2 05118f5a 2021-06-22 stsp * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
4 05118f5a 2021-06-22 stsp * Permission to use, copy, modify, and distribute this software for any
5 05118f5a 2021-06-22 stsp * purpose with or without fee is hereby granted, provided that the above
6 05118f5a 2021-06-22 stsp * copyright notice and this permission notice appear in all copies.
8 05118f5a 2021-06-22 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 05118f5a 2021-06-22 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 05118f5a 2021-06-22 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 05118f5a 2021-06-22 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 05118f5a 2021-06-22 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 05118f5a 2021-06-22 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 05118f5a 2021-06-22 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 05118f5a 2021-06-22 stsp #include <sys/types.h>
18 05118f5a 2021-06-22 stsp #include <sys/uio.h>
19 05118f5a 2021-06-22 stsp #include <sys/stat.h>
20 05118f5a 2021-06-22 stsp #include <sys/socket.h>
21 05118f5a 2021-06-22 stsp #include <sys/wait.h>
23 05118f5a 2021-06-22 stsp #include <dirent.h>
24 c8d1f14f 2021-06-23 naddy #include <endian.h>
25 05118f5a 2021-06-22 stsp #include <errno.h>
26 05118f5a 2021-06-22 stsp #include <fcntl.h>
27 05118f5a 2021-06-22 stsp #include <stdint.h>
28 05118f5a 2021-06-22 stsp #include <sha1.h>
29 05118f5a 2021-06-22 stsp #include <stdio.h>
30 05118f5a 2021-06-22 stsp #include <stdlib.h>
31 05118f5a 2021-06-22 stsp #include <string.h>
32 05118f5a 2021-06-22 stsp #include <limits.h>
33 05118f5a 2021-06-22 stsp #include <unistd.h>
35 05118f5a 2021-06-22 stsp #include "got_error.h"
36 05118f5a 2021-06-22 stsp #include "got_cancel.h"
37 05118f5a 2021-06-22 stsp #include "got_object.h"
38 05118f5a 2021-06-22 stsp #include "got_reference.h"
39 05118f5a 2021-06-22 stsp #include "got_repository.h"
40 05118f5a 2021-06-22 stsp #include "got_repository_admin.h"
41 05118f5a 2021-06-22 stsp #include "got_opentemp.h"
42 05118f5a 2021-06-22 stsp #include "got_path.h"
44 05118f5a 2021-06-22 stsp #include "got_lib_delta.h"
45 05118f5a 2021-06-22 stsp #include "got_lib_object.h"
46 b3d68e7f 2021-07-03 stsp #include "got_lib_object_idset.h"
47 05118f5a 2021-06-22 stsp #include "got_lib_object_cache.h"
48 05118f5a 2021-06-22 stsp #include "got_lib_pack.h"
49 05118f5a 2021-06-22 stsp #include "got_lib_privsep.h"
50 05118f5a 2021-06-22 stsp #include "got_lib_repository.h"
51 05118f5a 2021-06-22 stsp #include "got_lib_pack_create.h"
52 05118f5a 2021-06-22 stsp #include "got_lib_sha1.h"
53 b3d68e7f 2021-07-03 stsp #include "got_lib_lockfile.h"
55 05118f5a 2021-06-22 stsp #ifndef nitems
56 05118f5a 2021-06-22 stsp #define nitems(_a) (sizeof((_a)) / sizeof((_a)[0]))
59 05118f5a 2021-06-22 stsp static const struct got_error *
60 05118f5a 2021-06-22 stsp get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
61 05118f5a 2021-06-22 stsp unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
62 05118f5a 2021-06-22 stsp struct got_repository *repo,
63 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
65 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
66 05118f5a 2021-06-22 stsp const size_t alloc_chunksz = 256;
67 05118f5a 2021-06-22 stsp size_t nalloc;
68 05118f5a 2021-06-22 stsp struct got_reflist_entry *re;
71 05118f5a 2021-06-22 stsp *ids = NULL;
72 05118f5a 2021-06-22 stsp *nobjects = 0;
74 05118f5a 2021-06-22 stsp *ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
75 05118f5a 2021-06-22 stsp if (*ids == NULL)
76 05118f5a 2021-06-22 stsp return got_error_from_errno("reallocarray");
77 05118f5a 2021-06-22 stsp nalloc = alloc_chunksz;
79 05118f5a 2021-06-22 stsp TAILQ_FOREACH(re, refs, entry) {
80 05118f5a 2021-06-22 stsp struct got_object_id *id;
82 05118f5a 2021-06-22 stsp if (cancel_cb) {
83 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
88 05118f5a 2021-06-22 stsp err = got_ref_resolve(&id, repo, re->ref);
92 05118f5a 2021-06-22 stsp if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
93 05118f5a 2021-06-22 stsp int obj_type;
94 05118f5a 2021-06-22 stsp err = got_object_get_type(&obj_type, repo, id);
97 05118f5a 2021-06-22 stsp if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
104 ae23ce34 2021-07-01 stsp if (nalloc <= *nobjects) {
105 05118f5a 2021-06-22 stsp struct got_object_id **new;
106 05118f5a 2021-06-22 stsp new = recallocarray(*ids, nalloc,
107 05118f5a 2021-06-22 stsp nalloc + alloc_chunksz,
108 05118f5a 2021-06-22 stsp sizeof(struct got_object_id *));
109 05118f5a 2021-06-22 stsp if (new == NULL) {
110 05118f5a 2021-06-22 stsp err = got_error_from_errno(
111 05118f5a 2021-06-22 stsp "recallocarray");
114 05118f5a 2021-06-22 stsp *ids = new;
115 05118f5a 2021-06-22 stsp nalloc += alloc_chunksz;
117 05118f5a 2021-06-22 stsp (*ids)[*nobjects] = id;
118 05118f5a 2021-06-22 stsp if ((*ids)[*nobjects] == NULL) {
119 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
122 05118f5a 2021-06-22 stsp (*nobjects)++;
126 05118f5a 2021-06-22 stsp for (i = 0; i < *nobjects; i++)
127 05118f5a 2021-06-22 stsp free((*ids)[i]);
128 05118f5a 2021-06-22 stsp free(*ids);
129 05118f5a 2021-06-22 stsp *ids = NULL;
130 05118f5a 2021-06-22 stsp *nobjects = 0;
132 05118f5a 2021-06-22 stsp return err;
135 05118f5a 2021-06-22 stsp const struct got_error *
136 05118f5a 2021-06-22 stsp got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
137 05118f5a 2021-06-22 stsp struct got_reflist_head *include_refs,
138 05118f5a 2021-06-22 stsp struct got_reflist_head *exclude_refs, struct got_repository *repo,
139 05118f5a 2021-06-22 stsp int loose_obj_only, got_pack_progress_cb progress_cb, void *progress_arg,
140 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
142 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
143 05118f5a 2021-06-22 stsp struct got_object_id **ours = NULL, **theirs = NULL;
144 05118f5a 2021-06-22 stsp int nours = 0, ntheirs = 0, packfd = -1, i;
145 05118f5a 2021-06-22 stsp char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
146 05118f5a 2021-06-22 stsp char *sha1_str = NULL;
148 05118f5a 2021-06-22 stsp *packfile = NULL;
149 05118f5a 2021-06-22 stsp *pack_hash = NULL;
151 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/packing.pack",
152 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
153 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
156 05118f5a 2021-06-22 stsp err = got_opentemp_named_fd(&tmpfile_path, &packfd, path);
