2 05118f5a 2021-06-22 stsp * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
4 05118f5a 2021-06-22 stsp * Permission to use, copy, modify, and distribute this software for any
5 05118f5a 2021-06-22 stsp * purpose with or without fee is hereby granted, provided that the above
6 05118f5a 2021-06-22 stsp * copyright notice and this permission notice appear in all copies.
8 05118f5a 2021-06-22 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 05118f5a 2021-06-22 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 05118f5a 2021-06-22 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 05118f5a 2021-06-22 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 05118f5a 2021-06-22 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 05118f5a 2021-06-22 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 05118f5a 2021-06-22 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 05118f5a 2021-06-22 stsp #include <sys/types.h>
18 8b925c6c 2022-07-16 thomas #include <sys/queue.h>
19 05118f5a 2021-06-22 stsp #include <sys/uio.h>
20 05118f5a 2021-06-22 stsp #include <sys/stat.h>
21 05118f5a 2021-06-22 stsp #include <sys/socket.h>
22 05118f5a 2021-06-22 stsp #include <sys/wait.h>
24 05118f5a 2021-06-22 stsp #include <dirent.h>
25 05118f5a 2021-06-22 stsp #include <errno.h>
26 05118f5a 2021-06-22 stsp #include <fcntl.h>
27 05118f5a 2021-06-22 stsp #include <stdint.h>
28 05118f5a 2021-06-22 stsp #include <stdio.h>
29 05118f5a 2021-06-22 stsp #include <stdlib.h>
30 05118f5a 2021-06-22 stsp #include <string.h>
31 05118f5a 2021-06-22 stsp #include <limits.h>
32 05118f5a 2021-06-22 stsp #include <unistd.h>
34 05118f5a 2021-06-22 stsp #include "got_error.h"
35 05118f5a 2021-06-22 stsp #include "got_cancel.h"
36 05118f5a 2021-06-22 stsp #include "got_object.h"
37 05118f5a 2021-06-22 stsp #include "got_reference.h"
38 05118f5a 2021-06-22 stsp #include "got_repository.h"
39 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
40 05118f5a 2021-06-22 stsp #include "got_opentemp.h"
41 05118f5a 2021-06-22 stsp #include "got_path.h"
43 05118f5a 2021-06-22 stsp #include "got_lib_delta.h"
44 05118f5a 2021-06-22 stsp #include "got_lib_object.h"
45 b3d68e7f 2021-07-03 stsp #include "got_lib_object_idset.h"
46 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
47 05118f5a 2021-06-22 stsp #include "got_lib_pack.h"
48 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
49 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
50 abd46894 2022-10-18 thomas #include "got_lib_ratelimit.h"
51 05118f5a 2021-06-22 stsp #include "got_lib_pack_create.h"
52 be288a59 2023-02-23 thomas #include "got_lib_hash.h"
53 b3d68e7f 2021-07-03 stsp #include "got_lib_lockfile.h"
55 05118f5a 2021-06-22 stsp #ifndef nitems
56 05118f5a 2021-06-22 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
59 05118f5a 2021-06-22 stsp static const struct got_error *
60 05118f5a 2021-06-22 stsp get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
61 05118f5a 2021-06-22 stsp unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
62 05118f5a 2021-06-22 stsp struct got_repository *repo,
63 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
65 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
66 05118f5a 2021-06-22 stsp const size_t alloc_chunksz = 256;
67 05118f5a 2021-06-22 stsp size_t nalloc;
68 05118f5a 2021-06-22 stsp struct got_reflist_entry *re;
71 05118f5a 2021-06-22 stsp *ids = NULL;
72 05118f5a 2021-06-22 stsp *nobjects = 0;
74 bd00a07c 2022-04-16 thomas err = got_reflist_sort(refs,
75 bd00a07c 2022-04-16 thomas got_ref_cmp_by_commit_timestamp_descending, repo);
77 bd00a07c 2022-04-16 thomas return err;
79 05118f5a 2021-06-22 stsp *ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
80 05118f5a 2021-06-22 stsp if (*ids == NULL)
81 05118f5a 2021-06-22 stsp return got_error_from_errno("reallocarray");
82 05118f5a 2021-06-22 stsp nalloc = alloc_chunksz;
84 05118f5a 2021-06-22 stsp TAILQ_FOREACH(re, refs, entry) {
85 05118f5a 2021-06-22 stsp struct got_object_id *id;
87 05118f5a 2021-06-22 stsp if (cancel_cb) {
88 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
93 05118f5a 2021-06-22 stsp err = got_ref_resolve(&id, repo, re->ref);
97 05118f5a 2021-06-22 stsp if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
98 05118f5a 2021-06-22 stsp int obj_type;
99 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
102 05118f5a 2021-06-22 stsp if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
109 ae23ce34 2021-07-01 stsp if (nalloc <= *nobjects) {
110 05118f5a 2021-06-22 stsp struct got_object_id **new;
111 05118f5a 2021-06-22 stsp new = recallocarray(*ids, nalloc,
112 05118f5a 2021-06-22 stsp nalloc + alloc_chunksz,
113 05118f5a 2021-06-22 stsp sizeof(struct got_object_id *));
114 05118f5a 2021-06-22 stsp if (new == NULL) {
115 05118f5a 2021-06-22 stsp err = got_error_from_errno(
116 05118f5a 2021-06-22 stsp "recallocarray");
119 05118f5a 2021-06-22 stsp *ids = new;
120 05118f5a 2021-06-22 stsp nalloc += alloc_chunksz;
122 05118f5a 2021-06-22 stsp (*ids)[*nobjects] = id;
123 05118f5a 2021-06-22 stsp if ((*ids)[*nobjects] == NULL) {
124 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
127 05118f5a 2021-06-22 stsp (*nobjects)++;
131 05118f5a 2021-06-22 stsp for (i = 0; i < *nobjects; i++)
132 05118f5a 2021-06-22 stsp free((*ids)[i]);
133 05118f5a 2021-06-22 stsp free(*ids);
134 05118f5a 2021-06-22 stsp *ids = NULL;
135 05118f5a 2021-06-22 stsp *nobjects = 0;
137 05118f5a 2021-06-22 stsp return err;
140 05118f5a 2021-06-22 stsp const struct got_error *
141 05118f5a 2021-06-22 stsp got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
142 05118f5a 2021-06-22 stsp struct got_reflist_head *include_refs,
143 05118f5a 2021-06-22 stsp struct got_reflist_head *exclude_refs, struct got_repository *repo,
144 d8253374 2023-02-17 thomas int loose_obj_only, int force_refdelta,
145 f9c2e8e5 2022-02-13 thomas got_pack_progress_cb progress_cb, void *progress_arg,
146 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
148 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
149 05118f5a 2021-06-22 stsp struct got_object_id **ours = NULL, **theirs = NULL;
150 05118f5a 2021-06-22 stsp int nours = 0, ntheirs = 0, packfd = -1, i;
151 05118f5a 2021-06-22 stsp char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
152 05118f5a 2021-06-22 stsp char *sha1_str = NULL;
153 05fa7118 2022-10-16 thomas FILE *delta_cache = NULL;
154 abd46894 2022-10-18 thomas struct got_ratelimit rl;
156 05118f5a 2021-06-22 stsp *packfile = NULL;
157 05118f5a 2021-06-22 stsp *pack_hash = NULL;
159 abd46894 2022-10-18 thomas got_ratelimit_init(&rl, 0, 500);
161 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/packing.pack",
162 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
