Coverage Report

Created: 2026-04-09 11:41

next uncovered line (L), next uncovered region (R), next uncovered branch (B)
/work/workdir/UnpackedTarball/libxml2/dict.c
Line
Count
Source
1
/*
2
 * dict.c: dictionary of reusable strings, just used to avoid allocation
3
 *         and freeing operations.
4
 *
5
 * Copyright (C) 2003-2012 Daniel Veillard.
6
 *
7
 * Permission to use, copy, modify, and distribute this software for any
8
 * purpose with or without fee is hereby granted, provided that the above
9
 * copyright notice and this permission notice appear in all copies.
10
 *
11
 * THIS SOFTWARE IS PROVIDED ``AS IS'' AND WITHOUT ANY EXPRESS OR IMPLIED
12
 * WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF
13
 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE AUTHORS AND
14
 * CONTRIBUTORS ACCEPT NO RESPONSIBILITY IN ANY CONCEIVABLE MANNER.
15
 *
16
 * Author: Daniel Veillard
17
 */
18
19
#define IN_LIBXML
20
#include "libxml.h"
21
22
#include <errno.h>
23
#include <limits.h>
24
#include <stdlib.h>
25
#include <string.h>
26
27
#include "private/dict.h"
28
#include "private/error.h"
29
#include "private/globals.h"
30
#include "private/threads.h"
31
32
#include <libxml/parser.h>
33
#include <libxml/dict.h>
34
#include <libxml/xmlmemory.h>
35
#include <libxml/xmlstring.h>
36
37
#ifndef SIZE_MAX
38
  #define SIZE_MAX ((size_t) -1)
39
#endif
40
41
8.89M
#define MAX_FILL_NUM 7
42
8.89M
#define MAX_FILL_DENOM 8
43
218k
#define MIN_HASH_SIZE 8
44
81.0M
#define MAX_HASH_SIZE (1u << 31)
45
46
typedef struct _xmlDictStrings xmlDictStrings;
47
typedef xmlDictStrings *xmlDictStringsPtr;
48
struct _xmlDictStrings {
49
    xmlDictStringsPtr next;
50
    xmlChar *free;
51
    xmlChar *end;
52
    size_t size;
53
    size_t nbStrings;
54
    xmlChar array[1];
55
};
56
57
typedef xmlHashedString xmlDictEntry;
58
59
/*
60
 * The entire dictionary
61
 */
62
struct _xmlDict {
63
    int ref_counter;
64
65
    xmlDictEntry *table;
66
    size_t size;
67
    unsigned int nbElems;
68
    xmlDictStringsPtr strings;
69
70
    struct _xmlDict *subdict;
71
    /* used for randomization */
72
    unsigned seed;
73
    /* used to impose a limit on size */
74
    size_t limit;
75
};
76
77
/*
78
 * A mutex for modifying the reference counter for shared
79
 * dictionaries.
80
 */
81
static xmlMutex xmlDictMutex;
82
83
/**
84
 * @deprecated Alias for #xmlInitParser.
85
 *
86
 * @returns 0.
87
 */
88
int
89
0
xmlInitializeDict(void) {
90
0
    xmlInitParser();
91
0
    return(0);
92
0
}
93
94
/**
95
 * Initialize mutex.
96
 */
97
void
98
19
xmlInitDictInternal(void) {
99
19
    xmlInitMutex(&xmlDictMutex);
100
19
}
101
102
/**
103
 * @deprecated This function is a no-op. Call #xmlCleanupParser
104
 * to free global state but see the warnings there. #xmlCleanupParser
105
 * should be only called once at program exit. In most cases, you don't
106
 * have call cleanup functions at all.
107
 */
108
void
109
0
xmlDictCleanup(void) {
110
0
}
111
112
/**
113
 * Free the dictionary mutex.
114
 */
115
void
116
0
xmlCleanupDictInternal(void) {
117
0
    xmlCleanupMutex(&xmlDictMutex);
118
0
}
119
120
/*
121
 * @param dict  the dictionary
122
 * @param name  the name of the userdata
123
 * @param len  the length of the name
124
 *
125
 * Add the string to the array[s]
126
 *
127
 * @returns the pointer of the local string, or NULL in case of error.
