2
* diff-delta.c: generate a delta between two buffers
4
* This code was greatly inspired by parts of LibXDiff from Davide Libenzi
5
* http://www.xmailserver.org/xdiff-lib.html
7
* Rewritten for GIT by Nicolas Pitre <nico@fluxnic.net>, (C) 2005-2007
8
* Adapted for Bazaar by John Arbash Meinel <john@arbash-meinel.com> (C) 2009
10
* This program is free software; you can redistribute it and/or modify
11
* it under the terms of the GNU General Public License as published by
12
* the Free Software Foundation; either version 2 of the License, or
13
* (at your option) any later version.
15
* NB: The version in GIT is 'version 2 of the Licence only', however Nicolas
16
* has granted permission for use under 'version 2 or later' in private email
17
* to Robert Collins and Karl Fogel on the 6th April 2009.
27
/* maximum hash entry list for the same hash bucket */
30
#define RABIN_SHIFT 23
31
#define RABIN_WINDOW 16
33
/* The hash map is sized to put 4 entries per bucket, this gives us ~even room
34
* for more data. Tweaking this number above 4 doesn't seem to help much,
39
static const unsigned int T[256] = {
40
0x00000000, 0xab59b4d1, 0x56b369a2, 0xfdeadd73, 0x063f6795, 0xad66d344,
41
0x508c0e37, 0xfbd5bae6, 0x0c7ecf2a, 0xa7277bfb, 0x5acda688, 0xf1941259,
42
0x0a41a8bf, 0xa1181c6e, 0x5cf2c11d, 0xf7ab75cc, 0x18fd9e54, 0xb3a42a85,
43
0x4e4ef7f6, 0xe5174327, 0x1ec2f9c1, 0xb59b4d10, 0x48719063, 0xe32824b2,
44
0x1483517e, 0xbfdae5af, 0x423038dc, 0xe9698c0d, 0x12bc36eb, 0xb9e5823a,
45
0x440f5f49, 0xef56eb98, 0x31fb3ca8, 0x9aa28879, 0x6748550a, 0xcc11e1db,
46
0x37c45b3d, 0x9c9defec, 0x6177329f, 0xca2e864e, 0x3d85f382, 0x96dc4753,
47
0x6b369a20, 0xc06f2ef1, 0x3bba9417, 0x90e320c6, 0x6d09fdb5, 0xc6504964,
48
0x2906a2fc, 0x825f162d, 0x7fb5cb5e, 0xd4ec7f8f, 0x2f39c569, 0x846071b8,
49
0x798aaccb, 0xd2d3181a, 0x25786dd6, 0x8e21d907, 0x73cb0474, 0xd892b0a5,
50
0x23470a43, 0x881ebe92, 0x75f463e1, 0xdeadd730, 0x63f67950, 0xc8afcd81,
51
0x354510f2, 0x9e1ca423, 0x65c91ec5, 0xce90aa14, 0x337a7767, 0x9823c3b6,
52
0x6f88b67a, 0xc4d102ab, 0x393bdfd8, 0x92626b09, 0x69b7d1ef, 0xc2ee653e,
53
0x3f04b84d, 0x945d0c9c, 0x7b0be704, 0xd05253d5, 0x2db88ea6, 0x86e13a77,
54
0x7d348091, 0xd66d3440, 0x2b87e933, 0x80de5de2, 0x7775282e, 0xdc2c9cff,
55
0x21c6418c, 0x8a9ff55d, 0x714a4fbb, 0xda13fb6a, 0x27f92619, 0x8ca092c8,
56
0x520d45f8, 0xf954f129, 0x04be2c5a, 0xafe7988b, 0x5432226d, 0xff6b96bc,
57
0x02814bcf, 0xa9d8ff1e, 0x5e738ad2, 0xf52a3e03, 0x08c0e370, 0xa39957a1,
58
0x584ced47, 0xf3155996, 0x0eff84e5, 0xa5a63034, 0x4af0dbac, 0xe1a96f7d,
59
0x1c43b20e, 0xb71a06df, 0x4ccfbc39, 0xe79608e8, 0x1a7cd59b, 0xb125614a,
60
0x468e1486, 0xedd7a057, 0x103d7d24, 0xbb64c9f5, 0x40b17313, 0xebe8c7c2,
61
0x16021ab1, 0xbd5bae60, 0x6cb54671, 0xc7ecf2a0, 0x3a062fd3, 0x915f9b02,
62
0x6a8a21e4, 0xc1d39535, 0x3c394846, 0x9760fc97, 0x60cb895b, 0xcb923d8a,
63
0x3678e0f9, 0x9d215428, 0x66f4eece, 0xcdad5a1f, 0x3047876c, 0x9b1e33bd,
64
0x7448d825, 0xdf116cf4, 0x22fbb187, 0x89a20556, 0x7277bfb0, 0xd92e0b61,
65
0x24c4d612, 0x8f9d62c3, 0x7836170f, 0xd36fa3de, 0x2e857ead, 0x85dcca7c,
66
0x7e09709a, 0xd550c44b, 0x28ba1938, 0x83e3ade9, 0x5d4e7ad9, 0xf617ce08,
67
0x0bfd137b, 0xa0a4a7aa, 0x5b711d4c, 0xf028a99d, 0x0dc274ee, 0xa69bc03f,
68