163 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
166 fc2a50f2 2022-11-01 thomas err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
170 851a5b48 2023-02-03 thomas if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
171 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpfile_path);
175 05fa7118 2022-10-16 thomas delta_cache = got_opentemp();
176 05fa7118 2022-10-16 thomas if (delta_cache == NULL) {
177 05fa7118 2022-10-16 thomas err = got_error_from_errno("got_opentemp");
178 05fa7118 2022-10-16 thomas goto done;
181 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&ours, &nours,
182 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
183 abc59930 2021-09-05 naddy include_refs, repo, cancel_cb, cancel_arg);
187 05118f5a 2021-06-22 stsp if (nours == 0) {
188 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
192 05118f5a 2021-06-22 stsp if (!TAILQ_EMPTY(exclude_refs)) {
193 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&theirs, &ntheirs,
194 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
195 abc59930 2021-09-05 naddy exclude_refs, repo,
196 abc59930 2021-09-05 naddy cancel_cb, cancel_arg);
201 05118f5a 2021-06-22 stsp *pack_hash = calloc(1, sizeof(**pack_hash));
202 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
203 05118f5a 2021-06-22 stsp err = got_error_from_errno("calloc");
207 05fa7118 2022-10-16 thomas err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
208 d8253374 2023-02-17 thomas theirs, ntheirs, ours, nours, repo, loose_obj_only,
209 d8253374 2023-02-17 thomas 0, force_refdelta, progress_cb, progress_arg, &rl,
210 d8253374 2023-02-17 thomas cancel_cb, cancel_arg);
214 05118f5a 2021-06-22 stsp err = got_object_id_str(&sha1_str, *pack_hash);
217 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
218 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
219 05118f5a 2021-06-22 stsp sha1_str) == -1) {
220 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
224 20a7d452 2022-10-16 thomas if (lseek(packfd, 0L, SEEK_SET) == -1) {
225 20a7d452 2022-10-16 thomas err = got_error_from_errno("lseek");
228 05118f5a 2021-06-22 stsp if (rename(tmpfile_path, packfile_path) == -1) {
229 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpfile_path,
230 05118f5a 2021-06-22 stsp packfile_path);
233 05118f5a 2021-06-22 stsp free(tmpfile_path);
234 05118f5a 2021-06-22 stsp tmpfile_path = NULL;
236 20a7d452 2022-10-16 thomas *packfile = fdopen(packfd, "w");
237 20a7d452 2022-10-16 thomas if (*packfile == NULL) {
238 20a7d452 2022-10-16 thomas err = got_error_from_errno2("fdopen", tmpfile_path);
239 20a7d452 2022-10-16 thomas goto done;
241 20a7d452 2022-10-16 thomas packfd = -1;
243 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++)
244 05118f5a 2021-06-22 stsp free(ours[i]);
245 05118f5a 2021-06-22 stsp free(ours);
246 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++)
247 05118f5a 2021-06-22 stsp free(theirs[i]);
248 05118f5a 2021-06-22 stsp free(theirs);
249 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
250 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
251 05fa7118 2022-10-16 thomas if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
252 05fa7118 2022-10-16 thomas err = got_error_from_errno("fclose");
253 05118f5a 2021-06-22 stsp if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
254 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpfile_path);
255 05118f5a 2021-06-22 stsp free(tmpfile_path);
256 05118f5a 2021-06-22 stsp free(packfile_path);
257 05118f5a 2021-06-22 stsp free(sha1_str);
258 05118f5a 2021-06-22 stsp free(path);
260 05118f5a 2021-06-22 stsp free(*pack_hash);
261 05118f5a 2021-06-22 stsp *pack_hash = NULL;
262 05118f5a 2021-06-22 stsp if (*packfile)
263 05118f5a 2021-06-22 stsp fclose(*packfile);
264 05118f5a 2021-06-22 stsp *packfile = NULL;
266 05118f5a 2021-06-22 stsp return err;
269 05118f5a 2021-06-22 stsp const struct got_error *
270 05118f5a 2021-06-22 stsp got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
271 05118f5a 2021-06-22 stsp struct got_repository *repo,
272 05118f5a 2021-06-22 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
273 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
276 05118f5a 2021-06-22 stsp char *path;
277 05118f5a 2021-06-22 stsp int imsg_idxfds[2];
278 05118f5a 2021-06-22 stsp int npackfd = -1, idxfd = -1, nidxfd = -1;
279 05118f5a 2021-06-22 stsp int tmpfds[3];
280 05118f5a 2021-06-22 stsp int idxstatus, done = 0;
281 05118f5a 2021-06-22 stsp const struct got_error *err;
282 05118f5a 2021-06-22 stsp struct imsgbuf idxibuf;
283 05118f5a 2021-06-22 stsp pid_t idxpid;
284 05118f5a 2021-06-22 stsp char *tmpidxpath = NULL;
285 05118f5a 2021-06-22 stsp char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
286 05118f5a 2021-06-22 stsp const char *repo_path = got_repo_get_path_git_dir(repo);
287 05118f5a 2021-06-22 stsp struct stat sb;
289 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++)
290 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
292 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/indexing.idx",
293 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
294 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
297 fc2a50f2 2022-11-01 thomas err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
298 05118f5a 2021-06-22 stsp free(path);
301 851a5b48 2023-02-03 thomas if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
302 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpidxpath);
306 05118f5a 2021-06-22 stsp nidxfd = dup(idxfd);
307 05118f5a 2021-06-22 stsp if (nidxfd == -1) {
308 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
312 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
313 05118f5a 2021-06-22 stsp tmpfds[i] = got_opentempfd();
314 05118f5a 2021-06-22 stsp if (tmpfds[i] == -1) {
315 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_opentempfd");
320 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
324 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
325 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
326 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
330 05118f5a 2021-06-22 stsp if (fstat(fileno(packfile), &sb) == -1) {
331 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fstat", packfile_path);
335 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
336 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
337 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
341 05118f5a 2021-06-22 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
342 05118f5a 2021-06-22 stsp err = got_error_from_errno("socketpair");
345 05118f5a 2021-06-22 stsp idxpid = fork();
346 05118f5a 2021-06-22 stsp if (idxpid == -1) {
