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/queue.h>
19 f8b19efd 2021-10-13 stsp #include <sys/tree.h>
20 05118f5a 2021-06-22 stsp #include <sys/uio.h>
21 05118f5a 2021-06-22 stsp #include <sys/stat.h>
22 05118f5a 2021-06-22 stsp #include <sys/socket.h>
23 05118f5a 2021-06-22 stsp #include <sys/wait.h>
25 05118f5a 2021-06-22 stsp #include <dirent.h>
26 c8d1f14f 2021-06-23 naddy #include <endian.h>
27 05118f5a 2021-06-22 stsp #include <errno.h>
28 05118f5a 2021-06-22 stsp #include <fcntl.h>
29 05118f5a 2021-06-22 stsp #include <stdint.h>
30 05118f5a 2021-06-22 stsp #include <sha1.h>
31 5822e79e 2023-02-23 op #include <sha2.h>
32 05118f5a 2021-06-22 stsp #include <stdio.h>
33 05118f5a 2021-06-22 stsp #include <stdlib.h>
34 05118f5a 2021-06-22 stsp #include <string.h>
35 05118f5a 2021-06-22 stsp #include <limits.h>
36 05118f5a 2021-06-22 stsp #include <unistd.h>
37 05118f5a 2021-06-22 stsp #include <imsg.h>
39 05118f5a 2021-06-22 stsp #include "got_error.h"
40 05118f5a 2021-06-22 stsp #include "got_cancel.h"
41 05118f5a 2021-06-22 stsp #include "got_object.h"
42 05118f5a 2021-06-22 stsp #include "got_reference.h"
43 05118f5a 2021-06-22 stsp #include "got_repository.h"
44 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
45 05118f5a 2021-06-22 stsp #include "got_opentemp.h"
46 05118f5a 2021-06-22 stsp #include "got_path.h"
48 05118f5a 2021-06-22 stsp #include "got_lib_delta.h"
49 05118f5a 2021-06-22 stsp #include "got_lib_object.h"
50 b3d68e7f 2021-07-03 stsp #include "got_lib_object_idset.h"
51 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
52 05118f5a 2021-06-22 stsp #include "got_lib_pack.h"
53 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
54 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
55 cae60ab8 2022-10-18 stsp #include "got_lib_ratelimit.h"
56 05118f5a 2021-06-22 stsp #include "got_lib_pack_create.h"
57 53bf0b54 2023-02-23 op #include "got_lib_hash.h"
58 b3d68e7f 2021-07-03 stsp #include "got_lib_lockfile.h"
60 05118f5a 2021-06-22 stsp #ifndef nitems
61 05118f5a 2021-06-22 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
64 05118f5a 2021-06-22 stsp static const struct got_error *
65 05118f5a 2021-06-22 stsp get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
66 05118f5a 2021-06-22 stsp unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
67 05118f5a 2021-06-22 stsp struct got_repository *repo,
68 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
70 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
71 05118f5a 2021-06-22 stsp const size_t alloc_chunksz = 256;
72 05118f5a 2021-06-22 stsp size_t nalloc;
73 05118f5a 2021-06-22 stsp struct got_reflist_entry *re;
76 05118f5a 2021-06-22 stsp *ids = NULL;
77 05118f5a 2021-06-22 stsp *nobjects = 0;
79 7e4f461f 2022-04-13 stsp err = got_reflist_sort(refs,
80 7e4f461f 2022-04-13 stsp got_ref_cmp_by_commit_timestamp_descending, repo);
84 05118f5a 2021-06-22 stsp *ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
85 05118f5a 2021-06-22 stsp if (*ids == NULL)
86 05118f5a 2021-06-22 stsp return got_error_from_errno("reallocarray");
87 05118f5a 2021-06-22 stsp nalloc = alloc_chunksz;
89 05118f5a 2021-06-22 stsp TAILQ_FOREACH(re, refs, entry) {
90 05118f5a 2021-06-22 stsp struct got_object_id *id;
92 05118f5a 2021-06-22 stsp if (cancel_cb) {
93 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
98 05118f5a 2021-06-22 stsp err = got_ref_resolve(&id, repo, re->ref);
102 05118f5a 2021-06-22 stsp if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
103 05118f5a 2021-06-22 stsp int obj_type;
104 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
107 05118f5a 2021-06-22 stsp if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
114 ae23ce34 2021-07-01 stsp if (nalloc <= *nobjects) {
115 05118f5a 2021-06-22 stsp struct got_object_id **new;
116 05118f5a 2021-06-22 stsp new = recallocarray(*ids, nalloc,
117 05118f5a 2021-06-22 stsp nalloc + alloc_chunksz,
118 05118f5a 2021-06-22 stsp sizeof(struct got_object_id *));
119 05118f5a 2021-06-22 stsp if (new == NULL) {
120 05118f5a 2021-06-22 stsp err = got_error_from_errno(
121 05118f5a 2021-06-22 stsp "recallocarray");
124 05118f5a 2021-06-22 stsp *ids = new;
125 05118f5a 2021-06-22 stsp nalloc += alloc_chunksz;
127 05118f5a 2021-06-22 stsp (*ids)[*nobjects] = id;
128 05118f5a 2021-06-22 stsp if ((*ids)[*nobjects] == NULL) {
129 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
132 05118f5a 2021-06-22 stsp (*nobjects)++;
136 05118f5a 2021-06-22 stsp for (i = 0; i < *nobjects; i++)
137 05118f5a 2021-06-22 stsp free((*ids)[i]);
138 05118f5a 2021-06-22 stsp free(*ids);
139 05118f5a 2021-06-22 stsp *ids = NULL;
140 05118f5a 2021-06-22 stsp *nobjects = 0;
142 05118f5a 2021-06-22 stsp return err;
145 05118f5a 2021-06-22 stsp const struct got_error *
146 05118f5a 2021-06-22 stsp got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
147 05118f5a 2021-06-22 stsp struct got_reflist_head *include_refs,
148 05118f5a 2021-06-22 stsp struct got_reflist_head *exclude_refs, struct got_repository *repo,
149 c7a4fcc8 2023-02-17 op int loose_obj_only, int force_refdelta,
150 67fd6849 2022-02-13 stsp got_pack_progress_cb progress_cb, void *progress_arg,
151 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
153 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
154 05118f5a 2021-06-22 stsp struct got_object_id **ours = NULL, **theirs = NULL;
155 05118f5a 2021-06-22 stsp int nours = 0, ntheirs = 0, packfd = -1, i;
156 05118f5a 2021-06-22 stsp char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
157 05118f5a 2021-06-22 stsp char *sha1_str = NULL;
158 a32780aa 2022-10-15 stsp FILE *delta_cache = NULL;
159 cae60ab8 2022-10-18 stsp struct got_ratelimit rl;
161 05118f5a 2021-06-22 stsp *packfile = NULL;
162 05118f5a 2021-06-22 stsp *pack_hash = NULL;
164 cae60ab8 2022-10-18 stsp got_ratelimit_init(&rl, 0, 500);
166 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/packing.pack",
167 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
168 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
171 b90054ed 2022-11-01 stsp err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
175 839bbaae 2023-02-01 op if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
176 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpfile_path);
180 a32780aa 2022-10-15 stsp delta_cache = got_opentemp();
181 a32780aa 2022-10-15 stsp if (delta_cache == NULL) {
182 a32780aa 2022-10-15 stsp err = got_error_from_errno("got_opentemp");
186 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&ours, &nours,
