1 /* crc32.c -- compute the CRC-32 of a data stream 2 * Copyright (C) 1995-2022 Mark Adler 3 * For conditions of distribution and use, see copyright notice in zlib.h 4 * 5 * This interleaved implementation of a CRC makes use of pipelined multiple 6 * arithmetic-logic units, commonly found in modern CPU cores. It is due to 7 * Kadatch and Jenkins (2010). See doc/crc-doc.1.0.pdf in this distribution. 8 */ 9 10 /* @(#) $Id$ */ 11 12 /* 13 Note on the use of DYNAMIC_CRC_TABLE: there is no mutex or semaphore 14 protection on the static variables used to control the first-use generation 15 of the crc tables. Therefore, if you #define DYNAMIC_CRC_TABLE, you should 16 first call get_crc_table() to initialize the tables before allowing more than 17 one thread to use crc32(). 18 19 MAKECRCH can be #defined to write out crc32.h. A main() routine is also 20 produced, so that this one source file can be compiled to an executable. 21 */ 22 23 #ifdef MAKECRCH 24 # include <stdio.h> 25 # ifndef DYNAMIC_CRC_TABLE 26 # define DYNAMIC_CRC_TABLE 27 # endif /* !DYNAMIC_CRC_TABLE */ 28 #endif /* MAKECRCH */ 29 30 #include "zutil.h" /* for Z_U4, Z_U8, z_crc_t, and FAR definitions */ 31 32 /* 33 A CRC of a message is computed on N braids of words in the message, where 34 each word consists of W bytes (4 or 8). If N is 3, for example, then three 35 running sparse CRCs are calculated respectively on each braid, at these 36 indices in the array of words: 0, 3, 6, ..., 1, 4, 7, ..., and 2, 5, 8, ... 37 This is done starting at a word boundary, and continues until as many blocks 38 of N * W bytes as are available have been processed. The results are combined 39 into a single CRC at the end. For this code, N must be in the range 1..6 and 40 W must be 4 or 8. The upper limit on N can be increased if desired by adding 41 more #if blocks, extending the patterns apparent in the code. In addition, 42 crc32.h would need to be regenerated, if the maximum N value is increased. 43 44 N and W are chosen empirically by benchmarking the execution time on a given 45 processor. The choices for N and W below were based on testing on Intel Kaby 46 Lake i7, AMD Ryzen 7, ARM Cortex-A57, Sparc64-VII, PowerPC POWER9, and MIPS64 47 Octeon II processors. The Intel, AMD, and ARM processors were all fastest 48 with N=5, W=8. The Sparc, PowerPC, and MIPS64 were all fastest at N=5, W=4. 49 They were all tested with either gcc or clang, all using the -O3 optimization 50 level. Your mileage may vary. 51 */ 52 53 /* Define N */ 54 #ifdef Z_TESTN 55 # define N Z_TESTN 56 #else 57 # define N 5 58 #endif 59 #if N < 1 || N > 6 60 # error N must be in 1..6 61 #endif 62 63 /* 64 z_crc_t must be at least 32 bits. z_word_t must be at least as long as 65 z_crc_t. It is assumed here that z_word_t is either 32 bits or 64 bits, and 66 that bytes are eight bits. 67 */ 68 69 /* 70 Define W and the associated z_word_t type. If W is not defined, then a 71 braided calculation is not used, and the associated tables and code are not 72 compiled. 73 */ 74 #ifdef Z_TESTW 75 # if Z_TESTW-1 != -1 76 # define W Z_TESTW 77 # endif 78 #else 79 # ifdef MAKECRCH 80 # define W 8 /* required for MAKECRCH */ 81 # else 82 # if defined(__x86_64__) || defined(__aarch64__) 83 # define W 8 84 # else 85 # define W 4 86 # endif 87 # endif 88 #endif 89 #ifdef W 90 # if W == 8 && defined(Z_U8) 91 typedef Z_U8 z_word_t; 92 # elif defined(Z_U4) 93 # undef W 94 # define W 4 95 typedef Z_U4 z_word_t; 96 # else 97 # undef W 98 # endif 99 #endif 100 101 /* If available, use the ARM processor CRC32 instruction. */ 102 #if defined(__aarch64__) && defined(__ARM_FEATURE_CRC32) && W == 8 103 # define ARMCRC32 104 #endif 105 106 #if defined(W) && (!defined(ARMCRC32) || defined(DYNAMIC_CRC_TABLE)) 107 /* 108 Swap the bytes in a z_word_t to