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 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 const struct got_error *
642 cleanup_unlock(struct got_repository *repo, struct got_lockfile *lk)
643 {
644 	if (lk == NULL)
645 		return NULL;
646 
647 	return got_lockfile_unlock(lk, got_repo_get_fd(repo));
648 }
649 
650 static const struct got_error *
651 report_cleanup_progress(got_cleanup_progress_cb progress_cb,
652     void *progress_arg, struct got_ratelimit *rl,
653     int nloose, int ncommits, int npurged, int nredundant)
654 {
655 	const struct got_error *err;
656 	int elapsed;
657 
658 	if (progress_cb == NULL)
659 		return NULL;
660 
661 	err = got_ratelimit_check(&elapsed, rl);
662 	if (err || !elapsed)
663 		return err;
664 
665 	return progress_cb(progress_arg, nloose, ncommits, npurged, nredundant);
666 }
667 
668 static const struct got_error *
669 get_loose_object_ids(struct got_object_idset **loose_ids, off_t *ondisk_size,
670     got_cleanup_progress_cb progress_cb, void *progress_arg,
671     struct got_ratelimit *rl, struct got_repository *repo)
672 {
673 	const struct got_error *err = NULL;
674 	char *path_objects = NULL, *path = NULL;
675 	DIR *dir = NULL;
676 	struct got_object *obj = NULL;
677 	struct got_object_id id;
678 	int i, fd = -1;
679 	struct stat sb;
680 
681 	*ondisk_size = 0;
682 	*loose_ids = got_object_idset_alloc();
683 	if (*loose_ids == NULL)
684 		return got_error_from_errno("got_object_idset_alloc");
685 
686 	path_objects = got_repo_get_path_objects(repo);
687 	if (path_objects == NULL) {
688 		err = got_error_from_errno("got_repo_get_path_objects");
689 		goto done;
690 	}
691 
692 	for (i = 0; i <= 0xff; i++) {
693 		struct dirent *dent;
694 
695 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
696 			err = got_error_from_errno("asprintf");
697 			break;
698 		}
699 
700 		dir = opendir(path);
701 		if (dir == NULL) {
702 			if (errno == ENOENT) {
703 				err = NULL;
704 				continue;
705 			}
706 			err = got_error_from_errno2("opendir", path);
707 			break;
708 		}
709 
710 		while ((dent = readdir(dir)) != NULL) {
711 			char *id_str;
712 
713 			if (strcmp(dent->d_name, ".") == 0 ||
714 			    strcmp(dent->d_name, "..") == 0)
715 				continue;
716 
717 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
718 				err = got_error_from_errno("asprintf");
719 				goto done;
720 			}
721 
722 			if (!got_parse_object_id(&id, id_str, repo->algo)) {
723 				free(id_str);
724 				continue;
725 			}
726 			free(id_str);
727 
728 			err = got_object_open_loose_fd(&fd, &id, repo);
729 			if (err)
730 				goto done;
731 			if (fstat(fd, &sb) == -1) {
732 				err = got_error_from_errno("fstat");
733 				goto done;
734 			}
735 			err = got_object_read_header_privsep(&obj, &id, repo,
736 			    fd);
737 			if (err)
738 				goto done;
739 			fd = -1; /* already closed */
740 
741 			switch (obj->type) {
742 			case GOT_OBJ_TYPE_COMMIT:
743 			case GOT_OBJ_TYPE_TREE:
744 			case GOT_OBJ_TYPE_BLOB:
745 			case GOT_OBJ_TYPE_TAG:
746 				break;
747 			default:
748 				err = got_error_fmt(GOT_ERR_OBJ_TYPE,
749 				    "%d", obj->type);
750 				goto done;
751 			}
752 			got_object_close(obj);
753 			obj = NULL;
754 			(*ondisk_size) += sb.st_size;
755 			err = got_object_idset_add(*loose_ids, &id, NULL);
756 			if (err)
757 				goto done;
758 			err = report_cleanup_progress(progress_cb,
759 			    progress_arg, rl,
760 			    got_object_idset_num_elements(*loose_ids),
761 			    -1, -1, -1);
762 			if (err)
763 				goto done;
764 		}
765 
766 		if (closedir(dir) != 0) {
767 			err = got_error_from_errno("closedir");
768 			goto done;
769 		}
770 		dir = NULL;
771 
772 		free(path);
773 		path = NULL;
774 	}
775 done:
776 	if (dir && closedir(dir) != 0 && err == NULL)
777 		err = got_error_from_errno("closedir");
778 	if (fd != -1 && close(fd) == -1 && err == NULL)
779 		err = got_error_from_errno("close");
780 	if (err) {
781 		got_object_idset_free(*loose_ids);
782 		*loose_ids = NULL;
783 	}
784 	if (obj)
785 		got_object_close(obj);
786 	free(path_objects);
787 	free(path);
788 	return err;
789 }
790 
791 static const struct got_error *
792 preserve_loose_object(struct got_object_idset *loose_ids,
793     struct got_object_id *id, struct got_repository *repo, int *npacked)
794 {
795 	const struct got_error *err = NULL;
796 	struct got_object *obj;
797 
798 	if (!got_object_idset_contains(loose_ids, id))
799 		return NULL;
800 
801 	/*
802 	 * Try to open this object from a pack file. This ensures that
803 	 * we do in fact have a valid packed copy of the object. Otherwise
804 	 * we should not delete the loose representation of this object.
805 	 */
806 	err = got_object_open_packed(&obj, id, repo);
807 	if (err == NULL) {
808 		got_object_close(obj);
809 		/*
810 		 * The object is referenced and packed.
811 		 * We can purge the redundantly stored loose object.
812 		 */
813 		(*npacked)++;
814 		return NULL;
815 	} else if (err->code != GOT_ERR_NO_OBJ)
816 		return err;
817 
818 	/*
819 	 * This object is referenced and not packed.
820 	 * Remove it from our purge set.
821 	 */
822 	return got_object_idset_remove(NULL, loose_ids, id);
823 }
824 
825 static const struct got_error *
826 load_tree_entries(struct got_object_id_queue *ids,
827     struct got_object_idset *loose_ids,
828     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
829     const char *dpath, struct got_repository *repo, int *npacked,
830     got_cancel_cb cancel_cb, void *cancel_arg)
831 {
832 	const struct got_error *err;
833 	struct got_tree_object *tree;
834 	char *p = NULL;
835 	int i;
836 
837 	err = got_object_open_as_tree(&tree, repo, tree_id);
838 	if (err)
839 		return err;
840 
841 	for (i = 0; i < got_object_tree_get_nentries(tree); i++) {
842 		struct got_tree_entry *e = got_object_tree_get_entry(tree, i);
843 		struct got_object_id *id = got_tree_entry_get_id(e);
844 		mode_t mode = got_tree_entry_get_mode(e);
845 
846 		if (cancel_cb) {
847 			err = (*cancel_cb)(cancel_arg);
848 			if (err)
849 				break;
850 		}
851 
852 		if (got_object_tree_entry_is_symlink(e) ||
853 		    got_object_tree_entry_is_submodule(e) ||
854 		    got_object_idset_contains(traversed_ids, id))
855 			continue;
856 
857 		if (asprintf(&p, "%s%s%s", dpath, dpath[0] != '\0' ? "/" : "",
858 		    got_tree_entry_get_name(e)) == -1) {
859 			err = got_error_from_errno("asprintf");
860 			break;
861 		}
862 
863 		if (S_ISDIR(mode)) {
864 			struct got_object_qid *qid;
865 			err = got_object_qid_alloc(&qid, id);
866 			if (err)
867 				break;
868 			STAILQ_INSERT_TAIL(ids, qid, entry);
869 		} else if (S_ISREG(mode)) {
870 			/* This blob is referenced. */
871 			err = preserve_loose_object(loose_ids, id, repo,
872 			    npacked);
873 			if (err)
874 				break;
875 			err = got_object_idset_add(traversed_ids, id, NULL);
876 			if (err)
877 				break;
878 		}
879 		free(p);
880 		p = NULL;
881 	}
882 
883 	got_object_tree_close(tree);
884 	free(p);
885 	return err;
886 }
887 
888 static const struct got_error *
889 load_tree(struct got_object_idset *loose_ids,
890     struct got_object_idset *traversed_ids, struct got_object_id *tree_id,
891     const char *dpath, struct got_repository *repo, int *npacked,
892     got_cancel_cb cancel_cb, void *cancel_arg)
893 {
894 	const struct got_error *err = NULL;
895 	struct got_object_id_queue tree_ids;
896 	struct got_object_qid *qid;
897 
898 	err = got_object_qid_alloc(&qid, tree_id);