187 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
188 abc59930 2021-09-05 naddy include_refs, repo, cancel_cb, cancel_arg);
192 05118f5a 2021-06-22 stsp if (nours == 0) {
193 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
197 05118f5a 2021-06-22 stsp if (!TAILQ_EMPTY(exclude_refs)) {
198 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&theirs, &ntheirs,
199 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
200 abc59930 2021-09-05 naddy exclude_refs, repo,
201 abc59930 2021-09-05 naddy cancel_cb, cancel_arg);
206 05118f5a 2021-06-22 stsp *pack_hash = calloc(1, sizeof(**pack_hash));
207 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
208 05118f5a 2021-06-22 stsp err = got_error_from_errno("calloc");
212 a32780aa 2022-10-15 stsp err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
213 c7a4fcc8 2023-02-17 op theirs, ntheirs, ours, nours, repo, loose_obj_only,
214 c7a4fcc8 2023-02-17 op 0, force_refdelta, progress_cb, progress_arg, &rl,
215 c7a4fcc8 2023-02-17 op cancel_cb, cancel_arg);
219 05118f5a 2021-06-22 stsp err = got_object_id_str(&sha1_str, *pack_hash);
222 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
223 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
224 05118f5a 2021-06-22 stsp sha1_str) == -1) {
225 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
229 894e4711 2022-10-15 stsp if (lseek(packfd, 0L, SEEK_SET) == -1) {
230 894e4711 2022-10-15 stsp err = got_error_from_errno("lseek");
233 05118f5a 2021-06-22 stsp if (rename(tmpfile_path, packfile_path) == -1) {
234 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpfile_path,
235 05118f5a 2021-06-22 stsp packfile_path);
238 05118f5a 2021-06-22 stsp free(tmpfile_path);
239 05118f5a 2021-06-22 stsp tmpfile_path = NULL;
241 894e4711 2022-10-15 stsp *packfile = fdopen(packfd, "w");
242 894e4711 2022-10-15 stsp if (*packfile == NULL) {
243 894e4711 2022-10-15 stsp err = got_error_from_errno2("fdopen", tmpfile_path);
246 894e4711 2022-10-15 stsp packfd = -1;
248 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++)
249 05118f5a 2021-06-22 stsp free(ours[i]);
250 05118f5a 2021-06-22 stsp free(ours);
251 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++)
252 05118f5a 2021-06-22 stsp free(theirs[i]);
253 05118f5a 2021-06-22 stsp free(theirs);
254 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
255 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
256 a32780aa 2022-10-15 stsp if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
257 a32780aa 2022-10-15 stsp err = got_error_from_errno("fclose");
258 05118f5a 2021-06-22 stsp if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
259 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpfile_path);
260 05118f5a 2021-06-22 stsp free(tmpfile_path);
261 05118f5a 2021-06-22 stsp free(packfile_path);
262 05118f5a 2021-06-22 stsp free(sha1_str);
263 05118f5a 2021-06-22 stsp free(path);
265 05118f5a 2021-06-22 stsp free(*pack_hash);
266 05118f5a 2021-06-22 stsp *pack_hash = NULL;
267 05118f5a 2021-06-22 stsp if (*packfile)
268 05118f5a 2021-06-22 stsp fclose(*packfile);
269 05118f5a 2021-06-22 stsp *packfile = NULL;
271 05118f5a 2021-06-22 stsp return err;
274 05118f5a 2021-06-22 stsp const struct got_error *
275 05118f5a 2021-06-22 stsp got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
276 05118f5a 2021-06-22 stsp struct got_repository *repo,
277 05118f5a 2021-06-22 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
278 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
281 05118f5a 2021-06-22 stsp char *path;
282 05118f5a 2021-06-22 stsp int imsg_idxfds[2];
283 05118f5a 2021-06-22 stsp int npackfd = -1, idxfd = -1, nidxfd = -1;
284 05118f5a 2021-06-22 stsp int tmpfds[3];
285 05118f5a 2021-06-22 stsp int idxstatus, done = 0;
286 05118f5a 2021-06-22 stsp const struct got_error *err;
287 05118f5a 2021-06-22 stsp struct imsgbuf idxibuf;
288 05118f5a 2021-06-22 stsp pid_t idxpid;
289 05118f5a 2021-06-22 stsp char *tmpidxpath = NULL;
290 05118f5a 2021-06-22 stsp char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
291 05118f5a 2021-06-22 stsp const char *repo_path = got_repo_get_path_git_dir(repo);
292 05118f5a 2021-06-22 stsp struct stat sb;
294 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++)
295 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
297 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/indexing.idx",
298 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
299 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
302 b90054ed 2022-11-01 stsp err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
303 05118f5a 2021-06-22 stsp free(path);
306 839bbaae 2023-02-01 op if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
307 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpidxpath);
311 05118f5a 2021-06-22 stsp nidxfd = dup(idxfd);
312 05118f5a 2021-06-22 stsp if (nidxfd == -1) {
313 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
317 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
318 05118f5a 2021-06-22 stsp tmpfds[i] = got_opentempfd();
319 05118f5a 2021-06-22 stsp if (tmpfds[i] == -1) {
320 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_opentempfd");
325 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
329 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
330 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
331 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
335 05118f5a 2021-06-22 stsp if (fstat(fileno(packfile), &sb) == -1) {
336 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fstat", packfile_path);
340 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
341 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
342 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
346 05118f5a 2021-06-22 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
347 05118f5a 2021-06-22 stsp err = got_error_from_errno("socketpair");
350 05118f5a 2021-06-22 stsp idxpid = fork();
351 05118f5a 2021-06-22 stsp if (idxpid == -1) {
352 05118f5a 2021-06-22 stsp err= got_error_from_errno("fork");
354 05118f5a 2021-06-22 stsp } else if (idxpid == 0)
355 05118f5a 2021-06-22 stsp got_privsep_exec_child(imsg_idxfds,
356 05118f5a 2021-06-22 stsp GOT_PATH_PROG_INDEX_PACK, packfile_path);
357 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[1]) == -1) {
358 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
361 05118f5a 2021-06-22 stsp imsg_init(&idxibuf, imsg_idxfds[0]);
363 05118f5a 2021-06-22 stsp npackfd = dup(fileno(packfile));
