dolphin/Source/Core/Common/Hash.cpp

// Copyright 2008 Dolphin Emulator Project
// Licensed under GPLv2+
// Refer to the license.txt file included.

#include "Common/Hash.h"
#include <algorithm>
#include <cstring>
#include "Common/CPUDetect.h"
#include "Common/CommonFuncs.h"
#include "Common/Intrinsics.h"

static u64 (*ptrHashFunction)(const u8* src, u32 len, u32 samples) = &GetMurmurHash3;

// uint32_t
// WARNING - may read one more byte!
// Implementation from Wikipedia.
u32 HashFletcher(const u8* data_u8, size_t length)
{
  const u16* data = (const u16*)data_u8; /* Pointer to the data to be summed */
  size_t len = (length + 1) / 2;         /* Length in 16-bit words */
  u32 sum1 = 0xffff, sum2 = 0xffff;

  while (len)
  {
    size_t tlen = len > 360 ? 360 : len;
    len -= tlen;

    do
    {
      sum1 += *data++;
      sum2 += sum1;
    } while (--tlen);

    sum1 = (sum1 & 0xffff) + (sum1 >> 16);
    sum2 = (sum2 & 0xffff) + (sum2 >> 16);
  }

  // Second reduction step to reduce sums to 16 bits
  sum1 = (sum1 & 0xffff) + (sum1 >> 16);
  sum2 = (sum2 & 0xffff) + (sum2 >> 16);
  return (sum2 << 16 | sum1);
}

// Implementation from Wikipedia
// Slightly slower than Fletcher above, but slightly more reliable.
// data: Pointer to the data to be summed; len is in bytes
u32 HashAdler32(const u8* data, size_t len)
{
  static const u32 MOD_ADLER = 65521;
  u32 a = 1, b = 0;

  while (len)
  {
    size_t tlen = len > 5550 ? 5550 : len;
    len -= tlen;

    do
    {
      a += *data++;
      b += a;
    } while (--tlen);

    a = (a & 0xffff) + (a >> 16) * (65536 - MOD_ADLER);
    b = (b & 0xffff) + (b >> 16) * (65536 - MOD_ADLER);
  }

  // It can be shown that a <= 0x1013a here, so a single subtract will do.
  if (a >= MOD_ADLER)
  {
    a -= MOD_ADLER;
  }

  // It can be shown that b can reach 0xfff87 here.
  b = (b & 0xffff) + (b >> 16) * (65536 - MOD_ADLER);

  if (b >= MOD_ADLER)
  {
    b -= MOD_ADLER;
  }

  return ((b << 16) | a);
}

// Stupid hash - but can't go back now :)
// Don't use for new things. At least it's reasonably fast.
u32 HashEctor(const u8* ptr, int length)
{
  u32 crc = 0;

  for (int i = 0; i < length; i++)
  {
    crc ^= ptr[i];
    crc = (crc << 3) | (crc >> 29);
  }

  return (crc);
}

#if _ARCH_64

//-----------------------------------------------------------------------------
// Block read - if your platform needs to do endian-swapping or can only
// handle aligned reads, do the conversion here

inline u64 getblock(const u64* p, int i)
{
  return p[i];
}

//----------
// Block mix - combine the key bits with the hash bits and scramble everything

inline void bmix64(u64& h1, u64& h2, u64& k1, u64& k2, u64& c1, u64& c2)
{
  k1 *= c1;
  k1 = _rotl64(k1, 23);
  k1 *= c2;
  h1 ^= k1;
  h1 += h2;

  h2 = _rotl64(h2, 41);

  k2 *= c2;
  k2 = _rotl64(k2, 23);
  k2 *= c1;
  h2 ^= k2;
  h2 += h1;

  h1 = h1 * 3 + 0x52dce729;
  h2 = h2 * 3 + 0x38495ab5;

  c1 = c1 * 5 + 0x7b7d159c;
  c2 = c2 * 5 + 0x6bce6396;
}

//----------
// Finalization mix - avalanches all bits to within 0.05% bias

inline u64 fmix64(u64 k)
{
  k ^= k >> 33;
  k *= 0xff51afd7ed558ccd;
  k ^= k >> 33;
  k *= 0xc4ceb9fe1a85ec53;
  k ^= k >> 33;

  return k;
}

u64 GetMurmurHash3(const u8* src, u32 len, u32 samples)
{
  const u8* data = (const u8*)src;
  const int nblocks = len / 16;
  u32 Step = (len / 8);
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;

  u64 h1 = 0x9368e53c2f6af274;
  u64 h2 = 0x586dcd208f7cd3fd;

  u64 c1 = 0x87c37b91114253d5;
  u64 c2 = 0x4cf5ad432745937f;

