1 ab2f42e7 2019-11-10 stsp /*
  2 5aa81393 2020-01-06 stsp  * Copyright (c) 2019, 2020 Stefan Sperling <stsp@openbsd.org>
  3 ab2f42e7 2019-11-10 stsp  *
  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.
  7 ab2f42e7 2019-11-10 stsp  *
  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.
 15 ab2f42e7 2019-11-10 stsp  */
 16 ab2f42e7 2019-11-10 stsp 
 17 ab2f42e7 2019-11-10 stsp #include <sys/queue.h>
 18 ab2f42e7 2019-11-10 stsp 
 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 ab2f42e7 2019-11-10 stsp #include <stdio.h>
 24 ab2f42e7 2019-11-10 stsp #include <zlib.h>
 25 ab2f42e7 2019-11-10 stsp #include <limits.h>
 26 ab2f42e7 2019-11-10 stsp #include <time.h>
 27 dac5c75e 2022-06-04 stsp #include <errno.h>
 28 dac5c75e 2022-06-04 stsp #include <siphash.h>
 29 ab2f42e7 2019-11-10 stsp 
 30 ab2f42e7 2019-11-10 stsp #include "got_object.h"
 31 ab2f42e7 2019-11-10 stsp #include "got_error.h"
 32 ab2f42e7 2019-11-10 stsp 
 33 ab2f42e7 2019-11-10 stsp #include "got_lib_delta.h"
 34 ab2f42e7 2019-11-10 stsp #include "got_lib_inflate.h"
 35 ab2f42e7 2019-11-10 stsp #include "got_lib_object.h"
 36 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
 37 ab2f42e7 2019-11-10 stsp 
 38 ab2f42e7 2019-11-10 stsp #ifndef nitems
 39 ab2f42e7 2019-11-10 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
 40 ab2f42e7 2019-11-10 stsp #endif
 41 ab2f42e7 2019-11-10 stsp 
 42 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MIN_BUCKETS	64
 43 99f7567e 2022-11-08 stsp #define GOT_DELTA_CACHE_MAX_BUCKETS	2048
 44 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_MAX_CHAIN	2
 45 99f7567e 2022-11-08 stsp #define GOT_DELTA_CACHE_MAX_DELTA_SIZE	1024
 46 dac5c75e 2022-06-04 stsp 
 47 dac5c75e 2022-06-04 stsp struct got_cached_delta {
 48 dac5c75e 2022-06-04 stsp 	off_t offset;
 49 dac5c75e 2022-06-04 stsp 	uint8_t *data;
 50 dac5c75e 2022-06-04 stsp 	size_t len;
 51 ab2f42e7 2019-11-10 stsp };
 52 ab2f42e7 2019-11-10 stsp 
 53 dac5c75e 2022-06-04 stsp struct got_delta_cache_head {
 54 dac5c75e 2022-06-04 stsp 	struct got_cached_delta entries[GOT_DELTA_CACHE_MAX_CHAIN];
 55 dac5c75e 2022-06-04 stsp 	unsigned int nchain;
 56 dac5c75e 2022-06-04 stsp };
 57 ab2f42e7 2019-11-10 stsp 
 58 ab2f42e7 2019-11-10 stsp struct got_delta_cache {
 59 dac5c75e 2022-06-04 stsp 	struct got_delta_cache_head *buckets;
 60 dac5c75e 2022-06-04 stsp 	unsigned int nbuckets;
 61 dac5c75e 2022-06-04 stsp 	unsigned int totelem;
 62 c3b318d0 2019-11-10 stsp 	int cache_search;
 63 c3b318d0 2019-11-10 stsp 	int cache_hit;
 64 c3b318d0 2019-11-10 stsp 	int cache_miss;
 65 c3b318d0 2019-11-10 stsp 	int cache_evict;
 66 c3b318d0 2019-11-10 stsp 	int cache_toolarge;
 67 dac5c75e 2022-06-04 stsp 	unsigned int flags;