364 05118f5a 2021-06-22 stsp if (npackfd == -1) {
365 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
368 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
370 05118f5a 2021-06-22 stsp if (err != NULL)
372 05118f5a 2021-06-22 stsp npackfd = -1;
373 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
374 05118f5a 2021-06-22 stsp if (err != NULL)
376 05118f5a 2021-06-22 stsp nidxfd = -1;
377 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
378 05118f5a 2021-06-22 stsp err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
379 05118f5a 2021-06-22 stsp if (err != NULL)
381 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
384 05118f5a 2021-06-22 stsp while (!done) {
385 05118f5a 2021-06-22 stsp int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
387 05118f5a 2021-06-22 stsp if (cancel_cb) {
388 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
393 05118f5a 2021-06-22 stsp err = got_privsep_recv_index_progress(&done, &nobj_total,
394 05118f5a 2021-06-22 stsp &nobj_indexed, &nobj_loose, &nobj_resolved,
396 05118f5a 2021-06-22 stsp if (err != NULL)
398 05118f5a 2021-06-22 stsp if (nobj_indexed != 0) {
399 05118f5a 2021-06-22 stsp err = progress_cb(progress_arg, sb.st_size,
400 05118f5a 2021-06-22 stsp nobj_total, nobj_indexed, nobj_loose,
401 05118f5a 2021-06-22 stsp nobj_resolved);
406 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[0]) == -1) {
407 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
410 05118f5a 2021-06-22 stsp if (waitpid(idxpid, &idxstatus, 0) == -1) {
411 05118f5a 2021-06-22 stsp err = got_error_from_errno("waitpid");
415 05118f5a 2021-06-22 stsp if (rename(tmpidxpath, idxpath) == -1) {
416 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpidxpath, idxpath);
419 05118f5a 2021-06-22 stsp free(tmpidxpath);
420 05118f5a 2021-06-22 stsp tmpidxpath = NULL;
423 05118f5a 2021-06-22 stsp if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
424 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpidxpath);
425 05118f5a 2021-06-22 stsp if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
426 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
427 05118f5a 2021-06-22 stsp if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
428 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
429 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
430 05118f5a 2021-06-22 stsp if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
431 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
433 05118f5a 2021-06-22 stsp free(tmpidxpath);
434 05118f5a 2021-06-22 stsp free(idxpath);
435 05118f5a 2021-06-22 stsp free(packfile_path);
436 05118f5a 2021-06-22 stsp return err;
439 05118f5a 2021-06-22 stsp const struct got_error *
440 05118f5a 2021-06-22 stsp got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
441 05118f5a 2021-06-22 stsp struct got_repository *repo, const char *packfile_path)
443 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
444 05118f5a 2021-06-22 stsp const char *packdir_path = NULL;
445 05118f5a 2021-06-22 stsp char *packfile_name = NULL, *p, *dot;
446 05118f5a 2021-06-22 stsp struct got_object_id id;
447 05118f5a 2021-06-22 stsp int packfd = -1;
449 05118f5a 2021-06-22 stsp *packfile = NULL;
450 05118f5a 2021-06-22 stsp *pack_hash = NULL;
452 05118f5a 2021-06-22 stsp packdir_path = got_repo_get_path_objects_pack(repo);
453 05118f5a 2021-06-22 stsp if (packdir_path == NULL)
454 05118f5a 2021-06-22 stsp return got_error_from_errno("got_repo_get_path_objects_pack");
456 05118f5a 2021-06-22 stsp if (!got_path_is_child(packfile_path, packdir_path,
457 05118f5a 2021-06-22 stsp strlen(packdir_path))) {
458 05118f5a 2021-06-22 stsp err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
463 05118f5a 2021-06-22 stsp err = got_path_basename(&packfile_name, packfile_path);
466 05118f5a 2021-06-22 stsp p = packfile_name;
468 05118f5a 2021-06-22 stsp if (strncmp(p, "pack-", 5) != 0) {
469 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
470 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
471 abc59930 2021-09-05 naddy packfile_name);
475 05118f5a 2021-06-22 stsp dot = strchr(p, '.');
476 05118f5a 2021-06-22 stsp if (dot == NULL) {
477 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
478 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
479 abc59930 2021-09-05 naddy packfile_name);
482 05118f5a 2021-06-22 stsp if (strcmp(dot + 1, "pack") != 0) {
483 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
484 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
485 abc59930 2021-09-05 naddy packfile_name);
488 05118f5a 2021-06-22 stsp *dot = '\0';
489 b54930d5 2023-02-28 op if (!got_parse_object_id(&id, p, repo->algo)) {
490 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
491 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
492 abc59930 2021-09-05 naddy packfile_name);
496 05118f5a 2021-06-22 stsp *pack_hash = got_object_id_dup(&id);
497 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
498 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
502 8bd0cdad 2021-12-31 stsp packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
503 05118f5a 2021-06-22 stsp if (packfd == -1) {
504 05118f5a 2021-06-22 stsp err = got_error_from_errno2("open", packfile_path);
508 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "r");
509 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
510 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", packfile_path);
513 05118f5a 2021-06-22 stsp packfd = -1;
515 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
516 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
517 05118f5a 2021-06-22 stsp free(packfile_name);
519 05118f5a 2021-06-22 stsp free(*pack_hash);
520 05118f5a 2021-06-22 stsp *pack_hash = NULL;
522 05118f5a 2021-06-22 stsp return err;
525 05118f5a 2021-06-22 stsp const struct got_error *
526 05118f5a 2021-06-22 stsp got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
527 05118f5a 2021-06-22 stsp struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
528 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
530 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
531 05118f5a 2021-06-22 stsp char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
532 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
533 05118f5a 2021-06-22 stsp struct got_pack *pack = NULL;
534 05118f5a 2021-06-22 stsp uint32_t nobj, i;
536 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
540 05118f5a 2021-06-22 stsp if (asprintf(&packpath, "%s/pack-%s.pack",
