kanayago 0.1.1
Sign up to get free protection for your applications and to get access to all the features.
- checksums.yaml +7 -0
- data/.rubocop.yml +15 -0
- data/.rubocop_todo.yml +23 -0
- data/LICENSE.txt +21 -0
- data/README.md +79 -0
- data/Rakefile +182 -0
- data/ext/kanayago/ccan/check_type/check_type.h +63 -0
- data/ext/kanayago/ccan/container_of/container_of.h +142 -0
- data/ext/kanayago/ccan/list/list.h +791 -0
- data/ext/kanayago/ccan/str/str.h +17 -0
- data/ext/kanayago/constant.h +53 -0
- data/ext/kanayago/extconf.rb +21 -0
- data/ext/kanayago/id.h +347 -0
- data/ext/kanayago/id_table.h +39 -0
- data/ext/kanayago/internal/array.h +151 -0
- data/ext/kanayago/internal/basic_operators.h +64 -0
- data/ext/kanayago/internal/bignum.h +244 -0
- data/ext/kanayago/internal/bits.h +568 -0
- data/ext/kanayago/internal/compile.h +34 -0
- data/ext/kanayago/internal/compilers.h +107 -0
- data/ext/kanayago/internal/complex.h +29 -0
- data/ext/kanayago/internal/encoding.h +36 -0
- data/ext/kanayago/internal/error.h +218 -0
- data/ext/kanayago/internal/fixnum.h +184 -0
- data/ext/kanayago/internal/gc.h +322 -0
- data/ext/kanayago/internal/hash.h +191 -0
- data/ext/kanayago/internal/imemo.h +261 -0
- data/ext/kanayago/internal/io.h +140 -0
- data/ext/kanayago/internal/numeric.h +274 -0
- data/ext/kanayago/internal/parse.h +117 -0
- data/ext/kanayago/internal/rational.h +71 -0
- data/ext/kanayago/internal/re.h +28 -0
- data/ext/kanayago/internal/ruby_parser.h +125 -0
- data/ext/kanayago/internal/sanitizers.h +297 -0
- data/ext/kanayago/internal/serial.h +23 -0
- data/ext/kanayago/internal/static_assert.h +16 -0
- data/ext/kanayago/internal/string.h +186 -0
- data/ext/kanayago/internal/symbol.h +45 -0
- data/ext/kanayago/internal/thread.h +79 -0
- data/ext/kanayago/internal/variable.h +72 -0
- data/ext/kanayago/internal/vm.h +137 -0
- data/ext/kanayago/internal/warnings.h +16 -0
- data/ext/kanayago/internal.h +108 -0
- data/ext/kanayago/kanayago.c +420 -0
- data/ext/kanayago/kanayago.h +21 -0
- data/ext/kanayago/lex.c +302 -0
- data/ext/kanayago/method.h +255 -0
- data/ext/kanayago/node.c +440 -0
- data/ext/kanayago/node.h +111 -0
- data/ext/kanayago/node_name.inc +224 -0
- data/ext/kanayago/parse.c +26931 -0
- data/ext/kanayago/parse.h +244 -0
- data/ext/kanayago/parse.tmp.y +16145 -0
- data/ext/kanayago/parser_bits.h +564 -0
- data/ext/kanayago/parser_node.h +32 -0
- data/ext/kanayago/parser_st.c +164 -0
- data/ext/kanayago/parser_st.h +162 -0
- data/ext/kanayago/parser_value.h +106 -0
- data/ext/kanayago/probes.h +4 -0
- data/ext/kanayago/ruby_assert.h +14 -0
- data/ext/kanayago/ruby_atomic.h +23 -0
- data/ext/kanayago/ruby_parser.c +1165 -0
- data/ext/kanayago/rubyparser.h +1391 -0
- data/ext/kanayago/shape.h +234 -0
- data/ext/kanayago/st.c +2339 -0
- data/ext/kanayago/symbol.h +123 -0
- data/ext/kanayago/thread_pthread.h +168 -0
- data/ext/kanayago/universal_parser.c +230 -0
- data/ext/kanayago/vm_core.h +2215 -0
- data/ext/kanayago/vm_opts.h +67 -0
- data/lib/kanayago/version.rb +5 -0
- data/lib/kanayago.rb +11 -0
- data/sig/kanayago.rbs +4 -0
- metadata +116 -0
data/ext/kanayago/st.c
ADDED
@@ -0,0 +1,2339 @@
|
|
1
|
+
/* This is a public domain general purpose hash table package
|
2
|
+
originally written by Peter Moore @ UCB.
|
3
|
+
|
4
|
+
The hash table data structures were redesigned and the package was
|
5
|
+
rewritten by Vladimir Makarov <vmakarov@redhat.com>. */
|
6
|
+
|
7
|
+
/* The original package implemented classic bucket-based hash tables
|
8
|
+
with entries doubly linked for an access by their insertion order.
|
9
|
+
To decrease pointer chasing and as a consequence to improve a data
|
10
|
+
locality the current implementation is based on storing entries in
|
11
|
+
an array and using hash tables with open addressing. The current
|
12
|
+
entries are more compact in comparison with the original ones and
|
13
|
+
this also improves the data locality.
|
14
|
+
|
15
|
+
The hash table has two arrays called *bins* and *entries*.
|
16
|
+
|
17
|
+
bins:
|
18
|
+
-------
|
19
|
+
| | entries array:
|
20
|
+
|-------| --------------------------------
|
21
|
+
| index | | | entry: | | |
|
22
|
+
|-------| | | | | |
|
23
|
+
| ... | | ... | hash | ... | ... |
|
24
|
+
|-------| | | key | | |
|
25
|
+
| empty | | | record | | |
|
26
|
+
|-------| --------------------------------
|
27
|
+
| ... | ^ ^
|
28
|
+
|-------| |_ entries start |_ entries bound
|
29
|
+
|deleted|
|
30
|
+
-------
|
31
|
+
|
32
|
+
o The entry array contains table entries in the same order as they
|
33
|
+
were inserted.
|
34
|
+
|
35
|
+
When the first entry is deleted, a variable containing index of
|
36
|
+
the current first entry (*entries start*) is changed. In all
|
37
|
+
other cases of the deletion, we just mark the entry as deleted by
|
38
|
+
using a reserved hash value.
|
39
|
+
|
40
|
+
Such organization of the entry storage makes operations of the
|
41
|
+
table shift and the entries traversal very fast.
|
42
|
+
|
43
|
+
o The bins provide access to the entries by their keys. The
|
44
|
+
key hash is mapped to a bin containing *index* of the
|
45
|
+
corresponding entry in the entry array.
|
46
|
+
|
47
|
+
The bin array size is always power of two, it makes mapping very
|
48
|
+
fast by using the corresponding lower bits of the hash.
|
49
|
+
Generally it is not a good idea to ignore some part of the hash.
|
50
|
+
But alternative approach is worse. For example, we could use a
|
51
|
+
modulo operation for mapping and a prime number for the size of
|
52
|
+
the bin array. Unfortunately, the modulo operation for big
|
53
|
+
64-bit numbers are extremely slow (it takes more than 100 cycles
|
54
|
+
on modern Intel CPUs).
|
55
|
+
|
56
|
+
Still other bits of the hash value are used when the mapping
|
57
|
+
results in a collision. In this case we use a secondary hash
|
58
|
+
value which is a result of a function of the collision bin
|
59
|
+
index and the original hash value. The function choice
|
60
|
+
guarantees that we can traverse all bins and finally find the
|
61
|
+
corresponding bin as after several iterations the function
|
62
|
+
becomes a full cycle linear congruential generator because it
|
63
|
+
satisfies requirements of the Hull-Dobell theorem.
|
64
|
+
|
65
|
+
When an entry is removed from the table besides marking the
|
66
|
+
hash in the corresponding entry described above, we also mark
|
67
|
+
the bin by a special value in order to find entries which had
|
68
|
+
a collision with the removed entries.
|
69
|
+
|
70
|
+
There are two reserved values for the bins. One denotes an
|
71
|
+
empty bin, another one denotes a bin for a deleted entry.
|
72
|
+
|
73
|
+
o The length of the bin array is at least two times more than the
|
74
|
+
entry array length. This keeps the table load factor healthy.
|
75
|
+
The trigger of rebuilding the table is always a case when we can
|
76
|
+
not insert an entry anymore at the entries bound. We could
|
77
|
+
change the entries bound too in case of deletion but than we need
|
78
|
+
a special code to count bins with corresponding deleted entries
|
79
|
+
and reset the bin values when there are too many bins
|
80
|
+
corresponding deleted entries
|
81
|
+
|
82
|
+
Table rebuilding is done by creation of a new entry array and
|
83
|
+
bins of an appropriate size. We also try to reuse the arrays
|
84
|
+
in some cases by compacting the array and removing deleted
|
85
|
+
entries.
|
86
|
+
|
87
|
+
o To save memory very small tables have no allocated arrays
|
88
|
+
bins. We use a linear search for an access by a key.
|
89
|
+
|
90
|
+
o To save more memory we use 8-, 16-, 32- and 64- bit indexes in
|
91
|
+
bins depending on the current hash table size.
|
92
|
+
|
93
|
+
o The implementation takes into account that the table can be
|
94
|
+
rebuilt during hashing or comparison functions. It can happen if
|
95
|
+
the functions are implemented in Ruby and a thread switch occurs
|
96
|
+
during their execution.
|
97
|
+
|
98
|
+
This implementation speeds up the Ruby hash table benchmarks in
|
99
|
+
average by more 40% on Intel Haswell CPU.
|
100
|
+
|
101
|
+
*/
|
102
|
+
|
103
|
+
#ifdef NOT_RUBY
|
104
|
+
#include "regint.h"
|
105
|
+
#include "st.h"
|
106
|
+
#elif defined RUBY_EXPORT
|
107
|
+
#include "internal.h"
|
108
|
+
#include "internal/bits.h"
|
109
|
+
#include "internal/hash.h"
|
110
|
+
#include "internal/sanitizers.h"
|
111
|
+
#include "internal/st.h"
|
112
|
+
#endif
|
113
|
+
|
114
|
+
#include <stdio.h>
|
115
|
+
#ifdef HAVE_STDLIB_H
|
116
|
+
#include <stdlib.h>
|
117
|
+
#endif
|
118
|
+
#include <string.h>
|
119
|
+
#include <assert.h>
|
120
|
+
|
121
|
+
#ifdef __GNUC__
|
122
|
+
#define PREFETCH(addr, write_p) __builtin_prefetch(addr, write_p)
|
123
|
+
#define EXPECT(expr, val) __builtin_expect(expr, val)
|
124
|
+
#define ATTRIBUTE_UNUSED __attribute__((unused))
|
125
|
+
#else
|
126
|
+
#define PREFETCH(addr, write_p)
|
127
|
+
#define EXPECT(expr, val) (expr)
|
128
|
+
#define ATTRIBUTE_UNUSED
|
129
|
+
#endif
|
130
|
+
|
131
|
+
/* The type of hashes. */
|
132
|
+
typedef st_index_t st_hash_t;
|
133
|
+
|
134
|
+
struct st_table_entry {
|
135
|
+
st_hash_t hash;
|
136
|
+
st_data_t key;
|
137
|
+
st_data_t record;
|
138
|
+
};
|
139
|
+
|
140
|
+
#define type_numhash st_hashtype_num
|
141
|
+
static const struct st_hash_type st_hashtype_num = {
|
142
|
+
st_numcmp,
|
143
|
+
st_numhash,
|
144
|
+
};
|
145
|
+
|
146
|
+
static int st_strcmp(st_data_t, st_data_t);
|
147
|
+
static st_index_t strhash(st_data_t);
|
148
|
+
static const struct st_hash_type type_strhash = {
|
149
|
+
st_strcmp,
|
150
|
+
strhash,
|
151
|
+
};
|
152
|
+
|
153
|
+
static int st_locale_insensitive_strcasecmp_i(st_data_t lhs, st_data_t rhs);
|
154
|
+
static st_index_t strcasehash(st_data_t);
|
155
|
+
static const struct st_hash_type type_strcasehash = {
|
156
|
+
st_locale_insensitive_strcasecmp_i,
|
157
|
+
strcasehash,
|
158
|
+
};
|
159
|
+
|
160
|
+
/* Value used to catch uninitialized entries/bins during debugging.
