/Users/deen/code/yugabyte-db/src/yb/gutil/bits.h
| Line | Count | Source (jump to first uncovered line) | 
| 1 |  | // Copyright 2002 and onwards Google Inc. | 
| 2 |  | // | 
| 3 |  | // The following only applies to changes made to this file as part of YugaByte development. | 
| 4 |  | // | 
| 5 |  | // Portions Copyright (c) YugaByte, Inc. | 
| 6 |  | // | 
| 7 |  | // Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except | 
| 8 |  | // in compliance with the License.  You may obtain a copy of the License at | 
| 9 |  | // | 
| 10 |  | // http://www.apache.org/licenses/LICENSE-2.0 | 
| 11 |  | // | 
| 12 |  | // Unless required by applicable law or agreed to in writing, software distributed under the License | 
| 13 |  | // is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express | 
| 14 |  | // or implied.  See the License for the specific language governing permissions and limitations | 
| 15 |  | // under the License. | 
| 16 |  | // | 
| 17 |  | // A collection of useful (static) bit-twiddling functions. | 
| 18 |  |  | 
| 19 |  | #ifndef YB_GUTIL_BITS_H | 
| 20 |  | #define YB_GUTIL_BITS_H | 
| 21 |  |  | 
| 22 |  | #include <glog/logging.h> | 
| 23 |  |  | 
| 24 |  | #include "yb/gutil/integral_types.h" | 
| 25 |  | #include "yb/gutil/logging-inl.h" | 
| 26 |  | #include "yb/gutil/macros.h" | 
| 27 |  |  | 
| 28 |  | class Bits { | 
| 29 |  |  public: | 
| 30 |  |   // Return the number of one bits in the given integer. | 
| 31 |  |   static int CountOnesInByte(unsigned char n); | 
| 32 |  |  | 
| 33 | 0 |   static int CountOnes(uint32 n) { | 
| 34 | 0 |     n -= ((n >> 1) & 0x55555555); | 
| 35 | 0 |     n = ((n >> 2) & 0x33333333) + (n & 0x33333333); | 
| 36 | 0 |     return (((n + (n >> 4)) & 0xF0F0F0F) * 0x1010101) >> 24; | 
| 37 | 0 |   } | 
| 38 |  |  | 
| 39 |  |   // Count bits using sideways addition [WWG'57]. See Knuth TAOCP v4 7.1.3(59) | 
| 40 | 0 |   static inline int CountOnes64(uint64 n) { | 
| 41 | 0 | #if defined(__x86_64__) | 
| 42 | 0 |     n -= (n >> 1) & 0x5555555555555555ULL; | 
| 43 | 0 |     n = ((n >> 2) & 0x3333333333333333ULL) + (n & 0x3333333333333333ULL); | 
| 44 | 0 |     return (((n + (n >> 4)) & 0xF0F0F0F0F0F0F0FULL) | 
| 45 | 0 |             * 0x101010101010101ULL) >> 56; | 
| 46 | 0 | #else | 
| 47 | 0 |     return CountOnes(n >> 32) + CountOnes(n & 0xffffffff); | 
| 48 | 0 | #endif | 
| 49 | 0 |   } | 
| 50 |  |  | 
| 51 |  |   // Reverse the bits in the given integer. | 
| 52 |  |   static uint8 ReverseBits8(uint8 n); | 
| 53 |  |   static uint32 ReverseBits32(uint32 n); | 
| 54 |  |   static uint64 ReverseBits64(uint64 n); | 
| 55 |  |  | 
| 56 |  |   // Return the number of one bits in the byte sequence. | 
| 57 |  |   static int Count(const void *m, int num_bytes); | 
| 58 |  |  | 
| 59 |  |   // Return the number of different bits in the given byte sequences. | 
| 60 |  |   // (i.e., the Hamming distance) | 
| 61 |  |   static int Difference(const void *m1, const void *m2, int num_bytes); | 
| 62 |  |  | 
| 63 |  |   // Return the number of different bits in the given byte sequences, | 
| 64 |  |   // up to a maximum.  Values larger than the maximum may be returned | 
| 65 |  |   // (because multiple bits are checked at a time), but the function | 
| 66 |  |   // may exit early if the cap is exceeded. | 
