1 /*
2  * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/stat.h>
22 #include <sys/socket.h>
23 #include <sys/wait.h>
24 
25 #include <dirent.h>
26 #include <endian.h>
27 #include <errno.h>
28 #include <fcntl.h>
29 #include <stdint.h>
30 #include <sha1.h>
31 #include <sha2.h>
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <limits.h>
36 #include <unistd.h>
37 #include <imsg.h>
38 
39 #include "got_error.h"
40 #include "got_cancel.h"
41 #include "got_object.h"
42 #include "got_reference.h"
43 #include "got_repository.h"
44 #include "got_repository_admin.h"
45 #include "got_opentemp.h"
46 #include "got_path.h"
47 
48 #include "got_lib_delta.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_idset.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_privsep.h"
54 #include "got_lib_repository.h"
55 #include "got_lib_ratelimit.h"
56 #include "got_lib_pack_create.h"
57 #include "got_lib_hash.h"
58 #include "got_lib_lockfile.h"
59 
60 #ifndef nitems
61 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
62 #endif
63 
64 static const struct got_error *
65 get_reflist_object_ids(struct got_object_id ***ids, int *nobjects,
66     unsigned int wanted_obj_type_mask, struct got_reflist_head *refs,
67     struct got_repository *repo,
68     got_cancel_cb cancel_cb, void *cancel_arg)
69 {
70 	const struct got_error *err = NULL;
71 	const size_t alloc_chunksz = 256;
72 	size_t nalloc;
73 	struct got_reflist_entry *re;
74 	int i;
75 
76 	*ids = NULL;
77 	*nobjects = 0;
78 
79 	err = got_reflist_sort(refs,
80 	    got_ref_cmp_by_commit_timestamp_descending, repo);
81 	if (err)
82 		return err;
83 
84 	*ids = reallocarray(NULL, alloc_chunksz, sizeof(struct got_object_id *));
85 	if (*ids == NULL)
86 		return got_error_from_errno("reallocarray");
87 	nalloc = alloc_chunksz;
88 
89 	TAILQ_FOREACH(re, refs, entry) {
90 		struct got_object_id *id;
91 
92 		if (cancel_cb) {
93 			err = cancel_cb(cancel_arg);
94 			if (err)
95 				goto done;
96 		}
97 
98 		err = got_ref_resolve(&id, repo, re->ref);
99 		if (err)
100 			goto done;
101 
102 		if (wanted_obj_type_mask != GOT_OBJ_TYPE_ANY) {
103 			int obj_type;
104 			err = got_object_get_type(&obj_type, repo, id);
105 			if (err)
106 				goto done;
107 			if ((wanted_obj_type_mask & (1 << obj_type)) == 0) {
108 				free(id);
109 				id = NULL;
110 				continue;
111 			}
112 		}
113 
114 		if (nalloc <= *nobjects) {
115 			struct got_object_id **new;
116 			new = recallocarray(*ids, nalloc,
117 			    nalloc + alloc_chunksz,
118 			    sizeof(struct got_object_id *));
119 			if (new == NULL) {
120 				err = got_error_from_errno(
121 				    "recallocarray");
122 				goto done;
123 			}
124 			*ids = new;
125 			nalloc += alloc_chunksz;
126 		}
127 		(*ids)[*nobjects] = id;
128 		if ((*ids)[*nobjects] == NULL) {
129 			err = got_error_from_errno("got_object_id_dup");
130 			goto done;
131 		}
132 		(*nobjects)++;
133 	}
134 done:
135 	if (err) {
136 		for (i = 0; i < *nobjects; i++)
137 			free((*ids)[i]);
138 		free(*ids);
139 		*ids = NULL;
140 		*nobjects = 0;
141 	}
142 	return err;
143 }
144 
145 const struct got_error *
146 got_repo_pack_objects(FILE **packfile, struct got_object_id **pack_hash,
147     struct got_reflist_head *include_refs,
148     struct got_reflist_head *exclude_refs, struct got_repository *repo,
149     int loose_obj_only, int force_refdelta,
150     got_pack_progress_cb progress_cb, void *progress_arg,
151     got_cancel_cb cancel_cb, void *cancel_arg)
152 {
153 	const struct got_error *err = NULL;
154 	struct got_object_id **ours = NULL, **theirs = NULL;
155 	int nours = 0, ntheirs = 0, packfd = -1, i;
156 	char *tmpfile_path = NULL, *path = NULL, *packfile_path = NULL;
157 	char *sha1_str = NULL;
158 	FILE *delta_cache = NULL;
159 	struct got_ratelimit rl;
160 
161 	*packfile = NULL;
162 	*pack_hash = NULL;
163 
164 	got_ratelimit_init(&rl, 0, 500);
165 
166 	if (asprintf(&path, "%s/%s/packing.pack",
167 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR) == -1) {
168 		err = got_error_from_errno("asprintf");
169 		goto done;
170 	}
171 	err = got_opentemp_named_fd(&tmpfile_path, &packfd, path, "");
172 	if (err)
173 		goto done;
174 
175 	if (fchmod(packfd, GOT_DEFAULT_PACK_MODE) == -1) {
176 		err = got_error_from_errno2("fchmod", tmpfile_path);
177 		goto done;
178 	}
179 
180 	delta_cache = got_opentemp();
181 	if (delta_cache == NULL) {
182 		err = got_error_from_errno("got_opentemp");
183 		goto done;
184 	}
185 
186 	err = get_reflist_object_ids(&ours, &nours,
187 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
188 	    include_refs, repo, cancel_cb, cancel_arg);
189 	if (err)
190 		goto done;
191 
192 	if (nours == 0) {
193 		err = got_error(GOT_ERR_CANNOT_PACK);
194 		goto done;
195 	}
196 
197 	if (!TAILQ_EMPTY(exclude_refs)) {
198 		err = get_reflist_object_ids(&theirs, &ntheirs,
199 		    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
200 		    exclude_refs, repo,
201 		    cancel_cb, cancel_arg);
202 		if (err)
203 			goto done;
204 	}
205 
206 	*pack_hash = calloc(1, sizeof(**pack_hash));
207 	if (*pack_hash == NULL) {
208 		err = got_error_from_errno("calloc");
209 		goto done;
210 	}
211 
212 	err = got_pack_create((*pack_hash)->sha1, packfd, delta_cache,
213 	    theirs, ntheirs, ours, nours, repo, loose_obj_only,
214 	    0, force_refdelta, progress_cb, progress_arg, &rl,
215 	    cancel_cb, cancel_arg);
216 	if (err)
217 		goto done;
218 
219 	err = got_object_id_str(&sha1_str, *pack_hash);
220 	if (err)
221 		goto done;
222 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
223 	    got_repo_get_path_git_dir(repo), GOT_OBJECTS_PACK_DIR,
224 	    sha1_str) == -1) {
225 		err = got_error_from_errno("asprintf");
226 		goto done;
227 	}
228 
229 	if (lseek(packfd, 0L, SEEK_SET) == -1) {
