James Kuszmaul | 48dd4c8 | 2021-10-27 20:04:08 -0700 | [diff] [blame^] | 1 | // Copyright 2005 and onwards Google Inc. |
| 2 | // |
| 3 | // Redistribution and use in source and binary forms, with or without |
| 4 | // modification, are permitted provided that the following conditions are |
| 5 | // met: |
| 6 | // |
| 7 | // * Redistributions of source code must retain the above copyright |
| 8 | // notice, this list of conditions and the following disclaimer. |
| 9 | // * Redistributions in binary form must reproduce the above |
| 10 | // copyright notice, this list of conditions and the following disclaimer |
| 11 | // in the documentation and/or other materials provided with the |
| 12 | // distribution. |
| 13 | // * Neither the name of Google Inc. nor the names of its |
| 14 | // contributors may be used to endorse or promote products derived from |
| 15 | // this software without specific prior written permission. |
| 16 | // |
| 17 | // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 18 | // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 19 | // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| 20 | // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| 21 | // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| 22 | // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| 23 | // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| 24 | // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| 25 | // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 26 | // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 27 | // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 28 | |
| 29 | #include <algorithm> |
| 30 | #include <cmath> |
| 31 | #include <cstdlib> |
| 32 | #include <random> |
| 33 | #include <string> |
| 34 | #include <utility> |
| 35 | #include <vector> |
| 36 | |
| 37 | #include "snappy-test.h" |
| 38 | |
| 39 | #include "gtest/gtest.h" |
| 40 | |
| 41 | #include "snappy-internal.h" |
| 42 | #include "snappy-sinksource.h" |
| 43 | #include "snappy.h" |
| 44 | #include "snappy_test_data.h" |
| 45 | |
| 46 | SNAPPY_FLAG(bool, snappy_dump_decompression_table, false, |
| 47 | "If true, we print the decompression table during tests."); |
| 48 | |
| 49 | namespace snappy { |
| 50 | |
| 51 | namespace { |
| 52 | |
| 53 | #if HAVE_FUNC_MMAP && HAVE_FUNC_SYSCONF |
| 54 | |
| 55 | // To test against code that reads beyond its input, this class copies a |
| 56 | // string to a newly allocated group of pages, the last of which |
| 57 | // is made unreadable via mprotect. Note that we need to allocate the |
| 58 | // memory with mmap(), as POSIX allows mprotect() only on memory allocated |
| 59 | // with mmap(), and some malloc/posix_memalign implementations expect to |
| 60 | // be able to read previously allocated memory while doing heap allocations. |
| 61 | class DataEndingAtUnreadablePage { |
| 62 | public: |
| 63 | explicit DataEndingAtUnreadablePage(const std::string& s) { |
| 64 | const size_t page_size = sysconf(_SC_PAGESIZE); |
| 65 | const size_t size = s.size(); |
| 66 | // Round up space for string to a multiple of page_size. |
| 67 | size_t space_for_string = (size + page_size - 1) & ~(page_size - 1); |
| 68 | alloc_size_ = space_for_string + page_size; |
| 69 | mem_ = mmap(NULL, alloc_size_, |
| 70 | PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0); |
| 71 | CHECK_NE(MAP_FAILED, mem_); |
| 72 | protected_page_ = reinterpret_cast<char*>(mem_) + space_for_string; |
| 73 | char* dst = protected_page_ - size; |
| 74 | std::memcpy(dst, s.data(), size); |
| 75 | data_ = dst; |
| 76 | size_ = size; |
| 77 | // Make guard page unreadable. |
| 78 | CHECK_EQ(0, mprotect(protected_page_, page_size, PROT_NONE)); |
| 79 | } |
| 80 | |
| 81 | ~DataEndingAtUnreadablePage() { |
| 82 | const size_t page_size = sysconf(_SC_PAGESIZE); |
| 83 | // Undo the mprotect. |
| 84 | CHECK_EQ(0, mprotect(protected_page_, page_size, PROT_READ|PROT_WRITE)); |
| 85 | CHECK_EQ(0, munmap(mem_, alloc_size_)); |
| 86 | } |
| 87 | |
| 88 | const char* data() const { return data_; } |
| 89 | size_t size() const { return size_; } |
| 90 | |
| 91 | private: |
| 92 | size_t alloc_size_; |
| 93 | void* mem_; |
| 94 | char* protected_page_; |
| 95 | const char* data_; |
| 96 | size_t size_; |
| 97 | }; |
| 98 | |
| 99 | #else // HAVE_FUNC_MMAP) && HAVE_FUNC_SYSCONF |
| 100 | |
| 101 | // Fallback for systems without mmap. |
| 102 | using DataEndingAtUnreadablePage = std::string; |
| 103 | |
| 104 | #endif |
| 105 | |
| 106 | int VerifyString(const std::string& input) { |
| 107 | std::string compressed; |
| 108 | DataEndingAtUnreadablePage i(input); |
| 109 | const size_t written = snappy::Compress(i.data(), i.size(), &compressed); |
| 110 | CHECK_EQ(written, compressed.size()); |
| 111 | CHECK_LE(compressed.size(), |
| 112 | snappy::MaxCompressedLength(input.size())); |
| 113 | CHECK(snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 114 | |
| 115 | std::string uncompressed; |
| 116 | DataEndingAtUnreadablePage c(compressed); |