0x5130b5f3, 0xfa690122, 0x0783dc51, 0xacda6880, 0x570fd266, 0xfc5666b7,
69
0x01bcbbc4, 0xaae50f15, 0x45b3e48d, 0xeeea505c, 0x13008d2f, 0xb85939fe,
70
0x438c8318, 0xe8d537c9, 0x153feaba, 0xbe665e6b, 0x49cd2ba7, 0xe2949f76,
71
0x1f7e4205, 0xb427f6d4, 0x4ff24c32, 0xe4abf8e3, 0x19412590, 0xb2189141,
72
0x0f433f21, 0xa41a8bf0, 0x59f05683, 0xf2a9e252, 0x097c58b4, 0xa225ec65,
73
0x5fcf3116, 0xf49685c7, 0x033df00b, 0xa86444da, 0x558e99a9, 0xfed72d78,
74
0x0502979e, 0xae5b234f, 0x53b1fe3c, 0xf8e84aed, 0x17bea175, 0xbce715a4,
75
0x410dc8d7, 0xea547c06, 0x1181c6e0, 0xbad87231, 0x4732af42, 0xec6b1b93,
76
0x1bc06e5f, 0xb099da8e, 0x4d7307fd, 0xe62ab32c, 0x1dff09ca, 0xb6a6bd1b,
77
0x4b4c6068, 0xe015d4b9, 0x3eb80389, 0x95e1b758, 0x680b6a2b, 0xc352defa,
78
0x3887641c, 0x93ded0cd, 0x6e340dbe, 0xc56db96f, 0x32c6cca3, 0x999f7872,
79
0x6475a501, 0xcf2c11d0, 0x34f9ab36, 0x9fa01fe7, 0x624ac294, 0xc9137645,
80
0x26459ddd, 0x8d1c290c, 0x70f6f47f, 0xdbaf40ae, 0x207afa48, 0x8b234e99,
81
0x76c993ea, 0xdd90273b, 0x2a3b52f7, 0x8162e626, 0x7c883b55, 0xd7d18f84,
82
0x2c043562, 0x875d81b3, 0x7ab75cc0, 0xd1eee811
85
static const unsigned int U[256] = {
86
0x00000000, 0x7eb5200d, 0x5633f4cb, 0x2886d4c6, 0x073e5d47, 0x798b7d4a,
87
0x510da98c, 0x2fb88981, 0x0e7cba8e, 0x70c99a83, 0x584f4e45, 0x26fa6e48,
88
0x0942e7c9, 0x77f7c7c4, 0x5f711302, 0x21c4330f, 0x1cf9751c, 0x624c5511,
89
0x4aca81d7, 0x347fa1da, 0x1bc7285b, 0x65720856, 0x4df4dc90, 0x3341fc9d,
90
0x1285cf92, 0x6c30ef9f, 0x44b63b59, 0x3a031b54, 0x15bb92d5, 0x6b0eb2d8,
91
0x4388661e, 0x3d3d4613, 0x39f2ea38, 0x4747ca35, 0x6fc11ef3, 0x11743efe,
92
0x3eccb77f, 0x40799772, 0x68ff43b4, 0x164a63b9, 0x378e50b6, 0x493b70bb,
93
0x61bda47d, 0x1f088470, 0x30b00df1, 0x4e052dfc, 0x6683f93a, 0x1836d937,
94
0x250b9f24, 0x5bbebf29, 0x73386bef, 0x0d8d4be2, 0x2235c263, 0x5c80e26e,
95
0x740636a8, 0x0ab316a5, 0x2b7725aa, 0x55c205a7, 0x7d44d161, 0x03f1f16c,
96
0x2c4978ed, 0x52fc58e0, 0x7a7a8c26, 0x04cfac2b, 0x73e5d470, 0x0d50f47d,
97
0x25d620bb, 0x5b6300b6, 0x74db8937, 0x0a6ea93a, 0x22e87dfc, 0x5c5d5df1,
98
0x7d996efe, 0x032c4ef3, 0x2baa9a35, 0x551fba38, 0x7aa733b9, 0x041213b4,
99
0x2c94c772, 0x5221e77f, 0x6f1ca16c, 0x11a98161, 0x392f55a7, 0x479a75aa,
100
0x6822fc2b, 0x1697dc26, 0x3e1108e0, 0x40a428ed, 0x61601be2, 0x1fd53bef,
101
0x3753ef29, 0x49e6cf24, 0x665e46a5, 0x18eb66a8, 0x306db26e, 0x4ed89263,
102
0x4a173e48, 0x34a21e45, 0x1c24ca83, 0x6291ea8e, 0x4d29630f, 0x339c4302,
103
0x1b1a97c4, 0x65afb7c9, 0x446b84c6, 0x3adea4cb, 0x1258700d, 0x6ced5000,
104
0x4355d981, 0x3de0f98c, 0x15662d4a, 0x6bd30d47, 0x56ee4b54, 0x285b6b59,
105
0x00ddbf9f, 0x7e689f92, 0x51d01613, 0x2f65361e, 0x07e3e2d8, 0x7956c2d5,
106
0x5892f1da, 0x2627d1d7, 0x0ea10511, 0x7014251c, 0x5facac9d, 0x21198c90,
107
0x099f5856, 0x772a785b, 0x4c921c31, 0x32273c3c, 0x1aa1e8fa, 0x6414c8f7,
108
0x4bac4176, 0x3519617b, 0x1d9fb5bd, 0x632a95b0, 0x42eea6bf, 0x3c5b86b2,
109
0x14dd5274, 0x6a687279, 0x45d0fbf8, 0x3b65dbf5, 0x13e30f33, 0x6d562f3e,
110
0x506b692d, 0x2ede4920, 0x06589de6, 0x78edbdeb, 0x5755346a, 0x29e01467,
111
0x0166c0a1, 0x7fd3e0ac, 0x5e17d3a3, 0x20a2f3ae, 0x08242768, 0x76910765,
112
0x59298ee4, 0x279caee9, 0x0f1a7a2f, 0x71af5a22, 0x7560f609, 0x0bd5d604,
113
0x235302c2, 0x5de622cf, 0x725eab4e, 0x0ceb8b43, 0x246d5f85, 0x5ad87f88,
114
