LLVM  10.0.0svn
Hashing.h
Go to the documentation of this file.
1 //===-- llvm/ADT/Hashing.h - Utilities for hashing --------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // This file implements the newly proposed standard C++ interfaces for hashing
10 // arbitrary data and building hash functions for user-defined types. This
11 // interface was originally proposed in N3333[1] and is currently under review
12 // for inclusion in a future TR and/or standard.
13 //
14 // The primary interfaces provide are comprised of one type and three functions:
15 //
16 // -- 'hash_code' class is an opaque type representing the hash code for some
17 // data. It is the intended product of hashing, and can be used to implement
18 // hash tables, checksumming, and other common uses of hashes. It is not an
19 // integer type (although it can be converted to one) because it is risky
20 // to assume much about the internals of a hash_code. In particular, each
21 // execution of the program has a high probability of producing a different
22 // hash_code for a given input. Thus their values are not stable to save or
23 // persist, and should only be used during the execution for the
24 // construction of hashing datastructures.
25 //
26 // -- 'hash_value' is a function designed to be overloaded for each
27 // user-defined type which wishes to be used within a hashing context. It
28 // should be overloaded within the user-defined type's namespace and found
29 // via ADL. Overloads for primitive types are provided by this library.
30 //
31 // -- 'hash_combine' and 'hash_combine_range' are functions designed to aid
32 // programmers in easily and intuitively combining a set of data into
33 // a single hash_code for their object. They should only logically be used
34 // within the implementation of a 'hash_value' routine or similar context.
35 //
36 // Note that 'hash_combine_range' contains very special logic for hashing
37 // a contiguous array of integers or pointers. This logic is *extremely* fast,
38 // on a modern Intel "Gainestown" Xeon (Nehalem uarch) @2.2 GHz, these were
39 // benchmarked at over 6.5 GiB/s for large keys, and <20 cycles/hash for keys
40 // under 32-bytes.
41 //
42 //===----------------------------------------------------------------------===//
43 
44 #ifndef LLVM_ADT_HASHING_H
45 #define LLVM_ADT_HASHING_H
46 
47 #include "llvm/Support/DataTypes.h"
50 #include <algorithm>
51 #include <cassert>
52 #include <cstring>
53 #include <string>
54 #include <utility>
55 
56 namespace llvm {
57 
58 /// An opaque object representing a hash code.
59 ///
60 /// This object represents the result of hashing some entity. It is intended to
61 /// be used to implement hashtables or other hashing-based data structures.
62 /// While it wraps and exposes a numeric value, this value should not be
63 /// trusted to be stable or predictable across processes or executions.
64 ///
65 /// In order to obtain the hash_code for an object 'x':
66 /// \code
67 /// using llvm::hash_value;
68 /// llvm::hash_code code = hash_value(x);
69 /// \endcode
70 class hash_code {
71  size_t value;
72 
73 public:
74  /// Default construct a hash_code.
75  /// Note that this leaves the value uninitialized.
76  hash_code() = default;
77 
78  /// Form a hash code directly from a numerical value.
79  hash_code(size_t value) : value(value) {}
80 
81  /// Convert the hash code to its numerical value for use.
82  /*explicit*/ operator size_t() const { return value; }
83 
84  friend bool operator==(const hash_code &lhs, const hash_code &rhs) {
85  return lhs.value == rhs.value;
86  }
87  friend bool operator!=(const hash_code &lhs, const hash_code &rhs) {
88  return lhs.value != rhs.value;
89  }
90 
91  /// Allow a hash_code to be directly run through hash_value.
92  friend size_t hash_value(const hash_code &code) { return code.value; }
93 };
94 
95 /// Compute a hash_code for any integer value.
96 ///
97 /// Note that this function is intended to compute the same hash_code for
98 /// a particular value without regard to the pre-promotion type. This is in
99 /// contrast to hash_combine which may produce different hash_codes for
100 /// differing argument types even if they would implicit promote to a common
101 /// type without changing the value.
102 template <typename T>
103 typename std::enable_if<is_integral_or_enum<T>::value, hash_code>::type
104 hash_value(T value);
105 
106 /// Compute a hash_code for a pointer's address.