|
161
|
+
There is a possibility for a false alarm, but its probability is
|
162
|
+
extremely small. */
|
163
|
+
#define ST_INIT_VAL 0xafafafafafafafaf
|
164
|
+
#define ST_INIT_VAL_BYTE 0xafa
|
165
|
+
|
166
|
+
#ifdef RUBY
|
167
|
+
#undef malloc
|
168
|
+
#undef realloc
|
169
|
+
#undef calloc
|
170
|
+
#undef free
|
171
|
+
#define malloc ruby_xmalloc
|
172
|
+
#define calloc ruby_xcalloc
|
173
|
+
#define realloc ruby_xrealloc
|
174
|
+
#define free ruby_xfree
|
175
|
+
#endif
|
176
|
+
|
177
|
+
#define EQUAL(tab,x,y) ((x) == (y) || (*(tab)->type->compare)((x),(y)) == 0)
|
178
|
+
#define PTR_EQUAL(tab, ptr, hash_val, key_) \
|
179
|
+
((ptr)->hash == (hash_val) && EQUAL((tab), (key_), (ptr)->key))
|
180
|
+
|
181
|
+
/* As PTR_EQUAL only its result is returned in RES. REBUILT_P is set
|
182
|
+
up to TRUE if the table is rebuilt during the comparison. */
|
183
|
+
#define DO_PTR_EQUAL_CHECK(tab, ptr, hash_val, key, res, rebuilt_p) \
|
184
|
+
do { \
|
185
|
+
unsigned int _old_rebuilds_num = (tab)->rebuilds_num; \
|
186
|
+
res = PTR_EQUAL(tab, ptr, hash_val, key); \
|
187
|
+
rebuilt_p = _old_rebuilds_num != (tab)->rebuilds_num; \
|
188
|
+
} while (FALSE)
|
189
|
+
|
190
|
+
/* Features of a table. */
|
191
|
+
struct st_features {
|
192
|
+
/* Power of 2 used for number of allocated entries. */
|
193
|
+
unsigned char entry_power;
|
194
|
+
/* Power of 2 used for number of allocated bins. Depending on the
|
195
|
+
table size, the number of bins is 2-4 times more than the
|
196
|
+
number of entries. */
|
197
|
+
unsigned char bin_power;
|
198
|
+
/* Enumeration of sizes of bins (8-bit, 16-bit etc). */
|
199
|
+
unsigned char size_ind;
|
200
|
+
/* Bins are packed in words of type st_index_t. The following is
|
201
|
+
a size of bins counted by words. */
|
202
|
+
st_index_t bins_words;
|
203
|
+
};
|
204
|
+
|
205
|
+
/* Features of all possible size tables. */
|
206
|
+
#if SIZEOF_ST_INDEX_T == 8
|
207
|
+
#define MAX_POWER2 62
|
208
|
+
static const struct st_features features[] = {
|
209
|
+
{0, 1, 0, 0x0},
|
210
|
+
{1, 2, 0, 0x1},
|
211
|
+
{2, 3, 0, 0x1},
|
212
|
+
{3, 4, 0, 0x2},
|
213
|
+
{4, 5, 0, 0x4},
|
214
|
+
{5, 6, 0, 0x8},
|
215
|
+
{6, 7, 0, 0x10},
|
216
|
+
{7, 8, 0, 0x20},
|
217
|
+
{8, 9, 1, 0x80},
|
218
|
+
{9, 10, 1, 0x100},
|
219
|
+
{10, 11, 1, 0x200},
|
220
|
+
{11, 12, 1, 0x400},
|
221
|
+
{12, 13, 1, 0x800},
|
222
|
+
{13, 14, 1, 0x1000},
|
223
|
+
{14, 15, 1, 0x2000},
|
224
|
+
{15, 16, 1, 0x4000},
|
225
|
+
{16, 17, 2, 0x10000},
|
226
|
+
{17, 18, 2, 0x20000},
|
227
|
+
{18, 19, 2, 0x40000},
|
228
|
+
{19, 20, 2, 0x80000},
|
229
|
+
{20, 21, 2, 0x100000},
|
230
|
+
{21, 22, 2, 0x200000},
|
231
|
+
{22, 23, 2, 0x400000},
|
232
|
+
{23, 24, 2, 0x800000},
|
233
|
+
{24, 25, 2, 0x1000000},
|
234
|
+
{25, 26, 2, 0x2000000},
|
235
|
+
{26, 27, 2, 0x4000000},
|
236
|
+
{27, 28, 2, 0x8000000},
|
237
|
+
{28, 29, 2, 0x10000000},
|
238
|
+
{29, 30, 2, 0x20000000},
|
239
|
+
{30, 31, 2, 0x40000000},
|
240
|
+
{31, 32, 2, 0x80000000},
|
241
|
+
{32, 33, 3, 0x200000000},
|
242
|
+
{33, 34, 3, 0x400000000},
|
243
|
+
{34, 35, 3, 0x800000000},
|
244
|
+
{35, 36, 3, 0x1000000000},
|
245
|
+
{36, 37, 3, 0x2000000000},
|
246
|
+
{37, 38, 3, 0x4000000000},
|
247
|
+
{38, 39, 3, 0x8000000000},
|
248
|
+
{39, 40, 3, 0x10000000000},
|
249
|
+
{40, 41, 3, 0x20000000000},
|
250
|
+
{41, 42, 3, 0x40000000000},
|
251
|
+
{42, 43, 3, 0x80000000000},
|
252
|
+
{43, 44, 3, 0x100000000000},
|
253
|
+
{44, 45, 3, 0x200000000000},
|
254
|
+
{45, 46, 3, 0x400000000000},
|
255
|
+
{46, 47, 3, 0x800000000000},
|
256
|
+
{47, 48, 3, 0x1000000000000},
|
257
|
+
{48, 49, 3, 0x2000000000000},
|
258
|
+
{49, 50, 3, 0x4000000000000},
|
259
|
+
{50, 51, 3, 0x8000000000000},
|
260
|
+
{51, 52, 3, 0x10000000000000},
|
261
|
+
{52, 53, 3, 0x20000000000000},
|
262
|
+
{53, 54, 3, 0x40000000000000},
|
263
|
+
{54, 55, 3, 0x80000000000000},
|
264
|
+
{55, 56, 3, 0x100000000000000},
|
265
|
+
{56, 57, 3, 0x200000000000000},
|
266
|
+
{57, 58, 3, 0x400000000000000},
|
267
|
+
{58, 59, 3, 0x800000000000000},
|
268
|
+
{59, 60, 3, 0x1000000000000000},
|
269
|
+
{60, 61, 3, 0x2000000000000000},
|
270
|
+
{61, 62, 3, 0x4000000000000000},
|
271
|
+
{62, 63, 3, 0x8000000000000000},
|
272
|
+
};
|
273
|
+
|
274
|
+
#else
|
275
|
+
#define MAX_POWER2 30
|
276
|
+
|
277
|
+
static const struct st_features features[] = {
|
278
|
+
{0, 1, 0, 0x1},
|
279
|
+
{1, 2, 0, 0x1},
|
280
|
+
{2, 3, 0, 0x2},
|
281
|
+
{3, 4, 0, 0x4},
|
282
|
+
{4, 5, 0, 0x8},
|
283
|
+
{5, 6, 0, 0x10},
|
284
|
+
{6, 7, 0, 0x20},
|
285
|
+
{7, 8, 0, 0x40},
|
286
|
+
{8, 9, 1, 0x100},
|
287
|
+
{9, 10, 1, 0x200},
|
288
|
+
{10, 11, 1, 0x400},
|
289
|
+
{11, 12, 1, 0x800},
|
290
|
+
{12, 13, 1, 0x1000},
|
291
|
+
{13, 14, 1, 0x2000},
|
292
|
+
{14, 15, 1, 0x4000},
|
293
|
+
{15, 16, 1, 0x8000},
|
294
|
+
{16, 17, 2, 0x20000},
|
295
|
+
{17, 18, 2, 0x40000},
|
296
|
+
{18, 19, 2, 0x80000},
|
297
|
+
{19, 20, 2, 0x100000},
|
298
|
+
{20, 21, 2, 0x200000},
|
299
|
+
{21, 22, 2, 0x400000},
|
300
|
+
{22, 23, 2, 0x800000},
|
301
|
+
{23, 24, 2, 0x1000000},
|
302
|
+
{24, 25, 2, 0x2000000},
|
303
|
+
{25, 26, 2, 0x4000000},
|
304
|
+
{26, 27, 2, 0x8000000},
|
305
|
+
{27, 28, 2, 0x10000000},
|
306
|
+
{28, 29, 2, 0x20000000},
|
307
|
+
{29, 30, 2, 0x40000000},
|
308
|
+
{30, 31, 2, 0x80000000},
|
309
|
+
};
|
310
|
+
|
311
|
+
#endif
|
312
|
+
|
313
|
+
/* The reserved hash value and its substitution. */
|
314
|
+
#define RESERVED_HASH_VAL (~(st_hash_t) 0)
|
315
|
+
#define RESERVED_HASH_SUBSTITUTION_VAL ((st_hash_t) 0)
|
316
|
+
|
317
|
+
/* Return hash value of KEY for table TAB. */
|
318
|
+
static inline st_hash_t
|
319
|
+
do_hash(st_data_t key, st_table *tab)
|
320
|
+
{
|
321
|
+
st_hash_t hash = (st_hash_t)(tab->type->hash)(key);
|
322
|
+
|
323
|
+
/* RESERVED_HASH_VAL is used for a deleted entry. Map it into
|
324
|
+
another value. Such mapping should be extremely rare. */
|
325
|
+
return hash == RESERVED_HASH_VAL ? RESERVED_HASH_SUBSTITUTION_VAL : hash;
|
326
|
+
}
|
327
|
+
|
328
|
+
/* Power of 2 defining the minimal number of allocated entries. */
|
329
|
+
#define MINIMAL_POWER2 2
|
330
|
+
|
331
|
+
#if MINIMAL_POWER2 < 2
|
332
|
+
#error "MINIMAL_POWER2 should be >= 2"
|
333
|
+
#endif
|
334
|
+
|
335
|
+
/* If the power2 of the allocated `entries` is less than the following
|
336
|
+
value, don't allocate bins and use a linear search. */
|
337
|
+
#define MAX_POWER2_FOR_TABLES_WITHOUT_BINS 4
|
338
|
+
|
339
|
+
/* Return smallest n >= MINIMAL_POWER2 such 2^n > SIZE. */
|
340
|
+
static int
|
341
|
+
get_power2(st_index_t size)
|
342
|
+
{
|
343
|
+
unsigned int n = ST_INDEX_BITS - nlz_intptr(size);
|
344
|
+
if (n <= MAX_POWER2)
|
345
|
+
return n < MINIMAL_POWER2 ? MINIMAL_POWER2 : n;
|
346
|
+
#ifdef RUBY
|
347
|
+
/* Ran out of the table entries */
|
348
|
+
rb_raise(rb_eRuntimeError, "st_table too big");
|
349
|
+
#endif
|
350
|
+
/* should raise exception */
|
351
|
+
return -1;
|
352
|
+
}
|
353
|
+
|
354
|
+
/* Return value of N-th bin in array BINS of table with bins size
|
355
|
+
index S. */
|
356
|
+
static inline st_index_t
|
357
|
+
get_bin(st_index_t *bins, int s, st_index_t n)
|
358
|
+
{
|
359
|
+
return (s == 0 ? ((unsigned char *) bins)[n]
|
360
|
+
: s == 1 ? ((unsigned short *) bins)[n]
|
361
|
+
: s == 2 ? ((unsigned int *) bins)[n]
|
362
|
+
: ((st_index_t *) bins)[n]);
|
363
|
+
}
|
364
|
+
|
365
|
+
/* Set up N-th bin in array BINS of table with bins size index S to
|
366
|
+
value V. */
|
367
|
+
static inline void
|
368
|
+
set_bin(st_index_t *bins, int s, st_index_t n, st_index_t v)
|
369
|
+
{
|
370
|
+
if (s == 0) ((unsigned char *) bins)[n] = (unsigned char) v;
|
371
|
+
else if (s == 1) ((unsigned short *) bins)[n] = (unsigned short) v;
|
372
|
+
else if (s == 2) ((unsigned int *) bins)[n] = (unsigned int) v;
|
373
|
+
else ((st_index_t *) bins)[n] = v;
|
374
|
+
}
|
375
|
+
|
376
|
+
/* These macros define reserved values for empty table bin and table
|
377
|
+
bin which contains a deleted entry. We will never use such values
|
378
|
+
for an entry index in bins. */
|
379
|
+
#define EMPTY_BIN 0
|
380
|
+
#define DELETED_BIN 1
|
381
|
+
/* Base of a real entry index in the bins. */
|
382
|
+
#define ENTRY_BASE 2
|
383
|
+
|
384
|
+
/* Mark I-th bin of table TAB as empty, in other words not
|
385
|
+
corresponding to any entry. */
|
386
|
+
#define MARK_BIN_EMPTY(tab, i) (set_bin((tab)->bins, get_size_ind(tab), i, EMPTY_BIN))
|
387
|
+
|
388
|
+
/* Values used for not found entry and bin with given
|
389
|
+
characteristics. */
|
390
|
+
#define UNDEFINED_ENTRY_IND (~(st_index_t) 0)
|
391
|
+
#define UNDEFINED_BIN_IND (~(st_index_t) 0)
|
392
|
+
|
393
|
+
/* Entry and bin values returned when we found a table rebuild during
|
394
|
+
the search. */
|
395
|
+
#define REBUILT_TABLE_ENTRY_IND (~(st_index_t) 1)
|
396
|
+
#define REBUILT_TABLE_BIN_IND (~(st_index_t) 1)
|
397
|
+
|
398
|
+
/* Mark I-th bin of table TAB as corresponding to a deleted table
|
399
|
+
entry. Update number of entries in the table and number of bins
|
400
|
+
corresponding to deleted entries. */
|
401
|
+
#define MARK_BIN_DELETED(tab, i) \
|
402
|
+
do { \
|
403
|
+
set_bin((tab)->bins, get_size_ind(tab), i, DELETED_BIN); \
|
404
|
+
} while (0)
|
405
|
+
|
406
|
+
/* Macros to check that value B is used empty bins and bins
|
407
|
+
corresponding deleted entries. */
|
408
|
+
#define EMPTY_BIN_P(b) ((b) == EMPTY_BIN)
|
409
|
+
#define DELETED_BIN_P(b) ((b) == DELETED_BIN)
|
410
|
+
#define EMPTY_OR_DELETED_BIN_P(b) ((b) <= DELETED_BIN)
|
411
|
+
|
412
|
+
/* Macros to check empty bins and bins corresponding to deleted
|
413
|
+
entries. Bins are given by their index I in table TAB. */
|
414
|
+
#define IND_EMPTY_BIN_P(tab, i) (EMPTY_BIN_P(get_bin((tab)->bins, get_size_ind(tab), i)))
|
415
|
+
#define IND_DELETED_BIN_P(tab, i) (DELETED_BIN_P(get_bin((tab)->bins, get_size_ind(tab), i)))
|
416
|
+
#define IND_EMPTY_OR_DELETED_BIN_P(tab, i) (EMPTY_OR_DELETED_BIN_P(get_bin((tab)->bins, get_size_ind(tab), i)))
|
417
|
+
|
418
|
+
/* Macros for marking and checking deleted entries given by their
|
419
|
+
pointer E_PTR. */
|
420
|
+
#define MARK_ENTRY_DELETED(e_ptr) ((e_ptr)->hash = RESERVED_HASH_VAL)
|
421
|
+
#define DELETED_ENTRY_P(e_ptr) ((e_ptr)->hash == RESERVED_HASH_VAL)
|
422
|
+
|
423
|
+
/* Return bin size index of table TAB. */
|
424
|
+
static inline unsigned int
|
425
|
+
get_size_ind(const st_table *tab)
|
426
|
+
{
|
427
|
+
return tab->size_ind;
|
428
|
+
}
|
429
|
+
|
430
|
+
/* Return the number of allocated bins of table TAB. */
|
431
|
+
static inline st_index_t
|
432
|
+
get_bins_num(const st_table *tab)
|
433
|
+
{
|
434
|
+
return ((st_index_t) 1)<<tab->bin_power;
|
435
|
+
}
|
436
|
+
|
437
|
+
/* Return mask for a bin index in table TAB. */
|
438
|
+
static inline st_index_t
|
439
|
+
bins_mask(const st_table *tab)
|
440
|
+
{
|
441
|
+
return get_bins_num(tab) - 1;
|
442
|
+
}
|
443
|
+
|
444
|
+
/* Return the index of table TAB bin corresponding to
|
445
|
+
HASH_VALUE. */
|
446
|
+
static inline st_index_t
|
447
|
+
hash_bin(st_hash_t hash_value, st_table *tab)
|
448
|
+
{
|
449
|
+
return hash_value & bins_mask(tab);
|
450
|
+
}
|
451
|
+
|
452
|
+
/* Return the number of allocated entries of table TAB. */
|
453
|
+
static inline st_index_t
|
454
|
+
get_allocated_entries(const st_table *tab)
|
455
|
+
{
|
456
|
+
return ((st_index_t) 1)<<tab->entry_power;
|
457
|
+
}
|
458
|
+
|
459
|
+
/* Return size of the allocated bins of table TAB. */
|
460
|
+
static inline st_index_t
|
461
|
+
bins_size(const st_table *tab)
|
462
|
+
{
|
463
|
+
return features[tab->entry_power].bins_words * sizeof (st_index_t);
|
464
|
+
}
|
465
|
+
|
466
|
+
/* Mark all bins of table TAB as empty. */
|
467
|
+
static void
|
468
|
+
initialize_bins(st_table *tab)
|
469
|
+
{
|
470
|
+
memset(tab->bins, 0, bins_size(tab));
|
471
|
+
}
|
472
|
+
|
473
|
+
/* Make table TAB empty. */
|
474
|
+
static void
|
475
|
+
make_tab_empty(st_table *tab)
|
476
|
+
{
|
477
|
