2 6bef87be 2018-09-11 stsp * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
4 6bef87be 2018-09-11 stsp * Permission to use, copy, modify, and distribute this software for any
5 6bef87be 2018-09-11 stsp * purpose with or without fee is hereby granted, provided that the above
6 6bef87be 2018-09-11 stsp * copyright notice and this permission notice appear in all copies.
8 6bef87be 2018-09-11 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 6bef87be 2018-09-11 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 6bef87be 2018-09-11 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 6bef87be 2018-09-11 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 6bef87be 2018-09-11 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 6bef87be 2018-09-11 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 6bef87be 2018-09-11 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 6bef87be 2018-09-11 stsp #include <sys/time.h>
18 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
19 8ab9215c 2021-10-15 thomas #include <sys/resource.h>
21 6bef87be 2018-09-11 stsp #include <stdio.h>
22 6bef87be 2018-09-11 stsp #include <stdlib.h>
23 a60c9e77 2019-05-22 stsp #include <stdint.h>
24 6bef87be 2018-09-11 stsp #include <string.h>
25 56e0773d 2019-11-28 stsp #include <limits.h>
26 6bef87be 2018-09-11 stsp #include <zlib.h>
28 6bef87be 2018-09-11 stsp #include "got_error.h"
29 6bef87be 2018-09-11 stsp #include "got_object.h"
31 6bef87be 2018-09-11 stsp #include "got_lib_delta.h"
32 6bef87be 2018-09-11 stsp #include "got_lib_inflate.h"
33 6bef87be 2018-09-11 stsp #include "got_lib_object.h"
34 f054b67a 2018-11-05 stsp #include "got_lib_object_idset.h"
35 6bef87be 2018-09-11 stsp #include "got_lib_object_cache.h"
38 a60c9e77 2019-05-22 stsp * XXX This should be reworked to track cache size and usage in bytes,
39 a60c9e77 2019-05-22 stsp * rather than tracking N elements capped to a maximum element size.
41 9bccfa63 2018-11-05 stsp #define GOT_OBJECT_CACHE_SIZE_OBJ 256
42 9185b863 2018-11-05 stsp #define GOT_OBJECT_CACHE_SIZE_TREE 256
43 9bccfa63 2018-11-05 stsp #define GOT_OBJECT_CACHE_SIZE_COMMIT 64
44 f73df793 2020-01-07 stsp #define GOT_OBJECT_CACHE_SIZE_TAG 2048
45 8ab9215c 2021-10-15 thomas #define GOT_OBJECT_CACHE_SIZE_RAW 64
46 a60c9e77 2019-05-22 stsp #define GOT_OBJECT_CACHE_MAX_ELEM_SIZE 1048576 /* 1 MB */
48 6bef87be 2018-09-11 stsp const struct got_error *
49 6bef87be 2018-09-11 stsp got_object_cache_init(struct got_object_cache *cache,
50 6bef87be 2018-09-11 stsp enum got_object_cache_type type)
52 8ab9215c 2021-10-15 thomas struct rlimit rl;
54 dab9d9b6 2018-11-05 stsp memset(cache, 0, sizeof(*cache));
56 f054b67a 2018-11-05 stsp cache->idset = got_object_idset_alloc();
57 f054b67a 2018-11-05 stsp if (cache->idset == NULL)
58 638f9024 2019-05-13 stsp return got_error_from_errno("got_object_idset_alloc");
60 f054b67a 2018-11-05 stsp cache->type = type;
61 6bef87be 2018-09-11 stsp switch (type) {
62 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
63 f054b67a 2018-11-05 stsp cache->size = GOT_OBJECT_CACHE_SIZE_OBJ;
65 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
66 f054b67a 2018-11-05 stsp cache->size = GOT_OBJECT_CACHE_SIZE_TREE;
68 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