  //----------
  // body

  const u64* blocks = (const u64*)(data);

  for (int i = 0; i < nblocks; i += Step)
  {
    u64 k1 = getblock(blocks, i * 2 + 0);
    u64 k2 = getblock(blocks, i * 2 + 1);

    bmix64(h1, h2, k1, k2, c1, c2);
  }

  //----------
  // tail

  const u8* tail = (const u8*)(data + nblocks * 16);

  u64 k1 = 0;
  u64 k2 = 0;

  switch (len & 15)
  {
  case 15:
    k2 ^= u64(tail[14]) << 48;
  case 14:
    k2 ^= u64(tail[13]) << 40;
  case 13:
    k2 ^= u64(tail[12]) << 32;
  case 12:
    k2 ^= u64(tail[11]) << 24;
  case 11:
    k2 ^= u64(tail[10]) << 16;
  case 10:
    k2 ^= u64(tail[9]) << 8;
  case 9:
    k2 ^= u64(tail[8]) << 0;

  case 8:
    k1 ^= u64(tail[7]) << 56;
  case 7:
    k1 ^= u64(tail[6]) << 48;
  case 6:
    k1 ^= u64(tail[5]) << 40;
  case 5:
    k1 ^= u64(tail[4]) << 32;
  case 4:
    k1 ^= u64(tail[3]) << 24;
  case 3:
    k1 ^= u64(tail[2]) << 16;
  case 2:
    k1 ^= u64(tail[1]) << 8;
  case 1:
    k1 ^= u64(tail[0]) << 0;
    bmix64(h1, h2, k1, k2, c1, c2);
  };

  //----------
  // finalization

  h2 ^= len;

  h1 += h2;
  h2 += h1;

  h1 = fmix64(h1);
  h2 = fmix64(h2);

  h1 += h2;

  return h1;
}

// CRC32 hash using the SSE4.2 instruction
u64 GetCRC32(const u8* src, u32 len, u32 samples)
{
#if _M_SSE >= 0x402 || defined(_M_ARM_64)
  u64 h[4] = {len, 0, 0, 0};
  u32 Step = (len / 8);
  const u64* data = (const u64*)src;
  const u64* end = data + Step;
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;
#endif

#if _M_SSE >= 0x402
  while (data < end - Step * 3)
  {
    h[0] = _mm_crc32_u64(h[0], data[Step * 0]);
    h[1] = _mm_crc32_u64(h[1], data[Step * 1]);
    h[2] = _mm_crc32_u64(h[2], data[Step * 2]);
    h[3] = _mm_crc32_u64(h[3], data[Step * 3]);
    data += Step * 4;
  }
  if (data < end - Step * 0)
    h[0] = _mm_crc32_u64(h[0], data[Step * 0]);
  if (data < end - Step * 1)
    h[1] = _mm_crc32_u64(h[1], data[Step * 1]);
  if (data < end - Step * 2)
    h[2] = _mm_crc32_u64(h[2], data[Step * 2]);

  if (len & 7)
  {
    u64 temp = 0;
    memcpy(&temp, end, len & 7);
    h[0] = _mm_crc32_u64(h[0], temp);
  }
#elif defined(_M_ARM_64)
  // We should be able to use intrinsics for this
  // Too bad the intrinsics for this instruction was added in GCC 4.9.1
  // The Android NDK (as of r10e) only has GCC 4.9
  // Once the Android NDK has a newer GCC version, update these to use intrinsics
  while (data < end - Step * 3)
  {
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[0])
        : [two] "r"(h[0]), [three] "r"(data[Step * 0]));
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[1])
        : [two] "r"(h[1]), [three] "r"(data[Step * 1]));
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[2])
        : [two] "r"(h[2]), [three] "r"(data[Step * 2]));
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[3])
        : [two] "r"(h[3]), [three] "r"(data[Step * 3]));