160 05118f5a 2021-06-22 stsp if (fchmod(packfd, GOT_DEFAULT_FILE_MODE) != 0) {
161 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpfile_path);
165 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "w");
166 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
167 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", tmpfile_path);
170 05118f5a 2021-06-22 stsp packfd = -1;
172 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&ours, &nours,
173 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
174 abc59930 2021-09-05 naddy include_refs, repo, cancel_cb, cancel_arg);
178 05118f5a 2021-06-22 stsp if (nours == 0) {
179 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_CANNOT_PACK);
183 05118f5a 2021-06-22 stsp if (!TAILQ_EMPTY(exclude_refs)) {
184 05118f5a 2021-06-22 stsp err = get_reflist_object_ids(&theirs, &ntheirs,
185 abc59930 2021-09-05 naddy (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
186 abc59930 2021-09-05 naddy exclude_refs, repo,
187 abc59930 2021-09-05 naddy cancel_cb, cancel_arg);
192 05118f5a 2021-06-22 stsp *pack_hash = calloc(1, sizeof(**pack_hash));
193 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
194 05118f5a 2021-06-22 stsp err = got_error_from_errno("calloc");
198 05118f5a 2021-06-22 stsp err = got_pack_create((*pack_hash)->sha1, *packfile, theirs, ntheirs,
199 f8a36e22 2021-08-26 stsp ours, nours, repo, loose_obj_only, 0, progress_cb, progress_arg,
200 05118f5a 2021-06-22 stsp cancel_cb, cancel_arg);
204 05118f5a 2021-06-22 stsp err = got_object_id_str(&sha1_str, *pack_hash);
207 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
208 05118f5a 2021-06-22 stsp got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
209 05118f5a 2021-06-22 stsp sha1_str) == -1) {
210 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
214 05118f5a 2021-06-22 stsp if (fflush(*packfile) == -1) {
215 05118f5a 2021-06-22 stsp err = got_error_from_errno("fflush");
218 05118f5a 2021-06-22 stsp if (fseek(*packfile, 0L, SEEK_SET) == -1) {
219 05118f5a 2021-06-22 stsp err = got_error_from_errno("fseek");
222 05118f5a 2021-06-22 stsp if (rename(tmpfile_path, packfile_path) == -1) {
223 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpfile_path,
224 05118f5a 2021-06-22 stsp packfile_path);
227 05118f5a 2021-06-22 stsp free(tmpfile_path);
228 05118f5a 2021-06-22 stsp tmpfile_path = NULL;
230 05118f5a 2021-06-22 stsp for (i = 0; i < nours; i++)
231 05118f5a 2021-06-22 stsp free(ours[i]);
232 05118f5a 2021-06-22 stsp free(ours);
233 05118f5a 2021-06-22 stsp for (i = 0; i < ntheirs; i++)
234 05118f5a 2021-06-22 stsp free(theirs[i]);
235 05118f5a 2021-06-22 stsp free(theirs);
236 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
237 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
238 05118f5a 2021-06-22 stsp if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
239 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpfile_path);
240 05118f5a 2021-06-22 stsp free(tmpfile_path);
241 05118f5a 2021-06-22 stsp free(packfile_path);
242 05118f5a 2021-06-22 stsp free(sha1_str);
243 05118f5a 2021-06-22 stsp free(path);
245 05118f5a 2021-06-22 stsp free(*pack_hash);
246 05118f5a 2021-06-22 stsp *pack_hash = NULL;
247 05118f5a 2021-06-22 stsp if (*packfile)
248 05118f5a 2021-06-22 stsp fclose(*packfile);
249 05118f5a 2021-06-22 stsp *packfile = NULL;
251 05118f5a 2021-06-22 stsp return err;
254 05118f5a 2021-06-22 stsp const struct got_error *
255 05118f5a 2021-06-22 stsp got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
256 05118f5a 2021-06-22 stsp struct got_repository *repo,
257 05118f5a 2021-06-22 stsp got_pack_index_progress_cb progress_cb, void *progress_arg,
258 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
261 05118f5a 2021-06-22 stsp char *path;
262 05118f5a 2021-06-22 stsp int imsg_idxfds[2];
263 05118f5a 2021-06-22 stsp int npackfd = -1, idxfd = -1, nidxfd = -1;
264 05118f5a 2021-06-22 stsp int tmpfds[3];
265 05118f5a 2021-06-22 stsp int idxstatus, done = 0;
266 05118f5a 2021-06-22 stsp const struct got_error *err;
267 05118f5a 2021-06-22 stsp struct imsgbuf idxibuf;
268 05118f5a 2021-06-22 stsp pid_t idxpid;
269 05118f5a 2021-06-22 stsp char *tmpidxpath = NULL;
270 05118f5a 2021-06-22 stsp char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
271 05118f5a 2021-06-22 stsp const char *repo_path = got_repo_get_path_git_dir(repo);
272 05118f5a 2021-06-22 stsp struct stat sb;
274 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++)
275 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
277 05118f5a 2021-06-22 stsp if (asprintf(&path, "%s/%s/indexing.idx",
278 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
279 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
282 05118f5a 2021-06-22 stsp err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path);
283 05118f5a 2021-06-22 stsp free(path);
286 05118f5a 2021-06-22 stsp if (fchmod(idxfd, GOT_DEFAULT_FILE_MODE) != 0) {
287 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fchmod", tmpidxpath);
291 05118f5a 2021-06-22 stsp nidxfd = dup(idxfd);
292 05118f5a 2021-06-22 stsp if (nidxfd == -1) {
293 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
297 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
298 05118f5a 2021-06-22 stsp tmpfds[i] = got_opentempfd();
299 05118f5a 2021-06-22 stsp if (tmpfds[i] == -1) {
300 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_opentempfd");
305 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
309 05118f5a 2021-06-22 stsp if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
310 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
311 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
315 05118f5a 2021-06-22 stsp if (fstat(fileno(packfile), &sb) == -1) {
316 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fstat", packfile_path);
320 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
321 05118f5a 2021-06-22 stsp repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
322 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
326 05118f5a 2021-06-22 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
327 05118f5a 2021-06-22 stsp err = got_error_from_errno("socketpair");
330 05118f5a 2021-06-22 stsp idxpid = fork();
331 05118f5a 2021-06-22 stsp if (idxpid == -1) {
332 05118f5a 2021-06-22 stsp err= got_error_from_errno("fork");
334 05118f5a 2021-06-22 stsp } else if (idxpid == 0)
335 05118f5a 2021-06-22 stsp got_privsep_exec_child(imsg_idxfds,
