0x64E6, 0x74C7, 0x44A4, 0x5485, 0xA56A, 0xB54B,
367 0x8528, 0x9509, 0xE5EE, 0xF5CF, 0xC5AC, 0xD58D,
368 0x3653, 0x2672, 0x1611, 0x0630, 0x76D7, 0x66F6,
369 0x5695, 0x46B4, 0xB75B, 0xA77A, 0x9719, 0x8738,
370 0xF7DF, 0xE7FE, 0xD79D, 0xC7BC, 0x48C4, 0x58E5,
371 0x6886, 0x78A7, 0x0840, 0x1861, 0x2802, 0x3823,
372 0xC9CC, 0xD9ED, 0xE98E, 0xF9AF, 0x8948, 0x9969,
373 0xA90A, 0xB92B, 0x5AF5, 0x4AD4, 0x7AB7, 0x6A96,
374 0x1A71, 0x0A50, 0x3A33, 0x2A12, 0xDBFD, 0xCBDC,
375 0xFBBF, 0xEB9E, 0x9B79, 0x8B58, 0xBB3B, 0xAB1A,
376 0x6CA6, 0x7C87, 0x4CE4, 0x5CC5, 0x2C22, 0x3C03,
377 0x0C60, 0x1C41, 0xEDAE, 0xFD8F, 0xCDEC, 0xDDCD,
378 0xAD2A, 0xBD0B, 0x8D68, 0x9D49, 0x7E97, 0x6EB6,
379 0x5ED5, 0x4EF4, 0x3E13, 0x2E32, 0x1E51, 0x0E70,
380 0xFF9F, 0xEFBE, 0xDFDD, 0xCFFC, 0xBF1B, 0xAF3A,
381 0x9F59, 0x8F78, 0x9188, 0x81A9, 0xB1CA, 0xA1EB,
382 0xD10C, 0xC12D, 0xF14E, 0xE16F, 0x1080, 0x00A1,
383 0x30C2, 0x20E3, 0x5004, 0x4025, 0x7046, 0x6067,
384 0x83B9, 0x9398, 0xA3FB, 0xB3DA, 0xC33D, 0xD31C,
385 0xE37F, 0xF35E, 0x02B1, 0x1290, 0x22F3, 0x32D2,
386 0x4235, 0x5214, 0x6277, 0x7256, 0xB5EA, 0xA5CB,
387 0x95A8, 0x8589, 0xF56E, 0xE54F, 0xD52C, 0xC50D,
388 0x34E2, 0x24C3, 0x14A0, 0x0481, 0x7466, 0x6447,
389 0x5424, 0x4405, 0xA7DB, 0xB7FA, 0x8799, 0x97B8,
390 0xE75F, 0xF77E, 0xC71D, 0xD73C, 0x26D3, 0x36F2,
391 0x0691, 0x16B0, 0x6657, 0x7676, 0x4615, 0x5634,
392 0xD94C, 0xC96D, 0xF90E, 0xE92F, 0x99C8, 0x89E9,
393 0xB98A, 0xA9AB, 0x5844, 0x4865, 0x7806, 0x6827,
394 0x18C0, 0x08E1, 0x3882, 0x28A3, 0xCB7D, 0xDB5C,
395 0xEB3F, 0xFB1E, 0x8BF9, 0x9BD8, 0xABBB, 0xBB9A,
396 0x4A75, 0x5A54, 0x6A37, 0x7A16, 0x0AF1, 0x1AD0,
397 0x2AB3, 0x3A92, 0xFD2E, 0xED0F, 0xDD6C, 0xCD4D,
398 0xBDAA, 0xAD8B, 0x9DE8, 0x8DC9, 0x7C26, 0x6C07,
399 0x5C64, 0x4C45, 0x3CA2, 0x2C83, 0x1CE0, 0x0CC1,
400 0xEF1F, 0xFF3E, 0xCF5D, 0xDF7C, 0xAF9B, 0xBFBA,
401 0x8FD9, 0x9FF8, 0x6E17, 0x7E36, 0x4E55, 0x5E74,
402 0x2E93, 0x3EB2, 0x0ED1, 0x1EF0
405 /* Non - Reflected */
406 uint16_t util_crc16(uint16_t current, const char *k, size_t len) {
407 register uint16_t h = current;
408 for (; len; --len, ++k)
409 h = util_crc16_table[(h>>8)^((unsigned char)*k)]^(h<<8);
412 /* Reflective Varation (for reference) */
414 uint16_t util_crc16(const char *k, int len, const short clamp) {
415 register uint16_t h= (uint16_t)0xFFFFFFFF;
416 for (; len; --len, ++k)
417 h = util_crc16_table[(h^((unsigned char)*k))&0xFF]^(h>>8);
422 size_t util_strtocmd(const char *in, char *out, size_t outsz) {
424 for (; *in && sz < outsz; ++in, ++out, ++sz)
425 *out = (*in == '-') ? '_' : (isalpha(*in) && !isupper(*in)) ? *in + 'A' - 'a': *in;
430 size_t util_strtononcmd(const char *in, char *out, size_t outsz) {
432 for (; *in && sz < outsz; ++in, ++out, ++sz)
433 *out = (*in == '_') ? '-' : (isalpha(*in) && isupper(*in)) ? *in + 'a' - 'A' : *in;
438 /* TODO: rewrite ... when I redo the ve cleanup */
439 void _util_vec_grow(void **a, size_t i, size_t s) {
440 vector_t *d = vec_meta(*a);
445 m = 2 * d->allocated + i;
446 p = mem_r(d, s * m + sizeof(vector_t));
449 p = mem_a(s * m + sizeof(vector_t));
450 ((vector_t*)p)->used = 0;
454 *a = (vector_t*)p + 1;
455 vec_meta(*a)->allocated = m;
459 * Hash table for generic data, based on dynamic memory allocations
460 * all around. This is the internal interface, please look for
461 * EXPOSED INTERFACE comment below
463 typedef struct hash_node_t {
464 char *key; /* the key for this node in table */
465 void *value; /* pointer to the data as void* */
466 struct hash_node_t *next; /* next node (linked list) */