    data += Step * 4;
  }
  if (data < end - Step * 0)
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[0])
        : [two] "r"(h[0]), [three] "r"(data[Step * 0]));
  if (data < end - Step * 1)
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[1])
        : [two] "r"(h[1]), [three] "r"(data[Step * 1]));
  if (data < end - Step * 2)
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[2])
        : [two] "r"(h[2]), [three] "r"(data[Step * 2]));

  if (len & 7)
  {
    u64 temp = 0;
    memcpy(&temp, end, len & 7);
    asm("crc32x %w[res], %w[two], %x[three]"
        : [res] "=r"(h[0])
        : [two] "r"(h[0]), [three] "r"(temp));
  }
#endif

#if _M_SSE >= 0x402 || defined(_M_ARM_64)
  // FIXME: is there a better way to combine these partial hashes?
  return h[0] + (h[1] << 10) + (h[2] << 21) + (h[3] << 32);
#else
  return 0;
#endif
}

/*
 * NOTE: This hash function is used for custom texture loading/dumping, so
 * it should not be changed, which would require all custom textures to be
 * recalculated for their new hash values. If the hashing function is
 * changed, make sure this one is still used when the legacy parameter is
 * true.
 */
u64 GetHashHiresTexture(const u8* src, u32 len, u32 samples)
{
  const u64 m = 0xc6a4a7935bd1e995;
  u64 h = len * m;
  const int r = 47;
  u32 Step = (len / 8);
  const u64* data = (const u64*)src;
  const u64* end = data + Step;
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;
  while (data < end)
  {
    u64 k = data[0];
    data += Step;
    k *= m;
    k ^= k >> r;
    k *= m;
    h ^= k;
    h *= m;
  }

  const u8* data2 = (const u8*)end;

  switch (len & 7)
  {
  case 7:
    h ^= u64(data2[6]) << 48;
  case 6:
    h ^= u64(data2[5]) << 40;
  case 5:
    h ^= u64(data2[4]) << 32;
  case 4:
    h ^= u64(data2[3]) << 24;
  case 3:
    h ^= u64(data2[2]) << 16;
  case 2:
    h ^= u64(data2[1]) << 8;
  case 1:
    h ^= u64(data2[0]);
    h *= m;
  };

  h ^= h >> r;
  h *= m;
  h ^= h >> r;

  return h;
}
#else
// CRC32 hash using the SSE4.2 instruction
u64 GetCRC32(const u8* src, u32 len, u32 samples)
{
#if _M_SSE >= 0x402
  u32 h = len;
  u32 Step = (len / 4);
  const u32* data = (const u32*)src;
  const u32* end = data + Step;
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;
  while (data < end)
  {
    h = _mm_crc32_u32(h, data[0]);
    data += Step;
  }

  const u8* data2 = (const u8*)end;
  return (u64)_mm_crc32_u32(h, u32(data2[0]));
#else
  return 0;
#endif
}

//-----------------------------------------------------------------------------
// Block read - if your platform needs to do endian-swapping or can only
// handle aligned reads, do the conversion here

inline u32 getblock(const u32* p, int i)
{
  return p[i];
}

//----------
// Finalization mix - force all bits of a hash block to avalanche

// avalanches all bits to within 0.25% bias

inline u32 fmix32(u32 h)
{
  h ^= h >> 16;
  h *= 0x85ebca6b;
  h ^= h >> 13;
  h *= 0xc2b2ae35;
  h ^= h >> 16;

  return h;
}

inline void bmix32(u32& h1, u32& h2, u32& k1, u32& k2, u32& c1, u32& c2)
{
  k1 *= c1;
  k1 = _rotl(k1, 11);
  k1 *= c2;
  h1 ^= k1;
  h1 += h2;

  h2 = _rotl(h2, 17);

  k2 *= c2;
  k2 = _rotl(k2, 11);
  k2 *= c1;
  h2 ^= k2;
  h2 += h1;