336 05118f5a 2021-06-22 stsp GOT_PATH_PROG_INDEX_PACK, packfile_path);
337 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[1]) == -1) {
338 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
341 05118f5a 2021-06-22 stsp imsg_init(&idxibuf, imsg_idxfds[0]);
343 05118f5a 2021-06-22 stsp npackfd = dup(fileno(packfile));
344 05118f5a 2021-06-22 stsp if (npackfd == -1) {
345 05118f5a 2021-06-22 stsp err = got_error_from_errno("dup");
348 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
350 05118f5a 2021-06-22 stsp if (err != NULL)
352 05118f5a 2021-06-22 stsp npackfd = -1;
353 05118f5a 2021-06-22 stsp err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
354 05118f5a 2021-06-22 stsp if (err != NULL)
356 05118f5a 2021-06-22 stsp nidxfd = -1;
357 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
358 05118f5a 2021-06-22 stsp err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
359 05118f5a 2021-06-22 stsp if (err != NULL)
361 05118f5a 2021-06-22 stsp tmpfds[i] = -1;
364 05118f5a 2021-06-22 stsp while (!done) {
365 05118f5a 2021-06-22 stsp int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
367 05118f5a 2021-06-22 stsp if (cancel_cb) {
368 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
373 05118f5a 2021-06-22 stsp err = got_privsep_recv_index_progress(&done, &nobj_total,
374 05118f5a 2021-06-22 stsp &nobj_indexed, &nobj_loose, &nobj_resolved,
376 05118f5a 2021-06-22 stsp if (err != NULL)
378 05118f5a 2021-06-22 stsp if (nobj_indexed != 0) {
379 05118f5a 2021-06-22 stsp err = progress_cb(progress_arg, sb.st_size,
380 05118f5a 2021-06-22 stsp nobj_total, nobj_indexed, nobj_loose,
381 05118f5a 2021-06-22 stsp nobj_resolved);
385 05118f5a 2021-06-22 stsp imsg_clear(&idxibuf);
387 05118f5a 2021-06-22 stsp if (close(imsg_idxfds[0]) == -1) {
388 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
391 05118f5a 2021-06-22 stsp if (waitpid(idxpid, &idxstatus, 0) == -1) {
392 05118f5a 2021-06-22 stsp err = got_error_from_errno("waitpid");
396 05118f5a 2021-06-22 stsp if (rename(tmpidxpath, idxpath) == -1) {
397 05118f5a 2021-06-22 stsp err = got_error_from_errno3("rename", tmpidxpath, idxpath);
400 05118f5a 2021-06-22 stsp free(tmpidxpath);
401 05118f5a 2021-06-22 stsp tmpidxpath = NULL;
404 05118f5a 2021-06-22 stsp if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
405 05118f5a 2021-06-22 stsp err = got_error_from_errno2("unlink", tmpidxpath);
406 05118f5a 2021-06-22 stsp if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
407 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
408 05118f5a 2021-06-22 stsp if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
409 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
410 05118f5a 2021-06-22 stsp for (i = 0; i < nitems(tmpfds); i++) {
411 05118f5a 2021-06-22 stsp if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
412 05118f5a 2021-06-22 stsp err = got_error_from_errno("close");
414 05118f5a 2021-06-22 stsp free(tmpidxpath);
415 05118f5a 2021-06-22 stsp free(idxpath);
416 05118f5a 2021-06-22 stsp free(packfile_path);
417 05118f5a 2021-06-22 stsp return err;
420 05118f5a 2021-06-22 stsp const struct got_error *
421 05118f5a 2021-06-22 stsp got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
422 05118f5a 2021-06-22 stsp struct got_repository *repo, const char *packfile_path)
424 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
425 05118f5a 2021-06-22 stsp const char *packdir_path = NULL;
426 05118f5a 2021-06-22 stsp char *packfile_name = NULL, *p, *dot;
427 05118f5a 2021-06-22 stsp struct got_object_id id;
428 05118f5a 2021-06-22 stsp int packfd = -1;
430 05118f5a 2021-06-22 stsp *packfile = NULL;
431 05118f5a 2021-06-22 stsp *pack_hash = NULL;
433 05118f5a 2021-06-22 stsp packdir_path = got_repo_get_path_objects_pack(repo);
434 05118f5a 2021-06-22 stsp if (packdir_path == NULL)
435 05118f5a 2021-06-22 stsp return got_error_from_errno("got_repo_get_path_objects_pack");
437 05118f5a 2021-06-22 stsp if (!got_path_is_child(packfile_path, packdir_path,
438 05118f5a 2021-06-22 stsp strlen(packdir_path))) {
439 05118f5a 2021-06-22 stsp err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
444 05118f5a 2021-06-22 stsp err = got_path_basename(&packfile_name, packfile_path);
447 05118f5a 2021-06-22 stsp p = packfile_name;
449 05118f5a 2021-06-22 stsp if (strncmp(p, "pack-", 5) != 0) {
450 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
451 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
452 abc59930 2021-09-05 naddy packfile_name);
456 05118f5a 2021-06-22 stsp dot = strchr(p, '.');
457 05118f5a 2021-06-22 stsp if (dot == NULL) {
458 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
459 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
460 abc59930 2021-09-05 naddy packfile_name);
463 05118f5a 2021-06-22 stsp if (strcmp(dot + 1, "pack") != 0) {
464 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
465 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
466 abc59930 2021-09-05 naddy packfile_name);
469 05118f5a 2021-06-22 stsp *dot = '\0';
470 05118f5a 2021-06-22 stsp if (!got_parse_sha1_digest(id.sha1, p)) {
471 05118f5a 2021-06-22 stsp err = got_error_fmt(GOT_ERR_BAD_PATH,
472 abc59930 2021-09-05 naddy "'%s' is not a valid pack file name",
473 abc59930 2021-09-05 naddy packfile_name);
477 05118f5a 2021-06-22 stsp *pack_hash = got_object_id_dup(&id);
478 05118f5a 2021-06-22 stsp if (*pack_hash == NULL) {
479 05118f5a 2021-06-22 stsp err = got_error_from_errno("got_object_id_dup");
483 05118f5a 2021-06-22 stsp packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW);
484 05118f5a 2021-06-22 stsp if (packfd == -1) {
485 05118f5a 2021-06-22 stsp err = got_error_from_errno2("open", packfile_path);
489 05118f5a 2021-06-22 stsp *packfile = fdopen(packfd, "r");
490 05118f5a 2021-06-22 stsp if (*packfile == NULL) {
491 05118f5a 2021-06-22 stsp err = got_error_from_errno2("fdopen", packfile_path);
494 05118f5a 2021-06-22 stsp packfd = -1;
496 05118f5a 2021-06-22 stsp if (packfd != -1 && close(packfd) == -1 && err == NULL)
497 05118f5a 2021-06-22 stsp err = got_error_from_errno2("close", packfile_path);
498 05118f5a 2021-06-22 stsp free(packfile_name);
500 05118f5a 2021-06-22 stsp free(*pack_hash);
501 05118f5a 2021-06-22 stsp *pack_hash = NULL;
503 05118f5a 2021-06-22 stsp return err;
506 05118f5a 2021-06-22 stsp const struct got_error *
507 05118f5a 2021-06-22 stsp got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
508 05118f5a 2021-06-22 stsp struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
