snappy-stubs-internal.h 13.7 KB
Newer Older
gejun's avatar
gejun committed
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
// Copyright 2011 Google Inc. All Rights Reserved.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
//     * Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//     * Redistributions in binary form must reproduce the above
// copyright notice, this list of conditions and the following disclaimer
// in the documentation and/or other materials provided with the
// distribution.
//     * Neither the name of Google Inc. nor the names of its
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Various stubs for the open-source version of Snappy.

gejun's avatar
gejun committed
31 32
#ifndef BUTIL_THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_
#define BUTIL_THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_
gejun's avatar
gejun committed
33 34 35 36 37

#include <string>
#include <assert.h>
#include <stdlib.h>
#include <string.h>
38 39 40
#include "butil/compiler_specific.h"
#include "butil/basictypes.h"
#include "butil/sys_byteorder.h"
gejun's avatar
gejun committed
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

#define SNAPPY_MAJOR 1
#define SNAPPY_MINOR 1
#define SNAPPY_PATCHLEVEL 3
#define SNAPPY_VERSION                                                  \
    ((SNAPPY_MAJOR << 16) | (SNAPPY_MINOR << 8) | SNAPPY_PATCHLEVEL)

#if defined(__x86_64__)

// Enable 64-bit optimized versions of some routines.
#define ARCH_K8 1

#endif

// Needed by OS X, among others.
#ifndef MAP_ANONYMOUS
#define MAP_ANONYMOUS MAP_ANON
#endif

// This is only used for recomputing the tag byte table used during
// decompression; for simplicity we just remove it from the open-source
// version (anyone who wants to regenerate it can just do the call
// themselves within main()).
#define DEFINE_bool(flag_name, default_value, description)      \
    bool FLAGS_ ## flag_name = default_value
#define DECLARE_bool(flag_name)                 \
    extern bool FLAGS_ ## flag_name

69
namespace butil {
gejun's avatar
gejun committed
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
namespace snappy {

// x86 and PowerPC can simply do these loads and stores native.

#if defined(__i386__) || defined(__x86_64__) || defined(__powerpc__)

#define UNALIGNED_LOAD16(_p) (*reinterpret_cast<const uint16_t *>(_p))
#define UNALIGNED_LOAD32(_p) (*reinterpret_cast<const uint32_t *>(_p))
#define UNALIGNED_LOAD64(_p) (*reinterpret_cast<const uint64_t *>(_p))

#define UNALIGNED_STORE16(_p, _val) (*reinterpret_cast<uint16_t *>(_p) = (_val))
#define UNALIGNED_STORE32(_p, _val) (*reinterpret_cast<uint32_t *>(_p) = (_val))
#define UNALIGNED_STORE64(_p, _val) (*reinterpret_cast<uint64_t *>(_p) = (_val))

// ARMv7 and newer support native unaligned accesses, but only of 16-bit
// and 32-bit values (not 64-bit); older versions either raise a fatal signal,
// do an unaligned read and rotate the words around a bit, or do the reads very
// slowly (trip through kernel mode). There's no simple #define that says just
// “ARMv7 or higher”, so we have to filter away all ARMv5 and ARMv6
// sub-architectures.
//
// This is a mess, but there's not much we can do about it.

#elif defined(__arm__) &&                       \
    !defined(__ARM_ARCH_4__) &&                 \
    !defined(__ARM_ARCH_4T__) &&                \
    !defined(__ARM_ARCH_5__) &&                 \
    !defined(__ARM_ARCH_5T__) &&                \
    !defined(__ARM_ARCH_5TE__) &&               \
    !defined(__ARM_ARCH_5TEJ__) &&              \
    !defined(__ARM_ARCH_6__) &&                 \
    !defined(__ARM_ARCH_6J__) &&                \
    !defined(__ARM_ARCH_6K__) &&                \
    !defined(__ARM_ARCH_6Z__) &&                \
    !defined(__ARM_ARCH_6ZK__) &&               \
    !defined(__ARM_ARCH_6T2__)

#define UNALIGNED_LOAD16(_p) (*reinterpret_cast<const uint16_t *>(_p))
#define UNALIGNED_LOAD32(_p) (*reinterpret_cast<const uint32_t *>(_p))

#define UNALIGNED_STORE16(_p, _val) (*reinterpret_cast<uint16_t *>(_p) = (_val))
#define UNALIGNED_STORE32(_p, _val) (*reinterpret_cast<uint32_t *>(_p) = (_val))

// TODO(user): NEON supports unaligned 64-bit loads and stores.
// See if that would be more efficient on platforms supporting it,
// at least for copies.

inline uint64_tUNALIGNED_LOAD64(const void *p) {
    uint64_tt;
    memcpy(&t, p, sizeof t);
    return t;
}

inline void UNALIGNED_STORE64(void *p, uint64_tv) {
    memcpy(p, &v, sizeof v);
}

#else

// These functions are provided for architectures that don't support
// unaligned loads and stores.