convert between little and big endian. Any 109 self-respecting compiler will optimize this to a single machine byte-swap 110 instruction, if one is available. This assumes that word_t is either 32 bits 111 or 64 bits. 112 */ 113 local z_word_t byte_swap(z_word_t word) { 114 # if W == 8 115 return 116 (word & 0xff00000000000000) >> 56 | 117 (word & 0xff000000000000) >> 40 | 118 (word & 0xff0000000000) >> 24 | 119 (word & 0xff00000000) >> 8 | 120 (word & 0xff000000) << 8 | 121 (word & 0xff0000) << 24 | 122 (word & 0xff00) << 40 | 123 (word & 0xff) << 56; 124 # else /* W == 4 */ 125 return 126 (word & 0xff000000) >> 24 | 127 (word & 0xff0000) >> 8 | 128 (word & 0xff00) << 8 | 129 (word & 0xff) << 24; 130 # endif 131 } 132 #endif 133 134 #ifdef DYNAMIC_CRC_TABLE 135 /* ========================================================================= 136 * Table of powers of x for combining CRC-32s, filled in by make_crc_table() 137 * below. 138 */ 139 local z_crc_t FAR x2n_table[32]; 140 #else 141 /* ========================================================================= 142 * Tables for byte-wise and braided CRC-32 calculations, and a table of powers 143 * of x for combining CRC-32s, all made by make_crc_table(). 144 */ 145 # include "crc32.h" 146 #endif 147 148 /* CRC polynomial. */ 149 #define POLY 0xedb88320 /* p(x) reflected, with x^32 implied */ 150 151 /* 152 Return a(x) multiplied by b(x) modulo p(x), where p(x) is the CRC polynomial, 153 reflected. For speed, this requires that a not be zero. 154 */ 155 local z_crc_t multmodp(z_crc_t a, z_crc_t b) { 156 z_crc_t m, p; 157 158 m = (z_crc_t)1 << 31; 159 p = 0; 160 for (;;) { 161 if (a & m) { 162 p ^= b; 163 if ((a & (m - 1)) == 0) 164 break; 165 } 166 m >>= 1; 167 b = b & 1 ? (b >> 1) ^ POLY : b >> 1; 168 } 169 return p; 170 } 171 172 /* 173 Return x^(n * 2^k) modulo p(x). Requires that x2n_table[] has been 174 initialized. 175 */ 176 local z_crc_t x2nmodp(z_off64_t n, unsigned k) { 177 z_crc_t p; 178 179 p = (z_crc_t)1 << 31; /* x^0 == 1 */ 180 while (n) { 181 if (n & 1) 182 p = multmodp(x2n_table[k & 31], p); 183 n >>= 1; 184 k++; 185 } 186 return p; 187 } 188 189 #ifdef DYNAMIC_CRC_TABLE 190 /* ========================================================================= 191 * Build the tables for byte-wise and braided CRC-32 calculations, and a table 192 * of powers of x for combining CRC-32s. 193 */ 194 local z_crc_t FAR crc_table[256]; 195 #ifdef W 196 local z_word_t FAR crc_big_table[256]; 197 local z_crc_t FAR crc_braid_table[W][256]; 198 local z_word_t FAR crc_braid_big_table[W][256]; 199 local void braid(z_crc_t [][256], z_word_t [][256], int, int); 200 #endif 201 #ifdef MAKECRCH 202 local void write_table(FILE *, const z_crc_t FAR *, int); 203 local void write_table32hi(FILE *, const z_word_t FAR *, int); 204 local void write_table64(FILE *, const z_word_t FAR *, int); 205 #endif /* MAKECRCH */ 206 207 /* 208 Define a once() function depending on the availability of atomics. If this is 209 compiled with DYNAMIC_CRC_TABLE defined, and if CRCs will be computed in 210 multiple threads, and if atomics are not available, then get_crc_table() must 211 be called to initialize the tables and must return before any threads are 212 allowed to compute or combine CRCs. 213 */ 214 215 /* Definition of once functionality. */ 216 typedef struct once_s once_t; 217 218 /* Check for the availability of atomics. */ 219 #if defined(__STDC__) && __STDC_VERSION__ >= 201112L && \ 220 !defined(__STDC_NO_ATOMICS__) 221 222 #include <stdatomic.h> 223 224 /* Structure for once(), which must be initialized with ONCE_INIT. */ 225 struct once_s { 226 atomic_flag begun; 227 atomic_int done; 228 }; 229 #define ONCE_INIT {ATOMIC_FLAG_INIT, 0} 230 231 /* 232 Run the provided init() function exactly once, even if