509 05118f5a 2021-06-22 stsp got_cancel_cb cancel_cb, void *cancel_arg)
511 05118f5a 2021-06-22 stsp const struct got_error *err = NULL;
512 05118f5a 2021-06-22 stsp char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
513 05118f5a 2021-06-22 stsp struct got_packidx *packidx = NULL;
514 05118f5a 2021-06-22 stsp struct got_pack *pack = NULL;
515 05118f5a 2021-06-22 stsp uint32_t nobj, i;
517 05118f5a 2021-06-22 stsp err = got_object_id_str(&id_str, pack_hash);
521 05118f5a 2021-06-22 stsp if (asprintf(&packpath, "%s/pack-%s.pack",
522 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
523 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
526 05118f5a 2021-06-22 stsp if (asprintf(&idxpath, "%s/pack-%s.idx",
527 05118f5a 2021-06-22 stsp GOT_OBJECTS_PACK_DIR, id_str) == -1) {
528 05118f5a 2021-06-22 stsp err = got_error_from_errno("asprintf");
532 05118f5a 2021-06-22 stsp err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
536 05118f5a 2021-06-22 stsp err = got_repo_cache_pack(&pack, repo, packpath, packidx);
540 05118f5a 2021-06-22 stsp nobj = be32toh(packidx->hdr.fanout_table[0xff]);
541 05118f5a 2021-06-22 stsp for (i = 0; i < nobj; i++) {
542 05118f5a 2021-06-22 stsp struct got_packidx_object_id *oid;
543 05118f5a 2021-06-22 stsp struct got_object_id id, base_id;
544 05118f5a 2021-06-22 stsp off_t offset, base_offset = 0;
545 05118f5a 2021-06-22 stsp uint8_t type;
546 05118f5a 2021-06-22 stsp uint64_t size;
547 05118f5a 2021-06-22 stsp size_t tslen, len;
549 05118f5a 2021-06-22 stsp if (cancel_cb) {
550 05118f5a 2021-06-22 stsp err = cancel_cb(cancel_arg);
554 05118f5a 2021-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
555 05118f5a 2021-06-22 stsp memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
557 05118f5a 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, i);
558 05118f5a 2021-06-22 stsp if (offset == -1) {
559 05118f5a 2021-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
563 05118f5a 2021-06-22 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
564 05118f5a 2021-06-22 stsp pack, offset);
568 05118f5a 2021-06-22 stsp switch (type) {
569 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
570 05118f5a 2021-06-22 stsp err = got_pack_parse_offset_delta(&base_offset, &len,
571 05118f5a 2021-06-22 stsp pack, offset, tslen);
575 05118f5a 2021-06-22 stsp case GOT_OBJ_TYPE_REF_DELTA:
576 05118f5a 2021-06-22 stsp err = got_pack_parse_ref_delta(&base_id,
577 05118f5a 2021-06-22 stsp pack, offset, tslen);
582 05118f5a 2021-06-22 stsp err = (*list_cb)(list_arg, &id, type, offset, size,
583 05118f5a 2021-06-22 stsp base_offset, &base_id);
589 05118f5a 2021-06-22 stsp free(id_str);
590 05118f5a 2021-06-22 stsp free(idxpath);
591 05118f5a 2021-06-22 stsp free(packpath);
592 05118f5a 2021-06-22 stsp if (packidx)
593 05118f5a 2021-06-22 stsp got_packidx_close(packidx);
594 b3d68e7f 2021-07-03 stsp return err;
597 b3d68e7f 2021-07-03 stsp static const struct got_error *
598 b3d68e7f 2021-07-03 stsp get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
599 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg,
600 b3d68e7f 2021-07-03 stsp struct got_repository *repo)
602 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
603 b3d68e7f 2021-07-03 stsp char *path_objects = NULL, *path = NULL;
604 b3d68e7f 2021-07-03 stsp DIR *dir = NULL;
605 b3d68e7f 2021-07-03 stsp struct got_object *obj = NULL;
606 b3d68e7f 2021-07-03 stsp struct got_object_id id;
607 b3d68e7f 2021-07-03 stsp int i, fd = -1;
608 b3d68e7f 2021-07-03 stsp struct stat sb;
610 b3d68e7f 2021-07-03 stsp *ondisk_size = 0;
611 b3d68e7f 2021-07-03 stsp *loose_ids = got_object_idset_alloc();
612 b3d68e7f 2021-07-03 stsp if (*loose_ids == NULL)
613 b3d68e7f 2021-07-03 stsp return got_error_from_errno("got_object_idset_alloc");
615 b3d68e7f 2021-07-03 stsp path_objects = got_repo_get_path_objects(repo);
616 b3d68e7f 2021-07-03 stsp if (path_objects == NULL) {
617 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_repo_get_path_objects");
621 b3d68e7f 2021-07-03 stsp for (i = 0; i <= 0xff; i++) {
622 b3d68e7f 2021-07-03 stsp struct dirent *dent;
624 b3d68e7f 2021-07-03 stsp if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
625 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
629 b3d68e7f 2021-07-03 stsp dir = opendir(path);
630 b3d68e7f 2021-07-03 stsp if (dir == NULL) {
631 b3d68e7f 2021-07-03 stsp if (errno == ENOENT) {
632 b3d68e7f 2021-07-03 stsp err = NULL;
635 b3d68e7f 2021-07-03 stsp err = got_error_from_errno2("opendir", path);
639 b3d68e7f 2021-07-03 stsp while ((dent = readdir(dir)) != NULL) {
640 b3d68e7f 2021-07-03 stsp char *id_str;
642 b3d68e7f 2021-07-03 stsp if (strcmp(dent->d_name, ".") == 0 ||
643 b3d68e7f 2021-07-03 stsp strcmp(dent->d_name, "..") == 0)
646 b3d68e7f 2021-07-03 stsp if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
647 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
651 b3d68e7f 2021-07-03 stsp memset(&id, 0, sizeof(id));
652 b3d68e7f 2021-07-03 stsp if (!got_parse_sha1_digest(id.sha1, id_str)) {
653 b3d68e7f 2021-07-03 stsp free(id_str);
656 b3d68e7f 2021-07-03 stsp free(id_str);
658 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, &id, repo);
661 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
662 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
665 d5c81d44 2021-07-08 stsp err = got_object_read_header_privsep(&obj, &id, repo,
669 b3d68e7f 2021-07-03 stsp fd = -1; /* already closed */
671 b3d68e7f 2021-07-03 stsp switch (obj->type) {
672 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
673 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
674 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_BLOB:
675 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
678 b3d68e7f 2021-07-03 stsp err = got_error_fmt(GOT_ERR_OBJ_TYPE,
679 b3d68e7f 2021-07-03 stsp "%d", obj->type);
682 b3d68e7f 2021-07-03 stsp got_object_close(obj);
683 b3d68e7f 2021-07-03 stsp obj = NULL;
684 b3d68e7f 2021-07-03 stsp (*ondisk_size) += sb.st_size;
685 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(*loose_ids, &id, NULL);
688 b3d68e7f 2021-07-03 stsp if (progress_cb) {
689 b3d68e7f 2021-07-03 stsp err = progress_cb(progress_arg,
690 b3d68e7f 2021-07-03 stsp got_object_idset_num_elements(*loose_ids),
697 b3d68e7f 2021-07-03 stsp if (closedir(dir) != 0) {
698 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
701 b3d68e7f 2021-07-03 stsp dir = NULL;
703 b3d68e7f 2021-07-03 stsp free(path);
