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 3bf0e21f 2023-06-19 op const struct got_error *
617 3bf0e21f 2023-06-19 op got_repo_cleanup_prepare(struct got_repository *repo,
618 3bf0e21f 2023-06-19 op struct got_lockfile **lk)
620 3bf0e21f 2023-06-19 op const struct got_error *err;
621 3bf0e21f 2023-06-19 op char myname[HOST_NAME_MAX + 1];
623 3bf0e21f 2023-06-19 op if (gethostname(myname, sizeof(myname)) == -1)
624 3bf0e21f 2023-06-19 op return got_error_from_errno("gethostname");
626 3bf0e21f 2023-06-19 op err = got_lockfile_lock(lk, "gc.pid", got_repo_get_fd(repo));
631 3bf0e21f 2023-06-19 op * Git uses these info to provide some verbiage when finds a
632 3bf0e21f 2023-06-19 op * lock during `git gc --force' so don't try too hard to avoid
633 3bf0e21f 2023-06-19 op * short writes and don't care if a race happens between the
634 3bf0e21f 2023-06-19 op * lockfile creation and the write itself.
636 3bf0e21f 2023-06-19 op if (dprintf((*lk)->fd, "%d %s", getpid(), myname) < 0)
637 3bf0e21f 2023-06-19 op return got_error_from_errno("dprintf");
642 3bf0e21f 2023-06-19 op const struct got_error *
643 3bf0e21f 2023-06-19 op got_repo_cleanup_complete(struct got_repository *repo,
644 3bf0e21f 2023-06-19 op struct got_lockfile *lk)
646 3bf0e21f 2023-06-19 op if (lk == NULL)
649 3bf0e21f 2023-06-19 op return got_lockfile_unlock(lk, got_repo_get_fd(repo));
652 b3d68e7f 2021-07-03 stsp static const struct got_error *
653 211cfef0 2022-01-05 stsp report_cleanup_progress(got_cleanup_progress_cb progress_cb,
654 211cfef0 2022-01-05 stsp void *progress_arg, struct got_ratelimit *rl,
655 211cfef0 2022-01-05 stsp int nloose, int ncommits, int npurged)
657 211cfef0 2022-01-05 stsp const struct got_error *err;
658 211cfef0 2022-01-05 stsp int elapsed;
660 211cfef0 2022-01-05 stsp if (progress_cb == NULL)
661 211cfef0 2022-01-05 stsp return NULL;
663 211cfef0 2022-01-05 stsp err = got_ratelimit_check(&elapsed, rl);
664 211cfef0 2022-01-05 stsp if (err || !elapsed)
665 211cfef0 2022-01-05 stsp return err;
667 9a7c12cf 2023-06-18 op return progress_cb(progress_arg, nloose, ncommits, npurged, -1);
670 211cfef0 2022-01-05 stsp static const struct got_error *
671 b3d68e7f 2021-07-03 stsp get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
672 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
673 211cfef0 2022-01-05 stsp struct got_ratelimit *rl, struct got_repository *repo)
675 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
676 b3d68e7f 2021-07-03 stsp char *path_objects = NULL, *path = NULL;
677 b3d68e7f 2021-07-03 stsp DIR *dir = NULL;
678 b3d68e7f 2021-07-03 stsp struct got_object *obj = NULL;
679 b3d68e7f 2021-07-03 stsp struct got_object_id id;
680 b3d68e7f 2021-07-03 stsp int i, fd = -1;
681 b3d68e7f 2021-07-03 stsp struct stat sb;
683 b3d68e7f 2021-07-03 stsp *ondisk_size = 0;
684 b3d68e7f 2021-07-03 stsp *loose_ids = got_object_idset_alloc();
685 b3d68e7f 2021-07-03 stsp if (*loose_ids == NULL)
686 b3d68e7f 2021-07-03 stsp return got_error_from_errno("got_object_idset_alloc");
688 b3d68e7f 2021-07-03 stsp path_objects = got_repo_get_path_objects(repo);
689 b3d68e7f 2021-07-03 stsp if (path_objects == NULL) {
690 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_repo_get_path_objects");
694 b3d68e7f 2021-07-03 stsp for (i = 0; i <= 0xff; i++) {
695 b3d68e7f 2021-07-03 stsp struct dirent *dent;
697 b3d68e7f 2021-07-03 stsp if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
698 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
702 b3d68e7f 2021-07-03 stsp dir = opendir(path);
703 b3d68e7f 2021-07-03 stsp if (dir == NULL) {
704 b3d68e7f 2021-07-03 stsp if (errno == ENOENT) {
705 b3d68e7f 2021-07-03 stsp err = NULL;
708 b3d68e7f 2021-07-03 stsp err = got_error_from_errno2("opendir", path);
712 b3d68e7f 2021-07-03 stsp while ((dent = readdir(dir)) != NULL) {