107 ///
108 /// N.B.: This hashes the *address*. Not the value and not the type.
109 template <typename T> hash_code hash_value(const T *ptr);
110 
111 /// Compute a hash_code for a pair of objects.
112 template <typename T, typename U>
113 hash_code hash_value(const std::pair<T, U> &arg);
114 
115 /// Compute a hash_code for a standard string.
116 template <typename T>
117 hash_code hash_value(const std::basic_string<T> &arg);
118 
119 
120 /// Override the execution seed with a fixed value.
121 ///
122 /// This hashing library uses a per-execution seed designed to change on each
123 /// run with high probability in order to ensure that the hash codes are not
124 /// attackable and to ensure that output which is intended to be stable does
125 /// not rely on the particulars of the hash codes produced.
126 ///
127 /// That said, there are use cases where it is important to be able to
128 /// reproduce *exactly* a specific behavior. To that end, we provide a function
129 /// which will forcibly set the seed to a fixed value. This must be done at the
130 /// start of the program, before any hashes are computed. Also, it cannot be
131 /// undone. This makes it thread-hostile and very hard to use outside of
132 /// immediately on start of a simple program designed for reproducible
133 /// behavior.
134 void set_fixed_execution_hash_seed(uint64_t fixed_value);
135 
136 
137 // All of the implementation details of actually computing the various hash
138 // code values are held within this namespace. These routines are included in
139 // the header file mainly to allow inlining and constant propagation.
140 namespace hashing {
141 namespace detail {
142 
143 inline uint64_t fetch64(const char *p) {
144  uint64_t result;
145  memcpy(&result, p, sizeof(result));
147  sys::swapByteOrder(result);
148  return result;
149 }
150 
151 inline uint32_t fetch32(const char *p) {
152  uint32_t result;
153  memcpy(&result, p, sizeof(result));
155  sys::swapByteOrder(result);
156  return result;
157 }
158 
159 /// Some primes between 2^63 and 2^64 for various uses.
160 static const uint64_t k0 = 0xc3a5c85c97cb3127ULL;
161 static const uint64_t k1 = 0xb492b66fbe98f273ULL;
162 static const uint64_t k2 = 0x9ae16a3b2f90404fULL;
163 static const uint64_t k3 = 0xc949d7c7509e6557ULL;
164 
165 /// Bitwise right rotate.
166 /// Normally this will compile to a single instruction, especially if the
167 /// shift is a manifest constant.
168 inline uint64_t rotate(uint64_t val, size_t shift) {
169  // Avoid shifting by 64: doing so yields an undefined result.
170  return shift == 0 ? val : ((val >> shift) | (val << (64 - shift)));
171 }
172 
173 inline uint64_t shift_mix(uint64_t val) {
174  return val ^ (val >> 47);
175 }
176 
177 inline uint64_t hash_16_bytes(uint64_t low, uint64_t high) {
178  // Murmur-inspired hashing.