899 	if (err)
900 		return err;
901 
902 	STAILQ_INIT(&tree_ids);
903 	STAILQ_INSERT_TAIL(&tree_ids, qid, entry);
904 
905 	while (!STAILQ_EMPTY(&tree_ids)) {
906 		if (cancel_cb) {
907 			err = (*cancel_cb)(cancel_arg);
908 			if (err)
909 				break;
910 		}
911 
912 		qid = STAILQ_FIRST(&tree_ids);
913 		STAILQ_REMOVE_HEAD(&tree_ids, entry);
914 
915 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
916 			got_object_qid_free(qid);
917 			continue;
918 		}
919 
920 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
921 		if (err) {
922 			got_object_qid_free(qid);
923 			break;
924 		}
925 
926 		/* This tree is referenced. */
927 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
928 		if (err)
929 			break;
930 
931 		err = load_tree_entries(&tree_ids, loose_ids, traversed_ids,
932 		    &qid->id, dpath, repo, npacked, cancel_cb, cancel_arg);
933 		got_object_qid_free(qid);
934 		if (err)
935 			break;
936 	}
937 
938 	got_object_id_queue_free(&tree_ids);
939 	return err;
940 }
941 
942 static const struct got_error *
943 load_commit_or_tag(struct got_object_idset *loose_ids, int *ncommits,
944     int *npacked, struct got_object_idset *traversed_ids,
945     struct got_object_id *id, struct got_repository *repo,
946     got_cleanup_progress_cb progress_cb, void *progress_arg,
947     struct got_ratelimit *rl, int nloose,
948     got_cancel_cb cancel_cb, void *cancel_arg)
949 {
950 	const struct got_error *err;
951 	struct got_commit_object *commit = NULL;
952 	struct got_tag_object *tag = NULL;
953 	struct got_object_id *tree_id = NULL;
954 	struct got_object_id_queue ids;
955 	struct got_object_qid *qid;
956 	int obj_type;
957 
958 	err = got_object_qid_alloc(&qid, id);
959 	if (err)
960 		return err;
961 
962 	STAILQ_INIT(&ids);
963 	STAILQ_INSERT_TAIL(&ids, qid, entry);
964 
965 	while (!STAILQ_EMPTY(&ids)) {
966 		if (cancel_cb) {
967 			err = (*cancel_cb)(cancel_arg);
968 			if (err)
969 				break;
970 		}
971 
972 		qid = STAILQ_FIRST(&ids);
973 		STAILQ_REMOVE_HEAD(&ids, entry);
974 
975 		if (got_object_idset_contains(traversed_ids, &qid->id)) {
976 			got_object_qid_free(qid);
977 			qid = NULL;
978 			continue;
979 		}
980 
981 		err = got_object_idset_add(traversed_ids, &qid->id, NULL);
982 		if (err)
983 			break;
984 
985 		/* This commit or tag is referenced. */
986 		err = preserve_loose_object(loose_ids, &qid->id, repo, npacked);
987 		if (err)
988 			break;
989 
990 		err = got_object_get_type(&obj_type, repo, &qid->id);
991 		if (err)
992 			break;
993 		switch (obj_type) {
994 		case GOT_OBJ_TYPE_COMMIT:
995 			err = got_object_open_as_commit(&commit, repo,
996 			    &qid->id);
997 			if (err)
998 				goto done;
999 			break;
1000 		case GOT_OBJ_TYPE_TAG:
1001 			err = got_object_open_as_tag(&tag, repo, &qid->id);
1002 			if (err)
1003 				goto done;
1004 			break;
1005 		default:
1006 			/* should not happen */
1007 			err = got_error(GOT_ERR_OBJ_TYPE);
1008 			goto done;
1009 		}
1010 
1011 		/* Find a tree object to scan. */
1012 		if (commit) {
1013 			tree_id = got_object_commit_get_tree_id(commit);
1014 		} else if (tag) {
1015 			obj_type = got_object_tag_get_object_type(tag);
1016 			switch (obj_type) {
1017 			case GOT_OBJ_TYPE_COMMIT:
1018 				err = got_object_open_as_commit(&commit, repo,
1019 				    got_object_tag_get_object_id(tag));
1020 				if (err)
1021 					goto done;
1022 				tree_id = got_object_commit_get_tree_id(commit);
1023 				break;
1024 			case GOT_OBJ_TYPE_TREE:
1025 				tree_id = got_object_tag_get_object_id(tag);
1026 				break;
1027 			default:
1028 				/*
1029 				 * Tag points at something other than a
1030 				 * commit or tree. Leave this weird tag object
1031 				 * and the object it points to on disk.