multiple threads 233 invoke once() at the same time. The state must be a once_t initialized with 234 ONCE_INIT. 235 */ 236 local void once(once_t *state, void (*init)(void)) { 237 if (!atomic_load(&state->done)) { 238 if (atomic_flag_test_and_set(&state->begun)) 239 while (!atomic_load(&state->done)) 240 ; 241 else { 242 init(); 243 atomic_store(&state->done, 1); 244 } 245 } 246 } 247 248 #else /* no atomics */ 249 250 /* Structure for once(), which must be initialized with ONCE_INIT. */ 251 struct once_s { 252 volatile int begun; 253 volatile int done; 254 }; 255 #define ONCE_INIT {0, 0} 256 257 /* Test and set. Alas, not atomic, but tries to minimize the period of 258 vulnerability. */ 259 local int test_and_set(int volatile *flag) { 260 int was; 261 262 was = *flag; 263 *flag = 1; 264 return was; 265 } 266 267 /* Run the provided init() function once. This is not thread-safe. */ 268 local void once(once_t *state, void (*init)(void)) { 269 if (!state->done) { 270 if (test_and_set(&state->begun)) 271 while (!state->done) 272 ; 273 else { 274 init(); 275 state->done = 1; 276 } 277 } 278 } 279 280 #endif 281 282 /* State for once(). */ 283 local once_t made = ONCE_INIT; 284 285 /* 286 Generate tables for a byte-wise 32-bit CRC calculation on the polynomial: 287 x^32+x^26+x^23+x^22+x^16+x^12+x^11+x^10+x^8+x^7+x^5+x^4+x^2+x+1. 288 289 Polynomials over GF(2) are represented in binary, one bit per coefficient, 290 with the lowest powers in the most significant bit. Then adding polynomials 291 is just exclusive-or, and multiplying a polynomial by x is a right shift by 292 one. If we call the above polynomial p, and represent a byte as the 293 polynomial q, also with the lowest power in the most significant bit (so the 294 byte 0xb1 is the polynomial x^7+x^3+x^2+1), then the CRC is (q*x^32) mod p, 295 where a mod b means the remainder after dividing a by b. 296 297 This calculation is done using the shift-register method of multiplying and 298 taking the remainder. The register is initialized to zero, and for each 299 incoming bit, x^32 is added mod p to the register if the bit is a one (where 300 x^32 mod p is p+x^32 = x^26+...+1), and the register is multiplied mod p by x 301 (which is shifting right by one and adding x^32 mod p if the bit shifted out 302 is a one). We start with the highest power (least significant bit) of q and 303 repeat for all eight bits of q. 304 305 The table is simply the CRC of all possible eight bit values. This is all the 306 information needed to generate CRCs on data a byte at a time for all 307 combinations of CRC register values and incoming bytes. 308 */ 309 310 local void make_crc_table(void) { 311 unsigned i, j, n; 312 z_crc_t p; 313 314 /* initialize the CRC of bytes tables */ 315 for (i = 0; i < 256; i++) { 316 p = i; 317 for (j = 0; j < 8; j++) 318 p = p & 1 ? (p >> 1) ^ POLY : p >> 1; 319 crc_table[i] = p; 320 #ifdef W 321 crc_big_table[i] = byte_swap(p); 322 #endif 323 } 324 325 /* initialize the x^2^n mod p(x) table */ 326 p = (z_crc_t)1 << 30; /* x^1 */ 327 x2n_table[0] = p; 328 for (n = 1; n < 32; n++) 329 x2n_table[n] = p = multmodp(p, p); 330 331 #ifdef W 332 /* initialize the braiding tables -- needs x2n_table[] */ 333 braid(crc_braid_table, crc_braid_big_table, N, W); 334 #endif 335 336 #ifdef MAKECRCH 337 { 338 /* 339 The crc32.h header file contains tables for both 32-bit and 64-bit 340 z_word_t's, and so requires a 64-bit type be available. In that case, 341 z_word_t must be defined to be 64-bits. This code then also generates 342 and writes out the tables for the case that z_word_t is 32 bits. 343 */ 344 #if !defined(W) || W != 8 345 # error Need a 64-bit integer type in order to generate crc32.h. 346 #endif 347 FILE *out; 348 int k, n; 349 z_crc_t ltl[8][256]; 350 z_word_t big[8][256]; 