69 f054b67a 2018-11-05 stsp cache->size = GOT_OBJECT_CACHE_SIZE_COMMIT;
71 f4a881ce 2018-11-17 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
72 f4a881ce 2018-11-17 stsp cache->size = GOT_OBJECT_CACHE_SIZE_TAG;
74 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
75 8ab9215c 2021-10-15 thomas if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
76 8ab9215c 2021-10-15 thomas return got_error_from_errno("getrlimit");
77 8ab9215c 2021-10-15 thomas cache->size = GOT_OBJECT_CACHE_SIZE_RAW;
78 8ab9215c 2021-10-15 thomas if (cache->size > rl.rlim_cur / 16)
79 8ab9215c 2021-10-15 thomas cache->size = rl.rlim_cur / 16;
82 6bef87be 2018-09-11 stsp return NULL;
85 ef20f542 2022-06-26 thomas static size_t
86 a60c9e77 2019-05-22 stsp get_size_obj(struct got_object *obj)
88 a60c9e77 2019-05-22 stsp size_t size = sizeof(*obj);
89 a60c9e77 2019-05-22 stsp struct got_delta *delta;
91 a60c9e77 2019-05-22 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
92 a60c9e77 2019-05-22 stsp return size;
94 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &obj->deltas.entries, entry) {
95 42c69117 2019-11-10 stsp if (SIZE_MAX - sizeof(*delta) < size)
96 a60c9e77 2019-05-22 stsp return SIZE_MAX;
97 42c69117 2019-11-10 stsp size += sizeof(*delta);
100 a60c9e77 2019-05-22 stsp return size;
103 ef20f542 2022-06-26 thomas static size_t
104 a60c9e77 2019-05-22 stsp get_size_tree(struct got_tree_object *tree)
106 a60c9e77 2019-05-22 stsp size_t size = sizeof(*tree);
108 56e0773d 2019-11-28 stsp size += sizeof(struct got_tree_entry) * tree->nentries;
109 a60c9e77 2019-05-22 stsp return size;
112 ef20f542 2022-06-26 thomas static size_t
113 a60c9e77 2019-05-22 stsp get_size_commit(struct got_commit_object *commit)
115 a60c9e77 2019-05-22 stsp size_t size = sizeof(*commit);
116 a60c9e77 2019-05-22 stsp struct got_object_qid *qid;
118 a60c9e77 2019-05-22 stsp size += sizeof(*commit->tree_id);
119 a60c9e77 2019-05-22 stsp size += strlen(commit->author);
120 a60c9e77 2019-05-22 stsp size += strlen(commit->committer);
121 a60c9e77 2019-05-22 stsp size += strlen(commit->logmsg);
123 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(qid, &commit->parent_ids, entry)
124 ec242592 2022-04-22 thomas size += sizeof(*qid) + sizeof(qid->id);
126 a60c9e77 2019-05-22 stsp return size;
129 ef20f542 2022-06-26 thomas static size_t
130 a60c9e77 2019-05-22 stsp get_size_tag(struct got_tag_object *tag)
132 a60c9e77 2019-05-22 stsp size_t size = sizeof(*tag);
134 a60c9e77 2019-05-22 stsp size += strlen(tag->tag);
135 a60c9e77 2019-05-22 stsp size += strlen(tag->tagger);
136 a60c9e77 2019-05-22 stsp size += strlen(tag->tagmsg);
138 a60c9e77 2019-05-22 stsp return size;
141 ef20f542 2022-06-26 thomas static size_t
142 8ab9215c 2021-10-15 thomas get_size_raw(struct got_raw_object *raw)
144 8ab9215c 2021-10-15 thomas return sizeof(*raw);
147 6bef87be 2018-09-11 stsp const struct got_error *
148 48b4f239 2021-12-31 thomas got_object_cache_add(struct got_object_cache *cache, struct got_object_id *id,
149 48b4f239 2021-12-31 thomas void *item)
151 6bef87be 2018-09-11 stsp const struct got_error *err = NULL;
152 6bef87be 2018-09-11 stsp struct got_object_cache_entry *ce;
154 a60c9e77 2019-05-22 stsp size_t size;
156 a60c9e77 2019-05-22 stsp switch (cache->type) {
157 a60c9e77 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
