1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591
|
// SPDX-License-Identifier: 0BSD
#include <algorithm>
#include <atomic>
#include <clocale>
#include <cstdint>
#include <cstdio>
#include <cstring>
#include <endian.h>
#include <err.h>
#include <pthread.h>
#include <snappy-sinksource.h>
#include <snappy.h>
#include <string_view>
#include <sys/mman.h>
#include <unistd.h>
#if __has_include(<libintl.h>)
#include <libintl.h>
#else
#define gettext(s) (s)
#define ngettext(s, p, n) (n == 1 ? s : p)
#endif
using namespace std::literals;
#if __i386__ || __x86_64__ // using function multiversioning (or manual cpuid check)
#include <nmmintrin.h>
#if(__linux__ && !__GLIBC__) || __OpenBSD__ // cpuid
#include <cpuid.h>
#define CRC32C_GENERIC_ATTR
#define CRC32C_GENERIC_NAME crc32c_generic
#define PARALLEL_COMPRESSION_CRC32C parallel_compression_crc32c_generic
#else // multiversioning
#define MULTIVER 1
#define CRC32C_GENERIC_ATTR [[gnu::target("default")]]
#define CRC32C_GENERIC_NAME crc32c
#define PARALLEL_COMPRESSION_CRC32C parallel_compression_crc32c
#endif
#elif __aarch64__ // accelerated by default (CRC nominally optional on v8, required on v8.1)
#include <arm_acle.h>
#define CRC32C_GENERIC_ATTR [[maybe_unused]]
#define CRC32C_GENERIC_NAME crc32c_generic
#define PARALLEL_COMPRESSION_CRC32C parallel_compression_crc32c_generic
#elif __loongarch64 // manual cpucfg check in static init
#include <larchintrin.h>
#define CRC32C_GENERIC_ATTR
#define CRC32C_GENERIC_NAME crc32c_generic
#define PARALLEL_COMPRESSION_CRC32C parallel_compression_crc32c // TODO: measure on hardware! defaults to true, which wins in QEMU
#else
#define CRC32C_GENERIC_ATTR
#define CRC32C_GENERIC_NAME crc32c
#define PARALLEL_COMPRESSION_CRC32C parallel_compression_crc32c
#endif
namespace {
CRC32C_GENERIC_ATTR std::uint32_t CRC32C_GENERIC_NAME(std::uint32_t cur, const void * data, std::size_t data_len) {
static const constexpr std::uint32_t crc_lut[] = {
0x00000000, 0xF26B8303, 0xE13B70F7, 0x1350F3F4, 0xC79A971F, 0x35F1141C, 0x26A1E7E8, 0xD4CA64EB, 0x8AD958CF, 0x78B2DBCC, 0x6BE22838, 0x9989AB3B,
0x4D43CFD0, 0xBF284CD3, 0xAC78BF27, 0x5E133C24, 0x105EC76F, 0xE235446C, 0xF165B798, 0x030E349B, 0xD7C45070, 0x25AFD373, 0x36FF2087, 0xC494A384,
0x9A879FA0, 0x68EC1CA3, 0x7BBCEF57, 0x89D76C54, 0x5D1D08BF, 0xAF768BBC, 0xBC267848, 0x4E4DFB4B, 0x20BD8EDE, 0xD2D60DDD, 0xC186FE29, 0x33ED7D2A,
0xE72719C1, 0x154C9AC2, 0x061C6936, 0xF477EA35, 0xAA64D611, 0x580F5512, 0x4B5FA6E6, 0xB93425E5, 0x6DFE410E, 0x9F95C20D, 0x8CC531F9, 0x7EAEB2FA,
0x30E349B1, 0xC288CAB2, 0xD1D83946, 0x23B3BA45, 0xF779DEAE, 0x05125DAD, 0x1642AE59, 0xE4292D5A, 0xBA3A117E, 0x4851927D, 0x5B016189, 0xA96AE28A,
0x7DA08661, 0x8FCB0562, 0x9C9BF696, 0x6EF07595, 0x417B1DBC, 0xB3109EBF, 0xA0406D4B, 0x522BEE48, 0x86E18AA3, 0x748A09A0, 0x67DAFA54, 0x95B17957,