713 b3d68e7f 2021-07-03 stsp char *id_str;
715 b3d68e7f 2021-07-03 stsp if (strcmp(dent->d_name, ".") == 0 ||
716 b3d68e7f 2021-07-03 stsp strcmp(dent->d_name, "..") == 0)
719 b3d68e7f 2021-07-03 stsp if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
720 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
724 b54930d5 2023-02-28 op if (!got_parse_object_id(&id, id_str, repo->algo)) {
725 b3d68e7f 2021-07-03 stsp free(id_str);
728 b3d68e7f 2021-07-03 stsp free(id_str);
730 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, &id, repo);
733 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
734 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
737 d5c81d44 2021-07-08 stsp err = got_object_read_header_privsep(&obj, &id, repo,
741 b3d68e7f 2021-07-03 stsp fd = -1; /* already closed */
743 b3d68e7f 2021-07-03 stsp switch (obj->type) {
744 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
745 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
746 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_BLOB:
747 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
750 b3d68e7f 2021-07-03 stsp err = got_error_fmt(GOT_ERR_OBJ_TYPE,
751 b3d68e7f 2021-07-03 stsp "%d", obj->type);
754 b3d68e7f 2021-07-03 stsp got_object_close(obj);
755 b3d68e7f 2021-07-03 stsp obj = NULL;
756 b3d68e7f 2021-07-03 stsp (*ondisk_size) += sb.st_size;
757 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(*loose_ids, &id, NULL);
760 211cfef0 2022-01-05 stsp err = report_cleanup_progress(progress_cb,
761 211cfef0 2022-01-05 stsp progress_arg, rl,
762 211cfef0 2022-01-05 stsp got_object_idset_num_elements(*loose_ids), -1, -1);
767 b3d68e7f 2021-07-03 stsp if (closedir(dir) != 0) {
768 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
771 b3d68e7f 2021-07-03 stsp dir = NULL;
773 b3d68e7f 2021-07-03 stsp free(path);
774 b3d68e7f 2021-07-03 stsp path = NULL;
777 b3d68e7f 2021-07-03 stsp if (dir && closedir(dir) != 0 && err == NULL)
778 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
779 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
780 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
782 b3d68e7f 2021-07-03 stsp got_object_idset_free(*loose_ids);
783 b3d68e7f 2021-07-03 stsp *loose_ids = NULL;
786 b3d68e7f 2021-07-03 stsp got_object_close(obj);
787 b3d68e7f 2021-07-03 stsp free(path_objects);
788 b3d68e7f 2021-07-03 stsp free(path);
789 b3d68e7f 2021-07-03 stsp return err;
792 b3d68e7f 2021-07-03 stsp static const struct got_error *
793 b3d68e7f 2021-07-03 stsp preserve_loose_object(struct got_object_idset *loose_ids,
794 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo, int *npacked)
796 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
797 9614da0d 2021-07-04 stsp struct got_object *obj;
799 b3d68e7f 2021-07-03 stsp if (!got_object_idset_contains(loose_ids, id))
800 b3d68e7f 2021-07-03 stsp return NULL;
803 9614da0d 2021-07-04 stsp * Try to open this object from a pack file. This ensures that
804 9614da0d 2021-07-04 stsp * we do in fact have a valid packed copy of the object. Otherwise
805 9614da0d 2021-07-04 stsp * we should not delete the loose representation of this object.
807 abc59930 2021-09-05 naddy err = got_object_open_packed(&obj, id, repo);
808 abc59930 2021-09-05 naddy if (err == NULL) {
809 9614da0d 2021-07-04 stsp got_object_close(obj);
811 9614da0d 2021-07-04 stsp * The object is referenced and packed.
812 9614da0d 2021-07-04 stsp * We can purge the redundantly stored loose object.
814 9614da0d 2021-07-04 stsp (*npacked)++;
815 9614da0d 2021-07-04 stsp return NULL;
816 9614da0d 2021-07-04 stsp } else if (err->code != GOT_ERR_NO_OBJ)
817 9614da0d 2021-07-04 stsp return err;
820 b3d68e7f 2021-07-03 stsp * This object is referenced and not packed.
821 b3d68e7f 2021-07-03 stsp * Remove it from our purge set.