| 117 | CHECK(snappy::Uncompress(c.data(), c.size(), &uncompressed)); |
| 118 | CHECK_EQ(uncompressed, input); |
| 119 | return uncompressed.size(); |
| 120 | } |
| 121 | |
| 122 | void VerifyStringSink(const std::string& input) { |
| 123 | std::string compressed; |
| 124 | DataEndingAtUnreadablePage i(input); |
| 125 | const size_t written = snappy::Compress(i.data(), i.size(), &compressed); |
| 126 | CHECK_EQ(written, compressed.size()); |
| 127 | CHECK_LE(compressed.size(), |
| 128 | snappy::MaxCompressedLength(input.size())); |
| 129 | CHECK(snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 130 | |
| 131 | std::string uncompressed; |
| 132 | uncompressed.resize(input.size()); |
| 133 | snappy::UncheckedByteArraySink sink(string_as_array(&uncompressed)); |
| 134 | DataEndingAtUnreadablePage c(compressed); |
| 135 | snappy::ByteArraySource source(c.data(), c.size()); |
| 136 | CHECK(snappy::Uncompress(&source, &sink)); |
| 137 | CHECK_EQ(uncompressed, input); |
| 138 | } |
| 139 | |
| 140 | void VerifyIOVec(const std::string& input) { |
| 141 | std::string compressed; |
| 142 | DataEndingAtUnreadablePage i(input); |
| 143 | const size_t written = snappy::Compress(i.data(), i.size(), &compressed); |
| 144 | CHECK_EQ(written, compressed.size()); |
| 145 | CHECK_LE(compressed.size(), |
| 146 | snappy::MaxCompressedLength(input.size())); |
| 147 | CHECK(snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 148 | |
| 149 | // Try uncompressing into an iovec containing a random number of entries |
| 150 | // ranging from 1 to 10. |
| 151 | char* buf = new char[input.size()]; |
| 152 | std::minstd_rand0 rng(input.size()); |
| 153 | std::uniform_int_distribution<size_t> uniform_1_to_10(1, 10); |
| 154 | size_t num = uniform_1_to_10(rng); |
| 155 | if (input.size() < num) { |
| 156 | num = input.size(); |
| 157 | } |
| 158 | struct iovec* iov = new iovec[num]; |
| 159 | size_t used_so_far = 0; |
| 160 | std::bernoulli_distribution one_in_five(1.0 / 5); |
| 161 | for (size_t i = 0; i < num; ++i) { |
| 162 | assert(used_so_far < input.size()); |
| 163 | iov[i].iov_base = buf + used_so_far; |
| 164 | if (i == num - 1) { |
| 165 | iov[i].iov_len = input.size() - used_so_far; |
| 166 | } else { |
| 167 | // Randomly choose to insert a 0 byte entry. |
| 168 | if (one_in_five(rng)) { |
| 169 | iov[i].iov_len = 0; |
| 170 | } else { |
| 171 | std::uniform_int_distribution<size_t> uniform_not_used_so_far( |
| 172 | 0, input.size() - used_so_far - 1); |
| 173 | iov[i].iov_len = uniform_not_used_so_far(rng); |
| 174 | } |
| 175 | } |
| 176 | used_so_far += iov[i].iov_len; |
| 177 | } |
| 178 | CHECK(snappy::RawUncompressToIOVec( |
| 179 | compressed.data(), compressed.size(), iov, num)); |
| 180 | CHECK(!memcmp(buf, input.data(), input.size())); |
| 181 | delete[] iov; |
| 182 | delete[] buf; |
| 183 | } |
| 184 | |
| 185 | // Test that data compressed by a compressor that does not |
| 186 | // obey block sizes is uncompressed properly. |
| 187 | void VerifyNonBlockedCompression(const std::string& input) { |
| 188 | if (input.length() > snappy::kBlockSize) { |
| 189 | // We cannot test larger blocks than the maximum block size, obviously. |
| 190 | return; |
| 191 | } |
| 192 | |
| 193 | std::string prefix; |
| 194 | Varint::Append32(&prefix, input.size()); |
| 195 | |
| 196 | // Setup compression table |
| 197 | snappy::internal::WorkingMemory wmem(input.size()); |
| 198 | int table_size; |
| 199 | uint16_t* table = wmem.GetHashTable(input.size(), &table_size); |
| 200 | |
| 201 | // Compress entire input in one shot |
| 202 | std::string compressed; |
| 203 | compressed += prefix; |
| 204 | compressed.resize(prefix.size()+snappy::MaxCompressedLength(input.size())); |
| 205 | char* dest = string_as_array(&compressed) + prefix.size(); |
| 206 | char* end = snappy::internal::CompressFragment(input.data(), input.size(), |
| 207 | dest, table, table_size); |
| 208 | compressed.resize(end - compressed.data()); |
| 209 | |
| 210 | // Uncompress into std::string |
| 211 | std::string uncomp_str; |
| 212 | CHECK(snappy::Uncompress(compressed.data(), compressed.size(), &uncomp_str)); |
| 213 | CHECK_EQ(uncomp_str, input); |
| 214 | |
| 215 | // Uncompress using source/sink |
| 216 | std::string uncomp_str2; |
| 217 | uncomp_str2.resize(input.size()); |
| 218 | snappy::UncheckedByteArraySink sink(string_as_array(&uncomp_str2)); |
| 219 | snappy::ByteArraySource source(compressed.data(), compressed.size()); |
| 220 | CHECK(snappy::Uncompress(&source, &sink)); |
| 221 | CHECK_EQ(uncomp_str2, input); |
| 222 | |
| 223 | // Uncompress into iovec |
| 224 | { |
| 225 | static const int kNumBlocks = 10; |
| 226 | struct iovec vec[kNumBlocks]; |
| 227 | const int block_size = 1 + input.size() / kNumBlocks; |
| 228 | std::string iovec_data(block_size * kNumBlocks, 'x'); |
| 229 | for (int i = 0; i < kNumBlocks; ++i) { |
| 230 | vec[i].iov_base = string_as_array(&iovec_data) + i * block_size; |
| 231 | vec[i].iov_len = block_size; |
| 232 | } |
| 233 | CHECK(snappy::RawUncompressToIOVec(compressed.data(), compressed.size(), |