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");
545 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/pack-%s.idx",
546 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
547 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
551 05118f5a 2021-06-22 stsp err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
555 05118f5a 2021-06-22 stsp err = got_repo_cache_pack(&pack, repo, packpath, packidx);
559 05118f5a 2021-06-22 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
560 05118f5a 2021-06-22 stsp for (i = 0; i < nobj; i++) {
561 05118f5a 2021-06-22 stsp struct got_packidx_object_id *oid;
562 05118f5a 2021-06-22 stsp struct got_object_id id, base_id;
563 05118f5a 2021-06-22 stsp off_t offset, base_offset = 0;
564 05118f5a 2021-06-22 stsp uint8_t type;
565 05118f5a 2021-06-22 stsp uint64_t size;
566 05118f5a 2021-06-22 stsp size_t tslen, len;
568 05118f5a 2021-06-22 stsp if (cancel_cb) {
569 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
573 05118f5a 2021-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
574 05118f5a 2021-06-22 stsp memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
576 05118f5a 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, i);
577 05118f5a 2021-06-22 stsp if (offset == -1) {
578 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
582 05118f5a 2021-06-22 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
583 05118f5a 2021-06-22 stsp pack, offset);
587 05118f5a 2021-06-22 stsp switch (type) {
588 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
589 05118f5a 2021-06-22 stsp err = got_pack_parse_offset_delta(&base_offset, &len,
590 05118f5a 2021-06-22 stsp pack, offset, tslen);
594 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_REF_DELTA:
595 05118f5a 2021-06-22 stsp err = got_pack_parse_ref_delta(&base_id,
596 05118f5a 2021-06-22 stsp pack, offset, tslen);
601 05118f5a 2021-06-22 stsp err = (*list_cb)(list_arg, &id, type, offset, size,
602 05118f5a 2021-06-22 stsp base_offset, &base_id);
608 05118f5a 2021-06-22 stsp free(id_str);
609 05118f5a 2021-06-22 stsp free(idxpath);
610 05118f5a 2021-06-22 stsp free(packpath);
611 05118f5a 2021-06-22 stsp if (packidx)
612 05118f5a 2021-06-22 stsp got_packidx_close(packidx);
613 b3d68e7f 2021-07-03 stsp return err;
616 b3d68e7f 2021-07-03 stsp static const struct got_error *
617 211cfef0 2022-01-05 stsp report_cleanup_progress(got_cleanup_progress_cb progress_cb,
618 211cfef0 2022-01-05 stsp void *progress_arg, struct got_ratelimit *rl,
619 211cfef0 2022-01-05 stsp int nloose, int ncommits, int npurged)
621 211cfef0 2022-01-05 stsp const struct got_error *err;
622 211cfef0 2022-01-05 stsp int elapsed;
624 211cfef0 2022-01-05 stsp if (progress_cb == NULL)
625 211cfef0 2022-01-05 stsp return NULL;
627 211cfef0 2022-01-05 stsp err = got_ratelimit_check(&elapsed, rl);
628 211cfef0 2022-01-05 stsp if (err || !elapsed)
629 211cfef0 2022-01-05 stsp return err;
631 211cfef0 2022-01-05 stsp return progress_cb(progress_arg, nloose, ncommits, npurged);
634 211cfef0 2022-01-05 stsp static const struct got_error *
635 b3d68e7f 2021-07-03 stsp get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
636 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
637 211cfef0 2022-01-05 stsp struct got_ratelimit *rl, struct got_repository *repo)
639 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
640 b3d68e7f 2021-07-03 stsp char *path_objects = NULL, *path = NULL;
641 b3d68e7f 2021-07-03 stsp DIR *dir = NULL;
642 b3d68e7f 2021-07-03 stsp struct got_object *obj = NULL;
643 b3d68e7f 2021-07-03 stsp struct got_object_id id;
644 b3d68e7f 2021-07-03 stsp int i, fd = -1;
645 b3d68e7f 2021-07-03 stsp struct stat sb;
647 b3d68e7f 2021-07-03 stsp *ondisk_size = 0;
648 b3d68e7f 2021-07-03 stsp *loose_ids = got_object_idset_alloc();
649 b3d68e7f 2021-07-03 stsp if (*loose_ids == NULL)
650 b3d68e7f 2021-07-03 stsp return got_error_from_errno("got_object_idset_alloc");
652 b3d68e7f 2021-07-03 stsp path_objects = got_repo_get_path_objects(repo);
653 b3d68e7f 2021-07-03 stsp if (path_objects == NULL) {
654 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_repo_get_path_objects");
658 b3d68e7f 2021-07-03 stsp for (i = 0; i <= 0xff; i++) {
659 b3d68e7f 2021-07-03 stsp struct dirent *dent;
661 b3d68e7f 2021-07-03 stsp if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
662 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
666 b3d68e7f 2021-07-03 stsp dir = opendir(path);
667 b3d68e7f 2021-07-03 stsp if (dir == NULL) {
668 b3d68e7f 2021-07-03 stsp if (errno == ENOENT) {
669 b3d68e7f 2021-07-03 stsp err = NULL;
672 b3d68e7f 2021-07-03 stsp err = got_error_from_errno2("opendir", path);
676 b3d68e7f 2021-07-03 stsp while ((dent = readdir(dir)) != NULL) {
677 b3d68e7f 2021-07-03 stsp char *id_str;
679 b3d68e7f 2021-07-03 stsp if (strcmp(dent->d_name, ".") == 0 ||
680 b3d68e7f 2021-07-03 stsp strcmp(dent->d_name, "..") == 0)
683 b3d68e7f 2021-07-03 stsp if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
684 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
688 b54930d5 2023-02-28 op if (!got_parse_object_id(&id, id_str, repo->algo)) {
689 b3d68e7f 2021-07-03 stsp free(id_str);
692 b3d68e7f 2021-07-03 stsp free(id_str);
694 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, &id, repo);
697 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
698 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
701 d5c81d44 2021-07-08 stsp err = got_object_read_header_privsep(&obj, &id, repo,
705 b3d68e7f 2021-07-03 stsp fd = -1; /* already closed */
707 b3d68e7f 2021-07-03 stsp switch (obj->type) {
708 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
709 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
710 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_BLOB:
711 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
714 b3d68e7f 2021-07-03 stsp err = got_error_fmt(GOT_ERR_OBJ_TYPE,
715 b3d68e7f 2021-07-03 stsp "%d", obj->type);
718 b3d68e7f 2021-07-03 stsp got_object_close(obj);
719 b3d68e7f 2021-07-03 stsp obj = NULL;
720 b3d68e7f 2021-07-03 stsp (*ondisk_size) += sb.st_size;
721 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(*loose_ids, &id, NULL);
724 211cfef0 2022-01-05 stsp err = report_cleanup_progress(progress_cb,
725 211cfef0 2022-01-05 stsp progress_arg, rl,
726 211cfef0 2022-01-05 stsp got_object_idset_num_elements(*loose_ids), -1, -1);
731 b3d68e7f 2021-07-03 stsp if (closedir(dir) != 0) {
732 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