351 352 out = fopen("crc32.h", "w"); 353 if (out == NULL) return; 354 355 /* write out little-endian CRC table to crc32.h */ 356 fprintf(out, 357 "/* crc32.h -- tables for rapid CRC calculation\n" 358 " * Generated automatically by crc32.c\n */\n" 359 "\n" 360 "local const z_crc_t FAR crc_table[] = {\n" 361 " "); 362 write_table(out, crc_table, 256); 363 fprintf(out, 364 "};\n"); 365 366 /* write out big-endian CRC table for 64-bit z_word_t to crc32.h */ 367 fprintf(out, 368 "\n" 369 "#ifdef W\n" 370 "\n" 371 "#if W == 8\n" 372 "\n" 373 "local const z_word_t FAR crc_big_table[] = {\n" 374 " "); 375 write_table64(out, crc_big_table, 256); 376 fprintf(out, 377 "};\n"); 378 379 /* write out big-endian CRC table for 32-bit z_word_t to crc32.h */ 380 fprintf(out, 381 "\n" 382 "#else /* W == 4 */\n" 383 "\n" 384 "local const z_word_t FAR crc_big_table[] = {\n" 385 " "); 386 write_table32hi(out, crc_big_table, 256); 387 fprintf(out, 388 "};\n" 389 "\n" 390 "#endif\n"); 391 392 /* write out braid tables for each value of N */ 393 for (n = 1; n <= 6; n++) { 394 fprintf(out, 395 "\n" 396 "#if N == %d\n", n); 397 398 /* compute braid tables for this N and 64-bit word_t */ 399 braid(ltl, big, n, 8); 400 401 /* write out braid tables for 64-bit z_word_t to crc32.h */ 402 fprintf(out, 403 "\n" 404 "#if W == 8\n" 405 "\n" 406 "local const z_crc_t FAR crc_braid_table[][256] = {\n"); 407 for (k = 0; k < 8; k++) { 408 fprintf(out, " {"); 409 write_table(out, ltl[k], 256); 410 fprintf(out, "}%s", k < 7 ? ",\n" : ""); 411 } 412 fprintf(out, 413 "};\n" 414 "\n" 415 "local const z_word_t FAR crc_braid_big_table[][256] = {\n"); 416 for (k = 0; k < 8; k++) { 417 fprintf(out, " {"); 418 write_table64(out, big[k], 256); 419 fprintf(out, "}%s", k < 7 ? ",\n" : ""); 420 } 421 fprintf(out, 422 "};\n"); 423 424 /* compute braid tables for this N and 32-bit word_t */ 425 braid(ltl, big, n, 4); 426 427 /* write out braid tables for 32-bit z_word_t to crc32.h */ 428 fprintf(out, 429 "\n" 430 "#else /* W == 4 */\n" 431 "\n" 432 "local const z_crc_t FAR crc_braid_table[][256] = {\n"); 433 for (k = 0; k < 4; k++) { 434 fprintf(out, " {"); 435 write_table(out, ltl[k], 256); 436 fprintf(out, "}%s", k < 3 ? ",\n" : ""); 437 } 438 fprintf(out, 439 "};\n" 440 "\n" 441 "local const z_word_t FAR crc_braid_big_table[][256] = {\n"); 442 for (k = 0; k < 4; k++) { 443 fprintf(out, " {"); 444 write_table32hi(out, big[k], 256); 445 fprintf(out, "}%s", k < 3 ? ",\n" : ""); 446 } 447 fprintf(out, 448 "};\n" 449 "\n" 450 "#endif\n" 451 "\n" 452 "#endif\n"); 453 } 454 fprintf(out, 455 "\n" 456 "#endif\n"); 457 458 /* write out zeros operator table to crc32.h */ 459 fprintf(out, 460 "\n" 461 "local const z_crc_t FAR x2n_table[] = {\n" 462 " "); 463 write_table(out, x2n_table, 32); 464 fprintf(out, 465 "};\n"); 466 fclose(out); 467 } 468 #endif /* MAKECRCH */ 469 } 470 471 #ifdef MAKECRCH 472 473 /* 474 Write the 32-bit values in table[0..k-1] to out, five per line in 475 hexadecimal separated by commas. 476 */ 477 local void write_table(FILE *out, const z_crc_t FAR *table, int k) { 478 int n; 479 480 for (n = 0; n < k; n++) 481 fprintf(out, "%s0x%08lx%s", n == 0 || n % 5 ? "" : " ", 482 (unsigned long)(table[n]), 483 n == k - 1 ? "" : (n % 5 == 4 ? ",\n" : ", ")); 484 } 485 486 /* 487 Write the high 32-bits of each value in table[0..k-1] to out, five per line 488 in hexadecimal separated by commas. 489 */ 490 local void write_table32hi(FILE *out, const z_word_t FAR *table, int k) { 491 int n; 492 493 for (n = 0; n < k; n++) 494 fprintf(out, "%s0x%08lx%s", n == 0 || n % 5 ? "" : " ", 495 (unsigned long)(table[n] >> 32), 496 n == k - 1 ? "" : (n % 5 == 4 ? ",\n" : ", ")); 497 } 498 499 /* 500 Write the 64-bit values in table[0..k-1] to out, three per line in 501 hexadecimal separated by commas. This assumes that if there is a 64-bit 502 type, then there is also a long long integer type, and it is at least 64 503 bits. If not, then the type cast and format string can be adjusted 504 accordingly. 