0xCBA24573, 0x39C9C670, 0x2A993584, 0xD8F2B687, 0x0C38D26C, 0xFE53516F, 0xED03A29B, 0x1F682198, 0x5125DAD3, 0xA34E59D0, 0xB01EAA24, 0x42752927,
0x96BF4DCC, 0x64D4CECF, 0x77843D3B, 0x85EFBE38, 0xDBFC821C, 0x2997011F, 0x3AC7F2EB, 0xC8AC71E8, 0x1C661503, 0xEE0D9600, 0xFD5D65F4, 0x0F36E6F7,
0x61C69362, 0x93AD1061, 0x80FDE395, 0x72966096, 0xA65C047D, 0x5437877E, 0x4767748A, 0xB50CF789, 0xEB1FCBAD, 0x197448AE, 0x0A24BB5A, 0xF84F3859,
0x2C855CB2, 0xDEEEDFB1, 0xCDBE2C45, 0x3FD5AF46, 0x7198540D, 0x83F3D70E, 0x90A324FA, 0x62C8A7F9, 0xB602C312, 0x44694011, 0x5739B3E5, 0xA55230E6,
0xFB410CC2, 0x092A8FC1, 0x1A7A7C35, 0xE811FF36, 0x3CDB9BDD, 0xCEB018DE, 0xDDE0EB2A, 0x2F8B6829, 0x82F63B78, 0x709DB87B, 0x63CD4B8F, 0x91A6C88C,
0x456CAC67, 0xB7072F64, 0xA457DC90, 0x563C5F93, 0x082F63B7, 0xFA44E0B4, 0xE9141340, 0x1B7F9043, 0xCFB5F4A8, 0x3DDE77AB, 0x2E8E845F, 0xDCE5075C,
0x92A8FC17, 0x60C37F14, 0x73938CE0, 0x81F80FE3, 0x55326B08, 0xA759E80B, 0xB4091BFF, 0x466298FC, 0x1871A4D8, 0xEA1A27DB, 0xF94AD42F, 0x0B21572C,
0xDFEB33C7, 0x2D80B0C4, 0x3ED04330, 0xCCBBC033, 0xA24BB5A6, 0x502036A5, 0x4370C551, 0xB11B4652, 0x65D122B9, 0x97BAA1BA, 0x84EA524E, 0x7681D14D,
0x2892ED69, 0xDAF96E6A, 0xC9A99D9E, 0x3BC21E9D, 0xEF087A76, 0x1D63F975, 0x0E330A81, 0xFC588982, 0xB21572C9, 0x407EF1CA, 0x532E023E, 0xA145813D,
0x758FE5D6, 0x87E466D5, 0x94B49521, 0x66DF1622, 0x38CC2A06, 0xCAA7A905, 0xD9F75AF1, 0x2B9CD9F2, 0xFF56BD19, 0x0D3D3E1A, 0x1E6DCDEE, 0xEC064EED,
0xC38D26C4, 0x31E6A5C7, 0x22B65633, 0xD0DDD530, 0x0417B1DB, 0xF67C32D8, 0xE52CC12C, 0x1747422F, 0x49547E0B, 0xBB3FFD08, 0xA86F0EFC, 0x5A048DFF,
0x8ECEE914, 0x7CA56A17, 0x6FF599E3, 0x9D9E1AE0, 0xD3D3E1AB, 0x21B862A8, 0x32E8915C, 0xC083125F, 0x144976B4, 0xE622F5B7, 0xF5720643, 0x07198540,
0x590AB964, 0xAB613A67, 0xB831C993, 0x4A5A4A90, 0x9E902E7B, 0x6CFBAD78, 0x7FAB5E8C, 0x8DC0DD8F, 0xE330A81A, 0x115B2B19, 0x020BD8ED, 0xF0605BEE,
0x24AA3F05, 0xD6C1BC06, 0xC5914FF2, 0x37FACCF1, 0x69E9F0D5, 0x9B8273D6, 0x88D28022, 0x7AB90321, 0xAE7367CA, 0x5C18E4C9, 0x4F48173D, 0xBD23943E,
0xF36E6F75, 0x0105EC76, 0x12551F82, 0xE03E9C81, 0x34F4F86A, 0xC69F7B69, 0xD5CF889D, 0x27A40B9E, 0x79B737BA, 0x8BDCB4B9, 0x988C474D, 0x6AE7C44E,
0xBE2DA0A5, 0x4C4623A6, 0x5F16D052, 0xAD7D5351};
auto bytes = reinterpret_cast<const std::uint8_t *>(data), end = bytes + data_len;
for(; bytes != end; ++bytes)
cur = (cur >> 8) ^ crc_lut[(cur & 0xFF) ^ *bytes];
return cur;
}
CRC32C_GENERIC_ATTR bool PARALLEL_COMPRESSION_CRC32C() {
return true;
}
#if __i386__ || __x86_64__
#if MULTIVER
#define CRC32Q(sym) sym
#else
#define CRC32Q(sym) sym##_sse42
#endif
[[gnu::target("sse4.2")]] std::uint32_t CRC32Q(crc32c)(std::uint32_t cur_r, const void * data, std::size_t data_len) {
#if __x86_64__ // this isn't a simple std::size_t+#if _LP64 because x32 is amd64 ILP32
using sse42_t = std::uint64_t;
#define _mm_crc32_big _mm_crc32_u64
#else
using sse42_t = std::uint32_t;
#define _mm_crc32_big _mm_crc32_u32
#endif
sse42_t cur = cur_r;
