aboutsummaryrefslogtreecommitdiff
path: root/include/clang/Basic/OnDiskHashTable.h
blob: f54d67042c472a7f1c95afb73d0cc58e8d86e339 (plain) (blame)
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
31
32
33
34
35
36
37
38
39
40
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
69
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
//===--- OnDiskHashTable.h - On-Disk Hash Table Implementation --*- C++ -*-===//
//
//                     The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
//  This file defines facilities for reading and writing on-disk hash
//  tables.
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_CLANG_BASIC_ON_DISK_HASH_TABLE_H
#define LLVM_CLANG_BASIC_ON_DISK_HASH_TABLE_H

#include "llvm/Support/Allocator.h"
#include "llvm/Support/Compiler.h"
#include "llvm/Support/DataTypes.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include "llvm/System/Host.h"
#include <cassert>
#include <cstdlib>

namespace clang {

// Bernstein hash function:
// This is basically copy-and-paste from StringMap.  This likely won't
// stay here, which is why I didn't both to expose this function from
// String Map.
inline unsigned BernsteinHash(const char* x) {  
  unsigned int R = 0;
  for ( ; *x != '\0' ; ++x) R = R * 33 + *x;
  return R + (R >> 5);
}

inline unsigned BernsteinHash(const char* x, unsigned n) {
  unsigned int R = 0;
  for (unsigned i = 0 ; i < n ; ++i, ++x) R = R * 33 + *x;
  return R + (R >> 5);
}

inline unsigned BernsteinHashPartial(const char* x, unsigned n, unsigned R) {
  for (unsigned i = 0 ; i < n ; ++i, ++x) R = R * 33 + *x;
  return R + (R >> 5);
}

namespace io {

typedef uint32_t Offset;

inline void Emit8(llvm::raw_ostream& Out, uint32_t V) {
  Out << (unsigned char)(V);
}

inline void Emit16(llvm::raw_ostream& Out, uint32_t V) {
  Out << (unsigned char)(V);
  Out << (unsigned char)(V >>  8);
  assert((V >> 16) == 0);
}

inline void Emit32(llvm::raw_ostream& Out, uint32_t V) {
  Out << (unsigned char)(V);
  Out << (unsigned char)(V >>  8);
  Out << (unsigned char)(V >> 16);
  Out << (unsigned char)(V >> 24);
}

inline void Emit64(llvm::raw_ostream& Out, uint64_t V) {
  Out << (unsigned char)(V);
  Out << (unsigned char)(V >>  8);
  Out << (unsigned char)(V >> 16);
  Out << (unsigned char)(V >> 24);
  Out << (unsigned char)(V >> 32);
  Out << (unsigned char)(V >> 40);
  Out << (unsigned char)(V >> 48);
  Out << (unsigned char)(V >> 56);
}

inline void Pad(llvm::raw_ostream& Out, unsigned A) {
  Offset off = (Offset) Out.tell();
  uint32_t n = ((uintptr_t)(off+A-1) & ~(uintptr_t)(A-1)) - off;
  for (; n ; --n)
    Emit8(Out, 0);
}

inline uint16_t ReadUnalignedLE16(const unsigned char *&Data) {
  uint16_t V = ((uint16_t)Data[0]) |
               ((uint16_t)Data[1] <<  8);
  Data += 2;
  return V;
}

inline uint32_t ReadUnalignedLE32(const unsigned char *&Data) {
  uint32_t V = ((uint32_t)Data[0])  |
               ((uint32_t)Data[1] << 8)  |
               ((uint32_t)Data[2] << 16) |
               ((uint32_t)Data[3] << 24);
  Data += 4;
  return V;
}

inline uint64_t ReadUnalignedLE64(const unsigned char *&Data) {
  uint64_t V = ((uint64_t)Data[0])  |
    ((uint64_t)Data[1] << 8)  |
    ((uint64_t)Data[2] << 16) |
    ((uint64_t)Data[3] << 24) |
    ((uint64_t)Data[4] << 32) |
    ((uint64_t)Data[5] << 40) |
    ((uint64_t)Data[6] << 48) |
    ((uint64_t)Data[7] << 56);
  Data += 8;
  return V;
}

inline uint32_t ReadLE32(const unsigned char *&Data) {
  // Hosts that directly support little-endian 32-bit loads can just
  // use them.  Big-endian hosts need a bswap.
  uint32_t V = *((uint32_t*)Data);
  if (llvm::sys::isBigEndianHost())
    V = llvm::ByteSwap_32(V);
  Data += 4;
  return V;
}