823 b3d68e7f 2021-07-03 stsp return got_object_idset_remove(NULL, loose_ids, id);
826 b3d68e7f 2021-07-03 stsp static const struct got_error *
827 b3d68e7f 2021-07-03 stsp load_tree_entries(struct got_object_id_queue *ids,
828 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids,
829 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
830 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
831 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
833 b3d68e7f 2021-07-03 stsp const struct got_error *err;
834 b3d68e7f 2021-07-03 stsp struct got_tree_object *tree;
835 b3d68e7f 2021-07-03 stsp char *p = NULL;
838 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
840 b3d68e7f 2021-07-03 stsp return err;
842 b3d68e7f 2021-07-03 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
843 b3d68e7f 2021-07-03 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
844 b3d68e7f 2021-07-03 stsp struct got_object_id *id = got_tree_entry_get_id(e);
845 b3d68e7f 2021-07-03 stsp mode_t mode = got_tree_entry_get_mode(e);
847 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
848 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
853 b3d68e7f 2021-07-03 stsp if (got_object_tree_entry_is_symlink(e) ||
854 b3d68e7f 2021-07-03 stsp got_object_tree_entry_is_submodule(e) ||
855 b3d68e7f 2021-07-03 stsp got_object_idset_contains(traversed_ids, id))
858 b3d68e7f 2021-07-03 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
859 b3d68e7f 2021-07-03 stsp got_tree_entry_get_name(e)) == -1) {
860 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
864 b3d68e7f 2021-07-03 stsp if (S_ISDIR(mode)) {
865 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
866 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
869 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
870 b3d68e7f 2021-07-03 stsp } else if (S_ISREG(mode)) {
871 b3d68e7f 2021-07-03 stsp /* This blob is referenced. */
872 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, id, repo,
876 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, id, NULL);
884 b3d68e7f 2021-07-03 stsp got_object_tree_close(tree);
886 b3d68e7f 2021-07-03 stsp return err;
889 b3d68e7f 2021-07-03 stsp static const struct got_error *
890 b3d68e7f 2021-07-03 stsp load_tree(struct got_object_idset *loose_ids,
891 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
892 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
893 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
895 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
896 b3d68e7f 2021-07-03 stsp struct got_object_id_queue tree_ids;
897 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
899 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, tree_id);
901 b3d68e7f 2021-07-03 stsp return err;
903 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&tree_ids);
904 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
906 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&tree_ids)) {
907 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
908 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
913 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&tree_ids);
914 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&tree_ids, entry);
916 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(traversed_ids, &qid->id)) {
917 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
921 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(traversed_ids, &qid->id, NULL);
923 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
927 b3d68e7f 2021-07-03 stsp /* This tree is referenced. */
928 d7b5a0e8 2022-04-20 stsp err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
932 b3d68e7f 2021-07-03 stsp err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
933 d7b5a0e8 2022-04-20 stsp &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
934 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
939 b3d68e7f 2021-07-03 stsp got_object_id_queue_free(&tree_ids);
940 b3d68e7f 2021-07-03 stsp return err;
943 b3d68e7f 2021-07-03 stsp static const struct got_error *
944 b3d68e7f 2021-07-03 stsp load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
945 b3d68e7f 2021-07-03 stsp int *npacked, struct got_object_idset *traversed_ids,
946 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo,
947 211cfef0 2022-01-05 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
948 211cfef0 2022-01-05 stsp struct got_ratelimit *rl, int nloose,
949 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
951 b3d68e7f 2021-07-03 stsp const struct got_error *err;
952 b3d68e7f 2021-07-03 stsp struct got_commit_object *commit = NULL;
953 b3d68e7f 2021-07-03 stsp struct got_tag_object *tag = NULL;
954 b3d68e7f 2021-07-03 stsp struct got_object_id *tree_id = NULL;
955 b3d68e7f 2021-07-03 stsp struct got_object_id_queue ids;
956 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
957 b3d68e7f 2021-07-03 stsp int obj_type;
959 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
961 b3d68e7f 2021-07-03 stsp return err;
963 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&ids);
964 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&ids, qid, entry);
966 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&ids)) {
967 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
968 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
973 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&ids);
974 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&ids, entry);
976 d7b5a0e8 2022-04-20 stsp if (got_object_idset_contains(traversed_ids, &qid->id)) {
977 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
978 b3d68e7f 2021-07-03 stsp qid = NULL;
982 d7b5a0e8 2022-04-20 stsp err = got_object_idset_add(traversed_ids, &qid->id, NULL);
986 b3d68e7f 2021-07-03 stsp /* This commit or tag is referenced. */
987 d7b5a0e8 2022-04-20 stsp err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
991 d7b5a0e8 2022-04-20 stsp err = got_object_get_type(&obj_type, repo, &qid->id);
994 b3d68e7f 2021-07-03 stsp switch (obj_type) {
995 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
996 d7b5a0e8 2022-04-20 stsp err = got_object_open_as_commit(&commit, repo,
1001 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
1002 d7b5a0e8 2022-04-20 stsp err = got_object_open_as_tag(&tag, repo, &qid->id);
1004 b3d68e7f 2021-07-03 stsp goto done;
1007 b3d68e7f 2021-07-03 stsp /* should not happen */
1008 b3d68e7f 2021-07-03 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1009 b3d68e7f 2021-07-03 stsp goto done;
1012 b3d68e7f 2021-07-03 stsp /* Find a tree object to scan. */
1013 b3d68e7f 2021-07-03 stsp if (commit) {
1014 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
1015 b3d68e7f 2021-07-03 stsp } else if (tag) {
1016 b3d68e7f 2021-07-03 stsp obj_type = got_object_tag_get_object_type(tag);
1017 b3d68e7f 2021-07-03 stsp switch (obj_type) {
1018 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
1019 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo,
1020 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
1022 b3d68e7f 2021-07-03 stsp goto done;
1023 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
1025 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
1026 b3d68e7f 2021-07-03 stsp tree_id = got_object_tag_get_object_id(tag);
1030 b3d68e7f 2021-07-03 stsp * Tag points at something other than a
1031 b3d68e7f 2021-07-03 stsp * commit or tree. Leave this weird tag object
1032 b3d68e7f 2021-07-03 stsp * and the object it points to on disk.
