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 ab2f42e7 2019-11-10 stsp #include <sha1.h>
22 ab2f42e7 2019-11-10 stsp #include <stdio.h>
23 ab2f42e7 2019-11-10 stsp #include <zlib.h>
24 ab2f42e7 2019-11-10 stsp #include <limits.h>
25 ab2f42e7 2019-11-10 stsp #include <time.h>
26 dac5c75e 2022-06-04 stsp #include <errno.h>
27 dac5c75e 2022-06-04 stsp #include <siphash.h>
29 ab2f42e7 2019-11-10 stsp #include "got_object.h"
30 ab2f42e7 2019-11-10 stsp #include "got_error.h"
32 ab2f42e7 2019-11-10 stsp #include "got_lib_delta.h"
33 ab2f42e7 2019-11-10 stsp #include "got_lib_inflate.h"
34 ab2f42e7 2019-11-10 stsp #include "got_lib_object.h"
35 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
37 ab2f42e7 2019-11-10 stsp #ifndef nitems
38 ab2f42e7 2019-11-10 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
41 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MIN_BUCKETS 64
42 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MAX_BUCKETS 16384
43 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MAX_CHAIN 2
44 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MAX_DELTA_SIZE 2048
46 dac5c75e 2022-06-04 stsp struct got_cached_delta {
47 dac5c75e 2022-06-04 stsp off_t offset;
48 dac5c75e 2022-06-04 stsp uint8_t *data;
52 dac5c75e 2022-06-04 stsp struct got_delta_cache_head {
53 dac5c75e 2022-06-04 stsp struct got_cached_delta entries[GOT_DELTA_CACHE_MAX_CHAIN];
54 dac5c75e 2022-06-04 stsp unsigned int nchain;
57 ab2f42e7 2019-11-10 stsp struct got_delta_cache {
58 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *buckets;
59 dac5c75e 2022-06-04 stsp unsigned int nbuckets;
60 dac5c75e 2022-06-04 stsp unsigned int totelem;
61 c3b318d0 2019-11-10 stsp int cache_search;
62 c3b318d0 2019-11-10 stsp int cache_hit;
63 c3b318d0 2019-11-10 stsp int cache_miss;
64 c3b318d0 2019-11-10 stsp int cache_evict;
65 c3b318d0 2019-11-10 stsp int cache_toolarge;
66 dac5c75e 2022-06-04 stsp unsigned int flags;
67 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_F_NOMEM 0x01
68 dac5c75e 2022-06-04 stsp SIPHASH_KEY key;
71 dac5c75e 2022-06-04 stsp const struct got_error *
72 dac5c75e 2022-06-04 stsp got_delta_cache_alloc(struct got_delta_cache **new)
74 dac5c75e 2022-06-04 stsp const struct got_error *err;
75 ab2f42e7 2019-11-10 stsp struct got_delta_cache *cache;
77 dac5c75e 2022-06-04 stsp *new = NULL;
79 ab2f42e7 2019-11-10 stsp cache = calloc(1, sizeof(*cache));
80 ab2f42e7 2019-11-10 stsp if (cache == NULL)
81 dac5c75e 2022-06-04 stsp return got_error_from_errno("calloc");
83 dac5c75e 2022-06-04 stsp cache->buckets = calloc(GOT_DELTA_CACHE_MIN_BUCKETS,
84 dac5c75e 2022-06-04 stsp sizeof(cache->buckets[0]));
85 dac5c75e 2022-06-04 stsp if (cache->buckets == NULL) {
86 dac5c75e 2022-06-04 stsp err = got_error_from_errno("calloc");
87 dac5c75e 2022-06-04 stsp free(cache);
90 dac5c75e 2022-06-04 stsp cache->nbuckets = GOT_DELTA_CACHE_MIN_BUCKETS;
92 dac5c75e 2022-06-04 stsp arc4random_buf(&cache->key, sizeof(cache->key));
93 dac5c75e 2022-06-04 stsp *new = cache;
94 dac5c75e 2022-06-04 stsp return NULL;
98 ab2f42e7 2019-11-10 stsp got_delta_cache_free(struct got_delta_cache *cache)
100 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
101 dac5c75e 2022-06-04 stsp unsigned int i;
103 c3b318d0 2019-11-10 stsp #ifdef GOT_OBJ_CACHE_DEBUG
104 dac5c75e 2022-06-04 stsp fprintf(stderr, "%s: delta cache: %u elements, %d searches, %d hits, "
