X-Git-Url: https://git.xonotic.org/?a=blobdiff_plain;f=util.c;h=77753a9be006d3add446d1a3bde7c9d9924091fc;hb=c8c25ef6f7b6182c7fb45032fdc12493f12d9670;hp=136b12a6e92124ee107d1abd379a67f732efdbc3;hpb=cda507635685e2d5aa62a45d62fd4dc4c7515179;p=xonotic%2Fgmqcc.git diff --git a/util.c b/util.c index 136b12a..2478304 100644 --- a/util.c +++ b/util.c @@ -1,6 +1,7 @@ /* - * Copyright (C) 2012 + * Copyright (C) 2012, 2013 * Dale Weiler + * Wolfgang Bumiller * * Permission is hereby granted, free of charge, to any person obtaining a copy of * this software and associated documentation files (the "Software"), to deal in @@ -20,268 +21,550 @@ * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE * SOFTWARE. */ -#include -#include -#include "gmqcc.h" +#include +#include -unsigned long long mem_ab = 0; -unsigned long long mem_db = 0; -unsigned long long mem_at = 0; -unsigned long long mem_dt = 0; +#include "gmqcc.h" -struct memblock_t { - const char *file; - unsigned int line; - unsigned int byte; +/* + * Initially this was handled with a table in the gmqcc.h header, but + * much to my surprise the contents of the table was duplicated for + * each translation unit, causing all these strings to be duplicated + * for every .c file it was included into. This method culls back on + * it. This is a 'utility' function because the executor also depends + * on this for dissasembled bytecode. + */ +const char *util_instr_str[VINSTR_END] = { + "DONE", "MUL_F", "MUL_V", "MUL_FV", + "MUL_VF", "DIV_F", "ADD_F", "ADD_V", + "SUB_F", "SUB_V", "EQ_F", "EQ_V", + "EQ_S", "EQ_E", "EQ_FNC", "NE_F", + "NE_V", "NE_S", "NE_E", "NE_FNC", + "LE", "GE", "LT", "GT", + "LOAD_F", "LOAD_V", "LOAD_S", "LOAD_ENT", + "LOAD_FLD", "LOAD_FNC", "ADDRESS", "STORE_F", + "STORE_V", "STORE_S", "STORE_ENT", "STORE_FLD", + "STORE_FNC", "STOREP_F", "STOREP_V", "STOREP_S", + "STOREP_ENT", "STOREP_FLD", "STOREP_FNC", "RETURN", + "NOT_F", "NOT_V", "NOT_S", "NOT_ENT", + "NOT_FNC", "IF", "IFNOT", "CALL0", + "CALL1", "CALL2", "CALL3", "CALL4", + "CALL5", "CALL6", "CALL7", "CALL8", + "STATE", "GOTO", "AND", "OR", + "BITAND", "BITOR" }; -void *util_memory_a(unsigned int byte, unsigned int line, const char *file) { - struct memblock_t *info = malloc(sizeof(struct memblock_t) + byte); - void *data =(void*)((uintptr_t)info+sizeof(struct memblock_t)); - if (!data) return NULL; - info->line = line; - info->byte = byte; - info->file = file; - - util_debug("MEM", "allocation: % 8u (bytes) address 0x%08X @ %s:%u\n", byte, data, file, line); - mem_at++; - mem_ab += info->byte; - return data; -} - -void util_memory_d(void *ptrn, unsigned int line, const char *file) { - if (!ptrn) return; - void *data = (void*)((uintptr_t)ptrn-sizeof(struct memblock_t)); - struct memblock_t *info = (struct memblock_t*)data; - - util_debug("MEM", "released: % 8u (bytes) address 0x%08X @ %s:%u\n", info->byte, data, file, line); - mem_db += info->byte; - mem_dt++; - free(data); -} +void util_debug(const char *area, const char *ms, ...) { + va_list va; + if (!OPTS_OPTION_BOOL(OPTION_DEBUG)) + return; -void util_meminfo() { - if (!opts_memchk) + if (!strcmp(area, "MEM") && !OPTS_OPTION_BOOL(OPTION_MEMCHK)) return; - - util_debug("MEM", "Memory information:\n\ - Total allocations: %llu\n\ - Total deallocations: %llu\n\ - Total allocated: %llu (bytes)\n\ - Total deallocated: %llu (bytes)\n\ - Leaks found: lost %llu (bytes) in %d allocations\n", - mem_at, mem_dt, - mem_ab, mem_db, - (mem_ab - mem_db), - (mem_at - mem_dt) - ); + + va_start(va, ms); + con_out ("[%s] ", area); + con_vout(ms, va); + va_end (va); } /* - * Some string utility functions, because strdup uses malloc, and we want - * to track all memory (without replacing malloc). + * only required if big endian .. otherwise no need to swap + * data. */ -char *util_strdup(const char *s) { - size_t len = 0; - char *ptr = NULL; - - if (!s) - return NULL; - - if ((len = strlen(s)) && (ptr = mem_a(len+1))) { - memcpy(ptr, s, len); - ptr[len] = '\0'; +#if PLATFORM_BYTE_ORDER == GMQCC_BYTE_ORDER_BIG + static GMQCC_INLINE void util_swap16(uint16_t *d, size_t l) { + while (l--) { + d[l] = (d[l] << 8) | (d[l] >> 8); + } } - return ptr; -} -/* - * Remove quotes from a string, escapes from \ in string - * as well. This function shouldn't be used to create a - * char array that is later freed (it uses pointer arith) - */ -char *util_strrq(char *s) { - char *dst = s; - char *src = s; - char chr; - while ((chr = *src++) != '\0') { - if (chr == '\\') { - *dst++ = chr; - if ((chr = *src++) == '\0') - break; - *dst++ = chr; - } else if (chr != '"') - *dst++ = chr; + static GMQCC_INLINE void util_swap32(uint32_t *d, size_t l) { + while (l--) { + uint32_t v; + v = ((d[l] << 8) & 0xFF00FF00) | ((d[l] >> 8) & 0x00FF00FF); + d[l] = (v << 16) | (v >> 16); + } } - *dst = '\0'; - return dst; -} -/* - * Remove newline from a string (if it exists). This is - * done pointer wise instead of strlen(), and an array - * access. - */ -char *util_strrnl(char *src) { - if (!src) return NULL; - char *cpy = src; - while (*cpy && *cpy != '\n') - cpy++; - - *cpy = '\0'; - return src; -} + /* Some strange system doesn't like constants that big, AND doesn't recognize an ULL suffix + * so let's go the safe way + */ + static GMQCC_INLINE void util_swap64(uint32_t *d, size_t l) { + /* + while (l--) { + uint64_t v; + v = ((d[l] << 8) & 0xFF00FF00FF00FF00) | ((d[l] >> 8) & 0x00FF00FF00FF00FF); + v = ((v << 16) & 0xFFFF0000FFFF0000) | ((v >> 16) & 0x0000FFFF0000FFFF); + d[l] = (v << 32) | (v >> 32); + } + */ + size_t i; + for (i = 0; i < l; i += 2) { + uint32_t v1 = d[i]; + d[i] = d[i+1]; + d[i+1] = v1; + util_swap32(d+i, 2); + } + } +#endif -void util_debug(const char *area, const char *ms, ...) { - if (!opts_debug) +void util_endianswap(void *_data, size_t length, unsigned int typesize) { +# if PLATFORM_BYTE_ORDER == -1 /* runtime check */ + if (*((char*)&typesize)) return; - - va_list va; - va_start(va, ms); - fprintf (stdout, "DEBUG: "); - fputc ('[', stdout); - fprintf(stdout, "%s", area); - fputs ("] ", stdout); - vfprintf(stdout, ms, va); - va_end (va); -} +#else + /* prevent unused warnings */ + (void) _data; + (void) length; + (void) typesize; -/* - * Endianess swapping, all data must be stored little-endian. This - * reorders by stride and length, much nicer than other functions for - * certian-sized types like short or int. - */ -void util_endianswap(void *m, int s, int l) { - size_t w = 0; - size_t i = 0; - - /* ignore if we're already LE */ - if(*((char *)&s)) +# if PLATFORM_BYTE_ORDER == GMQCC_BYTE_ORDER_LITTLE return; +# else + switch (typesize) { + case 1: return; + case 2: + util_swap16((uint16_t*)_data, length>>1); + return; + case 4: + util_swap32((uint32_t*)_data, length>>2); + return; + case 8: + util_swap64((uint32_t*)_data, length>>3); + return; - for(; w < l; w++) { - for(; i < s << 1; i++) { - unsigned char *p = (unsigned char *)m+w*s; - unsigned char t = p[i]; - p[i] = p[s-i-1]; - p[s-i-1] = t; + default: exit(EXIT_FAILURE); /* please blow the fuck up! */ } - } +# endif +#endif } /* - * This is an implementation of CRC32. The polynomial has been offline - * computed for faster generation at the cost of larger code size. This - * is used to compute hashes for hash-table. This is not used for crc16 - * generation of progs header calculation. + * CRC algorithms vary in the width of the polynomial, the value of said polynomial, + * the initial value used for the register, weather the bits of each byte are reflected + * before being processed, weather the algorithm itself feeds input bytes through the + * register or XORs them with a byte from one end and then straight into the table, as + * well as (but not limited to the idea of reflected versions) where the final register + * value becomes reversed, and finally weather the value itself is used to XOR the final + * register value. AS such you can already imagine how painfully annoying CRCs are, + * of course we stand to target Quake, which expects it's certian set of rules for proper + * calculation of a CRC. + * + * In most traditional CRC algorithms on uses a reflected table driven method where a value + * or register is reflected if it's bits are swapped around it's center. For example: + * take the bits 0101 is the 4-bit reflection of 1010, and respectfully 0011 would be the + * reflection of 1100. Quake however expects a NON-Reflected CRC on the output, but still + * requires a final XOR on the values (0xFFFF and 0x0000) this is a standard CCITT CRC-16 + * which I respectfully as a programmer don't agree with. + * + * So now you know what we target, and why we target it, despite how unsettling it may seem + * but those are what Quake seems to request. */ -#define util_crc32_update(O,C) \ - (util_crc32_table[((C)^((unsigned char)O)) & 0xFF] ^ ((C)>>8)) - -/* CRC polynomial 0xEDB88320 */ -static const uint32_t util_crc32_table[] = { - 0x00000000, 0x77073096, 0xee0e612c, 0x990951ba, 0x076dc419, 0x706af48f, - 0xe963a535, 0x9e6495a3, 0x0edb8832, 0x79dcb8a4, 0xe0d5e91e, 0x97d2d988, - 0x09b64c2b, 0x7eb17cbd, 0xe7b82d07, 0x90bf1d91, 