1032 				 */
1033 				err = got_object_idset_remove(NULL, loose_ids,
1034 				    &qid->id);
1035 				if (err && err->code != GOT_ERR_NO_OBJ)
1036 					goto done;
1037 				err = got_object_idset_remove(NULL, loose_ids,
1038 				    got_object_tag_get_object_id(tag));
1039 				if (err && err->code != GOT_ERR_NO_OBJ)
1040 					goto done;
1041 				err = NULL;
1042 				break;
1043 			}
1044 		}
1045 
1046 		if (tree_id) {
1047 			err = load_tree(loose_ids, traversed_ids, tree_id, "",
1048 			    repo, npacked, cancel_cb, cancel_arg);
1049 			if (err)
1050 				break;
1051 		}
1052 
1053 		if (commit || tag)
1054 			(*ncommits)++; /* scanned tags are counted as commits */
1055 
1056 		err = report_cleanup_progress(progress_cb, progress_arg, rl,
1057 		    nloose, *ncommits, -1, -1);
1058 		if (err)
1059 			break;
1060 
1061 		if (commit) {
1062 			/* Find parent commits to scan. */
1063 			const struct got_object_id_queue *parent_ids;
1064 			parent_ids = got_object_commit_get_parent_ids(commit);
1065 			err = got_object_id_queue_copy(parent_ids, &ids);
1066 			if (err)
1067 				break;
1068 			got_object_commit_close(commit);
1069 			commit = NULL;
1070 		}
1071 		if (tag) {
1072 			got_object_tag_close(tag);
1073 			tag = NULL;
1074 		}
1075 		got_object_qid_free(qid);
1076 		qid = NULL;
1077 	}
1078 done:
1079 	if (qid)
1080 		got_object_qid_free(qid);
1081 	if (commit)
1082 		got_object_commit_close(commit);
1083 	if (tag)
1084 		got_object_tag_close(tag);
1085 	got_object_id_queue_free(&ids);
1086 	return err;
1087 }
1088 
1089 struct purge_loose_object_arg {
1090 	struct got_repository *repo;
1091 	got_cleanup_progress_cb progress_cb;
1092 	void *progress_arg;
1093 	struct got_ratelimit *rl;
1094 	int nloose;
1095 	int ncommits;
1096 	int npurged;
1097 	off_t size_purged;
1098 	int dry_run;
1099 	time_t max_mtime;
1100 	int ignore_mtime;
1101 };
1102 
1103 static const struct got_error *
1104 purge_loose_object(struct got_object_id *id, void *data, void *arg)
1105 {
1106 	struct purge_loose_object_arg *a = arg;
1107 	const struct got_error *err, *unlock_err = NULL;
1108 	char *path = NULL;
1109 	int fd = -1;
1110 	struct stat sb;
1111 	struct got_lockfile *lf = NULL;
1112 
1113 	err = got_object_get_path(&path, id, a->repo);
1114 	if (err)
1115 		return err;
1116 
1117 	err = got_object_open_loose_fd(&fd, id, a->repo);
1118 	if (err)
1119 		goto done;
1120 
1121 	if (fstat(fd, &sb) == -1) {
1122 		err = got_error_from_errno("fstat");
1123 		goto done;
1124 	}
1125 
1126 	/*
1127 	 * Do not delete objects which are younger than our maximum
1128 	 * modification time threshold. This prevents a race where
1129 	 * new objects which are being added to the repository
1130 	 * concurrently would be deleted.