+
tab->num_entries = 0;
|
478
|
+
tab->entries_start = tab->entries_bound = 0;
|
479
|
+
if (tab->bins != NULL)
|
480
|
+
initialize_bins(tab);
|
481
|
+
}
|
482
|
+
|
483
|
+
#ifdef HASH_LOG
|
484
|
+
#ifdef HAVE_UNISTD_H
|
485
|
+
#include <unistd.h>
|
486
|
+
#endif
|
487
|
+
static struct {
|
488
|
+
int all, total, num, str, strcase;
|
489
|
+
} collision;
|
490
|
+
|
491
|
+
/* Flag switching off output of package statistics at the end of
|
492
|
+
program. */
|
493
|
+
static int init_st = 0;
|
494
|
+
|
495
|
+
/* Output overall number of table searches and collisions into a
|
496
|
+
temporary file. */
|
497
|
+
static void
|
498
|
+
stat_col(void)
|
499
|
+
{
|
500
|
+
char fname[10+sizeof(long)*3];
|
501
|
+
FILE *f;
|
502
|
+
if (!collision.total) return;
|
503
|
+
f = fopen((snprintf(fname, sizeof(fname), "/tmp/col%ld", (long)getpid()), fname), "w");
|
504
|
+
if (f == NULL)
|
505
|
+
return;
|
506
|
+
fprintf(f, "collision: %d / %d (%6.2f)\n", collision.all, collision.total,
|
507
|
+
((double)collision.all / (collision.total)) * 100);
|
508
|
+
fprintf(f, "num: %d, str: %d, strcase: %d\n", collision.num, collision.str, collision.strcase);
|
509
|
+
fclose(f);
|
510
|
+
}
|
511
|
+
#endif
|
512
|
+
|
513
|
+
st_table *
|
514
|
+
st_init_existing_table_with_size(st_table *tab, const struct st_hash_type *type, st_index_t size)
|
515
|
+
{
|
516
|
+
int n;
|
517
|
+
|
518
|
+
#ifdef HASH_LOG
|
519
|
+
#if HASH_LOG+0 < 0
|
520
|
+
{
|
521
|
+
const char *e = getenv("ST_HASH_LOG");
|
522
|
+
if (!e || !*e) init_st = 1;
|
523
|
+
}
|
524
|
+
#endif
|
525
|
+
if (init_st == 0) {
|
526
|
+
init_st = 1;
|
527
|
+
atexit(stat_col);
|
528
|
+
}
|
529
|
+
#endif
|
530
|
+
|
531
|
+
n = get_power2(size);
|
532
|
+
#ifndef RUBY
|
533
|
+
if (n < 0)
|
534
|
+
return NULL;
|
535
|
+
#endif
|
536
|
+
|
537
|
+
tab->type = type;
|
538
|
+
tab->entry_power = n;
|
539
|
+
tab->bin_power = features[n].bin_power;
|
540
|
+
tab->size_ind = features[n].size_ind;
|
541
|
+
if (n <= MAX_POWER2_FOR_TABLES_WITHOUT_BINS)
|
542
|
+
tab->bins = NULL;
|
543
|
+
else {
|
544
|
+
tab->bins = (st_index_t *) malloc(bins_size(tab));
|
545
|
+
#ifndef RUBY
|
546
|
+
if (tab->bins == NULL) {
|
547
|
+
free(tab);
|
548
|
+
return NULL;
|
549
|
+
}
|
550
|
+
#endif
|
551
|
+
}
|
552
|
+
tab->entries = (st_table_entry *) malloc(get_allocated_entries(tab)
|
553
|
+
* sizeof(st_table_entry));
|
554
|
+
#ifndef RUBY
|
555
|
+
if (tab->entries == NULL) {
|
556
|
+
st_free_table(tab);
|
557
|
+
return NULL;
|
558
|
+
}
|
559
|
+
#endif
|
560
|
+
make_tab_empty(tab);
|
561
|
+
tab->rebuilds_num = 0;
|
562
|
+
return tab;
|
563
|
+
}
|
564
|
+
|
565
|
+
/* Create and return table with TYPE which can hold at least SIZE
|
566
|
+
entries. The real number of entries which the table can hold is
|
567
|
+
the nearest power of two for SIZE. */
|
568
|
+
st_table *
|
569
|
+
st_init_table_with_size(const struct st_hash_type *type, st_index_t size)
|
570
|
+
{
|
571
|
+
st_table *tab = malloc(sizeof(st_table));
|
572
|
+
#ifndef RUBY
|
573
|
+
if (tab == NULL)
|
574
|
+
return NULL;
|
575
|
+
#endif
|
576
|
+
|
577
|
+
#ifdef RUBY
|
578
|
+
st_init_existing_table_with_size(tab, type, size);
|
579
|
+
#else
|
580
|
+
if (st_init_existing_table_with_size(tab, type, size) == NULL) {
|
581
|
+
free(tab);
|
582
|
+
return NULL;
|
583
|
+
}
|
584
|
+
#endif
|
585
|
+
|
586
|
+
return tab;
|
587
|
+
}
|
588
|
+
|
589
|
+
size_t
|
590
|
+
st_table_size(const struct st_table *tbl)
|
591
|
+
{
|
592
|
+
return tbl->num_entries;
|
593
|
+
}
|
594
|
+
|
595
|
+
/* Create and return table with TYPE which can hold a minimal number
|
596
|
+
of entries (see comments for get_power2). */
|
597
|
+
st_table *
|
598
|
+
st_init_table(const struct st_hash_type *type)
|
599
|
+
{
|
600
|
+
return st_init_table_with_size(type, 0);
|
601
|
+
}
|
602
|
+
|
603
|
+
/* Create and return table which can hold a minimal number of
|
604
|
+
numbers. */
|
605
|
+
st_table *
|
606
|
+
st_init_numtable(void)
|
607
|
+
{
|
608
|
+
return st_init_table(&type_numhash);
|
609
|
+
}
|
610
|
+
|
611
|
+
/* Create and return table which can hold SIZE numbers. */
|
612
|
+
st_table *
|
613
|
+
st_init_numtable_with_size(st_index_t size)
|
614
|
+
{
|
615
|
+
return st_init_table_with_size(&type_numhash, size);
|
616
|
+
}
|
617
|
+
|
618
|
+
/* Create and return table which can hold a minimal number of
|
619
|
+
strings. */
|
620
|
+
st_table *
|
621
|
+
st_init_strtable(void)
|
622
|
+
{
|
623
|
+
return st_init_table(&type_strhash);
|
624
|
+
}
|
625
|
+
|
626
|
+
/* Create and return table which can hold SIZE strings. */
|
627
|
+
st_table *
|
628
|
+
st_init_strtable_with_size(st_index_t size)
|
629
|
+
{
|
630
|
+
return st_init_table_with_size(&type_strhash, size);
|
631
|
+
}
|
632
|
+
|
633
|
+
/* Create and return table which can hold a minimal number of strings
|
634
|
+
whose character case is ignored. */
|
635
|
+
st_table *
|
636
|
+
st_init_strcasetable(void)
|
637
|
+
{
|
638
|
+
return st_init_table(&type_strcasehash);
|
639
|
+
}
|
640
|
+
|
641
|
+
/* Create and return table which can hold SIZE strings whose character
|
642
|
+
case is ignored. */
|
643
|
+
st_table *
|
644
|
+
st_init_strcasetable_with_size(st_index_t size)
|
645
|
+
{
|
646
|
+
return st_init_table_with_size(&type_strcasehash, size);
|
647
|
+
}
|
648
|
+
|
649
|
+
/* Make table TAB empty. */
|
650
|
+
void
|
651
|
+
st_clear(st_table *tab)
|
652
|
+
{
|
653
|
+
make_tab_empty(tab);
|
654
|
+
tab->rebuilds_num++;
|
655
|
+
}
|
656
|
+
|
657
|
+
/* Free table TAB space. */
|
658
|
+
void
|
659
|
+
st_free_table(st_table *tab)
|
660
|
+
{
|
661
|
+
free(tab->bins);
|
662
|
+
free(tab->entries);
|
663
|
+
free(tab);
|
664
|
+
}
|
665
|
+
|
666
|
+
/* Return byte size of memory allocated for table TAB. */
|
667
|
+
size_t
|
668
|
+
st_memsize(const st_table *tab)
|
669
|
+
{
|
670
|
+
return(sizeof(st_table)
|
671
|
+
+ (tab->bins == NULL ? 0 : bins_size(tab))
|
672
|
+
+ get_allocated_entries(tab) * sizeof(st_table_entry));
|
673
|
+
}
|
674
|
+
|
675
|
+
static st_index_t
|
676
|
+
find_table_entry_ind(st_table *tab, st_hash_t hash_value, st_data_t key);
|
677
|
+
|
678
|
+
static st_index_t
|
679
|
+
find_table_bin_ind(st_table *tab, st_hash_t hash_value, st_data_t key);
|
680
|
+
|
681
|
+
static st_index_t
|
682
|
+
find_table_bin_ind_direct(st_table *table, st_hash_t hash_value, st_data_t key);
|
683
|
+
|
684
|
+
static st_index_t
|
685
|
+
find_table_bin_ptr_and_reserve(st_table *tab, st_hash_t *hash_value,
|
686
|
+
st_data_t key, st_index_t *bin_ind);
|
687
|
+
|
688
|
+
#ifdef HASH_LOG
|
689
|
+
static void
|
690
|
+
count_collision(const struct st_hash_type *type)
|
691
|
+
{
|
692
|
+
collision.all++;
|
693
|
+
if (type == &type_numhash) {
|
694
|
+
collision.num++;
|
695
|
+
}
|
696
|
+
else if (type == &type_strhash) {
|
697
|
+
collision.strcase++;
|
698
|
+
}
|
699
|
+
else if (type == &type_strcasehash) {
|
700
|
+
collision.str++;
|
701
|
+
}
|
702
|
+
}
|
703
|
+
|
704
|
+
#define COLLISION (collision_check ? count_collision(tab->type) : (void)0)
|
705
|
+
#define FOUND_BIN (collision_check ? collision.total++ : (void)0)
|
706
|
+
#define collision_check 0
|
707
|
+
#else
|
708
|
+
#define COLLISION
|
709
|
+
#define FOUND_BIN
|
710
|
+
#endif
|
711
|
+
|
712
|
+
/* If the number of entries in the table is at least REBUILD_THRESHOLD
|
713
|
+
times less than the entry array length, decrease the table
|
714
|
+
size. */
|
715
|
+
#define REBUILD_THRESHOLD 4
|
716
|
+
|
717
|
+
#if REBUILD_THRESHOLD < 2
|
718
|
+
#error "REBUILD_THRESHOLD should be >= 2"
|
719
|
+
#endif
|
720
|
+
|
721
|
+
static void rebuild_table_with(st_table *const new_tab, st_table *const tab);
|
722
|
+
static void rebuild_move_table(st_table *const new_tab, st_table *const tab);
|
723
|
+
static void rebuild_cleanup(st_table *const tab);
|
724
|
+
|
725
|
+
/* Rebuild table TAB. Rebuilding removes all deleted bins and entries
|
726
|
+
and can change size of the table entries and bins arrays.
|
727
|
+
Rebuilding is implemented by creation of a new table or by
|
728
|
+
compaction of the existing one. */
|
729
|
+
static void
|
730
|
+
rebuild_table(st_table *tab)
|
731
|
+
{
|
732
|
+
if ((2 * tab->num_entries <= get_allocated_entries(tab)
|
733
|
+
&& REBUILD_THRESHOLD * tab->num_entries > get_allocated_entries(tab))
|
734
|
+
|| tab->num_entries < (1 << MINIMAL_POWER2)) {
|
735
|
+
/* Compaction: */
|
736
|
+
tab->num_entries = 0;
|
737
|
+
if (tab->bins != NULL)
|
738
|
+
initialize_bins(tab);
|
739
|
+
rebuild_table_with(tab, tab);
|
740
|
+
}
|
741
|
+
else {
|
742
|
+
st_table *new_tab;
|
743
|
+
/* This allocation could trigger GC and compaction. If tab is the
|
744
|
+
* gen_iv_tbl, then tab could have changed in size due to objects being
|
745
|
+
* freed and/or moved. Do not store attributes of tab before this line. */
|
746
|
+
new_tab = st_init_table_with_size(tab->type,
|
747
|
+
2 * tab->num_entries - 1);
|
748
|
+
rebuild_table_with(new_tab, tab);
|
749
|
+
rebuild_move_table(new_tab, tab);
|
750
|
+
}
|
751
|
+
rebuild_cleanup(tab);
|
752
|
+
}
|
753
|
+
|
754
|
+
static void
|
755
|
+
rebuild_table_with(st_table *const new_tab, st_table *const tab)
|
756
|
+
{
|
757
|
+
st_index_t i, ni;
|
758
|
+
unsigned int size_ind;
|
759
|
+
st_table_entry *new_entries;
|
760
|
+
st_table_entry *curr_entry_ptr;
|
761
|
+
st_index_t *bins;
|
762
|
+
st_index_t bin_ind;
|
763
|
+
|
764
|
+
new_entries = new_tab->entries;
|
765
|
+
|
766
|
+
ni = 0;
|
767
|
+
bins = new_tab->bins;
|
768
|
+
size_ind = get_size_ind(new_tab);
|
769
|
+
st_index_t bound = tab->entries_bound;
|
770
|
+
st_table_entry *entries = tab->entries;
|
771
|
+
|
772
|
+
for (i = tab->entries_start; i < bound; i++) {
|
773
|
+
curr_entry_ptr = &entries[i];
|
774
|
+
PREFETCH(entries + i + 1, 0);
|
775
|
+
if (EXPECT(DELETED_ENTRY_P(curr_entry_ptr), 0))
|
776
|
+
continue;
|
777
|
+
if (&new_entries[ni] != curr_entry_ptr)
|
778
|
+
new_entries[ni] = *curr_entry_ptr;
|
779
|
+
if (EXPECT(bins != NULL, 1)) {
|
780
|
+
bin_ind = find_table_bin_ind_direct(new_tab, curr_entry_ptr->hash,
|
781
|
+
curr_entry_ptr->key);
|
782
|
+
set_bin(bins, size_ind, bin_ind, ni + ENTRY_BASE);
|
783
|
+
}
|
784
|
+
new_tab->num_entries++;
|
785
|
+
ni++;
|
786
|
+
}
|
787
|
+
}
|
788
|
+
|
789
|
+
static void
|
790
|
+
rebuild_move_table(st_table *const new_tab, st_table *const tab)
|
791
|
+
{
|
792
|
+
tab->entry_power = new_tab->entry_power;
|
793
|
+
tab->bin_power = new_tab->bin_power;
|
794
|
+
tab->size_ind = new_tab->size_ind;
|
795
|
+
free(tab->bins);
|
796
|
+
tab->bins = new_tab->bins;
|
797
|
+
free(tab->entries);
|
798
|
+
tab->entries = new_tab->entries;
|
799
|
+
free(new_tab);
|
800
|
+
}
|
801
|
+
|
802
|
+
static void
|
803
|
+
rebuild_cleanup(st_table *const tab)
|
804
|
+
{
|
805
|
+
tab->entries_start = 0;
|
806
|
+
tab->entries_bound = tab->num_entries;
|
807
|
+
tab->rebuilds_num++;
|
808
|
+
}
|
809
|
+
|
810
|
+
/* Return the next secondary hash index for table TAB using previous
|
811
|
+
index IND and PERTURB. Finally modulo of the function becomes a
|
812
|
+
full *cycle linear congruential generator*, in other words it
|
813
|
+
guarantees traversing all table bins in extreme case.
|
814
|
+
|
815
|
+
According the Hull-Dobell theorem a generator
|
816
|
+
"Xnext = (a*Xprev + c) mod m" is a full cycle generator if and only if
|
817
|
+
o m and c are relatively prime
|
818
|
+
o a-1 is divisible by all prime factors of m
|
819
|
+
o a-1 is divisible by 4 if m is divisible by 4.
|
820
|
+
|
821
|
+
For our case a is 5, c is 1, and m is a power of two. */
|
822
|
+
static inline st_index_t
|
823
|
+
secondary_hash(st_index_t ind, st_table *tab, st_index_t *perturb)
|
824
|
+
{
|
825
|
+
*perturb >>= 11;
|
826
|
+
ind = (ind << 2) + ind + *perturb + 1;
|
827
|
+
return hash_bin(ind, tab);
|
828
|
+
}
|
829
|
+
|
830
|
+
/* Find an entry with HASH_VALUE and KEY in TABLE using a linear
|
831
|
+
search. Return the index of the found entry in array `entries`.