0x1db71064, 0x6ab020f2, - 0xf3b97148, 0x84be41de, 0x1adad47d, 0x6ddde4eb, 0xf4d4b551, 0x83d385c7, - 0x136c9856, 0x646ba8c0, 0xfd62f97a, 0x8a65c9ec, 0x14015c4f, 0x63066cd9, - 0xfa0f3d63, 0x8d080df5, 0x3b6e20c8, 0x4c69105e, 0xd56041e4, 0xa2677172, - 0x3c03e4d1, 0x4b04d447, 0xd20d85fd, 0xa50ab56b, 0x35b5a8fa, 0x42b2986c, - 0xdbbbc9d6, 0xacbcf940, 0x32d86ce3, 0x45df5c75, 0xdcd60dcf, 0xabd13d59, - 0x26d930ac, 0x51de003a, 0xc8d75180, 0xbfd06116, 0x21b4f4b5, 0x56b3c423, - 0xcfba9599, 0xb8bda50f, 0x2802b89e, 0x5f058808, 0xc60cd9b2, 0xb10be924, - 0x2f6f7c87, 0x58684c11, 0xc1611dab, 0xb6662d3d, 0x76dc4190, 0x01db7106, - 0x98d220bc, 0xefd5102a, 0x71b18589, 0x06b6b51f, 0x9fbfe4a5, 0xe8b8d433, - 0x7807c9a2, 0x0f00f934, 0x9609a88e, 0xe10e9818, 0x7f6a0dbb, 0x086d3d2d, - 0x91646c97, 0xe6635c01, 0x6b6b51f4, 0x1c6c6162, 0x856530d8, 0xf262004e, - 0x6c0695ed, 0x1b01a57b, 0x8208f4c1, 0xf50fc457, 0x65b0d9c6, 0x12b7e950, - 0x8bbeb8ea, 0xfcb9887c, 0x62dd1ddf, 0x15da2d49, 0x8cd37cf3, 0xfbd44c65, - 0x4db26158, 0x3ab551ce, 0xa3bc0074, 0xd4bb30e2, 0x4adfa541, 0x3dd895d7, - 0xa4d1c46d, 0xd3d6f4fb, 0x4369e96a, 0x346ed9fc, 0xad678846, 0xda60b8d0, - 0x44042d73, 0x33031de5, 0xaa0a4c5f, 0xdd0d7cc9, 0x5005713c, 0x270241aa, - 0xbe0b1010, 0xc90c2086, 0x5768b525, 0x206f85b3, 0xb966d409, 0xce61e49f, - 0x5edef90e, 0x29d9c998, 0xb0d09822, 0xc7d7a8b4, 0x59b33d17, 0x2eb40d81, - 0xb7bd5c3b, 0xc0ba6cad, 0xedb88320, 0x9abfb3b6, 0x03b6e20c, 0x74b1d29a, - 0xead54739, 0x9dd277af, 0x04db2615, 0x73dc1683, 0xe3630b12, 0x94643b84, - 0x0d6d6a3e, 0x7a6a5aa8, 0xe40ecf0b, 0x9309ff9d, 0x0a00ae27, 0x7d079eb1, - 0xf00f9344, 0x8708a3d2, 0x1e01f268, 0x6906c2fe, 0xf762575d, 0x806567cb, - 0x196c3671, 0x6e6b06e7, 0xfed41b76, 0x89d32be0, 0x10da7a5a, 0x67dd4acc, - 0xf9b9df6f, 0x8ebeeff9, 0x17b7be43, 0x60b08ed5, 0xd6d6a3e8, 0xa1d1937e, - 0x38d8c2c4, 0x4fdff252, 0xd1bb67f1, 0xa6bc5767, 0x3fb506dd, 0x48b2364b, - 0xd80d2bda, 0xaf0a1b4c, 0x36034af6, 0x41047a60, 0xdf60efc3, 0xa867df55, - 0x316e8eef, 0x4669be79, 0xcb61b38c, 0xbc66831a, 0x256fd2a0, 0x5268e236, - 0xcc0c7795, 0xbb0b4703, 0x220216b9, 0x5505262f, 0xc5ba3bbe, 0xb2bd0b28, - 0x2bb45a92, 0x5cb36a04, 0xc2d7ffa7, 0xb5d0cf31, 0x2cd99e8b, 0x5bdeae1d, - 0x9b64c2b0, 0xec63f226, 0x756aa39c, 0x026d930a, 0x9c0906a9, 0xeb0e363f, - 0x72076785, 0x05005713, 0x95bf4a82, 0xe2b87a14, 0x7bb12bae, 0x0cb61b38, - 0x92d28e9b, 0xe5d5be0d, 0x7cdcefb7, 0x0bdbdf21, 0x86d3d2d4, 0xf1d4e242, - 0x68ddb3f8, 0x1fda836e, 0x81be16cd, 0xf6b9265b, 0x6fb077e1, 0x18b74777, - 0x88085ae6, 0xff0f6a70, 0x66063bca, 0x11010b5c, 0x8f659eff, 0xf862ae69, - 0x616bffd3, 0x166ccf45, 0xa00ae278, 0xd70dd2ee, 0x4e048354, 0x3903b3c2, - 0xa7672661, 0xd06016f7, 0x4969474d, 0x3e6e77db, 0xaed16a4a, 0xd9d65adc, - 0x40df0b66, 0x37d83bf0, 0xa9bcae53, 0xdebb9ec5, 0x47b2cf7f, 0x30b5ffe9, - 0xbdbdf21c, 0xcabac28a, 0x53b39330, 0x24b4a3a6, 0xbad03605, 0xcdd70693, - 0x54de5729, 0x23d967bf, 0xb3667a2e, 0xc4614ab8, 0x5d681b02, 0x2a6f2b94, - 0xb40bbe37, 0xc30c8ea1, 0x5a05df1b, 0x2d02ef8d + +static const uint16_t util_crc16_table[] = { + 0x0000, 0x1021, 