| 234 | vec, kNumBlocks)); |
| 235 | CHECK_EQ(std::string(iovec_data.data(), input.size()), input); |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | // Expand the input so that it is at least K times as big as block size |
| 240 | std::string Expand(const std::string& input) { |
| 241 | static const int K = 3; |
| 242 | std::string data = input; |
| 243 | while (data.size() < K * snappy::kBlockSize) { |
| 244 | data += input; |
| 245 | } |
| 246 | return data; |
| 247 | } |
| 248 | |
| 249 | int Verify(const std::string& input) { |
| 250 | VLOG(1) << "Verifying input of size " << input.size(); |
| 251 | |
| 252 | // Compress using string based routines |
| 253 | const int result = VerifyString(input); |
| 254 | |
| 255 | // Verify using sink based routines |
| 256 | VerifyStringSink(input); |
| 257 | |
| 258 | VerifyNonBlockedCompression(input); |
| 259 | VerifyIOVec(input); |
| 260 | if (!input.empty()) { |
| 261 | const std::string expanded = Expand(input); |
| 262 | VerifyNonBlockedCompression(expanded); |
| 263 | VerifyIOVec(input); |
| 264 | } |
| 265 | |
| 266 | return result; |
| 267 | } |
| 268 | |
| 269 | bool IsValidCompressedBuffer(const std::string& c) { |
| 270 | return snappy::IsValidCompressedBuffer(c.data(), c.size()); |
| 271 | } |
| 272 | bool Uncompress(const std::string& c, std::string* u) { |
| 273 | return snappy::Uncompress(c.data(), c.size(), u); |
| 274 | } |
| 275 | |
| 276 | // This test checks to ensure that snappy doesn't coredump if it gets |
| 277 | // corrupted data. |
| 278 | TEST(CorruptedTest, VerifyCorrupted) { |
| 279 | std::string source = "making sure we don't crash with corrupted input"; |
| 280 | VLOG(1) << source; |
| 281 | std::string dest; |
| 282 | std::string uncmp; |
| 283 | snappy::Compress(source.data(), source.size(), &dest); |
| 284 | |
| 285 | // Mess around with the data. It's hard to simulate all possible |
| 286 | // corruptions; this is just one example ... |
| 287 | CHECK_GT(dest.size(), 3); |
| 288 | dest[1]--; |
| 289 | dest[3]++; |
| 290 | // this really ought to fail. |
| 291 | CHECK(!IsValidCompressedBuffer(dest)); |
| 292 | CHECK(!Uncompress(dest, &uncmp)); |
| 293 | |
| 294 | // This is testing for a security bug - a buffer that decompresses to 100k |
| 295 | // but we lie in the snappy header and only reserve 0 bytes of memory :) |
| 296 | source.resize(100000); |
| 297 | for (char& source_char : source) { |
| 298 | source_char = 'A'; |
| 299 | } |
| 300 | snappy::Compress(source.data(), source.size(), &dest); |
| 301 | dest[0] = dest[1] = dest[2] = dest[3] = 0; |
| 302 | CHECK(!IsValidCompressedBuffer(dest)); |
| 303 | CHECK(!Uncompress(dest, &uncmp)); |
| 304 | |
| 305 | if (sizeof(void *) == 4) { |
| 306 | // Another security check; check a crazy big length can't DoS us with an |
| 307 | // over-allocation. |
| 308 | // Currently this is done only for 32-bit builds. On 64-bit builds, |
| 309 | // where 3 GB might be an acceptable allocation size, Uncompress() |
| 310 | // attempts to decompress, and sometimes causes the test to run out of |
| 311 | // memory. |
| 312 | dest[0] = dest[1] = dest[2] = dest[3] = '\xff'; |
| 313 | // This decodes to a really large size, i.e., about 3 GB. |
| 314 | dest[4] = 'k'; |
| 315 | CHECK(!IsValidCompressedBuffer(dest)); |
| 316 | CHECK(!Uncompress(dest, &uncmp)); |
| 317 | } else { |
| 318 | LOG(WARNING) << "Crazy decompression lengths not checked on 64-bit build"; |
| 319 | } |
| 320 | |
| 321 | // This decodes to about 2 MB; much smaller, but should still fail. |
| 322 | dest[0] = dest[1] = dest[2] = '\xff'; |
| 323 | dest[3] = 0x00; |
| 324 | CHECK(!IsValidCompressedBuffer(dest)); |
| 325 | CHECK(!Uncompress(dest, &uncmp)); |
| 326 | |
| 327 | // try reading stuff in from a bad file. |
| 328 | for (int i = 1; i <= 3; ++i) { |
| 329 | std::string data = |
| 330 | ReadTestDataFile(StrFormat("baddata%d.snappy", i).c_str(), 0); |
| 331 | std::string uncmp; |
| 332 | // check that we don't return a crazy length |
| 333 | size_t ulen; |
| 334 | CHECK(!snappy::GetUncompressedLength(data.data(), data.size(), &ulen) |
| 335 | || (ulen < (1<<20))); |
| 336 | uint32_t ulen2; |
| 337 | snappy::ByteArraySource source(data.data(), data.size()); |
| 338 | CHECK(!snappy::GetUncompressedLength(&source, &ulen2) || |
| 339 | (ulen2 < (1<<20))); |
| 340 | CHECK(!IsValidCompressedBuffer(data)); |
| 341 | CHECK(!Uncompress(data, &uncmp)); |
| 342 | } |
| 343 | } |
| 344 | |
| 345 | // Helper routines to construct arbitrary compressed strings. |
| 346 | // These mirror the compression code in snappy.cc, but are copied |
| 347 | // here so that we can bypass some limitations in the how snappy.cc |
| 348 | // invokes these routines. |
| 349 | void AppendLiteral(std::string* dst, const std::string& literal) { |
| 350 | if (literal.empty()) return; |
| 351 | int n = literal.size() - 1; |
| 352 | if (n < 60) { |
| 353 | // Fit length in tag byte |
| 354 | dst->push_back(0 | (n << 2)); |
| 355 | } else { |
| 356 | // Encode in upcoming bytes |
| 357 | char number[4]; |
