1 /*
2  * Copyright (c) 2018, 2019, 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/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 #include <sys/resource.h>
24 
25 #include <ctype.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <string.h>
33 #include <time.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 
40 #include "bloom.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_sha1.h"
58 #include "got_lib_object_cache.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_gotconfig.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
67 
68 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
69     got_packidx_bloom_filter_cmp);
70 
71 const char *
72 got_repo_get_path(struct got_repository *repo)
73 {
74 	return repo->path;
75 }
76 
77 const char *
78 got_repo_get_path_git_dir(struct got_repository *repo)
79 {
80 	return repo->path_git_dir;
81 }
82 
83 int
84 got_repo_get_fd(struct got_repository *repo)
85 {
86 	return repo->gitdir_fd;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_name(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_name;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_email(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_email;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_name;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_email;
111 }
112 
113 const char *
114 got_repo_get_gitconfig_owner(struct got_repository *repo)
115 {
116 	return repo->gitconfig_owner;
117 }
118 
119 void
120 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
121     struct got_repository *repo)
122 {
123 	*extensions = repo->extensions;
124 	*nextensions = repo->nextensions;
125 }
126 
127 int
128 got_repo_is_bare(struct got_repository *repo)
129 {
130 	return (strcmp(repo->path, repo->path_git_dir) == 0);
131 }
132 
133 static char *
134 get_path_git_child(struct got_repository *repo, const char *basename)
135 {
136 	char *path_child;
137 
138 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
139 	    basename) == -1)
140 		return NULL;
141 
142 	return path_child;
143 }
144 
145 char *
146 got_repo_get_path_objects(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
149 }
150 
151 char *
152 got_repo_get_path_objects_pack(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
155 }
156 
157 char *
158 got_repo_get_path_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_REFS_DIR);
161 }
162 
163 char *
164 got_repo_get_path_packed_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
167 }
168 
169 static char *
170 get_path_head(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_HEAD_FILE);
173 }
174 
175 char *
176 got_repo_get_path_gitconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GITCONFIG);
179 }
180 
181 char *
182 got_repo_get_path_gotconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
185 }
186 
187 const struct got_gotconfig *
188 got_repo_get_gotconfig(struct got_repository *repo)
189 {
190 	return repo->gotconfig;
191 }
192 
193 void
194 got_repo_get_gitconfig_remotes(int *nremotes,
195     const struct got_remote_repo **remotes, struct got_repository *repo)
196 {
197 	*nremotes = repo->ngitconfig_remotes;
198 	*remotes = repo->gitconfig_remotes;
199 }
200 
201 static int
202 is_git_repo(struct got_repository *repo)
203 {
204 	const char *path_git = got_repo_get_path_git_dir(repo);
205 	char *path_objects = got_repo_get_path_objects(repo);
206 	char *path_refs = got_repo_get_path_refs(repo);
207 	char *path_head = get_path_head(repo);
208 	int ret = 0;
209 	struct stat sb;
210 	struct got_reference *head_ref;
211 
212 	if (lstat(path_git, &sb) == -1)
213 		goto done;
214 	if (!S_ISDIR(sb.st_mode))
215 		goto done;
216 
217 	if (lstat(path_objects, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_refs, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_head, &sb) == -1)
228 		goto done;
229 	if (!S_ISREG(sb.st_mode))
230 		goto done;
231 
232 	/* Check if the HEAD reference can be opened. */
233 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
234 		goto done;
235 	got_ref_close(head_ref);
236 
237 	ret = 1;
238 done:
239 	free(path_objects);
240 	free(path_refs);
241 	free(path_head);
242 	return ret;
243 
244 }
245 
246 const struct got_error *
247 got_repo_pack_fds_open(int **pack_fds)
248 {
249 	const struct got_error *err = NULL;
250 	int i;
251 
252 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
253 	if (*pack_fds == NULL)
254 		return got_error_from_errno("calloc");
255 
256 	/*
257 	 * got_repo_pack_fds_close will try to close all of the
258 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
259 	 * a value from got_opentempfd(), which would result in a close(0) if
260 	 * we do not initialize to -1 here.