704 b3d68e7f 2021-07-03 stsp path = NULL;
707 b3d68e7f 2021-07-03 stsp if (dir && closedir(dir) != 0 && err == NULL)
708 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("closedir");
709 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
710 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
712 b3d68e7f 2021-07-03 stsp got_object_idset_free(*loose_ids);
713 b3d68e7f 2021-07-03 stsp *loose_ids = NULL;
716 b3d68e7f 2021-07-03 stsp got_object_close(obj);
717 b3d68e7f 2021-07-03 stsp free(path_objects);
718 b3d68e7f 2021-07-03 stsp free(path);
719 b3d68e7f 2021-07-03 stsp return err;
722 b3d68e7f 2021-07-03 stsp static const struct got_error *
723 b3d68e7f 2021-07-03 stsp preserve_loose_object(struct got_object_idset *loose_ids,
724 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo, int *npacked)
726 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
727 9614da0d 2021-07-04 stsp struct got_object *obj;
729 b3d68e7f 2021-07-03 stsp if (!got_object_idset_contains(loose_ids, id))
730 b3d68e7f 2021-07-03 stsp return NULL;
733 9614da0d 2021-07-04 stsp * Try to open this object from a pack file. This ensures that
734 9614da0d 2021-07-04 stsp * we do in fact have a valid packed copy of the object. Otherwise
735 9614da0d 2021-07-04 stsp * we should not delete the loose representation of this object.
737 abc59930 2021-09-05 naddy err = got_object_open_packed(&obj, id, repo);
738 abc59930 2021-09-05 naddy if (err == NULL) {
739 9614da0d 2021-07-04 stsp got_object_close(obj);
741 9614da0d 2021-07-04 stsp * The object is referenced and packed.
742 9614da0d 2021-07-04 stsp * We can purge the redundantly stored loose object.
744 9614da0d 2021-07-04 stsp (*npacked)++;
745 9614da0d 2021-07-04 stsp return NULL;
746 9614da0d 2021-07-04 stsp } else if (err->code != GOT_ERR_NO_OBJ)
747 9614da0d 2021-07-04 stsp return err;
750 b3d68e7f 2021-07-03 stsp * This object is referenced and not packed.
751 b3d68e7f 2021-07-03 stsp * Remove it from our purge set.
753 b3d68e7f 2021-07-03 stsp return got_object_idset_remove(NULL, loose_ids, id);
756 b3d68e7f 2021-07-03 stsp static const struct got_error *
757 b3d68e7f 2021-07-03 stsp load_tree_entries(struct got_object_id_queue *ids,
758 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids,
759 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
760 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
761 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
763 b3d68e7f 2021-07-03 stsp const struct got_error *err;
764 b3d68e7f 2021-07-03 stsp struct got_tree_object *tree;
765 b3d68e7f 2021-07-03 stsp char *p = NULL;
768 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tree(&tree, repo, tree_id);
770 b3d68e7f 2021-07-03 stsp return err;
772 b3d68e7f 2021-07-03 stsp for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
773 b3d68e7f 2021-07-03 stsp struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
774 b3d68e7f 2021-07-03 stsp struct got_object_id *id = got_tree_entry_get_id(e);
775 b3d68e7f 2021-07-03 stsp mode_t mode = got_tree_entry_get_mode(e);
777 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
778 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
783 b3d68e7f 2021-07-03 stsp if (got_object_tree_entry_is_symlink(e) ||
784 b3d68e7f 2021-07-03 stsp got_object_tree_entry_is_submodule(e) ||
785 b3d68e7f 2021-07-03 stsp got_object_idset_contains(traversed_ids, id))
788 b3d68e7f 2021-07-03 stsp if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
789 b3d68e7f 2021-07-03 stsp got_tree_entry_get_name(e)) == -1) {
790 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("asprintf");
794 b3d68e7f 2021-07-03 stsp if (S_ISDIR(mode)) {
795 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
796 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
799 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(ids, qid, entry);
800 b3d68e7f 2021-07-03 stsp } else if (S_ISREG(mode)) {
801 b3d68e7f 2021-07-03 stsp /* This blob is referenced. */
802 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, id, repo,
806 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, id, NULL);
815 b3d68e7f 2021-07-03 stsp got_object_tree_close(tree);
817 b3d68e7f 2021-07-03 stsp return err;
820 b3d68e7f 2021-07-03 stsp static const struct got_error *
821 b3d68e7f 2021-07-03 stsp load_tree(struct got_object_idset *loose_ids,
822 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
823 b3d68e7f 2021-07-03 stsp const char *dpath, struct got_repository *repo, int *npacked,
824 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
826 b3d68e7f 2021-07-03 stsp const struct got_error *err = NULL;
827 b3d68e7f 2021-07-03 stsp struct got_object_id_queue tree_ids;
828 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
830 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, tree_id);
832 b3d68e7f 2021-07-03 stsp return err;
834 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&tree_ids);
835 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
837 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&tree_ids)) {
838 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
839 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
844 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&tree_ids);
845 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&tree_ids, entry);
847 b3d68e7f 2021-07-03 stsp if (got_object_idset_contains(traversed_ids, qid->id)) {
848 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
852 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, qid->id, NULL);
854 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
858 b3d68e7f 2021-07-03 stsp /* This tree is referenced. */
859 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
863 b3d68e7f 2021-07-03 stsp err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
864 b3d68e7f 2021-07-03 stsp qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
865 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
870 b3d68e7f 2021-07-03 stsp got_object_id_queue_free(&tree_ids);
871 b3d68e7f 2021-07-03 stsp return err;
874 b3d68e7f 2021-07-03 stsp static const struct got_error *
875 b3d68e7f 2021-07-03 stsp load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
876 b3d68e7f 2021-07-03 stsp int *npacked, struct got_object_idset *traversed_ids,
877 b3d68e7f 2021-07-03 stsp struct got_object_id *id, struct got_repository *repo,
878 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb, void *progress_arg, int nloose,
879 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