| 358 | int count = 0; |
| 359 | while (n > 0) { |
| 360 | number[count++] = n & 0xff; |
| 361 | n >>= 8; |
| 362 | } |
| 363 | dst->push_back(0 | ((59+count) << 2)); |
| 364 | *dst += std::string(number, count); |
| 365 | } |
| 366 | *dst += literal; |
| 367 | } |
| 368 | |
| 369 | void AppendCopy(std::string* dst, int offset, int length) { |
| 370 | while (length > 0) { |
| 371 | // Figure out how much to copy in one shot |
| 372 | int to_copy; |
| 373 | if (length >= 68) { |
| 374 | to_copy = 64; |
| 375 | } else if (length > 64) { |
| 376 | to_copy = 60; |
| 377 | } else { |
| 378 | to_copy = length; |
| 379 | } |
| 380 | length -= to_copy; |
| 381 | |
| 382 | if ((to_copy >= 4) && (to_copy < 12) && (offset < 2048)) { |
| 383 | assert(to_copy-4 < 8); // Must fit in 3 bits |
| 384 | dst->push_back(1 | ((to_copy-4) << 2) | ((offset >> 8) << 5)); |
| 385 | dst->push_back(offset & 0xff); |
| 386 | } else if (offset < 65536) { |
| 387 | dst->push_back(2 | ((to_copy-1) << 2)); |
| 388 | dst->push_back(offset & 0xff); |
| 389 | dst->push_back(offset >> 8); |
| 390 | } else { |
| 391 | dst->push_back(3 | ((to_copy-1) << 2)); |
| 392 | dst->push_back(offset & 0xff); |
| 393 | dst->push_back((offset >> 8) & 0xff); |
| 394 | dst->push_back((offset >> 16) & 0xff); |
| 395 | dst->push_back((offset >> 24) & 0xff); |
| 396 | } |
| 397 | } |
| 398 | } |
| 399 | |
| 400 | TEST(Snappy, SimpleTests) { |
| 401 | Verify(""); |
| 402 | Verify("a"); |
| 403 | Verify("ab"); |
| 404 | Verify("abc"); |
| 405 | |
| 406 | Verify("aaaaaaa" + std::string(16, 'b') + std::string("aaaaa") + "abc"); |
| 407 | Verify("aaaaaaa" + std::string(256, 'b') + std::string("aaaaa") + "abc"); |
| 408 | Verify("aaaaaaa" + std::string(2047, 'b') + std::string("aaaaa") + "abc"); |
| 409 | Verify("aaaaaaa" + std::string(65536, 'b') + std::string("aaaaa") + "abc"); |
| 410 | Verify("abcaaaaaaa" + std::string(65536, 'b') + std::string("aaaaa") + "abc"); |
| 411 | } |
| 412 | |
| 413 | // Regression test for cr/345340892. |
| 414 | TEST(Snappy, AppendSelfPatternExtensionEdgeCases) { |
| 415 | Verify("abcabcabcabcabcabcab"); |
| 416 | Verify("abcabcabcabcabcabcab0123456789ABCDEF"); |
| 417 | |
| 418 | Verify("abcabcabcabcabcabcabcabcabcabcabcabc"); |
| 419 | Verify("abcabcabcabcabcabcabcabcabcabcabcabc0123456789ABCDEF"); |
| 420 | } |
| 421 | |
| 422 | // Regression test for cr/345340892. |
| 423 | TEST(Snappy, AppendSelfPatternExtensionEdgeCasesExhaustive) { |
| 424 | std::mt19937 rng; |
| 425 | std::uniform_int_distribution<int> uniform_byte(0, 255); |
| 426 | for (int pattern_size = 1; pattern_size <= 18; ++pattern_size) { |
| 427 | for (int length = 1; length <= 64; ++length) { |
| 428 | for (int extra_bytes_after_pattern : {0, 1, 15, 16, 128}) { |
| 429 | const int size = pattern_size + length + extra_bytes_after_pattern; |
| 430 | std::string input; |
| 431 | input.resize(size); |
| 432 | for (int i = 0; i < pattern_size; ++i) { |
| 433 | input[i] = 'a' + i; |
| 434 | } |
| 435 | for (int i = 0; i < length; ++i) { |
| 436 | input[pattern_size + i] = input[i]; |
| 437 | } |
| 438 | for (int i = 0; i < extra_bytes_after_pattern; ++i) { |
| 439 | input[pattern_size + length + i] = |
| 440 | static_cast<char>(uniform_byte(rng)); |
| 441 | } |
| 442 | Verify(input); |
| 443 | } |
| 444 | } |
| 445 | } |
| 446 | } |
| 447 | |
| 448 | // Verify max blowup (lots of four-byte copies) |
| 449 | TEST(Snappy, MaxBlowup) { |
| 450 | std::mt19937 rng; |
| 451 | std::uniform_int_distribution<int> uniform_byte(0, 255); |
| 452 | std::string input; |
| 453 | for (int i = 0; i < 80000; ++i) |
| 454 | input.push_back(static_cast<char>(uniform_byte(rng))); |
| 455 | |
| 456 | for (int i = 0; i < 80000; i += 4) { |
| 457 | std::string four_bytes(input.end() - i - 4, input.end() - i); |
| 458 | input.append(four_bytes); |
| 459 | } |
| 460 | Verify(input); |
| 461 | } |
| 462 | |
| 463 | TEST(Snappy, RandomData) { |
| 464 | std::minstd_rand0 rng(snappy::GetFlag(FLAGS_test_random_seed)); |
| 465 | std::uniform_int_distribution<int> uniform_0_to_3(0, 3); |
| 466 | std::uniform_int_distribution<int> uniform_0_to_8(0, 8); |
| 467 | std::uniform_int_distribution<int> uniform_byte(0, 255); |
| 468 | std::uniform_int_distribution<size_t> uniform_4k(0, 4095); |
| 469 | std::uniform_int_distribution<size_t> uniform_64k(0, 65535); |
| 470 | std::bernoulli_distribution one_in_ten(1.0 / 10); |
| 471 | |
| 472 | constexpr int num_ops = 20000; |
| 473 | for (int i = 0; i < num_ops; ++i) { |
| 474 | if ((i % 1000) == 0) { |
| 475 | VLOG(0) << "Random op " << i << " of " << num_ops; |
| 476 | } |
| 477 | |
| 478 | std::string x; |
| 479 | size_t len = uniform_4k(rng); |
| 480 | if (i < 100) { |
| 481 | len = 65536 + uniform_64k(rng); |
| 482 | } |
| 483 | while (x.size() < len) { |
| 484 | int run_len = 1; |
| 485 | if (one_in_ten(rng)) { |
| 486 | int skewed_bits = uniform_0_to_8(rng); |
| 487 | // int is guaranteed to hold at least 16 bits, this uses at most 8 bits. |
| 488 | std::uniform_int_distribution<int> skewed_low(0, |
| 489 | (1 << skewed_bits) - 1); |
