2 5aa81393 2020-01-06 stsp * Copyright (c) 2019, 2020 Stefan Sperling <stsp@openbsd.org>
4 ab2f42e7 2019-11-10 stsp * Permission to use, copy, modify, and distribute this software for any
5 ab2f42e7 2019-11-10 stsp * purpose with or without fee is hereby granted, provided that the above
6 ab2f42e7 2019-11-10 stsp * copyright notice and this permission notice appear in all copies.
8 ab2f42e7 2019-11-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 ab2f42e7 2019-11-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 ab2f42e7 2019-11-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 ab2f42e7 2019-11-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 ab2f42e7 2019-11-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 ab2f42e7 2019-11-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 ab2f42e7 2019-11-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 ab2f42e7 2019-11-10 stsp #include <sys/queue.h>
19 ab2f42e7 2019-11-10 stsp #include <stdlib.h>
20 ab2f42e7 2019-11-10 stsp #include <string.h>
21 d89a2828 2022-07-05 stsp #include <stdint.h>
22 ab2f42e7 2019-11-10 stsp #include <sha1.h>
23 e83f12a6 2023-02-12 op #include <sha2.h>
24 ab2f42e7 2019-11-10 stsp #include <stdio.h>
25 ab2f42e7 2019-11-10 stsp #include <zlib.h>
26 ab2f42e7 2019-11-10 stsp #include <limits.h>
27 ab2f42e7 2019-11-10 stsp #include <time.h>
28 dac5c75e 2022-06-04 stsp #include <errno.h>
29 dac5c75e 2022-06-04 stsp #include <siphash.h>
31 ab2f42e7 2019-11-10 stsp #include "got_object.h"
32 ab2f42e7 2019-11-10 stsp #include "got_error.h"
34 ab2f42e7 2019-11-10 stsp #include "got_lib_delta.h"
35 ab2f42e7 2019-11-10 stsp #include "got_lib_inflate.h"
36 ab2f42e7 2019-11-10 stsp #include "got_lib_object.h"
37 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
39 ab2f42e7 2019-11-10 stsp #ifndef nitems
40 ab2f42e7 2019-11-10 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
43 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MIN_BUCKETS 64
44 99f7567e 2022-11-08 stsp #define GOT_DELTA_CACHE_MAX_BUCKETS 2048
45 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MAX_CHAIN 2
46 99f7567e 2022-11-08 stsp #define GOT_DELTA_CACHE_MAX_DELTA_SIZE 1024
48 dac5c75e 2022-06-04 stsp struct got_cached_delta {
49 dac5c75e 2022-06-04 stsp off_t offset;
50 dac5c75e 2022-06-04 stsp uint8_t *data;
54 dac5c75e 2022-06-04 stsp struct got_delta_cache_head {
55 dac5c75e 2022-06-04 stsp struct got_cached_delta entries[GOT_DELTA_CACHE_MAX_CHAIN];
56 dac5c75e 2022-06-04 stsp unsigned int nchain;
59 ab2f42e7 2019-11-10 stsp struct got_delta_cache {
60 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *buckets;
61 dac5c75e 2022-06-04 stsp unsigned int nbuckets;
62 dac5c75e 2022-06-04 stsp unsigned int totelem;
63 c3b318d0 2019-11-10 stsp int cache_search;
64 c3b318d0 2019-11-10 stsp int cache_hit;
65 c3b318d0 2019-11-10 stsp int cache_miss;
66 c3b318d0 2019-11-10 stsp int cache_evict;
67 c3b318d0 2019-11-10 stsp int cache_toolarge;
68 dac5c75e 2022-06-04 stsp unsigned int flags;
69 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_F_NOMEM 0x01
70 dac5c75e 2022-06-04 stsp SIPHASH_KEY key;
73 dac5c75e 2022-06-04 stsp const struct got_error *
74 dac5c75e 2022-06-04 stsp got_delta_cache_alloc(struct got_delta_cache **new)
76 dac5c75e 2022-06-04 stsp const struct got_error *err;
77 ab2f42e7 2019-11-10 stsp struct got_delta_cache *cache;
79 dac5c75e 2022-06-04 stsp *new = NULL;
81 ab2f42e7 2019-11-10 stsp cache = calloc(1, sizeof(*cache));
82 ab2f42e7 2019-11-10 stsp if (cache == NULL)
83 dac5c75e 2022-06-04 stsp return got_error_from_errno("calloc");
85 dac5c75e 2022-06-04 stsp cache->buckets = calloc(GOT_DELTA_CACHE_MIN_BUCKETS,
86 dac5c75e 2022-06-04 stsp sizeof(cache->buckets[0]));
87 dac5c75e 2022-06-04 stsp if (cache->buckets == NULL) {