1034 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
1035 d7b5a0e8 2022-04-20 stsp &qid->id);
1036 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
1037 b3d68e7f 2021-07-03 stsp goto done;
1038 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
1039 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
1040 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
1041 b3d68e7f 2021-07-03 stsp goto done;
1042 b3d68e7f 2021-07-03 stsp err = NULL;
1047 b3d68e7f 2021-07-03 stsp if (tree_id) {
1048 b3d68e7f 2021-07-03 stsp err = load_tree(loose_ids, traversed_ids, tree_id, "",
1049 b3d68e7f 2021-07-03 stsp repo, npacked, cancel_cb, cancel_arg);
1054 b3d68e7f 2021-07-03 stsp if (commit || tag)
1055 b3d68e7f 2021-07-03 stsp (*ncommits)++; /* scanned tags are counted as commits */
1057 211cfef0 2022-01-05 stsp err = report_cleanup_progress(progress_cb, progress_arg, rl,
1058 211cfef0 2022-01-05 stsp nloose, *ncommits, -1);
1062 b3d68e7f 2021-07-03 stsp if (commit) {
1063 b3d68e7f 2021-07-03 stsp /* Find parent commits to scan. */
1064 b3d68e7f 2021-07-03 stsp const struct got_object_id_queue *parent_ids;
1065 b3d68e7f 2021-07-03 stsp parent_ids = got_object_commit_get_parent_ids(commit);
1066 b3d68e7f 2021-07-03 stsp err = got_object_id_queue_copy(parent_ids, &ids);
1069 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1070 b3d68e7f 2021-07-03 stsp commit = NULL;
1072 b3d68e7f 2021-07-03 stsp if (tag) {
1073 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1074 b3d68e7f 2021-07-03 stsp tag = NULL;
1076 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1077 b3d68e7f 2021-07-03 stsp qid = NULL;
1081 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1082 b3d68e7f 2021-07-03 stsp if (commit)
1083 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1085 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1086 4b2e47fb 2021-07-03 stsp got_object_id_queue_free(&ids);
1087 b3d68e7f 2021-07-03 stsp return err;
1090 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg {
1091 b3d68e7f 2021-07-03 stsp struct got_repository *repo;
1092 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb;
1093 b3d68e7f 2021-07-03 stsp void *progress_arg;
1094 211cfef0 2022-01-05 stsp struct got_ratelimit *rl;
1095 b3d68e7f 2021-07-03 stsp int nloose;
1096 b3d68e7f 2021-07-03 stsp int ncommits;
1097 b3d68e7f 2021-07-03 stsp int npurged;
1098 b3d68e7f 2021-07-03 stsp off_t size_purged;
1099 b3d68e7f 2021-07-03 stsp int dry_run;
1100 ef8ec606 2021-07-27 stsp time_t max_mtime;
1101 ef8ec606 2021-07-27 stsp int ignore_mtime;
1104 b3d68e7f 2021-07-03 stsp static const struct got_error *
1105 b3d68e7f 2021-07-03 stsp purge_loose_object(struct got_object_id *id, void *data, void *arg)
1107 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg *a = arg;
1108 b3d68e7f 2021-07-03 stsp const struct got_error *err, *unlock_err = NULL;
1109 b3d68e7f 2021-07-03 stsp char *path = NULL;
1110 b3d68e7f 2021-07-03 stsp int fd = -1;
1111 b3d68e7f 2021-07-03 stsp struct stat sb;
1112 b3d68e7f 2021-07-03 stsp struct got_lockfile *lf = NULL;
1114 b3d68e7f 2021-07-03 stsp err = got_object_get_path(&path, id, a->repo);
1116 b3d68e7f 2021-07-03 stsp return err;
1118 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, id, a->repo);
1120 b3d68e7f 2021-07-03 stsp goto done;
1122 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
1123 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
1124 b3d68e7f 2021-07-03 stsp goto done;
1128 ef8ec606 2021-07-27 stsp * Do not delete objects which are younger than our maximum
1129 ef8ec606 2021-07-27 stsp * modification time threshold. This prevents a race where
1130 ef8ec606 2021-07-27 stsp * new objects which are being added to the repository
1131 ef8ec606 2021-07-27 stsp * concurrently would be deleted.