179  const uint64_t kMul = 0x9ddfea08eb382d69ULL;
180  uint64_t a = (low ^ high) * kMul;
181  a ^= (a >> 47);
182  uint64_t b = (high ^ a) * kMul;
183  b ^= (b >> 47);
184  b *= kMul;
185  return b;
186 }
187 
188 inline uint64_t hash_1to3_bytes(const char *s, size_t len, uint64_t seed) {
189  uint8_t a = s[0];
190  uint8_t b = s[len >> 1];
191  uint8_t c = s[len - 1];
192  uint32_t y = static_cast<uint32_t>(a) + (static_cast<uint32_t>(b) << 8);
193  uint32_t z = static_cast<uint32_t>(len) + (static_cast<uint32_t>(c) << 2);
194  return shift_mix(y * k2 ^ z * k3 ^ seed) * k2;
195 }
196 
197 inline uint64_t hash_4to8_bytes(const char *s, size_t len, uint64_t seed) {
198  uint64_t a = fetch32(s);
199  return hash_16_bytes(len + (a << 3), seed ^ fetch32(s + len - 4));
200 }
201 
202 inline uint64_t hash_9to16_bytes(const char *s, size_t len, uint64_t seed) {
203  uint64_t a = fetch64(s);
204  uint64_t b = fetch64(s + len - 8);
205  return hash_16_bytes(seed ^ a, rotate(b + len, len)) ^ b;
206 }
207 
208 inline uint64_t hash_17to32_bytes(const char *s, size_t len, uint64_t seed) {
209  uint64_t a = fetch64(s) * k1;
210  uint64_t b = fetch64(s + 8);
211  uint64_t c = fetch64(s + len - 8) * k2;
212  uint64_t d = fetch64(s + len - 16) * k0;
213  return hash_16_bytes(rotate(a - b, 43) + rotate(c ^ seed, 30) + d,
214  a + rotate(b ^ k3, 20) - c + len + seed);
215 }
216 
217 inline uint64_t hash_33to64_bytes(const char *s, size_t len, uint64_t seed) {
218  uint64_t z = fetch64(s + 24);
219  uint64_t a = fetch64(s) + (len + fetch64(s + len - 16)) * k0;
220  uint64_t b = rotate(a + z, 52);
221  uint64_t c = rotate(a, 37);
222  a += fetch64(s + 8);
223  c += rotate(a, 7);
224  a += fetch64(s + 16);
225  uint64_t vf = a + z;
226  uint64_t vs = b + rotate(a, 31) + c;
227  a = fetch64(s + 16) + fetch64(s + len - 32);
228  z = fetch64(s + len - 8);
229  b = rotate(a + z, 52);
230  c = rotate(a, 37);
231  a += fetch64(s + len - 24);
232  c += rotate(a, 7);
233  a += fetch64(s + len - 16);
234  uint64_t wf = a + z;
235  uint64_t ws = b + rotate(a, 31) + c;
236  uint64_t r = shift_mix((vf + ws) * k2 + (wf + vs) * k0);
237  return shift_mix((seed ^ (r * k0)) + vs) * k2;
238 }
239 
240 inline uint64_t hash_short(const char *s, size_t length, uint64_t seed) {
241  if (length >= 4 && length <= 8)
242  return hash_4to8_bytes(s, length, seed);
243  if (length > 8 && length <= 16)
244  return hash_9to16_bytes(s, length, seed);
245  if (length > 16 && length <= 32)
246  return hash_17to32_bytes(s, length, seed);
247  if (length > 32)
248  return hash_33to64_bytes(s, length, seed);
249  if (length != 0)
250  return hash_1to3_bytes(s, length, seed);
251 
252  return k2 ^ seed;
253 }
254 
255 /// The intermediate state used during hashing.
256 /// Currently, the algorithm for computing hash codes is based on CityHash and
257 /// keeps 56 bytes of arbitrary state.
258 struct hash_state {
259  uint64_t h0, h1, h2, h3, h4, h5, h6;
260 
261  /// Create a new hash_state structure and initialize it based on the
262  /// seed and the first 64-byte chunk.
263  /// This effectively performs the initial mix.
264  static hash_state create(const char *s, uint64_t seed) {
265  hash_state state = {
266  0, seed, hash_16_bytes(seed, k1), rotate(seed ^ k1, 49),
267  seed * k1, shift_mix(seed), 0 };
268  state.h6 = hash_16_bytes(state.h4, state.h5);
269  state.mix(s);
270  return state;
271  }
272 
273  /// Mix 32-bytes from the input sequence into the 16-bytes of 'a'
274  /// and 'b', including whatever is already in 'a' and 'b'.
275  static void mix_32_bytes(const char *s, uint64_t &a, uint64_t &b) {
276  a += fetch64(s);
277  uint64_t c = fetch64(s + 24);
278  b = rotate(b + a + c, 21);
279  uint64_t d = a;
280  a += fetch64(s + 8) + fetch64(s + 16);
281  b += rotate(a, 44) + d;
282  a += c;
283  }
284 
285  /// Mix in a 64-byte buffer of data.
286  /// We mix all 64 bytes even when the chunk length is smaller, but we
287  /// record the actual length.