} // end namespace io

template<typename Info>
class OnDiskChainedHashTableGenerator {
  unsigned NumBuckets;
  unsigned NumEntries;
  llvm::BumpPtrAllocator BA;
  
  class Item {
  public:
    typename Info::key_type key;
    typename Info::data_type data;
    Item *next;
    const uint32_t hash;
    
    Item(typename Info::key_type_ref k, typename Info::data_type_ref d)
    : key(k), data(d), next(0), hash(Info::ComputeHash(k)) {}
  };
  
  class Bucket { 
  public:
    io::Offset off;
    Item*  head;
    unsigned length;
    
    Bucket() {}
  };
  
  Bucket* Buckets;
  
private:
  void insert(Bucket* b, size_t size, Item* E) {
    unsigned idx = E->hash & (size - 1);
    Bucket& B = b[idx];
    E->next = B.head;
    ++B.length;
    B.head = E;
  }
  
  void resize(size_t newsize) {
    Bucket* newBuckets = (Bucket*) std::calloc(newsize, sizeof(Bucket));
    // Populate newBuckets with the old entries.
    for (unsigned i = 0; i < NumBuckets; ++i)
      for (Item* E = Buckets[i].head; E ; ) {
        Item* N = E->next;
        E->next = 0;
        insert(newBuckets, newsize, E);
        E = N;
      }
    
    free(Buckets);
    NumBuckets = newsize;
    Buckets = newBuckets;
  }  
  
public:
  
  void insert(typename Info::key_type_ref key,
              typename Info::data_type_ref data) {

    ++NumEntries;
    if (4*NumEntries >= 3*NumBuckets) resize(NumBuckets*2);
    insert(Buckets, NumBuckets, new (BA.Allocate<Item>()) Item(key, data));
  }
  
  io::Offset Emit(llvm::raw_ostream &out) {
    Info InfoObj;
    return Emit(out, InfoObj);
  }

  io::Offset Emit(llvm::raw_ostream &out, Info &InfoObj) {
    using namespace clang::io;

    // Emit the payload of the table.
    for (unsigned i = 0; i < NumBuckets; ++i) {
      Bucket& B = Buckets[i];
      if (!B.head) continue;
      
      // Store the offset for the data of this bucket.
      B.off = out.tell();
      assert(B.off && "Cannot write a bucket at offset 0. Please add padding.");

      // Write out the number of items in the bucket.
      Emit16(out, B.length);
      
      // Write out the entries in the bucket.
      for (Item *I = B.head; I ; I = I->next) {
        Emit32(out, I->hash);
        const std::pair<unsigned, unsigned>& Len = 
          InfoObj.EmitKeyDataLength(out, I->key, I->data);
        InfoObj.EmitKey(out, I->key, Len.first);
        InfoObj.EmitData(out, I->key, I->data, Len.second);
      }
    }
    
    // Emit the hashtable itself.
    Pad(out, 4);
    io::Offset TableOff = out.tell();
    Emit32(out, NumBuckets);
    Emit32(out, NumEntries);
    for (unsigned i = 0; i < NumBuckets; ++i) Emit32(out, Buckets[i].off);
    
    return TableOff;
  }
  
  OnDiskChainedHashTableGenerator() {
    NumEntries = 0;
    NumBuckets = 64;    
    // Note that we do not need to run the constructors of the individual
    // Bucket objects since 'calloc' returns bytes that are all 0.
    Buckets = (Bucket*) std::calloc(NumBuckets, sizeof(Bucket));
  }
  