505 */ 506 local void write_table64(FILE *out, const z_word_t FAR *table, int k) { 507 int n; 508 509 for (n = 0; n < k; n++) 510 fprintf(out, "%s0x%016llx%s", n == 0 || n % 3 ? "" : " ", 511 (unsigned long long)(table[n]), 512 n == k - 1 ? "" : (n % 3 == 2 ? ",\n" : ", ")); 513 } 514 515 /* Actually do the deed. */ 516 int main(void) { 517 make_crc_table(); 518 return 0; 519 } 520 521 #endif /* MAKECRCH */ 522 523 #ifdef W 524 /* 525 Generate the little and big-endian braid tables for the given n and z_word_t 526 size w. Each array must have room for w blocks of 256 elements. 527 */ 528 local void braid(z_crc_t ltl[][256], z_word_t big[][256], int n, int w) { 529 int k; 530 z_crc_t i, p, q; 531 for (k = 0; k < w; k++) { 532 p = x2nmodp((n * w + 3 - k) << 3, 0); 533 ltl[k][0] = 0; 534 big[w - 1 - k][0] = 0; 535 for (i = 1; i < 256; i++) { 536 ltl[k][i] = q = multmodp(i << 24, p); 537 big[w - 1 - k][i] = byte_swap(q); 538 } 539 } 540 } 541 #endif 542 543 #endif /* DYNAMIC_CRC_TABLE */ 544 545 /* ========================================================================= 546 * This function can be used by asm versions of crc32(), and to force the 547 * generation of the CRC tables in a threaded application. 548 */ 549 const z_crc_t FAR * ZEXPORT get_crc_table(void) { 550 #ifdef DYNAMIC_CRC_TABLE 551 once(&made, make_crc_table); 552 #endif /* DYNAMIC_CRC_TABLE */ 553 return (const z_crc_t FAR *)crc_table; 554 } 555 556 /* ========================================================================= 557 * Use ARM machine instructions if available. This will compute the CRC about 558 * ten times faster than the braided calculation. This code does not check for 559 * the presence of the CRC instruction at run time. __ARM_FEATURE_CRC32 will 560 * only be defined if the compilation specifies an ARM processor architecture 561 * that has the instructions. For example, compiling with -march=armv8.1-a or 562 * -march=armv8-a+crc, or -march=native if the compile machine has the crc32 563 * instructions. 564 */ 565 #ifdef ARMCRC32 566 567 /* 568 Constants empirically determined to maximize speed. These values are from 569 measurements on a Cortex-A57. Your mileage may vary. 570 */ 571 #define Z_BATCH 3990 /* number of words in a batch */ 572 #define Z_BATCH_ZEROS 0xa10d3d0c /* computed from Z_BATCH = 3990 */ 573 #define Z_BATCH_MIN 800 /* fewest words in a final batch */ 574 575 unsigned long ZEXPORT crc32_z(unsigned long crc, const unsigned char FAR *buf, 576 z_size_t len) { 577 z_crc_t val; 578 z_word_t crc1, crc2; 579 const z_word_t *word; 580 z_word_t val0, val1, val2; 581 z_size_t last, last2, i; 582 z_size_t num; 583 584 /* Return initial CRC, if requested. */ 585 if (buf == Z_NULL) return 0; 586 587 #ifdef DYNAMIC_CRC_TABLE 588 once(&made, make_crc_table); 589 #endif /* DYNAMIC_CRC_TABLE */ 590 591 /* Pre-condition the CRC */ 592 crc = (~crc) & 0xffffffff; 593 594 /* Compute the CRC up to a word boundary. */ 595 while (len && ((z_size_t)buf & 7) != 0) { 596 len--; 597 val = *buf++; 598 __asm__ volatile("crc32b %w0, %w0, %w1" : "+r"(crc) : "r"(val)); 599 } 600 601 /* Prepare to compute the CRC on full 64-bit words word[0..num-1]. */ 602 word = (z_word_t const *)buf; 603 num = len >> 3; 604 len &= 7; 605 606 /* Do three interleaved CRCs to realize the throughput of one crc32x 607 instruction per cycle. Each CRC is calculated on Z_BATCH words. The 608 three CRCs are combined into a single CRC after each set of batches. */ 609 while (num >= 3 * Z_BATCH) { 610 crc1 = 0; 611 crc2 = 0; 612 for (i = 0; i < Z_BATCH; i++) { 613 val0 = word[i]; 