auto bulk_iter = reinterpret_cast<const sse42_t *>(data);
for(auto bulk = data_len / sizeof(sse42_t); bulk; --bulk, ++bulk_iter)
cur = _mm_crc32_big(cur, *bulk_iter);
auto single_iter = reinterpret_cast<const std::uint8_t *>(bulk_iter);
for(auto single = data_len % sizeof(sse42_t); single; --single, ++single_iter)
cur = _mm_crc32_u8(cur, *single_iter);
return cur;
}
[[gnu::target("sse4.2")]] bool CRC32Q(parallel_compression_crc32c)() {
return false;
}
#if !MULTIVER
const bool have_crc32q = ([] {
std::uint32_t r[4];
return __get_cpuid_count(1, 0, &r[0], &r[1], &r[2], &r[3]) && ((r[2] & (1 << 20)) == (1 << 20));
}());
const auto crc32c = have_crc32q ? CRC32Q(crc32c) : CRC32C_GENERIC_NAME;
const auto parallel_compression_crc32c = have_crc32q ? CRC32Q(parallel_compression_crc32c) : PARALLEL_COMPRESSION_CRC32C;
#endif
#elif __aarch64__
[[gnu::target("+crc")]] std::uint32_t crc32c(std::uint32_t cur, const void * data, std::size_t data_len) {
auto bulk_iter = reinterpret_cast<const std::uint64_t *>(data);
for(auto bulk = data_len / sizeof(std::uint64_t); bulk; --bulk, ++bulk_iter)
cur = __crc32cd(cur, *bulk_iter);
auto single_iter = reinterpret_cast<const std::uint8_t *>(bulk_iter);
for(auto single = data_len % sizeof(std::uint64_t); single; --single, ++single_iter)
cur = __crc32cb(cur, *single_iter);
return cur;
}
bool parallel_compression_crc32c() {
return false;
}
#elif __loongarch64
std::uint32_t crc32c_crcc(std::uint32_t cur_r, const void * data, std::size_t data_len) {
int cur;
std::memcpy(&cur, &cur_r, sizeof(std::uint32_t));
auto bulk_iter = reinterpret_cast<const long int *>(data);
for(auto bulk = data_len / sizeof(long int); bulk; --bulk, ++bulk_iter)
cur = __crcc_w_d_w(*bulk_iter, cur);
auto single_iter = reinterpret_cast<const char *>(bulk_iter);
for(auto single = data_len % sizeof(long int); single; --single, ++single_iter)
cur = __crcc_w_b_w(*single_iter, cur);
std::memcpy(&cur_r, &cur, sizeof(std::uint32_t));
return cur_r;
}
// bool parallel_compression_crc32c() {
// // TODO: measure on hardware! defaults to true, which wins in QEMU
// }
// As of 2024-01-14, the official upstream ISA manual (https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html#_cpucfg) says:
// Table 3. The configuration information accessible by the CPUCFG instruction
// Word number Bit number Annotation Implication
// 0x1 25 IOCSR_BRD 1 indicates that the string of processor product information is recorded at address 0 of the IOCSR access space
//
// That is, information such as “Loongson3A5000 @2.5GHz”
//
// This is corroborated by the manuals for the 3A5000/3B5000, 3C5000, and 3D5000
// https://github.com/loongson-community/docs/blob/master/3A5000/Loongson3A5000_3B5000%20user%20book_V1.3.pdf
// https://github.com/loongson-community/docs/blob/master/3C5000/龙芯3C5000寄存器及使用手册V1.0.pdf
// https://github.com/loongson-community/docs/blob/master/3D5000/2023061508513851030.龙芯3D5000处理器寄存器使用手册_V1.0.pdf