|
832
|
+
If it is not found, return UNDEFINED_ENTRY_IND. If the table was
|
833
|
+
rebuilt during the search, return REBUILT_TABLE_ENTRY_IND. */
|
834
|
+
static inline st_index_t
|
835
|
+
find_entry(st_table *tab, st_hash_t hash_value, st_data_t key)
|
836
|
+
{
|
837
|
+
int eq_p, rebuilt_p;
|
838
|
+
st_index_t i, bound;
|
839
|
+
st_table_entry *entries;
|
840
|
+
|
841
|
+
bound = tab->entries_bound;
|
842
|
+
entries = tab->entries;
|
843
|
+
for (i = tab->entries_start; i < bound; i++) {
|
844
|
+
DO_PTR_EQUAL_CHECK(tab, &entries[i], hash_value, key, eq_p, rebuilt_p);
|
845
|
+
if (EXPECT(rebuilt_p, 0))
|
846
|
+
return REBUILT_TABLE_ENTRY_IND;
|
847
|
+
if (eq_p)
|
848
|
+
return i;
|
849
|
+
}
|
850
|
+
return UNDEFINED_ENTRY_IND;
|
851
|
+
}
|
852
|
+
|
853
|
+
/* Use the quadratic probing. The method has a better data locality
|
854
|
+
but more collisions than the current approach. In average it
|
855
|
+
results in a bit slower search. */
|
856
|
+
/*#define QUADRATIC_PROBE*/
|
857
|
+
|
858
|
+
/* Return index of entry with HASH_VALUE and KEY in table TAB. If
|
859
|
+
there is no such entry, return UNDEFINED_ENTRY_IND. If the table
|
860
|
+
was rebuilt during the search, return REBUILT_TABLE_ENTRY_IND. */
|
861
|
+
static st_index_t
|
862
|
+
find_table_entry_ind(st_table *tab, st_hash_t hash_value, st_data_t key)
|
863
|
+
{
|
864
|
+
int eq_p, rebuilt_p;
|
865
|
+
st_index_t ind;
|
866
|
+
#ifdef QUADRATIC_PROBE
|
867
|
+
st_index_t d;
|
868
|
+
#else
|
869
|
+
st_index_t perturb;
|
870
|
+
#endif
|
871
|
+
st_index_t bin;
|
872
|
+
st_table_entry *entries = tab->entries;
|
873
|
+
|
874
|
+
ind = hash_bin(hash_value, tab);
|
875
|
+
#ifdef QUADRATIC_PROBE
|
876
|
+
d = 1;
|
877
|
+
#else
|
878
|
+
perturb = hash_value;
|
879
|
+
#endif
|
880
|
+
FOUND_BIN;
|
881
|
+
for (;;) {
|
882
|
+
bin = get_bin(tab->bins, get_size_ind(tab), ind);
|
883
|
+
if (! EMPTY_OR_DELETED_BIN_P(bin)) {
|
884
|
+
DO_PTR_EQUAL_CHECK(tab, &entries[bin - ENTRY_BASE], hash_value, key, eq_p, rebuilt_p);
|
885
|
+
if (EXPECT(rebuilt_p, 0))
|
886
|
+
return REBUILT_TABLE_ENTRY_IND;
|
887
|
+
if (eq_p)
|
888
|
+
break;
|
889
|
+
}
|
890
|
+
else if (EMPTY_BIN_P(bin))
|
891
|
+
return UNDEFINED_ENTRY_IND;
|
892
|
+
#ifdef QUADRATIC_PROBE
|
893
|
+
ind = hash_bin(ind + d, tab);
|
894
|
+
d++;
|
895
|
+
#else
|
896
|
+
ind = secondary_hash(ind, tab, &perturb);
|
897
|
+
#endif
|
898
|
+
COLLISION;
|
899
|
+
}
|
900
|
+
return bin;
|
901
|
+
}
|
902
|
+
|
903
|
+
/* Find and return index of table TAB bin corresponding to an entry
|
904
|
+
with HASH_VALUE and KEY. If there is no such bin, return
|
905
|
+
UNDEFINED_BIN_IND. If the table was rebuilt during the search,
|
906
|
+
return REBUILT_TABLE_BIN_IND. */
|
907
|
+
static st_index_t
|
908
|
+
find_table_bin_ind(st_table *tab, st_hash_t hash_value, st_data_t key)
|
909
|
+
{
|
910
|
+
int eq_p, rebuilt_p;
|
911
|
+
st_index_t ind;
|
912
|
+
#ifdef QUADRATIC_PROBE
|
913
|
+
st_index_t d;
|
914
|
+
#else
|
915
|
+
st_index_t perturb;
|
916
|
+
#endif
|
917
|
+
st_index_t bin;
|
918
|
+
st_table_entry *entries = tab->entries;
|
919
|
+
|
920
|
+
ind = hash_bin(hash_value, tab);
|
921
|
+
#ifdef QUADRATIC_PROBE
|
922
|
+
d = 1;
|
923
|
+
#else
|
924
|
+
perturb = hash_value;
|
925
|
+
#endif
|
926
|
+
FOUND_BIN;
|
927
|
+
for (;;) {
|
928
|
+
bin = get_bin(tab->bins, get_size_ind(tab), ind);
|
929
|
+
if (! EMPTY_OR_DELETED_BIN_P(bin)) {
|
930
|
+
DO_PTR_EQUAL_CHECK(tab, &entries[bin - ENTRY_BASE], hash_value, key, eq_p, rebuilt_p);
|
931
|
+
if (EXPECT(rebuilt_p, 0))
|
932
|
+
return REBUILT_TABLE_BIN_IND;
|
933
|
+
if (eq_p)
|
934
|
+
break;
|
935
|
+
}
|
936
|
+
else if (EMPTY_BIN_P(bin))
|
937
|
+
return UNDEFINED_BIN_IND;
|
938
|
+
#ifdef QUADRATIC_PROBE
|
939
|
+
ind = hash_bin(ind + d, tab);
|
940
|
+
d++;
|
941
|
+
#else
|
942
|
+
ind = secondary_hash(ind, tab, &perturb);
|
943
|
+
#endif
|
944
|
+
COLLISION;
|
945
|
+
}
|
946
|
+
return ind;
|
947
|
+
}
|
948
|
+
|
949
|
+
/* Find and return index of table TAB bin corresponding to an entry
|
950
|
+
with HASH_VALUE and KEY. The entry should be in the table
|
951
|
+
already. */
|
952
|
+
static st_index_t
|
953
|
+
find_table_bin_ind_direct(st_table *tab, st_hash_t hash_value, st_data_t key)
|
954
|
+
{
|
955
|
+
st_index_t ind;
|
956
|
+
#ifdef QUADRATIC_PROBE
|
957
|
+
st_index_t d;
|
958
|
+
#else
|
959
|
+
st_index_t perturb;
|
960
|
+
#endif
|
961
|
+
st_index_t bin;
|
962
|
+
|
963
|
+
ind = hash_bin(hash_value, tab);
|
964
|
+
#ifdef QUADRATIC_PROBE
|
965
|
+
d = 1;
|
966
|
+
#else
|
967
|
+
perturb = hash_value;
|
968
|
+
#endif
|
969
|
+
FOUND_BIN;
|
970
|
+
for (;;) {
|
971
|
+
bin = get_bin(tab->bins, get_size_ind(tab), ind);
|
972
|
+
if (EMPTY_OR_DELETED_BIN_P(bin))
|
973
|
+
return ind;
|
974
|
+
#ifdef QUADRATIC_PROBE
|
975
|
+
ind = hash_bin(ind + d, tab);
|
976
|
+
d++;
|
977
|
+
#else
|
978
|
+
ind = secondary_hash(ind, tab, &perturb);
|
979
|
+
#endif
|
980
|
+
COLLISION;
|
981
|
+
}
|
982
|
+
}
|
983
|
+
|
984
|
+
/* Return index of table TAB bin for HASH_VALUE and KEY through
|
985
|
+
BIN_IND and the pointed value as the function result. Reserve the
|
986
|
+
bin for inclusion of the corresponding entry into the table if it
|
987
|
+
is not there yet. We always find such bin as bins array length is
|
988
|
+
bigger entries array. Although we can reuse a deleted bin, the
|
989
|
+
result bin value is always empty if the table has no entry with
|
990
|
+
KEY. Return the entries array index of the found entry or
|
991
|
+
UNDEFINED_ENTRY_IND if it is not found. If the table was rebuilt
|
992
|
+
during the search, return REBUILT_TABLE_ENTRY_IND. */
|
993
|
+
static st_index_t
|
994
|
+
find_table_bin_ptr_and_reserve(st_table *tab, st_hash_t *hash_value,
|
995
|
+
st_data_t key, st_index_t *bin_ind)
|
996
|
+
{
|
997
|
+
int eq_p, rebuilt_p;
|
998
|
+
st_index_t ind;
|
999
|
+
st_hash_t curr_hash_value = *hash_value;
|
1000
|
+
#ifdef QUADRATIC_PROBE
|
1001
|
+
st_index_t d;
|
1002
|
+
#else
|
1003
|
+
st_index_t perturb;
|
1004
|
+
#endif
|
1005
|
+
st_index_t entry_index;
|
1006
|
+
st_index_t first_deleted_bin_ind;
|
1007
|
+
st_table_entry *entries;
|
1008
|
+
|
1009
|
+
ind = hash_bin(curr_hash_value, tab);
|
1010
|
+
#ifdef QUADRATIC_PROBE
|
1011
|
+
d = 1;
|
1012
|
+
#else
|
1013
|
+
perturb = curr_hash_value;
|
1014
|
+
#endif
|
1015
|
+
FOUND_BIN;
|
1016
|
+
first_deleted_bin_ind = UNDEFINED_BIN_IND;
|
1017
|
+
entries = tab->entries;
|
1018
|
+
for (;;) {
|
1019
|
+
entry_index = get_bin(tab->bins, get_size_ind(tab), ind);
|
1020
|
+
if (EMPTY_BIN_P(entry_index)) {
|
1021
|
+
tab->num_entries++;
|
1022
|
+
entry_index = UNDEFINED_ENTRY_IND;
|
1023
|
+
if (first_deleted_bin_ind != UNDEFINED_BIN_IND) {
|
1024
|
+
/* We can reuse bin of a deleted entry. */
|
1025
|
+
ind = first_deleted_bin_ind;
|
1026
|
+
MARK_BIN_EMPTY(tab, ind);
|
1027
|
+
}
|
1028
|
+
break;
|
1029
|
+
}
|
1030
|
+
else if (! DELETED_BIN_P(entry_index)) {
|
1031
|
+
DO_PTR_EQUAL_CHECK(tab, &entries[entry_index - ENTRY_BASE], curr_hash_value, key, eq_p, rebuilt_p);
|
1032
|
+
if (EXPECT(rebuilt_p, 0))
|
1033
|
+
return REBUILT_TABLE_ENTRY_IND;
|
1034
|
+
if (eq_p)
|
1035
|
+
break;
|
1036
|
+
}
|
1037
|
+
else if (first_deleted_bin_ind == UNDEFINED_BIN_IND)
|
1038
|
+
first_deleted_bin_ind = ind;
|
1039
|
+
#ifdef QUADRATIC_PROBE
|
1040
|
+
ind = hash_bin(ind + d, tab);
|
1041
|
+
d++;
|
1042
|
+
#else
|
1043
|
+
ind = secondary_hash(ind, tab, &perturb);
|
1044
|
+
#endif
|
1045
|
+
COLLISION;
|
1046
|
+
}
|
1047
|
+
*bin_ind = ind;
|
1048
|
+
return entry_index;
|
1049
|
+
}
|
1050
|
+
|
1051
|
+
/* Find an entry with KEY in table TAB. Return non-zero if we found
|
1052
|
+
it. Set up *RECORD to the found entry record. */
|
1053
|
+
int
|
1054
|
+
st_lookup(st_table *tab, st_data_t key, st_data_t *value)
|
1055
|
+
{
|
1056
|
+
st_index_t bin;
|
1057
|
+
st_hash_t hash = do_hash(key, tab);
|
1058
|
+
|
1059
|
+
retry:
|
1060
|
+
if (tab->bins == NULL) {
|
1061
|
+
bin = find_entry(tab, hash, key);
|
1062
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1063
|
+
goto retry;
|
1064
|
+
if (bin == UNDEFINED_ENTRY_IND)
|
1065
|
+
return 0;
|
1066
|
+
}
|
1067
|
+
else {
|
1068
|
+
bin = find_table_entry_ind(tab, hash, key);
|
1069
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1070
|
+
goto retry;
|
1071
|
+
if (bin == UNDEFINED_ENTRY_IND)
|
1072
|
+
return 0;
|
1073
|
+
bin -= ENTRY_BASE;
|
1074
|
+
}
|
1075
|
+
if (value != 0)
|
1076
|
+
*value = tab->entries[bin].record;
|
1077
|
+
return 1;
|
1078
|
+
}
|
1079
|
+
|
1080
|
+
/* Find an entry with KEY in table TAB. Return non-zero if we found
|
1081
|
+
it. Set up *RESULT to the found table entry key. */
|
1082
|
+
int
|
1083
|
+
st_get_key(st_table *tab, st_data_t key, st_data_t *result)
|
1084
|
+
{
|
1085
|
+
st_index_t bin;
|
1086
|
+
st_hash_t hash = do_hash(key, tab);
|
1087
|
+
|
1088
|
+
retry:
|
1089
|
+
if (tab->bins == NULL) {
|
1090
|
+
bin = find_entry(tab, hash, key);
|
1091
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1092
|
+
goto retry;
|
1093
|
+
if (bin == UNDEFINED_ENTRY_IND)
|
1094
|
+
return 0;
|
1095
|
+
}
|
1096
|
+
else {
|
1097
|
+
bin = find_table_entry_ind(tab, hash, key);
|
1098
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1099
|
+
goto retry;
|
1100
|
+
if (bin == UNDEFINED_ENTRY_IND)
|
1101
|
+
return 0;
|
1102
|
+
bin -= ENTRY_BASE;
|
1103
|
+
}
|
1104
|
+
if (result != 0)
|
1105
|
+
*result = tab->entries[bin].key;
|
1106
|
+
return 1;
|
1107
|
+
}
|
1108
|
+
|
1109
|
+
/* Check the table and rebuild it if it is necessary. */
|
1110
|
+
static inline void
|
1111
|
+
rebuild_table_if_necessary (st_table *tab)
|
1112
|
+
{
|
1113