1133 ef8ec606 2021-07-27 stsp if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1134 ef8ec606 2021-07-27 stsp if (!a->dry_run) {
1135 ef8ec606 2021-07-27 stsp err = got_lockfile_lock(&lf, path, -1);
1137 ef8ec606 2021-07-27 stsp goto done;
1138 ef8ec606 2021-07-27 stsp if (unlink(path) == -1) {
1139 ef8ec606 2021-07-27 stsp err = got_error_from_errno2("unlink", path);
1140 ef8ec606 2021-07-27 stsp goto done;
1144 ef8ec606 2021-07-27 stsp a->npurged++;
1145 ef8ec606 2021-07-27 stsp a->size_purged += sb.st_size;
1146 211cfef0 2022-01-05 stsp err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1147 211cfef0 2022-01-05 stsp a->rl, a->nloose, a->ncommits, a->npurged);
1149 211cfef0 2022-01-05 stsp goto done;
1152 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
1153 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
1154 b3d68e7f 2021-07-03 stsp free(path);
1156 5345b4c7 2021-07-06 stsp unlock_err = got_lockfile_unlock(lf, -1);
1157 b3d68e7f 2021-07-03 stsp return err ? err : unlock_err;
1160 b3d68e7f 2021-07-03 stsp const struct got_error *
1161 b3d68e7f 2021-07-03 stsp got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1162 900499fd 2023-06-23 stsp off_t *size_before, off_t *size_after, int *ncommits, int *nloose,
1163 900499fd 2023-06-23 stsp int *npacked, int dry_run, int ignore_mtime,
1164 900499fd 2023-06-23 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
1165 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1167 b3d68e7f 2021-07-03 stsp const struct got_error *err;
1168 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids;
1169 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids;
1170 b3d68e7f 2021-07-03 stsp struct got_object_id **referenced_ids;
1171 900499fd 2023-06-23 stsp int i, nreferenced;
1172 b3d68e7f 2021-07-03 stsp struct got_reflist_head refs;
1173 ef8ec606 2021-07-27 stsp struct got_reflist_entry *re;
1174 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg arg;
1175 ef8ec606 2021-07-27 stsp time_t max_mtime = 0;
1176 211cfef0 2022-01-05 stsp struct got_ratelimit rl;
1178 b3d68e7f 2021-07-03 stsp TAILQ_INIT(&refs);
1179 211cfef0 2022-01-05 stsp got_ratelimit_init(&rl, 0, 500);
1181 b3d68e7f 2021-07-03 stsp *size_before = 0;
1182 b3d68e7f 2021-07-03 stsp *size_after = 0;
1183 b3d68e7f 2021-07-03 stsp *npacked = 0;
1185 b3d68e7f 2021-07-03 stsp err = get_loose_object_ids(&loose_ids, size_before,
1186 211cfef0 2022-01-05 stsp progress_cb, progress_arg, &rl, repo);
1188 b3d68e7f 2021-07-03 stsp return err;
1189 900499fd 2023-06-23 stsp *nloose = got_object_idset_num_elements(loose_ids);
1190 900499fd 2023-06-23 stsp if (*nloose == 0) {
1191 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1192 dbe266a4 2022-01-05 stsp if (progress_cb) {
1193 9a7c12cf 2023-06-18 op err = progress_cb(progress_arg, 0, 0, 0, -1);
1195 dbe266a4 2022-01-05 stsp return err;
1197 b3d68e7f 2021-07-03 stsp return NULL;
1200 b3d68e7f 2021-07-03 stsp traversed_ids = got_object_idset_alloc();
1201 b3d68e7f 2021-07-03 stsp if (traversed_ids == NULL) {
1202 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_object_idset_alloc");
1203 b3d68e7f 2021-07-03 stsp goto done;
1206 b3d68e7f 2021-07-03 stsp err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1208 b3d68e7f 2021-07-03 stsp goto done;
1209 ef8ec606 2021-07-27 stsp if (!ignore_mtime) {
1210 ef8ec606 2021-07-27 stsp TAILQ_FOREACH(re, &refs, entry) {
1211 ef8ec606 2021-07-27 stsp time_t mtime = got_ref_get_mtime(re->ref);
1212 ef8ec606 2021-07-27 stsp if (mtime > max_mtime)
1213 ef8ec606 2021-07-27 stsp max_mtime = mtime;
1216 ef8ec606 2021-07-27 stsp * For safety, keep objects created within 10 minutes
1217 ef8ec606 2021-07-27 stsp * before the youngest reference was created.