// whereas the 3A4000 manual allocates cpucfg completely differently
// https://github.com/loongson-community/docs/blob/master/3A4000/3A4000_user_v1.5_20191220.pdf
//
// The 3A6000's allocation scheme agrees with the live doc and the 5000s except bit 25 is
// CRC32 为 1 表示支持 CRC32 加速指令。 1’b1
// https://github.com/loongson-community/docs/blob/master/3A6000/Loongson3A6000%20user%20book_V1.1.pdf
//
// Linux checks bit 25 for CRC/CRCC presence since
// https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=df830336045db1246d3245d3737fee9939c5f731
// saying
// LoongArch: Fix probing of the CRC32 feature
//
// Not all LoongArch processors support CRC32 instructions. This feature
// is indicated by CPUCFG1.CRC32 (Bit25) but it is wrongly defined in the
// previous versions of the ISA manual (and so does in loongarch.h). The
// CRC32 feature is set unconditionally now, so fix it.
#define LOONG64_CPUCFG1_CRC32 (1 << 25)
static const auto crc32c = __cpucfg(1) & LOONG64_CPUCFG1_CRC32 ? crc32c_crcc : crc32c_generic;
#endif
std::uint32_t crc32c_finish(std::uint32_t cur) {
cur = ~cur;
// https://github.com/google/snappy/blob/f82bff66afe0de4c9ae22f8c4ef84e3c2233e799/framing_format.txt#L53-L55
cur = ((cur >> 15) | (cur << 17)) + 0xA282EAD8;
return cur;
}
static union {
char buf[0xFFFFFF];
std::uint32_t bufsum;
};
// A Source is an interface that yields a sequence of bytes
struct fd_source : snappy::Source {
int fd;
const char * filename;
off_t size, off;
fd_source(int fd, const char * filename) : fd(fd), filename(filename) {
if((this->off = (filename == "-"sv ? lseek(this->fd, 0, SEEK_CUR) : 0)) == -1)
return;
if((this->size = lseek(this->fd, 0, SEEK_END)) == -1)
this->off = -1;
}
// Return the number of bytes left to read from the source
virtual size_t Available() const { return this->size - this->off; }
// Peek at the next flat region of the source. Does not reposition
// the source. The returned region is empty iff Available()==0.
//
// Returns a pointer to the beginning of the region and store its
// length in *len.
//
// The returned region is valid until the next call to Skip() or
// until this object is destroyed, whichever occurs first.
//
// The returned region may be larger than Available() (for example
// if this ByteSource is a view on a substring of a larger source).
// The caller is responsible for ensuring that it only reads the
// Available() bytes.
virtual const char * Peek(size_t * len) {
static_assert(snappy::kBlockSize <= sizeof(buf));
ssize_t rd;
while((rd = pread(this->fd, buf, std::min(static_cast<off_t>(snappy::kBlockSize), this->size), this->off)) == -1 && errno == EINTR)
;
if(rd == -1)
err(2, "%s", this->filename);
if(rd == 0)
this->off = this->size;
*len = rd;
return buf;
}
// Skip the next n bytes. Invalidates any buffer returned by
// a previous call to Peek().