1131 	 */
1132 	if (a->ignore_mtime || sb.st_mtime <= a->max_mtime) {
1133 		if (!a->dry_run) {
1134 			err = got_lockfile_lock(&lf, path, -1);
1135 			if (err)
1136 				goto done;
1137 			if (unlink(path) == -1) {
1138 				err = got_error_from_errno2("unlink", path);
1139 				goto done;
1140 			}
1141 		}
1142 
1143 		a->npurged++;
1144 		a->size_purged += sb.st_size;
1145 		err = report_cleanup_progress(a->progress_cb, a->progress_arg,
1146 		    a->rl, a->nloose, a->ncommits, a->npurged, -1);
1147 		if (err)
1148 			goto done;
1149 	}
1150 done:
1151 	if (fd != -1 && close(fd) == -1 && err == NULL)
1152 		err = got_error_from_errno("close");
1153 	free(path);
1154 	if (lf)
1155 		unlock_err = got_lockfile_unlock(lf, -1);
1156 	return err ? err : unlock_err;
1157 }
1158 
1159 const struct got_error *
1160 got_repo_cleanup(struct got_repository *repo,
1161     off_t *loose_before, off_t *loose_after,
1162     off_t *pack_before, off_t *pack_after,
1163     int *ncommits, int *nloose,
1164     int *npacked, int dry_run, int ignore_mtime,
1165     got_cleanup_progress_cb progress_cb, void *progress_arg,
1166     got_cancel_cb cancel_cb, void *cancel_arg)
1167 {
1168 	const struct got_error *err;
1169 	struct got_object_idset *loose_ids;
1170 	struct got_object_idset *traversed_ids;
1171 	struct got_object_id **referenced_ids;
1172 	int i, nreferenced;
1173 	struct got_reflist_head refs;
1174 	struct got_reflist_entry *re;
1175 	struct purge_loose_object_arg arg;
1176 	time_t max_mtime = 0;
1177 	struct got_ratelimit rl;
1178 
1179 	TAILQ_INIT(&refs);
1180 	got_ratelimit_init(&rl, 0, 500);
1181 
1182 	*loose_before = 0;
1183 	*loose_after = 0;
1184 	*npacked = 0;
1185 
1186 	err = get_loose_object_ids(&loose_ids, loose_before,
1187 	    progress_cb, progress_arg, &rl, repo);
1188 	if (err)
1189 		return err;
1190 	*nloose = got_object_idset_num_elements(loose_ids);
1191 	if (*nloose == 0) {
1192 		got_object_idset_free(loose_ids);
1193 		if (progress_cb) {
1194 			err = progress_cb(progress_arg, 0, 0, 0, -1);
1195 			if (err)
1196 				return err;
1197 		}
1198 		return NULL;
1199 	}
1200 
1201 	traversed_ids = got_object_idset_alloc();
1202 	if (traversed_ids == NULL) {
1203 		err = got_error_from_errno("got_object_idset_alloc");
1204 		goto done;
1205 	}
1206 
1207 	err = got_ref_list(&refs, repo, "", got_ref_cmp_by_name, NULL);
1208 	if (err)
1209 		goto done;
1210 	if (!ignore_mtime) {
1211 		TAILQ_FOREACH(re, &refs, entry) {
1212 			time_t mtime = got_ref_get_mtime(re->ref);
1213 			if (mtime > max_mtime)
1214 				max_mtime = mtime;
1215 		}
1216 		/*
1217 		 * For safety, keep objects created within 10 minutes
1218 		 * before the youngest reference was created.