261 	 */
262 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
263 		(*pack_fds)[i] = -1;
264 
265 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
266 		(*pack_fds)[i] = got_opentempfd();
267 		if ((*pack_fds)[i] == -1) {
268 			err = got_error_from_errno("got_opentempfd");
269 			got_repo_pack_fds_close(*pack_fds);
270 			*pack_fds = NULL;
271 			return err;
272 		}
273 	}
274 
275 	return NULL;
276 }
277 
278 const struct got_error *
279 got_repo_pack_fds_close(int *pack_fds)
280 {
281 	const struct got_error *err = NULL;
282 	int i;
283 
284 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
285 		if (pack_fds[i] == -1)
286 			continue;
287 		if (close(pack_fds[i]) == -1) {
288 			err = got_error_from_errno("close");
289 			break;
290 		}
291 	}
292 	free(pack_fds);
293 	return err;
294 }
295 
296 const struct got_error *
297 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
298     struct got_object *obj)
299 {
300 #ifndef GOT_NO_OBJ_CACHE
301 	const struct got_error *err = NULL;
302 	err = got_object_cache_add(&repo->objcache, id, obj);
303 	if (err) {
304 		if (err->code == GOT_ERR_OBJ_EXISTS ||
305 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
306 			err = NULL;
307 		return err;
308 	}
309 	obj->refcnt++;
310 #endif
311 	return NULL;
312 }
313 
314 struct got_object *
315 got_repo_get_cached_object(struct got_repository *repo,
316     struct got_object_id *id)
317 {
318 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
319 }
320 
321 const struct got_error *
322 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
323     struct got_tree_object *tree)
324 {
325 #ifndef GOT_NO_OBJ_CACHE
326 	const struct got_error *err = NULL;
327 	err = got_object_cache_add(&repo->treecache, id, tree);
328 	if (err) {
329 		if (err->code == GOT_ERR_OBJ_EXISTS ||
330 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
331 			err = NULL;
332 		return err;
333 	}
334 	tree->refcnt++;
335 #endif
336 	return NULL;
337 }
338 
339 struct got_tree_object *
340 got_repo_get_cached_tree(struct got_repository *repo,
341     struct got_object_id *id)
342 {
343 	return (struct got_tree_object *)got_object_cache_get(
344 	    &repo->treecache, id);
345 }
346 
347 const struct got_error *
348 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
349     struct got_commit_object *commit)
350 {
351 #ifndef GOT_NO_OBJ_CACHE
352 	const struct got_error *err = NULL;
353 	err = got_object_cache_add(&repo->commitcache, id, commit);
354 	if (err) {
355 		if (err->code == GOT_ERR_OBJ_EXISTS ||
356 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
357 			err = NULL;
358 		return err;
359 	}
360 	commit->refcnt++;
361 #endif
362 	return NULL;
363 }
364 
365 struct got_commit_object *
366 got_repo_get_cached_commit(struct got_repository *repo,
367     struct got_object_id *id)
368 {
369 	return (struct got_commit_object *)got_object_cache_get(
370 	    &repo->commitcache, id);
371 }
372 
373 const struct got_error *
374 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
375     struct got_tag_object *tag)
376 {
377 #ifndef GOT_NO_OBJ_CACHE
378 	const struct got_error *err = NULL;
379 	err = got_object_cache_add(&repo->tagcache, id, tag);
380 	if (err) {
381 		if (err->code == GOT_ERR_OBJ_EXISTS ||
382 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
383 			err = NULL;
384 		return err;
385 	}
386 	tag->refcnt++;
387 #endif
388 	return NULL;
389 }
390 
391 struct got_tag_object *
392 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