128
 */
129
static const xmlChar *
130
9.11M
xmlDictAddString(xmlDictPtr dict, const xmlChar *name, unsigned int namelen) {
131
9.11M
    xmlDictStringsPtr pool;
132
9.11M
    const xmlChar *ret;
133
9.11M
    size_t size = 0; /* + sizeof(_xmlDictStrings) == 1024 */
134
9.11M
    size_t limit = 0;
135
136
9.11M
    pool = dict->strings;
137
9.15M
    while (pool != NULL) {
138
8.89M
  if ((size_t)(pool->end - pool->free) > namelen)
139
8.85M
      goto found_pool;
140
42.2k
  if (pool->size > size) size = pool->size;
141
42.2k
        limit += pool->size;
142
42.2k
  pool = pool->next;
143
42.2k
    }
144
    /*
145
     * Not found, need to allocate
146
     */
147
259k
    if (pool == NULL) {
148
259k
        if ((dict->limit > 0) && (limit > dict->limit)) {
149
0
            return(NULL);
150
0
        }
151
152
259k
        if (size == 0) {
153
218k
            size = 1000;
154
218k
        } else {
155
40.4k
            if (size < (SIZE_MAX - sizeof(xmlDictStrings)) / 4)
156
40.4k
                size *= 4; /* exponential growth */
157
0
            else
158
0
                size = SIZE_MAX - sizeof(xmlDictStrings);
159
40.4k
        }
160
259k
        if (size / 4 < namelen) {
161
531
            if ((size_t) namelen + 0 < (SIZE_MAX - sizeof(xmlDictStrings)) / 4)
162
531
                size = 4 * (size_t) namelen; /* just in case ! */
163
0
            else
164
0
                return(NULL);
165
531
        }
166
259k
  pool = (xmlDictStringsPtr) xmlMalloc(sizeof(xmlDictStrings) + size);
167
259k
  if (pool == NULL)
168
0
      return(NULL);
169
259k
  pool->size = size;
170
259k
  pool->nbStrings = 0;
171
259k
  pool->free = &pool->array[0];
172
259k
  pool->end = &pool->array[size];
173
259k
  pool->next = dict->strings;
174
259k
  dict->strings = pool;
175
259k
    }
176
9.11M
found_pool:
177
9.11M
    ret = pool->free;
178
9.11M
    memcpy(pool->free, name, namelen);
179
9.11M
    pool->free += namelen;
180
9.11M
    *(pool->free++) = 0;
181
9.11M
    pool->nbStrings++;
182
9.11M
    return(ret);
183
259k
}
184
185
/*
186
 * @param dict  the dictionary
187
 * @param prefix  the prefix of the userdata
188
 * @param plen  the prefix length
189
 * @param name  the name of the userdata
190
 * @param len  the length of the name
191
 *
192
 * Add the QName to the array[s]
193
 *
194
 * @returns the pointer of the local string, or NULL in case of error.
195
 */
196
static const xmlChar *
197
xmlDictAddQString(xmlDictPtr dict, const xmlChar *prefix, unsigned int plen,
198
                 const xmlChar *name, unsigned int namelen)
199
383
{
200
383
    xmlDictStringsPtr pool;
201
383
    const xmlChar *ret;
202
383
    size_t size = 0; /* + sizeof(_xmlDictStrings) == 1024 */
203
383
    size_t limit = 0;
204
205
383
    pool = dict->strings;
206
383
    while (pool != NULL) {
207
383
  if ((size_t)(pool->end - pool->free) > namelen + plen + 1)
208
383
      goto found_pool;
209
0
  if (pool->size > size) size = pool->size;
210
0
        limit += pool->size;
211
0
  pool = pool->next;
212
0
    }
213
    /*
214
     * Not found, need to allocate
215
     */
216
0
    if (pool == NULL) {
217
0
        if ((dict->limit > 0) && (limit > dict->limit)) {
218
0
            return(NULL);
219
0
        }
220
221
0
        if (size == 0) size = 1000;
222
0
  else size *= 4; /* exponential growth */
223
0
        if (size < 4 * (namelen + plen + 1))
224
0
      size = 4 * (namelen + plen + 1); /* just in case ! */
225
0
  pool = (xmlDictStringsPtr) xmlMalloc(sizeof(xmlDictStrings) + size);
226
0
  if (pool == NULL)
227
0
      return(NULL);
228
0
  pool->size = size;
229
0
  pool->nbStrings = 0;
230
0
  pool->free = &pool->array[0];
231
0
  pool->end = &pool->array[size];
232
0
  pool->next = dict->strings;
233
0
  dict->strings = pool;
234
0
    }
235
383
found_pool:
236
383
    ret = pool->free;
237
383
    memcpy(pool->free, prefix, plen);
238
383
    pool->free += plen;
239
383
    *(pool->free++) = ':';
240
383
    memcpy(pool->free, name, namelen);
241
383
    pool->free += namelen;
242
383
    *(pool->free++) = 0;
243
383
    pool->nbStrings++;
244
383
    return(ret);
245
0
}
246
247
/**
248
 * Create a new dictionary
249
 *
250
 * @returns the newly created dictionary, or NULL if an error occurred.