158 a60c9e77 2019-05-22 stsp size = get_size_obj((struct got_object *)item);
160 a60c9e77 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
161 a60c9e77 2019-05-22 stsp size = get_size_tree((struct got_tree_object *)item);
163 a60c9e77 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
164 a60c9e77 2019-05-22 stsp size = get_size_commit((struct got_commit_object *)item);
166 a60c9e77 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
167 a60c9e77 2019-05-22 stsp size = get_size_tag((struct got_tag_object *)item);
169 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
170 8ab9215c 2021-10-15 thomas size = get_size_raw((struct got_raw_object *)item);
173 a0de39f3 2019-08-09 stsp return got_error(GOT_ERR_OBJ_TYPE);
176 01cd7614 2019-05-22 stsp if (size > GOT_OBJECT_CACHE_MAX_ELEM_SIZE) {
177 01cd7614 2019-05-22 stsp #ifdef GOT_OBJ_CACHE_DEBUG
178 01cd7614 2019-05-22 stsp char *id_str;
179 01cd7614 2019-05-22 stsp if (got_object_id_str(&id_str, id) != NULL)
180 01cd7614 2019-05-22 stsp return got_error_from_errno("got_object_id_str");
181 01cd7614 2019-05-22 stsp fprintf(stderr, "%s: not caching ", getprogname());
182 01cd7614 2019-05-22 stsp switch (cache->type) {
183 01cd7614 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
184 01cd7614 2019-05-22 stsp fprintf(stderr, "object");
186 01cd7614 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
187 01cd7614 2019-05-22 stsp fprintf(stderr, "tree");
189 01cd7614 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
190 01cd7614 2019-05-22 stsp fprintf(stderr, "commit");
192 01cd7614 2019-05-22 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
193 01cd7614 2019-05-22 stsp fprintf(stderr, "tag");
195 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
196 8ab9215c 2021-10-15 thomas fprintf(stderr, "raw");
199 01cd7614 2019-05-22 stsp fprintf(stderr, " %s (%zd bytes; %zd MB)\n", id_str, size,
200 01cd7614 2019-05-22 stsp size/1024/1024);
201 01cd7614 2019-05-22 stsp free(id_str);
203 01cd7614 2019-05-22 stsp cache->cache_toolarge++;
204 79c99a64 2019-05-23 stsp return got_error(GOT_ERR_OBJ_TOO_LARGE);
207 f054b67a 2018-11-05 stsp nelem = got_object_idset_num_elements(cache->idset);
208 6bef87be 2018-09-11 stsp if (nelem >= cache->size) {
209 f054b67a 2018-11-05 stsp err = got_object_idset_remove((void **)&ce,
210 f054b67a 2018-11-05 stsp cache->idset, NULL);
212 6bef87be 2018-09-11 stsp return err;
213 6bef87be 2018-09-11 stsp switch (cache->type) {
214 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
215 6bef87be 2018-09-11 stsp got_object_close(ce->data.obj);
217 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
218 6bef87be 2018-09-11 stsp got_object_tree_close(ce->data.tree);
220 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
221 6bef87be 2018-09-11 stsp got_object_commit_close(ce->data.commit);
223 f4a881ce 2018-11-17 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
224 f4a881ce 2018-11-17 stsp got_object_tag_close(ce->data.tag);
226 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
227 8ab9215c 2021-10-15 thomas got_object_raw_close(ce->data.raw);
230 f1461b8c 2022-08-30 thomas memset(ce, 0, sizeof(*ce));
231 315fa2b2 2018-09-15 stsp cache->cache_evict++;
233 f1461b8c 2022-08-30 thomas ce = malloc(sizeof(*ce));