393 {
394 	return (struct got_tag_object *)got_object_cache_get(
395 	    &repo->tagcache, id);
396 }
397 
398 const struct got_error *
399 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
400     struct got_raw_object *raw)
401 {
402 #ifndef GOT_NO_OBJ_CACHE
403 	const struct got_error *err = NULL;
404 	err = got_object_cache_add(&repo->rawcache, id, raw);
405 	if (err) {
406 		if (err->code == GOT_ERR_OBJ_EXISTS ||
407 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
408 			err = NULL;
409 		return err;
410 	}
411 	raw->refcnt++;
412 #endif
413 	return NULL;
414 }
415 
416 
417 struct got_raw_object *
418 got_repo_get_cached_raw_object(struct got_repository *repo,
419     struct got_object_id *id)
420 {
421 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
422 }
423 
424 
425 static const struct got_error *
426 open_repo(struct got_repository *repo, const char *path)
427 {
428 	const struct got_error *err = NULL;
429 
430 	repo->gitdir_fd = -1;
431 
432 	/* bare git repository? */
433 	repo->path_git_dir = strdup(path);
434 	if (repo->path_git_dir == NULL)
435 		return got_error_from_errno("strdup");
436 	if (is_git_repo(repo)) {
437 		repo->path = strdup(repo->path_git_dir);
438 		if (repo->path == NULL) {
439 			err = got_error_from_errno("strdup");
440 			goto done;
441 		}
442 		repo->gitdir_fd = open(repo->path_git_dir,
443 		    O_DIRECTORY | O_CLOEXEC);
444 		if (repo->gitdir_fd == -1) {
445 			err = got_error_from_errno2("open",
446 			    repo->path_git_dir);
447 			goto done;
448 		}
449 		return NULL;
450 	}
451 
452 	/* git repository with working tree? */
453 	free(repo->path_git_dir);
454 	repo->path_git_dir = NULL;
455 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
456 		err = got_error_from_errno("asprintf");
457 		goto done;
458 	}
459 	if (is_git_repo(repo)) {
460 		repo->path = strdup(path);
461 		if (repo->path == NULL) {
462 			err = got_error_from_errno("strdup");
463 			goto done;
464 		}
465 		repo->gitdir_fd = open(repo->path_git_dir,
466 		    O_DIRECTORY | O_CLOEXEC);
467 		if (repo->gitdir_fd == -1) {
468 			err = got_error_from_errno2("open",
469 			    repo->path_git_dir);
470 			goto done;
471 		}
472 		return NULL;
473 	}
474 
475 	err = got_error(GOT_ERR_NOT_GIT_REPO);
476 done:
477 	if (err) {
478 		free(repo->path);
479 		repo->path = NULL;
480 		free(repo->path_git_dir);
481 		repo->path_git_dir = NULL;
482 		if (repo->gitdir_fd != -1)
483 			close(repo->gitdir_fd);
484 		repo->gitdir_fd = -1;
485 
486 	}
487 	return err;
488 }
489 
490 static const struct got_error *
491 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
492 {
493 	const struct got_error *err = NULL;
494 	char *repo_gitconfig_path = NULL;
495 
496 	if (global_gitconfig_path) {
497 		/* Read settings from ~/.gitconfig. */
498 		int dummy_repo_version;
499 		err = got_repo_read_gitconfig(&dummy_repo_version,
500 		    &repo->global_gitconfig_author_name,
501 		    &repo->global_gitconfig_author_email,
502 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
503 		if (err)
504 			return err;
505 	}
506 
507 	/* Read repository's .git/config file. */
508 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
509 	if (repo_gitconfig_path == NULL)
510 		return got_error_from_errno("got_repo_get_path_gitconfig");
511 
512 	err = got_repo_read_gitconfig(
513 	    &repo->gitconfig_repository_format_version,
514 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
515 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