251
 */
252
xmlDict *
253
218k
xmlDictCreate(void) {
254
218k
    xmlDictPtr dict;
255
256
218k
    xmlInitParser();
257
258
218k
    dict = xmlMalloc(sizeof(xmlDict));
259
218k
    if (dict == NULL)
260
0
        return(NULL);
261
218k
    dict->ref_counter = 1;
262
218k
    dict->limit = 0;
263
264
218k
    dict->size = 0;
265
218k
    dict->nbElems = 0;
266
218k
    dict->table = NULL;
267
218k
    dict->strings = NULL;
268
218k
    dict->subdict = NULL;
269
218k
    dict->seed = xmlRandom();
270
218k
#ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
271
218k
    dict->seed = 0;
272
218k
#endif
273
218k
    return(dict);
274
218k
}
275
276
/**
277
 * Create a new dictionary, inheriting strings from the read-only
278
 * dictionary `sub`. On lookup, strings are first searched in the
279
 * new dictionary, then in `sub`, and if not found are created in the
280
 * new dictionary.
281
 *
282
 * @param sub  an existing dictionary
283
 * @returns the newly created dictionary, or NULL if an error occurred.
284
 */
285
xmlDict *
286
0
xmlDictCreateSub(xmlDict *sub) {
287
0
    xmlDictPtr dict = xmlDictCreate();
288
289
0
    if ((dict != NULL) && (sub != NULL)) {
290
0
        dict->seed = sub->seed;
291
0
        dict->subdict = sub;
292
0
  xmlDictReference(dict->subdict);
293
0
    }
294
0
    return(dict);
295
0
}
296
297
/**
298
 * Increment the reference counter of a dictionary
299
 *
300
 * @param dict  the dictionary
301
 * @returns 0 in case of success and -1 in case of error
302
 */
303
int
304
12.8k
xmlDictReference(xmlDict *dict) {
305
12.8k
    if (dict == NULL) return -1;
306
11.2k
    xmlMutexLock(&xmlDictMutex);
307
11.2k
    dict->ref_counter++;
308
11.2k
    xmlMutexUnlock(&xmlDictMutex);
309
11.2k
    return(0);
310
12.8k
}
311
312
/**
313
 * Free the hash `dict` and its contents. The userdata is
314
 * deallocated with `f` if provided.
315
 *
316
 * @param dict  the dictionary
317
 */
318
void
319
230k
xmlDictFree(xmlDict *dict) {
320
230k
    xmlDictStringsPtr pool, nextp;
321
322
230k
    if (dict == NULL)
323
0
  return;
324
325
    /* decrement the counter, it may be shared by a parser and docs */
326
230k
    xmlMutexLock(&xmlDictMutex);
327
230k
    dict->ref_counter--;
328
230k
    if (dict->ref_counter > 0) {
329
11.2k
        xmlMutexUnlock(&xmlDictMutex);
330
11.2k
        return;
331
11.2k
    }
332
333
218k
    xmlMutexUnlock(&xmlDictMutex);
334
335
218k
    if (dict->subdict != NULL) {
336
0
        xmlDictFree(dict->subdict);
337
0
    }
338
339
218k
    if (dict->table) {
340
218k
  xmlFree(dict->table);
341
218k
    }
342
218k
    pool = dict->strings;
343
478k
    while (pool != NULL) {
344
259k
        nextp = pool->next;
345
259k
  xmlFree(pool);
346
259k
  pool = nextp;
347
259k
    }
348
218k
    xmlFree(dict);
349
218k
}
350
351
/**
352
 * check if a string is owned by the dictionary
353
 *
354
 * @param dict  the dictionary
355
 * @param str  the string
356
 * @returns 1 if true, 0 if false and -1 in case of error
357
 * -1 in case of error
358
 */
359
int
360
3.13M
xmlDictOwns(xmlDict *dict, const xmlChar *str) {
361
3.13M
    xmlDictStringsPtr pool;
362
363
3.13M
    if ((dict == NULL) || (str == NULL))
364
0
  return(-1);
365
3.13M
    pool = dict->strings;
366
3.98M
    while (pool != NULL) {
367
3.29M
        if ((str >= &pool->array[0]) && (str <= pool->free))
368
2.43M
      return(1);
369
853k
  pool = pool->next;
370
853k
    }
371
695k
    if (dict->subdict)
372
0
        return(xmlDictOwns(dict->subdict, str));
373
695k
    return(0);
374
695k
}
375
376
/**
377
 * Query the number of elements installed in the hash `dict`.