0x7b1c4c87, 0x05a96c8a, 0x2d2fb84c, 0x539a9841, 0x7c2211c0, 0x029731cd,
115
0x2a11e50b, 0x54a4c506, 0x69998315, 0x172ca318, 0x3faa77de, 0x411f57d3,
116
0x6ea7de52, 0x1012fe5f, 0x38942a99, 0x46210a94, 0x67e5399b, 0x19501996,
117
0x31d6cd50, 0x4f63ed5d, 0x60db64dc, 0x1e6e44d1, 0x36e89017, 0x485db01a,
118
0x3f77c841, 0x41c2e84c, 0x69443c8a, 0x17f11c87, 0x38499506, 0x46fcb50b,
119
0x6e7a61cd, 0x10cf41c0, 0x310b72cf, 0x4fbe52c2, 0x67388604, 0x198da609,
120
0x36352f88, 0x48800f85, 0x6006db43, 0x1eb3fb4e, 0x238ebd5d, 0x5d3b9d50,
121
0x75bd4996, 0x0b08699b, 0x24b0e01a, 0x5a05c017, 0x728314d1, 0x0c3634dc,
122
0x2df207d3, 0x534727de, 0x7bc1f318, 0x0574d315, 0x2acc5a94, 0x54797a99,
123
0x7cffae5f, 0x024a8e52, 0x06852279, 0x78300274, 0x50b6d6b2, 0x2e03f6bf,
124
0x01bb7f3e, 0x7f0e5f33, 0x57888bf5, 0x293dabf8, 0x08f998f7, 0x764cb8fa,
125
0x5eca6c3c, 0x207f4c31, 0x0fc7c5b0, 0x7172e5bd, 0x59f4317b, 0x27411176,
126
0x1a7c5765, 0x64c97768, 0x4c4fa3ae, 0x32fa83a3, 0x1d420a22, 0x63f72a2f,
127
0x4b71fee9, 0x35c4dee4, 0x1400edeb, 0x6ab5cde6, 0x42331920, 0x3c86392d,
128
0x133eb0ac, 0x6d8b90a1, 0x450d4467, 0x3bb8646a
132
const unsigned char *ptr;
133
const struct source_info *src;
137
struct index_entry_linked_list {
138
struct index_entry *p_entry;
139
struct index_entry_linked_list *next;
142
struct unpacked_index_entry {
143
struct index_entry entry;
144
struct unpacked_index_entry *next;
148
unsigned long memsize; /* Total bytes pointed to by this index */
149
const struct source_info *last_src; /* Information about the referenced source */
150
unsigned int hash_mask; /* val & hash_mask gives the hash index for a given
152
unsigned int num_entries; /* The total number of entries in this index */
153
struct index_entry *last_entry; /* Pointer to the last valid entry */
154
struct index_entry *hash[];
158
limit_hash_buckets(struct unpacked_index_entry **hash,
159
unsigned int *hash_count, unsigned int hsize,
160
unsigned int entries)
162
struct unpacked_index_entry *entry;
165
* Determine a limit on the number of entries in the same hash
166
* bucket. This guards us against pathological data sets causing
167
* really bad hash distribution with most entries in the same hash
168
* bucket that would bring us to O(m*n) computing costs (m and n
169
* corresponding to reference and target buffer sizes).
171
* Make sure none of the hash buckets has more entries than
172
* we're willing to test. Otherwise we cull the entry list
173
* uniformly to still preserve a good repartition across
174
* the reference buffer.
176
for (i = 0; i < hsize; i++) {
179
if (hash_count[i] <= HASH_LIMIT)
182
/* We leave exactly HASH_LIMIT entries in the bucket */
183
entries -= hash_count[i] - HASH_LIMIT;
189
* Assume that this loop is gone through exactly
190
* HASH_LIMIT times and is entered and left with
191
* acc==0. So the first statement in the loop
192
* contributes (hash_count[i]-HASH_LIMIT)*HASH_LIMIT
193
* to the accumulator, and the inner loop consequently
194
* is run (hash_count[i]-HASH_LIMIT) times, removing
195
* one element from the list each time. Since acc
196
* balances out to 0 at the final run, the inner loop
197
* body can't be left with entry==NULL. So we indeed
198
* encounter entry==NULL in the outer loop only.