614 val1 = word[i + Z_BATCH]; 615 val2 = word[i + 2 * Z_BATCH]; 616 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc) : "r"(val0)); 617 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc1) : "r"(val1)); 618 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc2) : "r"(val2)); 619 } 620 word += 3 * Z_BATCH; 621 num -= 3 * Z_BATCH; 622 crc = multmodp(Z_BATCH_ZEROS, crc) ^ crc1; 623 crc = multmodp(Z_BATCH_ZEROS, crc) ^ crc2; 624 } 625 626 /* Do one last smaller batch with the remaining words, if there are enough 627 to pay for the combination of CRCs. */ 628 last = num / 3; 629 if (last >= Z_BATCH_MIN) { 630 last2 = last << 1; 631 crc1 = 0; 632 crc2 = 0; 633 for (i = 0; i < last; i++) { 634 val0 = word[i]; 635 val1 = word[i + last]; 636 val2 = word[i + last2]; 637 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc) : "r"(val0)); 638 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc1) : "r"(val1)); 639 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc2) : "r"(val2)); 640 } 641 word += 3 * last; 642 num -= 3 * last; 643 val = x2nmodp(last, 6); 644 crc = multmodp(val, crc) ^ crc1; 645 crc = multmodp(val, crc) ^ crc2; 646 } 647 648 /* Compute the CRC on any remaining words. */ 649 for (i = 0; i < num; i++) { 650 val0 = word[i]; 651 __asm__ volatile("crc32x %w0, %w0, %x1" : "+r"(crc) : "r"(val0)); 652 } 653 word += num; 654 655 /* Complete the CRC on any remaining bytes. */ 656 buf = (const unsigned char FAR *)word; 657 while (len) { 658 len--; 659 val = *buf++; 660 __asm__ volatile("crc32b %w0, %w0, %w1" : "+r"(crc) : "r"(val)); 661 } 662 663 /* Return the CRC, post-conditioned. */ 664 return crc ^ 0xffffffff; 665 } 666 667 #else 668 669 #ifdef W 670 671 /* 672 Return the CRC of the W bytes in the word_t data, taking the 673 least-significant byte of the word as the first byte of data, without any pre 674 or post conditioning. This is used to combine the CRCs of each braid. 675 */ 676 local z_crc_t crc_word(z_word_t data) { 677 int k; 678 for (k = 0; k < W; k++) 679 data = (data >> 8) ^ crc_table[data & 0xff]; 680 return (z_crc_t)data; 681 } 682 683 local z_word_t crc_word_big(z_word_t data) { 684 int k; 685 for (k = 0; k < W; k++) 686 data = (data << 8) ^ 687 crc_big_table[(data >> ((W - 1) << 3)) & 0xff]; 688 return data; 689 } 690 691 #endif 692 693 /* ========================================================================= */ 694 unsigned long ZEXPORT crc32_z(unsigned long crc, const unsigned char FAR *buf, 695 z_size_t len) { 696 /* Return initial CRC, if requested. */ 697 if (buf == Z_NULL) return 0; 698 699 #ifdef DYNAMIC_CRC_TABLE 700 once(&made, make_crc_table); 701 #endif /* DYNAMIC_CRC_TABLE */ 702 703 /* Pre-condition the CRC */ 704 crc = (~crc) & 0xffffffff; 705 706 #ifdef W 707 708 /* If provided enough bytes, do a braided CRC calculation. */ 709 if (len >= N * W + W - 1) { 710 z_size_t blks; 711 z_word_t const *words; 712 unsigned endian; 713 int k; 714 715 /* Compute the CRC up to a z_word_t boundary. */ 716 while (len && ((z_size_t)buf & (W - 1)) != 0) { 717 len--; 718 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 719 } 720 721 /* Compute the CRC on as many N z_word_t blocks as are available. */ 722 blks = len / (N * W); 723 len -= blks * N * W; 724 words = (z_word_t const *)buf; 725 726 /* Do endian check at execution time instead of compile time, since ARM 727 processors can change the endianness at execution time. If the 728 compiler knows what the endianness will be, it can optimize out the 729 check and the unused branch. */ 730 endian = 1; 731 if (*(unsigned char *)&endian) { 732 /* Little endian. */ 733 734 z_crc_t crc0; 735 z_word_t word0; 736 #if N > 1 737 z_crc_t crc1; 738 z_word_t word1; 739 #if N > 2 740 z_crc_t crc2; 741 z_word_t word2; 742 #if N > 3 743 z_crc_t crc3; 744 z_word_t word3; 745 #if N > 4 746 z_crc_t crc4; 747 z_word_t word4; 748 #if N > 5 749 z_crc_t crc5; 750 z_word_t word5; 751 #endif 752 #endif 753 #endif 754 #endif 755 #endif 756 757 /* Initialize the CRC for each braid. */ 758 crc0 = crc; 759 #if N > 1 760 crc1 = 0; 761 #if N > 2 762 crc2 = 0; 763 #if N > 3 764 crc3 = 0; 765 #if N > 4 766 crc4 = 0; 767 #if N > 5 768 crc5 = 0; 769 #endif 770 #endif 771 #endif 772 #endif 773 #endif 774 775 /* 776 Process the first blks-1 blocks, computing the CRCs on each braid 777 independently. 