  h1 = h1 * 3 + 0x52dce729;
  h2 = h2 * 3 + 0x38495ab5;

  c1 = c1 * 5 + 0x7b7d159c;
  c2 = c2 * 5 + 0x6bce6396;
}

//----------

u64 GetMurmurHash3(const u8* src, u32 len, u32 samples)
{
  const u8* data = (const u8*)src;
  u32 out[2];
  const int nblocks = len / 8;
  u32 Step = (len / 4);
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;

  u32 h1 = 0x8de1c3ac;
  u32 h2 = 0xbab98226;

  u32 c1 = 0x95543787;
  u32 c2 = 0x2ad7eb25;

  //----------
  // body

  const u32* blocks = (const u32*)(data + nblocks * 8);

  for (int i = -nblocks; i < 0; i += Step)
  {
    u32 k1 = getblock(blocks, i * 2 + 0);
    u32 k2 = getblock(blocks, i * 2 + 1);

    bmix32(h1, h2, k1, k2, c1, c2);
  }

  //----------
  // tail

  const u8* tail = (const u8*)(data + nblocks * 8);

  u32 k1 = 0;
  u32 k2 = 0;

  switch (len & 7)
  {
  case 7:
    k2 ^= tail[6] << 16;
  case 6:
    k2 ^= tail[5] << 8;
  case 5:
    k2 ^= tail[4] << 0;
  case 4:
    k1 ^= tail[3] << 24;
  case 3:
    k1 ^= tail[2] << 16;
  case 2:
    k1 ^= tail[1] << 8;
  case 1:
    k1 ^= tail[0] << 0;
    bmix32(h1, h2, k1, k2, c1, c2);
  };

  //----------
  // finalization

  h2 ^= len;

  h1 += h2;
  h2 += h1;

  h1 = fmix32(h1);
  h2 = fmix32(h2);

  h1 += h2;
  h2 += h1;

  out[0] = h1;
  out[1] = h2;

  return *((u64*)&out);
}

/*
 * FIXME: The old 32-bit version of this hash made different hashes than the
 * 64-bit version. Until someone can make a new version of the 32-bit one that
 * makes identical hashes, this is just a c/p of the 64-bit one.
 */
u64 GetHashHiresTexture(const u8* src, u32 len, u32 samples)
{
  const u64 m = 0xc6a4a7935bd1e995ULL;
  u64 h = len * m;
  const int r = 47;
  u32 Step = (len / 8);
  const u64* data = (const u64*)src;
  const u64* end = data + Step;
  if (samples == 0)
    samples = std::max(Step, 1u);
  Step = Step / samples;
  if (Step < 1)
    Step = 1;
  while (data < end)
  {
    u64 k = data[0];
    data += Step;
    k *= m;
    k ^= k >> r;
    k *= m;
    h ^= k;
    h *= m;
  }

  const u8* data2 = (const u8*)end;

  switch (len & 7)
  {
  case 7:
    h ^= u64(data2[6]) << 48;
  case 6:
    h ^= u64(data2[5]) << 40;
  case 5:
    h ^= u64(data2[4]) << 32;
  case 4:
    h ^= u64(data2[3]) << 24;
  case 3:
    h ^= u64(data2[2]) << 16;
  case 2:
    h ^= u64(data2[1]) << 8;
  case 1:
    h ^= u64(data2[0]);
    h *= m;
  };

  h ^= h >> r;
  h *= m;
  h ^= h >> r;

  return h;
}
#endif

u64 GetHash64(const u8* src, u32 len, u32 samples)
{
  return ptrHashFunction(src, len, samples);
}

// sets the hash function used for the texture cache
void SetHash64Function()
{
#if _M_SSE >= 0x402
  if (cpu_info.bSSE4_2)  // sse crc32 version
  {
    ptrHashFunction = &GetCRC32;
  }
  else
#elif defined(_M_ARM_64)
  if (cpu_info.bCRC32)
  {
    ptrHashFunction = &GetCRC32;
  }
  else
#endif
  {
    ptrHashFunction = &GetMurmurHash3;
  }
}