201
acc += hash_count[i] - HASH_LIMIT;
203
struct unpacked_index_entry *keep = entry;
208
keep->next = entry->next;
216
static struct delta_index *
217
pack_delta_index(struct unpacked_index_entry **hash, unsigned int hsize,
218
unsigned int num_entries, struct delta_index *old_index)
220
unsigned int i, j, hmask, memsize, fit_in_old, copied_count;
221
struct unpacked_index_entry *entry;
222
struct delta_index *index;
223
struct index_entry *packed_entry, **packed_hash, *old_entry;
224
struct index_entry null_entry = {0};
230
// fprintf(stderr, "Packing %d entries into %d for total of %d entries"
232
// num_entries - old_index->num_entries,
233
// old_index->num_entries, num_entries,
234
// old_index->hash_mask, hmask);
236
// fprintf(stderr, "Packing %d entries into a new index\n",
239
/* First, see if we can squeeze the new items into the existing structure.
243
if (old_index && old_index->hash_mask == hmask) {
245
for (i = 0; i < hsize; ++i) {
247
for (entry = hash[i]; entry; entry = entry->next) {
248
if (packed_entry == NULL) {
249
/* Find the last open spot */
250
packed_entry = old_index->hash[i + 1];
252
while (packed_entry >= old_index->hash[i]
253
&& packed_entry->ptr == NULL) {
258
if (packed_entry >= old_index->hash[i+1]
259
|| packed_entry->ptr != NULL) {
260
/* There are no free spots here :( */
264
/* We found an empty spot to put this entry
265
* Copy it over, and remove it from the linked list, just in
266
* case we end up running out of room later.
268
*packed_entry++ = entry->entry;
269
assert(entry == hash[i]);
270
hash[i] = entry->next;
272
old_index->num_entries++;
281
// fprintf(stderr, "Fit all %d entries into old index\n",
284
* No need to allocate a new buffer, but return old_index ptr so
285
* callers can distinguish this from an OOM failure.
289
// fprintf(stderr, "Fit only %d entries into old index,"
290
// " reallocating\n", copied_count);
294
* Now create the packed index in array form
295
* rather than linked lists.
296
* Leave a 2-entry gap for inserting more entries between the groups
298
memsize = sizeof(*index)
299
+ sizeof(*packed_hash) * (hsize+1)
300
+ sizeof(*packed_entry) * (num_entries + hsize * EXTRA_NULLS);
301
mem = malloc(memsize);
307
index->memsize = memsize;
308
index->hash_mask = hmask;
309
index->num_entries = num_entries;
311
if (hmask < old_index->hash_mask) {
312
fprintf(stderr, "hash mask was shrunk %x => %x\n",
313
old_index->hash_mask, hmask);
315
assert(hmask >= old_index->hash_mask);
320
mem = packed_hash + (hsize+1);
323
for (i = 0; i < hsize; i++) {
325
* Coalesce all entries belonging to one linked list
326
* into consecutive array entries.
328
packed_hash[i] = packed_entry;
329
/* Old comes earlier as a source, so it always comes first in a given
333
/* Could we optimize this to use memcpy when hmask ==
334
* old_index->hash_mask? Would it make any real difference?
336
j = i & old_index->hash_mask;
337
for (old_entry = old_index->hash[j];
338
old_entry < old_index->hash[j + 1] && old_entry->ptr != NULL;
340
if ((old_entry->val & hmask) == i) {
341
*packed_entry++ = *old_entry;
345
for (entry = hash[i]; entry; entry = entry->next) {
346
*packed_entry++ = entry->entry;
348
/* TODO: At this point packed_entry - packed_hash[i] is the number of
349
* records that we have inserted into this hash bucket.
350
* We should *really* consider doing some limiting along the
351
* lines of limit_hash_buckets() to avoid pathological behavior.
353
/* Now add extra 'NULL' entries that we can use for future expansion. */
354
for (j = 0; j < EXTRA_NULLS; ++j ) {
355
*packed_entry++ = null_entry;
359
/* Sentinel value to indicate the length of the last hash bucket */
360
packed_hash[hsize] = packed_entry;
362
if (packed_entry - (struct index_entry *)mem
363
!= num_entries + hsize*EXTRA_NULLS) {
364
fprintf(stderr, "We expected %d entries, but created %d\n",
365
num_entries + hsize*EXTRA_NULLS,
366
(int)(packed_entry - (struct index_entry*)mem));
368
assert(packed_entry - (struct index_entry *)mem
369
== num_entries + hsize*EXTRA_NULLS);
370
index->last_entry = (packed_entry - 1);
376
create_delta_index(const struct source_info *src,
377
struct delta_index *old,
378
struct delta_index **fresh,
379
int max_bytes_to_index)
381
unsigned int i, hsize, hmask, num_entries, prev_val, *hash_count;
382
unsigned int total_num_entries, stride, max_entries;
383
const unsigned char *data, *buffer;
384
struct delta_index *index;
385
struct unpacked_index_entry *entry, **hash;
387
unsigned long memsize;
389
if (!src->buf || !src->size)
390
return DELTA_SOURCE_EMPTY;
393
/* Determine index hash size. Note that indexing skips the
394
first byte so we subtract 1 to get the edge cases right.
396
stride = RABIN_WINDOW;
397
num_entries = (src->size - 1) / RABIN_WINDOW;
398
if (max_bytes_to_index > 0) {
399
max_entries = (unsigned int) (max_bytes_to_index / RABIN_WINDOW);
400
if (num_entries > max_entries) {
401
/* Limit the max number of matching entries. This reduces the 'best'
402
* possible match, but means we don't consume all of ram.