inline uint16_t UNALIGNED_LOAD16(const void *p) {
    uint16_t t;
    memcpy(&t, p, sizeof t);
    return t;
}

inline uint32_t UNALIGNED_LOAD32(const void *p) {
    uint32_t t;
    memcpy(&t, p, sizeof t);
    return t;
}

inline uint64_tUNALIGNED_LOAD64(const void *p) {
    uint64_tt;
    memcpy(&t, p, sizeof t);
    return t;
}

inline void UNALIGNED_STORE16(void *p, uint16_t v) {
    memcpy(p, &v, sizeof v);
}

inline void UNALIGNED_STORE32(void *p, uint32_t v) {
    memcpy(p, &v, sizeof v);
}

inline void UNALIGNED_STORE64(void *p, uint64_tv) {
    memcpy(p, &v, sizeof v);
}

#endif

// This can be more efficient than UNALIGNED_LOAD64 + UNALIGNED_STORE64
// on some platforms, in particular ARM.
inline void UnalignedCopy64(const void *src, void *dst) {
    if (sizeof(void *) == 8) {
        UNALIGNED_STORE64(dst, UNALIGNED_LOAD64(src));
    } else {
        const char *src_char = reinterpret_cast<const char *>(src);
        char *dst_char = reinterpret_cast<char *>(dst);

        UNALIGNED_STORE32(dst_char, UNALIGNED_LOAD32(src_char));
        UNALIGNED_STORE32(dst_char + 4, UNALIGNED_LOAD32(src_char + 4));
    }
}

// Convert to little-endian storage, opposite of network format.
// Convert x from host to little endian: x = LittleEndian.FromHost(x);
// convert x from little endian to host: x = LittleEndian.ToHost(x);
//
//  Store values into unaligned memory converting to little endian order:
//    LittleEndian.Store16(p, x);
//
//  Load unaligned values stored in little endian converting to host order:
//    x = LittleEndian.Load16(p);
class LittleEndian {
public:
    // Conversion functions.
#if defined(ARCH_CPU_LITTLE_ENDIAN)
    static uint16_t FromHost16(uint16_t x) { return x; }
    static uint16_t ToHost16(uint16_t x) { return x; }

    static uint32_t FromHost32(uint32_t x) { return x; }
    static uint32_t ToHost32(uint32_t x) { return x; }

    static bool IsLittleEndian() { return true; }

#else  // !defined(ARCH_CPU_LITTLE_ENDIAN)
    static uint16_t FromHost16(uint16_t x) { return ByteSwap(x); }
    static uint16_t ToHost16(uint16_t x) { return ByteSwap(x); }

    static uint32_t FromHost32(uint32_t x) { return ByteSwap(x); }
    static uint32_t ToHost32(uint32_t x) { return ByteSwap(x); }

    static bool IsLittleEndian() { return false; }
#endif  // !defined(ARCH_CPU_LITTLE_ENDIAN)

    // Functions to do unaligned loads and stores in little-endian order.
    static uint16_t Load16(const void *p) {
        return ToHost16(UNALIGNED_LOAD16(p));
    }

    static void Store16(void *p, uint16_t v) {
        UNALIGNED_STORE16(p, FromHost16(v));
    }

    static uint32_t Load32(const void *p) {
        return ToHost32(UNALIGNED_LOAD32(p));
    }

    static void Store32(void *p, uint32_t v) {
        UNALIGNED_STORE32(p, FromHost32(v));
    }
};

// Some bit-manipulation functions.
class Bits {
public:
    // Return floor(log2(n)) for positive integer n.  Returns -1 iff n == 0.
    static int Log2Floor(uint32_t n);

    // Return the first set least / most significant bit, 0-indexed.  Returns an
    // undefined value if n == 0.  FindLSBSetNonZero() is similar to ffs() except
    // that it's 0-indexed.
    static int FindLSBSetNonZero(uint32_t n);
    static int FindLSBSetNonZero64(uint64_t n);

private:
    DISALLOW_COPY_AND_ASSIGN(Bits);
};

#if defined(COMPILER_GCC)

inline int Bits::Log2Floor(uint32_t n) {
    return n == 0 ? -1 : 31 ^ __builtin_clz(n);
}

inline int Bits::FindLSBSetNonZero(uint32_t n) {
    return __builtin_ctz(n);
}

inline int Bits::FindLSBSetNonZero64(uint64_t n) {
    return __builtin_ctzll(n);
}

#else  // Portable versions.

inline int Bits::Log2Floor(uint32_t n) {
    if (n == 0)
        return -1;
    int log = 0;
    uint32_t value = n;
    for (int i = 4; i >= 0; --i) {
        int shift = (1 << i);
        uint32_t x = value >> shift;
        if (x != 0) {
            value = x;
            log += shift;
        }
    }
    assert(value == 1);
    return log;
}

inline int Bits::FindLSBSetNonZero(uint32_t n) {
    int rc = 31;
    for (int i = 4, shift = 1 << 4; i >= 0; --i) {
        const uint32_t x = n << shift;
        if (x != 0) {
            n = x;
            rc -= shift;
        }
        shift >>= 1;
    }
    return rc;
}

// FindLSBSetNonZero64() is defined in terms of FindLSBSetNonZero().
inline int Bits::FindLSBSetNonZero64(uint64_t n) {
    const uint32_t bottombits = static_cast<uint32_t>(n);
    if (bottombits == 0) {
        // Bottom bits are zero, so scan in top bits
        return 32 + FindLSBSetNonZero(static_cast<uint32_t>(n >> 32));
    } else {
        return FindLSBSetNonZero(bottombits);
    }
}

#endif  // End portable versions.