378
 *
379
 * @param dict  the dictionary
380
 * @returns the number of elements in the dictionary or
381
 * -1 in case of error
382
 */
383
int
384
0
xmlDictSize(xmlDict *dict) {
385
0
    if (dict == NULL)
386
0
  return(-1);
387
0
    if (dict->subdict)
388
0
        return(dict->nbElems + dict->subdict->nbElems);
389
0
    return(dict->nbElems);
390
0
}
391
392
/**
393
 * Set a size limit for the dictionary
394
 * Added in 2.9.0
395
 *
396
 * @param dict  the dictionary
397
 * @param limit  the limit in bytes
398
 * @returns the previous limit of the dictionary or 0
399
 */
400
size_t
401
218k
xmlDictSetLimit(xmlDict *dict, size_t limit) {
402
218k
    size_t ret;
403
404
218k
    if (dict == NULL)
405
0
  return(0);
406
218k
    ret = dict->limit;
407
218k
    dict->limit = limit;
408
218k
    return(ret);
409
218k
}
410
411
/**
412
 * Get how much memory is used by a dictionary for strings
413
 * Added in 2.9.0
414
 *
415
 * @param dict  the dictionary
416
 * @returns the amount of strings allocated
417
 */
418
size_t
419
0
xmlDictGetUsage(xmlDict *dict) {
420
0
    xmlDictStringsPtr pool;
421
0
    size_t limit = 0;
422
423
0
    if (dict == NULL)
424
0
  return(0);
425
0
    pool = dict->strings;
426
0
    while (pool != NULL) {
427
0
        limit += pool->size;
428
0
  pool = pool->next;
429
0
    }
430
0
    return(limit);
431
0
}
432
433
/*****************************************************************
434
 *
435
 * The code below was rewritten and is additionally licensed under
436
 * the main license in file 'Copyright'.
437
 *
438
 *****************************************************************/
439
440
ATTRIBUTE_NO_SANITIZE_INTEGER
441
static unsigned
442
xmlDictHashName(unsigned seed, const xmlChar* data, size_t maxLen,
443
80.4M
                size_t *plen) {
444
80.4M
    unsigned h1, h2;
445
80.4M
    size_t i;
446
447
80.4M
    HASH_INIT(h1, h2, seed);
448
449
646M
    for (i = 0; i < maxLen && data[i]; i++) {
450
566M
        HASH_UPDATE(h1, h2, data[i]);
451
566M
    }
452
453
80.4M
    HASH_FINISH(h1, h2);
454
455
80.4M
    *plen = i;
456
80.4M
    return(h2 | MAX_HASH_SIZE);
457
80.4M
}
458
459
ATTRIBUTE_NO_SANITIZE_INTEGER
460
static unsigned
461
xmlDictHashQName(unsigned seed, const xmlChar *prefix, const xmlChar *name,
462
655
                 size_t *pplen, size_t *plen) {
463
655
    unsigned h1, h2;
464
655
    size_t i;
465
466
655
    HASH_INIT(h1, h2, seed);
467
468
1.66k
    for (i = 0; prefix[i] != 0; i++) {
469
1.01k
        HASH_UPDATE(h1, h2, prefix[i]);
470
1.01k
    }
471
655
    *pplen = i;
472
473
655
    HASH_UPDATE(h1, h2, ':');
474
475
4.63k
    for (i = 0; name[i] != 0; i++) {
476
3.98k
        HASH_UPDATE(h1, h2, name[i]);
477
3.98k
    }
478
655
    *plen = i;
479
480
655
    HASH_FINISH(h1, h2);
481
482
    /*
483
     * Always set the upper bit of hash values since 0 means an unoccupied
484
     * bucket.
485
     */
486
655
    return(h2 | MAX_HASH_SIZE);
487
655
}
488
489
/**
490
 * Compute the hash value of a C string.
491
 *
492
 * @param dict  dictionary
493
 * @param string  C string
494
 * @returns the hash value.
495
 */
496
unsigned
497
810k
xmlDictComputeHash(const xmlDict *dict, const xmlChar *string) {
498
810k
    size_t len;
499
810k
    return(xmlDictHashName(dict->seed, string, SIZE_MAX, &len));
500
810k
}
501
502
16.1M
#define HASH_ROL31(x,n) ((x) << (n) | ((x) & 0x7FFFFFFF) >> (31 - (n)))
503
504
/**
505
 * Combine two hash values.
506
 *
507
 * @param v1  first hash value
508
 * @param v2  second hash value
509
 * @returns the combined hash value.
510
 */
511
ATTRIBUTE_NO_SANITIZE_INTEGER
512
unsigned
513
16.1M
xmlDictCombineHash(unsigned v1, unsigned v2) {
514
    /*
515
     * The upper bit of hash values is always set, so we have to operate on
516
     * 31-bit hashes here.
