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 05118f5a 2021-06-22 stsp #include <sys/uio.h>
19 05118f5a 2021-06-22 stsp #include <sys/stat.h>
20 05118f5a 2021-06-22 stsp #include <sys/socket.h>
21 05118f5a 2021-06-22 stsp #include <sys/wait.h>
23 05118f5a 2021-06-22 stsp #include <dirent.h>
24 05118f5a 2021-06-22 stsp #include <errno.h>
25 05118f5a 2021-06-22 stsp #include <fcntl.h>
26 05118f5a 2021-06-22 stsp #include <stdint.h>
27 05118f5a 2021-06-22 stsp #include <stdio.h>
28 05118f5a 2021-06-22 stsp #include <stdlib.h>
29 05118f5a 2021-06-22 stsp #include <string.h>
30 05118f5a 2021-06-22 stsp #include <limits.h>
31 05118f5a 2021-06-22 stsp #include <unistd.h>
33 05118f5a 2021-06-22 stsp #include "got_error.h"
34 05118f5a 2021-06-22 stsp #include "got_cancel.h"
35 05118f5a 2021-06-22 stsp #include "got_object.h"
36 05118f5a 2021-06-22 stsp #include "got_reference.h"
37 05118f5a 2021-06-22 stsp #include "got_repository.h"
38 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
39 05118f5a 2021-06-22 stsp #include "got_opentemp.h"
40 05118f5a 2021-06-22 stsp #include "got_path.h"
42 05118f5a 2021-06-22 stsp #include "got_lib_delta.h"
43 05118f5a 2021-06-22 stsp #include "got_lib_object.h"
44 b3d68e7f 2021-07-03 stsp #include "got_lib_object_idset.h"
45 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
46 05118f5a 2021-06-22 stsp #include "got_lib_pack.h"
47 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
48 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
49 05118f5a 2021-06-22 stsp #include "got_lib_pack_create.h"
50 05118f5a 2021-06-22 stsp #include "got_lib_sha1.h"
51 b3d68e7f 2021-07-03 stsp #include "got_lib_lockfile.h"
52 31ba2236 2022-01-05 thomas #include "got_lib_ratelimit.h"
54 05118f5a 2021-06-22 stsp #ifndef nitems
55 05118f5a 2021-06-22 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
58 05118f5a 2021-06-22 stsp static const struct got_error *
59 05118f5a 2021-06-22 stsp get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
60 05118f5a 2021-06-22 stsp unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
61 05118f5a 2021-06-22 stsp struct got_repository *repo,
62 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
64 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
65 05118f5a 2021-06-22 stsp const size_t alloc_chunksz = 256;
66 05118f5a 2021-06-22 stsp size_t nalloc;
67 05118f5a 2021-06-22 stsp struct got_reflist_entry *re;
70 05118f5a 2021-06-22 stsp *ids = NULL;
71 05118f5a 2021-06-22 stsp *nobjects = 0;
73 bd00a07c 2022-04-16 thomas err = got_reflist_sort(refs,
74 bd00a07c 2022-04-16 thomas got_ref_cmp_by_commit_timestamp_descending, repo);
76 bd00a07c 2022-04-16 thomas return err;
78 05118f5a 2021-06-22 stsp *ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
79 05118f5a 2021-06-22 stsp if (*ids == NULL)
80 05118f5a 2021-06-22 stsp return got_error_from_errno("reallocarray");
81 05118f5a 2021-06-22 stsp nalloc = alloc_chunksz;
83 05118f5a 2021-06-22 stsp TAILQ_FOREACH(re, refs, entry) {
84 05118f5a 2021-06-22 stsp struct got_object_id *id;
86 05118f5a 2021-06-22 stsp if (cancel_cb) {
87 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
92 05118f5a 2021-06-22 stsp err = got_ref_resolve(&id, repo, re->ref);
96 05118f5a 2021-06-22 stsp if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
97 05118f5a 2021-06-22 stsp int obj_type;
98 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
101 05118f5a 2021-06-22 stsp if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
108 ae23ce34 2021-07-01 stsp if (nalloc <= *nobjects) {
109 05118f5a 2021-06-22 stsp struct got_object_id **new;
110 05118f5a 2021-06-22 stsp new = recallocarray(*ids, nalloc,
111 05118f5a 2021-06-22 stsp nalloc + alloc_chunksz,
112 05118f5a 2021-06-22 stsp sizeof(struct got_object_id *));
113 05118f5a 2021-06-22 stsp if (new == NULL) {
114 05118f5a 2021-06-22 stsp err = got_error_from_errno(
115 05118f5a 2021-06-22 stsp "recallocarray");
118 05118f5a 2021-06-22 stsp *ids = new;
119 05118f5a 2021-06-22 stsp nalloc += alloc_chunksz;
121 05118f5a 2021-06-22 stsp (*ids)[*nobjects] = id;
122 05118f5a 2021-06-22 stsp if ((*ids)[*nobjects] == NULL) {
123 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
126 05118f5a 2021-06-22 stsp (*nobjects)++;
130 05118f5a 2021-06-22 stsp for (i = 0; i < *nobjects; i++)
131 05118f5a 2021-06-22 stsp free((*ids)[i]);
132 05118f5a 2021-06-22 stsp free(*ids);
133 05118f5a 2021-06-22 stsp *ids = NULL;
134 05118f5a 2021-06-22 stsp *nobjects = 0;
136 05118f5a 2021-06-22 stsp return err;
139 05118f5a 2021-06-22 stsp const struct got_error *
140 05118f5a 2021-06-22 stsp got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
141 05118f5a 2021-06-22 stsp struct got_reflist_head *include_refs,
142 05118f5a 2021-06-22 stsp struct got_reflist_head *exclude_refs, struct got_repository *repo,
143 f9c2e8e5 2022-02-13 thomas int loose_obj_only,
144 f9c2e8e5 2022-02-13 thomas got_pack_progress_cb progress_cb, void *progress_arg,
145 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
147 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
148 05118f5a 2021-06-22 stsp struct got_object_id **ours = NULL, **theirs = NULL;
149 05118f5a 2021-06-22 stsp int nours = 0, ntheirs = 0, packfd = -1, i;
150 05118f5a 2021-06-22 stsp char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
151 05118f5a 2021-06-22 stsp char *sha1_str = NULL;
153 05118f5a 2021-06-22 stsp *packfile = NULL;
154 05118f5a 2021-06-22 stsp *pack_hash = NULL;
156 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/packing.pack",
157 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
158 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
161 05118f5a 2021-06-22 stsp err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
165 05118f5a 2021-06-22 stsp if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
166 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpfile_path);
170 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "w");
171 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
172 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", tmpfile_path);
175 05118f5a 2021-06-22 stsp packfd = -1;
177 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&ours, &nours,
178 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
179 abc59930 2021-09-05 naddy include_refs, repo, cancel_cb, cancel_arg);
183 05118f5a 2021-06-22 stsp if (nours == 0) {
184 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