230 		err = got_error_from_errno("lseek");
231 		goto done;
232 	}
233 	if (rename(tmpfile_path, packfile_path) == -1) {
234 		err = got_error_from_errno3("rename", tmpfile_path,
235 		    packfile_path);
236 		goto done;
237 	}
238 	free(tmpfile_path);
239 	tmpfile_path = NULL;
240 
241 	*packfile = fdopen(packfd, "w");
242 	if (*packfile == NULL) {
243 		err = got_error_from_errno2("fdopen", tmpfile_path);
244 		goto done;
245 	}
246 	packfd = -1;
247 done:
248 	for (i = 0; i < nours; i++)
249 		free(ours[i]);
250 	free(ours);
251 	for (i = 0; i < ntheirs; i++)
252 		free(theirs[i]);
253 	free(theirs);
254 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
255 		err = got_error_from_errno2("close", packfile_path);
256 	if (delta_cache && fclose(delta_cache) == EOF && err == NULL)
257 		err = got_error_from_errno("fclose");
258 	if (tmpfile_path && unlink(tmpfile_path) == -1 && err == NULL)
259 		err = got_error_from_errno2("unlink", tmpfile_path);
260 	free(tmpfile_path);
261 	free(packfile_path);
262 	free(sha1_str);
263 	free(path);
264 	if (err) {
265 		free(*pack_hash);
266 		*pack_hash = NULL;
267 		if (*packfile)
268 			fclose(*packfile);
269 		*packfile = NULL;
270 	}
271 	return err;
272 }
273 
274 const struct got_error *
275 got_repo_index_pack(FILE *packfile, struct got_object_id *pack_hash,
276     struct got_repository *repo,
277     got_pack_index_progress_cb progress_cb, void *progress_arg,
278     got_cancel_cb cancel_cb, void *cancel_arg)
279 {
280 	size_t i;
281 	char *path;
282 	int imsg_idxfds[2];
283 	int npackfd = -1, idxfd = -1, nidxfd = -1;
284 	int tmpfds[3];
285 	int idxstatus, done = 0;
286 	const struct got_error *err;
287 	struct imsgbuf idxibuf;
288 	pid_t idxpid;
289 	char *tmpidxpath = NULL;
290 	char *packfile_path = NULL, *idxpath = NULL, *id_str = NULL;
291 	const char *repo_path = got_repo_get_path_git_dir(repo);
292 	struct stat sb;
293 
294 	for (i = 0; i < nitems(tmpfds); i++)
295 		tmpfds[i] = -1;
296 
297 	if (asprintf(&path, "%s/%s/indexing.idx",
298 	    repo_path, GOT_OBJECTS_PACK_DIR) == -1) {
299 		err = got_error_from_errno("asprintf");
300 		goto done;
301 	}
302 	err = got_opentemp_named_fd(&tmpidxpath, &idxfd, path, "");
303 	free(path);
304 	if (err)
305 		goto done;
306 	if (fchmod(idxfd, GOT_DEFAULT_PACK_MODE) == -1) {
307 		err = got_error_from_errno2("fchmod", tmpidxpath);
308 		goto done;
309 	}
310 
311 	nidxfd = dup(idxfd);
312 	if (nidxfd == -1) {
313 		err = got_error_from_errno("dup");
314 		goto done;
315 	}
316 
317 	for (i = 0; i < nitems(tmpfds); i++) {
318 		tmpfds[i] = got_opentempfd();
319 		if (tmpfds[i] == -1) {
320 			err = got_error_from_errno("got_opentempfd");
321 			goto done;
322 		}
323 	}
324 
325 	err = got_object_id_str(&id_str, pack_hash);
326 	if (err)
327 		goto done;
328 
329 	if (asprintf(&packfile_path, "%s/%s/pack-%s.pack",
330 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
331 		err = got_error_from_errno("asprintf");
332 		goto done;
333 	}
334 
335 	if (fstat(fileno(packfile), &sb) == -1) {
336 		err = got_error_from_errno2("fstat", packfile_path);
337 		goto done;
338 	}
339 
340 	if (asprintf(&idxpath, "%s/%s/pack-%s.idx",
341 	    repo_path, GOT_OBJECTS_PACK_DIR, id_str) == -1) {
342 		err = got_error_from_errno("asprintf");
343 		goto done;
344 	}
345 
346 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_idxfds) == -1) {
347 		err = got_error_from_errno("socketpair");
348 		goto done;
349 	}
350 	idxpid = fork();
351 	if (idxpid == -1) {
352 		err= got_error_from_errno("fork");
353 		goto done;
354 	} else if (idxpid == 0)
355 		got_privsep_exec_child(imsg_idxfds,
356 		    GOT_PATH_PROG_INDEX_PACK, packfile_path);
357 	if (close(imsg_idxfds[1]) == -1) {
358 		err = got_error_from_errno("close");
359 		goto done;
360 	}
361 	imsg_init(&idxibuf, imsg_idxfds[0]);
362 
363 	npackfd = dup(fileno(packfile));
364 	if (npackfd == -1) {
365 		err = got_error_from_errno("dup");
366 		goto done;
367 	}
368 	err = got_privsep_send_index_pack_req(&idxibuf, pack_hash->sha1,
369 	    npackfd);
370 	if (err != NULL)
371 		goto done;
372 	npackfd = -1;
373 	err = got_privsep_send_index_pack_outfd(&idxibuf, nidxfd);
374 	if (err != NULL)
375 		goto done;
376 	nidxfd = -1;
377 	for (i = 0; i < nitems(tmpfds); i++) {
378 		err = got_privsep_send_tmpfd(&idxibuf, tmpfds[i]);
379 		if (err != NULL)
380 			goto done;
381 		tmpfds[i] = -1;
382 	}
383 	done = 0;
384 	while (!done) {
385 		int nobj_total, nobj_indexed, nobj_loose, nobj_resolved;
386 
387 		if (cancel_cb) {
388 			err = cancel_cb(cancel_arg);
389 			if (err)
390 				goto done;
391 		}
392 
393 		err = got_privsep_recv_index_progress(&done, &nobj_total,
394 		    &nobj_indexed, &nobj_loose, &nobj_resolved,
395 		    &idxibuf);
396 		if (err != NULL)
397 			goto done;
398 		if (nobj_indexed != 0) {
399 			err = progress_cb(progress_arg, sb.st_size,
400 			    nobj_total, nobj_indexed, nobj_loose,
401 			    nobj_resolved);
402 			if (err)
403 				break;
404 		}
405 	}
406 	if (close(imsg_idxfds[0]) == -1) {
407 		err = got_error_from_errno("close");
408 		goto done;
409 	}
410 	if (waitpid(idxpid, &idxstatus, 0) == -1) {
411 		err = got_error_from_errno("waitpid");
412 		goto done;
413 	}
414 
415 	if (rename(tmpidxpath, idxpath) == -1) {
416 		err = got_error_from_errno3("rename", tmpidxpath, idxpath);
417 		goto done;
418 	}
419 	free(tmpidxpath);
420 	tmpidxpath = NULL;
421 
422 done:
423 	if (tmpidxpath && unlink(tmpidxpath) == -1 && err == NULL)
424 		err = got_error_from_errno2("unlink", tmpidxpath);
425 	if (npackfd != -1 && close(npackfd) == -1 && err == NULL)
426 		err = got_error_from_errno("close");
427 	if (idxfd != -1 && close(idxfd) == -1 && err == NULL)
428 		err = got_error_from_errno("close");
429 	for (i = 0; i < nitems(tmpfds); i++) {