288  void mix(const char *s) {
289  h0 = rotate(h0 + h1 + h3 + fetch64(s + 8), 37) * k1;
290  h1 = rotate(h1 + h4 + fetch64(s + 48), 42) * k1;
291  h0 ^= h6;
292  h1 += h3 + fetch64(s + 40);
293  h2 = rotate(h2 + h5, 33) * k1;
294  h3 = h4 * k1;
295  h4 = h0 + h5;
296  mix_32_bytes(s, h3, h4);
297  h5 = h2 + h6;
298  h6 = h1 + fetch64(s + 16);
299  mix_32_bytes(s + 32, h5, h6);
300  std::swap(h2, h0);
301  }
302 
303  /// Compute the final 64-bit hash code value based on the current
304  /// state and the length of bytes hashed.
305  uint64_t finalize(size_t length) {
306  return hash_16_bytes(hash_16_bytes(h3, h5) + shift_mix(h1) * k1 + h2,
307  hash_16_bytes(h4, h6) + shift_mix(length) * k1 + h0);
308  }
309 };
310 
311 
312 /// A global, fixed seed-override variable.
313 ///
314 /// This variable can be set using the \see llvm::set_fixed_execution_seed
315 /// function. See that function for details. Do not, under any circumstances,
316 /// set or read this variable.
317 extern uint64_t fixed_seed_override;
318 
319 inline uint64_t get_execution_seed() {
320  // FIXME: This needs to be a per-execution seed. This is just a placeholder
321  // implementation. Switching to a per-execution seed is likely to flush out
322  // instability bugs and so will happen as its own commit.
323  //
324  // However, if there is a fixed seed override set the first time this is
325  // called, return that instead of the per-execution seed.
326  const uint64_t seed_prime = 0xff51afd7ed558ccdULL;
327  static uint64_t seed = fixed_seed_override ? fixed_seed_override : seed_prime;
328  return seed;
329 }
330 
331 
332 /// Trait to indicate whether a type's bits can be hashed directly.
333 ///
334 /// A type trait which is true if we want to combine values for hashing by
335 /// reading the underlying data. It is false if values of this type must
336 /// first be passed to hash_value, and the resulting hash_codes combined.
337 //
338 // FIXME: We want to replace is_integral_or_enum and is_pointer here with
339 // a predicate which asserts that comparing the underlying storage of two
340 // values of the type for equality is equivalent to comparing the two values
341 // for equality. For all the platforms we care about, this holds for integers
342 // and pointers, but there are platforms where it doesn't and we would like to
343 // support user-defined types which happen to satisfy this property.
344 template <typename T> struct is_hashable_data
345  : std::integral_constant<bool, ((is_integral_or_enum<T>::value ||
346  std::is_pointer<T>::value) &&
347  64 % sizeof(T) == 0)> {};
348 
349 // Special case std::pair to detect when both types are viable and when there
350 // is no alignment-derived padding in the pair. This is a bit of a lie because
351 // std::pair isn't truly POD, but it's close enough in all reasonable
352 // implementations for our use case of hashing the underlying data.
353 template <typename T, typename U> struct is_hashable_data<std::pair<T, U> >
354  : std::integral_constant<bool, (is_hashable_data<T>::value &&
355  is_hashable_data<U>::value &&
356  (sizeof(T) + sizeof(U)) ==
357  sizeof(std::pair<T, U>))> {};
358 
359 /// Helper to get the hashable data representation for a type.
360 /// This variant is enabled when the type itself can be used.
361 template <typename T>
362 typename std::enable_if<is_hashable_data<T>::value, T>::type
363 get_hashable_data(const T &value) {
364  return value;
365 }
366 /// Helper to get the hashable data representation for a type.
367 /// This variant is enabled when we must first call hash_value and use the
368 /// result as our data.
369 template <typename T>
370 typename std::enable_if<!is_hashable_data<T>::value, size_t>::type
371 get_hashable_data(const T &value) {
373  return hash_value(value);
374 }
375 
376 /// Helper to store data from a value into a buffer and advance the
377 /// pointer into that buffer.