88 dac5c75e 2022-06-04 stsp err = got_error_from_errno("calloc");
89 dac5c75e 2022-06-04 stsp free(cache);
92 dac5c75e 2022-06-04 stsp cache->nbuckets = GOT_DELTA_CACHE_MIN_BUCKETS;
94 dac5c75e 2022-06-04 stsp arc4random_buf(&cache->key, sizeof(cache->key));
95 dac5c75e 2022-06-04 stsp *new = cache;
96 dac5c75e 2022-06-04 stsp return NULL;
100 ab2f42e7 2019-11-10 stsp got_delta_cache_free(struct got_delta_cache *cache)
102 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
103 dac5c75e 2022-06-04 stsp unsigned int i;
105 b5e1b8cd 2022-12-01 stsp #ifdef GOT_DELTA_CACHE_DEBUG
106 dac5c75e 2022-06-04 stsp fprintf(stderr, "%s: delta cache: %u elements, %d searches, %d hits, "
107 dac5c75e 2022-06-04 stsp "%d missed, %d evicted, %d too large\n", getprogname(),
108 dac5c75e 2022-06-04 stsp cache->totelem, cache->cache_search, cache->cache_hit,
109 dac5c75e 2022-06-04 stsp cache->cache_miss, cache->cache_evict, cache->cache_toolarge);
111 dac5c75e 2022-06-04 stsp for (i = 0; i < cache->nbuckets; i++) {
112 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
114 dac5c75e 2022-06-04 stsp head = &cache->buckets[i];
115 dac5c75e 2022-06-04 stsp for (j = 0; j < head->nchain; j++) {
116 dac5c75e 2022-06-04 stsp delta = &head->entries[j];
117 dac5c75e 2022-06-04 stsp free(delta->data);
120 dac5c75e 2022-06-04 stsp free(cache->buckets);
121 ab2f42e7 2019-11-10 stsp free(cache);
124 dac5c75e 2022-06-04 stsp static uint64_t
125 dac5c75e 2022-06-04 stsp delta_cache_hash(struct got_delta_cache *cache, off_t delta_offset)
127 dac5c75e 2022-06-04 stsp return SipHash24(&cache->key, &delta_offset, sizeof(delta_offset));
130 22f6beb0 2022-11-08 stsp #ifndef GOT_NO_DELTA_CACHE
131 dac5c75e 2022-06-04 stsp static const struct got_error *
132 dac5c75e 2022-06-04 stsp delta_cache_resize(struct got_delta_cache *cache, unsigned int nbuckets)
134 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *buckets;
137 dac5c75e 2022-06-04 stsp buckets = calloc(nbuckets, sizeof(buckets[0]));
138 dac5c75e 2022-06-04 stsp if (buckets == NULL) {
139 dac5c75e 2022-06-04 stsp if (errno != ENOMEM)
140 dac5c75e 2022-06-04 stsp return got_error_from_errno("calloc");
141 dac5c75e 2022-06-04 stsp /* Proceed with our current amount of hash buckets. */
142 dac5c75e 2022-06-04 stsp cache->flags |= GOT_DELTA_CACHE_F_NOMEM;
143 dac5c75e 2022-06-04 stsp return NULL;
146 dac5c75e 2022-06-04 stsp arc4random_buf(&cache->key, sizeof(cache->key));
148 dac5c75e 2022-06-04 stsp for (i = 0; i < cache->nbuckets; i++) {
149 dac5c75e 2022-06-04 stsp unsigned int j;
150 dac5c75e 2022-06-04 stsp for (j = 0; j < cache->buckets[i].nchain; j++) {
151 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
152 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
153 dac5c75e 2022-06-04 stsp uint64_t idx;
154 dac5c75e 2022-06-04 stsp delta = &cache->buckets[i].entries[j];
155 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta->offset) % nbuckets;
156 dac5c75e 2022-06-04 stsp head = &buckets[idx];
157 dac5c75e 2022-06-04 stsp if (head->nchain < nitems(head->entries)) {
158 dac5c75e 2022-06-04 stsp struct got_cached_delta *new_delta;
159 dac5c75e 2022-06-04 stsp new_delta = &head->entries[head->nchain];
160 dac5c75e 2022-06-04 stsp memcpy(new_delta, delta, sizeof(*new_delta));
161 dac5c75e 2022-06-04 stsp head->nchain++;
163 dac5c75e 2022-06-04 stsp free(delta->data);
164 b43d5a6c 2022-11-08 stsp cache->totelem--;
169 dac5c75e 2022-06-04 stsp free(cache->buckets);
170 dac5c75e 2022-06-04 stsp cache->buckets = buckets;
171 dac5c75e 2022-06-04 stsp cache->nbuckets = nbuckets;
172 dac5c75e 2022-06-04 stsp return NULL;
175 dac5c75e 2022-06-04 stsp static const struct got_error *
176 dac5c75e 2022-06-04 stsp delta_cache_grow(struct got_delta_cache *cache)
178 dac5c75e 2022-06-04 stsp unsigned int nbuckets;