430 		if (tmpfds[i] != -1 && close(tmpfds[i]) == -1 && err == NULL)
431 			err = got_error_from_errno("close");
432 	}
433 	free(tmpidxpath);
434 	free(idxpath);
435 	free(packfile_path);
436 	return err;
437 }
438 
439 const struct got_error *
440 got_repo_find_pack(FILE **packfile, struct got_object_id **pack_hash,
441     struct got_repository *repo, const char *packfile_path)
442 {
443 	const struct got_error *err = NULL;
444 	const char *packdir_path = NULL;
445 	char *packfile_name = NULL, *p, *dot;
446 	struct got_object_id id;
447 	int packfd = -1;
448 
449 	*packfile = NULL;
450 	*pack_hash = NULL;
451 
452 	packdir_path = got_repo_get_path_objects_pack(repo);
453 	if (packdir_path == NULL)
454 		return got_error_from_errno("got_repo_get_path_objects_pack");
455 
456 	if (!got_path_is_child(packfile_path, packdir_path,
457 	    strlen(packdir_path))) {
458 		err = got_error_path(packfile_path, GOT_ERR_BAD_PATH);
459 		goto done;
460 
461 	}
462 
463 	err = got_path_basename(&packfile_name, packfile_path);
464 	if (err)
465 		goto done;
466 	p = packfile_name;
467 
468 	if (strncmp(p, "pack-", 5) != 0) {
469 		err = got_error_fmt(GOT_ERR_BAD_PATH,
470 		    "'%s' is not a valid pack file name",
471 		    packfile_name);
472 		goto done;
473 	}
474 	p += 5;
475 	dot = strchr(p, '.');
476 	if (dot == NULL) {
477 		err = got_error_fmt(GOT_ERR_BAD_PATH,
478 		    "'%s' is not a valid pack file name",
479 		    packfile_name);
480 		goto done;
481 	}
482 	if (strcmp(dot + 1, "pack") != 0) {
483 		err = got_error_fmt(GOT_ERR_BAD_PATH,
484 		    "'%s' is not a valid pack file name",
485 		    packfile_name);
486 		goto done;
487 	}
488 	*dot = '\0';
489 	if (!got_parse_object_id(&id, p, repo->algo)) {
490 		err = got_error_fmt(GOT_ERR_BAD_PATH,
491 		    "'%s' is not a valid pack file name",
492 		    packfile_name);
493 		goto done;
494 	}
495 
496 	*pack_hash = got_object_id_dup(&id);
497 	if (*pack_hash == NULL) {
498 		err = got_error_from_errno("got_object_id_dup");
499 		goto done;
500 	}
501 
502 	packfd = open(packfile_path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
503 	if (packfd == -1) {
504 		err = got_error_from_errno2("open", packfile_path);
505 		goto done;
506 	}
507 
508 	*packfile = fdopen(packfd, "r");
509 	if (*packfile == NULL) {
510 		err = got_error_from_errno2("fdopen", packfile_path);
511 		goto done;
512 	}
513 	packfd = -1;
514 done:
515 	if (packfd != -1 && close(packfd) == -1 && err == NULL)
516 		err = got_error_from_errno2("close", packfile_path);
517 	free(packfile_name);
518 	if (err) {
519 		free(*pack_hash);
520 		*pack_hash = NULL;
521 	}
522 	return err;
523 }
524 
525 const struct got_error *
526 got_repo_list_pack(FILE *packfile, struct got_object_id *pack_hash,
527     struct got_repository *repo, got_pack_list_cb list_cb, void *list_arg,
528     got_cancel_cb cancel_cb, void *cancel_arg)
529 {
530 	const struct got_error *err = NULL;
531 	char *id_str = NULL, *idxpath = NULL, *packpath = NULL;
532 	struct got_packidx *packidx = NULL;
533 	struct got_pack *pack = NULL;
534 	uint32_t nobj, i;
535 
536 	err = got_object_id_str(&id_str, pack_hash);
537 	if (err)
538 		goto done;
539 
540 	if (asprintf(&packpath, "%s/pack-%s.pack",
541 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
542 		err = got_error_from_errno("asprintf");
543 		goto done;
544 	}
545 	if (asprintf(&idxpath, "%s/pack-%s.idx",
546 	    GOT_OBJECTS_PACK_DIR, id_str) == -1) {
547 		err = got_error_from_errno("asprintf");
548 		goto done;
549 	}
550 
551 	err = got_packidx_open(&packidx, got_repo_get_fd(repo), idxpath, 1);
552 	if (err)
553 		goto done;
554 
555 	err = got_repo_cache_pack(&pack, repo, packpath, packidx);
556 	if (err)
557 		goto done;
558 
559 	nobj = be32toh(packidx->hdr.fanout_table[0xff]);
560 	for (i = 0; i < nobj; i++) {
561 		struct got_packidx_object_id *oid;
562 		struct got_object_id id, base_id;
563 		off_t offset, base_offset = 0;
564 		uint8_t type;
565 		uint64_t size;
566 		size_t tslen, len;
567 
568 		if (cancel_cb) {
569 			err = cancel_cb(cancel_arg);
570 			if (err)
571 				break;
572 		}
573 		oid = &packidx->hdr.sorted_ids[i];
574 		memcpy(id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
575 
576 		offset = got_packidx_get_object_offset(packidx, i);
577 		if (offset == -1) {
578 			err = got_error(GOT_ERR_BAD_PACKIDX);
579 			goto done;
580 		}
581 
582 		err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
583 		    pack, offset);
584 		if (err)
585 			goto done;
586 
587 		switch (type) {
588 		case GOT_OBJ_TYPE_OFFSET_DELTA:
589 			err = got_pack_parse_offset_delta(&base_offset, &len,
590 			    pack, offset, tslen);
591 			if (err)
592 				goto done;
593 			break;
594 		case GOT_OBJ_TYPE_REF_DELTA:
595 			err = got_pack_parse_ref_delta(&base_id,
596 			    pack, offset, tslen);
597 			if (err)
598 				goto done;
599 			break;
600 		}
601 		err = (*list_cb)(list_arg, &id, type, offset, size,
602 		    base_offset, &base_id);
603 		if (err)
604 			goto done;
605 	}
606 
607 done:
608 	free(id_str);
609 	free(idxpath);
610 	free(packpath);
611 	if (packidx)
612 		got_packidx_close(packidx);
613 	return err;
614 }
615 
616 static const struct got_error *
617 repo_cleanup_lock(struct got_repository *repo, struct got_lockfile **lk)
618 {
619 	const struct got_error *err;
620 	char myname[_POSIX_HOST_NAME_MAX + 1];
621 
622 	if (gethostname(myname, sizeof(myname)) == -1)
623 		return got_error_from_errno("gethostname");
624 
625 	err = got_lockfile_lock(lk, "gc.pid", got_repo_get_fd(repo));
626 	if (err)
627 		return err;
628 
629 	/*
630 	 * Git uses these info to provide some verbiage when finds a
631 	 * lock during `git gc --force' so don't try too hard to avoid
632 	 * short writes and don't care if a race happens between the
633 	 * lockfile creation and the write itself.