469 GMQCC_INLINE size_t util_hthash(hash_table_t *ht, const char *key) {
470 const uint32_t mix = 0x5BD1E995;
471 const uint32_t rot = 24;
472 size_t size = strlen(key);
473 uint32_t hash = 0x1EF0 /* LICRC TAB */ ^ size;
475 const unsigned char *data = (const unsigned char*)key;
478 alias = (data[0] | (data[1] << 8) | (data[2] << 16) | (data[3] << 24));
480 alias ^= alias >> rot;
491 case 3: hash ^= data[2] << 16;
492 case 2: hash ^= data[1] << 8;
493 case 1: hash ^= data[0];
501 return (size_t) (hash % ht->size);
504 static hash_node_t *_util_htnewpair(const char *key, void *value) {
506 if (!(node = (hash_node_t*)mem_a(sizeof(hash_node_t))))
509 if (!(node->key = util_strdupe(key))) {
521 * EXPOSED INTERFACE for the hashtable implementation
522 * util_htnew(size) -- to make a new hashtable
523 * util_htset(table, key, value, sizeof(value)) -- to set something in the table
524 * util_htget(table, key) -- to get something from the table
525 * util_htdel(table) -- to delete the table
527 hash_table_t *util_htnew(size_t size) {
528 hash_table_t *hashtable = NULL;
532 if (!(hashtable = (hash_table_t*)mem_a(sizeof(hash_table_t))))
535 if (!(hashtable->table = (hash_node_t**)mem_a(sizeof(hash_node_t*) * size))) {
540 hashtable->size = size;
541 memset(hashtable->table, 0, sizeof(hash_node_t*) * size);
546 void util_htseth(hash_table_t *ht, const char *key, size_t bin, void *value) {
547 hash_node_t *newnode = NULL;
548 hash_node_t *next = NULL;
549 hash_node_t *last = NULL;
551 next = ht->table[bin];
553 while (next && next->key && strcmp(key, next->key) > 0)
554 last = next, next = next->next;
556 /* already in table, do a replace */
557 if (next && next->key && strcmp(key, next->key) == 0) {
560 /* not found, grow a pair man :P */
561 newnode = _util_htnewpair(key, value);
562 if (next == ht->table[bin]) {
563 newnode->next = next;
564 ht->table[bin] = newnode;
566 last->next = newnode;
568 newnode->next = next;
569 last->next = newnode;
574 void util_htset(hash_table_t *ht, const char *key, void *value) {
575 util_htseth(ht, key, util_hthash(ht, key), value);
578 void *util_htgeth(hash_table_t *ht, const char *key, size_t bin) {
579 hash_node_t *pair = ht->table[bin];
581 while (pair && pair->key && strcmp(key, pair->key) > 0)
584 if (!pair || !pair->key || strcmp(key, pair->key) != 0)
590 void *util_htget(hash_table_t *ht, const char *key) {
591 return util_htgeth(ht, key, util_hthash(ht, key));
594 void *code_util_str_htgeth(hash_table_t *ht, const char *key, size_t bin) {
599 keylen = strlen(key);
601 pair = ht->table[bin];
602 while (pair && pair->key) {
603 len = strlen(pair->key);
609 cmp = strcmp(key, pair->key);
617 cmp = strcmp(key, pair->key + len - keylen);
619 uintptr_t up = (uintptr_t)pair->value;
629 * Free all allocated data in a hashtable, this is quite the amount
632 void util_htrem(hash_table_t *ht, void (*callback)(void *data)) {
634 for (; i < ht->size; i++) {
635 hash_node_t *n = ht->table[i];
655 void util_htrmh(hash_table_t *ht, const char *key, size_t bin, void (*cb)(void*)) {
656 hash_node_t **pair = &ht->table[bin];
659 while (*pair && (*pair)->key && strcmp(key, (*pair)->key) > 0)
660 pair = &(*pair)->next;
663 if (!tmp || !tmp->key || strcmp(key, tmp->key) != 0)
674 void util_htrm(hash_table_t *ht, const char *key, void (*cb)(void*)) {
675 util_htrmh(ht, key, util_hthash(ht, key), cb);
678 void util_htdel(hash_table_t *ht) {
679 util_htrem(ht, NULL);
683 * Portable implementation of vasprintf/asprintf. Assumes vsnprintf
684 * exists, otherwise compiler error.