378 ///
379 /// This routine first checks whether there is enough space in the provided
380 /// buffer, and if not immediately returns false. If there is space, it
381 /// copies the underlying bytes of value into the buffer, advances the
382 /// buffer_ptr past the copied bytes, and returns true.
383 template <typename T>
384 bool store_and_advance(char *&buffer_ptr, char *buffer_end, const T& value,
385  size_t offset = 0) {
386  size_t store_size = sizeof(value) - offset;
387  if (buffer_ptr + store_size > buffer_end)
388  return false;
389  const char *value_data = reinterpret_cast<const char *>(&value);
390  memcpy(buffer_ptr, value_data + offset, store_size);
391  buffer_ptr += store_size;
392  return true;
393 }
394 
395 /// Implement the combining of integral values into a hash_code.
396 ///
397 /// This overload is selected when the value type of the iterator is
398 /// integral. Rather than computing a hash_code for each object and then
399 /// combining them, this (as an optimization) directly combines the integers.
400 template <typename InputIteratorT>
401 hash_code hash_combine_range_impl(InputIteratorT first, InputIteratorT last) {
402  const uint64_t seed = get_execution_seed();
403  char buffer[64], *buffer_ptr = buffer;
404  char *const buffer_end = std::end(buffer);
405  while (first != last && store_and_advance(buffer_ptr, buffer_end,
406  get_hashable_data(*first)))
407  ++first;
408  if (first == last)
409  return hash_short(buffer, buffer_ptr - buffer, seed);
410  assert(buffer_ptr == buffer_end);
411 
412  hash_state state = state.create(buffer, seed);
413  size_t length = 64;
414  while (first != last) {
415  // Fill up the buffer. We don't clear it, which re-mixes the last round
416  // when only a partial 64-byte chunk is left.
417  buffer_ptr = buffer;
418  while (first != last && store_and_advance(buffer_ptr, buffer_end,
419  get_hashable_data(*first)))
420  ++first;
421 
422  // Rotate the buffer if we did a partial fill in order to simulate doing
423  // a mix of the last 64-bytes. That is how the algorithm works when we
424  // have a contiguous byte sequence, and we want to emulate that here.
425  std::rotate(buffer, buffer_ptr, buffer_end);
426 
427  // Mix this chunk into the current state.
428  state.mix(buffer);
429  length += buffer_ptr - buffer;
430  };
431 
432  return state.finalize(length);
433 }
434 
435 /// Implement the combining of integral values into a hash_code.
436 ///
437 /// This overload is selected when the value type of the iterator is integral
438 /// and when the input iterator is actually a pointer. Rather than computing
439 /// a hash_code for each object and then combining them, this (as an
440 /// optimization) directly combines the integers. Also, because the integers
441 /// are stored in contiguous memory, this routine avoids copying each value
442 /// and directly reads from the underlying memory.
443 template <typename ValueT>
444 typename std::enable_if<is_hashable_data<ValueT>::value, hash_code>::type
446  const uint64_t seed = get_execution_seed();
447  const char *s_begin = reinterpret_cast<const char *>(first);
448  const char *s_end = reinterpret_cast<const char *>(last);
449  const size_t length = std::distance(s_begin, s_end);
450  if (length <= 64)
451  return hash_short(s_begin, length, seed);
452 
453  const char *s_aligned_end = s_begin + (length & ~63);
454  hash_state state = state.create(s_begin, seed);
455  s_begin += 64;
456  while (s_begin != s_aligned_end) {
457  state.mix(s_begin);
458  s_begin += 64;
459  }
460  if (length & 63)
461  state.mix(s_end - 64);
462 
463  return state.finalize(length);
464 }
465 
466 } // namespace detail
467 } // namespace hashing
468 
469 
470 /// Compute a hash_code for a sequence of values.
471 ///
472 /// This hashes a sequence of values. It produces the same hash_code as
473 /// 'hash_combine(a, b, c, ...)', but can run over arbitrary sized sequences
474 /// and is significantly faster given pointers and types which can be hashed as
475 /// a sequence of bytes.