881 b3d68e7f 2021-07-03 stsp const struct got_error *err;
882 b3d68e7f 2021-07-03 stsp struct got_commit_object *commit = NULL;
883 b3d68e7f 2021-07-03 stsp struct got_tag_object *tag = NULL;
884 b3d68e7f 2021-07-03 stsp struct got_object_id *tree_id = NULL;
885 b3d68e7f 2021-07-03 stsp struct got_object_id_queue ids;
886 b3d68e7f 2021-07-03 stsp struct got_object_qid *qid;
887 b3d68e7f 2021-07-03 stsp int obj_type;
889 b3d68e7f 2021-07-03 stsp err = got_object_qid_alloc(&qid, id);
891 b3d68e7f 2021-07-03 stsp return err;
893 b3d68e7f 2021-07-03 stsp STAILQ_INIT(&ids);
894 b3d68e7f 2021-07-03 stsp STAILQ_INSERT_TAIL(&ids, qid, entry);
896 b3d68e7f 2021-07-03 stsp while (!STAILQ_EMPTY(&ids)) {
897 b3d68e7f 2021-07-03 stsp if (cancel_cb) {
898 b3d68e7f 2021-07-03 stsp err = (*cancel_cb)(cancel_arg);
903 b3d68e7f 2021-07-03 stsp qid = STAILQ_FIRST(&ids);
904 b3d68e7f 2021-07-03 stsp STAILQ_REMOVE_HEAD(&ids, entry);
906 b3d68e7f 2021-07-03 stsp if (got_object_idset_contains(traversed_ids, qid->id)) {
907 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
908 b3d68e7f 2021-07-03 stsp qid = NULL;
912 b3d68e7f 2021-07-03 stsp err = got_object_idset_add(traversed_ids, qid->id, NULL);
916 b3d68e7f 2021-07-03 stsp /* This commit or tag is referenced. */
917 b3d68e7f 2021-07-03 stsp err = preserve_loose_object(loose_ids, qid->id, repo, npacked);
921 b3d68e7f 2021-07-03 stsp err = got_object_get_type(&obj_type, repo, qid->id);
924 b3d68e7f 2021-07-03 stsp switch (obj_type) {
925 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
926 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo, qid->id);
930 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TAG:
931 b3d68e7f 2021-07-03 stsp err = got_object_open_as_tag(&tag, repo, qid->id);
936 b3d68e7f 2021-07-03 stsp /* should not happen */
937 b3d68e7f 2021-07-03 stsp err = got_error(GOT_ERR_OBJ_TYPE);
941 b3d68e7f 2021-07-03 stsp /* Find a tree object to scan. */
942 b3d68e7f 2021-07-03 stsp if (commit) {
943 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
944 b3d68e7f 2021-07-03 stsp } else if (tag) {
945 b3d68e7f 2021-07-03 stsp obj_type = got_object_tag_get_object_type(tag);
946 b3d68e7f 2021-07-03 stsp switch (obj_type) {
947 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_COMMIT:
948 b3d68e7f 2021-07-03 stsp err = got_object_open_as_commit(&commit, repo,
949 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
952 b3d68e7f 2021-07-03 stsp tree_id = got_object_commit_get_tree_id(commit);
954 b3d68e7f 2021-07-03 stsp case GOT_OBJ_TYPE_TREE:
955 b3d68e7f 2021-07-03 stsp tree_id = got_object_tag_get_object_id(tag);
959 b3d68e7f 2021-07-03 stsp * Tag points at something other than a
960 b3d68e7f 2021-07-03 stsp * commit or tree. Leave this weird tag object
961 b3d68e7f 2021-07-03 stsp * and the object it points to on disk.
963 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
965 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
967 b3d68e7f 2021-07-03 stsp err = got_object_idset_remove(NULL, loose_ids,
968 b3d68e7f 2021-07-03 stsp got_object_tag_get_object_id(tag));
969 b3d68e7f 2021-07-03 stsp if (err && err->code != GOT_ERR_NO_OBJ)
971 b3d68e7f 2021-07-03 stsp err = NULL;
976 b3d68e7f 2021-07-03 stsp if (tree_id) {
977 b3d68e7f 2021-07-03 stsp err = load_tree(loose_ids, traversed_ids, tree_id, "",
978 b3d68e7f 2021-07-03 stsp repo, npacked, cancel_cb, cancel_arg);
983 b3d68e7f 2021-07-03 stsp if (commit || tag)
984 b3d68e7f 2021-07-03 stsp (*ncommits)++; /* scanned tags are counted as commits */
986 b3d68e7f 2021-07-03 stsp if (progress_cb) {
987 b3d68e7f 2021-07-03 stsp err = progress_cb(progress_arg, nloose, *ncommits, -1);
992 b3d68e7f 2021-07-03 stsp if (commit) {
993 b3d68e7f 2021-07-03 stsp /* Find parent commits to scan. */
994 b3d68e7f 2021-07-03 stsp const struct got_object_id_queue *parent_ids;
995 b3d68e7f 2021-07-03 stsp parent_ids = got_object_commit_get_parent_ids(commit);
996 b3d68e7f 2021-07-03 stsp err = got_object_id_queue_copy(parent_ids, &ids);
999 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1000 b3d68e7f 2021-07-03 stsp commit = NULL;
1002 b3d68e7f 2021-07-03 stsp if (tag) {
1003 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1004 b3d68e7f 2021-07-03 stsp tag = NULL;
1006 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1007 b3d68e7f 2021-07-03 stsp qid = NULL;
1011 b3d68e7f 2021-07-03 stsp got_object_qid_free(qid);
1012 b3d68e7f 2021-07-03 stsp if (commit)
1013 b3d68e7f 2021-07-03 stsp got_object_commit_close(commit);
1015 b3d68e7f 2021-07-03 stsp got_object_tag_close(tag);
1016 4b2e47fb 2021-07-03 stsp got_object_id_queue_free(&ids);
1017 b3d68e7f 2021-07-03 stsp return err;
1020 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg {
1021 b3d68e7f 2021-07-03 stsp struct got_repository *repo;
1022 b3d68e7f 2021-07-03 stsp got_cleanup_progress_cb progress_cb;
1023 b3d68e7f 2021-07-03 stsp void *progress_arg;
1024 b3d68e7f 2021-07-03 stsp int nloose;
1025 b3d68e7f 2021-07-03 stsp int ncommits;
1026 b3d68e7f 2021-07-03 stsp int npurged;
1027 b3d68e7f 2021-07-03 stsp off_t size_purged;
1028 b3d68e7f 2021-07-03 stsp int dry_run;
1029 ef8ec606 2021-07-27 stsp time_t max_mtime;
1030 ef8ec606 2021-07-27 stsp int ignore_mtime;
1033 b3d68e7f 2021-07-03 stsp static const struct got_error *
1034 b3d68e7f 2021-07-03 stsp purge_loose_object(struct got_object_id *id, void *data, void *arg)
1036 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg *a = arg;
1037 b3d68e7f 2021-07-03 stsp const struct got_error *err, *unlock_err = NULL;
1038 b3d68e7f 2021-07-03 stsp char *path = NULL;
1039 b3d68e7f 2021-07-03 stsp int fd = -1;
1040 b3d68e7f 2021-07-03 stsp struct stat sb;
1041 b3d68e7f 2021-07-03 stsp struct got_lockfile *lf = NULL;
1043 b3d68e7f 2021-07-03 stsp err = got_object_get_path(&path, id, a->repo);
1045 b3d68e7f 2021-07-03 stsp return err;
1047 b3d68e7f 2021-07-03 stsp err = got_object_open_loose_fd(&fd, id, a->repo);
1049 b3d68e7f 2021-07-03 stsp goto done;
1051 b3d68e7f 2021-07-03 stsp if (fstat(fd, &sb) == -1) {
1052 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("fstat");
1053 b3d68e7f 2021-07-03 stsp goto done;
1057 ef8ec606 2021-07-27 stsp * Do not delete objects which are younger than our maximum
1058 ef8ec606 2021-07-27 stsp * modification time threshold. This prevents a race where
1059 ef8ec606 2021-07-27 stsp * new objects which are being added to the repository
1060 ef8ec606 2021-07-27 stsp * concurrently would be deleted.