516 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
517 	    repo_gitconfig_path);
518 	if (err)
519 		goto done;
520 
521 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
522 		int i;
523 
524 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
525 			got_repo_free_remote_repo_data(
526 			    &repo->gitconfig_remotes[i]);
527 		}
528 		free(repo->gitconfig_remotes);
529 		repo->gitconfig_remotes = NULL;
530 		repo->ngitconfig_remotes = 0;
531 
532 		free(repo->gitconfig_author_name);
533 		repo->gitconfig_author_name = NULL;
534 		free(repo->gitconfig_author_email);
535 		repo->gitconfig_author_email = NULL;
536 
537 		free(repo->global_gitconfig_author_name);
538 		repo->global_gitconfig_author_name = NULL;
539 		free(repo->global_gitconfig_author_email);
540 		repo->global_gitconfig_author_email = NULL;
541 	}
542 
543 done:
544 	free(repo_gitconfig_path);
545 	return err;
546 }
547 
548 static const struct got_error *
549 read_gotconfig(struct got_repository *repo)
550 {
551 	const struct got_error *err = NULL;
552 	char *gotconfig_path;
553 
554 	gotconfig_path = got_repo_get_path_gotconfig(repo);
555 	if (gotconfig_path == NULL)
556 		return got_error_from_errno("got_repo_get_path_gotconfig");
557 
558 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
559 	free(gotconfig_path);
560 	return err;
561 }
562 
563 /* Supported repository format extensions. */
564 static const char *const repo_extensions[] = {
565 	"noop",			/* Got supports repository format version 1. */
566 	"preciousObjects",	/* Supported by gotadmin cleanup. */
567 	"worktreeConfig",	/* Got does not care about Git work trees. */
568 };
569 
570 const struct got_error *
571 got_repo_open(struct got_repository **repop, const char *path,
572     const char *global_gitconfig_path, int *pack_fds)
573 {
574 	struct got_repository *repo = NULL;
575 	const struct got_error *err = NULL;
576 	char *repo_path = NULL;
577 	size_t i, j = 0;
578 	struct rlimit rl;
579 
580 	*repop = NULL;
581 
582 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
583 		return got_error_from_errno("getrlimit");
584 
585 	repo = calloc(1, sizeof(*repo));
586 	if (repo == NULL)
587 		return got_error_from_errno("calloc");
588 
589 	RB_INIT(&repo->packidx_bloom_filters);
590 	TAILQ_INIT(&repo->packidx_paths);
591 
592 	for (i = 0; i < nitems(repo->privsep_children); i++) {
593 		memset(&repo->privsep_children[i], 0,
594 		    sizeof(repo->privsep_children[0]));
595 		repo->privsep_children[i].imsg_fd = -1;
596 	}
597 
598 	err = got_object_cache_init(&repo->objcache,
599 	    GOT_OBJECT_CACHE_TYPE_OBJ);
600 	if (err)
601 		goto done;
602 	err = got_object_cache_init(&repo->treecache,
603 	    GOT_OBJECT_CACHE_TYPE_TREE);
604 	if (err)
605 		goto done;
606 	err = got_object_cache_init(&repo->commitcache,
607 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
608 	if (err)
609 		goto done;
610 	err = got_object_cache_init(&repo->tagcache,
611 	    GOT_OBJECT_CACHE_TYPE_TAG);
612 	if (err)
613 		goto done;
614 	err = got_object_cache_init(&repo->rawcache,
615 	    GOT_OBJECT_CACHE_TYPE_RAW);
616 	if (err)
617 		goto done;
618 
619 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
620 	if (repo->pack_cache_size > rl.rlim_cur / 8)
621 		repo->pack_cache_size = rl.rlim_cur / 8;
622 	for (i = 0; i < nitems(repo->packs); i++) {
623 		if (i < repo->pack_cache_size) {
624 			repo->packs[i].basefd = pack_fds[j++];
625 			repo->packs[i].accumfd = pack_fds[j++];