476 template <typename InputIteratorT>
477 hash_code hash_combine_range(InputIteratorT first, InputIteratorT last) {
479 }
480 
481 
482 // Implementation details for hash_combine.
483 namespace hashing {
484 namespace detail {
485 
486 /// Helper class to manage the recursive combining of hash_combine
487 /// arguments.
488 ///
489 /// This class exists to manage the state and various calls involved in the
490 /// recursive combining of arguments used in hash_combine. It is particularly
491 /// useful at minimizing the code in the recursive calls to ease the pain
492 /// caused by a lack of variadic functions.
494  char buffer[64];
496  const uint64_t seed;
497 
498 public:
499  /// Construct a recursive hash combining helper.
500  ///
501  /// This sets up the state for a recursive hash combine, including getting
502  /// the seed and buffer setup.
504  : seed(get_execution_seed()) {}
505 
506  /// Combine one chunk of data into the current in-flight hash.
507  ///
508  /// This merges one chunk of data into the hash. First it tries to buffer
509  /// the data. If the buffer is full, it hashes the buffer into its
510  /// hash_state, empties it, and then merges the new chunk in. This also
511  /// handles cases where the data straddles the end of the buffer.
512  template <typename T>
513  char *combine_data(size_t &length, char *buffer_ptr, char *buffer_end, T data) {
514  if (!store_and_advance(buffer_ptr, buffer_end, data)) {
515  // Check for skew which prevents the buffer from being packed, and do
516  // a partial store into the buffer to fill it. This is only a concern
517  // with the variadic combine because that formation can have varying
518  // argument types.
519  size_t partial_store_size = buffer_end - buffer_ptr;
520  memcpy(buffer_ptr, &data, partial_store_size);
521 
522  // If the store fails, our buffer is full and ready to hash. We have to
523  // either initialize the hash state (on the first full buffer) or mix
524  // this buffer into the existing hash state. Length tracks the *hashed*
525  // length, not the buffered length.
526  if (length == 0) {
527  state = state.create(buffer, seed);
528  length = 64;
529  } else {
530  // Mix this chunk into the current state and bump length up by 64.
531  state.mix(buffer);
532  length += 64;
533  }
534  // Reset the buffer_ptr to the head of the buffer for the next chunk of
535  // data.
536  buffer_ptr = buffer;
537 
538  // Try again to store into the buffer -- this cannot fail as we only
539  // store types smaller than the buffer.
540  if (!store_and_advance(buffer_ptr, buffer_end, data,
541  partial_store_size))
542  abort();
543  }
544  return buffer_ptr;
545  }
546 
547  /// Recursive, variadic combining method.
548  ///
549  /// This function recurses through each argument, combining that argument
550  /// into a single hash.
551  template <typename T, typename ...Ts>
552  hash_code combine(size_t length, char *buffer_ptr, char *buffer_end,
553  const T &arg, const Ts &...args) {
554  buffer_ptr = combine_data(length, buffer_ptr, buffer_end, get_hashable_data(arg));
555 
556  // Recurse to the next argument.
557  return combine(length, buffer_ptr, buffer_end, args...);
558  }
559 
560  /// Base case for recursive, variadic combining.
561  ///
562  /// The base case when combining arguments recursively is reached when all
563  /// arguments have been handled. It flushes the remaining buffer and
564  /// constructs a hash_code.
565  hash_code combine(size_t length, char *buffer_ptr, char *buffer_end) {
566  // Check whether the entire set of values fit in the buffer. If so, we'll
567  // use the optimized short hashing routine and skip state entirely.
568  if (length == 0)
569  return hash_short(buffer, buffer_ptr - buffer, seed);
570 
571  // Mix the final buffer, rotating it if we did a partial fill in order to
572  // simulate doing a mix of the last 64-bytes. That is how the algorithm
573  // works when we have a contiguous byte sequence, and we want to emulate
574  // that here.
575  std::rotate(buffer, buffer_ptr, buffer_end);
576 
577  // Mix this chunk into the current state.
578  state.mix(buffer);
579  length += buffer_ptr - buffer;
580 
581  return state.finalize(length);
582  }
583 };
584 
585 } // namespace detail
586 } // namespace hashing
587 
588 /// Combine values into a single hash_code.