1219 ef8ec606 2021-07-27 stsp if (max_mtime >= 600)
1220 ef8ec606 2021-07-27 stsp max_mtime -= 600;
1223 b3d68e7f 2021-07-03 stsp err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1224 b3d68e7f 2021-07-03 stsp (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1225 b3d68e7f 2021-07-03 stsp &refs, repo, cancel_cb, cancel_arg);
1227 b3d68e7f 2021-07-03 stsp goto done;
1229 b3d68e7f 2021-07-03 stsp for (i = 0; i < nreferenced; i++) {
1230 b3d68e7f 2021-07-03 stsp struct got_object_id *id = referenced_ids[i];
1231 900499fd 2023-06-23 stsp err = load_commit_or_tag(loose_ids, ncommits, npacked,
1232 211cfef0 2022-01-05 stsp traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1233 900499fd 2023-06-23 stsp *nloose, cancel_cb, cancel_arg);
1235 b3d68e7f 2021-07-03 stsp goto done;
1238 b3d68e7f 2021-07-03 stsp /* Any remaining loose objects are unreferenced and can be purged. */
1239 b3d68e7f 2021-07-03 stsp arg.repo = repo;
1240 b3d68e7f 2021-07-03 stsp arg.progress_arg = progress_arg;
1241 b3d68e7f 2021-07-03 stsp arg.progress_cb = progress_cb;
1242 211cfef0 2022-01-05 stsp arg.rl = &rl;
1243 900499fd 2023-06-23 stsp arg.nloose = *nloose;
1244 b3d68e7f 2021-07-03 stsp arg.npurged = 0;
1245 b3d68e7f 2021-07-03 stsp arg.size_purged = 0;
1246 900499fd 2023-06-23 stsp arg.ncommits = *ncommits;
1247 b3d68e7f 2021-07-03 stsp arg.dry_run = dry_run;
1248 ef8ec606 2021-07-27 stsp arg.max_mtime = max_mtime;
1249 ef8ec606 2021-07-27 stsp arg.ignore_mtime = ignore_mtime;
1250 b3d68e7f 2021-07-03 stsp err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1252 b3d68e7f 2021-07-03 stsp goto done;
1253 b3d68e7f 2021-07-03 stsp *size_after = *size_before - arg.size_purged;
1255 211cfef0 2022-01-05 stsp /* Produce a final progress report. */
1256 211cfef0 2022-01-05 stsp if (progress_cb) {
1257 900499fd 2023-06-23 stsp err = progress_cb(progress_arg, *nloose, *ncommits, arg.npurged,
1260 211cfef0 2022-01-05 stsp goto done;
1263 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1264 b3d68e7f 2021-07-03 stsp got_object_idset_free(traversed_ids);
1268 9a7c12cf 2023-06-18 op static const struct got_error *
1269 9a7c12cf 2023-06-18 op purge_redundant_pack(struct got_repository *repo, const char *packidx_path,
1270 9a7c12cf 2023-06-18 op int dry_run, int *remove, off_t *size_before, off_t *size_after)
1272 9a7c12cf 2023-06-18 op static const char *ext[] = {".idx", ".pack", ".rev", ".bitmap",
1273 9a7c12cf 2023-06-18 op ".promisor", ".mtimes"};
1274 9a7c12cf 2023-06-18 op struct stat sb;
1275 9a7c12cf 2023-06-18 op char *dot, path[PATH_MAX];
1278 9a7c12cf 2023-06-18 op if (strlcpy(path, packidx_path, sizeof(path)) >=
1279 9a7c12cf 2023-06-18 op sizeof(path))
1280 9a7c12cf 2023-06-18 op return got_error(GOT_ERR_NO_SPACE);
1283 9a7c12cf 2023-06-18 op * For compatibility with Git, if a matching .keep file exist
1284 9a7c12cf 2023-06-18 op * don't delete the packfile.
1286 9a7c12cf 2023-06-18 op dot = strrchr(path, '.');
1287 9a7c12cf 2023-06-18 op *dot = '\0';
1288 9a7c12cf 2023-06-18 op if (strlcat(path, ".keep", sizeof(path)) >= sizeof(path))
1289 9a7c12cf 2023-06-18 op return got_error(GOT_ERR_NO_SPACE);
1290 9a7c12cf 2023-06-18 op if (faccessat(got_repo_get_fd(repo), path, F_OK, 0) == 0)
1291 9a7c12cf 2023-06-18 op *remove = 0;
1293 9a7c12cf 2023-06-18 op for (i = 0; i < nitems(ext); ++i) {
1294 9a7c12cf 2023-06-18 op *dot = '\0';
1296 9a7c12cf 2023-06-18 op if (strlcat(path, ext[i], sizeof(path)) >=
1297 9a7c12cf 2023-06-18 op sizeof(path))
1298 9a7c12cf 2023-06-18 op return got_error(GOT_ERR_NO_SPACE);
1300 9a7c12cf 2023-06-18 op if (fstatat(got_repo_get_fd(repo), path, &sb, 0) ==
1302 9a7c12cf 2023-06-18 op if (errno == ENOENT &&
1303 9a7c12cf 2023-06-18 op strcmp(ext[i], ".pack") != 0 &&
1304 9a7c12cf 2023-06-18 op strcmp(ext[i], ".idx") != 0)
1306 9a7c12cf 2023-06-18 op return got_error_from_errno2("fstatat", path);
1309 9a7c12cf 2023-06-18 op *size_before += sb.st_size;
1310 9a7c12cf 2023-06-18 op if (!*remove) {
1311 9a7c12cf 2023-06-18 op *size_after += sb.st_size;
1315 9a7c12cf 2023-06-18 op if (dry_run)
1318 9a7c12cf 2023-06-18 op if (unlinkat(got_repo_get_fd(repo), path, 0) == -1) {