188 05118f5a 2021-06-22 stsp if (!TAILQ_EMPTY(exclude_refs)) {
189 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&theirs, &ntheirs,
190 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
191 abc59930 2021-09-05 naddy exclude_refs, repo,
192 abc59930 2021-09-05 naddy cancel_cb, cancel_arg);
197 05118f5a 2021-06-22 stsp *pack_hash = calloc(1, sizeof(**pack_hash));
198 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
199 05118f5a 2021-06-22 stsp err = got_error_from_errno("calloc");
203 05118f5a 2021-06-22 stsp err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
204 f8a36e22 2021-08-26 stsp ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
205 05118f5a 2021-06-22 stsp cancel_cb, cancel_arg);
209 05118f5a 2021-06-22 stsp err = got_object_id_str(&sha1_str, *pack_hash);
212 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
213 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
214 05118f5a 2021-06-22 stsp sha1_str) == -1) {
215 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
219 05118f5a 2021-06-22 stsp if (fflush(*packfile) == -1) {
220 05118f5a 2021-06-22 stsp err = got_error_from_errno("fflush");
223 05118f5a 2021-06-22 stsp if (fseek(*packfile, 0L, SEEK_SET) == -1) {
224 05118f5a 2021-06-22 stsp err = got_error_from_errno("fseek");
227 05118f5a 2021-06-22 stsp if (rename(tmpfile_path, packfile_path) == -1) {
228 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpfile_path,
229 05118f5a 2021-06-22 stsp packfile_path);
232 05118f5a 2021-06-22 stsp free(tmpfile_path);
233 05118f5a 2021-06-22 stsp tmpfile_path = NULL;
235 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++)
236 05118f5a 2021-06-22 stsp free(ours[i]);
237 05118f5a 2021-06-22 stsp free(ours);
238 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++)
239 05118f5a 2021-06-22 stsp free(theirs[i]);
240 05118f5a 2021-06-22 stsp free(theirs);
241 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
242 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
243 05118f5a 2021-06-22 stsp if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
244 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpfile_path);
245 05118f5a 2021-06-22 stsp free(tmpfile_path);
246 05118f5a 2021-06-22 stsp free(packfile_path);
247 05118f5a 2021-06-22 stsp free(sha1_str);
248 05118f5a 2021-06-22 stsp free(path);
250 05118f5a 2021-06-22 stsp free(*pack_hash);
251 05118f5a 2021-06-22 stsp *pack_hash = NULL;
252 05118f5a 2021-06-22 stsp if (*packfile)
253 05118f5a 2021-06-22 stsp fclose(*packfile);
254 05118f5a 2021-06-22 stsp *packfile = NULL;
256 05118f5a 2021-06-22 stsp return err;
259 05118f5a 2021-06-22 stsp const struct got_error *
260 05118f5a 2021-06-22 stsp got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
261 05118f5a 2021-06-22 stsp struct got_repository *repo,
262 05118f5a 2021-06-22 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
263 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
266 05118f5a 2021-06-22 stsp char *path;
267 05118f5a 2021-06-22 stsp int imsg_idxfds[2];
268 05118f5a 2021-06-22 stsp int npackfd = -1, idxfd = -1, nidxfd = -1;
269 05118f5a 2021-06-22 stsp int tmpfds[3];
270 05118f5a 2021-06-22 stsp int idxstatus, done = 0;
271 05118f5a 2021-06-22 stsp const struct got_error *err;
272 05118f5a 2021-06-22 stsp struct imsgbuf idxibuf;
273 05118f5a 2021-06-22 stsp pid_t idxpid;
274 05118f5a 2021-06-22 stsp char *tmpidxpath = NULL;
275 05118f5a 2021-06-22 stsp char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
276 05118f5a 2021-06-22 stsp const char *repo_path = got_repo_get_path_git_dir(repo);
277 05118f5a 2021-06-22 stsp struct stat sb;
279 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++)
280 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
282 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/indexing.idx",
283 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
284 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
287 05118f5a 2021-06-22 stsp err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
288 05118f5a 2021-06-22 stsp free(path);
291 05118f5a 2021-06-22 stsp if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
292 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpidxpath);
296 05118f5a 2021-06-22 stsp nidxfd = dup(idxfd);
297 05118f5a 2021-06-22 stsp if (nidxfd == -1) {
298 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
302 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
303 05118f5a 2021-06-22 stsp tmpfds[i] = got_opentempfd();
304 05118f5a 2021-06-22 stsp if (tmpfds[i] == -1) {
305 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_opentempfd");
310 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
314 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
315 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
316 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
320 05118f5a 2021-06-22 stsp if (fstat(fileno(packfile), &sb) == -1) {
321 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fstat", packfile_path);
325 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
326 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
327 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
331 05118f5a 2021-06-22 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
332 05118f5a 2021-06-22 stsp err = got_error_from_errno("socketpair");
335 05118f5a 2021-06-22 stsp idxpid = fork();
336 05118f5a 2021-06-22 stsp if (idxpid == -1) {
337 05118f5a 2021-06-22 stsp err= got_error_from_errno("fork");
339 05118f5a 2021-06-22 stsp } else if (idxpid == 0)
340 05118f5a 2021-06-22 stsp got_privsep_exec_child(imsg_idxfds,
341 05118f5a 2021-06-22 stsp GOT_PATH_PROG_INDEX_PACK, packfile_path);
342 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[1]) == -1) {
343 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
346 05118f5a 2021-06-22 stsp imsg_init(&idxibuf, imsg_idxfds[0]);
348 05118f5a 2021-06-22 stsp npackfd = dup(fileno(packfile));
349 05118f5a 2021-06-22 stsp if (npackfd == -1) {
350 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
353 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
355 05118f5a 2021-06-22 stsp if (err != NULL)
357 05118f5a 2021-06-22 stsp npackfd = -1;
358 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
359 05118f5a 2021-06-22 stsp if (err != NULL)
361 05118f5a 2021-06-22 stsp nidxfd = -1;
362 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
363 05118f5a 2021-06-22 stsp err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