589 ///
590 /// This routine accepts a varying number of arguments of any type. It will
591 /// attempt to combine them into a single hash_code. For user-defined types it
592 /// attempts to call a \see hash_value overload (via ADL) for the type. For
593 /// integer and pointer types it directly combines their data into the
594 /// resulting hash_code.
595 ///
596 /// The result is suitable for returning from a user's hash_value
597 /// *implementation* for their user-defined type. Consumers of a type should
598 /// *not* call this routine, they should instead call 'hash_value'.
599 template <typename ...Ts> hash_code hash_combine(const Ts &...args) {
600  // Recursively hash each argument using a helper class.
602  return helper.combine(0, helper.buffer, helper.buffer + 64, args...);
603 }
604 
605 // Implementation details for implementations of hash_value overloads provided
606 // here.
607 namespace hashing {
608 namespace detail {
609 
610 /// Helper to hash the value of a single integer.
611 ///
612 /// Overloads for smaller integer types are not provided to ensure consistent
613 /// behavior in the presence of integral promotions. Essentially,
614 /// "hash_value('4')" and "hash_value('0' + 4)" should be the same.
615 inline hash_code hash_integer_value(uint64_t value) {
616  // Similar to hash_4to8_bytes but using a seed instead of length.
617  const uint64_t seed = get_execution_seed();
618  const char *s = reinterpret_cast<const char *>(&value);
619  const uint64_t a = fetch32(s);
620  return hash_16_bytes(seed + (a << 3), fetch32(s + 4));
621 }
622 
623 } // namespace detail
624 } // namespace hashing
625 
626 // Declared and documented above, but defined here so that any of the hashing
627 // infrastructure is available.
628 template <typename T>
629 typename std::enable_if<is_integral_or_enum<T>::value, hash_code>::type
630 hash_value(T value) {
632  static_cast<uint64_t>(value));
633 }
634 
635 // Declared and documented above, but defined here so that any of the hashing
636 // infrastructure is available.
637 template <typename T> hash_code hash_value(const T *ptr) {
639  reinterpret_cast<uintptr_t>(ptr));
640 }
641 
642 // Declared and documented above, but defined here so that any of the hashing
643 // infrastructure is available.
644 template <typename T, typename U>
645 hash_code hash_value(const std::pair<T, U> &arg) {
646  return hash_combine(arg.first, arg.second);
647 }
648 
649 // Declared and documented above, but defined here so that any of the hashing
650 // infrastructure is available.
651 template <typename T>
652 hash_code hash_value(const std::basic_string<T> &arg) {
653  return hash_combine_range(arg.begin(), arg.end());
654 }
655 
656 } // namespace llvm
657 
658 #endif
const_iterator end(StringRef path)
Get end iterator over path.
Definition: Path.cpp:233
void swapByteOrder(T &Value)
Helper class to manage the recursive combining of hash_combine arguments.
Definition: Hashing.h:493
This class represents lattice values for constants.
Definition: AllocatorList.h:23
static const uint64_t k0
Some primes between 2^63 and 2^64 for various uses.
Definition: Hashing.h:160
static const Metadata * get_hashable_data(const MDOperand &X)
Make MDOperand transparent for hashing.
uint64_t shift_mix(uint64_t val)
Definition: Hashing.h:173
constexpr bool IsBigEndianHost
Definition: SwapByteOrder.h:51
Definition: BitVector.h:937
Trait to indicate whether a type&#39;s bits can be hashed directly.
Definition: Hashing.h:344
uint64_t hash_17to32_bytes(const char *s, size_t len, uint64_t seed)
Definition: Hashing.h:208
hash_code(size_t value)
Form a hash code directly from a numerical value.
Definition: Hashing.h:79
uint64_t finalize(size_t length)
Compute the final 64-bit hash code value based on the current state and the length of bytes hashed...
