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