626 		} else {
627 			repo->packs[i].basefd = -1;
628 			repo->packs[i].accumfd = -1;
629 		}
630 	}
631 	repo->pinned_pack = -1;
632 	repo->pinned_packidx = -1;
633 	repo->pinned_pid = 0;
634 
635 	repo_path = realpath(path, NULL);
636 	if (repo_path == NULL) {
637 		err = got_error_from_errno2("realpath", path);
638 		goto done;
639 	}
640 
641 	for (;;) {
642 		char *parent_path;
643 
644 		err = open_repo(repo, repo_path);
645 		if (err == NULL)
646 			break;
647 		if (err->code != GOT_ERR_NOT_GIT_REPO)
648 			goto done;
649 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
650 			err = got_error(GOT_ERR_NOT_GIT_REPO);
651 			goto done;
652 		}
653 		err = got_path_dirname(&parent_path, repo_path);
654 		if (err)
655 			goto done;
656 		free(repo_path);
657 		repo_path = parent_path;
658 	}
659 
660 	err = read_gotconfig(repo);
661 	if (err)
662 		goto done;
663 
664 	err = read_gitconfig(repo, global_gitconfig_path);
665 	if (err)
666 		goto done;
667 	if (repo->gitconfig_repository_format_version != 0) {
668 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
669 		goto done;
670 	}
671 	for (i = 0; i < repo->nextensions; i++) {
672 		char *ext = repo->extensions[i];
673 		int j, supported = 0;
674 		for (j = 0; j < nitems(repo_extensions); j++) {
675 			if (strcmp(ext, repo_extensions[j]) == 0) {
676 				supported = 1;
677 				break;
678 			}
679 		}
680 		if (!supported) {
681 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
682 			goto done;
683 		}
684 	}
685 
686 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
687 done:
688 	if (err)
689 		got_repo_close(repo);
690 	else
691 		*repop = repo;
692 	free(repo_path);
693 	return err;
694 }
695 
696 const struct got_error *
697 got_repo_close(struct got_repository *repo)
698 {
699 	const struct got_error *err = NULL, *child_err;
700 	struct got_packidx_bloom_filter *bf;
701 	struct got_pathlist_entry *pe;
702 	size_t i;
703 
704 	for (i = 0; i < repo->pack_cache_size; i++) {
705 		if (repo->packidx_cache[i] == NULL)
706 			break;
707 		got_packidx_close(repo->packidx_cache[i]);
708 	}
709 
710 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
711 	    &repo->packidx_bloom_filters))) {
712 		RB_REMOVE(got_packidx_bloom_filter_tree,
713 		    &repo->packidx_bloom_filters, bf);
714 		bloom_free(bf->bloom);
715 		free(bf->bloom);
716 		free(bf);
717 	}
718 
719 	for (i = 0; i < repo->pack_cache_size; i++)
720 		if (repo->packs[i].path_packfile)
721 			if (repo->packs[i].path_packfile)
722 				got_pack_close(&repo->packs[i]);
723 
724 	free(repo->path);
725 	free(repo->path_git_dir);
726 
727 	got_object_cache_close(&repo->objcache);
728 	got_object_cache_close(&repo->treecache);
729 	got_object_cache_close(&repo->commitcache);
730 	got_object_cache_close(&repo->tagcache);
731 	got_object_cache_close(&repo->rawcache);
732 
733 	for (i = 0; i < nitems(repo->privsep_children); i++) {
734 		if (repo->privsep_children[i].imsg_fd == -1)
735 			continue;
736 		imsg_clear(repo->privsep_children[i].ibuf);
737 		free(repo->privsep_children[i].ibuf);
738 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
739 		child_err = got_privsep_wait_for_child(
740 		    repo->privsep_children[i].pid);
741 		if (child_err && err == NULL)
742 			err = child_err;
743 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
744 		    err == NULL)
745 			err = got_error_from_errno("close");
746 	}
747 
748 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
749 	    err == NULL)