| 490 | run_len = skewed_low(rng); |
| 491 | } |
| 492 | char c = static_cast<char>(uniform_byte(rng)); |
| 493 | if (i >= 100) { |
| 494 | int skewed_bits = uniform_0_to_3(rng); |
| 495 | // int is guaranteed to hold at least 16 bits, this uses at most 3 bits. |
| 496 | std::uniform_int_distribution<int> skewed_low(0, |
| 497 | (1 << skewed_bits) - 1); |
| 498 | c = static_cast<char>(skewed_low(rng)); |
| 499 | } |
| 500 | while (run_len-- > 0 && x.size() < len) { |
| 501 | x.push_back(c); |
| 502 | } |
| 503 | } |
| 504 | |
| 505 | Verify(x); |
| 506 | } |
| 507 | } |
| 508 | |
| 509 | TEST(Snappy, FourByteOffset) { |
| 510 | // The new compressor cannot generate four-byte offsets since |
| 511 | // it chops up the input into 32KB pieces. So we hand-emit the |
| 512 | // copy manually. |
| 513 | |
| 514 | // The two fragments that make up the input string. |
| 515 | std::string fragment1 = "012345689abcdefghijklmnopqrstuvwxyz"; |
| 516 | std::string fragment2 = "some other string"; |
| 517 | |
| 518 | // How many times each fragment is emitted. |
| 519 | const int n1 = 2; |
| 520 | const int n2 = 100000 / fragment2.size(); |
| 521 | const size_t length = n1 * fragment1.size() + n2 * fragment2.size(); |
| 522 | |
| 523 | std::string compressed; |
| 524 | Varint::Append32(&compressed, length); |
| 525 | |
| 526 | AppendLiteral(&compressed, fragment1); |
| 527 | std::string src = fragment1; |
| 528 | for (int i = 0; i < n2; ++i) { |
| 529 | AppendLiteral(&compressed, fragment2); |
| 530 | src += fragment2; |
| 531 | } |
| 532 | AppendCopy(&compressed, src.size(), fragment1.size()); |
| 533 | src += fragment1; |
| 534 | CHECK_EQ(length, src.size()); |
| 535 | |
| 536 | std::string uncompressed; |
| 537 | CHECK(snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 538 | CHECK(snappy::Uncompress(compressed.data(), compressed.size(), |
| 539 | &uncompressed)); |
| 540 | CHECK_EQ(uncompressed, src); |
| 541 | } |
| 542 | |
| 543 | TEST(Snappy, IOVecEdgeCases) { |
| 544 | // Test some tricky edge cases in the iovec output that are not necessarily |
| 545 | // exercised by random tests. |
| 546 | |
| 547 | // Our output blocks look like this initially (the last iovec is bigger |
| 548 | // than depicted): |
| 549 | // [ ] [ ] [ ] [ ] [ ] |
| 550 | static const int kLengths[] = { 2, 1, 4, 8, 128 }; |
| 551 | |
| 552 | struct iovec iov[ARRAYSIZE(kLengths)]; |
| 553 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 554 | iov[i].iov_base = new char[kLengths[i]]; |
| 555 | iov[i].iov_len = kLengths[i]; |
| 556 | } |
| 557 | |
| 558 | std::string compressed; |
| 559 | Varint::Append32(&compressed, 22); |
| 560 | |
| 561 | // A literal whose output crosses three blocks. |
| 562 | // [ab] [c] [123 ] [ ] [ ] |
| 563 | AppendLiteral(&compressed, "abc123"); |
| 564 | |
| 565 | // A copy whose output crosses two blocks (source and destination |
| 566 | // segments marked). |
| 567 | // [ab] [c] [1231] [23 ] [ ] |
| 568 | // ^--^ -- |
| 569 | AppendCopy(&compressed, 3, 3); |
| 570 | |
| 571 | // A copy where the input is, at first, in the block before the output: |
| 572 | // |
| 573 | // [ab] [c] [1231] [231231 ] [ ] |
| 574 | // ^--- ^--- |
| 575 | // Then during the copy, the pointers move such that the input and |
| 576 | // output pointers are in the same block: |
| 577 | // |
| 578 | // [ab] [c] [1231] [23123123] [ ] |
| 579 | // ^- ^- |
| 580 | // And then they move again, so that the output pointer is no longer |
| 581 | // in the same block as the input pointer: |
| 582 | // [ab] [c] [1231] [23123123] [123 ] |
| 583 | // ^-- ^-- |
| 584 | AppendCopy(&compressed, 6, 9); |
| 585 | |
| 586 | // Finally, a copy where the input is from several blocks back, |
| 587 | // and it also crosses three blocks: |
| 588 | // |
| 589 | // [ab] [c] [1231] [23123123] [123b ] |
| 590 | // ^ ^ |
| 591 | // [ab] [c] [1231] [23123123] [123bc ] |
| 592 | // ^ ^ |
| 593 | // [ab] [c] [1231] [23123123] [123bc12 ] |
| 594 | // ^- ^- |
| 595 | AppendCopy(&compressed, 17, 4); |
| 596 | |
| 597 | CHECK(snappy::RawUncompressToIOVec( |
| 598 | compressed.data(), compressed.size(), iov, ARRAYSIZE(iov))); |
| 599 | CHECK_EQ(0, memcmp(iov[0].iov_base, "ab", 2)); |
| 600 | CHECK_EQ(0, memcmp(iov[1].iov_base, "c", 1)); |
| 601 | CHECK_EQ(0, memcmp(iov[2].iov_base, "1231", 4)); |
| 602 | CHECK_EQ(0, memcmp(iov[3].iov_base, "23123123", 8)); |
| 603 | CHECK_EQ(0, memcmp(iov[4].iov_base, "123bc12", 7)); |
| 604 | |
| 605 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 606 | delete[] reinterpret_cast<char *>(iov[i].iov_base); |
| 607 | } |
| 608 | } |
| 609 | |
| 610 | TEST(Snappy, IOVecLiteralOverflow) { |
| 611 | static const int kLengths[] = { 3, 4 }; |
| 612 | |
| 613 | struct iovec iov[ARRAYSIZE(kLengths)]; |
| 614 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 615 | iov[i].iov_base = new char[kLengths[i]]; |
| 616 | iov[i].iov_len = kLengths[i]; |
| 617 | } |
| 618 | |
| 619 | std::string compressed; |
| 620 | Varint::Append32(&compressed, 8); |
| 621 | |
| 622 | AppendLiteral(&compressed, "12345678"); |