234 f1461b8c 2022-08-30 thomas if (ce == NULL)
235 f1461b8c 2022-08-30 thomas return got_error_from_errno("malloc");
238 6bef87be 2018-09-11 stsp memcpy(&ce->id, id, sizeof(ce->id));
239 6bef87be 2018-09-11 stsp switch (cache->type) {
240 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
241 6bef87be 2018-09-11 stsp ce->data.obj = (struct got_object *)item;
243 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
244 6bef87be 2018-09-11 stsp ce->data.tree = (struct got_tree_object *)item;
246 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
247 6bef87be 2018-09-11 stsp ce->data.commit = (struct got_commit_object *)item;
249 f4a881ce 2018-11-17 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
250 f4a881ce 2018-11-17 stsp ce->data.tag = (struct got_tag_object *)item;
252 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
253 8ab9215c 2021-10-15 thomas ce->data.raw = (struct got_raw_object *)item;
257 f054b67a 2018-11-05 stsp err = got_object_idset_add(cache->idset, id, ce);
260 cadc72f0 2022-08-30 thomas else if (size > cache->max_cached_size)
261 cadc72f0 2022-08-30 thomas cache->max_cached_size = size;
262 6bef87be 2018-09-11 stsp return err;
266 6bef87be 2018-09-11 stsp got_object_cache_get(struct got_object_cache *cache, struct got_object_id *id)
268 6bef87be 2018-09-11 stsp struct got_object_cache_entry *ce;
270 221e79cd 2018-09-16 stsp cache->cache_searches++;
271 f054b67a 2018-11-05 stsp ce = got_object_idset_get(cache->idset, id);
273 6bef87be 2018-09-11 stsp cache->cache_hit++;
274 6bef87be 2018-09-11 stsp switch (cache->type) {
275 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
276 6bef87be 2018-09-11 stsp return ce->data.obj;
277 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
278 6bef87be 2018-09-11 stsp return ce->data.tree;
279 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
280 6bef87be 2018-09-11 stsp return ce->data.commit;
281 f4a881ce 2018-11-17 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
282 f4a881ce 2018-11-17 stsp return ce->data.tag;
283 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
284 8ab9215c 2021-10-15 thomas return ce->data.raw;
288 6bef87be 2018-09-11 stsp cache->cache_miss++;
289 6bef87be 2018-09-11 stsp return NULL;
292 f4081577 2018-09-15 stsp #ifdef GOT_OBJ_CACHE_DEBUG
293 6bef87be 2018-09-11 stsp static void
294 6bef87be 2018-09-11 stsp print_cache_stats(struct got_object_cache *cache, const char *name)
296 221e79cd 2018-09-16 stsp fprintf(stderr, "%s: %s cache: %d elements, %d searches, %d hits, "
297 cadc72f0 2022-08-30 thomas "%d missed, %d evicted, %d too large, max cached %zd bytes\n",
298 cadc72f0 2022-08-30 thomas getprogname(), name,
299 ce971618 2022-08-31 thomas cache->idset ? got_object_idset_num_elements(cache->idset) : -1,
300 221e79cd 2018-09-16 stsp cache->cache_searches, cache->cache_hit,
301 cadc72f0 2022-08-30 thomas cache->cache_miss, cache->cache_evict, cache->cache_toolarge,
302 cadc72f0 2022-08-30 thomas cache->max_cached_size);
305 30b23f54 2022-08-30 thomas static const struct got_error *
306 cb103d04 2018-11-07 stsp check_refcount(struct got_object_id *id, void *data, void *arg)
308 6bef87be 2018-09-11 stsp struct got_object_cache *cache = arg;
309 6bef87be 2018-09-11 stsp struct got_object_cache_entry *ce = data;