750 		err = got_error_from_errno("close");
751 
752 	if (repo->gotconfig)
753 		got_gotconfig_free(repo->gotconfig);
754 	free(repo->gitconfig_author_name);
755 	free(repo->gitconfig_author_email);
756 	for (i = 0; i < repo->ngitconfig_remotes; i++)
757 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
758 	free(repo->gitconfig_remotes);
759 	for (i = 0; i < repo->nextensions; i++)
760 		free(repo->extensions[i]);
761 	free(repo->extensions);
762 
763 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
764 		free((void *)pe->path);
765 	got_pathlist_free(&repo->packidx_paths);
766 	free(repo);
767 
768 	return err;
769 }
770 
771 void
772 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
773 {
774 	int i;
775 
776 	free(repo->name);
777 	repo->name = NULL;
778 	free(repo->fetch_url);
779 	repo->fetch_url = NULL;
780 	free(repo->send_url);
781 	repo->send_url = NULL;
782 	for (i = 0; i < repo->nfetch_branches; i++)
783 		free(repo->fetch_branches[i]);
784 	free(repo->fetch_branches);
785 	repo->fetch_branches = NULL;
786 	repo->nfetch_branches = 0;
787 	for (i = 0; i < repo->nsend_branches; i++)
788 		free(repo->send_branches[i]);
789 	free(repo->send_branches);
790 	repo->send_branches = NULL;
791 	repo->nsend_branches = 0;
792 }
793 
794 const struct got_error *
795 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
796     const char *input_path)
797 {
798 	const struct got_error *err = NULL;
799 	const char *repo_abspath = NULL;
800 	size_t repolen, len;
801 	char *canonpath, *path = NULL;
802 
803 	*in_repo_path = NULL;
804 
805 	canonpath = strdup(input_path);
806 	if (canonpath == NULL) {
807 		err = got_error_from_errno("strdup");
808 		goto done;
809 	}
810 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
811 	if (err)
812 		goto done;
813 
814 	repo_abspath = got_repo_get_path(repo);
815 
816 	if (canonpath[0] == '\0') {
817 		path = strdup(canonpath);
818 		if (path == NULL) {
819 			err = got_error_from_errno("strdup");
820 			goto done;
821 		}
822 	} else {
823 		path = realpath(canonpath, NULL);
824 		if (path == NULL) {
825 			if (errno != ENOENT) {
826 				err = got_error_from_errno2("realpath",
827 				    canonpath);
828 				goto done;
829 			}
830 			/*
831 			 * Path is not on disk.
832 			 * Assume it is already relative to repository root.
833 			 */
834 			path = strdup(canonpath);
835 			if (path == NULL) {
836 				err = got_error_from_errno("strdup");
837 				goto done;
838 			}
839 		}
840 
841 		repolen = strlen(repo_abspath);
842 		len = strlen(path);
843 
844 
845 		if (strcmp(path, repo_abspath) == 0) {
846 			free(path);
847 			path = strdup("");
848 			if (path == NULL) {
849 				err = got_error_from_errno("strdup");
850 				goto done;
851 			}
852 		} else if (len > repolen &&
853 		    got_path_is_child(path, repo_abspath, repolen)) {
854 			/* Matched an on-disk path inside repository. */
855 			if (got_repo_is_bare(repo)) {
856 				/*
857 				 * Matched an on-disk path inside repository
858 				 * database. Treat input as repository-relative.
859 				 */
860 				free(path);
861 				path = canonpath;
862 				canonpath = NULL;
863 			} else {
864 				char *child;
865 				/* Strip common prefix with repository path. */
866 				err = got_path_skip_common_ancestor(&child,
867 				    repo_abspath, path);
868 				if (err)
869 					goto done;
870 				free(path);
871 				path = child;
872 			}
873 		} else {
874 			/*
875 			 * Matched unrelated on-disk path.
876 			 * Treat input as repository-relative.