| 623 | |
| 624 | CHECK(!snappy::RawUncompressToIOVec( |
| 625 | compressed.data(), compressed.size(), iov, ARRAYSIZE(iov))); |
| 626 | |
| 627 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 628 | delete[] reinterpret_cast<char *>(iov[i].iov_base); |
| 629 | } |
| 630 | } |
| 631 | |
| 632 | TEST(Snappy, IOVecCopyOverflow) { |
| 633 | static const int kLengths[] = { 3, 4 }; |
| 634 | |
| 635 | struct iovec iov[ARRAYSIZE(kLengths)]; |
| 636 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 637 | iov[i].iov_base = new char[kLengths[i]]; |
| 638 | iov[i].iov_len = kLengths[i]; |
| 639 | } |
| 640 | |
| 641 | std::string compressed; |
| 642 | Varint::Append32(&compressed, 8); |
| 643 | |
| 644 | AppendLiteral(&compressed, "123"); |
| 645 | AppendCopy(&compressed, 3, 5); |
| 646 | |
| 647 | CHECK(!snappy::RawUncompressToIOVec( |
| 648 | compressed.data(), compressed.size(), iov, ARRAYSIZE(iov))); |
| 649 | |
| 650 | for (int i = 0; i < ARRAYSIZE(kLengths); ++i) { |
| 651 | delete[] reinterpret_cast<char *>(iov[i].iov_base); |
| 652 | } |
| 653 | } |
| 654 | |
| 655 | bool CheckUncompressedLength(const std::string& compressed, size_t* ulength) { |
| 656 | const bool result1 = snappy::GetUncompressedLength(compressed.data(), |
| 657 | compressed.size(), |
| 658 | ulength); |
| 659 | |
| 660 | snappy::ByteArraySource source(compressed.data(), compressed.size()); |
| 661 | uint32_t length; |
| 662 | const bool result2 = snappy::GetUncompressedLength(&source, &length); |
| 663 | CHECK_EQ(result1, result2); |
| 664 | return result1; |
| 665 | } |
| 666 | |
| 667 | TEST(SnappyCorruption, TruncatedVarint) { |
| 668 | std::string compressed, uncompressed; |
| 669 | size_t ulength; |
| 670 | compressed.push_back('\xf0'); |
| 671 | CHECK(!CheckUncompressedLength(compressed, &ulength)); |
| 672 | CHECK(!snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 673 | CHECK(!snappy::Uncompress(compressed.data(), compressed.size(), |
| 674 | &uncompressed)); |
| 675 | } |
| 676 | |
| 677 | TEST(SnappyCorruption, UnterminatedVarint) { |
| 678 | std::string compressed, uncompressed; |
| 679 | size_t ulength; |
| 680 | compressed.push_back('\x80'); |
| 681 | compressed.push_back('\x80'); |
| 682 | compressed.push_back('\x80'); |
| 683 | compressed.push_back('\x80'); |
| 684 | compressed.push_back('\x80'); |
| 685 | compressed.push_back(10); |
| 686 | CHECK(!CheckUncompressedLength(compressed, &ulength)); |
| 687 | CHECK(!snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 688 | CHECK(!snappy::Uncompress(compressed.data(), compressed.size(), |
| 689 | &uncompressed)); |
| 690 | } |
| 691 | |
| 692 | TEST(SnappyCorruption, OverflowingVarint) { |
| 693 | std::string compressed, uncompressed; |
| 694 | size_t ulength; |
| 695 | compressed.push_back('\xfb'); |
| 696 | compressed.push_back('\xff'); |
| 697 | compressed.push_back('\xff'); |
| 698 | compressed.push_back('\xff'); |
| 699 | compressed.push_back('\x7f'); |
| 700 | CHECK(!CheckUncompressedLength(compressed, &ulength)); |
| 701 | CHECK(!snappy::IsValidCompressedBuffer(compressed.data(), compressed.size())); |
| 702 | CHECK(!snappy::Uncompress(compressed.data(), compressed.size(), |
| 703 | &uncompressed)); |
| 704 | } |
| 705 | |
| 706 | TEST(Snappy, ReadPastEndOfBuffer) { |
| 707 | // Check that we do not read past end of input |
| 708 | |
| 709 | // Make a compressed string that ends with a single-byte literal |
| 710 | std::string compressed; |
| 711 | Varint::Append32(&compressed, 1); |
| 712 | AppendLiteral(&compressed, "x"); |
| 713 | |
| 714 | std::string uncompressed; |
| 715 | DataEndingAtUnreadablePage c(compressed); |
| 716 | CHECK(snappy::Uncompress(c.data(), c.size(), &uncompressed)); |
| 717 | CHECK_EQ(uncompressed, std::string("x")); |
| 718 | } |
| 719 | |
| 720 | // Check for an infinite loop caused by a copy with offset==0 |
| 721 | TEST(Snappy, ZeroOffsetCopy) { |
| 722 | const char* compressed = "\x40\x12\x00\x00"; |
| 723 | // \x40 Length (must be > kMaxIncrementCopyOverflow) |
| 724 | // \x12\x00\x00 Copy with offset==0, length==5 |
| 725 | char uncompressed[100]; |
| 726 | EXPECT_FALSE(snappy::RawUncompress(compressed, 4, uncompressed)); |
| 727 | } |
| 728 | |
| 729 | TEST(Snappy, ZeroOffsetCopyValidation) { |
| 730 | const char* compressed = "\x05\x12\x00\x00"; |
| 731 | // \x05 Length |
| 732 | // \x12\x00\x00 Copy with offset==0, length==5 |
| 733 | EXPECT_FALSE(snappy::IsValidCompressedBuffer(compressed, 4)); |
| 734 | } |
| 735 | |
| 736 | int TestFindMatchLength(const char* s1, const char *s2, unsigned length) { |
| 737 | uint64_t data; |
| 738 | std::pair<size_t, bool> p = |
| 739 | snappy::internal::FindMatchLength(s1, s2, s2 + length, &data); |
| 740 | CHECK_EQ(p.first < 8, p.second); |
| 741 | return p.first; |
| 742 | } |
| 743 | |
| 744 | TEST(Snappy, FindMatchLength) { |
| 745 | // Exercise all different code paths through the function. |
| 746 | // 64-bit version: |
| 747 | |
| 748 | // Hit s1_limit in 64-bit loop, hit s1_limit in single-character loop. |