517
     */
518
16.1M
    v1 ^= v2;
519
16.1M
    v1 += HASH_ROL31(v2, 5);
520
521
16.1M
    return((v1 & 0xFFFFFFFF) | 0x80000000);
522
16.1M
}
523
524
/**
525
 * Try to find a matching hash table entry. If an entry was found, set
526
 * `found` to 1 and return the entry. Otherwise, set `found` to 0 and return
527
 * the location where a new entry should be inserted.
528
 *
529
 * @param dict  dict
530
 * @param prefix  optional QName prefix
531
 * @param name  string
532
 * @param len  length of string
533
 * @param hashValue  valid hash value of string
534
 * @param pfound  result of search
535
 */
536
ATTRIBUTE_NO_SANITIZE_INTEGER
537
static xmlDictEntry *
538
xmlDictFindEntry(const xmlDict *dict, const xmlChar *prefix,
539
                 const xmlChar *name, int len, unsigned hashValue,
540
79.4M
                 int *pfound) {
541
79.4M
    xmlDictEntry *entry;
542
79.4M
    unsigned mask, pos, displ;
543
79.4M
    int found = 0;
544
545
79.4M
    mask = dict->size - 1;
546
79.4M
    pos = hashValue & mask;
547
79.4M
    entry = &dict->table[pos];
548
549
79.4M
    if (entry->hashValue != 0) {
550
        /*
551
         * Robin hood hashing: abort if the displacement of the entry
552
         * is smaller than the displacement of the key we look for.
553
         * This also stops at the correct position when inserting.
554
         */
555
75.8M
        displ = 0;
556
557
135M
        do {
558
135M
            if (entry->hashValue == hashValue) {
559
70.5M
                if (prefix == NULL) {
560
                    /*
561
                     * name is not necessarily null-terminated.
562
                     */
563
70.5M
                    if ((strncmp((const char *) entry->name,
564
70.5M
                                 (const char *) name, len) == 0) &&
565
70.5M
                        (entry->name[len] == 0)) {
566
70.5M
                        found = 1;
567
70.5M
                        break;
568
70.5M
                    }
569
70.5M
                } else {
570
272
                    if (xmlStrQEqual(prefix, name, entry->name)) {
571
272
                        found = 1;
572
272
                        break;
573
272
                    }
574
223
                }
575
70.5M
            }
576
577
65.3M
            displ++;
578
65.3M
            pos++;
579
65.3M
            entry++;
580
65.3M
            if ((pos & mask) == 0)
581
797k
                entry = dict->table;
582
65.3M
        } while ((entry->hashValue != 0) &&
583
63.1M
                 (((pos - entry->hashValue) & mask) >= displ));
584
75.8M
    }
585
586
79.4M
    *pfound = found;
587
79.4M
    return(entry);
588
79.4M
}
589
590
/**
591
 * Resize the dictionary hash table.
592
 *
593
 * @param dict  dictionary
594
 * @param size  new size of the dictionary
595
 * @returns 0 in case of success, -1 if a memory allocation failed.
596
 */
597
static int
598
735k
xmlDictGrow(xmlDictPtr dict, unsigned size) {
599
735k
    const xmlDictEntry *oldentry, *oldend, *end;
600
735k
    xmlDictEntry *table;
601
735k
    unsigned oldsize, i;
602
603
    /* Add 0 to avoid spurious -Wtype-limits warning on 64-bit GCC */
604
735k
    if ((size_t) size + 0 > SIZE_MAX / sizeof(table[0]))
605
0
        return(-1);
606
735k
    table = xmlMalloc(size * sizeof(table[0]));
607
735k
    if (table == NULL)
608
0
        return(-1);
609
735k
    memset(table, 0, size * sizeof(table[0]));
610
611
735k
    oldsize = dict->size;
612
735k
    if (oldsize == 0)
613
218k
        goto done;
614
615
516k
    oldend = &dict->table[oldsize];
616
516k
    end = &table[size];
617
618
    /*
619
     * Robin Hood sorting order is maintained if we
620
     *
621
     * - compute dict indices with modulo
622
     * - resize by an integer factor
623
     * - start to copy from the beginning of a probe sequence
624
     */
625
516k
    oldentry = dict->table;
626
3.66M
    while (oldentry->hashValue != 0) {
627
3.14M
        if (++oldentry >= oldend)
628
0
            oldentry = dict->table;
629
3.14M
    }
630
631
13.8M
    for (i = 0; i < oldsize; i++) {
632
13.2M
        if (oldentry->hashValue != 0) {
633
11.6M
            xmlDictEntry *entry = &table[oldentry->hashValue & (size - 1)];
634
635
15.2M
            while (entry->hashValue != 0) {
636
3.63M
                if (++entry >= end)
637
82.5k
                    entry = table;
638
3.63M
            }
639
11.6M
            *entry = *oldentry;
640
11.6M
        }
641
642
13.2M
        if (++oldentry >= oldend)
643
516k
            oldentry = dict->table;
644
13.2M
    }
645
646
516k
    xmlFree(dict->table);
647
648
735k
done:
649
735k
    dict->table = table;
650
735k
    dict->size = size;
651
652
735k
    return(0);
653
516k
}
654
655
/**
656
 * Internal lookup and update function.