Definition: Hashing.h:305
uint64_t fetch64(const char *p)
Definition: Hashing.h:143
static void mix_32_bytes(const char *s, uint64_t &a, uint64_t &b)
Mix 32-bytes from the input sequence into the 16-bytes of &#39;a&#39; and &#39;b&#39;, including whatever is already ...
Definition: Hashing.h:275
void set_fixed_execution_hash_seed(uint64_t fixed_value)
Override the execution seed with a fixed value.
Definition: Hashing.cpp:26
uint64_t hash_1to3_bytes(const char *s, size_t len, uint64_t seed)
Definition: Hashing.h:188
hash_code hash_value(const RegisterBankInfo::PartialMapping &PartMapping)
Hashing function for PartialMapping.
uint32_t fetch32(const char *p)
Definition: Hashing.h:151
uint64_t hash_16_bytes(uint64_t low, uint64_t high)
Definition: Hashing.h:177
hash_code combine(size_t length, char *buffer_ptr, char *buffer_end)
Base case for recursive, variadic combining.
Definition: Hashing.h:565
friend bool operator==(const hash_code &lhs, const hash_code &rhs)
Definition: Hashing.h:84
void mix(const char *s)
Mix in a 64-byte buffer of data.
Definition: Hashing.h:288
static hash_state create(const char *s, uint64_t seed)
Create a new hash_state structure and initialize it based on the seed and the first 64-byte chunk...
Definition: Hashing.h:264
uint64_t hash_4to8_bytes(const char *s, size_t len, uint64_t seed)
Definition: Hashing.h:197
hash_code hash_integer_value(uint64_t value)
Helper to hash the value of a single integer.
Definition: Hashing.h:615
hash_code()=default
Default construct a hash_code.
std::enable_if< is_hashable_data< ValueT >::value, hash_code >::type hash_combine_range_impl(ValueT *first, ValueT *last)
Implement the combining of integral values into a hash_code.
Definition: Hashing.h:445
unsigned first
static const uint64_t k1
Definition: Hashing.h:161
uint64_t hash_9to16_bytes(const char *s, size_t len, uint64_t seed)
Definition: Hashing.h:202
loop rotate
hash_combine_recursive_helper()
Construct a recursive hash combining helper.
Definition: Hashing.h:503
uint64_t hash_short(const char *s, size_t length, uint64_t seed)
Definition: Hashing.h:240
hash_code combine(size_t length, char *buffer_ptr, char *buffer_end, const T &arg, const Ts &...args)
Recursive, variadic combining method.
Definition: Hashing.h:552
char * combine_data(size_t &length, char *buffer_ptr, char *buffer_end, T data)
Combine one chunk of data into the current in-flight hash.
Definition: Hashing.h:513
uint64_t fixed_seed_override
A global, fixed seed-override variable.
Definition: Hashing.cpp:22
void swap(llvm::BitVector &LHS, llvm::BitVector &RHS)
Implement std::swap in terms of BitVector swap.
Definition: BitVector.h:940
static const uint64_t k2
Definition: Hashing.h:162
hash_code hash_combine(const Ts &...args)
Combine values into a single hash_code.
Definition: Hashing.h:599
hash_code hash_combine_range(InputIteratorT first, InputIteratorT last)
Compute a hash_code for a sequence of values.
Definition: Hashing.h:477
An opaque object representing a hash code.
Definition: Hashing.h:70
uint64_t get_execution_seed()
Definition: Hashing.h:319
bool store_and_advance(char *&buffer_ptr, char *buffer_end, const T &value, size_t offset=0)
Helper to store data from a value into a buffer and advance the pointer into that buffer...
Definition: Hashing.h:384
static const uint64_t k3
Definition: Hashing.h:163
The intermediate state used during hashing.
Definition: Hashing.h:258
assert(ImpDefSCC.getReg()==AMDGPU::SCC &&ImpDefSCC.isDef())
friend size_t hash_value(const hash_code &code)
Allow a hash_code to be directly run through hash_value.
Definition: Hashing.h:92
uint64_t hash_33to64_bytes(const char *s, size_t len, uint64_t seed)
Definition: Hashing.h:217
friend bool operator!=(const hash_code &lhs, const hash_code &rhs)
Definition: Hashing.h:87