364 05118f5a 2021-06-22 stsp if (err != NULL)
366 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
369 05118f5a 2021-06-22 stsp while (!done) {
370 05118f5a 2021-06-22 stsp int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
372 05118f5a 2021-06-22 stsp if (cancel_cb) {
373 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
378 05118f5a 2021-06-22 stsp err = got_privsep_recv_index_progress(&done, &nobj_total,
379 05118f5a 2021-06-22 stsp &nobj_indexed, &nobj_loose, &nobj_resolved,
381 05118f5a 2021-06-22 stsp if (err != NULL)
383 05118f5a 2021-06-22 stsp if (nobj_indexed != 0) {
384 05118f5a 2021-06-22 stsp err = progress_cb(progress_arg, sb.st_size,
385 05118f5a 2021-06-22 stsp nobj_total, nobj_indexed, nobj_loose,
386 05118f5a 2021-06-22 stsp nobj_resolved);
391 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[0]) == -1) {
392 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
395 05118f5a 2021-06-22 stsp if (waitpid(idxpid, &idxstatus, 0) == -1) {
396 05118f5a 2021-06-22 stsp err = got_error_from_errno("waitpid");
400 05118f5a 2021-06-22 stsp if (rename(tmpidxpath, idxpath) == -1) {
401 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpidxpath, idxpath);
404 05118f5a 2021-06-22 stsp free(tmpidxpath);
405 05118f5a 2021-06-22 stsp tmpidxpath = NULL;
408 05118f5a 2021-06-22 stsp if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
409 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpidxpath);
410 05118f5a 2021-06-22 stsp if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
411 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
412 05118f5a 2021-06-22 stsp if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
413 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
414 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
415 05118f5a 2021-06-22 stsp if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
416 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
418 05118f5a 2021-06-22 stsp free(tmpidxpath);
419 05118f5a 2021-06-22 stsp free(idxpath);
420 05118f5a 2021-06-22 stsp free(packfile_path);
421 05118f5a 2021-06-22 stsp return err;
424 05118f5a 2021-06-22 stsp const struct got_error *
425 05118f5a 2021-06-22 stsp got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
426 05118f5a 2021-06-22 stsp struct got_repository *repo, const char *packfile_path)
428 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
429 05118f5a 2021-06-22 stsp const char *packdir_path = NULL;
430 05118f5a 2021-06-22 stsp char *packfile_name = NULL, *p, *dot;
431 05118f5a 2021-06-22 stsp struct got_object_id id;
432 05118f5a 2021-06-22 stsp int packfd = -1;
434 05118f5a 2021-06-22 stsp *packfile = NULL;
435 05118f5a 2021-06-22 stsp *pack_hash = NULL;
437 05118f5a 2021-06-22 stsp packdir_path = got_repo_get_path_objects_pack(repo);
438 05118f5a 2021-06-22 stsp if (packdir_path == NULL)
439 05118f5a 2021-06-22 stsp return got_error_from_errno("got_repo_get_path_objects_pack");
441 05118f5a 2021-06-22 stsp if (!got_path_is_child(packfile_path, packdir_path,
442 05118f5a 2021-06-22 stsp strlen(packdir_path))) {
443 05118f5a 2021-06-22 stsp err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
448 05118f5a 2021-06-22 stsp err = got_path_basename(&packfile_name, packfile_path);
451 05118f5a 2021-06-22 stsp p = packfile_name;
453 05118f5a 2021-06-22 stsp if (strncmp(p, "pack-", 5) != 0) {
454 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
455 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
456 abc59930 2021-09-05 naddy packfile_name);
460 05118f5a 2021-06-22 stsp dot = strchr(p, '.');
461 05118f5a 2021-06-22 stsp if (dot == NULL) {
462 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
463 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
464 abc59930 2021-09-05 naddy packfile_name);
467 05118f5a 2021-06-22 stsp if (strcmp(dot + 1, "pack") != 0) {
468 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
469 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
470 abc59930 2021-09-05 naddy packfile_name);
473 05118f5a 2021-06-22 stsp *dot = '\0';
474 05118f5a 2021-06-22 stsp if (!got_parse_sha1_digest(id.sha1, p)) {
475 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
476 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
477 abc59930 2021-09-05 naddy packfile_name);
481 05118f5a 2021-06-22 stsp *pack_hash = got_object_id_dup(&id);
482 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
483 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
487 06340621 2021-12-31 thomas packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
488 05118f5a 2021-06-22 stsp if (packfd == -1) {
489 05118f5a 2021-06-22 stsp err = got_error_from_errno2("open", packfile_path);
493 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "r");
494 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
495 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", packfile_path);
498 05118f5a 2021-06-22 stsp packfd = -1;
500 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
501 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
502 05118f5a 2021-06-22 stsp free(packfile_name);
504 05118f5a 2021-06-22 stsp free(*pack_hash);
505 05118f5a 2021-06-22 stsp *pack_hash = NULL;
507 05118f5a 2021-06-22 stsp return err;
510 05118f5a 2021-06-22 stsp const struct got_error *
511 05118f5a 2021-06-22 stsp got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
512 05118f5a 2021-06-22 stsp struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
513 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
515 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
516 05118f5a 2021-06-22 stsp char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
517 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
518 05118f5a 2021-06-22 stsp struct got_pack *pack = NULL;
519 05118f5a 2021-06-22 stsp uint32_t nobj, i;
521 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
525 05118f5a 2021-06-22 stsp if (asprintf(&packpath, "%s/pack-%s.pack",
526 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
527 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
530 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/pack-%s.idx",
531 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
532 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
536 05118f5a 2021-06-22 stsp err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
540 05118f5a 2021-06-22 stsp err = got_repo_cache_pack(&pack, repo, packpath, packidx);