404
num_entries = max_entries;
405
stride = (src->size - 1) / num_entries;
409
total_num_entries = num_entries + old->num_entries;
411
total_num_entries = num_entries;
412
hsize = total_num_entries / 4;
413
for (i = 4; (1u << i) < hsize && i < 31; i++);
416
if (old && old->hash_mask > hmask) {
417
hmask = old->hash_mask;
421
/* allocate lookup index */
422
memsize = sizeof(*hash) * hsize +
423
sizeof(*entry) * total_num_entries;
424
mem = malloc(memsize);
426
return DELTA_OUT_OF_MEMORY;
431
memset(hash, 0, hsize * sizeof(*hash));
433
/* allocate an array to count hash num_entries */
434
hash_count = calloc(hsize, sizeof(*hash_count));
437
return DELTA_OUT_OF_MEMORY;
440
/* then populate the index for the new data */
442
for (data = buffer + num_entries * stride - RABIN_WINDOW;
445
unsigned int val = 0;
446
for (i = 1; i <= RABIN_WINDOW; i++)
447
val = ((val << 8) | data[i]) ^ T[val >> RABIN_SHIFT];
448
if (val == prev_val) {
449
/* keep the lowest of consecutive identical blocks */
450
entry[-1].entry.ptr = data + RABIN_WINDOW;
456
entry->entry.ptr = data + RABIN_WINDOW;
457
entry->entry.val = val;
458
entry->entry.src = src;
459
entry->next = hash[i];
464
/* TODO: It would be nice to limit_hash_buckets at a better time. */
465
total_num_entries = limit_hash_buckets(hash, hash_count, hsize,
468
index = pack_delta_index(hash, hsize, total_num_entries, old);
470
/* pack_delta_index only returns NULL on malloc failure */
472
return DELTA_OUT_OF_MEMORY;
474
index->last_src = src;
479
/* Take some entries, and put them into a custom hash.
480
* @param entries A list of entries, sorted by position in file
481
* @param num_entries Length of entries
482
* @param out_hsize The maximum size of the hash, the final size will be
485
struct index_entry_linked_list **
486
_put_entries_into_hash(struct index_entry *entries, unsigned int num_entries,
489
unsigned int hash_offset, hmask, memsize;
490
struct index_entry *entry;
491
struct index_entry_linked_list *out_entry, **hash;
496
memsize = sizeof(*hash) * hsize +
497
sizeof(*out_entry) * num_entries;
498
mem = malloc(memsize);
505
memset(hash, 0, sizeof(*hash)*(hsize+1));
507
/* We know that entries are in the order we want in the output, but they
508
* aren't "grouped" by hash bucket yet.
510
for (entry = entries + num_entries - 1; entry >= entries; --entry) {
511
hash_offset = entry->val & hmask;
512
out_entry->p_entry = entry;
513
out_entry->next = hash[hash_offset];
514
/* TODO: Remove entries that have identical vals, or at least filter
515
* the map a little bit.
516
* if (hash[i] != NULL) {
519
hash[hash_offset] = out_entry;
527
create_index_from_old_and_new_entries(const struct delta_index *old_index,
528
struct index_entry *entries,
529
unsigned int num_entries)
531
unsigned int i, j, hsize, hmask, total_num_entries;
532
struct delta_index *index;
533
struct index_entry *entry, *packed_entry, **packed_hash;
534
struct index_entry null_entry = {0};
536
unsigned long memsize;
537
struct index_entry_linked_list *unpacked_entry, **mini_hash;
539
/* Determine index hash size. Note that indexing skips the
540
first byte to allow for optimizing the Rabin's polynomial
541
initialization in create_delta(). */
542
total_num_entries = num_entries + old_index->num_entries;
543
hsize = total_num_entries / 4;
544
for (i = 4; (1u << i) < hsize && i < 31; i++);
546
if (hsize < old_index->hash_mask) {
547
/* For some reason, there was a code path that would actually *shrink*
548
* the hash size. This screws with some later code, and in general, I
549
* think it better to make the hash bigger, rather than smaller. So
550
* we'll just force the size here.
551
* Possibly done by create_delta_index running into a
552
* limit_hash_buckets call, that ended up transitioning across a
553
* power-of-2. The cause isn't 100% clear, though.
555
hsize = old_index->hash_mask + 1;
558
// fprintf(stderr, "resizing index to insert %d entries into array"
559
// " with %d entries: %x => %x\n",
560
// num_entries, old_index->num_entries, old_index->hash_mask, hmask);
562
memsize = sizeof(*index)
563
+ sizeof(*packed_hash) * (hsize+1)
564
+ sizeof(*packed_entry) * (total_num_entries + hsize*EXTRA_NULLS);
565
mem = malloc(memsize);
570
index->memsize = memsize;
571
index->hash_mask = hmask;
572
index->num_entries = total_num_entries;
573
index->last_src = old_index->last_src;
577
mem = packed_hash + (hsize+1);
580
mini_hash = _put_entries_into_hash(entries, num_entries, hsize);
581
if (mini_hash == NULL) {
585
for (i = 0; i < hsize; i++) {
587
* Coalesce all entries belonging in one hash bucket
588
* into consecutive array entries.
589
* The entries in old_index all come before 'entries'.