| 67 |  |   static int CappedDifference(const void *m1, const void *m2, | 
| 68 |  |                               int num_bytes, int cap); | 
| 69 |  |  | 
| 70 |  |   // Return floor(log2(n)) for positive integer n.  Returns -1 iff n == 0. | 
| 71 |  |   static int Log2Floor(uint32 n); | 
| 72 |  |   static int Log2Floor64(uint64 n); | 
| 73 |  |  | 
| 74 |  |   // Potentially faster version of Log2Floor() that returns an | 
| 75 |  |   // undefined value if n == 0 | 
| 76 |  |   static int Log2FloorNonZero(uint32 n); | 
| 77 |  |   static int Log2FloorNonZero64(uint64 n); | 
| 78 |  |  | 
| 79 |  |   // Return ceiling(log2(n)) for positive integer n.  Returns -1 iff n == 0. | 
| 80 |  |   static int Log2Ceiling(uint32 n); | 
| 81 |  |   static int Log2Ceiling64(uint64 n); | 
| 82 |  |  | 
| 83 |  |   // Return the first set least / most significant bit, 0-indexed.  Returns an | 
| 84 |  |   // undefined value if n == 0.  FindLSBSetNonZero() is similar to ffs() except | 
| 85 |  |   // that it's 0-indexed, while FindMSBSetNonZero() is the same as | 
| 86 |  |   // Log2FloorNonZero(). | 
| 87 |  |   static int FindLSBSetNonZero(uint32 n); | 
| 88 |  |   static int FindLSBSetNonZero64(uint64 n); | 
| 89 | 0 |   static int FindMSBSetNonZero(uint32 n) { return Log2FloorNonZero(n); } | 
| 90 | 0 |   static int FindMSBSetNonZero64(uint64 n) { return Log2FloorNonZero64(n); } | 
| 91 |  |  | 
| 92 |  |   // Portable implementations | 
| 93 |  |   static int Log2Floor_Portable(uint32 n); | 
| 94 |  |   static int Log2FloorNonZero_Portable(uint32 n); | 
| 95 |  |   static int FindLSBSetNonZero_Portable(uint32 n); | 
| 96 |  |   static int Log2Floor64_Portable(uint64 n); | 
| 97 |  |   static int Log2FloorNonZero64_Portable(uint64 n); | 
| 98 |  |   static int FindLSBSetNonZero64_Portable(uint64 n); | 
| 99 |  |  | 
| 100 |  |   // Viewing bytes as a stream of unsigned bytes, does that stream | 
| 101 |  |   // contain any byte equal to c? | 
| 102 |  |   template <class T> static bool BytesContainByte(T bytes, uint8 c); | 
| 103 |  |  | 
| 104 |  |   // Viewing bytes as a stream of unsigned bytes, does that stream | 
| 105 |  |   // contain any byte b < c? | 
| 106 |  |   template <class T> static bool BytesContainByteLessThan(T bytes, uint8 c); | 
| 107 |  |  | 
| 108 |  |   // Viewing bytes as a stream of unsigned bytes, are all elements of that | 
| 109 |  |   // stream in [lo, hi]? | 
| 110 |  |   template <class T> static bool BytesAllInRange(T bytes, uint8 lo, uint8 hi); | 
| 111 |  |  | 
| 112 |  |  private: | 
| 113 |  |   static const char num_bits[]; | 
| 114 |  |   static const unsigned char bit_reverse_table[]; | 
| 115 |  |   DISALLOW_COPY_AND_ASSIGN(Bits); | 
| 116 |  | }; | 
| 117 |  |  | 
| 118 |  | // A utility class for some handy bit patterns.  The names l and h | 
| 119 |  | // were chosen to match Knuth Volume 4: l is 0x010101... and h is 0x808080...; | 
| 120 |  | // half_ones is ones in the lower half only.  We assume sizeof(T) is 1 or even. | 
| 121 |  | template <class T> struct BitPattern { | 
| 122 |  |   static const T half_ones = (static_cast<T>(1) << (sizeof(T)*4)) - 1; | 
| 123 |  |   static const T l = (sizeof(T) == 1) ? 1 : | 
| 124 |  |                        (half_ones / 0xff * (half_ones + 2)); | 