180 dac5c75e 2022-06-04 stsp if ((cache->flags & GOT_DELTA_CACHE_F_NOMEM) ||
181 dac5c75e 2022-06-04 stsp cache->nbuckets == GOT_DELTA_CACHE_MAX_BUCKETS)
182 dac5c75e 2022-06-04 stsp return NULL;
184 dac5c75e 2022-06-04 stsp if (cache->nbuckets >= GOT_DELTA_CACHE_MAX_BUCKETS / 2)
185 dac5c75e 2022-06-04 stsp nbuckets = GOT_DELTA_CACHE_MAX_BUCKETS;
187 dac5c75e 2022-06-04 stsp nbuckets = cache->nbuckets * 2;
189 dac5c75e 2022-06-04 stsp return delta_cache_resize(cache, nbuckets);
193 ab2f42e7 2019-11-10 stsp const struct got_error *
194 ab2f42e7 2019-11-10 stsp got_delta_cache_add(struct got_delta_cache *cache,
195 ab2f42e7 2019-11-10 stsp off_t delta_data_offset, uint8_t *delta_data, size_t delta_len)
197 22f6beb0 2022-11-08 stsp #ifdef GOT_NO_DELTA_CACHE
198 fa7a529e 2020-01-06 stsp return got_error(GOT_ERR_NO_SPACE);
200 dac5c75e 2022-06-04 stsp const struct got_error *err = NULL;
201 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
202 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
203 dac5c75e 2022-06-04 stsp uint64_t idx;
205 4264b841 2022-06-04 stsp if (delta_len > GOT_DELTA_CACHE_MAX_DELTA_SIZE) {
206 c3b318d0 2019-11-10 stsp cache->cache_toolarge++;
207 ab2f42e7 2019-11-10 stsp return got_error(GOT_ERR_NO_SPACE);
210 dac5c75e 2022-06-04 stsp if (cache->nbuckets * 3 < cache->totelem * 4) {
211 dac5c75e 2022-06-04 stsp err = delta_cache_grow(cache);
213 dac5c75e 2022-06-04 stsp return err;
216 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
217 dac5c75e 2022-06-04 stsp head = &cache->buckets[idx];
218 dac5c75e 2022-06-04 stsp if (head->nchain >= nitems(head->entries)) {
219 dac5c75e 2022-06-04 stsp delta = &head->entries[head->nchain - 1];
220 dac5c75e 2022-06-04 stsp free(delta->data);
221 dac5c75e 2022-06-04 stsp memset(delta, 0, sizeof(*delta));
222 dac5c75e 2022-06-04 stsp head->nchain--;
223 b43d5a6c 2022-11-08 stsp cache->totelem--;
224 b43d5a6c 2022-11-08 stsp cache->cache_evict++;
227 dac5c75e 2022-06-04 stsp delta = &head->entries[head->nchain];
228 dac5c75e 2022-06-04 stsp delta->offset = delta_data_offset;
229 dac5c75e 2022-06-04 stsp delta->data = delta_data;
230 dac5c75e 2022-06-04 stsp delta->len = delta_len;
231 dac5c75e 2022-06-04 stsp head->nchain++;
232 dac5c75e 2022-06-04 stsp cache->totelem++;
234 ab2f42e7 2019-11-10 stsp return NULL;
239 ab2f42e7 2019-11-10 stsp got_delta_cache_get(uint8_t **delta_data, size_t *delta_len,
240 ab2f42e7 2019-11-10 stsp struct got_delta_cache *cache, off_t delta_data_offset)
242 dac5c75e 2022-06-04 stsp uint64_t idx;
243 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
244 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
247 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
248 dac5c75e 2022-06-04 stsp head = &cache->buckets[idx];
250 c3b318d0 2019-11-10 stsp cache->cache_search++;
251 ab2f42e7 2019-11-10 stsp *delta_data = NULL;
252 ab2f42e7 2019-11-10 stsp *delta_len = 0;
253 dac5c75e 2022-06-04 stsp for (i = 0; i < head->nchain; i++) {
254 dac5c75e 2022-06-04 stsp delta = &head->entries[i];
255 dac5c75e 2022-06-04 stsp if (delta->offset != delta_data_offset)
257 c3b318d0 2019-11-10 stsp cache->cache_hit++;
258 dac5c75e 2022-06-04 stsp if (i > 0) {
259 dac5c75e 2022-06-04 stsp struct got_cached_delta tmp;
260 dac5c75e 2022-06-04 stsp memcpy(&tmp, &head->entries[0], sizeof(tmp));
261 dac5c75e 2022-06-04 stsp memcpy(&head->entries[0], &head->entries[i],
262 dac5c75e 2022-06-04 stsp sizeof(head->entries[0]));
263 dac5c75e 2022-06-04 stsp memcpy(&head->entries[i], &tmp,
264 dac5c75e 2022-06-04 stsp sizeof(head->entries[i]));
265 dac5c75e 2022-06-04 stsp delta = &head->entries[0];
267 dac5c75e 2022-06-04 stsp *delta_data = delta->data;
268 dac5c75e 2022-06-04 stsp *delta_len = delta->len;
272 c3b318d0 2019-11-10 stsp cache->cache_miss++;