1219 		 */
1220 		if (max_mtime >= 600)
1221 			max_mtime -= 600;
1222 	}
1223 
1224 	err = get_reflist_object_ids(&referenced_ids, &nreferenced,
1225 	    (1 << GOT_OBJ_TYPE_COMMIT) | (1 << GOT_OBJ_TYPE_TAG),
1226 	    &refs, repo, cancel_cb, cancel_arg);
1227 	if (err)
1228 		goto done;
1229 
1230 	for (i = 0; i < nreferenced; i++) {
1231 		struct got_object_id *id = referenced_ids[i];
1232 		err = load_commit_or_tag(loose_ids, ncommits, npacked,
1233 		    traversed_ids, id, repo, progress_cb, progress_arg, &rl,
1234 		    *nloose, cancel_cb, cancel_arg);
1235 		if (err)
1236 			goto done;
1237 	}
1238 
1239 	/* Any remaining loose objects are unreferenced and can be purged. */
1240 	arg.repo = repo;
1241 	arg.progress_arg = progress_arg;
1242 	arg.progress_cb = progress_cb;
1243 	arg.rl = &rl;
1244 	arg.nloose = *nloose;
1245 	arg.npurged = 0;
1246 	arg.size_purged = 0;
1247 	arg.ncommits = *ncommits;
1248 	arg.dry_run = dry_run;
1249 	arg.max_mtime = max_mtime;
1250 	arg.ignore_mtime = ignore_mtime;
1251 	err = got_object_idset_for_each(loose_ids, purge_loose_object, &arg);
1252 	if (err)
1253 		goto done;
1254 	*loose_after = *loose_before - arg.size_purged;
1255 
1256 	/* Produce a final progress report. */
1257 	if (progress_cb) {
1258 		err = progress_cb(progress_arg, *nloose, *ncommits, arg.npurged,
1259 		    -1);
1260 		if (err)
1261 			goto done;
1262 	}
1263 done:
1264 	got_object_idset_free(loose_ids);
1265 	got_object_idset_free(traversed_ids);
1266 	return err;
1267 }
1268 
1269 static const struct got_error *
1270 purge_redundant_pack(struct got_repository *repo, const char *packidx_path,
1271     int dry_run, int *remove, off_t *size_before, off_t *size_after)
1272 {
1273 	static const char *ext[] = {".idx", ".pack", ".rev", ".bitmap",
1274 	    ".promisor", ".mtimes"};
1275 	struct stat sb;
1276 	char *dot, path[PATH_MAX];
1277 	size_t i;
1278 
1279 	if (strlcpy(path, packidx_path, sizeof(path)) >= sizeof(path))
1280 		return got_error(GOT_ERR_NO_SPACE);
1281 
1282 	/*
1283 	 * For compatibility with Git, if a matching .keep file exist
1284 	 * don't delete the packfile.
1285 	 */
1286 	dot = strrchr(path, '.');
1287 	*dot = '\0';
1288 	if (strlcat(path, ".keep", sizeof(path)) >= sizeof(path))
1289 		return got_error(GOT_ERR_NO_SPACE);
1290 	if (faccessat(got_repo_get_fd(repo), path, F_OK, 0) == 0)
1291 		*remove = 0;
1292 
1293 	for (i = 0; i < nitems(ext); ++i) {
1294 		*dot = '\0';
1295 
1296 		if (strlcat(path, ext[i], sizeof(path)) >=
1297 		    sizeof(path))
1298 			return got_error(GOT_ERR_NO_SPACE);
1299 
1300 		if (fstatat(got_repo_get_fd(repo), path, &sb, 0) ==
1301 		    -1) {
1302 			if (errno == ENOENT &&
1303 			    strcmp(ext[i], ".pack") != 0 &&
1304 			    strcmp(ext[i], ".idx") != 0)
1305 				continue;
1306 			return got_error_from_errno2("fstatat", path);
1307 		}
1308 
1309 		*size_before += sb.st_size;
1310 		if (!*remove) {
1311 			*size_after += sb.st_size;
1312 			continue;
1313 		}
1314 
1315 		if (dry_run)
1316 			continue;
1317 
1318 		if (unlinkat(got_repo_get_fd(repo), path, 0) == -1) {
1319 			if (errno == ENOENT)
1320 				continue;
1321 			return got_error_from_errno2("unlinkat",