544 05118f5a 2021-06-22 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
545 05118f5a 2021-06-22 stsp for (i = 0; i < nobj; i++) {
546 05118f5a 2021-06-22 stsp struct got_packidx_object_id *oid;
547 05118f5a 2021-06-22 stsp struct got_object_id id, base_id;
548 05118f5a 2021-06-22 stsp off_t offset, base_offset = 0;
549 05118f5a 2021-06-22 stsp uint8_t type;
550 05118f5a 2021-06-22 stsp uint64_t size;
551 05118f5a 2021-06-22 stsp size_t tslen, len;
553 05118f5a 2021-06-22 stsp if (cancel_cb) {
554 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
558 05118f5a 2021-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
559 05118f5a 2021-06-22 stsp memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
561 05118f5a 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, i);
562 05118f5a 2021-06-22 stsp if (offset == -1) {
563 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
567 05118f5a 2021-06-22 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
568 05118f5a 2021-06-22 stsp pack, offset);
572 05118f5a 2021-06-22 stsp switch (type) {
573 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
574 05118f5a 2021-06-22 stsp err = got_pack_parse_offset_delta(&base_offset, &len,
575 05118f5a 2021-06-22 stsp pack, offset, tslen);
579 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_REF_DELTA:
580 05118f5a 2021-06-22 stsp err = got_pack_parse_ref_delta(&base_id,
581 05118f5a 2021-06-22 stsp pack, offset, tslen);
586 05118f5a 2021-06-22 stsp err = (*list_cb)(list_arg, &id, type, offset, size,
587 05118f5a 2021-06-22 stsp base_offset, &base_id);
593 05118f5a 2021-06-22 stsp free(id_str);
594 05118f5a 2021-06-22 stsp free(idxpath);
595 05118f5a 2021-06-22 stsp free(packpath);
596 05118f5a 2021-06-22 stsp if (packidx)
597 05118f5a 2021-06-22 stsp got_packidx_close(packidx);
598 b3d68e7f 2021-07-03 stsp return err;
601 b3d68e7f 2021-07-03 stsp static const struct got_error *
602 31ba2236 2022-01-05 thomas report_cleanup_progress(got_cleanup_progress_cb progress_cb,
603 31ba2236 2022-01-05 thomas void *progress_arg, struct got_ratelimit *rl,
604 31ba2236 2022-01-05 thomas int nloose, int ncommits, int npurged)
606 31ba2236 2022-01-05 thomas const struct got_error *err;
607 31ba2236 2022-01-05 thomas int elapsed;
609 31ba2236 2022-01-05 thomas if (progress_cb == NULL)
610 31ba2236 2022-01-05 thomas return NULL;
612 31ba2236 2022-01-05 thomas err = got_ratelimit_check(&elapsed, rl);
613 31ba2236 2022-01-05 thomas if (err || !elapsed)
614 31ba2236 2022-01-05 thomas return err;
616 31ba2236 2022-01-05 thomas return progress_cb(progress_arg, nloose, ncommits, npurged);
619 31ba2236 2022-01-05 thomas static const struct got_error *
620 b3d68e7f 2021-07-03 stsp get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
621 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
622 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, struct got_repository *repo)
624 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
625 b3d68e7f 2021-07-03 stsp char *path_objects = NULL, *path = NULL;
626 b3d68e7f 2021-07-03 stsp DIR *dir = NULL;
627 b3d68e7f 2021-07-03 stsp struct got_object *obj = NULL;
628 b3d68e7f 2021-07-03 stsp struct got_object_id id;
629 b3d68e7f 2021-07-03 stsp int i, fd = -1;
630 b3d68e7f 2021-07-03 stsp struct stat sb;
632 b3d68e7f 2021-07-03 stsp *ondisk_size = 0;
633 b3d68e7f 2021-07-03 stsp *loose_ids = got_object_idset_alloc();
634 b3d68e7f 2021-07-03 stsp if (*loose_ids == NULL)
635 b3d68e7f 2021-07-03 stsp return got_error_from_errno("got_object_idset_alloc");
637 b3d68e7f 2021-07-03 stsp path_objects = got_repo_get_path_objects(repo);
638 b3d68e7f 2021-07-03 stsp if (path_objects == NULL) {
639 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_repo_get_path_objects");
643 b3d68e7f 2021-07-03 stsp for (i = 0; i <= 0xff; i++) {
644 b3d68e7f 2021-07-03 stsp struct dirent *dent;
646 b3d68e7f 2021-07-03 stsp if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
647 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
651 b3d68e7f 2021-07-03 stsp dir = opendir(path);
652 b3d68e7f 2021-07-03 stsp if (dir == NULL) {
653 b3d68e7f 2021-07-03 stsp if (errno == ENOENT) {
654 b3d68e7f 2021-07-03 stsp err = NULL;
657 b3d68e7f 2021-07-03 stsp err = got_error_from_errno2("opendir", path);
661 b3d68e7f 2021-07-03 stsp while ((dent = readdir(dir)) != NULL) {
662 b3d68e7f 2021-07-03 stsp char *id_str;
664 b3d68e7f 2021-07-03 stsp if (strcmp(dent->d_name, ".") == 0 ||
665 b3d68e7f 2021-07-03 stsp strcmp(dent->d_name, "..") == 0)
668 b3d68e7f 2021-07-03 stsp if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
669 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
673 b3d68e7f 2021-07-03 stsp memset(&id, 0, sizeof(id));
674 b3d68e7f 2021-07-03 stsp if (!got_parse_sha1_digest(id.sha1, id_str)) {
675 b3d68e7f 2021-07-03 stsp free(id_str);
678 b3d68e7f 2021-07-03 stsp free(id_str);
680 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, &id, repo);
683 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
684 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
687 d5c81d44 2021-07-08 stsp err = got_object_read_header_privsep(&obj, &id, repo,
691 b3d68e7f 2021-07-03 stsp fd = -1; /* already closed */
693 b3d68e7f 2021-07-03 stsp switch (obj->type) {
694 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
695 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
696 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_BLOB:
697 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
700 b3d68e7f 2021-07-03 stsp err = got_error_fmt(GOT_ERR_OBJ_TYPE,
701 b3d68e7f 2021-07-03 stsp "%d", obj->type);
704 b3d68e7f 2021-07-03 stsp got_object_close(obj);
705 b3d68e7f 2021-07-03 stsp obj = NULL;
706 b3d68e7f 2021-07-03 stsp (*ondisk_size) += sb.st_size;
707 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(*loose_ids, &id, NULL);
710 31ba2236 2022-01-05 thomas err = report_cleanup_progress(progress_cb,
711 31ba2236 2022-01-05 thomas progress_arg, rl,
712 31ba2236 2022-01-05 thomas got_object_idset_num_elements(*loose_ids), -1, -1);
714 31ba2236 2022-01-05 thomas goto done;
717 b3d68e7f 2021-07-03 stsp if (closedir(dir) != 0) {
718 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
721 b3d68e7f 2021-07-03 stsp dir = NULL;
723 b3d68e7f 2021-07-03 stsp free(path);
724 b3d68e7f 2021-07-03 stsp path = NULL;
727 b3d68e7f 2021-07-03 stsp if (dir && closedir(dir) != 0 && err == NULL)
728 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