634 	 */
635 	if (dprintf((*lk)->fd, "%d %s", getpid(), myname) < 0)
636 		return got_error_from_errno("dprintf");
637 
638 	return NULL;
639 }
640 
641 static const struct got_error *
642 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
643     void *progress_arg, struct got_ratelimit *rl,
644     int ncommits, int nloose, int npurged, int nredundant)
645 {
646 	const struct got_error *err;
647 	int elapsed;
648 
649 	if (progress_cb == NULL)
650 		return NULL;
651 
652 	err = got_ratelimit_check(&elapsed, rl);
653 	if (err || !elapsed)
654 		return err;
655 
656 	return progress_cb(progress_arg, ncommits, nloose, npurged,
657 	    nredundant);
658 }
659 
660 static const struct got_error *
661 get_loose_object_ids(struct got_object_idset **loose_ids,
662     off_t *ondisk_size, int ncommits,
663     got_cleanup_progress_cb progress_cb, void *progress_arg,
664     struct got_ratelimit *rl, struct got_repository *repo)
665 {
666 	const struct got_error *err = NULL;
667 	char *path_objects = NULL, *path = NULL;
668 	DIR *dir = NULL;
669 	struct got_object *obj = NULL;
670 	struct got_object_id id;
671 	int i, fd = -1;
672 	struct stat sb;
673 
674 	*ondisk_size = 0;
675 	*loose_ids = got_object_idset_alloc();
676 	if (*loose_ids == NULL)
677 		return got_error_from_errno("got_object_idset_alloc");
678 
679 	path_objects = got_repo_get_path_objects(repo);
680 	if (path_objects == NULL) {
681 		err = got_error_from_errno("got_repo_get_path_objects");
682 		goto done;
683 	}
684 
685 	for (i = 0; i <= 0xff; i++) {
686 		struct dirent *dent;
687 
688 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
689 			err = got_error_from_errno("asprintf");
690 			break;
691 		}
692 
693 		dir = opendir(path);
694 		if (dir == NULL) {
695 			if (errno == ENOENT) {
696 				err = NULL;
697 				continue;
698 			}
699 			err = got_error_from_errno2("opendir", path);
700 			break;
701 		}
702 
703 		while ((dent = readdir(dir)) != NULL) {
704 			char *id_str;
705 
706 			if (strcmp(dent->d_name, ".") == 0 ||
707 			    strcmp(dent->d_name, "..") == 0)
708 				continue;
709 
710 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
711 				err = got_error_from_errno("asprintf");
712 				goto done;
713 			}
714 
715 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
716 				free(id_str);
717 				continue;
718 			}
719 			free(id_str);
720 
721 			err = got_object_open_loose_fd(&fd, &id, repo);
722 			if (err)
723 				goto done;
724 			if (fstat(fd, &sb) == -1) {
725 				err = got_error_from_errno("fstat");
726 				goto done;
727 			}
728 			err = got_object_read_header_privsep(&obj, &id, repo,
729 			    fd);
730 			if (err)
731 				goto done;
732 			fd = -1; /* already closed */
733 
734 			switch (obj->type) {
735 			case GOT_OBJ_TYPE_COMMIT:
736 			case GOT_OBJ_TYPE_TREE:
737 			case GOT_OBJ_TYPE_BLOB:
738 			case GOT_OBJ_TYPE_TAG:
739 				break;
740 			default:
741 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
742 				    "%d", obj->type);
743 				goto done;
744 			}
745 			got_object_close(obj);
746 			obj = NULL;
747 			(*ondisk_size) += sb.st_size;
748 			err = got_object_idset_add(*loose_ids, &id, NULL);
749 			if (err)
750 				goto done;
751 			err = report_cleanup_progress(progress_cb,
752 			    progress_arg, rl, ncommits,
753 			    got_object_idset_num_elements(*loose_ids),
754 			    -1, -1);
755 			if (err)
756 				goto done;
757 		}
758 
759 		if (closedir(dir) != 0) {
760 			err = got_error_from_errno("closedir");
761 			goto done;
762 		}
763 		dir = NULL;
764 
765 		free(path);
766 		path = NULL;
767 	}
768 done:
769 	if (dir && closedir(dir) != 0 && err == NULL)
770 		err = got_error_from_errno("closedir");
771 	if (fd != -1 && close(fd) == -1 && err == NULL)
772 		err = got_error_from_errno("close");
773 	if (err) {
774 		got_object_idset_free(*loose_ids);
775 		*loose_ids = NULL;
776 	}
777 	if (obj)
778 		got_object_close(obj);
779 	free(path_objects);
780 	free(path);
781 	return err;
782 }
783 
784 static const struct got_error *
785 load_tree_entries(struct got_object_id_queue *ids,
786     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
787     const char *dpath, struct got_repository *repo,
788     got_cancel_cb cancel_cb, void *cancel_arg)
789 {
790 	const struct got_error *err;
791 	struct got_tree_object *tree;
792 	char *p = NULL;
793 	int i;
794 
795 	err = got_object_open_as_tree(&tree, repo, tree_id);
796 	if (err)
797 		return err;
798 
799 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
800 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
801 		struct got_object_id *id = got_tree_entry_get_id(e);
802 		mode_t mode = got_tree_entry_get_mode(e);