735 b3d68e7f 2021-07-03 stsp dir = NULL;
737 b3d68e7f 2021-07-03 stsp free(path);
738 b3d68e7f 2021-07-03 stsp path = NULL;
741 b3d68e7f 2021-07-03 stsp if (dir && closedir(dir) != 0 && err == NULL)
742 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
743 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
744 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
746 b3d68e7f 2021-07-03 stsp got_object_idset_free(*loose_ids);
747 b3d68e7f 2021-07-03 stsp *loose_ids = NULL;
750 b3d68e7f 2021-07-03 stsp got_object_close(obj);
751 b3d68e7f 2021-07-03 stsp free(path_objects);
752 b3d68e7f 2021-07-03 stsp free(path);
753 b3d68e7f 2021-07-03 stsp return err;
756 b3d68e7f 2021-07-03 stsp static const struct got_error *
757 b3d68e7f 2021-07-03 stsp preserve_loose_object(struct got_object_idset *loose_ids,
758 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo, int *npacked)
760 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
761 9614da0d 2021-07-04 stsp struct got_object *obj;
763 b3d68e7f 2021-07-03 stsp if (!got_object_idset_contains(loose_ids, id))
764 b3d68e7f 2021-07-03 stsp return NULL;
767 9614da0d 2021-07-04 stsp * Try to open this object from a pack file. This ensures that
768 9614da0d 2021-07-04 stsp * we do in fact have a valid packed copy of the object. Otherwise
769 9614da0d 2021-07-04 stsp * we should not delete the loose representation of this object.
771 abc59930 2021-09-05 naddy err = got_object_open_packed(&obj, id, repo);
772 abc59930 2021-09-05 naddy if (err == NULL) {
773 9614da0d 2021-07-04 stsp got_object_close(obj);
775 9614da0d 2021-07-04 stsp * The object is referenced and packed.
776 9614da0d 2021-07-04 stsp * We can purge the redundantly stored loose object.
778 9614da0d 2021-07-04 stsp (*npacked)++;
779 9614da0d 2021-07-04 stsp return NULL;
780 9614da0d 2021-07-04 stsp } else if (err->code != GOT_ERR_NO_OBJ)
781 9614da0d 2021-07-04 stsp return err;
784 b3d68e7f 2021-07-03 stsp * This object is referenced and not packed.
785 b3d68e7f 2021-07-03 stsp * Remove it from our purge set.
787 b3d68e7f 2021-07-03 stsp return got_object_idset_remove(NULL, loose_ids, id);
790 b3d68e7f 2021-07-03 stsp static const struct got_error *
791 b3d68e7f 2021-07-03 stsp load_tree_entries(struct got_object_id_queue *ids,
792 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids,
793 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
794 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
795 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
797 b3d68e7f 2021-07-03 stsp const struct got_error *err;
798 b3d68e7f 2021-07-03 stsp struct got_tree_object *tree;
799 b3d68e7f 2021-07-03 stsp char *p = NULL;
802 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
804 b3d68e7f 2021-07-03 stsp return err;
806 b3d68e7f 2021-07-03 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
807 b3d68e7f 2021-07-03 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
808 b3d68e7f 2021-07-03 stsp struct got_object_id *id = got_tree_entry_get_id(e);
809 b3d68e7f 2021-07-03 stsp mode_t mode = got_tree_entry_get_mode(e);
811 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
812 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
817 b3d68e7f 2021-07-03 stsp if (got_object_tree_entry_is_symlink(e) ||
818 b3d68e7f 2021-07-03 stsp got_object_tree_entry_is_submodule(e) ||
819 b3d68e7f 2021-07-03 stsp got_object_idset_contains(traversed_ids, id))
822 b3d68e7f 2021-07-03 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
823 b3d68e7f 2021-07-03 stsp got_tree_entry_get_name(e)) == -1) {
824 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
828 b3d68e7f 2021-07-03 stsp if (S_ISDIR(mode)) {
829 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
830 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
833 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
834 b3d68e7f 2021-07-03 stsp } else if (S_ISREG(mode)) {
835 b3d68e7f 2021-07-03 stsp /* This blob is referenced. */
836 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, id, repo,
840 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, id, NULL);
848 b3d68e7f 2021-07-03 stsp got_object_tree_close(tree);
850 b3d68e7f 2021-07-03 stsp return err;
853 b3d68e7f 2021-07-03 stsp static const struct got_error *
854 b3d68e7f 2021-07-03 stsp load_tree(struct got_object_idset *loose_ids,
855 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
856 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
857 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
859 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
860 b3d68e7f 2021-07-03 stsp struct got_object_id_queue tree_ids;
861 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
863 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, tree_id);
865 b3d68e7f 2021-07-03 stsp return err;
867 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&tree_ids);
868 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
870 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&tree_ids)) {
871 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
872 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
877 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&tree_ids);
878 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&tree_ids, entry);
880 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(traversed_ids, &qid->id)) {
881 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
885 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(traversed_ids, &qid->id, NULL);
887 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
891 b3d68e7f 2021-07-03 stsp /* This tree is referenced. */
892 d7b5a0e8 2022-04-20 stsp err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
896 b3d68e7f 2021-07-03 stsp err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
897 d7b5a0e8 2022-04-20 stsp &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
898 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
903 b3d68e7f 2021-07-03 stsp got_object_id_queue_free(&tree_ids);
904 b3d68e7f 2021-07-03 stsp return err;
907 b3d68e7f 2021-07-03 stsp static const struct got_error *
908 b3d68e7f 2021-07-03 stsp load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
909 b3d68e7f 2021-07-03 stsp int *npacked, struct got_object_idset *traversed_ids,
910 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo,
911 211cfef0 2022-01-05 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
912 211cfef0 2022-01-05 stsp struct got_ratelimit *rl, int nloose,
913 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