// REQUIRES: Available() >= n
virtual void Skip(size_t n) { this->off += n; }
};
struct FILE_sink : snappy::Sink {
FILE * f;
FILE_sink(FILE * f) : f(f) {}
// Append "bytes[0,n-1]" to this.
virtual void Append(const char * bytes, size_t n) {
if(std::fwrite(bytes, 1, n, this->f) != n)
err(2, gettext("write error"));
}
};
struct FILE_crc32_sink : FILE_sink {
std::uint32_t sum = ~0;
using FILE_sink::FILE_sink;
// Append "bytes[0,n-1]" to this.
virtual void Append(const char * bytes, size_t n) {
FILE_sink::Append(bytes, n);
this->sum = crc32c(this->sum, bytes, n);
}
};
bool uncompress_unframed(const void * data, std::size_t data_len, snappy::Sink * sink, auto filename, bool ignore_errors) {
std::uint32_t len = -1;
bool err{};
{
snappy::ByteArraySource src{reinterpret_cast<const char *>(data), data_len};
if(!snappy::GetUncompressedLength(&src, &len)) {
// not understood by the decompressor
warnx(gettext("%s: compressed block of length %zu: invalid data"), filename, data_len), err = true;
if(!ignore_errors)
return err;
}
}
{
snappy::ByteArraySource src{reinterpret_cast<const char *>(data), data_len};
auto un = snappy::UncompressAsMuchAsPossible(&src, sink);
if(un != len)
warnx(gettext("%s: compressed block of length %zu: expecting %zu bytes, got %zu"), filename, data_len, static_cast<std::size_t>(len), un), err = true;
}
return err;
}
// The stream ends when the file ends -- there is no explicit end-of-file marker.
int uncompress_framed(FILE * f, const char * filename, bool ignore_errors) {
bool err{};
for(;;) {
auto tp = getc(f);
if(tp == EOF) {
checkerr:
if(std::ferror(f))
::err(2, "%s", filename);
else
return err;
}
std::uint32_t chunk_len{};
for(int i = 0; i < 3; ++i) {
auto l = getc(f);
if(l == EOF)
goto checkerr;
chunk_len |= l << (i * 8);
}
if(std::fread(buf, 1, chunk_len, f) != chunk_len)
goto checkerr;
FILE_crc32_sink sink{stdout};
switch(tp) {
case 0xFF: // 4.1. Stream identifier (chunk type 0xff)
if(chunk_len != std::strlen("sNaPpY"))
warnx(gettext("%s: stream identifier chunk: length %zu != %zu"), filename, static_cast<std::size_t>(chunk_len), std::strlen("sNaPpY")), err = true;
if(std::memcmp(buf, "sNaPpY", std::strlen("sNaPpY")))
warnx(gettext("%s: stream identifier chunk: content %.*s != %s"), filename, (int)chunk_len, buf, "sNaPpY"), err = true;
break;
case 0x00: // 4.2. Compressed data (chunk type 0x00)
err |= uncompress_unframed(buf + 4, chunk_len - 4, &sink, filename, ignore_errors);
sink.f = nullptr;
break;
case 0x01: // 4.3. Uncompressed data (chunk type 0x01)
sink.Append(buf + 4, chunk_len - 4);
sink.f = nullptr;
break;
case 0xFE: // 4.4. Padding (chunk type 0xfe)
break;
default:
if(tp >= 0x02 && tp <= 0x7F) // 4.5. Reserved unskippable chunks (chunk types 0x02-0x7f)
warnx(gettext("%s: chunk of length %zu: unknown type 0x%02X"), filename, static_cast<std::size_t>(chunk_len), tp), err = true;
else // 4.6. Reserved skippable chunks (chunk types 0x80-0xfd)
;
}
if(err && !ignore_errors)
return err;
if(!sink.f) {
sink.sum = crc32c_finish(sink.sum);
if(le32toh(bufsum) != sink.sum) {
warnx(gettext("%s: chunk of length %zu: checksum 0x%08X != 0x%08X"), filename, static_cast<std::size_t>(chunk_len), sink.sum, le32toh(bufsum)),
err = true;
if(!ignore_errors)
return err;
}
}
}