657
 *
658
 * @param dict  dict
659
 * @param prefix  optional QName prefix
660
 * @param name  string
661
 * @param maybeLen  length of string or -1 if unknown
662
 * @param update  whether the string should be added
663
 */
664
ATTRIBUTE_NO_SANITIZE_INTEGER
665
static const xmlDictEntry *
666
xmlDictLookupInternal(xmlDict *dict, const xmlChar *prefix,
667
79.6M
                      const xmlChar *name, int maybeLen, int update) {
668
79.6M
    xmlDictEntry *entry = NULL;
669
79.6M
    const xmlChar *ret;
670
79.6M
    unsigned hashValue, newSize;
671
79.6M
    size_t maxLen, len, plen, klen;
672
79.6M
    int found = 0;
673
674
79.6M
    if ((dict == NULL) || (name == NULL))
675
0
  return(NULL);
676
677
79.6M
    maxLen = (maybeLen < 0) ? SIZE_MAX : (size_t) maybeLen;
678
679
79.6M
    if (prefix == NULL) {
680
79.6M
        hashValue = xmlDictHashName(dict->seed, name, maxLen, &len);
681
79.6M
        if (len > INT_MAX / 2)
682
0
            return(NULL);
683
79.6M
        klen = len;
684
79.6M
    } else {
685
587
        hashValue = xmlDictHashQName(dict->seed, prefix, name, &plen, &len);
686
655
        if ((len > INT_MAX / 2) || (plen >= INT_MAX / 2 - len))
687
0
            return(NULL);
688
587
        klen = plen + 1 + len;
689
587
    }
690
691
79.6M
    if ((dict->limit > 0) && (klen >= dict->limit))
692
0
        return(NULL);
693
694
    /*
695
     * Check for an existing entry
696
     */
697
79.6M
    if (dict->size == 0) {
698
218k
        newSize = MIN_HASH_SIZE;
699
79.4M
    } else {
700
79.4M
        entry = xmlDictFindEntry(dict, prefix, name, klen, hashValue, &found);
701
79.4M
        if (found)
702
70.5M
            return(entry);
703
704
8.89M
        if (dict->nbElems + 1 > dict->size / MAX_FILL_DENOM * MAX_FILL_NUM) {
705
516k
            if (dict->size >= MAX_HASH_SIZE)
706
0
                return(NULL);
707
516k
            newSize = dict->size * 2;
708
8.37M
        } else {
709
8.37M
            newSize = 0;
710
8.37M
        }
711
8.89M
    }
712
713
9.11M
    if ((dict->subdict != NULL) && (dict->subdict->size > 0)) {
714
0
        xmlDictEntry *subEntry;
715
0
        unsigned subHashValue;
716
717
0
        if (prefix == NULL)
718
0
            subHashValue = xmlDictHashName(dict->subdict->seed, name, len,
719
0
                                           &len);
720
0
        else
721
0
            subHashValue = xmlDictHashQName(dict->subdict->seed, prefix, name,
722
0
                                            &plen, &len);
723
0
        subEntry = xmlDictFindEntry(dict->subdict, prefix, name, klen,
724
0
                                    subHashValue, &found);
725
0
        if (found)
726
0
            return(subEntry);
727
0
    }
728
729
9.11M
    if (!update)
730
0
        return(NULL);
731
732
    /*
733
     * Grow the hash table if needed
734
     */
735
9.11M
    if (newSize > 0) {
736
735k
        unsigned mask, displ, pos;
737
738
735k
        if (xmlDictGrow(dict, newSize) != 0)
739
0
            return(NULL);
740
741
        /*
742
         * Find new entry
743
         */
744
735k
        mask = dict->size - 1;
745
735k
        displ = 0;
746
735k
        pos = hashValue & mask;
747
735k
        entry = &dict->table[pos];
748
749
1.03M
        while ((entry->hashValue != 0) &&
750
410k
               ((pos - entry->hashValue) & mask) >= displ) {
751
301k
            displ++;
752
301k
            pos++;
753
301k
            entry++;
754
301k
            if ((pos & mask) == 0)
755
10.9k
                entry = dict->table;
756
301k
        }
757
735k
    }
758
759
9.11M
    if (prefix == NULL)
760
9.11M
        ret = xmlDictAddString(dict, name, len);
761
1.70k
    else
762
1.70k
        ret = xmlDictAddQString(dict, prefix, plen, name, len);
763
9.11M
    if (ret == NULL)
764
0
        return(NULL);
765
766
    /*
767
     * Shift the remainder of the probe sequence to the right
768
     */
769
9.11M
    if (entry->hashValue != 0) {
770
2.89M
        const xmlDictEntry *end = &dict->table[dict->size];
771
2.89M
        const xmlDictEntry *cur = entry;
772
773
13.8M
        do {
774
13.8M
            cur++;
775
13.8M
            if (cur >= end)
776
246k
                cur = dict->table;
777
13.8M
        } while (cur->hashValue != 0);
778
779
2.89M
        if (cur < entry) {
780
            /*
781
             * If we traversed the end of the buffer, handle the part
782
             * at the start of the buffer.