 68 dac5c75e 2022-06-04 stsp #define GOT_DELTA_CACHE_F_NOMEM	0x01
 69 dac5c75e 2022-06-04 stsp 	SIPHASH_KEY key;
 70 ab2f42e7 2019-11-10 stsp };
 71 ab2f42e7 2019-11-10 stsp 
 72 dac5c75e 2022-06-04 stsp const struct got_error *
 73 dac5c75e 2022-06-04 stsp got_delta_cache_alloc(struct got_delta_cache **new)
 74 ab2f42e7 2019-11-10 stsp {
 75 dac5c75e 2022-06-04 stsp 	const struct got_error *err;
 76 ab2f42e7 2019-11-10 stsp 	struct got_delta_cache *cache;
 77 ab2f42e7 2019-11-10 stsp 
 78 dac5c75e 2022-06-04 stsp 	*new = NULL;
 79 dac5c75e 2022-06-04 stsp 
 80 ab2f42e7 2019-11-10 stsp 	cache = calloc(1, sizeof(*cache));
 81 ab2f42e7 2019-11-10 stsp 	if (cache == NULL)
 82 dac5c75e 2022-06-04 stsp 		return got_error_from_errno("calloc");
 83 20282b02 2022-06-16 tracey 
 84 dac5c75e 2022-06-04 stsp 	cache->buckets = calloc(GOT_DELTA_CACHE_MIN_BUCKETS,
 85 dac5c75e 2022-06-04 stsp 	    sizeof(cache->buckets[0]));
 86 dac5c75e 2022-06-04 stsp 	if (cache->buckets == NULL) {
 87 dac5c75e 2022-06-04 stsp 		err = got_error_from_errno("calloc");
 88 dac5c75e 2022-06-04 stsp 		free(cache);
 89 dac5c75e 2022-06-04 stsp 		return err;
 90 dac5c75e 2022-06-04 stsp 	}
 91 dac5c75e 2022-06-04 stsp 	cache->nbuckets = GOT_DELTA_CACHE_MIN_BUCKETS;
 92 ab2f42e7 2019-11-10 stsp 
 93 dac5c75e 2022-06-04 stsp 	arc4random_buf(&cache->key, sizeof(cache->key));
 94 dac5c75e 2022-06-04 stsp 	*new = cache;
 95 dac5c75e 2022-06-04 stsp 	return NULL;
 96 ab2f42e7 2019-11-10 stsp }
 97 ab2f42e7 2019-11-10 stsp 
 98 ab2f42e7 2019-11-10 stsp void
 99 ab2f42e7 2019-11-10 stsp got_delta_cache_free(struct got_delta_cache *cache)
100 ab2f42e7 2019-11-10 stsp {
101 dac5c75e 2022-06-04 stsp 	struct got_cached_delta *delta;
102 dac5c75e 2022-06-04 stsp 	unsigned int i;
103 ab2f42e7 2019-11-10 stsp 
104 b5e1b8cd 2022-12-01 stsp #ifdef GOT_DELTA_CACHE_DEBUG
105 dac5c75e 2022-06-04 stsp 	fprintf(stderr, "%s: delta cache: %u elements, %d searches, %d hits, "
106 dac5c75e 2022-06-04 stsp 	    "%d missed, %d evicted, %d too large\n", getprogname(),
107 dac5c75e 2022-06-04 stsp 	    cache->totelem, cache->cache_search, cache->cache_hit,
108 dac5c75e 2022-06-04 stsp 	    cache->cache_miss, cache->cache_evict, cache->cache_toolarge);
109 c3b318d0 2019-11-10 stsp #endif
110 dac5c75e 2022-06-04 stsp 	for (i = 0; i < cache->nbuckets; i++) {
111 dac5c75e 2022-06-04 stsp 		struct got_delta_cache_head *head;
112 dac5c75e 2022-06-04 stsp 		int j;
113 dac5c75e 2022-06-04 stsp 		head = &cache->buckets[i];
114 dac5c75e 2022-06-04 stsp 		for (j = 0; j < head->nchain; j++) {
115 dac5c75e 2022-06-04 stsp 			delta = &head->entries[j];
116 dac5c75e 2022-06-04 stsp 			free(delta->data);
117 dac5c75e 2022-06-04 stsp 		}
118 ab2f42e7 2019-11-10 stsp 	}