return err;
}
std::pair<void *, std::size_t> ingest(FILE * f, const char * filename, const void * prefix = nullptr, std::size_t prefix_len = 0) {
char * data;
std::size_t data_size;
auto mf = open_memstream(&data, &data_size);
if(!mf)
mferr:
err(2, nullptr);
std::fwrite(prefix, 1, prefix_len, mf);
for(auto rd = sizeof(buf); rd == sizeof(buf);) {
if((rd = std::fread(buf, 1, sizeof(buf), f)) != sizeof(buf) && std::ferror(f))
err(2, "%s", filename);
if(std::fwrite(buf, 1, rd, mf) != rd)
goto mferr;
}
if(std::fclose(mf))
goto mferr;
return {data, data_size};
}
struct stdout_flush {
~stdout_flush() {
if(std::fflush(stdout))
err(2, gettext("write error"));
}
};
}
int main(int argc, char * const * argv) {
setlocale(LC_ALL, "");
#if __has_include(<libintl.h>)
bindtextdomain("snappy-tools", TEXTDOMAIN_DIRNAME);
textdomain("snappy-tools");
#endif
auto bn = std::strrchr(argv[0] ?: "", '/') ?: (argv[0] ?: "");
auto uncompress_argv0 = *bn == 'u' || (*bn == '/' && *(bn + 1) == 'u');
bool uncompress = uncompress_argv0;
bool uncompress_ignore_errors{};
bool compress_frame{};
for(int arg; (arg = getopt(argc, argv, uncompress_argv0 ? "i" : "dif")) != -1;)
switch(arg) {
case 'd':
uncompress = true;
break;
case 'i':
uncompress_ignore_errors = true;
break;
case 'f':
compress_frame = true;
break;
default:
usage:
// Also in README!
return std::fprintf(stderr,
uncompress_argv0 ? gettext("usage: %1$s [-i] snappy.sn|.sz\n"
" %1$s [-i] < snappy.sn|.sz\n")
: gettext("usage: %1$s [-f] data > snappy.sn|.sz\n"
" %1$s [-f] < data > snappy.sn|.sz\n"
" %1$s -d [-i] snappy.sn|.sz\n"
" %1$s -d [-i] < snappy.sn|.sz\n"),
argv[0]),
1;
}
if(*(argv + optind) && *(argv + optind + 1))
goto usage;
auto filename = *(argv + optind) ?: "-";
if(filename != "-"sv)
if(!std::freopen(filename, "r", stdin))
err(2, "%s", filename);
const int fd = 0;
FILE * const f = stdin;
stdout_flush _flusher{};
if(uncompress) {
auto hdrlen = std::fread(buf, 1, sizeof("\xFF\x06\x00\x00sNaPpY") - 1, f);
if(hdrlen != sizeof("\xFF\x06\x00\x00sNaPpY") - 1) {
if(std::ferror(f))
err(2, "%s", filename);
FILE_sink sink{stdout};
return uncompress_unframed(buf, hdrlen, &sink, filename, uncompress_ignore_errors);
} else if(!std::memcmp(buf, "\xFF\x06\x00\x00sNaPpY", sizeof("\xFF\x06\x00\x00sNaPpY") - 1))
return uncompress_framed(f, filename, uncompress_ignore_errors);
// Ideally we'd just ungetc but realistically glibc/musl give us 8 bytes and we want 9 :/
// while(hdrlen--)
// assert(std::ungetc(buf[hdrlen], stdin) != EOF);
off_t orig_pos, len;
const void * mapping;
if((orig_pos = ftello(f)) == -1 || fseeko(f, 0, SEEK_END) == -1 || (len = ftello(f)) == -1 ||
((mapping = mmap(nullptr, len, PROT_READ, MAP_PRIVATE, fd, 0)) == MAP_FAILED && (fseeko(f, orig_pos, SEEK_SET), true))) {
auto dt = ingest(f, filename, buf, hdrlen);
FILE_sink sink{stdout};
return uncompress_unframed(dt.first, dt.second, &sink, filename, uncompress_ignore_errors);
}
FILE_sink sink{stdout};
return uncompress_unframed(reinterpret_cast<const std::uint8_t *>(mapping) + (orig_pos - hdrlen), len - (orig_pos - hdrlen), &sink, filename,
uncompress_ignore_errors);
} else {
unsigned long long read{}, written{};
// No thread: {read → compress → cksum → write}...
//
// Yes thread: {read 🚧→ compress 🚧→ write}...