803 
804 		if (cancel_cb) {
805 			err = (*cancel_cb)(cancel_arg);
806 			if (err)
807 				break;
808 		}
809 
810 		if (got_object_tree_entry_is_symlink(e) ||
811 		    got_object_tree_entry_is_submodule(e) ||
812 		    got_object_idset_contains(traversed_ids, id))
813 			continue;
814 
815 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
816 		    got_tree_entry_get_name(e)) == -1) {
817 			err = got_error_from_errno("asprintf");
818 			break;
819 		}
820 
821 		if (S_ISDIR(mode)) {
822 			struct got_object_qid *qid;
823 			err = got_object_qid_alloc(&qid, id);
824 			if (err)
825 				break;
826 			STAILQ_INSERT_TAIL(ids, qid, entry);
827 		} else if (S_ISREG(mode)) {
828 			/* This blob is referenced. */
829 			err = got_object_idset_add(traversed_ids, id, NULL);
830 			if (err)
831 				break;
832 		}
833 		free(p);
834 		p = NULL;
835 	}
836 
837 	got_object_tree_close(tree);
838 	free(p);
839 	return err;
840 }
841 
842 static const struct got_error *
843 load_tree(struct got_object_idset *traversed_ids,
844     struct got_object_id *tree_id,
845     const char *dpath, struct got_repository *repo,
846     got_cancel_cb cancel_cb, void *cancel_arg)
847 {
848 	const struct got_error *err = NULL;
849 	struct got_object_id_queue tree_ids;
850 	struct got_object_qid *qid;
851 
852 	err = got_object_qid_alloc(&qid, tree_id);
853 	if (err)
854 		return err;
855 
856 	STAILQ_INIT(&tree_ids);
857 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
858 
859 	while (!STAILQ_EMPTY(&tree_ids)) {
860 		if (cancel_cb) {
861 			err = (*cancel_cb)(cancel_arg);
862 			if (err)
863 				break;
864 		}
865 
866 		qid = STAILQ_FIRST(&tree_ids);
867 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
868 
869 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
870 			got_object_qid_free(qid);
871 			continue;
872 		}
873 
874 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
875 		if (err) {
876 			got_object_qid_free(qid);
877 			break;
878 		}
879 
880 		err = load_tree_entries(&tree_ids, traversed_ids,
881 		    &qid->id, dpath, repo, cancel_cb, cancel_arg);
882 		got_object_qid_free(qid);
883 		if (err)
884 			break;
885 	}
886 
887 	got_object_id_queue_free(&tree_ids);
888 	return err;
889 }
890 
891 static const struct got_error *
892 load_commit_or_tag(int *ncommits, struct got_object_idset *traversed_ids,
893     struct got_object_id *id, struct got_repository *repo,
894     got_cleanup_progress_cb progress_cb, void *progress_arg,
895     struct got_ratelimit *rl, got_cancel_cb cancel_cb, void *cancel_arg)
896 {
897 	const struct got_error *err;
898 	struct got_commit_object *commit = NULL;
899 	struct got_tag_object *tag = NULL;
900 	struct got_object_id *tree_id = NULL;
901 	struct got_object_id_queue ids;
902 	struct got_object_qid *qid;
903 	int obj_type;
904 
905 	err = got_object_qid_alloc(&qid, id);
906 	if (err)
907 		return err;
908 
909 	STAILQ_INIT(&ids);
910 	STAILQ_INSERT_TAIL(&ids, qid, entry);
911 
912 	while (!STAILQ_EMPTY(&ids)) {
913 		if (cancel_cb) {
914 			err = (*cancel_cb)(cancel_arg);
915 			if (err)
916 				break;
917 		}
918 
919 		qid = STAILQ_FIRST(&ids);
920 		STAILQ_REMOVE_HEAD(&ids, entry);
921 
922 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
923 			got_object_qid_free(qid);
924 			qid = NULL;
925 			continue;
926 		}
927 
928 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
929 		if (err)
930 			break;
931 
932 		err = got_object_get_type(&obj_type, repo, &qid->id);
933 		if (err)
934 			break;
935 		switch (obj_type) {
936 		case GOT_OBJ_TYPE_COMMIT:
937 			err = got_object_open_as_commit(&commit, repo,
938 			    &qid->id);
939 			if (err)
940 				goto done;
941 			break;
942 		case GOT_OBJ_TYPE_TAG:
943 			err = got_object_open_as_tag(&tag, repo, &qid->id);
944 			if (err)
945 				goto done;
946 			break;
947 		default:
948 			/* should not happen */
949 			err = got_error(GOT_ERR_OBJ_TYPE);
950 			goto done;
951 		}
952 
953 		/* Find a tree object to scan. */
954 		if (commit) {
955 			tree_id = got_object_commit_get_tree_id(commit);
956 		} else if (tag) {
957 			obj_type = got_object_tag_get_object_type(tag);
958 			switch (obj_type) {
959 			case GOT_OBJ_TYPE_COMMIT:
960 				err = got_object_open_as_commit(&commit, repo,
961 				    got_object_tag_get_object_id(tag));
962 				if (err)
963 					goto done;
964 				tree_id = got_object_commit_get_tree_id(commit);
965 				break;
966 			case GOT_OBJ_TYPE_TREE:
967 				tree_id = got_object_tag_get_object_id(tag);
968 				break;
969 			default:
970 				/*
971 				 * Tag points at something other than a
972 				 * commit or tree. Leave this weird tag object
973 				 * and the object it points to.