686 * TODO: fix for MSVC ....
688 int util_vasprintf(char **dat, const char *fmt, va_list args) {
694 * For visuals tido _vsnprintf doesn't tell you the length of a
695 * formatted string if it overflows. However there is a MSVC
696 * intrinsic (which is documented wrong) called _vcsprintf which
697 * will return the required amount to allocate.
700 if ((len = _vscprintf(fmt, args)) < 0) {
705 tmp = (char*)mem_a(len + 1);
706 if ((ret = _vsnprintf_s(tmp, len+1, len+1, fmt, args)) != len) {
715 * For everything else we have a decent conformint vsnprintf that
716 * returns the number of bytes needed. We give it a try though on
717 * a short buffer, since efficently speaking, it could be nice to
718 * above a second vsnprintf call.
723 len = vsnprintf(buf, sizeof(buf), fmt, cpy);
726 if (len < (int)sizeof(buf)) {
727 *dat = util_strdup(buf);
731 /* not large enough ... */
732 tmp = (char*)mem_a(len + 1);
733 if ((ret = vsnprintf(tmp, len + 1, fmt, args)) != len) {
743 int util_asprintf(char **ret, const char *fmt, ...) {
747 read = util_vasprintf(ret, fmt, args);
754 * These are various re-implementations (wrapping the real ones) of
755 * string functions that MSVC consideres unsafe. We wrap these up and
756 * use the safe varations on MSVC.
759 static char **util_strerror_allocated() {
760 static char **data = NULL;
764 static void util_strerror_cleanup(void) {
766 char **data = util_strerror_allocated();
767 for (i = 0; i < vec_size(data); i++)
772 const char *util_strerror(int num) {
773 char *allocated = NULL;
774 static bool install = false;
775 static size_t tries = 0;
776 char **vector = util_strerror_allocated();
778 /* try installing cleanup handler */
783 install = !atexit(&util_strerror_cleanup);
787 allocated = (char*)mem_a(4096); /* A page must be enough */
788 strerror_s(allocated, 4096, num);
790 vec_push(vector, allocated);
791 return (const char *)allocated;
794 int util_snprintf(char *src, size_t bytes, const char *format, ...) {
797 va_start(va, format);
799 rt = vsprintf_s(src, bytes, format, va);
805 char *util_strcat(char *dest, const char *src) {
806 strcat_s(dest, strlen(src), src);
810 char *util_strncpy(char *dest, const char *src, size_t num) {
811 strncpy_s(dest, num, src, num);
815 const char *util_strerror(int num) {
816 return strerror(num);
819 int util_snprintf(char *src, size_t bytes, const char *format, ...) {
822 va_start(va, format);
823 rt = vsnprintf(src, bytes, format, va);
829 char *util_strcat(char *dest, const char *src) {
830 return strcat(dest, src);
833 char *util_strncpy(char *dest, const char *src, size_t num) {
834 return strncpy(dest, src, num);
837 #endif /*! _MSC_VER */
840 * Implementation of the Mersenne twister PRNG (pseudo random numer
841 * generator). Implementation of MT19937. Has a period of 2^19937-1
842 * which is a Mersenne Prime (hence the name).
844 * Implemented from specification and original paper:
845 * http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/ARTICLES/mt.pdf
847 * This code is placed in the public domain by me personally
848 * (Dale Weiler, a.k.a graphitemaster).