| 125 |  |   static const T h = ~(l * 0x7f); | 
| 126 |  | }; | 
| 127 |  |  | 
| 128 |  | // ------------------------------------------------------------------------ | 
| 129 |  | // Implementation details follow | 
| 130 |  | // ------------------------------------------------------------------------ | 
| 131 |  |  | 
| 132 |  | // use GNU builtins where available | 
| 133 |  | #if defined(__GNUC__) && \ | 
| 134 |  |     ((__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || __GNUC__ >= 4) | 
| 135 | 9.22M | inline int Bits::Log2Floor(uint32 n) { | 
| 136 | 9.22M |   return n == 0 ? -10: 31 ^ __builtin_clz(n); | 
| 137 | 9.22M | } | 
| 138 |  |  | 
| 139 | 0 | inline int Bits::Log2FloorNonZero(uint32 n) { | 
| 140 | 0 |   return 31 ^ __builtin_clz(n); | 
| 141 | 0 | } | 
| 142 |  |  | 
| 143 |  | inline int Bits::FindLSBSetNonZero(uint32 n) { | 
| 144 |  |   return __builtin_ctz(n); | 
| 145 |  | } | 
| 146 |  |  | 
| 147 | 689M | inline int Bits::Log2Floor64(uint64 n) { | 
| 148 | 689M |   return n == 0 ? -10: 63 ^ __builtin_clzll(n); | 
| 149 | 689M | } | 
| 150 |  |  | 
| 151 | 0 | inline int Bits::Log2FloorNonZero64(uint64 n) { | 
| 152 | 0 |   return 63 ^ __builtin_clzll(n); | 
| 153 | 0 | } | 
| 154 |  |  | 
| 155 | 0 | inline int Bits::FindLSBSetNonZero64(uint64 n) { | 
| 156 | 0 |   return __builtin_ctzll(n); | 
| 157 | 0 | } | 
| 158 |  | #elif defined(_MSC_VER) | 
| 159 |  | #include "yb/gutil/bits-internal-windows.h" | 
| 160 |  | #else | 
| 161 |  | #include "yb/gutil/bits-internal-unknown.h" | 
| 162 |  | #endif | 
| 163 |  |  | 
| 164 | 0 | inline int Bits::CountOnesInByte(unsigned char n) { | 
| 165 | 0 |   return num_bits[n]; | 
| 166 | 0 | } | 
| 167 |  |  | 
| 168 | 0 | inline uint8 Bits::ReverseBits8(unsigned char n) { | 
| 169 | 0 |   n = ((n >> 1) & 0x55) | ((n & 0x55) << 1); | 
| 170 | 0 |   n = ((n >> 2) & 0x33) | ((n & 0x33) << 2); | 
| 171 | 0 |   return ((n >> 4) & 0x0f)  | ((n & 0x0f) << 4); | 
| 172 | 0 | } | 
| 173 |  |  | 
| 174 | 0 | inline uint32 Bits::ReverseBits32(uint32 n) { | 
| 175 | 0 |   n = ((n >> 1) & 0x55555555) | ((n & 0x55555555) << 1); | 
| 176 | 0 |   n = ((n >> 2) & 0x33333333) | ((n & 0x33333333) << 2); | 
| 177 | 0 |   n = ((n >> 4) & 0x0F0F0F0F) | ((n & 0x0F0F0F0F) << 4); | 
| 178 | 0 |   n = ((n >> 8) & 0x00FF00FF) | ((n & 0x00FF00FF) << 8); | 
| 179 | 0 |   return ( n >> 16 ) | ( n << 16); | 
| 180 | 0 | } | 
| 181 |  |  | 
| 182 | 0 | inline uint64 Bits::ReverseBits64(uint64 n) { | 
| 183 | 0 | #if defined(__x86_64__) | 
| 184 | 0 |   n = ((n >> 1) & 0x5555555555555555ULL) | ((n & 0x5555555555555555ULL) << 1); | 
| 185 | 0 |   n = ((n >> 2) & 0x3333333333333333ULL) | ((n & 0x3333333333333333ULL) << 2); | 
| 186 | 0 |   n = ((n >> 4) & 0x0F0F0F0F0F0F0F0FULL) | ((n & 0x0F0F0F0F0F0F0F0FULL) << 4); | 
| 187 | 0 |   n = ((n >> 8) & 0x00FF00FF00FF00FFULL) | ((n & 0x00FF00FF00FF00FFULL) << 8); | 
| 188 | 0 |   n = ((n >> 16) & 0x0000FFFF0000FFFFULL) | ((n & 0x0000FFFF0000FFFFULL) << 16); | 
| 189 | 0 |   return ( n >> 32 ) | ( n << 32); | 
| 190 | 0 | #else | 
| 191 | 0 |   return ReverseBits32( n >> 32 ) | | 
| 192 | 0 |          (static_cast<uint64>(ReverseBits32( n &  0xffffffff )) << 32); | 
| 193 | 0 | #endif | 
| 194 | 0 | } | 
| 195 |  |  | 