974 				 */
975 				if (got_object_idset_contains(traversed_ids,
976 				    got_object_tag_get_object_id(tag)))
977 					break;
978 				err = got_object_idset_add(traversed_ids,
979 				    got_object_tag_get_object_id(tag), NULL);
980 				if (err)
981 					goto done;
982 				break;
983 			}
984 		}
985 
986 		if (tree_id) {
987 			err = load_tree(traversed_ids, tree_id, "",
988 			    repo, cancel_cb, cancel_arg);
989 			if (err)
990 				break;
991 		}
992 
993 		if (commit || tag)
994 			(*ncommits)++; /* scanned tags are counted as commits */
995 
996 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
997 		    *ncommits, -1, -1, -1);
998 		if (err)
999 			break;
1000 
1001 		if (commit) {
1002 			/* Find parent commits to scan. */
1003 			const struct got_object_id_queue *parent_ids;
1004 			parent_ids = got_object_commit_get_parent_ids(commit);
1005 			err = got_object_id_queue_copy(parent_ids, &ids);
1006 			if (err)
1007 				break;
1008 			got_object_commit_close(commit);
1009 			commit = NULL;
1010 		}
1011 		if (tag) {
1012 			got_object_tag_close(tag);
1013 			tag = NULL;
1014 		}
1015 		got_object_qid_free(qid);
1016 		qid = NULL;
1017 	}
1018 done:
1019 	if (qid)
1020 		got_object_qid_free(qid);
1021 	if (commit)
1022 		got_object_commit_close(commit);
1023 	if (tag)
1024 		got_object_tag_close(tag);
1025 	got_object_id_queue_free(&ids);
1026 	return err;
1027 }
1028 
1029 static const struct got_error *
1030 is_object_packed(int *packed, struct got_repository *repo,
1031     struct got_object_id *id)
1032 {
1033 	const struct got_error *err;
1034 	struct got_object *obj;
1035 
1036 	*packed = 0;
1037 
1038 	err = got_object_open_packed(&obj, id, repo);
1039 	if (err) {
1040 		if (err->code == GOT_ERR_NO_OBJ)
1041 			err = NULL;
1042 		return err;
1043 	}
1044 	got_object_close(obj);
1045 	*packed = 1;
1046 	return NULL;
1047 }
1048 
1049 struct purge_loose_object_arg {
1050 	struct got_repository *repo;
1051 	got_cleanup_progress_cb progress_cb;
1052 	void *progress_arg;
1053 	struct got_ratelimit *rl;
1054 	struct got_object_idset *traversed_ids;
1055 	int nloose;
1056 	int ncommits;
1057 	int npacked;
1058 	int npurged;
1059 	off_t size_purged;
1060 	int dry_run;
1061 	time_t max_mtime;
1062 	int ignore_mtime;
1063 };
1064 
1065 static const struct got_error *
1066 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1067 {
1068 	struct purge_loose_object_arg *a = arg;
1069 	const struct got_error *err, *unlock_err = NULL;
1070 	char *path = NULL;
1071 	int packed, fd = -1;
1072 	struct stat sb;
1073 	struct got_lockfile *lf = NULL;
1074 
1075 	err = is_object_packed(&packed, a->repo, id);
1076 	if (err)
1077 		return err;
1078 
1079 	if (!packed && got_object_idset_contains(a->traversed_ids, id))
1080 		return NULL;
1081 
1082 	if (packed)
1083 		a->npacked++;
1084 
1085 	err = got_object_get_path(&path, id, a->repo);
1086 	if (err)
1087 		return err;
1088 
1089 	err = got_object_open_loose_fd(&fd, id, a->repo);
1090 	if (err)
1091 		goto done;
1092 
1093 	if (fstat(fd, &sb) == -1) {
1094 		err = got_error_from_errno("fstat");
1095 		goto done;
1096 	}
1097 
1098 	/*
1099 	 * Do not delete objects which are younger than our maximum
1100 	 * modification time threshold. This prevents a race where
1101 	 * new objects which are being added to the repository
1102 	 * concurrently would be deleted.
1103 	 */
1104 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1105 		if (!a->dry_run) {
1106 			err = got_lockfile_lock(&lf, path, -1);
1107 			if (err)
1108 				goto done;
1109 			if (unlink(path) == -1) {
1110 				err = got_error_from_errno2("unlink", path);
1111 				goto done;
1112 			}
1113 		}
1114 
1115 		a->npurged++;
1116 		a->size_purged += sb.st_size;
1117 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1118 		    a->rl, a->ncommits, a->nloose, a->npurged, -1);
1119 		if (err)
1120 			goto done;
1121 	}
1122 done:
1123 	if (fd != -1 && close(fd) == -1 && err == NULL)
1124 		err = got_error_from_errno("close");
1125 	free(path);
1126 	if (lf)
1127 		unlock_err = got_lockfile_unlock(lf, -1);
1128 	return err ? err : unlock_err;
1129 }
1130 
1131 static const struct got_error *
1132 repo_purge_unreferenced_loose_objects(struct got_repository *repo,
1133     struct got_object_idset *traversed_ids,
1134     off_t *size_before, off_t *size_after, int ncommits, int *nloose,
1135     int *npacked, int *npurged, int dry_run, int ignore_mtime,
1136     time_t max_mtime, struct got_ratelimit *rl,
1137     got_cleanup_progress_cb progress_cb, void *progress_arg,
1138     got_cancel_cb cancel_cb, void *cancel_arg)
1139 {
1140 	const struct got_error *err;
1141 	struct got_object_idset *loose_ids;
1142 	struct purge_loose_object_arg arg;
1143 
1144 	err = get_loose_object_ids(&loose_ids, size_before, ncommits,
1145 	    progress_cb, progress_arg, rl, repo);
1146 	if (err)
1147 		return err;
1148 	*nloose = got_object_idset_num_elements(loose_ids);
1149 	if (*nloose == 0) {
1150 		got_object_idset_free(loose_ids);
1151 		if (progress_cb) {
1152 			err = progress_cb(progress_arg, 0, 0, 0, -1);
1153 			if (err)
1154 				return err;
1155 		}
1156 		return NULL;
1157 	}
1158 
1159 	memset(&arg, 0, sizeof(arg));
1160 	arg.repo = repo;
1161 	arg.progress_arg = progress_arg;
1162 	arg.progress_cb = progress_cb;
1163 	arg.rl = rl;
1164 	arg.traversed_ids = traversed_ids;
1165 	arg.nloose = *nloose;
1166 	arg.npacked = 0;
1167 	arg.npurged = 0;
1168 	arg.size_purged = 0;
1169 	arg.dry_run = dry_run;
1170 	arg.max_mtime = max_mtime;
1171 	arg.ignore_mtime = ignore_mtime;
1172 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1173 	if (err)
1174 		goto done;
1175 
1176 	*size_after = *size_before - arg.size_purged;
1177 	*npacked = arg.npacked;
1178 	*npurged = arg.npurged;
1179 
1180 	/* Produce a final progress report. */
1181 	if (progress_cb) {
1182 		err = progress_cb(progress_arg, ncommits, *nloose,
1183 		    arg.npurged, -1);
1184 		if (err)
1185 			goto done;
1186 	}
1187 done:
1188 	got_object_idset_free(loose_ids);
1189 	return err;
1190 }
1191 
1192 static const struct got_error *
1193 purge_redundant_pack(struct got_repository *repo, const char *packidx_path,
1194     int dry_run, int ignore_mtime, time_t max_mtime,
1195     int *remove, off_t *size_before, off_t *size_after)
1196 {
1197 	static const char *ext[] = {".idx", ".pack", ".rev", ".bitmap",
1198 	    ".promisor", ".mtimes"};
1199 	struct stat sb;
1200 	char *dot, path[PATH_MAX];
1201 	size_t i;
1202 
1203 	if (strlcpy(path, packidx_path, sizeof(path)) >= sizeof(path))
1204 		return got_error(GOT_ERR_NO_SPACE);
1205 
1206 	/*
1207 	 * Do not delete pack files which are younger than our maximum
1208 	 * modification time threshold.  This prevents a race where a
1209 	 * new pack file which is being added to the repository
1210 	 * concurrently would be deleted.