|
+
st_index_t bound = tab->entries_bound;
|
1114
|
+
|
1115
|
+
if (bound == get_allocated_entries(tab))
|
1116
|
+
rebuild_table(tab);
|
1117
|
+
}
|
1118
|
+
|
1119
|
+
/* Insert (KEY, VALUE) into table TAB and return zero. If there is
|
1120
|
+
already entry with KEY in the table, return nonzero and update
|
1121
|
+
the value of the found entry. */
|
1122
|
+
int
|
1123
|
+
st_insert(st_table *tab, st_data_t key, st_data_t value)
|
1124
|
+
{
|
1125
|
+
st_table_entry *entry;
|
1126
|
+
st_index_t bin;
|
1127
|
+
st_index_t ind;
|
1128
|
+
st_hash_t hash_value;
|
1129
|
+
st_index_t bin_ind;
|
1130
|
+
int new_p;
|
1131
|
+
|
1132
|
+
hash_value = do_hash(key, tab);
|
1133
|
+
retry:
|
1134
|
+
rebuild_table_if_necessary(tab);
|
1135
|
+
if (tab->bins == NULL) {
|
1136
|
+
bin = find_entry(tab, hash_value, key);
|
1137
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1138
|
+
goto retry;
|
1139
|
+
new_p = bin == UNDEFINED_ENTRY_IND;
|
1140
|
+
if (new_p)
|
1141
|
+
tab->num_entries++;
|
1142
|
+
bin_ind = UNDEFINED_BIN_IND;
|
1143
|
+
}
|
1144
|
+
else {
|
1145
|
+
bin = find_table_bin_ptr_and_reserve(tab, &hash_value,
|
1146
|
+
key, &bin_ind);
|
1147
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1148
|
+
goto retry;
|
1149
|
+
new_p = bin == UNDEFINED_ENTRY_IND;
|
1150
|
+
bin -= ENTRY_BASE;
|
1151
|
+
}
|
1152
|
+
if (new_p) {
|
1153
|
+
ind = tab->entries_bound++;
|
1154
|
+
entry = &tab->entries[ind];
|
1155
|
+
entry->hash = hash_value;
|
1156
|
+
entry->key = key;
|
1157
|
+
entry->record = value;
|
1158
|
+
if (bin_ind != UNDEFINED_BIN_IND)
|
1159
|
+
set_bin(tab->bins, get_size_ind(tab), bin_ind, ind + ENTRY_BASE);
|
1160
|
+
return 0;
|
1161
|
+
}
|
1162
|
+
tab->entries[bin].record = value;
|
1163
|
+
return 1;
|
1164
|
+
}
|
1165
|
+
|
1166
|
+
/* Insert (KEY, VALUE, HASH) into table TAB. The table should not have
|
1167
|
+
entry with KEY before the insertion. */
|
1168
|
+
static inline void
|
1169
|
+
st_add_direct_with_hash(st_table *tab,
|
1170
|
+
st_data_t key, st_data_t value, st_hash_t hash)
|
1171
|
+
{
|
1172
|
+
st_table_entry *entry;
|
1173
|
+
st_index_t ind;
|
1174
|
+
st_index_t bin_ind;
|
1175
|
+
|
1176
|
+
rebuild_table_if_necessary(tab);
|
1177
|
+
ind = tab->entries_bound++;
|
1178
|
+
entry = &tab->entries[ind];
|
1179
|
+
entry->hash = hash;
|
1180
|
+
entry->key = key;
|
1181
|
+
entry->record = value;
|
1182
|
+
tab->num_entries++;
|
1183
|
+
if (tab->bins != NULL) {
|
1184
|
+
bin_ind = find_table_bin_ind_direct(tab, hash, key);
|
1185
|
+
set_bin(tab->bins, get_size_ind(tab), bin_ind, ind + ENTRY_BASE);
|
1186
|
+
}
|
1187
|
+
}
|
1188
|
+
|
1189
|
+
void
|
1190
|
+
rb_st_add_direct_with_hash(st_table *tab,
|
1191
|
+
st_data_t key, st_data_t value, st_hash_t hash)
|
1192
|
+
{
|
1193
|
+
st_add_direct_with_hash(tab, key, value, hash);
|
1194
|
+
}
|
1195
|
+
|
1196
|
+
/* Insert (KEY, VALUE) into table TAB. The table should not have
|
1197
|
+
entry with KEY before the insertion. */
|
1198
|
+
void
|
1199
|
+
st_add_direct(st_table *tab, st_data_t key, st_data_t value)
|
1200
|
+
{
|
1201
|
+
st_hash_t hash_value;
|
1202
|
+
|
1203
|
+
hash_value = do_hash(key, tab);
|
1204
|
+
st_add_direct_with_hash(tab, key, value, hash_value);
|
1205
|
+
}
|
1206
|
+
|
1207
|
+
/* Insert (FUNC(KEY), VALUE) into table TAB and return zero. If
|
1208
|
+
there is already entry with KEY in the table, return nonzero and
|
1209
|
+
update the value of the found entry. */
|
1210
|
+
int
|
1211
|
+
st_insert2(st_table *tab, st_data_t key, st_data_t value,
|
1212
|
+
st_data_t (*func)(st_data_t))
|
1213
|
+
{
|
1214
|
+
st_table_entry *entry;
|
1215
|
+
st_index_t bin;
|
1216
|
+
st_index_t ind;
|
1217
|
+
st_hash_t hash_value;
|
1218
|
+
st_index_t bin_ind;
|
1219
|
+
int new_p;
|
1220
|
+
|
1221
|
+
hash_value = do_hash(key, tab);
|
1222
|
+
retry:
|
1223
|
+
rebuild_table_if_necessary (tab);
|
1224
|
+
if (tab->bins == NULL) {
|
1225
|
+
bin = find_entry(tab, hash_value, key);
|
1226
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1227
|
+
goto retry;
|
1228
|
+
new_p = bin == UNDEFINED_ENTRY_IND;
|
1229
|
+
if (new_p)
|
1230
|
+
tab->num_entries++;
|
1231
|
+
bin_ind = UNDEFINED_BIN_IND;
|
1232
|
+
}
|
1233
|
+
else {
|
1234
|
+
bin = find_table_bin_ptr_and_reserve(tab, &hash_value,
|
1235
|
+
key, &bin_ind);
|
1236
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1237
|
+
goto retry;
|
1238
|
+
new_p = bin == UNDEFINED_ENTRY_IND;
|
1239
|
+
bin -= ENTRY_BASE;
|
1240
|
+
}
|
1241
|
+
if (new_p) {
|
1242
|
+
key = (*func)(key);
|
1243
|
+
ind = tab->entries_bound++;
|
1244
|
+
entry = &tab->entries[ind];
|
1245
|
+
entry->hash = hash_value;
|
1246
|
+
entry->key = key;
|
1247
|
+
entry->record = value;
|
1248
|
+
if (bin_ind != UNDEFINED_BIN_IND)
|
1249
|
+
set_bin(tab->bins, get_size_ind(tab), bin_ind, ind + ENTRY_BASE);
|
1250
|
+
return 0;
|
1251
|
+
}
|
1252
|
+
tab->entries[bin].record = value;
|
1253
|
+
return 1;
|
1254
|
+
}
|
1255
|
+
|
1256
|
+
/* Create a copy of old_tab into new_tab. */
|
1257
|
+
st_table *
|
1258
|
+
st_replace(st_table *new_tab, st_table *old_tab)
|
1259
|
+
{
|
1260
|
+
*new_tab = *old_tab;
|
1261
|
+
if (old_tab->bins == NULL)
|
1262
|
+
new_tab->bins = NULL;
|
1263
|
+
else {
|
1264
|
+
new_tab->bins = (st_index_t *) malloc(bins_size(old_tab));
|
1265
|
+
#ifndef RUBY
|
1266
|
+
if (new_tab->bins == NULL) {
|
1267
|
+
return NULL;
|
1268
|
+
}
|
1269
|
+
#endif
|
1270
|
+
}
|
1271
|
+
new_tab->entries = (st_table_entry *) malloc(get_allocated_entries(old_tab)
|
1272
|
+
* sizeof(st_table_entry));
|
1273
|
+
#ifndef RUBY
|
1274
|
+
if (new_tab->entries == NULL) {
|
1275
|
+
return NULL;
|
1276
|
+
}
|
1277
|
+
#endif
|
1278
|
+
MEMCPY(new_tab->entries, old_tab->entries, st_table_entry,
|
1279
|
+
get_allocated_entries(old_tab));
|
1280
|
+
if (old_tab->bins != NULL)
|
1281
|
+
MEMCPY(new_tab->bins, old_tab->bins, char, bins_size(old_tab));
|
1282
|
+
|
1283
|
+
return new_tab;
|
1284
|
+
}
|
1285
|
+
|
1286
|
+
/* Create and return a copy of table OLD_TAB. */
|
1287
|
+
st_table *
|
1288
|
+
st_copy(st_table *old_tab)
|
1289
|
+
{
|
1290
|
+
st_table *new_tab;
|
1291
|
+
|
1292
|
+
new_tab = (st_table *) malloc(sizeof(st_table));
|
1293
|
+
#ifndef RUBY
|
1294
|
+
if (new_tab == NULL)
|
1295
|
+
return NULL;
|
1296
|
+
#endif
|
1297
|
+
|
1298
|
+
if (st_replace(new_tab, old_tab) == NULL) {
|
1299
|
+
st_free_table(new_tab);
|
1300
|
+
return NULL;
|
1301
|
+
}
|
1302
|
+
|
1303
|
+
return new_tab;
|
1304
|
+
}
|
1305
|
+
|
1306
|
+
/* Update the entries start of table TAB after removing an entry
|
1307
|
+
with index N in the array entries. */
|
1308
|
+
static inline void
|
1309
|
+
update_range_for_deleted(st_table *tab, st_index_t n)
|
1310
|
+
{
|
1311
|
+
/* Do not update entries_bound here. Otherwise, we can fill all
|
1312
|
+
bins by deleted entry value before rebuilding the table. */
|
1313
|
+
if (tab->entries_start == n) {
|
1314
|
+
st_index_t start = n + 1;
|
1315
|
+
st_index_t bound = tab->entries_bound;
|
1316
|
+
st_table_entry *entries = tab->entries;
|
1317
|
+
while (start < bound && DELETED_ENTRY_P(&entries[start])) start++;
|
1318
|
+
tab->entries_start = start;
|
1319
|
+
}
|
1320
|
+
}
|
1321
|
+
|
1322
|
+
/* Delete entry with KEY from table TAB, set up *VALUE (unless
|
1323
|
+
VALUE is zero) from deleted table entry, and return non-zero. If
|
1324
|
+
there is no entry with KEY in the table, clear *VALUE (unless VALUE
|
1325
|
+
is zero), and return zero. */
|
1326
|
+
static int
|
1327
|
+
st_general_delete(st_table *tab, st_data_t *key, st_data_t *value)
|
1328
|
+
{
|
1329
|
+
st_table_entry *entry;
|
1330
|
+
st_index_t bin;
|
1331
|
+
st_index_t bin_ind;
|
1332
|
+
st_hash_t hash;
|
1333
|
+
|
1334
|
+
hash = do_hash(*key, tab);
|
1335
|
+
retry:
|
1336
|
+
if (tab->bins == NULL) {
|
1337
|
+
bin = find_entry(tab, hash, *key);
|
1338
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1339
|
+
goto retry;
|
1340
|
+
if (bin == UNDEFINED_ENTRY_IND) {
|
1341
|
+
if (value != 0) *value = 0;
|
1342
|
+
return 0;
|
1343
|
+
}
|
1344
|
+
}
|
1345
|
+
else {
|
1346
|
+
bin_ind = find_table_bin_ind(tab, hash, *key);
|
1347
|
+
if (EXPECT(bin_ind == REBUILT_TABLE_BIN_IND, 0))
|
1348
|
+
goto retry;
|
1349
|
+
if (bin_ind == UNDEFINED_BIN_IND) {
|
1350
|
+
if (value != 0) *value = 0;
|
1351
|
+
return 0;
|
1352
|
+
}
|
1353
|
+
bin = get_bin(tab->bins, get_size_ind(tab), bin_ind) - ENTRY_BASE;
|
1354
|
+
MARK_BIN_DELETED(tab, bin_ind);
|
1355
|
+
}
|
1356
|
+
entry = &tab->entries[bin];
|
1357
|
+
*key = entry->key;
|
1358
|
+
if (value != 0) *value = entry->record;
|
1359
|
+
MARK_ENTRY_DELETED(entry);
|
1360
|
+
tab->num_entries--;
|
1361
|
+
update_range_for_deleted(tab, bin);
|
1362
|
+
return 1;
|
1363
|
+
}
|
1364
|
+
|
1365
|
+
int
|
1366
|
+
st_delete(st_table *tab, st_data_t *key, st_data_t *value)
|
1367
|
+
{
|
1368
|
+
return st_general_delete(tab, key, value);
|
1369
|
+
}
|
1370
|
+
|
1371
|
+
/* The function and other functions with suffix '_safe' or '_check'
|
1372
|
+
are originated from the previous implementation of the hash tables.
|
1373
|
+
It was necessary for correct deleting entries during traversing
|
1374
|
+
tables. The current implementation permits deletion during
|
1375
|
+
traversing without a specific way to do this. */
|
1376
|
+
int
|
1377
|
+
st_delete_safe(st_table *tab, st_data_t *key, st_data_t *value,
|
1378
|
+
st_data_t never ATTRIBUTE_UNUSED)
|
1379
|
+
{
|
1380
|
+
return st_general_delete(tab, key, value);
|
1381
|
+
}
|
1382
|
+
|
1383
|
+
/* If table TAB is empty, clear *VALUE (unless VALUE is zero), and
|
1384
|
+
return zero. Otherwise, remove the first entry in the table.