591
packed_hash[i] = packed_entry;
592
/* Copy any of the old entries across */
593
/* Would we rather use memcpy? */
594
if (hmask == old_index->hash_mask) {
595
for (entry = old_index->hash[i];
596
entry < old_index->hash[i+1] && entry->ptr != NULL;
598
assert((entry->val & hmask) == i);
599
*packed_entry++ = *entry;
602
/* If we resized the index from this action, all of the old values
603
* will be found in the previous location, but they will end up
604
* spread across the new locations.
606
j = i & old_index->hash_mask;
607
for (entry = old_index->hash[j];
608
entry < old_index->hash[j+1] && entry->ptr != NULL;
610
assert((entry->val & old_index->hash_mask) == j);
611
if ((entry->val & hmask) == i) {
612
/* Any entries not picked up here will be picked up on the
615
*packed_entry++ = *entry;
619
/* Now see if we need to insert any of the new entries.
620
* Note that loop ends up O(hsize*num_entries), so we expect that
621
* num_entries is always small.
622
* We also help a little bit by collapsing the entry range when the
623
* endpoints are inserted. However, an alternative would be to build a
624
* quick hash lookup for just the new entries.
625
* Testing shows that this list can easily get up to about 100
626
* entries, the tradeoff is a malloc, 1 pass over the entries, copying
627
* them into a sorted buffer, and a free() when done,
629
for (unpacked_entry = mini_hash[i];
631
unpacked_entry = unpacked_entry->next) {
632
assert((unpacked_entry->p_entry->val & hmask) == i);
633
*packed_entry++ = *(unpacked_entry->p_entry);
635
/* Now insert some extra nulls */
636
for (j = 0; j < EXTRA_NULLS; ++j) {
637
*packed_entry++ = null_entry;
642
/* Sentinel value to indicate the length of the last hash bucket */
643
packed_hash[hsize] = packed_entry;
645
if ((packed_entry - (struct index_entry *)mem)
646
!= (total_num_entries + hsize*EXTRA_NULLS)) {
647
fprintf(stderr, "We expected %d entries, but created %d\n",
648
total_num_entries + hsize*EXTRA_NULLS,
649
(int)(packed_entry - (struct index_entry*)mem));
652
assert((packed_entry - (struct index_entry *)mem)
653
== (total_num_entries + hsize * EXTRA_NULLS));
654
index->last_entry = (packed_entry - 1);
660
get_text(char buff[128], const unsigned char *ptr)
663
const unsigned char *start;
665
start = (ptr-RABIN_WINDOW-1);
667
if (cmd < 0x80) {// This is likely to be an insert instruction
668
if (cmd < RABIN_WINDOW) {
672
/* This was either a copy [should never be] or it
673
* was a longer insert so the insert start happened at 16 more
676
cmd = RABIN_WINDOW + 1;
679
cmd = 60; /* Be friendly to 80char terms */
681
/* Copy the 1 byte command, and 4 bytes after the insert */
683
memcpy(buff, start, cmd);
685
for (i = 0; i < cmd; ++i) {
686
if (buff[i] == '\n') {
688
} else if (buff[i] == '\t') {
695
create_delta_index_from_delta(const struct source_info *src,
696
struct delta_index *old_index,
697
struct delta_index **fresh)
699
unsigned int i, num_entries, max_num_entries, prev_val, num_inserted;
700
unsigned int hash_offset;
701
const unsigned char *data, *buffer, *top;
703
struct delta_index *new_index;
704
struct index_entry *entry, *entries;
707
return DELTA_INDEX_NEEDED;
708
if (!src->buf || !src->size)
709
return DELTA_SOURCE_EMPTY;
711
top = buffer + src->size;
713
/* Determine index hash size. Note that indexing skips the
714
first byte to allow for optimizing the Rabin's polynomial
715
initialization in create_delta().
716
This computes the maximum number of entries that could be held. The
717
actual number will be recomputed during processing.
720
max_num_entries = (src->size - 1) / RABIN_WINDOW;
722
if (!max_num_entries) {
727
/* allocate an array to hold whatever entries we find */
728
entries = malloc(sizeof(*entry) * max_num_entries);
729
if (!entries) /* malloc failure */
730
return DELTA_OUT_OF_MEMORY;
732
/* then populate the index for the new data */
736
/* get_delta_hdr_size doesn't mutate the content, just moves the
737
* start-of-data pointer, so it is safe to do the cast.
739
get_delta_hdr_size((unsigned char**)&data, top);
740
entry = entries; /* start at the first slot */
741
num_entries = 0; /* calculate the real number of entries */
745
/* Copy instruction, skip it */
746
if (cmd & 0x01) data++;
747
if (cmd & 0x02) data++;
748
if (cmd & 0x04) data++;
749
if (cmd & 0x08) data++;
750
if (cmd & 0x10) data++;
751
if (cmd & 0x20) data++;
752
if (cmd & 0x40) data++;
754
/* Insert instruction, we want to index these bytes */
755
if (data + cmd > top) {
756
/* Invalid insert, not enough bytes in the delta */
759
/* The create_delta code requires a match at least 4 characters
760
* (including only the last char of the RABIN_WINDOW) before it
761
* will consider it something worth copying rather than inserting.