| 749 | EXPECT_EQ(6, TestFindMatchLength("012345", "012345", 6)); |
| 750 | EXPECT_EQ(11, TestFindMatchLength("01234567abc", "01234567abc", 11)); |
| 751 | |
| 752 | // Hit s1_limit in 64-bit loop, find a non-match in single-character loop. |
| 753 | EXPECT_EQ(9, TestFindMatchLength("01234567abc", "01234567axc", 9)); |
| 754 | |
| 755 | // Same, but edge cases. |
| 756 | EXPECT_EQ(11, TestFindMatchLength("01234567abc!", "01234567abc!", 11)); |
| 757 | EXPECT_EQ(11, TestFindMatchLength("01234567abc!", "01234567abc?", 11)); |
| 758 | |
| 759 | // Find non-match at once in first loop. |
| 760 | EXPECT_EQ(0, TestFindMatchLength("01234567xxxxxxxx", "?1234567xxxxxxxx", 16)); |
| 761 | EXPECT_EQ(1, TestFindMatchLength("01234567xxxxxxxx", "0?234567xxxxxxxx", 16)); |
| 762 | EXPECT_EQ(4, TestFindMatchLength("01234567xxxxxxxx", "01237654xxxxxxxx", 16)); |
| 763 | EXPECT_EQ(7, TestFindMatchLength("01234567xxxxxxxx", "0123456?xxxxxxxx", 16)); |
| 764 | |
| 765 | // Find non-match in first loop after one block. |
| 766 | EXPECT_EQ(8, TestFindMatchLength("abcdefgh01234567xxxxxxxx", |
| 767 | "abcdefgh?1234567xxxxxxxx", 24)); |
| 768 | EXPECT_EQ(9, TestFindMatchLength("abcdefgh01234567xxxxxxxx", |
| 769 | "abcdefgh0?234567xxxxxxxx", 24)); |
| 770 | EXPECT_EQ(12, TestFindMatchLength("abcdefgh01234567xxxxxxxx", |
| 771 | "abcdefgh01237654xxxxxxxx", 24)); |
| 772 | EXPECT_EQ(15, TestFindMatchLength("abcdefgh01234567xxxxxxxx", |
| 773 | "abcdefgh0123456?xxxxxxxx", 24)); |
| 774 | |
| 775 | // 32-bit version: |
| 776 | |
| 777 | // Short matches. |
| 778 | EXPECT_EQ(0, TestFindMatchLength("01234567", "?1234567", 8)); |
| 779 | EXPECT_EQ(1, TestFindMatchLength("01234567", "0?234567", 8)); |
| 780 | EXPECT_EQ(2, TestFindMatchLength("01234567", "01?34567", 8)); |
| 781 | EXPECT_EQ(3, TestFindMatchLength("01234567", "012?4567", 8)); |
| 782 | EXPECT_EQ(4, TestFindMatchLength("01234567", "0123?567", 8)); |
| 783 | EXPECT_EQ(5, TestFindMatchLength("01234567", "01234?67", 8)); |
| 784 | EXPECT_EQ(6, TestFindMatchLength("01234567", "012345?7", 8)); |
| 785 | EXPECT_EQ(7, TestFindMatchLength("01234567", "0123456?", 8)); |
| 786 | EXPECT_EQ(7, TestFindMatchLength("01234567", "0123456?", 7)); |
| 787 | EXPECT_EQ(7, TestFindMatchLength("01234567!", "0123456??", 7)); |
| 788 | |
| 789 | // Hit s1_limit in 32-bit loop, hit s1_limit in single-character loop. |
| 790 | EXPECT_EQ(10, TestFindMatchLength("xxxxxxabcd", "xxxxxxabcd", 10)); |
| 791 | EXPECT_EQ(10, TestFindMatchLength("xxxxxxabcd?", "xxxxxxabcd?", 10)); |
| 792 | EXPECT_EQ(13, TestFindMatchLength("xxxxxxabcdef", "xxxxxxabcdef", 13)); |
| 793 | |
| 794 | // Same, but edge cases. |
| 795 | EXPECT_EQ(12, TestFindMatchLength("xxxxxx0123abc!", "xxxxxx0123abc!", 12)); |
| 796 | EXPECT_EQ(12, TestFindMatchLength("xxxxxx0123abc!", "xxxxxx0123abc?", 12)); |
| 797 | |
| 798 | // Hit s1_limit in 32-bit loop, find a non-match in single-character loop. |
| 799 | EXPECT_EQ(11, TestFindMatchLength("xxxxxx0123abc", "xxxxxx0123axc", 13)); |
| 800 | |
| 801 | // Find non-match at once in first loop. |
| 802 | EXPECT_EQ(6, TestFindMatchLength("xxxxxx0123xxxxxxxx", |
| 803 | "xxxxxx?123xxxxxxxx", 18)); |
| 804 | EXPECT_EQ(7, TestFindMatchLength("xxxxxx0123xxxxxxxx", |
| 805 | "xxxxxx0?23xxxxxxxx", 18)); |
| 806 | EXPECT_EQ(8, TestFindMatchLength("xxxxxx0123xxxxxxxx", |
| 807 | "xxxxxx0132xxxxxxxx", 18)); |
| 808 | EXPECT_EQ(9, TestFindMatchLength("xxxxxx0123xxxxxxxx", |
| 809 | "xxxxxx012?xxxxxxxx", 18)); |
| 810 | |
| 811 | // Same, but edge cases. |
| 812 | EXPECT_EQ(6, TestFindMatchLength("xxxxxx0123", "xxxxxx?123", 10)); |
| 813 | EXPECT_EQ(7, TestFindMatchLength("xxxxxx0123", "xxxxxx0?23", 10)); |
| 814 | EXPECT_EQ(8, TestFindMatchLength("xxxxxx0123", "xxxxxx0132", 10)); |
| 815 | EXPECT_EQ(9, TestFindMatchLength("xxxxxx0123", "xxxxxx012?", 10)); |
| 816 | |
| 817 | // Find non-match in first loop after one block. |
| 818 | EXPECT_EQ(10, TestFindMatchLength("xxxxxxabcd0123xx", |
| 819 | "xxxxxxabcd?123xx", 16)); |
| 820 | EXPECT_EQ(11, TestFindMatchLength("xxxxxxabcd0123xx", |
| 821 | "xxxxxxabcd0?23xx", 16)); |
| 822 | EXPECT_EQ(12, TestFindMatchLength("xxxxxxabcd0123xx", |
| 823 | "xxxxxxabcd0132xx", 16)); |
| 824 | EXPECT_EQ(13, TestFindMatchLength("xxxxxxabcd0123xx", |
| 825 | "xxxxxxabcd012?xx", 16)); |
| 826 | |
| 827 | // Same, but edge cases. |
| 828 | EXPECT_EQ(10, TestFindMatchLength("xxxxxxabcd0123", "xxxxxxabcd?123", 14)); |
| 829 | EXPECT_EQ(11, TestFindMatchLength("xxxxxxabcd0123", "xxxxxxabcd0?23", 14)); |
| 830 | EXPECT_EQ(12, TestFindMatchLength("xxxxxxabcd0123", "xxxxxxabcd0132", 14)); |
| 831 | EXPECT_EQ(13, TestFindMatchLength("xxxxxxabcd0123", "xxxxxxabcd012?", 14)); |
| 832 | } |
| 833 | |
| 834 | TEST(Snappy, FindMatchLengthRandom) { |
| 835 | constexpr int kNumTrials = 10000; |
| 836 | constexpr int kTypicalLength = 10; |
| 837 | std::minstd_rand0 rng(snappy::GetFlag(FLAGS_test_random_seed)); |
| 838 | std::uniform_int_distribution<int> uniform_byte(0, 255); |
| 839 | std::bernoulli_distribution one_in_two(1.0 / 2); |