|
1385
|
+
Return its key through KEY and its record through VALUE (unless
|
1386
|
+
VALUE is zero). */
|
1387
|
+
int
|
1388
|
+
st_shift(st_table *tab, st_data_t *key, st_data_t *value)
|
1389
|
+
{
|
1390
|
+
st_index_t i, bound;
|
1391
|
+
st_index_t bin;
|
1392
|
+
st_table_entry *entries, *curr_entry_ptr;
|
1393
|
+
st_index_t bin_ind;
|
1394
|
+
|
1395
|
+
entries = tab->entries;
|
1396
|
+
bound = tab->entries_bound;
|
1397
|
+
for (i = tab->entries_start; i < bound; i++) {
|
1398
|
+
curr_entry_ptr = &entries[i];
|
1399
|
+
if (! DELETED_ENTRY_P(curr_entry_ptr)) {
|
1400
|
+
st_hash_t entry_hash = curr_entry_ptr->hash;
|
1401
|
+
st_data_t entry_key = curr_entry_ptr->key;
|
1402
|
+
|
1403
|
+
if (value != 0) *value = curr_entry_ptr->record;
|
1404
|
+
*key = entry_key;
|
1405
|
+
retry:
|
1406
|
+
if (tab->bins == NULL) {
|
1407
|
+
bin = find_entry(tab, entry_hash, entry_key);
|
1408
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0)) {
|
1409
|
+
entries = tab->entries;
|
1410
|
+
goto retry;
|
1411
|
+
}
|
1412
|
+
curr_entry_ptr = &entries[bin];
|
1413
|
+
}
|
1414
|
+
else {
|
1415
|
+
bin_ind = find_table_bin_ind(tab, entry_hash, entry_key);
|
1416
|
+
if (EXPECT(bin_ind == REBUILT_TABLE_BIN_IND, 0)) {
|
1417
|
+
entries = tab->entries;
|
1418
|
+
goto retry;
|
1419
|
+
}
|
1420
|
+
curr_entry_ptr = &entries[get_bin(tab->bins, get_size_ind(tab), bin_ind)
|
1421
|
+
- ENTRY_BASE];
|
1422
|
+
MARK_BIN_DELETED(tab, bin_ind);
|
1423
|
+
}
|
1424
|
+
MARK_ENTRY_DELETED(curr_entry_ptr);
|
1425
|
+
tab->num_entries--;
|
1426
|
+
update_range_for_deleted(tab, i);
|
1427
|
+
return 1;
|
1428
|
+
}
|
1429
|
+
}
|
1430
|
+
if (value != 0) *value = 0;
|
1431
|
+
return 0;
|
1432
|
+
}
|
1433
|
+
|
1434
|
+
/* See comments for function st_delete_safe. */
|
1435
|
+
void
|
1436
|
+
st_cleanup_safe(st_table *tab ATTRIBUTE_UNUSED,
|
1437
|
+
st_data_t never ATTRIBUTE_UNUSED)
|
1438
|
+
{
|
1439
|
+
}
|
1440
|
+
|
1441
|
+
/* Find entry with KEY in table TAB, call FUNC with pointers to copies
|
1442
|
+
of the key and the value of the found entry, and non-zero as the
|
1443
|
+
3rd argument. If the entry is not found, call FUNC with a pointer
|
1444
|
+
to KEY, a pointer to zero, and a zero argument. If the call
|
1445
|
+
returns ST_CONTINUE, the table will have an entry with key and
|
1446
|
+
value returned by FUNC through the 1st and 2nd parameters. If the
|
1447
|
+
call of FUNC returns ST_DELETE, the table will not have entry with
|
1448
|
+
KEY. The function returns flag of that the entry with KEY was in
|
1449
|
+
the table before the call. */
|
1450
|
+
int
|
1451
|
+
st_update(st_table *tab, st_data_t key,
|
1452
|
+
st_update_callback_func *func, st_data_t arg)
|
1453
|
+
{
|
1454
|
+
st_table_entry *entry = NULL; /* to avoid uninitialized value warning */
|
1455
|
+
st_index_t bin = 0; /* Ditto */
|
1456
|
+
st_table_entry *entries;
|
1457
|
+
st_index_t bin_ind;
|
1458
|
+
st_data_t value = 0, old_key;
|
1459
|
+
int retval, existing;
|
1460
|
+
st_hash_t hash = do_hash(key, tab);
|
1461
|
+
|
1462
|
+
retry:
|
1463
|
+
entries = tab->entries;
|
1464
|
+
if (tab->bins == NULL) {
|
1465
|
+
bin = find_entry(tab, hash, key);
|
1466
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1467
|
+
goto retry;
|
1468
|
+
existing = bin != UNDEFINED_ENTRY_IND;
|
1469
|
+
entry = &entries[bin];
|
1470
|
+
bin_ind = UNDEFINED_BIN_IND;
|
1471
|
+
}
|
1472
|
+
else {
|
1473
|
+
bin_ind = find_table_bin_ind(tab, hash, key);
|
1474
|
+
if (EXPECT(bin_ind == REBUILT_TABLE_BIN_IND, 0))
|
1475
|
+
goto retry;
|
1476
|
+
existing = bin_ind != UNDEFINED_BIN_IND;
|
1477
|
+
if (existing) {
|
1478
|
+
bin = get_bin(tab->bins, get_size_ind(tab), bin_ind) - ENTRY_BASE;
|
1479
|
+
entry = &entries[bin];
|
1480
|
+
}
|
1481
|
+
}
|
1482
|
+
if (existing) {
|
1483
|
+
key = entry->key;
|
1484
|
+
value = entry->record;
|
1485
|
+
}
|
1486
|
+
old_key = key;
|
1487
|
+
retval = (*func)(&key, &value, arg, existing);
|
1488
|
+
switch (retval) {
|
1489
|
+
case ST_CONTINUE:
|
1490
|
+
if (! existing) {
|
1491
|
+
st_add_direct_with_hash(tab, key, value, hash);
|
1492
|
+
break;
|
1493
|
+
}
|
1494
|
+
if (old_key != key) {
|
1495
|
+
entry->key = key;
|
1496
|
+
}
|
1497
|
+
entry->record = value;
|
1498
|
+
break;
|
1499
|
+
case ST_DELETE:
|
1500
|
+
if (existing) {
|
1501
|
+
if (bin_ind != UNDEFINED_BIN_IND)
|
1502
|
+
MARK_BIN_DELETED(tab, bin_ind);
|
1503
|
+
MARK_ENTRY_DELETED(entry);
|
1504
|
+
tab->num_entries--;
|
1505
|
+
update_range_for_deleted(tab, bin);
|
1506
|
+
}
|
1507
|
+
break;
|
1508
|
+
}
|
1509
|
+
return existing;
|
1510
|
+
}
|
1511
|
+
|
1512
|
+
/* Traverse all entries in table TAB calling FUNC with current entry
|
1513
|
+
key and value and zero. If the call returns ST_STOP, stop
|
1514
|
+
traversing. If the call returns ST_DELETE, delete the current
|
1515
|
+
entry from the table. In case of ST_CHECK or ST_CONTINUE, continue
|
1516
|
+
traversing. The function returns zero unless an error is found.
|
1517
|
+
CHECK_P is flag of st_foreach_check call. The behavior is a bit
|
1518
|
+
different for ST_CHECK and when the current element is removed
|
1519
|
+
during traversing. */
|
1520
|
+
static inline int
|
1521
|
+
st_general_foreach(st_table *tab, st_foreach_check_callback_func *func, st_update_callback_func *replace, st_data_t arg,
|
1522
|
+
int check_p)
|
1523
|
+
{
|
1524
|
+
st_index_t bin;
|
1525
|
+
st_index_t bin_ind;
|
1526
|
+
st_table_entry *entries, *curr_entry_ptr;
|
1527
|
+
enum st_retval retval;
|
1528
|
+
st_index_t i, rebuilds_num;
|
1529
|
+
st_hash_t hash;
|
1530
|
+
st_data_t key;
|
1531
|
+
int error_p, packed_p = tab->bins == NULL;
|
1532
|
+
|
1533
|
+
entries = tab->entries;
|
1534
|
+
/* The bound can change inside the loop even without rebuilding
|
1535
|
+
the table, e.g. by an entry insertion. */
|
1536
|
+
for (i = tab->entries_start; i < tab->entries_bound; i++) {
|
1537
|
+
curr_entry_ptr = &entries[i];
|
1538
|
+
if (EXPECT(DELETED_ENTRY_P(curr_entry_ptr), 0))
|
1539
|
+
continue;
|
1540
|
+
key = curr_entry_ptr->key;
|
1541
|
+
rebuilds_num = tab->rebuilds_num;
|
1542
|
+
hash = curr_entry_ptr->hash;
|
1543
|
+
retval = (*func)(key, curr_entry_ptr->record, arg, 0);
|
1544
|
+
|
1545
|
+
if (retval == ST_REPLACE && replace) {
|
1546
|
+
st_data_t value;
|
1547
|
+
value = curr_entry_ptr->record;
|
1548
|
+
retval = (*replace)(&key, &value, arg, TRUE);
|
1549
|
+
curr_entry_ptr->key = key;
|
1550
|
+
curr_entry_ptr->record = value;
|
1551
|
+
}
|
1552
|
+
|
1553
|
+
if (rebuilds_num != tab->rebuilds_num) {
|
1554
|
+
retry:
|
1555
|
+
entries = tab->entries;
|
1556
|
+
packed_p = tab->bins == NULL;
|
1557
|
+
if (packed_p) {
|
1558
|
+
i = find_entry(tab, hash, key);
|
1559
|
+
if (EXPECT(i == REBUILT_TABLE_ENTRY_IND, 0))
|
1560
|
+
goto retry;
|
1561
|
+
error_p = i == UNDEFINED_ENTRY_IND;
|
1562
|
+
}
|
1563
|
+
else {
|
1564
|
+
i = find_table_entry_ind(tab, hash, key);
|
1565
|
+
if (EXPECT(i == REBUILT_TABLE_ENTRY_IND, 0))
|
1566
|
+
goto retry;
|
1567
|
+
error_p = i == UNDEFINED_ENTRY_IND;
|
1568
|
+
i -= ENTRY_BASE;
|
1569
|
+
}
|
1570
|
+
if (error_p && check_p) {
|
1571
|
+
/* call func with error notice */
|
1572
|
+
retval = (*func)(0, 0, arg, 1);
|
1573
|
+
return 1;
|
1574
|
+
}
|
1575
|
+
curr_entry_ptr = &entries[i];
|
1576
|
+
}
|
1577
|
+
switch (retval) {
|
1578
|
+
case ST_REPLACE:
|
1579
|
+
break;
|
1580
|
+
case ST_CONTINUE:
|
1581
|
+
break;
|
1582
|
+
case ST_CHECK:
|
1583
|
+
if (check_p)
|
1584
|
+
break;
|
1585
|
+
case ST_STOP:
|
1586
|
+
return 0;
|
1587
|
+
case ST_DELETE: {
|
1588
|
+
st_data_t key = curr_entry_ptr->key;
|
1589
|
+
|
1590
|
+
again:
|
1591
|
+
if (packed_p) {
|
1592
|
+
bin = find_entry(tab, hash, key);
|
1593
|
+
if (EXPECT(bin == REBUILT_TABLE_ENTRY_IND, 0))
|
1594
|
+
goto again;
|
1595
|
+
if (bin == UNDEFINED_ENTRY_IND)
|
1596
|
+
break;
|
1597
|
+
}
|
1598
|
+
else {
|
1599
|
+
bin_ind = find_table_bin_ind(tab, hash, key);
|
1600
|
+
if (EXPECT(bin_ind == REBUILT_TABLE_BIN_IND, 0))
|
1601
|
+
goto again;
|
1602
|
+
if (bin_ind == UNDEFINED_BIN_IND)
|
1603
|
+
break;
|
1604
|
+
bin = get_bin(tab->bins, get_size_ind(tab), bin_ind) - ENTRY_BASE;
|
1605
|
+
MARK_BIN_DELETED(tab, bin_ind);
|
1606
|
+
}
|
1607
|
+
curr_entry_ptr = &entries[bin];
|
1608
|
+
MARK_ENTRY_DELETED(curr_entry_ptr);
|
1609
|
+
tab->num_entries--;
|
1610
|
+
update_range_for_deleted(tab, bin);
|
1611
|
+
break;
|
1612
|
+
}
|
1613
|
+
}
|
1614
|
+
}
|
1615
|
+
return 0;
|
1616
|
+
}
|
1617
|
+
|
1618
|
+
int
|
1619
|
+
st_foreach_with_replace(st_table *tab, st_foreach_check_callback_func *func, st_update_callback_func *replace, st_data_t arg)
|
1620
|
+
{
|
1621
|
+
return st_general_foreach(tab, func, replace, arg, TRUE);
|
1622
|
+
}
|
1623
|
+
|
1624
|
+
struct functor {
|
1625
|
+
st_foreach_callback_func *func;
|
1626
|
+
st_data_t arg;
|
1627
|
+
};
|
1628
|
+
|
1629
|
+
static int
|
1630
|
+
apply_functor(st_data_t k, st_data_t v, st_data_t d, int _)
|
1631
|
+
{
|
1632
|
+
const struct functor *f = (void *)d;
|
1633
|
+
return f->func(k, v, f->arg);
|
1634
|
+
}
|
1635
|
+
|
1636
|
+
int
|
1637
|
+
st_foreach(st_table *tab, st_foreach_callback_func *func, st_data_t arg)
|
1638
|
+
{
|
1639
|
+
const struct functor f = { func, arg };
|
1640
|
+
return st_general_foreach(tab, apply_functor, 0, (st_data_t)&f, FALSE);
|
1641
|
+
}
|
1642
|
+
|
1643
|
+
/* See comments for function st_delete_safe. */
|
1644
|
+
int
|
1645
|
+
st_foreach_check(st_table *tab, st_foreach_check_callback_func *func, st_data_t arg,
|
1646
|
+
st_data_t never ATTRIBUTE_UNUSED)
|
1647
|
+
{
|
1648
|
+
return st_general_foreach(tab, func, 0, arg, TRUE);
|
1649
|
+
}
|
1650
|
+
|
1651
|
+
/* Set up array KEYS by at most SIZE keys of head table TAB entries.
|
1652
|
+
Return the number of keys set up in array KEYS. */
|
1653
|
+
static inline st_index_t
|
1654
|
+
st_general_keys(st_table *tab, st_data_t *keys, st_index_t size)
|
1655
|
+
{
|
1656
|
+
st_index_t i, bound;
|
1657
|
+
st_data_t key, *keys_start, *keys_end;
|
1658
|
+
st_table_entry *curr_entry_ptr, *entries = tab->entries;
|
1659
|
+
|
1660
|
+
bound = tab->entries_bound;
|
1661
|
+
keys_start = keys;
|
1662
|
+
keys_end = keys + size;
|
1663
|
+
for (i = tab->entries_start; i < bound; i++) {
|
1664
|
+
if (keys == keys_end)
|
1665
|
+
break;
|
1666
|
+
curr_entry_ptr = &entries[i];
|
1667
|
+
key = curr_entry_ptr->key;
|
1668
|
+
if (! DELETED_ENTRY_P(curr_entry_ptr))
|
1669
|
+
*keys++ = key;
|
1670
|
+
}
|
1671
|
+
|
1672
|
+
return keys - keys_start;
|
1673
|
+
}
|
1674
|
+
|
1675
|
+
st_index_t
|
1676
|
+
st_keys(st_table *tab, st_data_t *keys, st_index_t size)
|
1677
|
+
{
|
1678
|
+
return st_general_keys(tab, keys, size);
|
1679
|
+
}
|
1680
|
+
|
1681
|
+
/* See comments for function st_delete_safe. */
|
1682
|
+
st_index_t
|
1683
|
+
st_keys_check(st_table *tab, st_data_t *keys, st_index_t size,
|
1684
|
+
st_data_t never ATTRIBUTE_UNUSED)
|
1685
|
+
{
|
1686
|
+
return st_general_keys(tab, keys, size);
|
1687
|
+
}
|
1688
|
+
|
1689
|
+
/* Set up array VALUES by at most SIZE values of head table TAB
|
1690
|
+
entries. Return the number of values set up in array VALUES. */
|
1691
|
+
static inline st_index_t
|
1692
|
+
st_general_values(st_table *tab, st_data_t *values, st_index_t size)
|
1693
|
+
{
|
1694
|
+
st_index_t i, bound;
|
1695
|
+
st_data_t *values_start, *values_end;
|
1696
|
+
st_table_entry *curr_entry_ptr, *entries = tab->entries;
|
1697
|
+
|
1698
|
+
values_start = values;
|
1699
|
+
values_end = values + size;
|
1700
|
+
bound = tab->entries_bound;
|
1701
|
+
for (i = tab->entries_start; i < bound; i++) {
|
1702
|
+
if (values == values_end)
|
1703
|
+
break;
|
1704
|
+
curr_entry_ptr = &entries[i];
|
1705
|
+
if (! DELETED_ENTRY_P(curr_entry_ptr))
|
1706
|
+
*values++ = curr_entry_ptr->record;
|
1707
|
+
}
|
1708
|
+
|
1709
|
+
return values - values_start;
|
1710
|
+
}
|
1711
|
+
|
1712
|
+
st_index_t
|
1713
|
+
st_values(st_table *tab, st_data_t *values, st_index_t size)
|
1714
|
+
{
|
1715
|
+
return st_general_values(tab, values, size);
|
1716
|
+
}
|
1717
|
+
|
1718
|
+
/* See comments for function st_delete_safe. */
|
1719
|
+
st_index_t
|
1720
|
+
st_values_check(st_table *tab, st_data_t *values, st_index_t size,
|
1721
|
+
st_data_t never ATTRIBUTE_UNUSED)
|
1722
|
+
{
|
1723
|
+
return st_general_values(tab, values, size);
|
1724
|
+
}
|
1725
|
+
|
1726
|
+
#define FNV1_32A_INIT 0x811c9dc5
|
1727
|
+
|
1728
|
+
/*
|
1729
|
+
* 32 bit magic FNV-1a prime
|
1730
|
+
*/
|
1731
|
+
#define FNV_32_PRIME 0x01000193
|
1732
|
+
|
1733
|
+
/* __POWERPC__ added to accommodate Darwin case. */
|
1734
|
+
#ifndef UNALIGNED_WORD_ACCESS
|
1735
|
+
# if defined(__i386) || defined(__i386__) || defined(_M_IX86) || \
|
1736
|
+
defined(__x86_64) || defined(__x86_64__) || defined(_M_AMD64) || \
|
1737
|
+
defined(__powerpc64__) || defined(__POWERPC__) || defined(__aarch64__) || \
|
1738
|
+
defined(__mc68020__)
|
1739
|
+
# define UNALIGNED_WORD_ACCESS 1
|
1740
|
+
# endif
|
1741
|
+
#endif
|
1742
|
+
#ifndef UNALIGNED_WORD_ACCESS
|
1743
|
+
# define UNALIGNED_WORD_ACCESS 0
|
1744
|
+
#endif
|
1745
|
+
|
1746
|
+
/* This hash function is quite simplified MurmurHash3
|
1747
|
+
* Simplification is legal, cause most of magic still happens in finalizator.