347 05118f5a 2021-06-22 stsp err= got_error_from_errno("fork");
349 05118f5a 2021-06-22 stsp } else if (idxpid == 0)
350 05118f5a 2021-06-22 stsp got_privsep_exec_child(imsg_idxfds,
351 05118f5a 2021-06-22 stsp GOT_PATH_PROG_INDEX_PACK, packfile_path);
352 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[1]) == -1) {
353 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
356 05118f5a 2021-06-22 stsp imsg_init(&idxibuf, imsg_idxfds[0]);
358 05118f5a 2021-06-22 stsp npackfd = dup(fileno(packfile));
359 05118f5a 2021-06-22 stsp if (npackfd == -1) {
360 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
363 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
365 05118f5a 2021-06-22 stsp if (err != NULL)
367 05118f5a 2021-06-22 stsp npackfd = -1;
368 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
369 05118f5a 2021-06-22 stsp if (err != NULL)
371 05118f5a 2021-06-22 stsp nidxfd = -1;
372 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
373 05118f5a 2021-06-22 stsp err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
374 05118f5a 2021-06-22 stsp if (err != NULL)
376 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
379 05118f5a 2021-06-22 stsp while (!done) {
380 05118f5a 2021-06-22 stsp int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
382 05118f5a 2021-06-22 stsp if (cancel_cb) {
383 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
388 05118f5a 2021-06-22 stsp err = got_privsep_recv_index_progress(&done, &nobj_total,
389 05118f5a 2021-06-22 stsp &nobj_indexed, &nobj_loose, &nobj_resolved,
391 05118f5a 2021-06-22 stsp if (err != NULL)
393 05118f5a 2021-06-22 stsp if (nobj_indexed != 0) {
394 05118f5a 2021-06-22 stsp err = progress_cb(progress_arg, sb.st_size,
395 05118f5a 2021-06-22 stsp nobj_total, nobj_indexed, nobj_loose,
396 05118f5a 2021-06-22 stsp nobj_resolved);
401 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[0]) == -1) {
402 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
405 05118f5a 2021-06-22 stsp if (waitpid(idxpid, &idxstatus, 0) == -1) {
406 05118f5a 2021-06-22 stsp err = got_error_from_errno("waitpid");
410 05118f5a 2021-06-22 stsp if (rename(tmpidxpath, idxpath) == -1) {
411 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpidxpath, idxpath);
414 05118f5a 2021-06-22 stsp free(tmpidxpath);
415 05118f5a 2021-06-22 stsp tmpidxpath = NULL;
418 05118f5a 2021-06-22 stsp if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
419 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpidxpath);
420 05118f5a 2021-06-22 stsp if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
421 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
422 05118f5a 2021-06-22 stsp if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
423 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
424 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
425 05118f5a 2021-06-22 stsp if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
426 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
428 05118f5a 2021-06-22 stsp free(tmpidxpath);
429 05118f5a 2021-06-22 stsp free(idxpath);
430 05118f5a 2021-06-22 stsp free(packfile_path);
431 05118f5a 2021-06-22 stsp return err;
434 05118f5a 2021-06-22 stsp const struct got_error *
435 05118f5a 2021-06-22 stsp got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
436 05118f5a 2021-06-22 stsp struct got_repository *repo, const char *packfile_path)
438 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
439 05118f5a 2021-06-22 stsp const char *packdir_path = NULL;
440 05118f5a 2021-06-22 stsp char *packfile_name = NULL, *p, *dot;
441 05118f5a 2021-06-22 stsp struct got_object_id id;
442 05118f5a 2021-06-22 stsp int packfd = -1;
444 05118f5a 2021-06-22 stsp *packfile = NULL;
445 05118f5a 2021-06-22 stsp *pack_hash = NULL;
447 05118f5a 2021-06-22 stsp packdir_path = got_repo_get_path_objects_pack(repo);
448 05118f5a 2021-06-22 stsp if (packdir_path == NULL)
449 05118f5a 2021-06-22 stsp return got_error_from_errno("got_repo_get_path_objects_pack");
451 05118f5a 2021-06-22 stsp if (!got_path_is_child(packfile_path, packdir_path,
452 05118f5a 2021-06-22 stsp strlen(packdir_path))) {
453 05118f5a 2021-06-22 stsp err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
458 05118f5a 2021-06-22 stsp err = got_path_basename(&packfile_name, packfile_path);
461 05118f5a 2021-06-22 stsp p = packfile_name;
463 05118f5a 2021-06-22 stsp if (strncmp(p, "pack-", 5) != 0) {
464 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
465 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
466 abc59930 2021-09-05 naddy packfile_name);
470 05118f5a 2021-06-22 stsp dot = strchr(p, '.');
471 05118f5a 2021-06-22 stsp if (dot == NULL) {
472 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
473 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
474 abc59930 2021-09-05 naddy packfile_name);
477 05118f5a 2021-06-22 stsp if (strcmp(dot + 1, "pack") != 0) {
478 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
479 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
480 abc59930 2021-09-05 naddy packfile_name);
483 05118f5a 2021-06-22 stsp *dot = '\0';
484 c8ae092d 2023-02-23 thomas if (!got_parse_object_id(&id, p, GOT_HASH_SHA1)) {
485 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
486 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
487 abc59930 2021-09-05 naddy packfile_name);
491 05118f5a 2021-06-22 stsp *pack_hash = got_object_id_dup(&id);
492 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
493 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
497 06340621 2021-12-31 thomas packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
498 05118f5a 2021-06-22 stsp if (packfd == -1) {
499 05118f5a 2021-06-22 stsp err = got_error_from_errno2("open", packfile_path);
503 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "r");
504 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
505 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", packfile_path);
508 05118f5a 2021-06-22 stsp packfd = -1;
510 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
511 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
512 05118f5a 2021-06-22 stsp free(packfile_name);
514 05118f5a 2021-06-22 stsp free(*pack_hash);
515 05118f5a 2021-06-22 stsp *pack_hash = NULL;
517 05118f5a 2021-06-22 stsp return err;
520 05118f5a 2021-06-22 stsp const struct got_error *
521 05118f5a 2021-06-22 stsp got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
522 05118f5a 2021-06-22 stsp struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
523 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
525 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
526 05118f5a 2021-06-22 stsp char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