1211 	 */
1212 	if (fstatat(got_repo_get_fd(repo), path, &sb, 0) == -1) {
1213 		if (errno == ENOENT)
1214 			return NULL;
1215 		return got_error_from_errno2("fstatat", path);
1216 	}
1217 	if (!ignore_mtime && sb.st_mtime > max_mtime)
1218 		*remove = 0;
1219 
1220 	/*
1221 	 * For compatibility with Git, if a matching .keep file exist
1222 	 * don't delete the packfile.
1223 	 */
1224 	dot = strrchr(path, '.');
1225 	*dot = '\0';
1226 	if (strlcat(path, ".keep", sizeof(path)) >= sizeof(path))
1227 		return got_error(GOT_ERR_NO_SPACE);
1228 	if (faccessat(got_repo_get_fd(repo), path, F_OK, 0) == 0)
1229 		*remove = 0;
1230 
1231 	for (i = 0; i < nitems(ext); ++i) {
1232 		*dot = '\0';
1233 
1234 		if (strlcat(path, ext[i], sizeof(path)) >=
1235 		    sizeof(path))
1236 			return got_error(GOT_ERR_NO_SPACE);
1237 
1238 		if (fstatat(got_repo_get_fd(repo), path, &sb, 0) ==
1239 		    -1) {
1240 			if (errno == ENOENT)
1241 				continue;
1242 			return got_error_from_errno2("fstatat", path);
1243 		}
1244 
1245 		*size_before += sb.st_size;
1246 		if (!*remove) {
1247 			*size_after += sb.st_size;
1248 			continue;
1249 		}
1250 
1251 		if (dry_run)
1252 			continue;
1253 
1254 		if (unlinkat(got_repo_get_fd(repo), path, 0) == -1) {
1255 			if (errno == ENOENT)
1256 				continue;
1257 			return got_error_from_errno2("unlinkat",
1258 			    path);
1259 		}
1260 	}
1261 
1262 	return NULL;
1263 }
1264 
1265 static const struct got_error *
1266 pack_is_redundant(int *redundant, struct got_repository *repo,
1267     struct got_object_idset *traversed_ids,
1268     const char *packidx_path, struct got_object_idset *idset)
1269 {
1270 	const struct got_error *err;
1271 	struct got_packidx *packidx;
1272 	struct got_packidx_object_id *pid;
1273 	struct got_object_id id;
1274 	size_t i, nobjects;
1275 
1276 	*redundant = 1;
1277 
1278 	err = got_repo_get_packidx(&packidx, packidx_path, repo);
1279 	if (err)
1280 		return err;
1281 
1282 	nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1283 	for (i = 0; i < nobjects; ++i) {
1284 		pid = &packidx->hdr.sorted_ids[i];
1285 
1286 		memset(&id, 0, sizeof(id));
1287 		memcpy(&id.sha1, pid->sha1, sizeof(id.sha1));
1288 
1289 		if (got_object_idset_contains(idset, &id))
1290 			continue;
1291 
1292 		if (!got_object_idset_contains(traversed_ids, &id))
1293 			continue;
1294 
1295 		*redundant = 0;
1296 		err = got_object_idset_add(idset, &id, NULL);
1297 		if (err)
1298 			return err;
1299 	}
1300 
1301 	return NULL;
1302 }
1303 
1304 struct pack_info {
1305 	const char	*path;
1306 	size_t		 nobjects;
1307 };
1308 
1309 static int
1310 pack_info_cmp(const void *a, const void *b)
1311 {
1312 	const struct pack_info	*pa, *pb;
1313 
1314 	pa = a;
1315 	pb = b;
1316 	if (pa->nobjects == pb->nobjects)
1317 		return strcmp(pa->path, pb->path);
1318 	if (pa->nobjects > pb->nobjects)
1319 		return -1;
1320 	return 1;
1321 }
1322 
1323 static const struct got_error *
1324 repo_purge_redundant_packfiles(struct got_repository *repo,
1325     struct got_object_idset *traversed_ids,
1326     off_t *size_before, off_t *size_after, int dry_run, int ignore_mtime,
1327     time_t max_mtime, int nloose, int ncommits, int npurged,
1328     struct got_ratelimit *rl,
1329     got_cleanup_progress_cb progress_cb, void *progress_arg,
1330     got_cancel_cb cancel_cb, void *cancel_arg)
1331 {
1332 	const struct got_error *err;
1333 	struct pack_info *pinfo, *sorted = NULL;
1334 	struct got_packidx *packidx;
1335 	struct got_object_idset *idset = NULL;
1336 	struct got_pathlist_entry *pe;
1337 	size_t i, npacks;
1338 	int remove, redundant_packs = 0;
1339 
1340 	npacks = 0;
1341 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
1342 		npacks++;
1343 
1344 	if (npacks == 0)
1345 		return NULL;
1346 
1347 	sorted = calloc(npacks, sizeof(*sorted));
1348 	if (sorted == NULL)
1349 		return got_error_from_errno("calloc");
1350 
1351 	i = 0;
1352 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1353 		err = got_repo_get_packidx(&packidx, pe->path, repo);
1354 		if (err)
1355 			goto done;
1356 
1357 		pinfo = &sorted[i++];
1358 		pinfo->path = pe->path;
1359 		pinfo->nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1360 	}
1361 	qsort(sorted, npacks, sizeof(*sorted), pack_info_cmp);
1362 
1363 	idset = got_object_idset_alloc();
1364 	if (idset == NULL) {
1365 		err = got_error_from_errno("got_object_idset_alloc");
1366 		goto done;
1367 	}
1368 
1369 	for (i = 0; i < npacks; ++i) {
1370 		if (cancel_cb) {
1371 			err = (*cancel_cb)(cancel_arg);
1372 			if (err)
1373 				break;
1374 		}
1375 
1376 		err = pack_is_redundant(&remove, repo, traversed_ids,
1377 		    sorted[i].path, idset);
1378 		if (err)
1379 			goto done;
1380 		err = purge_redundant_pack(repo, sorted[i].path, dry_run,
1381 		    ignore_mtime, max_mtime, &remove, size_before, size_after);
1382 		if (err)
1383 			goto done;
1384 		if (!remove)
1385 			continue;
1386 		err = report_cleanup_progress(progress_cb, progress_arg,
1387 		    rl, ncommits, nloose, npurged, ++redundant_packs);
1388 		if (err)
1389 			goto done;
1390 	}
1391 
1392 	/* Produce a final progress report. */
1393 	if (progress_cb) {
1394 		err = progress_cb(progress_arg, ncommits, nloose, npurged,
1395 		    redundant_packs);
1396 		if (err)
1397 			goto done;
1398 	}
1399  done:
1400 	free(sorted);
1401 	if (idset)
1402 		got_object_idset_free(idset);
1403 	return err;
1404 }
1405 
1406 const struct got_error *