|
1748
|
+
* And finalizator is almost the same as in MurmurHash3 */
|
1749
|
+
#define BIG_CONSTANT(x,y) ((st_index_t)(x)<<32|(st_index_t)(y))
|
1750
|
+
#define ROTL(x,n) ((x)<<(n)|(x)>>(SIZEOF_ST_INDEX_T*CHAR_BIT-(n)))
|
1751
|
+
|
1752
|
+
#if ST_INDEX_BITS <= 32
|
1753
|
+
#define C1 (st_index_t)0xcc9e2d51
|
1754
|
+
#define C2 (st_index_t)0x1b873593
|
1755
|
+
#else
|
1756
|
+
#define C1 BIG_CONSTANT(0x87c37b91,0x114253d5);
|
1757
|
+
#define C2 BIG_CONSTANT(0x4cf5ad43,0x2745937f);
|
1758
|
+
#endif
|
1759
|
+
NO_SANITIZE("unsigned-integer-overflow", static inline st_index_t murmur_step(st_index_t h, st_index_t k));
|
1760
|
+
NO_SANITIZE("unsigned-integer-overflow", static inline st_index_t murmur_finish(st_index_t h));
|
1761
|
+
NO_SANITIZE("unsigned-integer-overflow", extern st_index_t st_hash(const void *ptr, size_t len, st_index_t h));
|
1762
|
+
|
1763
|
+
static inline st_index_t
|
1764
|
+
murmur_step(st_index_t h, st_index_t k)
|
1765
|
+
{
|
1766
|
+
#if ST_INDEX_BITS <= 32
|
1767
|
+
#define r1 (17)
|
1768
|
+
#define r2 (11)
|
1769
|
+
#else
|
1770
|
+
#define r1 (33)
|
1771
|
+
#define r2 (24)
|
1772
|
+
#endif
|
1773
|
+
k *= C1;
|
1774
|
+
h ^= ROTL(k, r1);
|
1775
|
+
h *= C2;
|
1776
|
+
h = ROTL(h, r2);
|
1777
|
+
return h;
|
1778
|
+
}
|
1779
|
+
#undef r1
|
1780
|
+
#undef r2
|
1781
|
+
|
1782
|
+
static inline st_index_t
|
1783
|
+
murmur_finish(st_index_t h)
|
1784
|
+
{
|
1785
|
+
#if ST_INDEX_BITS <= 32
|
1786
|
+
#define r1 (16)
|
1787
|
+
#define r2 (13)
|
1788
|
+
#define r3 (16)
|
1789
|
+
const st_index_t c1 = 0x85ebca6b;
|
1790
|
+
const st_index_t c2 = 0xc2b2ae35;
|
1791
|
+
#else
|
1792
|
+
/* values are taken from Mix13 on http://zimbry.blogspot.ru/2011/09/better-bit-mixing-improving-on.html */
|
1793
|
+
#define r1 (30)
|
1794
|
+
#define r2 (27)
|
1795
|
+
#define r3 (31)
|
1796
|
+
const st_index_t c1 = BIG_CONSTANT(0xbf58476d,0x1ce4e5b9);
|
1797
|
+
const st_index_t c2 = BIG_CONSTANT(0x94d049bb,0x133111eb);
|
1798
|
+
#endif
|
1799
|
+
#if ST_INDEX_BITS > 64
|
1800
|
+
h ^= h >> 64;
|
1801
|
+
h *= c2;
|
1802
|
+
h ^= h >> 65;
|
1803
|
+
#endif
|
1804
|
+
h ^= h >> r1;
|
1805
|
+
h *= c1;
|
1806
|
+
h ^= h >> r2;
|
1807
|
+
h *= c2;
|
1808
|
+
h ^= h >> r3;
|
1809
|
+
return h;
|
1810
|
+
}
|
1811
|
+
#undef r1
|
1812
|
+
#undef r2
|
1813
|
+
#undef r3
|
1814
|
+
|
1815
|
+
st_index_t
|
1816
|
+
st_hash(const void *ptr, size_t len, st_index_t h)
|
1817
|
+
{
|
1818
|
+
const char *data = ptr;
|
1819
|
+
st_index_t t = 0;
|
1820
|
+
size_t l = len;
|
1821
|
+
|
1822
|
+
#define data_at(n) (st_index_t)((unsigned char)data[(n)])
|
1823
|
+
#define UNALIGNED_ADD_4 UNALIGNED_ADD(2); UNALIGNED_ADD(1); UNALIGNED_ADD(0)
|
1824
|
+
#if SIZEOF_ST_INDEX_T > 4
|
1825
|
+
#define UNALIGNED_ADD_8 UNALIGNED_ADD(6); UNALIGNED_ADD(5); UNALIGNED_ADD(4); UNALIGNED_ADD(3); UNALIGNED_ADD_4
|
1826
|
+
#if SIZEOF_ST_INDEX_T > 8
|
1827
|
+
#define UNALIGNED_ADD_16 UNALIGNED_ADD(14); UNALIGNED_ADD(13); UNALIGNED_ADD(12); UNALIGNED_ADD(11); \
|
1828
|
+
UNALIGNED_ADD(10); UNALIGNED_ADD(9); UNALIGNED_ADD(8); UNALIGNED_ADD(7); UNALIGNED_ADD_8
|
1829
|
+
#define UNALIGNED_ADD_ALL UNALIGNED_ADD_16
|
1830
|
+
#endif
|
1831
|
+
#define UNALIGNED_ADD_ALL UNALIGNED_ADD_8
|
1832
|
+
#else
|
1833
|
+
#define UNALIGNED_ADD_ALL UNALIGNED_ADD_4
|
1834
|
+
#endif
|
1835
|
+
#undef SKIP_TAIL
|
1836
|
+
if (len >= sizeof(st_index_t)) {
|
1837
|
+
#if !UNALIGNED_WORD_ACCESS
|
1838
|
+
int align = (int)((st_data_t)data % sizeof(st_index_t));
|
1839
|
+
if (align) {
|
1840
|
+
st_index_t d = 0;
|
1841
|
+
int sl, sr, pack;
|
1842
|
+
|
1843
|
+
switch (align) {
|
1844
|
+
#ifdef WORDS_BIGENDIAN
|
1845
|
+
# define UNALIGNED_ADD(n) case SIZEOF_ST_INDEX_T - (n) - 1: \
|
1846
|
+
t |= data_at(n) << CHAR_BIT*(SIZEOF_ST_INDEX_T - (n) - 2)
|
1847
|
+
#else
|
1848
|
+
# define UNALIGNED_ADD(n) case SIZEOF_ST_INDEX_T - (n) - 1: \
|
1849
|
+
t |= data_at(n) << CHAR_BIT*(n)
|
1850
|
+
#endif
|
1851
|
+
UNALIGNED_ADD_ALL;
|
1852
|
+
#undef UNALIGNED_ADD
|
1853
|
+
}
|
1854
|
+
|
1855
|
+
#ifdef WORDS_BIGENDIAN
|
1856
|
+
t >>= (CHAR_BIT * align) - CHAR_BIT;
|
1857
|
+
#else
|
1858
|
+
t <<= (CHAR_BIT * align);
|
1859
|
+
#endif
|
1860
|
+
|
1861
|
+
data += sizeof(st_index_t)-align;
|
1862
|
+
len -= sizeof(st_index_t)-align;
|
1863
|
+
|
1864
|
+
sl = CHAR_BIT * (SIZEOF_ST_INDEX_T-align);
|
1865
|
+
sr = CHAR_BIT * align;
|
1866
|
+
|
1867
|
+
while (len >= sizeof(st_index_t)) {
|
1868
|
+
d = *(st_index_t *)data;
|
1869
|
+
#ifdef WORDS_BIGENDIAN
|
1870
|
+
t = (t << sr) | (d >> sl);
|
1871
|
+
#else
|
1872
|
+
t = (t >> sr) | (d << sl);
|
1873
|
+
#endif
|
1874
|
+
h = murmur_step(h, t);
|
1875
|
+
t = d;
|
1876
|
+
data += sizeof(st_index_t);
|
1877
|
+
len -= sizeof(st_index_t);
|
1878
|
+
}
|
1879
|
+
|
1880
|
+
pack = len < (size_t)align ? (int)len : align;
|
1881
|
+
d = 0;
|
1882
|
+
switch (pack) {
|
1883
|
+
#ifdef WORDS_BIGENDIAN
|
1884
|
+
# define UNALIGNED_ADD(n) case (n) + 1: \
|
1885
|
+
d |= data_at(n) << CHAR_BIT*(SIZEOF_ST_INDEX_T - (n) - 1)
|
1886
|
+
#else
|
1887
|
+
# define UNALIGNED_ADD(n) case (n) + 1: \
|
1888
|
+
d |= data_at(n) << CHAR_BIT*(n)
|
1889
|
+
#endif
|
1890
|
+
UNALIGNED_ADD_ALL;
|
1891
|
+
#undef UNALIGNED_ADD
|
1892
|
+
}
|
1893
|
+
#ifdef WORDS_BIGENDIAN
|
1894
|
+
t = (t << sr) | (d >> sl);
|
1895
|
+
#else
|
1896
|
+
t = (t >> sr) | (d << sl);
|
1897
|
+
#endif
|
1898
|
+
|
1899
|
+
if (len < (size_t)align) goto skip_tail;
|
1900
|
+
# define SKIP_TAIL 1
|
1901
|
+
h = murmur_step(h, t);
|
1902
|
+
data += pack;
|
1903
|
+
len -= pack;
|
1904
|
+
}
|
1905
|
+
else
|
1906
|
+
#endif
|
1907
|
+
#ifdef HAVE_BUILTIN___BUILTIN_ASSUME_ALIGNED
|
1908
|
+
#define aligned_data __builtin_assume_aligned(data, sizeof(st_index_t))
|
1909
|
+
#else
|
1910
|
+
#define aligned_data data
|
1911
|
+
#endif
|
1912
|
+
{
|
1913
|
+
do {
|
1914
|
+
h = murmur_step(h, *(st_index_t *)aligned_data);
|
1915
|
+
data += sizeof(st_index_t);
|
1916
|
+
len -= sizeof(st_index_t);
|
1917
|
+
} while (len >= sizeof(st_index_t));
|
1918
|
+
}
|
1919
|
+
}
|
1920
|
+
|
1921
|
+
t = 0;
|
1922
|
+
switch (len) {
|
1923
|
+
#if UNALIGNED_WORD_ACCESS && SIZEOF_ST_INDEX_T <= 8 && CHAR_BIT == 8
|
1924
|
+
/* in this case byteorder doesn't really matter */
|
1925
|
+
#if SIZEOF_ST_INDEX_T > 4
|
1926
|
+
case 7: t |= data_at(6) << 48;
|
1927
|
+
case 6: t |= data_at(5) << 40;
|
1928
|
+
case 5: t |= data_at(4) << 32;
|
1929
|
+
case 4:
|
1930
|
+
t |= (st_index_t)*(uint32_t*)aligned_data;
|
1931
|
+
goto skip_tail;
|
1932
|
+
# define SKIP_TAIL 1
|
1933
|
+
#endif
|
1934
|
+
case 3: t |= data_at(2) << 16;
|
1935
|
+
case 2: t |= data_at(1) << 8;
|
1936
|
+
case 1: t |= data_at(0);
|
1937
|
+
#else
|
1938
|
+
#ifdef WORDS_BIGENDIAN
|
1939
|
+
# define UNALIGNED_ADD(n) case (n) + 1: \
|
1940
|
+
t |= data_at(n) << CHAR_BIT*(SIZEOF_ST_INDEX_T - (n) - 1)
|
1941
|
+
#else
|
1942
|
+
# define UNALIGNED_ADD(n) case (n) + 1: \
|
1943
|
+
t |= data_at(n) << CHAR_BIT*(n)
|
1944
|
+
#endif
|
1945
|
+
UNALIGNED_ADD_ALL;
|
1946
|
+
#undef UNALIGNED_ADD
|
1947
|
+
#endif
|
1948
|
+
#ifdef SKIP_TAIL
|
1949
|
+
skip_tail:
|
1950
|
+
#endif
|
1951
|
+
h ^= t; h -= ROTL(t, 7);
|
1952
|
+
h *= C2;
|
1953
|
+
}
|
1954
|
+
h ^= l;
|
1955
|
+
#undef aligned_data
|
1956
|
+
|
1957
|
+
return murmur_finish(h);
|
1958
|
+
}
|
1959
|
+
|
1960
|
+
st_index_t
|
1961
|
+
st_hash_uint32(st_index_t h, uint32_t i)
|
1962
|
+
{
|
1963
|
+
return murmur_step(h, i);
|
1964
|
+
}
|
1965
|
+
|
1966
|
+
NO_SANITIZE("unsigned-integer-overflow", extern st_index_t st_hash_uint(st_index_t h, st_index_t i));
|
1967
|
+
st_index_t
|
1968
|
+
st_hash_uint(st_index_t h, st_index_t i)
|
1969
|
+
{
|
1970
|
+
i += h;
|
1971
|
+
/* no matter if it is BigEndian or LittleEndian,
|
1972
|
+
* we hash just integers */
|
1973
|
+
#if SIZEOF_ST_INDEX_T*CHAR_BIT > 8*8
|
1974
|
+
h = murmur_step(h, i >> 8*8);
|
1975
|
+
#endif
|
1976
|
+
h = murmur_step(h, i);
|
1977
|
+
return h;
|
1978
|
+
}
|
1979
|
+
|
1980
|
+
st_index_t
|
1981
|
+
st_hash_end(st_index_t h)
|
1982
|
+
{
|
1983
|
+
h = murmur_finish(h);
|
1984
|
+
return h;
|
1985
|
+
}
|
1986
|
+
|
1987
|
+
#undef st_hash_start
|
1988
|
+
st_index_t
|
1989
|
+
rb_st_hash_start(st_index_t h)
|
1990
|
+
{
|
1991
|
+
return h;
|
1992
|
+
}
|
1993
|
+
|
1994
|
+
static st_index_t
|
1995
|
+
strhash(st_data_t arg)
|
1996
|
+
{
|
1997
|
+
register const char *string = (const char *)arg;
|
1998
|
+
return st_hash(string, strlen(string), FNV1_32A_INIT);
|
1999
|
+
}
|
2000
|
+
|
2001
|
+
int
|
2002
|
+
st_locale_insensitive_strcasecmp(const char *s1, const char *s2)
|
2003
|
+
{
|
2004
|
+
char c1, c2;
|
2005
|
+
|
2006
|
+
while (1) {
|
2007
|
+
c1 = *s1++;
|
2008
|
+
c2 = *s2++;
|
2009
|
+
if (c1 == '\0' || c2 == '\0') {
|
2010
|
+
if (c1 != '\0') return 1;
|
2011
|
+
if (c2 != '\0') return -1;
|
2012
|
+
return 0;
|
2013
|
+
}
|
2014
|
+
if (('A' <= c1) && (c1 <= 'Z')) c1 += 'a' - 'A';
|
2015
|
+
if (('A' <= c2) && (c2 <= 'Z')) c2 += 'a' - 'A';
|
2016
|
+
if (c1 != c2) {
|
2017
|
+
if (c1 > c2)
|
2018
|
+
return 1;
|
2019
|
+
else
|
2020
|
+
return -1;
|
2021
|
+
}
|
2022
|
+
}
|
2023
|
+
}
|
2024
|
+
|
2025
|
+
int
|
2026
|
+
st_locale_insensitive_strncasecmp(const char *s1, const char *s2, size_t n)
|
2027
|
+
{
|
2028
|
+
char c1, c2;
|
2029
|
+
size_t i;
|
2030
|
+
|
2031
|
+
for (i = 0; i < n; i++) {
|
2032
|
+
c1 = *s1++;
|
2033
|
+
c2 = *s2++;
|
2034
|
+
if (c1 == '\0' || c2 == '\0') {
|
2035
|
+
if (c1 != '\0') return 1;
|
2036
|
+
if (c2 != '\0') return -1;
|
2037
|
+
return 0;
|
2038
|
+
}
|
2039
|
+
if (('A' <= c1) && (c1 <= 'Z')) c1 += 'a' - 'A';
|
2040
|
+
if (('A' <= c2) && (c2 <= 'Z')) c2 += 'a' - 'A';
|
2041
|
+
if (c1 != c2) {
|
2042
|
+
if (c1 > c2)
|
2043
|
+
return 1;
|
2044
|
+
else
|
2045
|
+
return -1;
|
2046
|
+
}
|
2047
|
+
}
|
2048
|
+
return 0;
|
2049
|
+
}
|
2050
|
+
|
2051
|
+
static int
|
2052
|
+
st_strcmp(st_data_t lhs, st_data_t rhs)
|
2053
|
+
{
|
2054
|
+
const char *s1 = (char *)lhs;
|
2055
|
+
const char *s2 = (char *)rhs;
|
2056
|
+
return strcmp(s1, s2);
|
2057
|
+
}
|
2058
|
+
|
2059
|
+
static int
|
2060
|
+
st_locale_insensitive_strcasecmp_i(st_data_t lhs, st_data_t rhs)
|
2061
|
+
{
|
2062
|
+
const char *s1 = (char *)lhs;
|
2063
|
+
const char *s2 = (char *)rhs;
|
2064
|
+
return st_locale_insensitive_strcasecmp(s1, s2);
|
2065
|
+
}
|
2066
|
+
|
2067
|
+
NO_SANITIZE("unsigned-integer-overflow", PUREFUNC(static st_index_t strcasehash(st_data_t)));
|
2068
|
+
static st_index_t
|
2069
|
+
strcasehash(st_data_t arg)
|
2070
|
+
{
|
2071
|
+
register const char *string = (const char *)arg;
|
2072
|
+
register st_index_t hval = FNV1_32A_INIT;
|
2073
|
+
|
2074
|
+
/*
|
2075
|
+
* FNV-1a hash each octet in the buffer
|
2076
|
+
*/
|
2077
|
+
while (*string) {
|
2078
|
+
unsigned int c = (unsigned char)*string++;
|
2079
|
+
if ((unsigned int)(c - 'A') <= ('Z' - 'A')) c += 'a' - 'A';
|
2080
|
+
hval ^= c;
|
2081
|
+
|
2082
|
+
/* multiply by the 32 bit FNV magic prime mod 2^32 */
|
2083
|
+
hval *= FNV_32_PRIME;
|
2084
|
+
}
|
2085
|
+
return hval;
|
2086
|
+
}
|
2087
|
+
|
2088
|
+
int
|
2089
|
+
st_numcmp(st_data_t x, st_data_t y)
|
2090
|
+
{
|
2091
|
+
return x != y;
|
2092
|
+
}
|
2093
|
+
|
2094
|
+
st_index_t
|
2095
|
+
st_numhash(st_data_t n)
|
2096
|
+
{
|
2097
|
+
enum {s1 = 11, s2 = 3};
|
2098
|
+
return (st_index_t)((n>>s1|(n<<s2)) ^ (n>>s2));
|
2099
|
+
}
|
2100
|
+
|
2101
|
+
#ifdef RUBY
|
2102
|
+
/* Expand TAB to be suitable for holding SIZ entries in total.