1319 9a7c12cf 2023-06-18 op if (errno == ENOENT)
1321 9a7c12cf 2023-06-18 op return got_error_from_errno2("unlinkat",
1326 9a7c12cf 2023-06-18 op return NULL;
1329 9a7c12cf 2023-06-18 op static const struct got_error *
1330 9a7c12cf 2023-06-18 op pack_is_redundant(int *redundant, struct got_repository *repo,
1331 9a7c12cf 2023-06-18 op const char *packidx_path, struct got_object_idset *idset)
1333 9a7c12cf 2023-06-18 op const struct got_error *err;
1334 9a7c12cf 2023-06-18 op struct got_packidx *packidx;
1335 9a7c12cf 2023-06-18 op struct got_packidx_object_id *pid;
1336 9a7c12cf 2023-06-18 op struct got_object_id id;
1337 9a7c12cf 2023-06-18 op size_t i, nobjects;
1339 9a7c12cf 2023-06-18 op *redundant = 1;
1341 9a7c12cf 2023-06-18 op err = got_repo_get_packidx(&packidx, packidx_path, repo);
1345 9a7c12cf 2023-06-18 op nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1346 9a7c12cf 2023-06-18 op for (i = 0; i < nobjects; ++i) {
1347 9a7c12cf 2023-06-18 op pid = &packidx->hdr.sorted_ids[i];
1349 9a7c12cf 2023-06-18 op memset(&id, 0, sizeof(id));
1350 9a7c12cf 2023-06-18 op memcpy(&id.sha1, pid->sha1, sizeof(id.sha1));
1352 9a7c12cf 2023-06-18 op if (got_object_idset_contains(idset, &id))
1355 9a7c12cf 2023-06-18 op *redundant = 0;
1356 9a7c12cf 2023-06-18 op err = got_object_idset_add(idset, &id, NULL);
1361 9a7c12cf 2023-06-18 op return NULL;
1364 9a7c12cf 2023-06-18 op struct pack_info {
1365 9a7c12cf 2023-06-18 op const char *path;
1366 9a7c12cf 2023-06-18 op size_t nobjects;
1370 9a7c12cf 2023-06-18 op pack_info_cmp(const void *a, const void *b)
1372 9a7c12cf 2023-06-18 op const struct pack_info *pa, *pb;
1376 9a7c12cf 2023-06-18 op if (pa->nobjects == pb->nobjects)
1377 9a7c12cf 2023-06-18 op return strcmp(pa->path, pb->path);
1378 9a7c12cf 2023-06-18 op if (pa->nobjects > pb->nobjects)
1383 9a7c12cf 2023-06-18 op const struct got_error *
1384 9a7c12cf 2023-06-18 op got_repo_purge_redundant_packfiles(struct got_repository *repo,
1385 9a7c12cf 2023-06-18 op off_t *size_before, off_t *size_after, int dry_run,
1386 900499fd 2023-06-23 stsp int nloose, int ncommits, int npurged,
1387 9a7c12cf 2023-06-18 op got_cleanup_progress_cb progress_cb, void *progress_arg,
1388 9a7c12cf 2023-06-18 op got_cancel_cb cancel_cb, void *cancel_arg)
1390 9a7c12cf 2023-06-18 op const struct got_error *err;
1391 9a7c12cf 2023-06-18 op struct pack_info *pinfo, *sorted = NULL;
1392 9a7c12cf 2023-06-18 op struct got_packidx *packidx;
1393 9a7c12cf 2023-06-18 op struct got_object_idset *idset = NULL;
1394 9a7c12cf 2023-06-18 op struct got_pathlist_entry *pe;
1395 9a7c12cf 2023-06-18 op size_t i, npacks;
1396 9a7c12cf 2023-06-18 op int remove, redundant_packs = 0;
1398 9a7c12cf 2023-06-18 op *size_before = 0;
1399 9a7c12cf 2023-06-18 op *size_after = 0;
1402 9a7c12cf 2023-06-18 op TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
1405 9a7c12cf 2023-06-18 op if (npacks == 0)
1406 9a7c12cf 2023-06-18 op return NULL;
1408 9a7c12cf 2023-06-18 op sorted = calloc(npacks, sizeof(*sorted));
1409 9a7c12cf 2023-06-18 op if (sorted == NULL)
1410 9a7c12cf 2023-06-18 op return got_error_from_errno("calloc");
1413 9a7c12cf 2023-06-18 op TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1414 9a7c12cf 2023-06-18 op err = got_repo_get_packidx(&packidx, pe->path, repo);
1418 9a7c12cf 2023-06-18 op pinfo = &sorted[i++];
1419 9a7c12cf 2023-06-18 op pinfo->path = pe->path;
1420 9a7c12cf 2023-06-18 op pinfo->nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1422 9a7c12cf 2023-06-18 op qsort(sorted, npacks, sizeof(*sorted), pack_info_cmp);
1424 9a7c12cf 2023-06-18 op idset = got_object_idset_alloc();
1425 9a7c12cf 2023-06-18 op if (idset == NULL) {
1426 9a7c12cf 2023-06-18 op err = got_error_from_errno("got_object_idset_alloc");
1430 9a7c12cf 2023-06-18 op for (i = 0; i < npacks; ++i) {
1431 9a7c12cf 2023-06-18 op if (cancel_cb) {
1432 9a7c12cf 2023-06-18 op err = (*cancel_cb)(cancel_arg);
1437 9a7c12cf 2023-06-18 op err = pack_is_redundant(&remove, repo, sorted[i].path, idset);
1440 9a7c12cf 2023-06-18 op err = purge_redundant_pack(repo, sorted[i].path, dry_run,
1441 9a7c12cf 2023-06-18 op &remove, size_before, size_after);