119 dac5c75e 2022-06-04 stsp 	free(cache->buckets);
120 ab2f42e7 2019-11-10 stsp 	free(cache);
121 ab2f42e7 2019-11-10 stsp }
122 ab2f42e7 2019-11-10 stsp 
123 dac5c75e 2022-06-04 stsp static uint64_t
124 dac5c75e 2022-06-04 stsp delta_cache_hash(struct got_delta_cache *cache, off_t delta_offset)
125 ab2f42e7 2019-11-10 stsp {
126 dac5c75e 2022-06-04 stsp 	return SipHash24(&cache->key, &delta_offset, sizeof(delta_offset));
127 dac5c75e 2022-06-04 stsp }
128 ab2f42e7 2019-11-10 stsp 
129 22f6beb0 2022-11-08 stsp #ifndef GOT_NO_DELTA_CACHE
130 dac5c75e 2022-06-04 stsp static const struct got_error *
131 dac5c75e 2022-06-04 stsp delta_cache_resize(struct got_delta_cache *cache, unsigned int nbuckets)
132 dac5c75e 2022-06-04 stsp {
133 dac5c75e 2022-06-04 stsp 	struct got_delta_cache_head *buckets;
134 dac5c75e 2022-06-04 stsp 	size_t i;
135 ab2f42e7 2019-11-10 stsp 
136 dac5c75e 2022-06-04 stsp 	buckets = calloc(nbuckets, sizeof(buckets[0]));
137 dac5c75e 2022-06-04 stsp 	if (buckets == NULL) {
138 dac5c75e 2022-06-04 stsp 		if (errno != ENOMEM)
139 dac5c75e 2022-06-04 stsp 			return got_error_from_errno("calloc");
140 dac5c75e 2022-06-04 stsp 		/* Proceed with our current amount of hash buckets. */
141 dac5c75e 2022-06-04 stsp 		cache->flags |= GOT_DELTA_CACHE_F_NOMEM;
142 dac5c75e 2022-06-04 stsp 		return NULL;
143 dac5c75e 2022-06-04 stsp 	}
144 dac5c75e 2022-06-04 stsp 
145 dac5c75e 2022-06-04 stsp 	arc4random_buf(&cache->key, sizeof(cache->key));
146 dac5c75e 2022-06-04 stsp 
147 dac5c75e 2022-06-04 stsp 	for (i = 0; i < cache->nbuckets; i++) {
148 dac5c75e 2022-06-04 stsp 		unsigned int j;
149 dac5c75e 2022-06-04 stsp 		for (j = 0; j < cache->buckets[i].nchain; j++) {
150 dac5c75e 2022-06-04 stsp 			struct got_delta_cache_head *head;
151 dac5c75e 2022-06-04 stsp 			struct got_cached_delta *delta;
152 dac5c75e 2022-06-04 stsp 			uint64_t idx;
153 dac5c75e 2022-06-04 stsp 			delta = &cache->buckets[i].entries[j];
154 dac5c75e 2022-06-04 stsp 			idx = delta_cache_hash(cache, delta->offset) % nbuckets;
155 dac5c75e 2022-06-04 stsp 			head = &buckets[idx];
156 dac5c75e 2022-06-04 stsp 			if (head->nchain < nitems(head->entries)) {
157 dac5c75e 2022-06-04 stsp 				struct got_cached_delta *new_delta;
158 dac5c75e 2022-06-04 stsp 				new_delta = &head->entries[head->nchain];
159 dac5c75e 2022-06-04 stsp 				memcpy(new_delta, delta, sizeof(*new_delta));
160 dac5c75e 2022-06-04 stsp 				head->nchain++;
161 b43d5a6c 2022-11-08 stsp 			} else {
162 dac5c75e 2022-06-04 stsp 				free(delta->data);
163 b43d5a6c 2022-11-08 stsp 				cache->totelem--;
164 b43d5a6c 2022-11-08 stsp 			}
165 dac5c75e 2022-06-04 stsp 		}
166 dac5c75e 2022-06-04 stsp 	}
167 dac5c75e 2022-06-04 stsp 
168 dac5c75e 2022-06-04 stsp 	free(cache->buckets);