527 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
528 05118f5a 2021-06-22 stsp struct got_pack *pack = NULL;
529 05118f5a 2021-06-22 stsp uint32_t nobj, i;
531 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
535 05118f5a 2021-06-22 stsp if (asprintf(&packpath, "%s/pack-%s.pack",
536 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
537 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
540 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/pack-%s.idx",
541 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
542 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
546 05118f5a 2021-06-22 stsp err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
550 05118f5a 2021-06-22 stsp err = got_repo_cache_pack(&pack, repo, packpath, packidx);
554 05118f5a 2021-06-22 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
555 05118f5a 2021-06-22 stsp for (i = 0; i < nobj; i++) {
556 05118f5a 2021-06-22 stsp struct got_packidx_object_id *oid;
557 05118f5a 2021-06-22 stsp struct got_object_id id, base_id;
558 05118f5a 2021-06-22 stsp off_t offset, base_offset = 0;
559 05118f5a 2021-06-22 stsp uint8_t type;
560 05118f5a 2021-06-22 stsp uint64_t size;
561 05118f5a 2021-06-22 stsp size_t tslen, len;
563 05118f5a 2021-06-22 stsp if (cancel_cb) {
564 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
568 05118f5a 2021-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
569 05118f5a 2021-06-22 stsp memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
571 05118f5a 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, i);
572 05118f5a 2021-06-22 stsp if (offset == -1) {
573 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
577 05118f5a 2021-06-22 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
578 05118f5a 2021-06-22 stsp pack, offset);
582 05118f5a 2021-06-22 stsp switch (type) {
583 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
584 05118f5a 2021-06-22 stsp err = got_pack_parse_offset_delta(&base_offset, &len,
585 05118f5a 2021-06-22 stsp pack, offset, tslen);
589 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_REF_DELTA:
590 05118f5a 2021-06-22 stsp err = got_pack_parse_ref_delta(&base_id,
591 05118f5a 2021-06-22 stsp pack, offset, tslen);
596 05118f5a 2021-06-22 stsp err = (*list_cb)(list_arg, &id, type, offset, size,
597 05118f5a 2021-06-22 stsp base_offset, &base_id);
603 05118f5a 2021-06-22 stsp free(id_str);
604 05118f5a 2021-06-22 stsp free(idxpath);
605 05118f5a 2021-06-22 stsp free(packpath);
606 05118f5a 2021-06-22 stsp if (packidx)
607 05118f5a 2021-06-22 stsp got_packidx_close(packidx);
608 b3d68e7f 2021-07-03 stsp return err;
611 b3d68e7f 2021-07-03 stsp static const struct got_error *
612 31ba2236 2022-01-05 thomas report_cleanup_progress(got_cleanup_progress_cb progress_cb,
613 31ba2236 2022-01-05 thomas void *progress_arg, struct got_ratelimit *rl,
614 31ba2236 2022-01-05 thomas int nloose, int ncommits, int npurged)
616 31ba2236 2022-01-05 thomas const struct got_error *err;
617 31ba2236 2022-01-05 thomas int elapsed;
619 31ba2236 2022-01-05 thomas if (progress_cb == NULL)
620 31ba2236 2022-01-05 thomas return NULL;
622 31ba2236 2022-01-05 thomas err = got_ratelimit_check(&elapsed, rl);
623 31ba2236 2022-01-05 thomas if (err || !elapsed)
624 31ba2236 2022-01-05 thomas return err;
626 31ba2236 2022-01-05 thomas return progress_cb(progress_arg, nloose, ncommits, npurged);
629 31ba2236 2022-01-05 thomas static const struct got_error *
630 b3d68e7f 2021-07-03 stsp get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
631 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
632 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, struct got_repository *repo)
634 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
635 b3d68e7f 2021-07-03 stsp char *path_objects = NULL, *path = NULL;
636 b3d68e7f 2021-07-03 stsp DIR *dir = NULL;
637 b3d68e7f 2021-07-03 stsp struct got_object *obj = NULL;
638 b3d68e7f 2021-07-03 stsp struct got_object_id id;
639 b3d68e7f 2021-07-03 stsp int i, fd = -1;
640 b3d68e7f 2021-07-03 stsp struct stat sb;
642 b3d68e7f 2021-07-03 stsp *ondisk_size = 0;
643 b3d68e7f 2021-07-03 stsp *loose_ids = got_object_idset_alloc();
644 b3d68e7f 2021-07-03 stsp if (*loose_ids == NULL)
645 b3d68e7f 2021-07-03 stsp return got_error_from_errno("got_object_idset_alloc");
647 b3d68e7f 2021-07-03 stsp path_objects = got_repo_get_path_objects(repo);
648 b3d68e7f 2021-07-03 stsp if (path_objects == NULL) {
649 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_repo_get_path_objects");
653 b3d68e7f 2021-07-03 stsp for (i = 0; i <= 0xff; i++) {
654 b3d68e7f 2021-07-03 stsp struct dirent *dent;
656 b3d68e7f 2021-07-03 stsp if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
657 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
661 b3d68e7f 2021-07-03 stsp dir = opendir(path);
662 b3d68e7f 2021-07-03 stsp if (dir == NULL) {
663 b3d68e7f 2021-07-03 stsp if (errno == ENOENT) {
664 b3d68e7f 2021-07-03 stsp err = NULL;
667 b3d68e7f 2021-07-03 stsp err = got_error_from_errno2("opendir", path);
671 b3d68e7f 2021-07-03 stsp while ((dent = readdir(dir)) != NULL) {
672 b3d68e7f 2021-07-03 stsp char *id_str;
674 b3d68e7f 2021-07-03 stsp if (strcmp(dent->d_name, ".") == 0 ||
675 b3d68e7f 2021-07-03 stsp strcmp(dent->d_name, "..") == 0)
678 b3d68e7f 2021-07-03 stsp if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
679 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
683 c8ae092d 2023-02-23 thomas if (!got_parse_object_id(&id, id_str,
684 c8ae092d 2023-02-23 thomas GOT_HASH_SHA1)) {
685 b3d68e7f 2021-07-03 stsp free(id_str);
688 b3d68e7f 2021-07-03 stsp free(id_str);
690 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, &id, repo);
693 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
694 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
697 d5c81d44 2021-07-08 stsp err = got_object_read_header_privsep(&obj, &id, repo,
701 b3d68e7f 2021-07-03 stsp fd = -1; /* already closed */
703 b3d68e7f 2021-07-03 stsp switch (obj->type) {
704 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
705 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
706 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_BLOB:
707 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
710 b3d68e7f 2021-07-03 stsp err = got_error_fmt(GOT_ERR_OBJ_TYPE,
711 b3d68e7f 2021-07-03 stsp "%d", obj->type);
714 b3d68e7f 2021-07-03 stsp got_object_close(obj);
715 b3d68e7f 2021-07-03 stsp obj = NULL;
716 b3d68e7f 2021-07-03 stsp (*ondisk_size) += sb.st_size;
717 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(*loose_ids, &id, NULL);
720 31ba2236 2022-01-05 thomas err = report_cleanup_progress(progress_cb,
721 31ba2236 2022-01-05 thomas progress_arg, rl,