1062 ef8ec606 2021-07-27 stsp if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1063 ef8ec606 2021-07-27 stsp if (!a->dry_run) {
1064 ef8ec606 2021-07-27 stsp err = got_lockfile_lock(&lf, path, -1);
1066 ef8ec606 2021-07-27 stsp goto done;
1067 ef8ec606 2021-07-27 stsp if (unlink(path) == -1) {
1068 ef8ec606 2021-07-27 stsp err = got_error_from_errno2("unlink", path);
1069 ef8ec606 2021-07-27 stsp goto done;
1073 ef8ec606 2021-07-27 stsp a->npurged++;
1074 ef8ec606 2021-07-27 stsp a->size_purged += sb.st_size;
1075 ef8ec606 2021-07-27 stsp if (a->progress_cb) {
1076 ef8ec606 2021-07-27 stsp err = a->progress_cb(a->progress_arg, a->nloose,
1077 ef8ec606 2021-07-27 stsp a->ncommits, a->npurged);
1081 b3d68e7f 2021-07-03 stsp if (fd != -1 && close(fd) == -1 && err == NULL)
1082 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("close");
1083 b3d68e7f 2021-07-03 stsp free(path);
1085 5345b4c7 2021-07-06 stsp unlock_err = got_lockfile_unlock(lf, -1);
1086 b3d68e7f 2021-07-03 stsp return err ? err : unlock_err;
1089 b3d68e7f 2021-07-03 stsp const struct got_error *
1090 b3d68e7f 2021-07-03 stsp got_repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1091 b3d68e7f 2021-07-03 stsp off_t *size_before, off_t *size_after, int *npacked, int dry_run,
1092 ef8ec606 2021-07-27 stsp int ignore_mtime, got_cleanup_progress_cb progress_cb, void *progress_arg,
1093 b3d68e7f 2021-07-03 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1095 b3d68e7f 2021-07-03 stsp const struct got_error *err;
1096 b3d68e7f 2021-07-03 stsp struct got_object_idset *loose_ids;
1097 b3d68e7f 2021-07-03 stsp struct got_object_idset *traversed_ids;
1098 b3d68e7f 2021-07-03 stsp struct got_object_id **referenced_ids;
1099 b3d68e7f 2021-07-03 stsp int i, nreferenced, nloose, ncommits = 0;
1100 b3d68e7f 2021-07-03 stsp struct got_reflist_head refs;
1101 ef8ec606 2021-07-27 stsp struct got_reflist_entry *re;
1102 b3d68e7f 2021-07-03 stsp struct purge_loose_object_arg arg;
1103 ef8ec606 2021-07-27 stsp time_t max_mtime = 0;
1105 b3d68e7f 2021-07-03 stsp TAILQ_INIT(&refs);
1107 b3d68e7f 2021-07-03 stsp *size_before = 0;
1108 b3d68e7f 2021-07-03 stsp *size_after = 0;
1109 b3d68e7f 2021-07-03 stsp *npacked = 0;
1111 b3d68e7f 2021-07-03 stsp err = get_loose_object_ids(&loose_ids, size_before,
1112 b3d68e7f 2021-07-03 stsp progress_cb, progress_arg, repo);
1114 b3d68e7f 2021-07-03 stsp return err;
1115 b3d68e7f 2021-07-03 stsp nloose = got_object_idset_num_elements(loose_ids);
1116 b3d68e7f 2021-07-03 stsp if (nloose == 0) {
1117 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1118 b3d68e7f 2021-07-03 stsp return NULL;
1121 b3d68e7f 2021-07-03 stsp traversed_ids = got_object_idset_alloc();
1122 b3d68e7f 2021-07-03 stsp if (traversed_ids == NULL) {
1123 b3d68e7f 2021-07-03 stsp err = got_error_from_errno("got_object_idset_alloc");
1124 b3d68e7f 2021-07-03 stsp goto done;
1127 b3d68e7f 2021-07-03 stsp err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1129 b3d68e7f 2021-07-03 stsp goto done;
1130 ef8ec606 2021-07-27 stsp if (!ignore_mtime) {
1131 ef8ec606 2021-07-27 stsp TAILQ_FOREACH(re, &refs, entry) {
1132 ef8ec606 2021-07-27 stsp time_t mtime = got_ref_get_mtime(re->ref);
1133 ef8ec606 2021-07-27 stsp if (mtime > max_mtime)
1134 ef8ec606 2021-07-27 stsp max_mtime = mtime;
1137 ef8ec606 2021-07-27 stsp * For safety, keep objects created within 10 minutes
1138 ef8ec606 2021-07-27 stsp * before the youngest reference was created.