783
             */
784
246k
            memmove(&dict->table[1], dict->table,
785
246k
                    (char *) cur - (char *) dict->table);
786
246k
            cur = end - 1;
787
246k
            dict->table[0] = *cur;
788
246k
        }
789
790
2.89M
        memmove(&entry[1], entry, (char *) cur - (char *) entry);
791
2.89M
    }
792
793
    /*
794
     * Populate entry
795
     */
796
9.11M
    entry->hashValue = hashValue;
797
9.11M
    entry->name = ret;
798
799
9.11M
    dict->nbElems++;
800
801
9.11M
    return(entry);
802
9.11M
}
803
804
/**
805
 * Lookup a string and add it to the dictionary if it wasn't found.
806
 *
807
 * @param dict  dictionary
808
 * @param name  string key
809
 * @param len  length of the key, if -1 it is recomputed
810
 * @returns the interned copy of the string or NULL if a memory allocation
811
 * failed.
812
 */
813
const xmlChar *
814
2.09M
xmlDictLookup(xmlDict *dict, const xmlChar *name, int len) {
815
2.09M
    const xmlDictEntry *entry;
816
817
2.09M
    entry = xmlDictLookupInternal(dict, NULL, name, len, 1);
818
2.09M
    if (entry == NULL)
819
0
        return(NULL);
820
2.09M
    return(entry->name);
821
2.09M
}
822
823
/**
824
 * Lookup a dictionary entry and add the string to the dictionary if
825
 * it wasn't found.
826
 *
827
 * @param dict  dictionary
828
 * @param name  string key
829
 * @param len  length of the key, if -1 it is recomputed
830
 * @returns the dictionary entry.
831
 */
832
xmlHashedString
833
77.5M
xmlDictLookupHashed(xmlDict *dict, const xmlChar *name, int len) {
834
77.5M
    const xmlDictEntry *entry;
835
77.5M
    xmlHashedString ret;
836
837
77.5M
    entry = xmlDictLookupInternal(dict, NULL, name, len, 1);
838
839
77.5M
    if (entry == NULL) {
840
0
        ret.name = NULL;
841
0
        ret.hashValue = 0;
842
77.5M
    } else {
843
77.5M
        ret = *entry;
844
77.5M
    }
845
846
77.5M
    return(ret);
847
77.5M
}
848
849
/**
850
 * Check if a string exists in the dictionary.
851
 *
852
 * @param dict  the dictionary
853
 * @param name  the name of the userdata
854
 * @param len  the length of the name, if -1 it is recomputed
855
 * @returns the internal copy of the name or NULL if not found.
856
 */
857
const xmlChar *
858
0
xmlDictExists(xmlDict *dict, const xmlChar *name, int len) {
859
0
    const xmlDictEntry *entry;
860
861
0
    entry = xmlDictLookupInternal(dict, NULL, name, len, 0);
862
0
    if (entry == NULL)
863
0
        return(NULL);
864
0
    return(entry->name);
865
0
}
866
867
/**
868
 * Lookup the QName `prefix:name` and add it to the dictionary if
869
 * it wasn't found.
870
 *
871
 * @param dict  the dictionary
872
 * @param prefix  the prefix
873
 * @param name  the name
874
 * @returns the interned copy of the string or NULL if a memory allocation
875
 * failed.
876
 */
877
const xmlChar *
878
655
xmlDictQLookup(xmlDict *dict, const xmlChar *prefix, const xmlChar *name) {
879
655
    const xmlDictEntry *entry;
880
881
655
    entry = xmlDictLookupInternal(dict, prefix, name, -1, 1);
882
655
    if (entry == NULL)
883
0
        return(NULL);
884
655
    return(entry->name);
885
655
}
886
887
/*
888
 * Pseudo-random generator
889
 */
890
891
#ifdef _WIN32
892
  #define WIN32_LEAN_AND_MEAN
893
  #include <windows.h>
894
  #include <bcrypt.h>
895
#else
896
  #if HAVE_DECL_GETENTROPY
897
    /* POSIX 2024 */
898
    #include <unistd.h>
899
    /* Older platforms */
900
    #include <sys/random.h>
901
  #endif
902
  #include <time.h>
903
#endif
904
905
static xmlMutex xmlRngMutex;
906
907
static unsigned globalRngState[2];
908
909
/*
910
 *
911
 * Initialize the PRNG.