722 31ba2236 2022-01-05 thomas got_object_idset_num_elements(*loose_ids), -1, -1);
724 31ba2236 2022-01-05 thomas goto done;
727 b3d68e7f 2021-07-03 stsp if (closedir(dir) != 0) {
728 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
731 b3d68e7f 2021-07-03 stsp dir = NULL;
733 b3d68e7f 2021-07-03 stsp free(path);
734 b3d68e7f 2021-07-03 stsp path = NULL;
737 b3d68e7f 2021-07-03 stsp if (dir && closedir(dir) != 0 && err == NULL)
738 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
739 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
740 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
742 b3d68e7f 2021-07-03 stsp got_object_idset_free(*loose_ids);
743 b3d68e7f 2021-07-03 stsp *loose_ids = NULL;
746 b3d68e7f 2021-07-03 stsp got_object_close(obj);
747 b3d68e7f 2021-07-03 stsp free(path_objects);
748 b3d68e7f 2021-07-03 stsp free(path);
749 b3d68e7f 2021-07-03 stsp return err;
752 b3d68e7f 2021-07-03 stsp static const struct got_error *
753 b3d68e7f 2021-07-03 stsp preserve_loose_object(struct got_object_idset *loose_ids,
754 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo, int *npacked)
756 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
757 9614da0d 2021-07-04 stsp struct got_object *obj;
759 b3d68e7f 2021-07-03 stsp if (!got_object_idset_contains(loose_ids, id))
760 b3d68e7f 2021-07-03 stsp return NULL;
763 9614da0d 2021-07-04 stsp * Try to open this object from a pack file. This ensures that
764 9614da0d 2021-07-04 stsp * we do in fact have a valid packed copy of the object. Otherwise
765 9614da0d 2021-07-04 stsp * we should not delete the loose representation of this object.
767 abc59930 2021-09-05 naddy err = got_object_open_packed(&obj, id, repo);
768 abc59930 2021-09-05 naddy if (err == NULL) {
769 9614da0d 2021-07-04 stsp got_object_close(obj);
771 9614da0d 2021-07-04 stsp * The object is referenced and packed.
772 9614da0d 2021-07-04 stsp * We can purge the redundantly stored loose object.
774 9614da0d 2021-07-04 stsp (*npacked)++;
775 9614da0d 2021-07-04 stsp return NULL;
776 9614da0d 2021-07-04 stsp } else if (err->code != GOT_ERR_NO_OBJ)
777 9614da0d 2021-07-04 stsp return err;
780 b3d68e7f 2021-07-03 stsp * This object is referenced and not packed.
781 b3d68e7f 2021-07-03 stsp * Remove it from our purge set.
783 b3d68e7f 2021-07-03 stsp return got_object_idset_remove(NULL, loose_ids, id);
786 b3d68e7f 2021-07-03 stsp static const struct got_error *
787 b3d68e7f 2021-07-03 stsp load_tree_entries(struct got_object_id_queue *ids,
788 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids,
789 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
790 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
791 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
793 b3d68e7f 2021-07-03 stsp const struct got_error *err;
794 b3d68e7f 2021-07-03 stsp struct got_tree_object *tree;
795 b3d68e7f 2021-07-03 stsp char *p = NULL;
798 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
800 b3d68e7f 2021-07-03 stsp return err;
802 b3d68e7f 2021-07-03 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
803 b3d68e7f 2021-07-03 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
804 b3d68e7f 2021-07-03 stsp struct got_object_id *id = got_tree_entry_get_id(e);
805 b3d68e7f 2021-07-03 stsp mode_t mode = got_tree_entry_get_mode(e);
807 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
808 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
813 b3d68e7f 2021-07-03 stsp if (got_object_tree_entry_is_symlink(e) ||
814 b3d68e7f 2021-07-03 stsp got_object_tree_entry_is_submodule(e) ||
815 b3d68e7f 2021-07-03 stsp got_object_idset_contains(traversed_ids, id))
818 b3d68e7f 2021-07-03 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
819 b3d68e7f 2021-07-03 stsp got_tree_entry_get_name(e)) == -1) {
820 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
824 b3d68e7f 2021-07-03 stsp if (S_ISDIR(mode)) {
825 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
826 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
829 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
830 b3d68e7f 2021-07-03 stsp } else if (S_ISREG(mode)) {
831 b3d68e7f 2021-07-03 stsp /* This blob is referenced. */
832 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, id, repo,
836 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, id, NULL);
844 b3d68e7f 2021-07-03 stsp got_object_tree_close(tree);
846 b3d68e7f 2021-07-03 stsp return err;
849 b3d68e7f 2021-07-03 stsp static const struct got_error *
850 b3d68e7f 2021-07-03 stsp load_tree(struct got_object_idset *loose_ids,
851 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
852 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
853 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
855 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
856 b3d68e7f 2021-07-03 stsp struct got_object_id_queue tree_ids;
857 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
859 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, tree_id);
861 b3d68e7f 2021-07-03 stsp return err;
863 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&tree_ids);
864 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
866 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&tree_ids)) {
867 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
868 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
873 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&tree_ids);
874 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&tree_ids, entry);
876 ec242592 2022-04-22 thomas if (got_object_idset_contains(traversed_ids, &qid->id)) {
877 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
881 ec242592 2022-04-22 thomas err = got_object_idset_add(traversed_ids, &qid->id, NULL);
883 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
887 b3d68e7f 2021-07-03 stsp /* This tree is referenced. */
888 ec242592 2022-04-22 thomas err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
892 b3d68e7f 2021-07-03 stsp err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
893 ec242592 2022-04-22 thomas &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
894 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
899 b3d68e7f 2021-07-03 stsp got_object_id_queue_free(&tree_ids);
900 b3d68e7f 2021-07-03 stsp return err;
903 b3d68e7f 2021-07-03 stsp static const struct got_error *
904 b3d68e7f 2021-07-03 stsp load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
905 b3d68e7f 2021-07-03 stsp int *npacked, struct got_object_idset *traversed_ids,
906 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo,
907 31ba2236 2022-01-05 thomas got_cleanup_progress_cb progress_cb, void *progress_arg,
908 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, int nloose,