915 b3d68e7f 2021-07-03 stsp const struct got_error *err;
916 b3d68e7f 2021-07-03 stsp struct got_commit_object *commit = NULL;
917 b3d68e7f 2021-07-03 stsp struct got_tag_object *tag = NULL;
918 b3d68e7f 2021-07-03 stsp struct got_object_id *tree_id = NULL;
919 b3d68e7f 2021-07-03 stsp struct got_object_id_queue ids;
920 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
921 b3d68e7f 2021-07-03 stsp int obj_type;
923 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
925 b3d68e7f 2021-07-03 stsp return err;
927 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&ids);
928 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&ids, qid, entry);
930 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&ids)) {
931 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
932 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
937 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&ids);
938 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&ids, entry);
940 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(traversed_ids, &qid->id)) {
941 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
942 b3d68e7f 2021-07-03 stsp qid = NULL;
946 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(traversed_ids, &qid->id, NULL);
950 b3d68e7f 2021-07-03 stsp /* This commit or tag is referenced. */
951 d7b5a0e8 2022-04-20 stsp err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
955 d7b5a0e8 2022-04-20 stsp err = got_object_get_type(&obj_type, repo, &qid->id);
958 b3d68e7f 2021-07-03 stsp switch (obj_type) {
959 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
960 d7b5a0e8 2022-04-20 stsp err = got_object_open_as_commit(&commit, repo,
965 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
966 d7b5a0e8 2022-04-20 stsp err = got_object_open_as_tag(&tag, repo, &qid->id);
971 b3d68e7f 2021-07-03 stsp /* should not happen */
972 b3d68e7f 2021-07-03 stsp err = got_error(GOT_ERR_OBJ_TYPE);
976 b3d68e7f 2021-07-03 stsp /* Find a tree object to scan. */
977 b3d68e7f 2021-07-03 stsp if (commit) {
978 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
979 b3d68e7f 2021-07-03 stsp } else if (tag) {
980 b3d68e7f 2021-07-03 stsp obj_type = got_object_tag_get_object_type(tag);
981 b3d68e7f 2021-07-03 stsp switch (obj_type) {
982 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
983 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo,
984 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
987 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
989 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
990 b3d68e7f 2021-07-03 stsp tree_id = got_object_tag_get_object_id(tag);
994 b3d68e7f 2021-07-03 stsp * Tag points at something other than a
995 b3d68e7f 2021-07-03 stsp * commit or tree. Leave this weird tag object
996 b3d68e7f 2021-07-03 stsp * and the object it points to on disk.
998 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
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 = got_object_idset_remove(NULL, loose_ids,
1003 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
1004 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
1005 b3d68e7f 2021-07-03 stsp goto done;
1006 b3d68e7f 2021-07-03 stsp err = NULL;
1011 b3d68e7f 2021-07-03 stsp if (tree_id) {
1012 b3d68e7f 2021-07-03 stsp err = load_tree(loose_ids, traversed_ids, tree_id, "",
1013 b3d68e7f 2021-07-03 stsp repo, npacked, cancel_cb, cancel_arg);
1018 b3d68e7f 2021-07-03 stsp if (commit || tag)
1019 b3d68e7f 2021-07-03 stsp (*ncommits)++; /* scanned tags are counted as commits */
1021 211cfef0 2022-01-05 stsp err = report_cleanup_progress(progress_cb, progress_arg, rl,
1022 211cfef0 2022-01-05 stsp nloose, *ncommits, -1);
1026 b3d68e7f 2021-07-03 stsp if (commit) {
1027 b3d68e7f 2021-07-03 stsp /* Find parent commits to scan. */
1028 b3d68e7f 2021-07-03 stsp const struct got_object_id_queue *parent_ids;
1029 b3d68e7f 2021-07-03 stsp parent_ids = got_object_commit_get_parent_ids(commit);
1030 b3d68e7f 2021-07-03 stsp err = got_object_id_queue_copy(parent_ids, &ids);
1033 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1034 b3d68e7f 2021-07-03 stsp commit = NULL;
1036 b3d68e7f 2021-07-03 stsp if (tag) {
1037 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1038 b3d68e7f 2021-07-03 stsp tag = NULL;
1040 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1041 b3d68e7f 2021-07-03 stsp qid = NULL;
1045 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1046 b3d68e7f 2021-07-03 stsp if (commit)
1047 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1049 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1050 4b2e47fb 2021-07-03 stsp got_object_id_queue_free(&ids);
1051 b3d68e7f 2021-07-03 stsp return err;
1054 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg {
1055 b3d68e7f 2021-07-03 stsp struct got_repository *repo;
1056 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb;
1057 b3d68e7f 2021-07-03 stsp void *progress_arg;
1058 211cfef0 2022-01-05 stsp struct got_ratelimit *rl;
1059 b3d68e7f 2021-07-03 stsp int nloose;
1060 b3d68e7f 2021-07-03 stsp int ncommits;
1061 b3d68e7f 2021-07-03 stsp int npurged;
1062 b3d68e7f 2021-07-03 stsp off_t size_purged;
1063 b3d68e7f 2021-07-03 stsp int dry_run;
1064 ef8ec606 2021-07-27 stsp time_t max_mtime;
1065 ef8ec606 2021-07-27 stsp int ignore_mtime;
1068 b3d68e7f 2021-07-03 stsp static const struct got_error *
1069 b3d68e7f 2021-07-03 stsp purge_loose_object(struct got_object_id *id, void *data, void *arg)
1071 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg *a = arg;
1072 b3d68e7f 2021-07-03 stsp const struct got_error *err, *unlock_err = NULL;
1073 b3d68e7f 2021-07-03 stsp char *path = NULL;
1074 b3d68e7f 2021-07-03 stsp int fd = -1;
1075 b3d68e7f 2021-07-03 stsp struct stat sb;
1076 b3d68e7f 2021-07-03 stsp struct got_lockfile *lf = NULL;
1078 b3d68e7f 2021-07-03 stsp err = got_object_get_path(&path, id, a->repo);
1080 b3d68e7f 2021-07-03 stsp return err;
1082 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, id, a->repo);
1084 b3d68e7f 2021-07-03 stsp goto done;
1086 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
1087 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
1088 b3d68e7f 2021-07-03 stsp goto done;
1092 ef8ec606 2021-07-27 stsp * Do not delete objects which are younger than our maximum
1093 ef8ec606 2021-07-27 stsp * modification time threshold. This prevents a race where
1094 ef8ec606 2021-07-27 stsp * new objects which are being added to the repository
1095 ef8ec606 2021-07-27 stsp * concurrently would be deleted.