778 */ 779 while (--blks) { 780 /* Load the word for each braid into registers. */ 781 word0 = crc0 ^ words[0]; 782 #if N > 1 783 word1 = crc1 ^ words[1]; 784 #if N > 2 785 word2 = crc2 ^ words[2]; 786 #if N > 3 787 word3 = crc3 ^ words[3]; 788 #if N > 4 789 word4 = crc4 ^ words[4]; 790 #if N > 5 791 word5 = crc5 ^ words[5]; 792 #endif 793 #endif 794 #endif 795 #endif 796 #endif 797 words += N; 798 799 /* Compute and update the CRC for each word. The loop should 800 get unrolled. */ 801 crc0 = crc_braid_table[0][word0 & 0xff]; 802 #if N > 1 803 crc1 = crc_braid_table[0][word1 & 0xff]; 804 #if N > 2 805 crc2 = crc_braid_table[0][word2 & 0xff]; 806 #if N > 3 807 crc3 = crc_braid_table[0][word3 & 0xff]; 808 #if N > 4 809 crc4 = crc_braid_table[0][word4 & 0xff]; 810 #if N > 5 811 crc5 = crc_braid_table[0][word5 & 0xff]; 812 #endif 813 #endif 814 #endif 815 #endif 816 #endif 817 for (k = 1; k < W; k++) { 818 crc0 ^= crc_braid_table[k][(word0 >> (k << 3)) & 0xff]; 819 #if N > 1 820 crc1 ^= crc_braid_table[k][(word1 >> (k << 3)) & 0xff]; 821 #if N > 2 822 crc2 ^= crc_braid_table[k][(word2 >> (k << 3)) & 0xff]; 823 #if N > 3 824 crc3 ^= crc_braid_table[k][(word3 >> (k << 3)) & 0xff]; 825 #if N > 4 826 crc4 ^= crc_braid_table[k][(word4 >> (k << 3)) & 0xff]; 827 #if N > 5 828 crc5 ^= crc_braid_table[k][(word5 >> (k << 3)) & 0xff]; 829 #endif 830 #endif 831 #endif 832 #endif 833 #endif 834 } 835 } 836 837 /* 838 Process the last block, combining the CRCs of the N braids at the 839 same time. 840 */ 841 crc = crc_word(crc0 ^ words[0]); 842 #if N > 1 843 crc = crc_word(crc1 ^ words[1] ^ crc); 844 #if N > 2 845 crc = crc_word(crc2 ^ words[2] ^ crc); 846 #if N > 3 847 crc = crc_word(crc3 ^ words[3] ^ crc); 848 #if N > 4 849 crc = crc_word(crc4 ^ words[4] ^ crc); 850 #if N > 5 851 crc = crc_word(crc5 ^ words[5] ^ crc); 852 #endif 853 #endif 854 #endif 855 #endif 856 #endif 857 words += N; 858 } 859 else { 860 /* Big endian. */ 861 862 z_word_t crc0, word0, comb; 863 #if N > 1 864 z_word_t crc1, word1; 865 #if N > 2 866 z_word_t crc2, word2; 867 #if N > 3 868 z_word_t crc3, word3; 869 #if N > 4 870 z_word_t crc4, word4; 871 #if N > 5 872 z_word_t crc5, word5; 873 #endif 874 #endif 875 #endif 876 #endif 877 #endif 878 879 /* Initialize the CRC for each braid. */ 880 crc0 = byte_swap(crc); 881 #if N > 1 882 crc1 = 0; 883 #if N > 2 884 crc2 = 0; 885 #if N > 3 886 crc3 = 0; 887 #if N > 4 888 crc4 = 0; 889 #if N > 5 890 crc5 = 0; 891 #endif 892 #endif 893 #endif 894 #endif 895 #endif 896 897 /* 898 Process the first blks-1 blocks, computing the CRCs on each braid 899 independently. 900 */ 901 while (--blks) { 902 /* Load the word for each braid into registers. */ 903 word0 = crc0 ^ words[0]; 904 #if N > 1 905 word1 = crc1 ^ words[1]; 906 #if N > 2 907 word2 = crc2 ^ words[2]; 908 #if N > 3 909 word3 = crc3 ^ words[3]; 910 #if N > 4 911 word4 = crc4 ^ words[4]; 912 #if N > 5 913 word5 = crc5 ^ words[5]; 914 #endif 915 #endif 916 #endif 917 #endif 918 #endif 919 words += N; 920 921 /* Compute and update the CRC for each word. The loop should 922 get unrolled. */ 923 crc0 = crc_braid_big_table[0][word0 & 0xff]; 924 #if N > 1 925 crc1 = crc_braid_big_table[0][word1 & 0xff]; 926 #if N > 2 927 crc2 = crc_braid_big_table[0][word2 & 0xff]; 