909 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
911 b3d68e7f 2021-07-03 stsp const struct got_error *err;
912 b3d68e7f 2021-07-03 stsp struct got_commit_object *commit = NULL;
913 b3d68e7f 2021-07-03 stsp struct got_tag_object *tag = NULL;
914 b3d68e7f 2021-07-03 stsp struct got_object_id *tree_id = NULL;
915 b3d68e7f 2021-07-03 stsp struct got_object_id_queue ids;
916 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
917 b3d68e7f 2021-07-03 stsp int obj_type;
919 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
921 b3d68e7f 2021-07-03 stsp return err;
923 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&ids);
924 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&ids, qid, entry);
926 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&ids)) {
927 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
928 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
933 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&ids);
934 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&ids, entry);
936 ec242592 2022-04-22 thomas if (got_object_idset_contains(traversed_ids, &qid->id)) {
937 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
938 b3d68e7f 2021-07-03 stsp qid = NULL;
942 ec242592 2022-04-22 thomas err = got_object_idset_add(traversed_ids, &qid->id, NULL);
946 b3d68e7f 2021-07-03 stsp /* This commit or tag is referenced. */
947 ec242592 2022-04-22 thomas err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
951 ec242592 2022-04-22 thomas err = got_object_get_type(&obj_type, repo, &qid->id);
954 b3d68e7f 2021-07-03 stsp switch (obj_type) {
955 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
956 ec242592 2022-04-22 thomas err = got_object_open_as_commit(&commit, repo,
957 ec242592 2022-04-22 thomas &qid->id);
961 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
962 ec242592 2022-04-22 thomas err = got_object_open_as_tag(&tag, repo, &qid->id);
967 b3d68e7f 2021-07-03 stsp /* should not happen */
968 b3d68e7f 2021-07-03 stsp err = got_error(GOT_ERR_OBJ_TYPE);
972 b3d68e7f 2021-07-03 stsp /* Find a tree object to scan. */
973 b3d68e7f 2021-07-03 stsp if (commit) {
974 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
975 b3d68e7f 2021-07-03 stsp } else if (tag) {
976 b3d68e7f 2021-07-03 stsp obj_type = got_object_tag_get_object_type(tag);
977 b3d68e7f 2021-07-03 stsp switch (obj_type) {
978 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
979 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo,
980 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
983 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
985 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
986 b3d68e7f 2021-07-03 stsp tree_id = got_object_tag_get_object_id(tag);
990 b3d68e7f 2021-07-03 stsp * Tag points at something other than a
991 b3d68e7f 2021-07-03 stsp * commit or tree. Leave this weird tag object
992 b3d68e7f 2021-07-03 stsp * and the object it points to on disk.
994 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
995 ec242592 2022-04-22 thomas &qid->id);
996 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
998 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
999 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
1000 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
1001 b3d68e7f 2021-07-03 stsp goto done;
1002 b3d68e7f 2021-07-03 stsp err = NULL;
1007 b3d68e7f 2021-07-03 stsp if (tree_id) {
1008 b3d68e7f 2021-07-03 stsp err = load_tree(loose_ids, traversed_ids, tree_id, "",
1009 b3d68e7f 2021-07-03 stsp repo, npacked, cancel_cb, cancel_arg);
1014 b3d68e7f 2021-07-03 stsp if (commit || tag)
1015 b3d68e7f 2021-07-03 stsp (*ncommits)++; /* scanned tags are counted as commits */
1017 31ba2236 2022-01-05 thomas err = report_cleanup_progress(progress_cb, progress_arg, rl,
1018 31ba2236 2022-01-05 thomas nloose, *ncommits, -1);
1019 31ba2236 2022-01-05 thomas if (err)
1022 b3d68e7f 2021-07-03 stsp if (commit) {
1023 b3d68e7f 2021-07-03 stsp /* Find parent commits to scan. */
1024 b3d68e7f 2021-07-03 stsp const struct got_object_id_queue *parent_ids;
1025 b3d68e7f 2021-07-03 stsp parent_ids = got_object_commit_get_parent_ids(commit);
1026 b3d68e7f 2021-07-03 stsp err = got_object_id_queue_copy(parent_ids, &ids);
1029 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1030 b3d68e7f 2021-07-03 stsp commit = NULL;
1032 b3d68e7f 2021-07-03 stsp if (tag) {
1033 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1034 b3d68e7f 2021-07-03 stsp tag = NULL;
1036 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1037 b3d68e7f 2021-07-03 stsp qid = NULL;
1041 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1042 b3d68e7f 2021-07-03 stsp if (commit)
1043 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1045 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1046 4b2e47fb 2021-07-03 stsp got_object_id_queue_free(&ids);
1047 b3d68e7f 2021-07-03 stsp return err;
1050 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg {
1051 b3d68e7f 2021-07-03 stsp struct got_repository *repo;
1052 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb;
1053 b3d68e7f 2021-07-03 stsp void *progress_arg;
1054 31ba2236 2022-01-05 thomas struct got_ratelimit *rl;
1055 b3d68e7f 2021-07-03 stsp int nloose;
1056 b3d68e7f 2021-07-03 stsp int ncommits;
1057 b3d68e7f 2021-07-03 stsp int npurged;
1058 b3d68e7f 2021-07-03 stsp off_t size_purged;
1059 b3d68e7f 2021-07-03 stsp int dry_run;
1060 ef8ec606 2021-07-27 stsp time_t max_mtime;
1061 ef8ec606 2021-07-27 stsp int ignore_mtime;
1064 b3d68e7f 2021-07-03 stsp static const struct got_error *
1065 b3d68e7f 2021-07-03 stsp purge_loose_object(struct got_object_id *id, void *data, void *arg)
1067 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg *a = arg;
1068 b3d68e7f 2021-07-03 stsp const struct got_error *err, *unlock_err = NULL;
1069 b3d68e7f 2021-07-03 stsp char *path = NULL;
1070 b3d68e7f 2021-07-03 stsp int fd = -1;
1071 b3d68e7f 2021-07-03 stsp struct stat sb;
1072 b3d68e7f 2021-07-03 stsp struct got_lockfile *lf = NULL;
1074 b3d68e7f 2021-07-03 stsp err = got_object_get_path(&path, id, a->repo);
1076 b3d68e7f 2021-07-03 stsp return err;
1078 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, id, a->repo);
1080 b3d68e7f 2021-07-03 stsp goto done;
1082 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
1083 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
1084 b3d68e7f 2021-07-03 stsp goto done;
1088 ef8ec606 2021-07-27 stsp * Do not delete objects which are younger than our maximum
1089 ef8ec606 2021-07-27 stsp * modification time threshold. This prevents a race where
1090 ef8ec606 2021-07-27 stsp * new objects which are being added to the repository
1091 ef8ec606 2021-07-27 stsp * concurrently would be deleted.