1444 9a7c12cf 2023-06-18 op if (!remove)
1446 900499fd 2023-06-23 stsp if (progress_cb) {
1447 900499fd 2023-06-23 stsp err = progress_cb(progress_arg, nloose, ncommits,
1448 900499fd 2023-06-23 stsp npurged, ++redundant_packs);
1450 900499fd 2023-06-23 stsp goto done;
1454 900499fd 2023-06-23 stsp /* Produce a final progress report. */
1455 900499fd 2023-06-23 stsp if (progress_cb) {
1456 900499fd 2023-06-23 stsp err = progress_cb(progress_arg, nloose, ncommits, npurged,
1457 900499fd 2023-06-23 stsp redundant_packs);
1462 9a7c12cf 2023-06-18 op free(sorted);
1464 9a7c12cf 2023-06-18 op got_object_idset_free(idset);
1465 1124fe40 2021-07-07 stsp return err;
1468 1124fe40 2021-07-07 stsp const struct got_error *
1469 1124fe40 2021-07-07 stsp got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1470 1124fe40 2021-07-07 stsp got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1471 1124fe40 2021-07-07 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1473 2edc2f9d 2022-07-21 florian const struct got_error *err = NULL;
1474 1124fe40 2021-07-07 stsp DIR *packdir = NULL;
1475 1124fe40 2021-07-07 stsp struct dirent *dent;
1476 1124fe40 2021-07-07 stsp char *pack_relpath = NULL;
1477 1124fe40 2021-07-07 stsp int packdir_fd;
1478 1124fe40 2021-07-07 stsp struct stat sb;
1480 1124fe40 2021-07-07 stsp packdir_fd = openat(got_repo_get_fd(repo),
1481 e7ae0baf 2021-12-31 stsp GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1482 1124fe40 2021-07-07 stsp if (packdir_fd == -1) {
1483 1124fe40 2021-07-07 stsp if (errno == ENOENT)
1484 1124fe40 2021-07-07 stsp return NULL;
1485 1124fe40 2021-07-07 stsp return got_error_from_errno_fmt("openat: %s/%s",
1486 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1487 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR);
1490 1124fe40 2021-07-07 stsp packdir = fdopendir(packdir_fd);
1491 1124fe40 2021-07-07 stsp if (packdir == NULL) {
1492 1124fe40 2021-07-07 stsp err = got_error_from_errno("fdopendir");
1493 1124fe40 2021-07-07 stsp goto done;
1496 1124fe40 2021-07-07 stsp while ((dent = readdir(packdir)) != NULL) {
1497 1124fe40 2021-07-07 stsp if (cancel_cb) {
1498 1124fe40 2021-07-07 stsp err = cancel_cb(cancel_arg);
1500 1124fe40 2021-07-07 stsp goto done;
1503 1124fe40 2021-07-07 stsp if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1506 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath,
1507 1124fe40 2021-07-07 stsp dent->d_name);
1509 1124fe40 2021-07-07 stsp goto done;
1511 1124fe40 2021-07-07 stsp if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1512 1124fe40 2021-07-07 stsp free(pack_relpath);
1513 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1516 1124fe40 2021-07-07 stsp if (errno != ENOENT) {
1517 1124fe40 2021-07-07 stsp err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1518 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1519 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1520 1124fe40 2021-07-07 stsp pack_relpath);
1521 1124fe40 2021-07-07 stsp goto done;
1524 1124fe40 2021-07-07 stsp if (!dry_run) {
1525 3bf0e21f 2023-06-19 op if (unlinkat(packdir_fd, dent->d_name, 0) == -1) {
1526 3bf0e21f 2023-06-19 op err = got_error_from_errno("unlinkat");
1527 1124fe40 2021-07-07 stsp goto done;
1530 1124fe40 2021-07-07 stsp if (progress_cb) {
1531 1124fe40 2021-07-07 stsp char *path;
1532 1124fe40 2021-07-07 stsp if (asprintf(&path, "%s/%s/%s",
1533 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1534 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1535 1124fe40 2021-07-07 stsp dent->d_name) == -1) {
1536 1124fe40 2021-07-07 stsp err = got_error_from_errno("asprintf");
1537 1124fe40 2021-07-07 stsp goto done;
1539 1124fe40 2021-07-07 stsp err = progress_cb(progress_arg, path);
1540 1124fe40 2021-07-07 stsp free(path);
1542 1124fe40 2021-07-07 stsp goto done;
1544 1124fe40 2021-07-07 stsp free(pack_relpath);
1545 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1548 1124fe40 2021-07-07 stsp if (packdir && closedir(packdir) != 0 && err == NULL)
1549 1124fe40 2021-07-07 stsp err = got_error_from_errno("closedir");
1550 1124fe40 2021-07-07 stsp free(pack_relpath);
1551 b3d68e7f 2021-07-03 stsp return err;