729 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
730 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
732 b3d68e7f 2021-07-03 stsp got_object_idset_free(*loose_ids);
733 b3d68e7f 2021-07-03 stsp *loose_ids = NULL;
736 b3d68e7f 2021-07-03 stsp got_object_close(obj);
737 b3d68e7f 2021-07-03 stsp free(path_objects);
738 b3d68e7f 2021-07-03 stsp free(path);
739 b3d68e7f 2021-07-03 stsp return err;
742 b3d68e7f 2021-07-03 stsp static const struct got_error *
743 b3d68e7f 2021-07-03 stsp preserve_loose_object(struct got_object_idset *loose_ids,
744 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo, int *npacked)
746 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
747 9614da0d 2021-07-04 stsp struct got_object *obj;
749 b3d68e7f 2021-07-03 stsp if (!got_object_idset_contains(loose_ids, id))
750 b3d68e7f 2021-07-03 stsp return NULL;
753 9614da0d 2021-07-04 stsp * Try to open this object from a pack file. This ensures that
754 9614da0d 2021-07-04 stsp * we do in fact have a valid packed copy of the object. Otherwise
755 9614da0d 2021-07-04 stsp * we should not delete the loose representation of this object.
757 abc59930 2021-09-05 naddy err = got_object_open_packed(&obj, id, repo);
758 abc59930 2021-09-05 naddy if (err == NULL) {
759 9614da0d 2021-07-04 stsp got_object_close(obj);
761 9614da0d 2021-07-04 stsp * The object is referenced and packed.
762 9614da0d 2021-07-04 stsp * We can purge the redundantly stored loose object.
764 9614da0d 2021-07-04 stsp (*npacked)++;
765 9614da0d 2021-07-04 stsp return NULL;
766 9614da0d 2021-07-04 stsp } else if (err->code != GOT_ERR_NO_OBJ)
767 9614da0d 2021-07-04 stsp return err;
770 b3d68e7f 2021-07-03 stsp * This object is referenced and not packed.
771 b3d68e7f 2021-07-03 stsp * Remove it from our purge set.
773 b3d68e7f 2021-07-03 stsp return got_object_idset_remove(NULL, loose_ids, id);
776 b3d68e7f 2021-07-03 stsp static const struct got_error *
777 b3d68e7f 2021-07-03 stsp load_tree_entries(struct got_object_id_queue *ids,
778 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids,
779 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
780 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
781 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
783 b3d68e7f 2021-07-03 stsp const struct got_error *err;
784 b3d68e7f 2021-07-03 stsp struct got_tree_object *tree;
785 b3d68e7f 2021-07-03 stsp char *p = NULL;
788 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
790 b3d68e7f 2021-07-03 stsp return err;
792 b3d68e7f 2021-07-03 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
793 b3d68e7f 2021-07-03 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
794 b3d68e7f 2021-07-03 stsp struct got_object_id *id = got_tree_entry_get_id(e);
795 b3d68e7f 2021-07-03 stsp mode_t mode = got_tree_entry_get_mode(e);
797 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
798 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
803 b3d68e7f 2021-07-03 stsp if (got_object_tree_entry_is_symlink(e) ||
804 b3d68e7f 2021-07-03 stsp got_object_tree_entry_is_submodule(e) ||
805 b3d68e7f 2021-07-03 stsp got_object_idset_contains(traversed_ids, id))
808 b3d68e7f 2021-07-03 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
809 b3d68e7f 2021-07-03 stsp got_tree_entry_get_name(e)) == -1) {
810 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
814 b3d68e7f 2021-07-03 stsp if (S_ISDIR(mode)) {
815 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
816 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
819 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
820 b3d68e7f 2021-07-03 stsp } else if (S_ISREG(mode)) {
821 b3d68e7f 2021-07-03 stsp /* This blob is referenced. */
822 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, id, repo,
826 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, id, NULL);
835 b3d68e7f 2021-07-03 stsp got_object_tree_close(tree);
837 b3d68e7f 2021-07-03 stsp return err;
840 b3d68e7f 2021-07-03 stsp static const struct got_error *
841 b3d68e7f 2021-07-03 stsp load_tree(struct got_object_idset *loose_ids,
842 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
843 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
844 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
846 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
847 b3d68e7f 2021-07-03 stsp struct got_object_id_queue tree_ids;
848 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
850 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, tree_id);
852 b3d68e7f 2021-07-03 stsp return err;
854 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&tree_ids);
855 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
857 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&tree_ids)) {
858 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
859 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
864 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&tree_ids);
865 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&tree_ids, entry);
867 ec242592 2022-04-22 thomas if (got_object_idset_contains(traversed_ids, &qid->id)) {
868 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
872 ec242592 2022-04-22 thomas err = got_object_idset_add(traversed_ids, &qid->id, NULL);
874 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
878 b3d68e7f 2021-07-03 stsp /* This tree is referenced. */
879 ec242592 2022-04-22 thomas err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
883 b3d68e7f 2021-07-03 stsp err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
884 ec242592 2022-04-22 thomas &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
885 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
890 b3d68e7f 2021-07-03 stsp got_object_id_queue_free(&tree_ids);
891 b3d68e7f 2021-07-03 stsp return err;
894 b3d68e7f 2021-07-03 stsp static const struct got_error *
895 b3d68e7f 2021-07-03 stsp load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
896 b3d68e7f 2021-07-03 stsp int *npacked, struct got_object_idset *traversed_ids,
897 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo,
898 31ba2236 2022-01-05 thomas got_cleanup_progress_cb progress_cb, void *progress_arg,
899 31ba2236 2022-01-05 thomas struct got_ratelimit *rl, int nloose,
900 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
902 b3d68e7f 2021-07-03 stsp const struct got_error *err;
903 b3d68e7f 2021-07-03 stsp struct got_commit_object *commit = NULL;
904 b3d68e7f 2021-07-03 stsp struct got_tag_object *tag = NULL;