1097 ef8ec606 2021-07-27 stsp if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1098 ef8ec606 2021-07-27 stsp if (!a->dry_run) {
1099 ef8ec606 2021-07-27 stsp err = got_lockfile_lock(&lf, path, -1);
1101 ef8ec606 2021-07-27 stsp goto done;
1102 ef8ec606 2021-07-27 stsp if (unlink(path) == -1) {
1103 ef8ec606 2021-07-27 stsp err = got_error_from_errno2("unlink", path);
1104 ef8ec606 2021-07-27 stsp goto done;
1108 ef8ec606 2021-07-27 stsp a->npurged++;
1109 ef8ec606 2021-07-27 stsp a->size_purged += sb.st_size;
1110 211cfef0 2022-01-05 stsp err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1111 211cfef0 2022-01-05 stsp a->rl, a->nloose, a->ncommits, a->npurged);
1113 211cfef0 2022-01-05 stsp goto done;
1116 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
1117 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
1118 b3d68e7f 2021-07-03 stsp free(path);
1120 5345b4c7 2021-07-06 stsp unlock_err = got_lockfile_unlock(lf, -1);
1121 b3d68e7f 2021-07-03 stsp return err ? err : unlock_err;
1124 b3d68e7f 2021-07-03 stsp const struct got_error *
1125 b3d68e7f 2021-07-03 stsp got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1126 b3d68e7f 2021-07-03 stsp off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1127 ef8ec606 2021-07-27 stsp int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1128 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1130 b3d68e7f 2021-07-03 stsp const struct got_error *err;
1131 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids;
1132 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids;
1133 b3d68e7f 2021-07-03 stsp struct got_object_id **referenced_ids;
1134 b3d68e7f 2021-07-03 stsp int i, nreferenced, nloose, ncommits = 0;
1135 b3d68e7f 2021-07-03 stsp struct got_reflist_head refs;
1136 ef8ec606 2021-07-27 stsp struct got_reflist_entry *re;
1137 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg arg;
1138 ef8ec606 2021-07-27 stsp time_t max_mtime = 0;
1139 211cfef0 2022-01-05 stsp struct got_ratelimit rl;
1141 b3d68e7f 2021-07-03 stsp TAILQ_INIT(&refs);
1142 211cfef0 2022-01-05 stsp got_ratelimit_init(&rl, 0, 500);
1144 b3d68e7f 2021-07-03 stsp *size_before = 0;
1145 b3d68e7f 2021-07-03 stsp *size_after = 0;
1146 b3d68e7f 2021-07-03 stsp *npacked = 0;
1148 b3d68e7f 2021-07-03 stsp err = get_loose_object_ids(&loose_ids, size_before,
1149 211cfef0 2022-01-05 stsp progress_cb, progress_arg, &rl, repo);
1151 b3d68e7f 2021-07-03 stsp return err;
1152 b3d68e7f 2021-07-03 stsp nloose = got_object_idset_num_elements(loose_ids);
1153 b3d68e7f 2021-07-03 stsp if (nloose == 0) {
1154 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1155 dbe266a4 2022-01-05 stsp if (progress_cb) {
1156 dbe266a4 2022-01-05 stsp err = progress_cb(progress_arg, 0, 0, 0);
1158 dbe266a4 2022-01-05 stsp return err;
1160 b3d68e7f 2021-07-03 stsp return NULL;
1163 b3d68e7f 2021-07-03 stsp traversed_ids = got_object_idset_alloc();
1164 b3d68e7f 2021-07-03 stsp if (traversed_ids == NULL) {
1165 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_object_idset_alloc");
1166 b3d68e7f 2021-07-03 stsp goto done;
1169 b3d68e7f 2021-07-03 stsp err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1171 b3d68e7f 2021-07-03 stsp goto done;
1172 ef8ec606 2021-07-27 stsp if (!ignore_mtime) {
1173 ef8ec606 2021-07-27 stsp TAILQ_FOREACH(re, &refs, entry) {
1174 ef8ec606 2021-07-27 stsp time_t mtime = got_ref_get_mtime(re->ref);
1175 ef8ec606 2021-07-27 stsp if (mtime > max_mtime)
1176 ef8ec606 2021-07-27 stsp max_mtime = mtime;
1179 ef8ec606 2021-07-27 stsp * For safety, keep objects created within 10 minutes
1180 ef8ec606 2021-07-27 stsp * before the youngest reference was created.