// ↓ len ↑ sum
// thread: {🚧→ cksum 🚧→}...
const auto crc_in_thread = parallel_compression_crc32c();
struct ipc {
pthread_barrier_t barrier;
std::atomic<std::uint32_t> len_sum;
} ipc;
pthread_t cksum_thread;
if(crc_in_thread) {
while(pthread_barrier_init(&ipc.barrier, nullptr, 2))
;
while(pthread_create(
&cksum_thread, nullptr,
[](void * ipcp) -> void * {
auto & ipc = *reinterpret_cast<struct ipc *>(ipcp);
for(;;) {
pthread_barrier_wait(&ipc.barrier);
ipc.len_sum.store(htole32(crc32c_finish(crc32c(~0, buf, ipc.len_sum.load(std::memory_order::relaxed)))), std::memory_order::relaxed);
pthread_barrier_wait(&ipc.barrier);
}
__builtin_unreachable();
},
&ipc))
;
}
if(compress_frame) {
if(std::fwrite("\xFF\x06\x00\x00sNaPpY", 1, sizeof("\xFF\x06\x00\x00sNaPpY") - 1, stdout) != sizeof("\xFF\x06\x00\x00sNaPpY") - 1)
we:
err(2, gettext("write error"));
written = sizeof("\xFF\x06\x00\x00sNaPpY") - 1;
// However, we place an additional restriction that the uncompressed data in a chunk must be no longer than 65536 bytes.
for(std::size_t rd = 65536; rd == 65536;) {
if((rd = std::fread(buf, 1, 65536, f)) != 65536 && std::ferror(f))
err(2, "%s", filename);
read += rd;
if(!rd)
break;
if(crc_in_thread) {
ipc.len_sum.store(rd, std::memory_order::relaxed);
pthread_barrier_wait(&ipc.barrier);
}
auto compbuf = buf + 65536;
std::size_t compbuf_len = -1;
snappy::RawCompress(buf, rd, compbuf, &compbuf_len);
// Both the uncompressed and the compressed chunks have the same final size: [1 + 3 ] + 4 + data
// [type + len] + cksum + data
// [header ] + body
// 4.2. Compressed data (chunk type 0x00)
// 4.3. Uncompressed data (chunk type 0x01)
std::uint8_t tp = (compbuf_len < rd) ? 0x00 : 0x01;
if(putc(tp, stdout) == EOF)
goto we;
auto outsize = 4 + (tp ? rd : compbuf_len);
if(putc((outsize & 0x0000FF) >> 0, stdout) == EOF || //
putc((outsize & 0x00FF00) >> 8, stdout) == EOF || //
putc((outsize & 0xFF0000) >> 16, stdout) == EOF)
goto we;
std::uint32_t crc;
if(crc_in_thread) {
pthread_barrier_wait(&ipc.barrier);
crc = ipc.len_sum.load(std::memory_order::relaxed);
} else
crc = htole32(crc32c_finish(crc32c(~0, buf, rd)));
if(std::fwrite(&crc, 1, 4, stdout) != 4)
goto we;
if(std::fwrite(tp ? buf : compbuf, 1, tp ? rd : compbuf_len, stdout) != (tp ? rd : compbuf_len))
goto we;
written += 1 + 3 + 4 + (tp ? rd : compbuf_len);
}
} else {
FILE_sink sink{stdout};
if(fd_source source{fd, filename}; source.size && source.off != -1) {
read = source.size - source.off;
if(read != source.Available())
errno = EOVERFLOW, err(2, "%s", filename);
written = Compress(&source, &sink);
} else {
auto dt = ingest(f, filename);
read = dt.second;
snappy::ByteArraySource src{reinterpret_cast<const char *>(dt.first), dt.second};
written = Compress(&src, &sink);
}
}
// filename, byte count read, byte count written, ratio
std::fprintf(stderr, gettext("%s: %llu -> %llu (%.2f%%)\n"), filename, read, written, 100. * ((double)written / read));
if(!compress_frame && read >= 4ull * 1024 * 1024 * 1024)
return std::fprintf(stderr, gettext("%s: %s: sized %lluB >= 4GiB w/o -f: output stream may be broken!\n"), argv[0], filename, read), 3;
}
}
|