169 dac5c75e 2022-06-04 stsp 	cache->buckets = buckets;
170 dac5c75e 2022-06-04 stsp 	cache->nbuckets = nbuckets;
171 dac5c75e 2022-06-04 stsp 	return NULL;
172 ab2f42e7 2019-11-10 stsp }
173 ab2f42e7 2019-11-10 stsp 
174 dac5c75e 2022-06-04 stsp static const struct got_error *
175 dac5c75e 2022-06-04 stsp delta_cache_grow(struct got_delta_cache *cache)
176 dac5c75e 2022-06-04 stsp {
177 dac5c75e 2022-06-04 stsp 	unsigned int nbuckets;
178 dac5c75e 2022-06-04 stsp 
179 dac5c75e 2022-06-04 stsp 	if ((cache->flags & GOT_DELTA_CACHE_F_NOMEM) ||
180 dac5c75e 2022-06-04 stsp 	    cache->nbuckets == GOT_DELTA_CACHE_MAX_BUCKETS)
181 dac5c75e 2022-06-04 stsp 		return NULL;
182 dac5c75e 2022-06-04 stsp 
183 dac5c75e 2022-06-04 stsp 	if (cache->nbuckets >= GOT_DELTA_CACHE_MAX_BUCKETS / 2)
184 dac5c75e 2022-06-04 stsp 		nbuckets = GOT_DELTA_CACHE_MAX_BUCKETS;
185 dac5c75e 2022-06-04 stsp 	else
186 dac5c75e 2022-06-04 stsp 		nbuckets = cache->nbuckets * 2;
187 dac5c75e 2022-06-04 stsp 
188 dac5c75e 2022-06-04 stsp 	return delta_cache_resize(cache, nbuckets);
189 dac5c75e 2022-06-04 stsp }
190 20282b02 2022-06-16 tracey #endif
191 dac5c75e 2022-06-04 stsp 
192 ab2f42e7 2019-11-10 stsp const struct got_error *
193 ab2f42e7 2019-11-10 stsp got_delta_cache_add(struct got_delta_cache *cache,
194 ab2f42e7 2019-11-10 stsp     off_t delta_data_offset, uint8_t *delta_data, size_t delta_len)
195 ab2f42e7 2019-11-10 stsp {
196 22f6beb0 2022-11-08 stsp #ifdef GOT_NO_DELTA_CACHE
197 fa7a529e 2020-01-06 stsp 	return got_error(GOT_ERR_NO_SPACE);
198 fa7a529e 2020-01-06 stsp #else
199 dac5c75e 2022-06-04 stsp 	const struct got_error *err = NULL;
200 dac5c75e 2022-06-04 stsp 	struct got_cached_delta *delta;
201 dac5c75e 2022-06-04 stsp 	struct got_delta_cache_head *head;
202 dac5c75e 2022-06-04 stsp 	uint64_t idx;
203 ab2f42e7 2019-11-10 stsp 
204 4264b841 2022-06-04 stsp 	if (delta_len > GOT_DELTA_CACHE_MAX_DELTA_SIZE) {
205 c3b318d0 2019-11-10 stsp 		cache->cache_toolarge++;
206 ab2f42e7 2019-11-10 stsp 		return got_error(GOT_ERR_NO_SPACE);
207 c3b318d0 2019-11-10 stsp 	}
208 ab2f42e7 2019-11-10 stsp 
209 dac5c75e 2022-06-04 stsp 	if (cache->nbuckets * 3 < cache->totelem * 4) {
210 dac5c75e 2022-06-04 stsp 		err = delta_cache_grow(cache);
211 dac5c75e 2022-06-04 stsp 		if (err)
212 dac5c75e 2022-06-04 stsp 			return err;
213 dac5c75e 2022-06-04 stsp 	}
214 ab2f42e7 2019-11-10 stsp 
215 dac5c75e 2022-06-04 stsp 	idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
216 dac5c75e 2022-06-04 stsp 	head = &cache->buckets[idx];
217 dac5c75e 2022-06-04 stsp 	if (head->nchain >= nitems(head->entries)) {
218 dac5c75e 2022-06-04 stsp 		delta = &head->entries[head->nchain - 1];
219 dac5c75e 2022-06-04 stsp 		free(delta->data);