1182 ef8ec606 2021-07-27 stsp if (max_mtime >= 600)
1183 ef8ec606 2021-07-27 stsp max_mtime -= 600;
1186 b3d68e7f 2021-07-03 stsp err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1187 b3d68e7f 2021-07-03 stsp (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1188 b3d68e7f 2021-07-03 stsp &refs, repo, cancel_cb, cancel_arg);
1190 b3d68e7f 2021-07-03 stsp goto done;
1192 b3d68e7f 2021-07-03 stsp for (i = 0; i < nreferenced; i++) {
1193 b3d68e7f 2021-07-03 stsp struct got_object_id *id = referenced_ids[i];
1194 b3d68e7f 2021-07-03 stsp err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1195 211cfef0 2022-01-05 stsp traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1196 211cfef0 2022-01-05 stsp nloose, cancel_cb, cancel_arg);
1198 b3d68e7f 2021-07-03 stsp goto done;
1201 b3d68e7f 2021-07-03 stsp /* Any remaining loose objects are unreferenced and can be purged. */
1202 b3d68e7f 2021-07-03 stsp arg.repo = repo;
1203 b3d68e7f 2021-07-03 stsp arg.progress_arg = progress_arg;
1204 b3d68e7f 2021-07-03 stsp arg.progress_cb = progress_cb;
1205 211cfef0 2022-01-05 stsp arg.rl = &rl;
1206 b3d68e7f 2021-07-03 stsp arg.nloose = nloose;
1207 b3d68e7f 2021-07-03 stsp arg.npurged = 0;
1208 b3d68e7f 2021-07-03 stsp arg.size_purged = 0;
1209 b3d68e7f 2021-07-03 stsp arg.ncommits = ncommits;
1210 b3d68e7f 2021-07-03 stsp arg.dry_run = dry_run;
1211 ef8ec606 2021-07-27 stsp arg.max_mtime = max_mtime;
1212 ef8ec606 2021-07-27 stsp arg.ignore_mtime = ignore_mtime;
1213 b3d68e7f 2021-07-03 stsp err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1215 b3d68e7f 2021-07-03 stsp goto done;
1216 b3d68e7f 2021-07-03 stsp *size_after = *size_before - arg.size_purged;
1218 211cfef0 2022-01-05 stsp /* Produce a final progress report. */
1219 211cfef0 2022-01-05 stsp if (progress_cb) {
1220 9cd5f067 2022-01-05 stsp err = progress_cb(progress_arg, nloose, ncommits, arg.npurged);
1222 211cfef0 2022-01-05 stsp goto done;
1225 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1226 b3d68e7f 2021-07-03 stsp got_object_idset_free(traversed_ids);
1227 1124fe40 2021-07-07 stsp return err;
1230 1124fe40 2021-07-07 stsp static const struct got_error *
1231 1124fe40 2021-07-07 stsp remove_packidx(int dir_fd, const char *relpath)
1233 1124fe40 2021-07-07 stsp const struct got_error *err, *unlock_err;
1234 1124fe40 2021-07-07 stsp struct got_lockfile *lf;
1236 1124fe40 2021-07-07 stsp err = got_lockfile_lock(&lf, relpath, dir_fd);
1238 1124fe40 2021-07-07 stsp return err;
1239 1124fe40 2021-07-07 stsp if (unlinkat(dir_fd, relpath, 0) == -1)
1240 1124fe40 2021-07-07 stsp err = got_error_from_errno("unlinkat");
1241 1124fe40 2021-07-07 stsp unlock_err = got_lockfile_unlock(lf, dir_fd);
1242 1124fe40 2021-07-07 stsp return err ? err : unlock_err;
1245 1124fe40 2021-07-07 stsp const struct got_error *
1246 1124fe40 2021-07-07 stsp got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1247 1124fe40 2021-07-07 stsp got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1248 1124fe40 2021-07-07 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1250 2edc2f9d 2022-07-21 florian const struct got_error *err = NULL;
1251 1124fe40 2021-07-07 stsp DIR *packdir = NULL;
1252 1124fe40 2021-07-07 stsp struct dirent *dent;
1253 1124fe40 2021-07-07 stsp char *pack_relpath = NULL;
1254 1124fe40 2021-07-07 stsp int packdir_fd;
1255 1124fe40 2021-07-07 stsp struct stat sb;
1257 1124fe40 2021-07-07 stsp packdir_fd = openat(got_repo_get_fd(repo),
1258 e7ae0baf 2021-12-31 stsp GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1259 1124fe40 2021-07-07 stsp if (packdir_fd == -1) {
1260 1124fe40 2021-07-07 stsp if (errno == ENOENT)
1261 1124fe40 2021-07-07 stsp return NULL;
1262 1124fe40 2021-07-07 stsp return got_error_from_errno_fmt("openat: %s/%s",
1263 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1264 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR);
1267 1124fe40 2021-07-07 stsp packdir = fdopendir(packdir_fd);
1268 1124fe40 2021-07-07 stsp if (packdir == NULL) {
1269 1124fe40 2021-07-07 stsp err = got_error_from_errno("fdopendir");
1270 1124fe40 2021-07-07 stsp goto done;
1273 1124fe40 2021-07-07 stsp while ((dent = readdir(packdir)) != NULL) {
1274 1124fe40 2021-07-07 stsp if (cancel_cb) {
1275 1124fe40 2021-07-07 stsp err = cancel_cb(cancel_arg);
1277 1124fe40 2021-07-07 stsp goto done;
1280 1124fe40 2021-07-07 stsp if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1283 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath,
1284 1124fe40 2021-07-07 stsp dent->d_name);
1286 1124fe40 2021-07-07 stsp goto done;
1288 1124fe40 2021-07-07 stsp if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1289 1124fe40 2021-07-07 stsp free(pack_relpath);
1290 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1293 1124fe40 2021-07-07 stsp if (errno != ENOENT) {
1294 1124fe40 2021-07-07 stsp err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1295 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1296 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1297 1124fe40 2021-07-07 stsp pack_relpath);
1298 1124fe40 2021-07-07 stsp goto done;
1301 1124fe40 2021-07-07 stsp if (!dry_run) {
1302 1124fe40 2021-07-07 stsp err = remove_packidx(packdir_fd, dent->d_name);
1304 1124fe40 2021-07-07 stsp goto done;
1306 1124fe40 2021-07-07 stsp if (progress_cb) {
1307 1124fe40 2021-07-07 stsp char *path;
1308 1124fe40 2021-07-07 stsp if (asprintf(&path, "%s/%s/%s",
1309 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1310 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1311 1124fe40 2021-07-07 stsp dent->d_name) == -1) {
1312 1124fe40 2021-07-07 stsp err = got_error_from_errno("asprintf");
1313 1124fe40 2021-07-07 stsp goto done;
1315 1124fe40 2021-07-07 stsp err = progress_cb(progress_arg, path);
1316 1124fe40 2021-07-07 stsp free(path);
1318 1124fe40 2021-07-07 stsp goto done;
1320 1124fe40 2021-07-07 stsp free(pack_relpath);
1321 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1324 1124fe40 2021-07-07 stsp if (packdir && closedir(packdir) != 0 && err == NULL)
1325 1124fe40 2021-07-07 stsp err = got_error_from_errno("closedir");
1326 1124fe40 2021-07-07 stsp free(pack_relpath);
1327 b3d68e7f 2021-07-03 stsp return err;