  ~OnDiskChainedHashTableGenerator() {
    std::free(Buckets);
  }
};

template<typename Info>
class OnDiskChainedHashTable {
  const unsigned NumBuckets;
  const unsigned NumEntries;
  const unsigned char* const Buckets;
  const unsigned char* const Base;
  Info InfoObj;

public:
  typedef typename Info::internal_key_type internal_key_type;
  typedef typename Info::external_key_type external_key_type;
  typedef typename Info::data_type         data_type;
  
  OnDiskChainedHashTable(unsigned numBuckets, unsigned numEntries,
                         const unsigned char* buckets,
                         const unsigned char* base,
                         const Info &InfoObj = Info())
    : NumBuckets(numBuckets), NumEntries(numEntries),
      Buckets(buckets), Base(base), InfoObj(InfoObj) {
        assert((reinterpret_cast<uintptr_t>(buckets) & 0x3) == 0 &&
               "'buckets' must have a 4-byte alignment");
      }

  unsigned getNumBuckets() const { return NumBuckets; }
  unsigned getNumEntries() const { return NumEntries; }
  const unsigned char* getBase() const { return Base; }
  const unsigned char* getBuckets() const { return Buckets; }

  bool isEmpty() const { return NumEntries == 0; }
  
  class iterator {
    internal_key_type key;
    const unsigned char* const data;
    const unsigned len;
    Info *InfoObj;
  public:
    iterator() : data(0), len(0) {}
    iterator(const internal_key_type k, const unsigned char* d, unsigned l,
             Info *InfoObj)
      : key(k), data(d), len(l), InfoObj(InfoObj) {}
    
    data_type operator*() const { return InfoObj->ReadData(key, data, len); }    
    bool operator==(const iterator& X) const { return X.data == data; }    
    bool operator!=(const iterator& X) const { return X.data != data; }
  };    
  
  iterator find(const external_key_type& eKey, Info *InfoPtr = 0) {
    if (!InfoPtr)
      InfoPtr = &InfoObj;

    using namespace io;
    const internal_key_type& iKey = Info::GetInternalKey(eKey);
    unsigned key_hash = Info::ComputeHash(iKey);
    
    // Each bucket is just a 32-bit offset into the hash table file.
    unsigned idx = key_hash & (NumBuckets - 1);
    const unsigned char* Bucket = Buckets + sizeof(uint32_t)*idx;
    
    unsigned offset = ReadLE32(Bucket);
    if (offset == 0) return iterator(); // Empty bucket.
    const unsigned char* Items = Base + offset;
    
    // 'Items' starts with a 16-bit unsigned integer representing the
    // number of items in this bucket.
    unsigned len = ReadUnalignedLE16(Items);
    
    for (unsigned i = 0; i < len; ++i) {
      // Read the hash.
      uint32_t item_hash = ReadUnalignedLE32(Items);
      
      // Determine the length of the key and the data.
      const std::pair<unsigned, unsigned>& L = Info::ReadKeyDataLength(Items);      
      unsigned item_len = L.first + L.second;

      // Compare the hashes.  If they are not the same, skip the entry entirely.
      if (item_hash != key_hash) {
        Items += item_len;
        continue;
      }
      
      // Read the key.
      const internal_key_type& X =
        InfoPtr->ReadKey((const unsigned char* const) Items, L.first);

      // If the key doesn't match just skip reading the value.
      if (!Info::EqualKey(X, iKey)) {
        Items += item_len;
        continue;
      }
      
      // The key matches!
      return iterator(X, Items + L.first, L.second, InfoPtr);
    }
    
    return iterator();
  }
  
  iterator end() const { return iterator(); }
  
  
  static OnDiskChainedHashTable* Create(const unsigned char* buckets,
                                        const unsigned char* const base,
                                        const Info &InfoObj = Info()) {
    using namespace io;
    assert(buckets > base);
    assert((reinterpret_cast<uintptr_t>(buckets) & 0x3) == 0 &&
           "buckets should be 4-byte aligned.");
    
    unsigned numBuckets = ReadLE32(buckets);
    unsigned numEntries = ReadLE32(buckets);
    return new OnDiskChainedHashTable<Info>(numBuckets, numEntries, buckets,
                                            base, InfoObj);
  }  
};

} // end namespace clang

#endif