1322 			    path);
1323 		}
1324 	}
1325 
1326 	return NULL;
1327 }
1328 
1329 static const struct got_error *
1330 pack_is_redundant(int *redundant, struct got_repository *repo,
1331     const char *packidx_path, struct got_object_idset *idset)
1332 {
1333 	const struct got_error *err;
1334 	struct got_packidx *packidx;
1335 	struct got_packidx_object_id *pid;
1336 	struct got_object_id id;
1337 	size_t i, nobjects;
1338 
1339 	*redundant = 1;
1340 
1341 	err = got_repo_get_packidx(&packidx, packidx_path, repo);
1342 	if (err)
1343 		return err;
1344 
1345 	nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1346 	for (i = 0; i < nobjects; ++i) {
1347 		pid = &packidx->hdr.sorted_ids[i];
1348 
1349 		memset(&id, 0, sizeof(id));
1350 		memcpy(&id.sha1, pid->sha1, sizeof(id.sha1));
1351 
1352 		if (got_object_idset_contains(idset, &id))
1353 			continue;
1354 
1355 		*redundant = 0;
1356 		err = got_object_idset_add(idset, &id, NULL);
1357 		if (err)
1358 			return err;
1359 	}
1360 
1361 	return NULL;
1362 }
1363 
1364 struct pack_info {
1365 	const char	*path;
1366 	size_t		 nobjects;
1367 };
1368 
1369 static int
1370 pack_info_cmp(const void *a, const void *b)
1371 {
1372 	const struct pack_info	*pa, *pb;
1373 
1374 	pa = a;
1375 	pb = b;
1376 	if (pa->nobjects == pb->nobjects)
1377 		return strcmp(pa->path, pb->path);
1378 	if (pa->nobjects > pb->nobjects)
1379 		return -1;
1380 	return 1;
1381 }
1382 
1383 const struct got_error *
1384 got_repo_purge_redundant_packfiles(struct got_repository *repo,
1385     off_t *size_before, off_t *size_after, int dry_run,
1386     int nloose, int ncommits, int npurged,
1387     got_cleanup_progress_cb progress_cb, void *progress_arg,
1388     got_cancel_cb cancel_cb, void *cancel_arg)
1389 {
1390 	const struct got_error *err;
1391 	struct pack_info *pinfo, *sorted = NULL;
1392 	struct got_packidx *packidx;
1393 	struct got_object_idset *idset = NULL;
1394 	struct got_pathlist_entry *pe;
1395 	size_t i, npacks;
1396 	int remove, redundant_packs = 0;
1397 	struct got_ratelimit rl;
1398 
1399 	got_ratelimit_init(&rl, 0, 500);
1400 
1401 	*size_before = 0;
1402 	*size_after = 0;
1403 
1404 	npacks = 0;
1405 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
1406 		npacks++;
1407 
1408 	if (npacks == 0)
1409 		return NULL;
1410 
1411 	sorted = calloc(npacks, sizeof(*sorted));
1412 	if (sorted == NULL)
1413 		return got_error_from_errno("calloc");
1414 
1415 	i = 0;
1416 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1417 		err = got_repo_get_packidx(&packidx, pe->path, repo);
1418 		if (err)
1419 			goto done;
1420 
1421 		pinfo = &sorted[i++];
1422 		pinfo->path = pe->path;
1423 		pinfo->nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1424 	}
1425 	qsort(sorted, npacks, sizeof(*sorted), pack_info_cmp);
1426 
1427 	idset = got_object_idset_alloc();
1428 	if (idset == NULL) {
1429 		err = got_error_from_errno("got_object_idset_alloc");
1430 		goto done;
1431 	}
1432 
1433 	for (i = 0; i < npacks; ++i) {
1434 		if (cancel_cb) {
1435 			err = (*cancel_cb)(cancel_arg);
1436 			if (err)
1437 				break;
1438 		}
1439 
1440 		err = pack_is_redundant(&remove, repo, sorted[i].path, idset);