912
 */
913
ATTRIBUTE_NO_SANITIZE_INTEGER
914
void
915
19
xmlInitRandom(void) {
916
19
    xmlInitMutex(&xmlRngMutex);
917
918
19
    {
919
#ifdef _WIN32
920
        NTSTATUS status;
921
922
        /*
923
         * You can find many (recent as of 2025) discussions how
924
         * to get a pseudo-random seed on Windows in projects like
925
         * Golang, Rust, Chromium and Firefox.
926
         *
927
         * TODO: Support ProcessPrng available since Windows 10.
928
         */
929
        status = BCryptGenRandom(NULL, (unsigned char *) globalRngState,
930
                                 sizeof(globalRngState),
931
                                 BCRYPT_USE_SYSTEM_PREFERRED_RNG);
932
        if (!BCRYPT_SUCCESS(status))
933
            xmlAbort("libxml2: BCryptGenRandom failed with error code %lu\n",
934
                     GetLastError());
935
#else
936
19
        int var;
937
938
19
#if HAVE_DECL_GETENTROPY
939
19
        while (1) {
940
19
            if (getentropy(globalRngState, sizeof(globalRngState)) == 0)
941
19
                return;
942
943
            /*
944
             * This most likely means that libxml2 was compiled on
945
             * a system supporting certain system calls and is running
946
             * on a system that doesn't support these calls, as can
947
             * be the case on Linux.
948
             */
949
0
            if (errno == ENOSYS)
950
0
                break;
951
952
            /*
953
             * We really don't want to fallback to the unsafe PRNG
954
             * for possibly accidental reasons, so we abort on any
955
             * unknown error.
956
             */
957
0
            if (errno != EINTR)
958
0
                xmlAbort("libxml2: getentropy failed with error code %d\n",
959
0
                         errno);
960
0
        }
961
0
#endif
962
963
        /*
964
         * TODO: Fallback to /dev/urandom for older POSIX systems.
965
         */
966
0
        globalRngState[0] =
967
0
                (unsigned) time(NULL) ^
968
0
                HASH_ROL((unsigned) ((size_t) &xmlInitRandom & 0xFFFFFFFF), 8);
969
0
        globalRngState[1] =
970
0
                HASH_ROL((unsigned) ((size_t) &xmlRngMutex & 0xFFFFFFFF), 16) ^
971
0
                HASH_ROL((unsigned) ((size_t) &var & 0xFFFFFFFF), 24);
972
0
#endif
973
0
    }
974
0
}
975
976
/*
977
 *
978
 * Clean up PRNG globals.
979
 */
980
void
981
0
xmlCleanupRandom(void) {
982
0
    xmlCleanupMutex(&xmlRngMutex);
983
0
}
984
985
ATTRIBUTE_NO_SANITIZE_INTEGER
986
static unsigned
987
241k
xoroshiro64ss(unsigned *s) {
988
241k
    unsigned s0 = s[0];
989
241k
    unsigned s1 = s[1];
990
241k
    unsigned result = HASH_ROL(s0 * 0x9E3779BB, 5) * 5;
991
992
241k
    s1 ^= s0;
993
241k
    s[0] = HASH_ROL(s0, 26) ^ s1 ^ (s1 << 9);
994
241k
    s[1] = HASH_ROL(s1, 13);
995
996
241k
    return(result & 0xFFFFFFFF);
997
241k
}
998
999
/*
1000
 *
1001
 * Generate a pseudo-random value using the global PRNG.
1002
 *
1003
 * @returns a random value.
1004
 */
1005
unsigned
1006
13.7k
xmlGlobalRandom(void) {
1007
13.7k
    unsigned ret;
1008
1009
13.7k
    xmlMutexLock(&xmlRngMutex);
1010
13.7k
    ret = xoroshiro64ss(globalRngState);
1011
13.7k
    xmlMutexUnlock(&xmlRngMutex);
1012
1013
13.7k
    return(ret);
1014
13.7k
}
1015
1016
/*
1017
 *
1018
 * Generate a pseudo-random value using the thread-local PRNG.
1019
 *
1020
 * @returns a random value.
1021
 */
1022
unsigned
1023
227k
xmlRandom(void) {
1024
227k
    return(xoroshiro64ss(xmlGetLocalRngState()));
1025
227k
}
1026