762
* So we don't want to index anything that we know won't ever be a
765
for (; cmd > RABIN_WINDOW + 3; cmd -= RABIN_WINDOW,
766
data += RABIN_WINDOW) {
767
unsigned int val = 0;
768
for (i = 1; i <= RABIN_WINDOW; i++)
769
val = ((val << 8) | data[i]) ^ T[val >> RABIN_SHIFT];
770
if (val != prev_val) {
771
/* Only keep the first of consecutive data */
774
entry->ptr = data + RABIN_WINDOW;
778
if (num_entries > max_num_entries) {
779
/* We ran out of entry room, something is really wrong
785
/* Move the data pointer by whatever remainder is left */
789
* cmd == 0 is reserved for future encoding
790
* extensions. In the mean time we must fail when
791
* encountering them (might be data corruption).
797
/* The source_info data passed was corrupted or otherwise invalid */
799
return DELTA_SOURCE_BAD;
801
if (num_entries == 0) {
802
/** Nothing to index **/
807
old_index->last_src = src;
808
/* See if we can fill in these values into the holes in the array */
811
for (; num_entries > 0; --num_entries, ++entry) {
812
struct index_entry *next_bucket_entry, *cur_entry, *bucket_first_entry;
813
hash_offset = (entry->val & old_index->hash_mask);
814
/* The basic structure is a hash => packed_entries that fit in that
815
* hash bucket. Things are structured such that the hash-pointers are
816
* strictly ordered. So we start by pointing to the next pointer, and
817
* walk back until we stop getting NULL targets, and then go back
818
* forward. If there are no NULL targets, then we know because
819
* entry->ptr will not be NULL.
821
// The start of the next bucket, this may point past the end of the
822
// entry table if hash_offset is the last bucket.
823
next_bucket_entry = old_index->hash[hash_offset + 1];
824
// First entry in this bucket
825
bucket_first_entry = old_index->hash[hash_offset];
826
cur_entry = next_bucket_entry - 1;
827
while (cur_entry->ptr == NULL && cur_entry >= bucket_first_entry) {
830
// cur_entry now either points at the first NULL, or it points to
831
// next_bucket_entry if there were no blank spots.
833
if (cur_entry >= next_bucket_entry || cur_entry->ptr != NULL) {
834
/* There is no room for this entry, we have to resize */
836
// get_text(buff, entry->ptr);
837
// fprintf(stderr, "Failed to find an opening @%x for %8x:\n '%s'\n",
838
// hash_offset, entry->val, buff);
839
// for (old_entry = old_index->hash[hash_offset];
840
// old_entry < old_index->hash[hash_offset+1];
842
// get_text(buff, old_entry->ptr);
843
// fprintf(stderr, " [%2d] %8x %8x: '%s'\n",
844
// (int)(old_entry - old_index->hash[hash_offset]),
845
// old_entry->val, old_entry->ptr, buff);
851
/* For entries which we *do* manage to insert into old_index, we don't
852
* want them double copied into the final output.
854
old_index->num_entries++;
856
if (num_entries > 0) {
857
/* We couldn't fit the new entries into the old index, so allocate a
858
* new one, and fill it with stuff.
860
// fprintf(stderr, "inserted %d before resize\n", num_inserted);
861
new_index = create_index_from_old_and_new_entries(old_index,
864
new_index = old_index;
865
// fprintf(stderr, "inserted %d without resizing\n", num_inserted);
868
/* create_index_from_old_and_new_entries returns NULL on malloc failure */
870
return DELTA_OUT_OF_MEMORY;
875
void free_delta_index(struct delta_index *index)
881
sizeof_delta_index(struct delta_index *index)
884
return index->memsize;
890
* The maximum size for any opcode sequence, including the initial header
891
* plus Rabin window plus biggest copy.
893
#define MAX_OP_SIZE (5 + 5 + 1 + RABIN_WINDOW + 7)
896
create_delta(const struct delta_index *index,
897
const void *trg_buf, unsigned long trg_size,
898
unsigned long *delta_size, unsigned long max_size,
901
unsigned int i, outpos, outsize, moff, val;
903
const struct source_info *msource;
905
const unsigned char *ref_data, *ref_top, *data, *top;
908
if (!trg_buf || !trg_size)
909
return DELTA_BUFFER_EMPTY;
911
return DELTA_INDEX_NEEDED;
915
if (max_size && outsize >= max_size)
916
outsize = max_size + MAX_OP_SIZE + 1;
917
out = malloc(outsize);
919
return DELTA_OUT_OF_MEMORY;
921
/* store target buffer size */
924
out[outpos++] = i | 0x80;
930
top = (const unsigned char *) trg_buf + trg_size;
932
/* Start the matching by filling out with a simple 'insert' instruction, of
933
* the first RABIN_WINDOW bytes of the input.
935
outpos++; /* leave a byte for the insert command */
937
for (i = 0; i < RABIN_WINDOW && data < top; i++, data++) {
938
out[outpos++] = *data;
939
val = ((val << 8) | *data) ^ T[val >> RABIN_SHIFT];
941
/* we are now setup with an insert of 'i' bytes and val contains the RABIN
942
* hash for those bytes, and data points to the RABIN_WINDOW+1 byte of
952
/* we don't have a 'worthy enough' match yet, so let's look for
955
struct index_entry *entry;
956
/* Shift the window by one byte. */
957
val ^= U[data[-RABIN_WINDOW]];
958
val = ((val << 8) | *data) ^ T[val >> RABIN_SHIFT];
959
i = val & index->hash_mask;
960
/* TODO: When using multiple indexes like this, the hash tables
961
* mapping val => index_entry become less efficient.