220 dac5c75e 2022-06-04 stsp 		memset(delta, 0, sizeof(*delta));
221 dac5c75e 2022-06-04 stsp 		head->nchain--;
222 b43d5a6c 2022-11-08 stsp 		cache->totelem--;
223 b43d5a6c 2022-11-08 stsp 		cache->cache_evict++;
224 dac5c75e 2022-06-04 stsp 	}
225 ab2f42e7 2019-11-10 stsp 
226 dac5c75e 2022-06-04 stsp 	delta = &head->entries[head->nchain];
227 dac5c75e 2022-06-04 stsp 	delta->offset = delta_data_offset;
228 dac5c75e 2022-06-04 stsp 	delta->data = delta_data;
229 dac5c75e 2022-06-04 stsp 	delta->len = delta_len;
230 dac5c75e 2022-06-04 stsp 	head->nchain++;
231 dac5c75e 2022-06-04 stsp 	cache->totelem++;
232 ab2f42e7 2019-11-10 stsp 
233 ab2f42e7 2019-11-10 stsp 	return NULL;
234 fa7a529e 2020-01-06 stsp #endif
235 ab2f42e7 2019-11-10 stsp }
236 ab2f42e7 2019-11-10 stsp 
237 ab2f42e7 2019-11-10 stsp void
238 ab2f42e7 2019-11-10 stsp got_delta_cache_get(uint8_t **delta_data, size_t *delta_len,
239 ab2f42e7 2019-11-10 stsp     struct got_delta_cache *cache, off_t delta_data_offset)
240 ab2f42e7 2019-11-10 stsp {
241 dac5c75e 2022-06-04 stsp 	uint64_t idx;
242 dac5c75e 2022-06-04 stsp 	struct got_delta_cache_head *head;
243 dac5c75e 2022-06-04 stsp 	struct got_cached_delta *delta;
244 dac5c75e 2022-06-04 stsp 	int i;
245 ab2f42e7 2019-11-10 stsp 
246 dac5c75e 2022-06-04 stsp 	idx = delta_cache_hash(cache, delta_data_offset) % cache->nbuckets;
247 dac5c75e 2022-06-04 stsp 	head = &cache->buckets[idx];
248 dac5c75e 2022-06-04 stsp 
249 c3b318d0 2019-11-10 stsp 	cache->cache_search++;
250 ab2f42e7 2019-11-10 stsp 	*delta_data = NULL;
251 ab2f42e7 2019-11-10 stsp 	*delta_len = 0;
252 dac5c75e 2022-06-04 stsp 	for (i = 0; i < head->nchain; i++) {
253 dac5c75e 2022-06-04 stsp 		delta = &head->entries[i];
254 dac5c75e 2022-06-04 stsp 		if (delta->offset != delta_data_offset)
255 ab2f42e7 2019-11-10 stsp 			continue;
256 c3b318d0 2019-11-10 stsp 		cache->cache_hit++;
257 dac5c75e 2022-06-04 stsp 		if (i > 0) {
258 dac5c75e 2022-06-04 stsp 			struct got_cached_delta tmp;
259 dac5c75e 2022-06-04 stsp 			memcpy(&tmp, &head->entries[0], sizeof(tmp));
260 dac5c75e 2022-06-04 stsp 			memcpy(&head->entries[0], &head->entries[i],
261 dac5c75e 2022-06-04 stsp 			    sizeof(head->entries[0]));
262 dac5c75e 2022-06-04 stsp 			memcpy(&head->entries[i], &tmp,
263 dac5c75e 2022-06-04 stsp 			    sizeof(head->entries[i]));
264 dac5c75e 2022-06-04 stsp 			delta = &head->entries[0];
265 ab2f42e7 2019-11-10 stsp 		}
266 dac5c75e 2022-06-04 stsp 		*delta_data = delta->data;
267 dac5c75e 2022-06-04 stsp 		*delta_len = delta->len;
268 ab2f42e7 2019-11-10 stsp 		return;
269 ab2f42e7 2019-11-10 stsp 	}
270 c3b318d0 2019-11-10 stsp 
271 c3b318d0 2019-11-10 stsp 	cache->cache_miss++;
272 ab2f42e7 2019-11-10 stsp }