|
2103
|
+
Pre-existing entries remain not deleted inside of TAB, but its bins
|
2104
|
+
are cleared to expect future reconstruction. See rehash below. */
|
2105
|
+
static void
|
2106
|
+
st_expand_table(st_table *tab, st_index_t siz)
|
2107
|
+
{
|
2108
|
+
st_table *tmp;
|
2109
|
+
st_index_t n;
|
2110
|
+
|
2111
|
+
if (siz <= get_allocated_entries(tab))
|
2112
|
+
return; /* enough room already */
|
2113
|
+
|
2114
|
+
tmp = st_init_table_with_size(tab->type, siz);
|
2115
|
+
n = get_allocated_entries(tab);
|
2116
|
+
MEMCPY(tmp->entries, tab->entries, st_table_entry, n);
|
2117
|
+
free(tab->entries);
|
2118
|
+
free(tab->bins);
|
2119
|
+
free(tmp->bins);
|
2120
|
+
tab->entry_power = tmp->entry_power;
|
2121
|
+
tab->bin_power = tmp->bin_power;
|
2122
|
+
tab->size_ind = tmp->size_ind;
|
2123
|
+
tab->entries = tmp->entries;
|
2124
|
+
tab->bins = NULL;
|
2125
|
+
tab->rebuilds_num++;
|
2126
|
+
free(tmp);
|
2127
|
+
}
|
2128
|
+
|
2129
|
+
/* Rehash using linear search. Return TRUE if we found that the table
|
2130
|
+
was rebuilt. */
|
2131
|
+
static int
|
2132
|
+
st_rehash_linear(st_table *tab)
|
2133
|
+
{
|
2134
|
+
int eq_p, rebuilt_p;
|
2135
|
+
st_index_t i, j;
|
2136
|
+
st_table_entry *p, *q;
|
2137
|
+
|
2138
|
+
free(tab->bins);
|
2139
|
+
tab->bins = NULL;
|
2140
|
+
|
2141
|
+
for (i = tab->entries_start; i < tab->entries_bound; i++) {
|
2142
|
+
p = &tab->entries[i];
|
2143
|
+
if (DELETED_ENTRY_P(p))
|
2144
|
+
continue;
|
2145
|
+
for (j = i + 1; j < tab->entries_bound; j++) {
|
2146
|
+
q = &tab->entries[j];
|
2147
|
+
if (DELETED_ENTRY_P(q))
|
2148
|
+
continue;
|
2149
|
+
DO_PTR_EQUAL_CHECK(tab, p, q->hash, q->key, eq_p, rebuilt_p);
|
2150
|
+
if (EXPECT(rebuilt_p, 0))
|
2151
|
+
return TRUE;
|
2152
|
+
if (eq_p) {
|
2153
|
+
*p = *q;
|
2154
|
+
MARK_ENTRY_DELETED(q);
|
2155
|
+
tab->num_entries--;
|
2156
|
+
update_range_for_deleted(tab, j);
|
2157
|
+
}
|
2158
|
+
}
|
2159
|
+
}
|
2160
|
+
return FALSE;
|
2161
|
+
}
|
2162
|
+
|
2163
|
+
/* Rehash using index. Return TRUE if we found that the table was
|
2164
|
+
rebuilt. */
|
2165
|
+
static int
|
2166
|
+
st_rehash_indexed(st_table *tab)
|
2167
|
+
{
|
2168
|
+
int eq_p, rebuilt_p;
|
2169
|
+
st_index_t i;
|
2170
|
+
st_index_t const n = bins_size(tab);
|
2171
|
+
unsigned int const size_ind = get_size_ind(tab);
|
2172
|
+
st_index_t *bins = realloc(tab->bins, n);
|
2173
|
+
tab->bins = bins;
|
2174
|
+
initialize_bins(tab);
|
2175
|
+
for (i = tab->entries_start; i < tab->entries_bound; i++) {
|
2176
|
+
st_table_entry *p = &tab->entries[i];
|
2177
|
+
st_index_t ind;
|
2178
|
+
#ifdef QUADRATIC_PROBE
|
2179
|
+
st_index_t d = 1;
|
2180
|
+
#else
|
2181
|
+
st_index_t perturb = p->hash;
|
2182
|
+
#endif
|
2183
|
+
|
2184
|
+
if (DELETED_ENTRY_P(p))
|
2185
|
+
continue;
|
2186
|
+
|
2187
|
+
ind = hash_bin(p->hash, tab);
|
2188
|
+
for (;;) {
|
2189
|
+
st_index_t bin = get_bin(bins, size_ind, ind);
|
2190
|
+
if (EMPTY_OR_DELETED_BIN_P(bin)) {
|
2191
|
+
/* ok, new room */
|
2192
|
+
set_bin(bins, size_ind, ind, i + ENTRY_BASE);
|
2193
|
+
break;
|
2194
|
+
}
|
2195
|
+
else {
|
2196
|
+
st_table_entry *q = &tab->entries[bin - ENTRY_BASE];
|
2197
|
+
DO_PTR_EQUAL_CHECK(tab, q, p->hash, p->key, eq_p, rebuilt_p);
|
2198
|
+
if (EXPECT(rebuilt_p, 0))
|
2199
|
+
return TRUE;
|
2200
|
+
if (eq_p) {
|
2201
|
+
/* duplicated key; delete it */
|
2202
|
+
q->record = p->record;
|
2203
|
+
MARK_ENTRY_DELETED(p);
|
2204
|
+
tab->num_entries--;
|
2205
|
+
update_range_for_deleted(tab, bin);
|
2206
|
+
break;
|
2207
|
+
}
|
2208
|
+
else {
|
2209
|
+
/* hash collision; skip it */
|
2210
|
+
#ifdef QUADRATIC_PROBE
|
2211
|
+
ind = hash_bin(ind + d, tab);
|
2212
|
+
d++;
|
2213
|
+
#else
|
2214
|
+
ind = secondary_hash(ind, tab, &perturb);
|
2215
|
+
#endif
|
2216
|
+
}
|
2217
|
+
}
|
2218
|
+
}
|
2219
|
+
}
|
2220
|
+
return FALSE;
|
2221
|
+
}
|
2222
|
+
|
2223
|
+
/* Reconstruct TAB's bins according to TAB's entries. This function
|
2224
|
+
permits conflicting keys inside of entries. No errors are reported
|
2225
|
+
then. All but one of them are discarded silently. */
|
2226
|
+
static void
|
2227
|
+
st_rehash(st_table *tab)
|
2228
|
+
{
|
2229
|
+
int rebuilt_p;
|
2230
|
+
|
2231
|
+
do {
|
2232
|
+
if (tab->bin_power <= MAX_POWER2_FOR_TABLES_WITHOUT_BINS)
|
2233
|
+
rebuilt_p = st_rehash_linear(tab);
|
2234
|
+
else
|
2235
|
+
rebuilt_p = st_rehash_indexed(tab);
|
2236
|
+
} while (rebuilt_p);
|
2237
|
+
}
|
2238
|
+
|
2239
|
+
static st_data_t
|
2240
|
+
st_stringify(VALUE key)
|
2241
|
+
{
|
2242
|
+
return (rb_obj_class(key) == rb_cString && !RB_OBJ_FROZEN(key)) ?
|
2243
|
+
rb_hash_key_str(key) : key;
|
2244
|
+
}
|
2245
|
+
|
2246
|
+
static void
|
2247
|
+
st_insert_single(st_table *tab, VALUE hash, VALUE key, VALUE val)
|
2248
|
+
{
|
2249
|
+
st_data_t k = st_stringify(key);
|
2250
|
+
st_table_entry e;
|
2251
|
+
e.hash = do_hash(k, tab);
|
2252
|
+
e.key = k;
|
2253
|
+
e.record = val;
|
2254
|
+
|
2255
|
+
tab->entries[tab->entries_bound++] = e;
|
2256
|
+
tab->num_entries++;
|
2257
|
+
RB_OBJ_WRITTEN(hash, Qundef, k);
|
2258
|
+
RB_OBJ_WRITTEN(hash, Qundef, val);
|
2259
|
+
}
|
2260
|
+
|
2261
|
+
static void
|
2262
|
+
st_insert_linear(st_table *tab, long argc, const VALUE *argv, VALUE hash)
|
2263
|
+
{
|
2264
|
+
long i;
|
2265
|
+
|
2266
|
+
for (i = 0; i < argc; /* */) {
|
2267
|
+
st_data_t k = st_stringify(argv[i++]);
|
2268
|
+
st_data_t v = argv[i++];
|
2269
|
+
st_insert(tab, k, v);
|
2270
|
+
RB_OBJ_WRITTEN(hash, Qundef, k);
|
2271
|
+
RB_OBJ_WRITTEN(hash, Qundef, v);
|
2272
|
+
}
|
2273
|
+
}
|
2274
|
+
|
2275
|
+
static void
|
2276
|
+
st_insert_generic(st_table *tab, long argc, const VALUE *argv, VALUE hash)
|
2277
|
+
{
|
2278
|
+
long i;
|
2279
|
+
|
2280
|
+
/* push elems */
|
2281
|
+
for (i = 0; i < argc; /* */) {
|
2282
|
+
VALUE key = argv[i++];
|
2283
|
+
VALUE val = argv[i++];
|
2284
|
+
st_insert_single(tab, hash, key, val);
|
2285
|
+
}
|
2286
|
+
|
2287
|
+
/* reindex */
|
2288
|
+
st_rehash(tab);
|
2289
|
+
}
|
2290
|
+
|
2291
|
+
/* Mimics ruby's { foo => bar } syntax. This function is subpart
|
2292
|
+
of rb_hash_bulk_insert. */
|
2293
|
+
void
|
2294
|
+
rb_hash_bulk_insert_into_st_table(long argc, const VALUE *argv, VALUE hash)
|
2295
|
+
{
|
2296
|
+
st_index_t n, size = argc / 2;
|
2297
|
+
st_table *tab = RHASH_ST_TABLE(hash);
|
2298
|
+
|
2299
|
+
tab = RHASH_TBL_RAW(hash);
|
2300
|
+
n = tab->entries_bound + size;
|
2301
|
+
st_expand_table(tab, n);
|
2302
|
+
if (UNLIKELY(tab->num_entries))
|
2303
|
+
st_insert_generic(tab, argc, argv, hash);
|
2304
|
+
else if (argc <= 2)
|
2305
|
+
st_insert_single(tab, hash, argv[0], argv[1]);
|
2306
|
+
else if (tab->bin_power <= MAX_POWER2_FOR_TABLES_WITHOUT_BINS)
|
2307
|
+
st_insert_linear(tab, argc, argv, hash);
|
2308
|
+
else
|
2309
|
+
st_insert_generic(tab, argc, argv, hash);
|
2310
|
+
}
|
2311
|
+
|
2312
|
+
// to iterate iv_index_tbl
|
2313
|
+
st_data_t
|
2314
|
+
rb_st_nth_key(st_table *tab, st_index_t index)
|
2315
|
+
{
|
2316
|
+
if (LIKELY(tab->entries_start == 0 &&
|
2317
|
+
tab->num_entries == tab->entries_bound &&
|
2318
|
+
index < tab->num_entries)) {
|
2319
|
+
return tab->entries[index].key;
|
2320
|
+
}
|
2321
|
+
else {
|
2322
|
+
rb_bug("unreachable");
|
2323
|
+
}
|
2324
|
+
}
|
2325
|
+
|
2326
|
+
void
|
2327
|
+
rb_st_compact_table(st_table *tab)
|
2328
|
+
{
|
2329
|
+
st_index_t num = tab->num_entries;
|
2330
|
+
if (REBUILD_THRESHOLD * num <= get_allocated_entries(tab)) {
|
2331
|
+
/* Compaction: */
|
2332
|
+
st_table *new_tab = st_init_table_with_size(tab->type, 2 * num);
|
2333
|
+
rebuild_table_with(new_tab, tab);
|
2334
|
+
rebuild_move_table(new_tab, tab);
|
2335
|
+
rebuild_cleanup(tab);
|
2336
|
+
}
|
2337
|
+
}
|
2338
|
+
|
2339
|
+
#endif
|