| 840 | std::bernoulli_distribution one_in_typical_length(1.0 / kTypicalLength); |
| 841 | |
| 842 | for (int i = 0; i < kNumTrials; ++i) { |
| 843 | std::string s, t; |
| 844 | char a = static_cast<char>(uniform_byte(rng)); |
| 845 | char b = static_cast<char>(uniform_byte(rng)); |
| 846 | while (!one_in_typical_length(rng)) { |
| 847 | s.push_back(one_in_two(rng) ? a : b); |
| 848 | t.push_back(one_in_two(rng) ? a : b); |
| 849 | } |
| 850 | DataEndingAtUnreadablePage u(s); |
| 851 | DataEndingAtUnreadablePage v(t); |
| 852 | size_t matched = TestFindMatchLength(u.data(), v.data(), t.size()); |
| 853 | if (matched == t.size()) { |
| 854 | EXPECT_EQ(s, t); |
| 855 | } else { |
| 856 | EXPECT_NE(s[matched], t[matched]); |
| 857 | for (size_t j = 0; j < matched; ++j) { |
| 858 | EXPECT_EQ(s[j], t[j]); |
| 859 | } |
| 860 | } |
| 861 | } |
| 862 | } |
| 863 | |
| 864 | uint16_t MakeEntry(unsigned int extra, unsigned int len, |
| 865 | unsigned int copy_offset) { |
| 866 | // Check that all of the fields fit within the allocated space |
| 867 | assert(extra == (extra & 0x7)); // At most 3 bits |
| 868 | assert(copy_offset == (copy_offset & 0x7)); // At most 3 bits |
| 869 | assert(len == (len & 0x7f)); // At most 7 bits |
| 870 | return len | (copy_offset << 8) | (extra << 11); |
| 871 | } |
| 872 | |
| 873 | // Check that the decompression table is correct, and optionally print out |
| 874 | // the computed one. |
| 875 | TEST(Snappy, VerifyCharTable) { |
| 876 | using snappy::internal::LITERAL; |
| 877 | using snappy::internal::COPY_1_BYTE_OFFSET; |
| 878 | using snappy::internal::COPY_2_BYTE_OFFSET; |
| 879 | using snappy::internal::COPY_4_BYTE_OFFSET; |
| 880 | using snappy::internal::char_table; |
| 881 | |
| 882 | uint16_t dst[256]; |
| 883 | |
| 884 | // Place invalid entries in all places to detect missing initialization |
| 885 | int assigned = 0; |
| 886 | for (int i = 0; i < 256; ++i) { |
| 887 | dst[i] = 0xffff; |
| 888 | } |
| 889 | |
| 890 | // Small LITERAL entries. We store (len-1) in the top 6 bits. |
| 891 | for (uint8_t len = 1; len <= 60; ++len) { |
| 892 | dst[LITERAL | ((len - 1) << 2)] = MakeEntry(0, len, 0); |
| 893 | assigned++; |
| 894 | } |
| 895 | |
| 896 | // Large LITERAL entries. We use 60..63 in the high 6 bits to |
| 897 | // encode the number of bytes of length info that follow the opcode. |
| 898 | for (uint8_t extra_bytes = 1; extra_bytes <= 4; ++extra_bytes) { |
| 899 | // We set the length field in the lookup table to 1 because extra |
| 900 | // bytes encode len-1. |
| 901 | dst[LITERAL | ((extra_bytes + 59) << 2)] = MakeEntry(extra_bytes, 1, 0); |
| 902 | assigned++; |
| 903 | } |
| 904 | |
| 905 | // COPY_1_BYTE_OFFSET. |
| 906 | // |
| 907 | // The tag byte in the compressed data stores len-4 in 3 bits, and |
| 908 | // offset/256 in 3 bits. offset%256 is stored in the next byte. |
| 909 | // |
| 910 | // This format is used for length in range [4..11] and offset in |
| 911 | // range [0..2047] |
| 912 | for (uint8_t len = 4; len < 12; ++len) { |
| 913 | for (uint16_t offset = 0; offset < 2048; offset += 256) { |
| 914 | uint8_t offset_high = static_cast<uint8_t>(offset >> 8); |
| 915 | dst[COPY_1_BYTE_OFFSET | ((len - 4) << 2) | (offset_high << 5)] = |
| 916 | MakeEntry(1, len, offset_high); |
| 917 | assigned++; |
| 918 | } |
| 919 | } |
| 920 | |
| 921 | // COPY_2_BYTE_OFFSET. |
| 922 | // Tag contains len-1 in top 6 bits, and offset in next two bytes. |
| 923 | for (uint8_t len = 1; len <= 64; ++len) { |
| 924 | dst[COPY_2_BYTE_OFFSET | ((len - 1) << 2)] = MakeEntry(2, len, 0); |
| 925 | assigned++; |
| 926 | } |
| 927 | |
| 928 | // COPY_4_BYTE_OFFSET. |
| 929 | // Tag contents len-1 in top 6 bits, and offset in next four bytes. |
| 930 | for (uint8_t len = 1; len <= 64; ++len) { |
| 931 | dst[COPY_4_BYTE_OFFSET | ((len - 1) << 2)] = MakeEntry(4, len, 0); |
| 932 | assigned++; |
| 933 | } |
| 934 | |
| 935 | // Check that each entry was initialized exactly once. |
| 936 | EXPECT_EQ(256, assigned) << "Assigned only " << assigned << " of 256"; |
| 937 | for (int i = 0; i < 256; ++i) { |
| 938 | EXPECT_NE(0xffff, dst[i]) << "Did not assign byte " << i; |
| 939 | } |
| 940 | |
| 941 | if (snappy::GetFlag(FLAGS_snappy_dump_decompression_table)) { |
| 942 | std::printf("static const uint16_t char_table[256] = {\n "); |
| 943 | for (int i = 0; i < 256; ++i) { |
| 944 | std::printf("0x%04x%s", |
| 945 | dst[i], |
| 946 | ((i == 255) ? "\n" : (((i % 8) == 7) ? ",\n " : ", "))); |
| 947 | } |
| 948 | std::printf("};\n"); |
| 949 | } |
| 950 | |
| 951 | // Check that computed table matched recorded table. |
| 952 | for (int i = 0; i < 256; ++i) { |
| 953 | EXPECT_EQ(dst[i], char_table[i]) << "Mismatch in byte " << i; |
| 954 | } |
| 955 | } |
| 956 | |
| 957 | TEST(Snappy, TestBenchmarkFiles) { |
| 958 | for (int i = 0; i < ARRAYSIZE(kTestDataFiles); ++i) { |
| 959 | Verify(ReadTestDataFile(kTestDataFiles[i].filename, |
| 960 | kTestDataFiles[i].size_limit)); |
| 961 | } |
| 962 | } |
| 963 | |
| 964 | } // namespace |
| 965 | |
| 966 | } // namespace snappy |