1140 ef8ec606 2021-07-27 stsp if (max_mtime >= 600)
1141 ef8ec606 2021-07-27 stsp max_mtime -= 600;
1144 b3d68e7f 2021-07-03 stsp err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1145 b3d68e7f 2021-07-03 stsp (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1146 b3d68e7f 2021-07-03 stsp &refs, repo, cancel_cb, cancel_arg);
1148 b3d68e7f 2021-07-03 stsp goto done;
1150 b3d68e7f 2021-07-03 stsp for (i = 0; i < nreferenced; i++) {
1151 b3d68e7f 2021-07-03 stsp struct got_object_id *id = referenced_ids[i];
1152 b3d68e7f 2021-07-03 stsp err = load_commit_or_tag(loose_ids, &ncommits, npacked,
1153 b3d68e7f 2021-07-03 stsp traversed_ids, id, repo, progress_cb, progress_arg, nloose,
1154 b3d68e7f 2021-07-03 stsp cancel_cb, cancel_arg);
1156 b3d68e7f 2021-07-03 stsp goto done;
1159 b3d68e7f 2021-07-03 stsp /* Produce a final progress report in case no objects can be purged. */
1160 b3d68e7f 2021-07-03 stsp if (got_object_idset_num_elements(loose_ids) == 0 && progress_cb) {
1161 b3d68e7f 2021-07-03 stsp err = progress_cb(progress_arg, nloose, ncommits, 0);
1163 b3d68e7f 2021-07-03 stsp goto done;
1166 b3d68e7f 2021-07-03 stsp /* Any remaining loose objects are unreferenced and can be purged. */
1167 b3d68e7f 2021-07-03 stsp arg.repo = repo;
1168 b3d68e7f 2021-07-03 stsp arg.progress_arg = progress_arg;
1169 b3d68e7f 2021-07-03 stsp arg.progress_cb = progress_cb;
1170 b3d68e7f 2021-07-03 stsp arg.nloose = nloose;
1171 b3d68e7f 2021-07-03 stsp arg.npurged = 0;
1172 b3d68e7f 2021-07-03 stsp arg.size_purged = 0;
1173 b3d68e7f 2021-07-03 stsp arg.ncommits = ncommits;
1174 b3d68e7f 2021-07-03 stsp arg.dry_run = dry_run;
1175 ef8ec606 2021-07-27 stsp arg.max_mtime = max_mtime;
1176 ef8ec606 2021-07-27 stsp arg.ignore_mtime = ignore_mtime;
1177 b3d68e7f 2021-07-03 stsp err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1179 b3d68e7f 2021-07-03 stsp goto done;
1180 b3d68e7f 2021-07-03 stsp *size_after = *size_before - arg.size_purged;
1182 b3d68e7f 2021-07-03 stsp got_object_idset_free(loose_ids);
1183 b3d68e7f 2021-07-03 stsp got_object_idset_free(traversed_ids);
1184 1124fe40 2021-07-07 stsp return err;
1187 1124fe40 2021-07-07 stsp static const struct got_error *
1188 1124fe40 2021-07-07 stsp remove_packidx(int dir_fd, const char *relpath)
1190 1124fe40 2021-07-07 stsp const struct got_error *err, *unlock_err;
1191 1124fe40 2021-07-07 stsp struct got_lockfile *lf;
1193 1124fe40 2021-07-07 stsp err = got_lockfile_lock(&lf, relpath, dir_fd);
1195 1124fe40 2021-07-07 stsp return err;
1196 1124fe40 2021-07-07 stsp if (unlinkat(dir_fd, relpath, 0) == -1)
1197 1124fe40 2021-07-07 stsp err = got_error_from_errno("unlinkat");
1198 1124fe40 2021-07-07 stsp unlock_err = got_lockfile_unlock(lf, dir_fd);
1199 1124fe40 2021-07-07 stsp return err ? err : unlock_err;
1202 1124fe40 2021-07-07 stsp const struct got_error *
1203 1124fe40 2021-07-07 stsp got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1204 1124fe40 2021-07-07 stsp got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1205 1124fe40 2021-07-07 stsp got_cancel_cb cancel_cb, void *cancel_arg)
1207 1124fe40 2021-07-07 stsp const struct got_error *err;
1208 1124fe40 2021-07-07 stsp DIR *packdir = NULL;
1209 1124fe40 2021-07-07 stsp struct dirent *dent;
1210 1124fe40 2021-07-07 stsp char *pack_relpath = NULL;
1211 1124fe40 2021-07-07 stsp int packdir_fd;
1212 1124fe40 2021-07-07 stsp struct stat sb;
1214 1124fe40 2021-07-07 stsp packdir_fd = openat(got_repo_get_fd(repo),
1215 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1216 1124fe40 2021-07-07 stsp if (packdir_fd == -1) {
1217 1124fe40 2021-07-07 stsp if (errno == ENOENT)
1218 1124fe40 2021-07-07 stsp return NULL;
1219 1124fe40 2021-07-07 stsp return got_error_from_errno_fmt("openat: %s/%s",
1220 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1221 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR);
1224 1124fe40 2021-07-07 stsp packdir = fdopendir(packdir_fd);
1225 1124fe40 2021-07-07 stsp if (packdir == NULL) {
1226 1124fe40 2021-07-07 stsp err = got_error_from_errno("fdopendir");
1227 1124fe40 2021-07-07 stsp goto done;
1230 1124fe40 2021-07-07 stsp while ((dent = readdir(packdir)) != NULL) {
1231 1124fe40 2021-07-07 stsp if (cancel_cb) {
1232 1124fe40 2021-07-07 stsp err = cancel_cb(cancel_arg);
1234 1124fe40 2021-07-07 stsp goto done;
1237 dd038bc6 2021-09-21 thomas.ad if (!got_repo_is_packidx_filename(dent->d_name,
1238 dd038bc6 2021-09-21 thomas.ad strlen(dent->d_name)))
1241 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath,
1242 1124fe40 2021-07-07 stsp dent->d_name);
1244 1124fe40 2021-07-07 stsp goto done;
1246 1124fe40 2021-07-07 stsp if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1247 1124fe40 2021-07-07 stsp free(pack_relpath);
1248 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1251 1124fe40 2021-07-07 stsp if (errno != ENOENT) {
1252 1124fe40 2021-07-07 stsp err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1253 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1254 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1255 1124fe40 2021-07-07 stsp pack_relpath);
1256 1124fe40 2021-07-07 stsp goto done;
1259 1124fe40 2021-07-07 stsp if (!dry_run) {
1260 1124fe40 2021-07-07 stsp err = remove_packidx(packdir_fd, dent->d_name);
1262 1124fe40 2021-07-07 stsp goto done;
1264 1124fe40 2021-07-07 stsp if (progress_cb) {
1265 1124fe40 2021-07-07 stsp char *path;
1266 1124fe40 2021-07-07 stsp if (asprintf(&path, "%s/%s/%s",
1267 1124fe40 2021-07-07 stsp got_repo_get_path_git_dir(repo),
1268 1124fe40 2021-07-07 stsp GOT_OBJECTS_PACK_DIR,
1269 1124fe40 2021-07-07 stsp dent->d_name) == -1) {
1270 1124fe40 2021-07-07 stsp err = got_error_from_errno("asprintf");
1271 1124fe40 2021-07-07 stsp goto done;
1273 1124fe40 2021-07-07 stsp err = progress_cb(progress_arg, path);
1274 1124fe40 2021-07-07 stsp free(path);
1276 1124fe40 2021-07-07 stsp goto done;
1278 1124fe40 2021-07-07 stsp free(pack_relpath);
1279 1124fe40 2021-07-07 stsp pack_relpath = NULL;
1282 1124fe40 2021-07-07 stsp if (packdir && closedir(packdir) != 0 && err == NULL)
1283 1124fe40 2021-07-07 stsp err = got_error_from_errno("closedir");
1284 1124fe40 2021-07-07 stsp free(pack_relpath);
1285 b3d68e7f 2021-07-03 stsp return err;