310 6bef87be 2018-09-11 stsp struct got_object *obj;
311 6bef87be 2018-09-11 stsp struct got_tree_object *tree;
312 6bef87be 2018-09-11 stsp struct got_commit_object *commit;
313 f4a881ce 2018-11-17 stsp struct got_tag_object *tag;
314 8ab9215c 2021-10-15 thomas struct got_raw_object *raw;
315 6bef87be 2018-09-11 stsp char *id_str;
317 6bef87be 2018-09-11 stsp if (got_object_id_str(&id_str, id) != NULL)
318 cb103d04 2018-11-07 stsp return NULL;
320 6bef87be 2018-09-11 stsp switch (cache->type) {
321 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
322 6bef87be 2018-09-11 stsp obj = ce->data.obj;
323 6bef87be 2018-09-11 stsp if (obj->refcnt == 1)
325 6bef87be 2018-09-11 stsp fprintf(stderr, "object %s has %d unclaimed references\n",
326 6bef87be 2018-09-11 stsp id_str, obj->refcnt - 1);
328 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
329 6bef87be 2018-09-11 stsp tree = ce->data.tree;
330 6bef87be 2018-09-11 stsp if (tree->refcnt == 1)
332 6bef87be 2018-09-11 stsp fprintf(stderr, "tree %s has %d unclaimed references\n",
333 6bef87be 2018-09-11 stsp id_str, tree->refcnt - 1);
335 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
336 6bef87be 2018-09-11 stsp commit = ce->data.commit;
337 6bef87be 2018-09-11 stsp if (commit->refcnt == 1)
339 6bef87be 2018-09-11 stsp fprintf(stderr, "commit %s has %d unclaimed references\n",
340 414611d9 2018-09-19 stsp id_str, commit->refcnt - 1);
342 f4a881ce 2018-11-17 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
343 f4a881ce 2018-11-17 stsp tag = ce->data.tag;
344 f4a881ce 2018-11-17 stsp if (tag->refcnt == 1)
346 f4a881ce 2018-11-17 stsp fprintf(stderr, "tag %s has %d unclaimed references\n",
347 f4a881ce 2018-11-17 stsp id_str, tag->refcnt - 1);
349 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
350 8ab9215c 2021-10-15 thomas raw = ce->data.raw;
351 8ab9215c 2021-10-15 thomas if (raw->refcnt == 1)
353 8ab9215c 2021-10-15 thomas fprintf(stderr, "raw %s has %d unclaimed references\n",
354 8ab9215c 2021-10-15 thomas id_str, raw->refcnt - 1);
357 6bef87be 2018-09-11 stsp free(id_str);
358 cb103d04 2018-11-07 stsp return NULL;
363 6bef87be 2018-09-11 stsp got_object_cache_close(struct got_object_cache *cache)
365 f4081577 2018-09-15 stsp #ifdef GOT_OBJ_CACHE_DEBUG
366 6bef87be 2018-09-11 stsp switch (cache->type) {
367 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_OBJ:
368 6bef87be 2018-09-11 stsp print_cache_stats(cache, "object");
370 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_TREE:
371 6bef87be 2018-09-11 stsp print_cache_stats(cache, "tree");
373 6bef87be 2018-09-11 stsp case GOT_OBJECT_CACHE_TYPE_COMMIT:
374 6bef87be 2018-09-11 stsp print_cache_stats(cache, "commit");
376 2ee32b23 2019-05-15 stsp case GOT_OBJECT_CACHE_TYPE_TAG:
377 2ee32b23 2019-05-15 stsp print_cache_stats(cache, "tag");
379 8ab9215c 2021-10-15 thomas case GOT_OBJECT_CACHE_TYPE_RAW:
380 8ab9215c 2021-10-15 thomas print_cache_stats(cache, "raw");
384 ce971618 2022-08-31 thomas if (cache->idset)
385 ce971618 2022-08-31 thomas got_object_idset_for_each(cache->idset, check_refcount, cache);
388 f054b67a 2018-11-05 stsp if (cache->idset) {
389 f054b67a 2018-11-05 stsp got_object_idset_free(cache->idset);
390 f054b67a 2018-11-05 stsp cache->idset = NULL;
392 6bef87be 2018-09-11 stsp cache->size = 0;