// Variable-length integer encoding.
class Varint {
public:
    // Maximum lengths of varint encoding of uint32.
    static const int kMax32 = 5;

    // Attempts to parse a varint32 from a prefix of the bytes in [ptr,limit-1].
    // Never reads a character at or beyond limit.  If a valid/terminated varint32
    // was found in the range, stores it in *OUTPUT and returns a pointer just
    // past the last byte of the varint32. Else returns NULL.  On success,
    // "result <= limit".
    static const char* Parse32WithLimit(const char* ptr, const char* limit,
                                        uint32_t* OUTPUT);

    // REQUIRES   "ptr" points to a buffer of length sufficient to hold "v".
    // EFFECTS    Encodes "v" into "ptr" and returns a pointer to the
    //            byte just past the last encoded byte.
    static char* Encode32(char* ptr, uint32_t v);

    // EFFECTS    Appends the varint representation of "value" to "*s".
    static void Append32(std::string* s, uint32_t value);
};

inline const char* Varint::Parse32WithLimit(const char* p,
                                            const char* l,
                                            uint32_t* OUTPUT) {
    const unsigned char* ptr = reinterpret_cast<const unsigned char*>(p);
    const unsigned char* limit = reinterpret_cast<const unsigned char*>(l);
    uint32_t b, result;
    if (ptr >= limit) return NULL;
    b = *(ptr++); result = b & 127;          if (b < 128) goto done;
    if (ptr >= limit) return NULL;
    b = *(ptr++); result |= (b & 127) <<  7; if (b < 128) goto done;
    if (ptr >= limit) return NULL;
    b = *(ptr++); result |= (b & 127) << 14; if (b < 128) goto done;
    if (ptr >= limit) return NULL;
    b = *(ptr++); result |= (b & 127) << 21; if (b < 128) goto done;
    if (ptr >= limit) return NULL;
    b = *(ptr++); result |= (b & 127) << 28; if (b < 16) goto done;
    return NULL;       // Value is too long to be a varint32
done:
    *OUTPUT = result;
    return reinterpret_cast<const char*>(ptr);
}

inline char* Varint::Encode32(char* sptr, uint32_t v) {
    // Operate on characters as unsigneds
    unsigned char* ptr = reinterpret_cast<unsigned char*>(sptr);
    static const int B = 128;
    if (v < (1<<7)) {
        *(ptr++) = v;
    } else if (v < (1<<14)) {
        *(ptr++) = v | B;
        *(ptr++) = v>>7;
    } else if (v < (1<<21)) {
        *(ptr++) = v | B;
        *(ptr++) = (v>>7) | B;
        *(ptr++) = v>>14;
    } else if (v < (1<<28)) {
        *(ptr++) = v | B;
        *(ptr++) = (v>>7) | B;
        *(ptr++) = (v>>14) | B;
        *(ptr++) = v>>21;
    } else {
        *(ptr++) = v | B;
        *(ptr++) = (v>>7) | B;
        *(ptr++) = (v>>14) | B;
        *(ptr++) = (v>>21) | B;
        *(ptr++) = v>>28;
    }
    return reinterpret_cast<char*>(ptr);
}

// If you know the internal layout of the std::string in use, you can
// replace this function with one that resizes the string without
// filling the new space with zeros (if applicable) --
// it will be non-portable but faster.
inline void STLStringResizeUninitialized(std::string* s, size_t new_size) {
    s->resize(new_size);
}

// Return a mutable char* pointing to a string's internal buffer,
// which may not be null-terminated. Writing through this pointer will
// modify the string.
//
// string_as_array(&str)[i] is valid for 0 <= i < str.size() until the
// next call to a string method that invalidates iterators.
//
// As of 2006-04, there is no standard-blessed way of getting a
// mutable reference to a string's internal buffer. However, issue 530
// (http://www.open-std.org/JTC1/SC22/WG21/docs/lwg-defects.html#530)
// proposes this as the method. It will officially be part of the standard
// for C++0x. This should already work on all current implementations.
inline char* string_as_array(std::string* str) {
    return str->empty() ? NULL : &*str->begin();
}

}  // namespace snappy
400
}  // namespace butil
gejun's avatar
gejun committed
401

gejun's avatar
gejun committed
402
#endif  // BUTIL_THIRD_PARTY_SNAPPY_OPENSOURCE_SNAPPY_STUBS_INTERNAL_H_