905 b3d68e7f 2021-07-03 stsp struct got_object_id *tree_id = NULL;
906 b3d68e7f 2021-07-03 stsp struct got_object_id_queue ids;
907 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
908 b3d68e7f 2021-07-03 stsp int obj_type;
910 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
912 b3d68e7f 2021-07-03 stsp return err;
914 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&ids);
915 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&ids, qid, entry);
917 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&ids)) {
918 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
919 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
924 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&ids);
925 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&ids, entry);
927 ec242592 2022-04-22 thomas if (got_object_idset_contains(traversed_ids, &qid->id)) {
928 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
929 b3d68e7f 2021-07-03 stsp qid = NULL;
933 ec242592 2022-04-22 thomas err = got_object_idset_add(traversed_ids, &qid->id, NULL);
937 b3d68e7f 2021-07-03 stsp /* This commit or tag is referenced. */
938 ec242592 2022-04-22 thomas err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
942 ec242592 2022-04-22 thomas err = got_object_get_type(&obj_type, repo, &qid->id);
945 b3d68e7f 2021-07-03 stsp switch (obj_type) {
946 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
947 ec242592 2022-04-22 thomas err = got_object_open_as_commit(&commit, repo,
948 ec242592 2022-04-22 thomas &qid->id);
952 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
953 ec242592 2022-04-22 thomas err = got_object_open_as_tag(&tag, repo, &qid->id);
958 b3d68e7f 2021-07-03 stsp /* should not happen */
959 b3d68e7f 2021-07-03 stsp err = got_error(GOT_ERR_OBJ_TYPE);
963 b3d68e7f 2021-07-03 stsp /* Find a tree object to scan. */
964 b3d68e7f 2021-07-03 stsp if (commit) {
965 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
966 b3d68e7f 2021-07-03 stsp } else if (tag) {
967 b3d68e7f 2021-07-03 stsp obj_type = got_object_tag_get_object_type(tag);
968 b3d68e7f 2021-07-03 stsp switch (obj_type) {
969 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
970 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo,
971 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
974 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
976 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
977 b3d68e7f 2021-07-03 stsp tree_id = got_object_tag_get_object_id(tag);
981 b3d68e7f 2021-07-03 stsp * Tag points at something other than a
982 b3d68e7f 2021-07-03 stsp * commit or tree. Leave this weird tag object
983 b3d68e7f 2021-07-03 stsp * and the object it points to on disk.
985 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
986 ec242592 2022-04-22 thomas &qid->id);
987 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
989 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
990 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
991 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
993 b3d68e7f 2021-07-03 stsp err = NULL;
998 b3d68e7f 2021-07-03 stsp if (tree_id) {
999 b3d68e7f 2021-07-03 stsp err = load_tree(loose_ids, traversed_ids, tree_id, "",
1000 b3d68e7f 2021-07-03 stsp repo, npacked, cancel_cb, cancel_arg);
1005 b3d68e7f 2021-07-03 stsp if (commit || tag)
1006 b3d68e7f 2021-07-03 stsp (*ncommits)++; /* scanned tags are counted as commits */
1008 31ba2236 2022-01-05 thomas err = report_cleanup_progress(progress_cb, progress_arg, rl,
1009 31ba2236 2022-01-05 thomas nloose, *ncommits, -1);
1010 31ba2236 2022-01-05 thomas if (err)
1013 b3d68e7f 2021-07-03 stsp if (commit) {
1014 b3d68e7f 2021-07-03 stsp /* Find parent commits to scan. */
1015 b3d68e7f 2021-07-03 stsp const struct got_object_id_queue *parent_ids;
1016 b3d68e7f 2021-07-03 stsp parent_ids = got_object_commit_get_parent_ids(commit);
1017 b3d68e7f 2021-07-03 stsp err = got_object_id_queue_copy(parent_ids, &ids);
1020 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1021 b3d68e7f 2021-07-03 stsp commit = NULL;
1023 b3d68e7f 2021-07-03 stsp if (tag) {
1024 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1025 b3d68e7f 2021-07-03 stsp tag = NULL;
1027 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1028 b3d68e7f 2021-07-03 stsp qid = NULL;
1032 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1033 b3d68e7f 2021-07-03 stsp if (commit)
1034 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1036 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1037 4b2e47fb 2021-07-03 stsp got_object_id_queue_free(&ids);
1038 b3d68e7f 2021-07-03 stsp return err;
1041 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg {
1042 b3d68e7f 2021-07-03 stsp struct got_repository *repo;
1043 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb;
1044 b3d68e7f 2021-07-03 stsp void *progress_arg;
1045 31ba2236 2022-01-05 thomas struct got_ratelimit *rl;
1046 b3d68e7f 2021-07-03 stsp int nloose;
1047 b3d68e7f 2021-07-03 stsp int ncommits;
1048 b3d68e7f 2021-07-03 stsp int npurged;
1049 b3d68e7f 2021-07-03 stsp off_t size_purged;
1050 b3d68e7f 2021-07-03 stsp int dry_run;
1051 ef8ec606 2021-07-27 stsp time_t max_mtime;
1052 ef8ec606 2021-07-27 stsp int ignore_mtime;
1055 b3d68e7f 2021-07-03 stsp static const struct got_error *
1056 b3d68e7f 2021-07-03 stsp purge_loose_object(struct got_object_id *id, void *data, void *arg)
1058 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg *a = arg;
1059 b3d68e7f 2021-07-03 stsp const struct got_error *err, *unlock_err = NULL;
1060 b3d68e7f 2021-07-03 stsp char *path = NULL;
1061 b3d68e7f 2021-07-03 stsp int fd = -1;
1062 b3d68e7f 2021-07-03 stsp struct stat sb;
1063 b3d68e7f 2021-07-03 stsp struct got_lockfile *lf = NULL;
1065 b3d68e7f 2021-07-03 stsp err = got_object_get_path(&path, id, a->repo);
1067 b3d68e7f 2021-07-03 stsp return err;
1069 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, id, a->repo);
1071 b3d68e7f 2021-07-03 stsp goto done;
1073 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
1074 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
1075 b3d68e7f 2021-07-03 stsp goto done;
1079 ef8ec606 2021-07-27 stsp * Do not delete objects which are younger than our maximum
1080 ef8ec606 2021-07-27 stsp * modification time threshold. This prevents a race where
1081 ef8ec606 2021-07-27 stsp * new objects which are being added to the repository
1082 ef8ec606 2021-07-27 stsp * concurrently would be deleted.
