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