1093 ef8ec606 2021-07-27 stsp if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1094 ef8ec606 2021-07-27 stsp if (!a->dry_run) {
1095 ef8ec606 2021-07-27 stsp err = got_lockfile_lock(&lf, path, -1);
1097 ef8ec606 2021-07-27 stsp goto done;
1098 ef8ec606 2021-07-27 stsp if (unlink(path) == -1) {
1099 ef8ec606 2021-07-27 stsp err = got_error_from_errno2("unlink", path);
1100 ef8ec606 2021-07-27 stsp goto done;
1104 ef8ec606 2021-07-27 stsp a->npurged++;
1105 ef8ec606 2021-07-27 stsp a->size_purged += sb.st_size;
1106 31ba2236 2022-01-05 thomas err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1107 31ba2236 2022-01-05 thomas a->rl, a->nloose, a->ncommits, a->npurged);
1108 31ba2236 2022-01-05 thomas if (err)
1109 31ba2236 2022-01-05 thomas goto done;
1112 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
1113 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
1114 b3d68e7f 2021-07-03 stsp free(path);
1116 5345b4c7 2021-07-06 stsp unlock_err = got_lockfile_unlock(lf, -1);
1117 b3d68e7f 2021-07-03 stsp return err ? err : unlock_err;
1120 b3d68e7f 2021-07-03 stsp const struct got_error *
1121 b3d68e7f 2021-07-03 stsp got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1122 b3d68e7f 2021-07-03 stsp off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1123 ef8ec606 2021-07-27 stsp int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1124 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1126 b3d68e7f 2021-07-03 stsp const struct got_error *err;
1127 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids;
1128 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids;
1129 b3d68e7f 2021-07-03 stsp struct got_object_id **referenced_ids;
1130 b3d68e7f 2021-07-03 stsp int i, nreferenced, nloose, ncommits = 0;
1131 b3d68e7f 2021-07-03 stsp struct got_reflist_head refs;
1132 ef8ec606 2021-07-27 stsp struct got_reflist_entry *re;
1133 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg arg;
1134 ef8ec606 2021-07-27 stsp time_t max_mtime = 0;
1135 31ba2236 2022-01-05 thomas struct got_ratelimit rl;
1137 b3d68e7f 2021-07-03 stsp TAILQ_INIT(&refs);
1138 31ba2236 2022-01-05 thomas got_ratelimit_init(&rl, 0, 500);
1140 b3d68e7f 2021-07-03 stsp *size_before = 0;
1141 b3d68e7f 2021-07-03 stsp *size_after = 0;
1142 b3d68e7f 2021-07-03 stsp *npacked = 0;
1144 b3d68e7f 2021-07-03 stsp err = get_loose_object_ids(&loose_ids, size_before,
1145 31ba2236 2022-01-05 thomas progress_cb, progress_arg, &rl, repo);
1147 b3d68e7f 2021-07-03 stsp return err;
1148 b3d68e7f 2021-07-03 stsp nloose = got_object_idset_num_elements(loose_ids);
1149 b3d68e7f 2021-07-03 stsp if (nloose == 0) {
1150 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1151 39a089ac 2022-01-05 thomas if (progress_cb) {
1152 39a089ac 2022-01-05 thomas err = progress_cb(progress_arg, 0, 0, 0);
1153 39a089ac 2022-01-05 thomas if (err)
1154 39a089ac 2022-01-05 thomas return err;
1156 b3d68e7f 2021-07-03 stsp return NULL;
1159 b3d68e7f 2021-07-03 stsp traversed_ids = got_object_idset_alloc();
1160 b3d68e7f 2021-07-03 stsp if (traversed_ids == NULL) {
1161 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_object_idset_alloc");
1162 b3d68e7f 2021-07-03 stsp goto done;
1165 b3d68e7f 2021-07-03 stsp err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1167 b3d68e7f 2021-07-03 stsp goto done;
1168 ef8ec606 2021-07-27 stsp if (!ignore_mtime) {
1169 ef8ec606 2021-07-27 stsp TAILQ_FOREACH(re, &refs, entry) {
1170 ef8ec606 2021-07-27 stsp time_t mtime = got_ref_get_mtime(re->ref);
1171 ef8ec606 2021-07-27 stsp if (mtime > max_mtime)
1172 ef8ec606 2021-07-27 stsp max_mtime = mtime;
1175 ef8ec606 2021-07-27 stsp * For safety, keep objects created within 10 minutes
1176 ef8ec606 2021-07-27 stsp * before the youngest reference was created.