928 #if N > 3 929 crc3 = crc_braid_big_table[0][word3 & 0xff]; 930 #if N > 4 931 crc4 = crc_braid_big_table[0][word4 & 0xff]; 932 #if N > 5 933 crc5 = crc_braid_big_table[0][word5 & 0xff]; 934 #endif 935 #endif 936 #endif 937 #endif 938 #endif 939 for (k = 1; k < W; k++) { 940 crc0 ^= crc_braid_big_table[k][(word0 >> (k << 3)) & 0xff]; 941 #if N > 1 942 crc1 ^= crc_braid_big_table[k][(word1 >> (k << 3)) & 0xff]; 943 #if N > 2 944 crc2 ^= crc_braid_big_table[k][(word2 >> (k << 3)) & 0xff]; 945 #if N > 3 946 crc3 ^= crc_braid_big_table[k][(word3 >> (k << 3)) & 0xff]; 947 #if N > 4 948 crc4 ^= crc_braid_big_table[k][(word4 >> (k << 3)) & 0xff]; 949 #if N > 5 950 crc5 ^= crc_braid_big_table[k][(word5 >> (k << 3)) & 0xff]; 951 #endif 952 #endif 953 #endif 954 #endif 955 #endif 956 } 957 } 958 959 /* 960 Process the last block, combining the CRCs of the N braids at the 961 same time. 962 */ 963 comb = crc_word_big(crc0 ^ words[0]); 964 #if N > 1 965 comb = crc_word_big(crc1 ^ words[1] ^ comb); 966 #if N > 2 967 comb = crc_word_big(crc2 ^ words[2] ^ comb); 968 #if N > 3 969 comb = crc_word_big(crc3 ^ words[3] ^ comb); 970 #if N > 4 971 comb = crc_word_big(crc4 ^ words[4] ^ comb); 972 #if N > 5 973 comb = crc_word_big(crc5 ^ words[5] ^ comb); 974 #endif 975 #endif 976 #endif 977 #endif 978 #endif 979 words += N; 980 crc = byte_swap(comb); 981 } 982 983 /* 984 Update the pointer to the remaining bytes to process. 985 */ 986 buf = (unsigned char const *)words; 987 } 988 989 #endif /* W */ 990 991 /* Complete the computation of the CRC on any remaining bytes. */ 992 while (len >= 8) { 993 len -= 8; 994 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 995 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 996 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 997 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 998 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 999 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 1000 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 1001 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 1002 } 1003 while (len) { 1004 len--; 1005 crc = (crc >> 8) ^ crc_table[(crc ^ *buf++) & 0xff]; 1006 } 1007 1008 /* Return the CRC, post-conditioned. */ 1009 return crc ^ 0xffffffff; 1010 } 1011 1012 #endif 1013 1014 /* ========================================================================= */ 1015 unsigned long ZEXPORT crc32(unsigned long crc, const unsigned char FAR *buf, 1016 uInt len) { 1017 return crc32_z(crc, buf, len); 1018 } 1019 1020 /* ========================================================================= */ 1021 uLong ZEXPORT crc32_combine64(uLong crc1, uLong crc2, z_off64_t len2) { 1022 #ifdef DYNAMIC_CRC_TABLE 1023 once(&made, make_crc_table); 1024 #endif /* DYNAMIC_CRC_TABLE */ 1025 return multmodp(x2nmodp(len2, 3), crc1) ^ (crc2 & 0xffffffff); 1026 } 1027 1028 /* ========================================================================= */ 1029 uLong ZEXPORT crc32_combine(uLong crc1, uLong crc2, z_off_t len2) { 1030 return crc32_combine64(crc1, crc2, (z_off64_t)len2); 1031 } 1032 1033 /* ========================================================================= */ 1034 uLong ZEXPORT crc32_combine_gen64(z_off64_t len2) { 1035 #ifdef DYNAMIC_CRC_TABLE 1036 once(&made, make_crc_table); 1037 #endif /* DYNAMIC_CRC_TABLE */ 1038 return x2nmodp(len2, 3); 1039 } 1040 1041 /* ========================================================================= */ 1042 uLong ZEXPORT crc32_combine_gen(z_off_t len2) { 1043 return crc32_combine_gen64((z_off64_t)len2); 1044 } 1045 1046 /* ========================================================================= */ 1047 uLong ZEXPORT crc32_combine_op(uLong crc1, uLong crc2, uLong op) { 1048 return multmodp(op, crc1) ^ (crc2 & 0xffffffff); 1049 } 1050