| 196 | 0 | inline int Bits::Log2FloorNonZero_Portable(uint32 n) { | 
| 197 | 0 |   // Just use the common routine | 
| 198 | 0 |   return Log2Floor(n); | 
| 199 | 0 | } | 
| 200 |  |  | 
| 201 |  | // Log2Floor64() is defined in terms of Log2Floor32(), Log2FloorNonZero32() | 
| 202 | 0 | inline int Bits::Log2Floor64_Portable(uint64 n) { | 
| 203 | 0 |   const uint32 topbits = static_cast<uint32>(n >> 32); | 
| 204 | 0 |   if (topbits == 0) { | 
| 205 | 0 |     // Top bits are zero, so scan in bottom bits | 
| 206 | 0 |     return Log2Floor(static_cast<uint32>(n)); | 
| 207 | 0 |   } else { | 
| 208 | 0 |     return 32 + Log2FloorNonZero(topbits); | 
| 209 | 0 |   } | 
| 210 | 0 | } | 
| 211 |  |  | 
| 212 |  | // Log2FloorNonZero64() is defined in terms of Log2FloorNonZero32() | 
| 213 | 0 | inline int Bits::Log2FloorNonZero64_Portable(uint64 n) { | 
| 214 | 0 |   const uint32 topbits = static_cast<uint32>(n >> 32); | 
| 215 | 0 |   if (topbits == 0) { | 
| 216 | 0 |     // Top bits are zero, so scan in bottom bits | 
| 217 | 0 |     return Log2FloorNonZero(static_cast<uint32>(n)); | 
| 218 | 0 |   } else { | 
| 219 | 0 |     return 32 + Log2FloorNonZero(topbits); | 
| 220 | 0 |   } | 
| 221 | 0 | } | 
| 222 |  |  | 
| 223 |  | // FindLSBSetNonZero64() is defined in terms of FindLSBSetNonZero() | 
| 224 | 0 | inline int Bits::FindLSBSetNonZero64_Portable(uint64 n) { | 
| 225 | 0 |   const uint32 bottombits = static_cast<uint32>(n); | 
| 226 | 0 |   if (bottombits == 0) { | 
| 227 | 0 |     // Bottom bits are zero, so scan in top bits | 
| 228 | 0 |     return 32 + FindLSBSetNonZero(static_cast<uint32>(n >> 32)); | 
| 229 | 0 |   } else { | 
| 230 | 0 |     return FindLSBSetNonZero(bottombits); | 
| 231 | 0 |   } | 
| 232 | 0 | } | 
| 233 |  |  | 
| 234 |  | template <class T> | 
| 235 |  | inline bool Bits::BytesContainByteLessThan(T bytes, uint8 c) { | 
| 236 |  |   T l = BitPattern<T>::l; | 
| 237 |  |   T h = BitPattern<T>::h; | 
| 238 |  |   // The c <= 0x80 code is straight out of Knuth Volume 4. | 
| 239 |  |   // Usually c will be manifestly constant. | 
| 240 |  |   return c <= 0x80 ? | 
| 241 |  |       ((h & (bytes - l * c) & ~bytes) != 0) : | 
| 242 |  |       ((((bytes - l * c) | (bytes ^ h)) & h) != 0); | 
| 243 |  | } | 
| 244 |  |  | 
| 245 |  | template <class T> inline bool Bits::BytesContainByte(T bytes, uint8 c) { | 
| 246 |  |   // Usually c will be manifestly constant. | 
| 247 |  |   return Bits::BytesContainByteLessThan<T>(bytes ^ (c * BitPattern<T>::l), 1); | 
| 248 |  | } | 
| 249 |  |  | 
| 250 |  | template <class T> | 
| 251 |  | inline bool Bits::BytesAllInRange(T bytes, uint8 lo, uint8 hi) { | 
| 252 |  |   T l = BitPattern<T>::l; | 
| 253 |  |   T h = BitPattern<T>::h; | 
| 254 |  |   // In the common case, lo and hi are manifest constants. | 
| 255 |  |   if (lo > hi) { | 
| 256 |  |     return false; | 
| 257 |  |   } | 
| 258 |  |   if (hi - lo < 128) { | 
| 259 |  |     T x = bytes - l * lo; | 
| 260 |  |     T y = bytes + l * (127 - hi); | 
| 261 |  |     return ((x | y) & h) == 0; | 
| 262 |  |   } | 
| 263 |  |   return !Bits::BytesContainByteLessThan(bytes + (255 - hi) * l, | 
| 264 |  |                                          lo + (255 - hi)); | 
| 265 |  | } | 
| 266 |  |  | 
| 267 |  | #endif // YB_GUTIL_BITS_H |