0x2042, 0x3063, 0x4084, 0x50A5, + 0x60C6, 0x70E7, 0x8108, 0x9129, 0xA14A, 0xB16B, + 0xC18C, 0xD1AD, 0xE1CE, 0xF1EF, 0x1231, 0x0210, + 0x3273, 0x2252, 0x52B5, 0x4294, 0x72F7, 0x62D6, + 0x9339, 0x8318, 0xB37B, 0xA35A, 0xD3BD, 0xC39C, + 0xF3FF, 0xE3DE, 0x2462, 0x3443, 0x0420, 0x1401, + 0x64E6, 0x74C7, 0x44A4, 0x5485, 0xA56A, 0xB54B, + 0x8528, 0x9509, 0xE5EE, 0xF5CF, 0xC5AC, 0xD58D, + 0x3653, 0x2672, 0x1611, 0x0630, 0x76D7, 0x66F6, + 0x5695, 0x46B4, 0xB75B, 0xA77A, 0x9719, 0x8738, + 0xF7DF, 0xE7FE, 0xD79D, 0xC7BC, 0x48C4, 0x58E5, + 0x6886, 0x78A7, 0x0840, 0x1861, 0x2802, 0x3823, + 0xC9CC, 0xD9ED, 0xE98E, 0xF9AF, 0x8948, 0x9969, + 0xA90A, 0xB92B, 0x5AF5, 0x4AD4, 0x7AB7, 0x6A96, + 0x1A71, 0x0A50, 0x3A33, 0x2A12, 0xDBFD, 0xCBDC, + 0xFBBF, 0xEB9E, 0x9B79, 0x8B58, 0xBB3B, 0xAB1A, + 0x6CA6, 0x7C87, 0x4CE4, 0x5CC5, 0x2C22, 0x3C03, + 0x0C60, 0x1C41, 0xEDAE, 0xFD8F, 0xCDEC, 0xDDCD, + 0xAD2A, 0xBD0B, 0x8D68, 0x9D49, 0x7E97, 0x6EB6, + 0x5ED5, 0x4EF4, 0x3E13, 0x2E32, 0x1E51, 0x0E70, + 0xFF9F, 0xEFBE, 0xDFDD, 0xCFFC, 0xBF1B, 0xAF3A, + 0x9F59, 0x8F78, 0x9188, 0x81A9, 0xB1CA, 0xA1EB, + 0xD10C, 0xC12D, 0xF14E, 0xE16F, 0x1080, 0x00A1, + 0x30C2, 0x20E3, 0x5004, 0x4025, 0x7046, 0x6067, + 0x83B9, 0x9398, 0xA3FB, 0xB3DA, 0xC33D, 0xD31C, + 0xE37F, 0xF35E, 0x02B1, 0x1290, 0x22F3, 0x32D2, + 0x4235, 0x5214, 0x6277, 0x7256, 0xB5EA, 0xA5CB, + 0x95A8, 0x8589, 0xF56E, 0xE54F, 0xD52C, 0xC50D, + 0x34E2, 0x24C3, 0x14A0, 0x0481, 0x7466, 0x6447, + 0x5424, 0x4405, 0xA7DB, 0xB7FA, 0x8799, 0x97B8, + 0xE75F, 0xF77E, 0xC71D, 0xD73C, 0x26D3, 0x36F2, + 0x0691, 0x16B0, 0x6657, 0x7676, 0x4615, 0x5634, + 0xD94C, 0xC96D, 0xF90E, 0xE92F, 0x99C8, 0x89E9, + 0xB98A, 0xA9AB, 0x5844, 0x4865, 0x7806, 0x6827, + 0x18C0, 0x08E1, 0x3882, 0x28A3, 0xCB7D, 0xDB5C, + 0xEB3F, 0xFB1E, 0x8BF9, 0x9BD8, 0xABBB, 0xBB9A, + 0x4A75, 0x5A54, 0x6A37, 0x7A16, 0x0AF1, 0x1AD0, + 0x2AB3, 0x3A92, 0xFD2E, 0xED0F, 0xDD6C, 0xCD4D, + 0xBDAA, 0xAD8B, 0x9DE8, 0x8DC9, 0x7C26, 0x6C07, + 0x5C64, 0x4C45, 0x3CA2, 0x2C83, 0x1CE0, 0x0CC1, + 0xEF1F, 0xFF3E, 0xCF5D, 0xDF7C, 0xAF9B, 0xBFBA, + 0x8FD9, 0x9FF8, 0x6E17, 0x7E36, 0x4E55, 0x5E74, + 0x2E93, 0x3EB2, 0x0ED1, 0x1EF0 }; -uint32_t util_crc32(const char *key, int len, register const short clamp) { - register uint32_t hash = 0xFFFFFFFF; - for (; len; --len, ++key) - hash = util_crc32_update(*key, hash); - return (~hash) % clamp; +/* Non - Reflected */ +uint16_t util_crc16(uint16_t current, const char *k, size_t len) { + register uint16_t h = current; + for (; len; --len, ++k) + h = util_crc16_table[(h>>8)^((unsigned char)*k)]^(h<<8); + return h; +} +/* Reflective Varation (for reference) */ +#if 0 +uint16_t util_crc16(const char *k, int len, const short clamp) { + register uint16_t h= (uint16_t)0xFFFFFFFF; + for (; len; --len, ++k) + h = util_crc16_table[(h^((unsigned char)*k))&0xFF]^(h>>8); + return (~h)%clamp; +} +#endif + +size_t util_strtocmd(const char *in, char *out, size_t outsz) { + size_t sz = 1; + for (; *in && sz < outsz; ++in, ++out, ++sz) + *out = (*in == '-') ? '_' : (util_isalpha(*in) && !util_isupper(*in)) ? *in + 'A' - 'a': *in; + *out = 0; + return sz-1; +} + +size_t util_strtononcmd(const char *in, char *out, size_t outsz) { + size_t sz = 1; + for (; *in && sz < outsz; ++in, ++out, ++sz) + *out = (*in == '_') ? '-' : (util_isalpha(*in) && util_isupper(*in)) ? *in + 'a' - 'A' : *in; + *out = 0; + return sz-1; } /* - * Implements libc getline for systems that don't have it, which is - * assmed all. This works the same as getline(). + * Portable implementation of vasprintf/asprintf. Assumes vsnprintf + * exists, otherwise compiler error. + * + * TODO: fix for MSVC .... */ -int util_getline(char **lineptr, size_t *n, FILE *stream) { - int chr; +int util_vasprintf(char **dat, const char *fmt, va_list args) { int ret; - char *pos; + int len; + char *tmp = NULL; - if (!lineptr || !n || !stream) - return -1; - if (!*lineptr) { - if (!(*lineptr = mem_a((*n = 64)))) + /* + * For visuals tido _vsnprintf doesn't tell you the length of a + * formatted string if it overflows. However there is a MSVC + * intrinsic (which is documented wrong) called _vcsprintf which + * will return the required amount to allocate. + */ + #ifdef _MSC_VER + if ((len = _vscprintf(fmt, args)) < 0) { + *dat = NULL; return -1; - } + } - chr = *n; - pos = *lineptr; - - for (;;) { - int c = getc(stream); - - if (chr < 2) { - char *tmp = mem_a((*n+=(*n>16)?*n:64)); - if (!tmp) - return -1; - - chr = *n + *lineptr - pos; - strcpy(tmp,*lineptr); - if (!(*lineptr = tmp)) { - mem_d (tmp); - return -1; - } - pos = *n - chr + *lineptr; + tmp = (char*)mem_a(len + 1); + if ((ret = _vsnprintf_s(tmp, len+1, len+1, fmt, args)) != len) { + mem_d(tmp); + *dat = NULL; + return -1; } + *dat = tmp; + return len; + #else + /* + * For everything else we have a decent conformint vsnprintf that + * returns the number of bytes needed. We give it a try though on + * a short buffer, since efficently speaking, it could be nice to + * above a second vsnprintf call. + */ + char buf[128]; + va_list cpy; + va_copy(cpy, args); + len = vsnprintf(buf, sizeof(buf), fmt, cpy); + va_end (cpy); - if (ferror(stream)) + if (len < (int)sizeof(buf)) { + *dat = util_strdup(buf); + return len; + } + + /* not large enough ... */ + tmp = (char*)mem_a(len + 1); + if ((ret = vsnprintf(tmp, len + 1, fmt, args)) != len) { + mem_d(tmp); + *dat = NULL; return -1; - if (c == EOF) { - if (pos == *lineptr) - return -1; - else - break; } - *pos++ = c; - chr--; - if (c == '\n') - break; + *dat = tmp; + return len; + #endif +} +int util_asprintf(char **ret, const char *fmt, ...) { + va_list args; + int read; + va_start(args, fmt); + read = util_vasprintf(ret, fmt, args); + va_end (args); + + return read; +} + +/* + * These are various re-implementations (wrapping the real ones) of + * string functions that MSVC consideres unsafe. We wrap these up and + * use the safe varations on MSVC. + */ +#ifdef _MSC_VER + static char **util_strerror_allocated() { + static char **data = NULL; + return data; + } + + static void util_strerror_cleanup(void) { + size_t i; + char **data = util_strerror_allocated(); + for (i = 0; i < vec_size(data); i++) + mem_d(data[i]); + vec_free(data); + } + + const char *util_strerror(int num) { + char *allocated = NULL; + static bool install = false; + static size_t tries = 0; + char **vector = util_strerror_allocated(); + + /* try installing cleanup handler */ + while (!install) { + if (tries == 32) + return "(unknown)"; + + install = !atexit(&util_strerror_cleanup); + tries ++; + } + + allocated = (char*)mem_a(4096); /* A page must be enough */ + strerror_s(allocated, 4096, num); + + vec_push(vector, allocated); + return (const char *)allocated; + } + + int util_snprintf(char *src, size_t bytes, const char *format, ...) { + int rt; + va_list va; + va_start(va, format); + + rt = vsprintf_s(src, bytes, format, va); + va_end (va); + + return rt; + } + + char *util_strcat(char *dest, const char *src) { + strcat_s(dest, strlen(src), src); + return dest; + } + + char *util_strncpy(char *dest, const char *src, size_t num) { + strncpy_s(dest, num, src, num); + return dest; + } +#else + const char *util_strerror(int num) { + return strerror(num); + } + + int util_snprintf(char *src, size_t bytes, const char *format, ...) { + int rt; + va_list va; + va_start(va, format); + rt = vsnprintf(src, bytes, format, va); + va_end (va); + + return rt; + } + + char *util_strcat(char *dest, const char *src) { + return strcat(dest, src); + } + + char *util_strncpy(char *dest, const char *src, size_t num) { + return strncpy(dest, src, num); } - *pos = '\0'; - return (ret = pos - *lineptr); + +#endif /*! _MSC_VER */ + +/* + * Implementation of the Mersenne twister PRNG (pseudo random numer + * generator). Implementation of MT19937. Has a period of 2^19937-1 + * which is a Mersenne Prime (hence the name). + * + * Implemented from specification and original paper: + * http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/ARTICLES/mt.pdf + * + * This code is placed in the public domain by me personally + * (Dale Weiler, a.k.a graphitemaster). + */ + +#define MT_SIZE 624 +#define MT_PERIOD 397 +#define MT_SPACE (MT_SIZE - MT_PERIOD) + +static uint32_t mt_state[MT_SIZE]; +static size_t mt_index = 0; + +static GMQCC_INLINE void mt_generate(void) { + /* + * The loop has been unrolled here: the original paper and implemenation + * Called for the following code: + * for (register unsigned i = 0; i < MT_SIZE; ++i) { + * register uint32_t load; + * load = (0x80000000 & mt_state[i]) // most significant 32nd bit + * load |= (0x7FFFFFFF & mt_state[(i + 1) % MT_SIZE]) // least significant 31nd bit + * + * mt_state[i] = mt_state[(i + MT_PERIOD) % MT_SIZE] ^ (load >> 1); + * + * if (load & 1) mt_state[i] ^= 0x9908B0DF; + * } + * + * This essentially is a waste: we have two modulus operations, and + * a branch that is executed every iteration from [0, MT_SIZE). + * + * Please see: http://www.quadibloc.com/crypto/co4814.htm for more + * information on how this clever trick works. + */ + static const uint32_t matrix[2] = { + 0x00000000, + 0x9908B0Df + }; + /* + * This register gives up a little more speed by instructing the compiler + * to force these into CPU registers (they're counters for indexing mt_state + * which we can force the compiler to generate prefetch instructions for) + */ + register uint32_t y; + register uint32_t i; + + /* + * Said loop has been unrolled for MT_SPACE (226 iterations), opposed + * to [0, MT_SIZE) (634 iterations). + */ + for (i = 0; i < MT_SPACE-1; ++i) { + y = (0x80000000 & mt_state[i]) | (0x7FFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i + MT_PERIOD] ^ (y >> 1) ^ matrix[y & 1]; + + i ++; /* loop unroll */ + + y = (0x80000000 & mt_state[i]) | (0x7FFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i + MT_PERIOD] ^ (y >> 1) ^ matrix[y & 1]; + } + + /* + * collapsing the walls unrolled (evenly dividing 396 [632-227 = 396 + * = 2*2*3*3*11]) + */ + i = MT_SPACE; + while (i < MT_SIZE-2) { + /* + * We expand this 11 times .. manually, no macros are required + * here. This all fits in the CPU cache. + */ + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + y = (0x80000000 & mt_state[i]) | (0x7FFFFFFF & mt_state[i + 1]); + mt_state[i] = mt_state[i - MT_SPACE] ^ (y >> 1) ^ matrix[y & 1]; + ++i; + } + + /* i = mt_state[623] */ + y = (0x80000000 & mt_state[MT_SIZE - 1]) | (0x7FFFFFFF & mt_state[MT_SIZE - 1]); + mt_state[MT_SIZE - 1] = mt_state[MT_PERIOD - 1] ^ (y >> 1) ^ matrix[y & 1]; +} + +void util_seed(uint32_t value) { + /* + * We seed the mt_state with a LCG (linear congruential generator) + * We're operating exactly on exactly m=32, so there is no need to + * use modulus. + * + * The multipler of choice is 0x6C07865, also knows as the Borosh- + * Niederreiter multipler used for modulus 2^32. More can be read + * about this in Knuth's TAOCP Volume 2, page 106. + * + * If you don't own TAOCP something is wrong with you :-) .. so I + * also provided a link to the original paper by Borosh and + * Niederreiter. It's called "Optional Multipliers for PRNG by The + * Linear Congruential Method" (1983). + * http://en.wikipedia.org/wiki/Linear_congruential_generator + * + * From said page, it says the following: + * "A common Mersenne twister implementation, interestingly enough + * used an LCG to generate seed data." + * + * Remarks: + * The data we're operating on is 32-bits for the mt_state array, so + * there is no masking required with 0xFFFFFFFF + */ + register size_t i; + + mt_state[0] = value; + for (i = 1; i < MT_SIZE; ++i) + mt_state[i] = 0x6C078965 * (mt_state[i - 1] ^ mt_state[i - 1] >> 30) + i; +} + +uint32_t util_rand() { + register uint32_t y; + + /* + * This is inlined with any sane compiler (I checked) + * for some reason though, SubC seems to be generating invalid + * code when it inlines this. + */ + if (!mt_index) + mt_generate(); + + y = mt_state[mt_index]; + + /* Standard tempering */ + y ^= y >> 11; /* +7 */ + y ^= y << 7 & 0x9D2C5680; /* +4 */ + y ^= y << 15 & 0xEFC60000; /* -4 */ + y ^= y >> 18; /* -7 */ + + if(++mt_index == MT_SIZE) + mt_index = 0; + + return y; }