1084 ef8ec606 2021-07-27 stsp if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1085 ef8ec606 2021-07-27 stsp if (!a->dry_run) {
1086 ef8ec606 2021-07-27 stsp err = got_lockfile_lock(&lf, path, -1);
1088 ef8ec606 2021-07-27 stsp goto done;
1089 ef8ec606 2021-07-27 stsp if (unlink(path) == -1) {
1090 ef8ec606 2021-07-27 stsp err = got_error_from_errno2("unlink", path);
1091 ef8ec606 2021-07-27 stsp goto done;
1095 ef8ec606 2021-07-27 stsp a->npurged++;
1096 ef8ec606 2021-07-27 stsp a->size_purged += sb.st_size;
1097 31ba2236 2022-01-05 thomas err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1098 31ba2236 2022-01-05 thomas a->rl, a->nloose, a->ncommits, a->npurged);
1099 31ba2236 2022-01-05 thomas if (err)
1100 31ba2236 2022-01-05 thomas goto done;
1103 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
1104 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
1105 b3d68e7f 2021-07-03 stsp free(path);
1107 5345b4c7 2021-07-06 stsp unlock_err = got_lockfile_unlock(lf, -1);
1108 b3d68e7f 2021-07-03 stsp return err ? err : unlock_err;
1111 b3d68e7f 2021-07-03 stsp const struct got_error *
1112 b3d68e7f 2021-07-03 stsp got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1113 b3d68e7f 2021-07-03 stsp off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1114 ef8ec606 2021-07-27 stsp int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1115 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1117 b3d68e7f 2021-07-03 stsp const struct got_error *err;
1118 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids;
1119 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids;
1120 b3d68e7f 2021-07-03 stsp struct got_object_id **referenced_ids;
1121 b3d68e7f 2021-07-03 stsp int i, nreferenced, nloose, ncommits = 0;
1122 b3d68e7f 2021-07-03 stsp struct got_reflist_head refs;
1123 ef8ec606 2021-07-27 stsp struct got_reflist_entry *re;
1124 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg arg;
1125 ef8ec606 2021-07-27 stsp time_t max_mtime = 0;
1126 31ba2236 2022-01-05 thomas struct got_ratelimit rl;
1128 b3d68e7f 2021-07-03 stsp TAILQ_INIT(&refs);
1129 31ba2236 2022-01-05 thomas got_ratelimit_init(&rl, 0, 500);
1131 b3d68e7f 2021-07-03 stsp *size_before = 0;
1132 b3d68e7f 2021-07-03 stsp *size_after = 0;
1133 b3d68e7f 2021-07-03 stsp *npacked = 0;
1135 b3d68e7f 2021-07-03 stsp err = get_loose_object_ids(&loose_ids, size_before,
1136 31ba2236 2022-01-05 thomas progress_cb, progress_arg, &rl, repo);
1138 b3d68e7f 2021-07-03 stsp return err;
1139 b3d68e7f 2021-07-03 stsp nloose = got_object_idset_num_elements(loose_ids);
1140 b3d68e7f 2021-07-03 stsp if (nloose == 0) {
1141 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1142 39a089ac 2022-01-05 thomas if (progress_cb) {
1143 39a089ac 2022-01-05 thomas err = progress_cb(progress_arg, 0, 0, 0);
1144 39a089ac 2022-01-05 thomas if (err)
1145 39a089ac 2022-01-05 thomas return err;
1147 b3d68e7f 2021-07-03 stsp return NULL;
1150 b3d68e7f 2021-07-03 stsp traversed_ids = got_object_idset_alloc();
1151 b3d68e7f 2021-07-03 stsp if (traversed_ids == NULL) {
1152 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_object_idset_alloc");
1153 b3d68e7f 2021-07-03 stsp goto done;
1156 b3d68e7f 2021-07-03 stsp err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1158 b3d68e7f 2021-07-03 stsp goto done;
1159 ef8ec606 2021-07-27 stsp if (!ignore_mtime) {
1160 ef8ec606 2021-07-27 stsp TAILQ_FOREACH(re, &refs, entry) {
1161 ef8ec606 2021-07-27 stsp time_t mtime = got_ref_get_mtime(re->ref);
1162 ef8ec606 2021-07-27 stsp if (mtime > max_mtime)
1163 ef8ec606 2021-07-27 stsp max_mtime = mtime;
1166 ef8ec606 2021-07-27 stsp * For safety, keep objects created within 10 minutes
1167 ef8ec606 2021-07-27 stsp * before the youngest reference was created.