105 dac5c75e 2022-06-04 stsp "%d missed, %d evicted, %d too large\n", getprogname(),
106 dac5c75e 2022-06-04 stsp cache->totelem, cache->cache_search, cache->cache_hit,
107 dac5c75e 2022-06-04 stsp cache->cache_miss, cache->cache_evict, cache->cache_toolarge);
109 dac5c75e 2022-06-04 stsp for (i = 0; i < cache->nbuckets; i++) {
110 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
112 dac5c75e 2022-06-04 stsp head = &cache->buckets[i];
113 dac5c75e 2022-06-04 stsp for (j = 0; j < head->nchain; j++) {
114 dac5c75e 2022-06-04 stsp delta = &head->entries[j];
115 dac5c75e 2022-06-04 stsp free(delta->data);
118 dac5c75e 2022-06-04 stsp free(cache->buckets);
119 ab2f42e7 2019-11-10 stsp free(cache);
122 dac5c75e 2022-06-04 stsp static uint64_t
123 dac5c75e 2022-06-04 stsp delta_cache_hash(struct got_delta_cache *cache, off_t delta_offset)
125 dac5c75e 2022-06-04 stsp return SipHash24(&cache->key, &delta_offset, sizeof(delta_offset));
128 20282b02 2022-06-16 tracey #ifndef GOT_NO_OBJ_CACHE
129 dac5c75e 2022-06-04 stsp static const struct got_error *
130 dac5c75e 2022-06-04 stsp delta_cache_resize(struct got_delta_cache *cache, unsigned int nbuckets)
132 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *buckets;
135 dac5c75e 2022-06-04 stsp buckets = calloc(nbuckets, sizeof(buckets[0]));
136 dac5c75e 2022-06-04 stsp if (buckets == NULL) {
137 dac5c75e 2022-06-04 stsp if (errno != ENOMEM)
138 dac5c75e 2022-06-04 stsp return got_error_from_errno("calloc");
139 dac5c75e 2022-06-04 stsp /* Proceed with our current amount of hash buckets. */
140 dac5c75e 2022-06-04 stsp cache->flags |= GOT_DELTA_CACHE_F_NOMEM;
141 dac5c75e 2022-06-04 stsp return NULL;
144 dac5c75e 2022-06-04 stsp arc4random_buf(&cache->key, sizeof(cache->key));
146 dac5c75e 2022-06-04 stsp for (i = 0; i < cache->nbuckets; i++) {
147 dac5c75e 2022-06-04 stsp unsigned int j;
148 dac5c75e 2022-06-04 stsp for (j = 0; j < cache->buckets[i].nchain; j++) {
149 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
150 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
151 dac5c75e 2022-06-04 stsp uint64_t idx;
152 dac5c75e 2022-06-04 stsp delta = &cache->buckets[i].entries[j];
153 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta->offset) % nbuckets;
154 dac5c75e 2022-06-04 stsp head = &buckets[idx];
155 dac5c75e 2022-06-04 stsp if (head->nchain < nitems(head->entries)) {
156 dac5c75e 2022-06-04 stsp struct got_cached_delta *new_delta;
157 dac5c75e 2022-06-04 stsp new_delta = &head->entries[head->nchain];
158 dac5c75e 2022-06-04 stsp memcpy(new_delta, delta, sizeof(*new_delta));
159 dac5c75e 2022-06-04 stsp head->nchain++;
161 dac5c75e 2022-06-04 stsp free(delta->data);
165 dac5c75e 2022-06-04 stsp free(cache->buckets);
166 dac5c75e 2022-06-04 stsp cache->buckets = buckets;
167 dac5c75e 2022-06-04 stsp cache->nbuckets = nbuckets;
168 dac5c75e 2022-06-04 stsp return NULL;
171 dac5c75e 2022-06-04 stsp static const struct got_error *
172 dac5c75e 2022-06-04 stsp delta_cache_grow(struct got_delta_cache *cache)
174 dac5c75e 2022-06-04 stsp unsigned int nbuckets;
176 dac5c75e 2022-06-04 stsp if ((cache->flags & GOT_DELTA_CACHE_F_NOMEM) ||
177 dac5c75e 2022-06-04 stsp cache->nbuckets == GOT_DELTA_CACHE_MAX_BUCKETS)
178 dac5c75e 2022-06-04 stsp return NULL;
180 dac5c75e 2022-06-04 stsp if (cache->nbuckets >= GOT_DELTA_CACHE_MAX_BUCKETS / 2)
181 dac5c75e 2022-06-04 stsp nbuckets = GOT_DELTA_CACHE_MAX_BUCKETS;