1441 		if (err)
1442 			goto done;
1443 		err = purge_redundant_pack(repo, sorted[i].path, dry_run,
1444 		    &remove, size_before, size_after);
1445 		if (err)
1446 			goto done;
1447 		if (!remove)
1448 			continue;
1449 		err = report_cleanup_progress(progress_cb, progress_arg,
1450 		    &rl, nloose, ncommits, npurged, ++redundant_packs);
1451 		if (err)
1452 			goto done;
1453 	}
1454 
1455 	/* Produce a final progress report. */
1456 	if (progress_cb) {
1457 		err = progress_cb(progress_arg, nloose, ncommits, npurged,
1458 		    redundant_packs);
1459 		if (err)
1460 			goto done;
1461 	}
1462  done:
1463 	free(sorted);
1464 	if (idset)
1465 		got_object_idset_free(idset);
1466 	return err;
1467 }
1468 
1469 const struct got_error *
1470 got_repo_remove_lonely_packidx(struct got_repository *repo, int dry_run,
1471     got_lonely_packidx_progress_cb progress_cb, void *progress_arg,
1472     got_cancel_cb cancel_cb, void *cancel_arg)
1473 {
1474 	const struct got_error *err = NULL;
1475 	DIR *packdir = NULL;
1476 	struct dirent *dent;
1477 	char *pack_relpath = NULL;
1478 	int packdir_fd;
1479 	struct stat sb;
1480 
1481 	packdir_fd = openat(got_repo_get_fd(repo),
1482 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1483 	if (packdir_fd == -1) {
1484 		if (errno == ENOENT)
1485 			return NULL;
1486 		return got_error_from_errno_fmt("openat: %s/%s",
1487 		    got_repo_get_path_git_dir(repo),
1488 		    GOT_OBJECTS_PACK_DIR);
1489 	}
1490 
1491 	packdir = fdopendir(packdir_fd);
1492 	if (packdir == NULL) {
1493 		err = got_error_from_errno("fdopendir");
1494 		goto done;
1495 	}
1496 
1497 	while ((dent = readdir(packdir)) != NULL) {
1498 		if (cancel_cb) {
1499 			err = cancel_cb(cancel_arg);
1500 			if (err)
1501 				goto done;
1502 		}
1503 
1504 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1505 			continue;
1506 
1507 		err = got_packidx_get_packfile_path(&pack_relpath,
1508 		    dent->d_name);
1509 		if (err)
1510 			goto done;
1511 
1512 		if (fstatat(packdir_fd, pack_relpath, &sb, 0) != -1) {
1513 			free(pack_relpath);
1514 			pack_relpath = NULL;
1515 			continue;
1516 		}
1517 		if (errno != ENOENT) {
1518 			err = got_error_from_errno_fmt("fstatat: %s/%s/%s",
1519 			    got_repo_get_path_git_dir(repo),
1520 			    GOT_OBJECTS_PACK_DIR,
1521 			    pack_relpath);
1522 			goto done;
1523 		}
1524 
1525 		if (!dry_run) {
1526 			if (unlinkat(packdir_fd, dent->d_name, 0) == -1) {
1527 				err = got_error_from_errno("unlinkat");
1528 				goto done;
1529 			}
1530 		}
1531 		if (progress_cb) {
1532 			char *path;
1533 			if (asprintf(&path, "%s/%s/%s",
1534 			    got_repo_get_path_git_dir(repo),
1535 			    GOT_OBJECTS_PACK_DIR,
1536 			    dent->d_name) == -1) {
1537 				err = got_error_from_errno("asprintf");
1538 				goto done;
1539 			}
1540 			err = progress_cb(progress_arg, path);
1541 			free(path);
1542 			if (err)
1543 				goto done;
1544 		}
1545 		free(pack_relpath);
1546 		pack_relpath = NULL;
1547 	}
1548 done:
1549 	if (packdir && closedir(packdir) != 0 && err == NULL)
1550 		err = got_error_from_errno("closedir");
1551 	free(pack_relpath);
1552 	return err;
1553 }