1169 ef8ec606 2021-07-27 stsp if (max_mtime >= 600)
1170 ef8ec606 2021-07-27 stsp max_mtime -= 600;
1173 b3d68e7f 2021-07-03 stsp err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1174 b3d68e7f 2021-07-03 stsp (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1175 b3d68e7f 2021-07-03 stsp &refs, repo, cancel_cb, cancel_arg);
1177 b3d68e7f 2021-07-03 stsp goto done;
1179 b3d68e7f 2021-07-03 stsp for (i = 0; i < nreferenced; i++) {
1180 b3d68e7f 2021-07-03 stsp struct got_object_id *id = referenced_ids[i];
1181 b3d68e7f 2021-07-03 stsp err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1182 31ba2236 2022-01-05 thomas traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1183 31ba2236 2022-01-05 thomas nloose, cancel_cb, cancel_arg);
1185 b3d68e7f 2021-07-03 stsp goto done;
1188 b3d68e7f 2021-07-03 stsp /* Any remaining loose objects are unreferenced and can be purged. */
1189 b3d68e7f 2021-07-03 stsp arg.repo = repo;
1190 b3d68e7f 2021-07-03 stsp arg.progress_arg = progress_arg;
1191 b3d68e7f 2021-07-03 stsp arg.progress_cb = progress_cb;
1192 31ba2236 2022-01-05 thomas arg.rl = &rl;
1193 b3d68e7f 2021-07-03 stsp arg.nloose = nloose;
1194 b3d68e7f 2021-07-03 stsp arg.npurged = 0;
1195 b3d68e7f 2021-07-03 stsp arg.size_purged = 0;
1196 b3d68e7f 2021-07-03 stsp arg.ncommits = ncommits;
1197 b3d68e7f 2021-07-03 stsp arg.dry_run = dry_run;
1198 ef8ec606 2021-07-27 stsp arg.max_mtime = max_mtime;
1199 ef8ec606 2021-07-27 stsp arg.ignore_mtime = ignore_mtime;
1200 b3d68e7f 2021-07-03 stsp err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1202 b3d68e7f 2021-07-03 stsp goto done;
1203 b3d68e7f 2021-07-03 stsp *size_after = *size_before - arg.size_purged;
1205 31ba2236 2022-01-05 thomas /* Produce a final progress report. */
1206 31ba2236 2022-01-05 thomas if (progress_cb) {
1207 654306ef 2022-01-05 thomas err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1208 31ba2236 2022-01-05 thomas if (err)
1209 31ba2236 2022-01-05 thomas goto done;
1212 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1213 b3d68e7f 2021-07-03 stsp got_object_idset_free(traversed_ids);
1214 1124fe40 2021-07-07 stsp return err;
1217 1124fe40 2021-07-07 stsp static const struct got_error *
1218 1124fe40 2021-07-07 stsp remove_packidx(int dir_fd, const char *relpath)
1220 1124fe40 2021-07-07 stsp const struct got_error *err, *unlock_err;
1221 1124fe40 2021-07-07 stsp struct got_lockfile *lf;
1223 1124fe40 2021-07-07 stsp err = got_lockfile_lock(&lf, relpath, dir_fd);
1225 1124fe40 2021-07-07 stsp return err;
1226 1124fe40 2021-07-07 stsp if (unlinkat(dir_fd, relpath, 0) == -1)
1227 1124fe40 2021-07-07 stsp err = got_error_from_errno("unlinkat");
1228 1124fe40 2021-07-07 stsp unlock_err = got_lockfile_unlock(lf, dir_fd);
1229 1124fe40 2021-07-07 stsp return err ? err : unlock_err;
1232 1124fe40 2021-07-07 stsp const struct got_error *
1233 1124fe40 2021-07-07 stsp got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1234 1124fe40 2021-07-07 stsp got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1235 1124fe40 2021-07-07 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1237 1124fe40 2021-07-07 stsp const struct got_error *err;
1238 1124fe40 2021-07-07 stsp DIR *packdir = NULL;
1239 1124fe40 2021-07-07 stsp struct dirent *dent;
1240 1124fe40 2021-07-07 stsp char *pack_relpath = NULL;
1241 1124fe40 2021-07-07 stsp int packdir_fd;
1242 1124fe40 2021-07-07 stsp struct stat sb;
1244 1124fe40 2021-07-07 stsp packdir_fd = openat(got_repo_get_fd(repo),
1245 fc63f50d 2021-12-31 thomas GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1246 1124fe40 2021-07-07 stsp if (packdir_fd == -1) {
1247 1124fe40 2021-07-07 stsp if (errno == ENOENT)
1248 1124fe40 2021-07-07 stsp return NULL;
1249 1124fe40 2021-07-07 stsp return got_error_from_errno_fmt("openat: %s/%s",
1250 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1251 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR);
1254 1124fe40 2021-07-07 stsp packdir = fdopendir(packdir_fd);
1255 1124fe40 2021-07-07 stsp if (packdir == NULL) {
1256 1124fe40 2021-07-07 stsp err = got_error_from_errno("fdopendir");
1257 1124fe40 2021-07-07 stsp goto done;
1260 1124fe40 2021-07-07 stsp while ((dent = readdir(packdir)) != NULL) {
1261 1124fe40 2021-07-07 stsp if (cancel_cb) {
1262 1124fe40 2021-07-07 stsp err = cancel_cb(cancel_arg);
1264 1124fe40 2021-07-07 stsp goto done;
1267 dd038bc6 2021-09-21 thomas.ad if (!got_repo_is_packidx_filename(dent->d_name,
1268 dd038bc6 2021-09-21 thomas.ad strlen(dent->d_name)))
1271 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath,
1272 1124fe40 2021-07-07 stsp dent->d_name);
1274 1124fe40 2021-07-07 stsp goto done;
1276 1124fe40 2021-07-07 stsp if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1277 1124fe40 2021-07-07 stsp free(pack_relpath);
1278 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1281 1124fe40 2021-07-07 stsp if (errno != ENOENT) {
1282 1124fe40 2021-07-07 stsp err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1283 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1284 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1285 1124fe40 2021-07-07 stsp pack_relpath);
1286 1124fe40 2021-07-07 stsp goto done;
1289 1124fe40 2021-07-07 stsp if (!dry_run) {
1290 1124fe40 2021-07-07 stsp err = remove_packidx(packdir_fd, dent->d_name);
1292 1124fe40 2021-07-07 stsp goto done;
1294 1124fe40 2021-07-07 stsp if (progress_cb) {
1295 1124fe40 2021-07-07 stsp char *path;
1296 1124fe40 2021-07-07 stsp if (asprintf(&path, "%s/%s/%s",
1297 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1298 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1299 1124fe40 2021-07-07 stsp dent->d_name) == -1) {
1300 1124fe40 2021-07-07 stsp err = got_error_from_errno("asprintf");
1301 1124fe40 2021-07-07 stsp goto done;
1303 1124fe40 2021-07-07 stsp err = progress_cb(progress_arg, path);
1304 1124fe40 2021-07-07 stsp free(path);
1306 1124fe40 2021-07-07 stsp goto done;
1308 1124fe40 2021-07-07 stsp free(pack_relpath);
1309 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1312 1124fe40 2021-07-07 stsp if (packdir && closedir(packdir) != 0 && err == NULL)
1313 1124fe40 2021-07-07 stsp err = got_error_from_errno("closedir");
1314 1124fe40 2021-07-07 stsp free(pack_relpath);
1315 b3d68e7f 2021-07-03 stsp return err;