852 #define MT_PERIOD 397
853 #define MT_SPACE (MT_SIZE - MT_PERIOD)
855 static uint32_t mt_state[MT_SIZE];
856 static size_t mt_index = 0;
858 static GMQCC_INLINE void mt_generate() {
860 * The loop has been unrolled here: the original paper and implemenation
861 * Called for the following code:
862 * for (register unsigned i = 0; i < MT_SIZE; ++i) {
863 * register uint32_t load;
864 * load = (0x80000000 & mt_state[i]) // most significant 32nd bit
865 * load |= (0x7FFFFFFF & mt_state[(i + 1) % MT_SIZE]) // least significant 31nd bit
867 * mt_state[i] = mt_state[(i + MT_PERIOD) % MT_SIZE] ^ (load >> 1);
869 * if (load & 1) mt_state[i] ^= 0x9908B0DF;
872 * This essentially is a waste: we have two modulus operations, and
873 * a branch that is executed every iteration from [0, MT_SIZE).
875 * Please see: http://www.quadibloc.com/crypto/co4814.htm for more
876 * information on how this clever trick works.
878 static const uint32_t matrix[2] = {
883 * This register gives up a little more speed by instructing the compiler
884 * to force these into CPU registers (they're counters for indexing mt_state
885 * which we can force the compiler to generate prefetch instructions for)
891 * Said loop has been unrolled for MT_SPACE (226 iterations), opposed
892 * to [0, MT_SIZE) (634 iterations).
894 for (i = 0; i < MT_SPACE; ++i) {
895 y = (0x80000000 & mt_state[i]) | (0x7FFFFFF & mt_state[i + 1]);
896 mt_state[i] = mt_state[i + MT_PERIOD] ^ (y >> 1) ^ matrix[y & 1];
898 i ++; /* loop unroll */
900 y = (0x80000000 & mt_state[i]) | (0x7FFFFFF & mt_state[i + 1]);
901 mt_state[i] = mt_state[i + MT_PERIOD] ^ (y >> 1) ^ matrix[y & 1];
905 * collapsing the walls unrolled (evenly dividing 396 [632-227 = 396
909 while (i < MT_SIZE - 1) {
911 * We expand this 11 times .. manually, no macros are required
912 * here. This all fits in the CPU cache.
914 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
915 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
917 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
918 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
920 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
921 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
923 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
924 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
926 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
927 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
929 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
930 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
932 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
933 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
935 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
936 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
938 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
939 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
941 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
942 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
944 y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]);
945 mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1];
949 /* i = mt_state[623] */
950 y = (0x80000000 & mt_state[MT_SIZE - 1]) | (0x7FFFFFFF & mt_state[MT_SIZE - 1]);
951 mt_state[MT_SIZE - 1] = mt_state[MT_PERIOD - 1] ^ (y >> 1) ^ matrix[y & 1];
954 void util_seed(uint32_t value) {
956 * We seed the mt_state with a LCG (linear congruential generator)
957 * We're operating exactly on exactly m=32, so there is no need to
960 * The multipler of choice is 0x6C07865, also knows as the Borosh-
961 * Niederreiter multipler used for modulus 2^32. More can be read
962 * about this in Knuth's TAOCP Volume 2, page 106.
964 * If you don't own TAOCP something is wrong with you :-) .. so I
965 * also provided a link to the original paper by Borosh and
966 * Niederreiter. It's called "Optional Multipliers for PRNG by The
967 * Linear Congruential Method" (1983).
968 * http://en.wikipedia.org/wiki/Linear_congruential_generator
970 * From said page, it says the following:
971 * "A common Mersenne twister implementation, interestingly enough
972 * used an LCG to generate seed data."
975 * The data we're operating on is 32-bits for the mt_state array, so
976 * there is no masking required with 0xFFFFFFFF
981 for (i = 1; i < MT_SIZE; ++i)
982 mt_state[i] = 0x6C078965 * (mt_state[i - 1] ^ mt_state[i - 1] >> 30) + i;
985 uint32_t util_rand() {
989 * This is inlined with any sane compiler (I checked)
990 * for some reason though, SubC seems to be generating invalid
991 * code when it inlines this.
996 y = mt_state[mt_index];
998 /* Standard tempering */
999 y ^= y >> 11; /* +7 */
1000 y ^= y << 7 & 0x9D2C5680; /* +4 */
1001 y ^= y << 15 & 0xEFC60000; /* -4 */
1002 y ^= y >> 18; /* -7 */
1004 if(++mt_index == MT_SIZE)