1407 got_repo_cleanup(struct got_repository *repo,
1408     off_t *loose_before, off_t *loose_after,
1409     off_t *pack_before, off_t *pack_after,
1410     int *ncommits, int *nloose, int *npacked, int dry_run, int ignore_mtime,
1411     got_cleanup_progress_cb progress_cb, void *progress_arg,
1412     got_cancel_cb cancel_cb, void *cancel_arg)
1413 {
1414 	const struct got_error *unlock_err, *err = NULL;
1415 	struct got_lockfile *lk = NULL;
1416 	struct got_ratelimit rl;
1417 	struct got_reflist_head refs;
1418 	struct got_object_idset *traversed_ids = NULL;
1419 	struct got_reflist_entry *re;
1420 	struct got_object_id **referenced_ids;
1421 	int i, nreferenced;
1422 	int npurged = 0;
1423 	time_t max_mtime = 0;
1424 
1425 	TAILQ_INIT(&refs);
1426 	got_ratelimit_init(&rl, 0, 500);
1427 
1428 	*loose_before = 0;
1429 	*loose_after = 0;
1430 	*pack_before = 0;
1431 	*pack_after = 0;
1432 	*ncommits = 0;
1433 	*nloose = 0;
1434 	*npacked = 0;
1435 
1436 	err = repo_cleanup_lock(repo, &lk);
1437 	if (err)
1438 		return err;
1439 
1440 	traversed_ids = got_object_idset_alloc();
1441 	if (traversed_ids == NULL) {
1442 		err = got_error_from_errno("got_object_idset_alloc");
1443 		goto done;
1444 	}
1445 
1446 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1447 	if (err)
1448 		goto done;
1449 	if (!ignore_mtime) {
1450 		TAILQ_FOREACH(re, &refs, entry) {
1451 			time_t mtime = got_ref_get_mtime(re->ref);
1452 			if (mtime > max_mtime)
1453 				max_mtime = mtime;
1454 		}
1455 		/*
1456 		 * For safety, keep objects created within 10 minutes
1457 		 * before the youngest reference was created.
1458 		 */
1459 		if (max_mtime >= 600)
1460 			max_mtime -= 600;
1461 	}
1462 
1463 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1464 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1465 	    &refs, repo, cancel_cb, cancel_arg);
1466 	if (err)
1467 		goto done;
1468 
1469 	for (i = 0; i < nreferenced; i++) {
1470 		struct got_object_id *id = referenced_ids[i];
1471 		err = load_commit_or_tag(ncommits, traversed_ids,
1472 		    id, repo, progress_cb, progress_arg, &rl,
1473 		    cancel_cb, cancel_arg);
1474 		if (err)
1475 			goto done;
1476 	}
1477 
1478 	err = repo_purge_unreferenced_loose_objects(repo, traversed_ids,
1479 	    loose_before, loose_after, *ncommits, nloose, npacked, &npurged,
1480 	    dry_run, ignore_mtime, max_mtime, &rl, progress_cb, progress_arg,
1481 	    cancel_cb, cancel_arg);
1482 	if (err)
1483 		goto done;
1484 
1485 	err = repo_purge_redundant_packfiles(repo, traversed_ids,
1486 	    pack_before, pack_after, dry_run, ignore_mtime, max_mtime,
1487 	    *nloose, *ncommits, npurged, &rl, progress_cb, progress_arg,
1488 	    cancel_cb, cancel_arg);
1489 	if (err)
1490 		goto done;
1491 
1492  done:
1493 	if (lk) {
1494 		unlock_err = got_lockfile_unlock(lk, got_repo_get_fd(repo));
1495 		if (err == NULL)
1496 			err = unlock_err;
1497 	}
1498 	if (traversed_ids)
1499 		got_object_idset_free(traversed_ids);
1500 	return err;
1501 }
1502 
1503 const struct got_error *
1504 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1505     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1506     got_cancel_cb cancel_cb, void *cancel_arg)
1507 {
1508 	const struct got_error *err = NULL;
1509 	DIR *packdir = NULL;
1510 	struct dirent *dent;
1511 	char *pack_relpath = NULL;
1512 	int packdir_fd;
1513 	struct stat sb;
1514 
1515 	packdir_fd = openat(got_repo_get_fd(repo),
1516 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1517 	if (packdir_fd == -1) {
1518 		if (errno == ENOENT)
1519 			return NULL;
1520 		return got_error_from_errno_fmt("openat: %s/%s",
1521 		    got_repo_get_path_git_dir(repo),
1522 		    GOT_OBJECTS_PACK_DIR);
1523 	}
1524 
1525 	packdir = fdopendir(packdir_fd);
1526 	if (packdir == NULL) {
1527 		err = got_error_from_errno("fdopendir");
1528 		close(packdir_fd);
1529 		goto done;
1530 	}
1531 
1532 	while ((dent = readdir(packdir)) != NULL) {
1533 		if (cancel_cb) {
1534 			err = cancel_cb(cancel_arg);
1535 			if (err)
1536 				goto done;
1537 		}
1538 
1539 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1540 			continue;
1541 
1542 		err = got_packidx_get_packfile_path(&pack_relpath,
1543 		    dent->d_name);
1544 		if (err)
1545 			goto done;
1546 
1547 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1548 			free(pack_relpath);
1549 			pack_relpath = NULL;
1550 			continue;
1551 		}
1552 		if (errno != ENOENT) {
1553 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1554 			    got_repo_get_path_git_dir(repo),
1555 			    GOT_OBJECTS_PACK_DIR,
1556 			    pack_relpath);
1557 			goto done;
1558 		}
1559 
1560 		if (!dry_run) {
1561 			if (unlinkat(packdir_fd, dent->d_name, 0) == -1) {
1562 				err = got_error_from_errno("unlinkat");
1563 				goto done;
1564 			}
1565 		}
1566 		if (progress_cb) {
1567 			char *path;
1568 			if (asprintf(&path, "%s/%s/%s",
1569 			    got_repo_get_path_git_dir(repo),
1570 			    GOT_OBJECTS_PACK_DIR,
1571 			    dent->d_name) == -1) {
1572 				err = got_error_from_errno("asprintf");
1573 				goto done;
1574 			}
1575 			err = progress_cb(progress_arg, path);
1576 			free(path);
1577 			if (err)
1578 				goto done;
1579 		}
1580 		free(pack_relpath);
1581 		pack_relpath = NULL;
1582 	}
1583 done:
1584 	if (packdir && closedir(packdir) != 0 && err == NULL)
1585 		err = got_error_from_errno("closedir");
1586 	free(pack_relpath);
1587 	return err;
1588 }