183 dac5c75e 2022-06-04 stsp nbuckets = cache->nbuckets * 2;
185 dac5c75e 2022-06-04 stsp return delta_cache_resize(cache, nbuckets);
189 ab2f42e7 2019-11-10 stsp const struct got_error *
190 ab2f42e7 2019-11-10 stsp got_delta_cache_add(struct got_delta_cache *cache,
191 ab2f42e7 2019-11-10 stsp off_t delta_data_offset, uint8_t *delta_data, size_t delta_len)
193 fa7a529e 2020-01-06 stsp #ifdef GOT_NO_OBJ_CACHE
194 fa7a529e 2020-01-06 stsp return got_error(GOT_ERR_NO_SPACE);
196 dac5c75e 2022-06-04 stsp const struct got_error *err = NULL;
197 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
198 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
199 dac5c75e 2022-06-04 stsp uint64_t idx;
201 4264b841 2022-06-04 stsp if (delta_len > GOT_DELTA_CACHE_MAX_DELTA_SIZE) {
202 c3b318d0 2019-11-10 stsp cache->cache_toolarge++;
203 ab2f42e7 2019-11-10 stsp return got_error(GOT_ERR_NO_SPACE);
206 dac5c75e 2022-06-04 stsp if (cache->nbuckets * 3 < cache->totelem * 4) {
207 dac5c75e 2022-06-04 stsp err = delta_cache_grow(cache);
209 dac5c75e 2022-06-04 stsp return err;
212 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
213 dac5c75e 2022-06-04 stsp head = &cache->buckets[idx];
214 dac5c75e 2022-06-04 stsp if (head->nchain >= nitems(head->entries)) {
215 dac5c75e 2022-06-04 stsp delta = &head->entries[head->nchain - 1];
216 dac5c75e 2022-06-04 stsp free(delta->data);
217 dac5c75e 2022-06-04 stsp memset(delta, 0, sizeof(*delta));
218 dac5c75e 2022-06-04 stsp head->nchain--;
221 dac5c75e 2022-06-04 stsp delta = &head->entries[head->nchain];
222 dac5c75e 2022-06-04 stsp delta->offset = delta_data_offset;
223 dac5c75e 2022-06-04 stsp delta->data = delta_data;
224 dac5c75e 2022-06-04 stsp delta->len = delta_len;
225 dac5c75e 2022-06-04 stsp head->nchain++;
226 dac5c75e 2022-06-04 stsp cache->totelem++;
228 ab2f42e7 2019-11-10 stsp return NULL;
233 ab2f42e7 2019-11-10 stsp got_delta_cache_get(uint8_t **delta_data, size_t *delta_len,
234 ab2f42e7 2019-11-10 stsp struct got_delta_cache *cache, off_t delta_data_offset)
236 dac5c75e 2022-06-04 stsp uint64_t idx;
237 dac5c75e 2022-06-04 stsp struct got_delta_cache_head *head;
238 dac5c75e 2022-06-04 stsp struct got_cached_delta *delta;
241 dac5c75e 2022-06-04 stsp idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
242 dac5c75e 2022-06-04 stsp head = &cache->buckets[idx];
244 c3b318d0 2019-11-10 stsp cache->cache_search++;
245 ab2f42e7 2019-11-10 stsp *delta_data = NULL;
246 ab2f42e7 2019-11-10 stsp *delta_len = 0;
247 dac5c75e 2022-06-04 stsp for (i = 0; i < head->nchain; i++) {
248 dac5c75e 2022-06-04 stsp delta = &head->entries[i];
249 dac5c75e 2022-06-04 stsp if (delta->offset != delta_data_offset)
251 c3b318d0 2019-11-10 stsp cache->cache_hit++;
252 dac5c75e 2022-06-04 stsp if (i > 0) {
253 dac5c75e 2022-06-04 stsp struct got_cached_delta tmp;
254 dac5c75e 2022-06-04 stsp memcpy(&tmp, &head->entries[0], sizeof(tmp));
255 dac5c75e 2022-06-04 stsp memcpy(&head->entries[0], &head->entries[i],
256 dac5c75e 2022-06-04 stsp sizeof(head->entries[0]));
257 dac5c75e 2022-06-04 stsp memcpy(&head->entries[i], &tmp,
258 dac5c75e 2022-06-04 stsp sizeof(head->entries[i]));
259 dac5c75e 2022-06-04 stsp delta = &head->entries[0];
261 dac5c75e 2022-06-04 stsp *delta_data = delta->data;
262 dac5c75e 2022-06-04 stsp *delta_len = delta->len;
266 c3b318d0 2019-11-10 stsp cache->cache_miss++;