962
* You end up getting a lot more collisions in the hash,
963
* which doesn't actually lead to a entry->val match.
965
for (entry = index->hash[i];
966
entry < index->hash[i+1] && entry->src != NULL;
968
const unsigned char *ref;
969
const unsigned char *src;
971
if (entry->val != val)
975
ref_data = entry->src->buf;
976
ref_top = ref_data + entry->src->size;
977
ref_size = ref_top - ref;
978
/* ref_size is the longest possible match that we could make
979
* here. If ref_size <= msize, then we know that we cannot
980
* match more bytes with this location that we have already
983
if (ref_size > (top - src))
984
ref_size = top - src;
985
if (ref_size <= msize)
987
/* See how many bytes actually match at this location. */
988
while (ref_size-- && *src++ == *ref)
990
if (msize < (ref - entry->ptr)) {
991
/* this is our best match so far */
992
msize = ref - entry->ptr;
993
msource = entry->src;
994
moff = entry->ptr - ref_data;
995
if (msize >= 4096) /* good enough */
1002
/* The best match right now is less than 4 bytes long. So just add
1003
* the current byte to the insert instruction. Increment the insert
1004
* counter, and copy the byte of data into the output buffer.
1008
out[outpos++] = *data++;
1010
if (inscnt == 0x7f) {
1011
/* We have a max length insert instruction, finalize it in the
1014
out[outpos - inscnt - 1] = inscnt;
1023
ref_data = msource->buf;
1024
while (moff && ref_data[moff-1] == data[-1]) {
1025
/* we can match one byte back */
1032
outpos--; /* remove count slot */
1033
inscnt--; /* make it -1 */
1036
out[outpos - inscnt - 1] = inscnt;
1040
/* A copy op is currently limited to 64KB (pack v2) */
1041
left = (msize < 0x10000) ? 0 : (msize - 0x10000);
1044
op = out + outpos++;
1047
/* moff is the offset in the local structure, for encoding, we need
1048
* to push it into the global offset
1050
assert(moff < msource->size);
1051
moff += msource->agg_offset;
1053
<= index->last_src->size + index->last_src->agg_offset);
1054
if (moff & 0x000000ff)
1055
out[outpos++] = moff >> 0, i |= 0x01;
1056
if (moff & 0x0000ff00)
1057
out[outpos++] = moff >> 8, i |= 0x02;
1058
if (moff & 0x00ff0000)
1059
out[outpos++] = moff >> 16, i |= 0x04;
1060
if (moff & 0xff000000)
1061
out[outpos++] = moff >> 24, i |= 0x08;
1062
/* Put it back into local coordinates, in case we have multiple
1065
moff -= msource->agg_offset;
1068
out[outpos++] = msize >> 0, i |= 0x10;
1070
out[outpos++] = msize >> 8, i |= 0x20;
1081
for (j = -RABIN_WINDOW; j < 0; j++)
1082
val = ((val << 8) | data[j])
1083
^ T[val >> RABIN_SHIFT];
1087
if (outpos >= outsize - MAX_OP_SIZE) {
1089
outsize = outsize * 3 / 2;
1090
if (max_size && outsize >= max_size)
1091
outsize = max_size + MAX_OP_SIZE + 1;
1092
if (max_size && outpos > max_size)
1094
out = realloc(out, outsize);
1097
return DELTA_OUT_OF_MEMORY;
1103
out[outpos - inscnt - 1] = inscnt;
1105
if (max_size && outpos > max_size) {
1107
return DELTA_SIZE_TOO_BIG;
1110
*delta_size = outpos;
1117
get_entry_summary(const struct delta_index *index, int pos,
1118
unsigned int *text_offset, unsigned int *hash_val)
1121
const struct index_entry *entry;
1122
const struct index_entry *start_of_entries;
1123
unsigned int offset;
1124
if (pos < 0 || text_offset == NULL || hash_val == NULL
1129
hsize = index->hash_mask + 1;
1130
start_of_entries = (struct index_entry *)(((struct index_entry **)index->hash) + (hsize + 1));
1131
entry = start_of_entries + pos;
1132
if (entry > index->last_entry) {
1135
if (entry->ptr == NULL) {
1139
offset = entry->src->agg_offset;
1140
offset += (entry->ptr - ((unsigned char *)entry->src->buf));
1141
*text_offset = offset;
1142
*hash_val = entry->val;
1149
get_hash_offset(const struct delta_index *index, int pos,
1150
unsigned int *entry_offset)
1153
const struct index_entry *entry;
1154
const struct index_entry *start_of_entries;
1155
if (pos < 0 || index == NULL || entry_offset == NULL)
1159
hsize = index->hash_mask + 1;
1160
start_of_entries = (struct index_entry *)(((struct index_entry **)index->hash) + (hsize + 1));
1164
entry = index->hash[pos];
1165
if (entry == NULL) {
1168
*entry_offset = (entry - start_of_entries);
1175
rabin_hash(const unsigned char *data)
1178
unsigned int val = 0;
1179
for (i = 0; i < RABIN_WINDOW; i++)
1180
val = ((val << 8) | data[i]) ^ T[val >> RABIN_SHIFT];
1184
/* vim: et ts=4 sw=4 sts=4
added
removed
breezy/bzr/diff-delta.c