1178 ef8ec606 2021-07-27 stsp if (max_mtime >= 600)
1179 ef8ec606 2021-07-27 stsp max_mtime -= 600;
1182 b3d68e7f 2021-07-03 stsp err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1183 b3d68e7f 2021-07-03 stsp (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1184 b3d68e7f 2021-07-03 stsp &refs, repo, cancel_cb, cancel_arg);
1186 b3d68e7f 2021-07-03 stsp goto done;
1188 b3d68e7f 2021-07-03 stsp for (i = 0; i < nreferenced; i++) {
1189 b3d68e7f 2021-07-03 stsp struct got_object_id *id = referenced_ids[i];
1190 b3d68e7f 2021-07-03 stsp err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1191 31ba2236 2022-01-05 thomas traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1192 31ba2236 2022-01-05 thomas nloose, cancel_cb, cancel_arg);
1194 b3d68e7f 2021-07-03 stsp goto done;
1197 b3d68e7f 2021-07-03 stsp /* Any remaining loose objects are unreferenced and can be purged. */
1198 b3d68e7f 2021-07-03 stsp arg.repo = repo;
1199 b3d68e7f 2021-07-03 stsp arg.progress_arg = progress_arg;
1200 b3d68e7f 2021-07-03 stsp arg.progress_cb = progress_cb;
1201 31ba2236 2022-01-05 thomas arg.rl = &rl;
1202 b3d68e7f 2021-07-03 stsp arg.nloose = nloose;
1203 b3d68e7f 2021-07-03 stsp arg.npurged = 0;
1204 b3d68e7f 2021-07-03 stsp arg.size_purged = 0;
1205 b3d68e7f 2021-07-03 stsp arg.ncommits = ncommits;
1206 b3d68e7f 2021-07-03 stsp arg.dry_run = dry_run;
1207 ef8ec606 2021-07-27 stsp arg.max_mtime = max_mtime;
1208 ef8ec606 2021-07-27 stsp arg.ignore_mtime = ignore_mtime;
1209 b3d68e7f 2021-07-03 stsp err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1211 b3d68e7f 2021-07-03 stsp goto done;
1212 b3d68e7f 2021-07-03 stsp *size_after = *size_before - arg.size_purged;
1214 31ba2236 2022-01-05 thomas /* Produce a final progress report. */
1215 31ba2236 2022-01-05 thomas if (progress_cb) {
1216 654306ef 2022-01-05 thomas err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1217 31ba2236 2022-01-05 thomas if (err)
1218 31ba2236 2022-01-05 thomas goto done;
1221 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1222 b3d68e7f 2021-07-03 stsp got_object_idset_free(traversed_ids);
1223 1124fe40 2021-07-07 stsp return err;
1226 1124fe40 2021-07-07 stsp static const struct got_error *
1227 1124fe40 2021-07-07 stsp remove_packidx(int dir_fd, const char *relpath)
1229 1124fe40 2021-07-07 stsp const struct got_error *err, *unlock_err;
1230 1124fe40 2021-07-07 stsp struct got_lockfile *lf;
1232 1124fe40 2021-07-07 stsp err = got_lockfile_lock(&lf, relpath, dir_fd);
1234 1124fe40 2021-07-07 stsp return err;
1235 1124fe40 2021-07-07 stsp if (unlinkat(dir_fd, relpath, 0) == -1)
1236 1124fe40 2021-07-07 stsp err = got_error_from_errno("unlinkat");
1237 1124fe40 2021-07-07 stsp unlock_err = got_lockfile_unlock(lf, dir_fd);
1238 1124fe40 2021-07-07 stsp return err ? err : unlock_err;
1241 1124fe40 2021-07-07 stsp const struct got_error *
1242 1124fe40 2021-07-07 stsp got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1243 1124fe40 2021-07-07 stsp got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1244 1124fe40 2021-07-07 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1246 dee47c87 2022-07-21 thomas const struct got_error *err = NULL;
1247 1124fe40 2021-07-07 stsp DIR *packdir = NULL;
1248 1124fe40 2021-07-07 stsp struct dirent *dent;
1249 1124fe40 2021-07-07 stsp char *pack_relpath = NULL;
1250 1124fe40 2021-07-07 stsp int packdir_fd;
1251 1124fe40 2021-07-07 stsp struct stat sb;
1253 1124fe40 2021-07-07 stsp packdir_fd = openat(got_repo_get_fd(repo),
1254 fc63f50d 2021-12-31 thomas GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1255 1124fe40 2021-07-07 stsp if (packdir_fd == -1) {
1256 1124fe40 2021-07-07 stsp if (errno == ENOENT)
1257 1124fe40 2021-07-07 stsp return NULL;
1258 1124fe40 2021-07-07 stsp return got_error_from_errno_fmt("openat: %s/%s",
1259 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1260 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR);
1263 1124fe40 2021-07-07 stsp packdir = fdopendir(packdir_fd);
1264 1124fe40 2021-07-07 stsp if (packdir == NULL) {
1265 1124fe40 2021-07-07 stsp err = got_error_from_errno("fdopendir");
1266 1124fe40 2021-07-07 stsp goto done;
1269 1124fe40 2021-07-07 stsp while ((dent = readdir(packdir)) != NULL) {
1270 1124fe40 2021-07-07 stsp if (cancel_cb) {
1271 1124fe40 2021-07-07 stsp err = cancel_cb(cancel_arg);
1273 1124fe40 2021-07-07 stsp goto done;
1276 dd038bc6 2021-09-21 thomas.ad if (!got_repo_is_packidx_filename(dent->d_name,
1277 dd038bc6 2021-09-21 thomas.ad strlen(dent->d_name)))
1280 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath,
1281 1124fe40 2021-07-07 stsp dent->d_name);
1283 1124fe40 2021-07-07 stsp goto done;
1285 1124fe40 2021-07-07 stsp if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1286 1124fe40 2021-07-07 stsp free(pack_relpath);
1287 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1290 1124fe40 2021-07-07 stsp if (errno != ENOENT) {
1291 1124fe40 2021-07-07 stsp err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1292 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1293 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1294 1124fe40 2021-07-07 stsp pack_relpath);
1295 1124fe40 2021-07-07 stsp goto done;
1298 1124fe40 2021-07-07 stsp if (!dry_run) {
1299 1124fe40 2021-07-07 stsp err = remove_packidx(packdir_fd, dent->d_name);
1301 1124fe40 2021-07-07 stsp goto done;
1303 1124fe40 2021-07-07 stsp if (progress_cb) {
1304 1124fe40 2021-07-07 stsp char *path;
1305 1124fe40 2021-07-07 stsp if (asprintf(&path, "%s/%s/%s",
1306 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1307 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1308 1124fe40 2021-07-07 stsp dent->d_name) == -1) {
1309 1124fe40 2021-07-07 stsp err = got_error_from_errno("asprintf");
1310 1124fe40 2021-07-07 stsp goto done;
1312 1124fe40 2021-07-07 stsp err = progress_cb(progress_arg, path);
1313 1124fe40 2021-07-07 stsp free(path);
1315 1124fe40 2021-07-07 stsp goto done;
1317 1124fe40 2021-07-07 stsp free(pack_relpath);
1318 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1321 1124fe40 2021-07-07 stsp if (packdir && closedir(packdir) != 0 && err == NULL)
1322 1124fe40 2021-07-07 stsp err = got_error_from_errno("closedir");
1323 1124fe40 2021-07-07 stsp free(pack_relpath);
1324 b3d68e7f 2021-07-03 stsp return err;