877 			 */
878 			free(path);
879 			path = canonpath;
880 			canonpath = NULL;
881 		}
882 	}
883 
884 	/* Make in-repository path absolute */
885 	if (path[0] != '/') {
886 		char *abspath;
887 		if (asprintf(&abspath, "/%s", path) == -1) {
888 			err = got_error_from_errno("asprintf");
889 			goto done;
890 		}
891 		free(path);
892 		path = abspath;
893 	}
894 
895 done:
896 	free(canonpath);
897 	if (err)
898 		free(path);
899 	else
900 		*in_repo_path = path;
901 	return err;
902 }
903 
904 static const struct got_error *
905 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
906     const char *path_packidx)
907 {
908 	const struct got_error *err = NULL;
909 	size_t i;
910 
911 	for (i = 0; i < repo->pack_cache_size; i++) {
912 		if (repo->packidx_cache[i] == NULL)
913 			break;
914 		if (strcmp(repo->packidx_cache[i]->path_packidx,
915 		    path_packidx) == 0) {
916 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
917 		}
918 	}
919 	if (i == repo->pack_cache_size) {
920 		do {
921 			i--;
922 		} while (i > 0 && repo->pinned_packidx >= 0 &&
923 		    i == repo->pinned_packidx);
924 		err = got_packidx_close(repo->packidx_cache[i]);
925 		if (err)
926 			return err;
927 	}
928 
929 	repo->packidx_cache[i] = packidx;
930 
931 	return NULL;
932 }
933 
934 int
935 got_repo_is_packidx_filename(const char *name, size_t len)
936 {
937 	if (len != GOT_PACKIDX_NAMELEN)
938 		return 0;
939 
940 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
941 		return 0;
942 
943 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
944 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
945 		return 0;
946 
947 	return 1;
948 }
949 
950 static struct got_packidx_bloom_filter *
951 get_packidx_bloom_filter(struct got_repository *repo,
952     const char *path, size_t path_len)
953 {
954 	struct got_packidx_bloom_filter key;
955 
956 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
957 		return NULL; /* XXX */
958 	key.path_len = path_len;
959 
960 	return RB_FIND(got_packidx_bloom_filter_tree,
961 	    &repo->packidx_bloom_filters, &key);
962 }
963 
964 int
965 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
966     const char *path_packidx, struct got_object_id *id)
967 {
968 	struct got_packidx_bloom_filter *bf;
969 
970 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
971 	if (bf)
972 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
973 
974 	/* No bloom filter means this pack index must be searched. */
975 	return 1;
976 }
977 
978 static const struct got_error *
979 add_packidx_bloom_filter(struct got_repository *repo,
980     struct got_packidx *packidx, const char *path_packidx)
981 {
982 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
983 	struct got_packidx_bloom_filter *bf;
984 	size_t len;
985 
986 	/*
987 	 * Don't use bloom filters for very large pack index files.
988 	 * Large pack files will contain a relatively large fraction
989 	 * of our objects so we will likely need to visit them anyway.
990 	 * The more objects a pack file contains the higher the probability
991 	 * of a false-positive match from the bloom filter. And reading
992 	 * all object IDs from a large pack index file can be expensive.
993 	 */
994 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
995 		return NULL;
996 
997 	/* Do we already have a filter for this pack index? */
998 	if (get_packidx_bloom_filter(repo, path_packidx,
999 	    strlen(path_packidx)) != NULL)
1000 		return NULL;
1001 
1002 	bf = calloc(1, sizeof(*bf));
1003 	if (bf == NULL)
1004 		return got_error_from_errno("calloc");
1005 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1006 	if (bf->bloom == NULL) {
1007 		free(bf);
1008 		return got_error_from_errno("calloc");
1009 	}
1010 
1011 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1012 	if (len >= sizeof(bf->path)) {
1013 		free(bf->bloom);
1014 		free(bf);
1015 		return got_error(GOT_ERR_NO_SPACE);
1016 	}
1017 	bf->path_len = len;
1018 
1019 	/* Minimum size supported by our bloom filter is 1000 entries. */
1020 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1021 	for (i = 0; i < nobjects; i++) {
1022 		struct got_packidx_object_id *id;
1023 		id = &packidx->hdr.sorted_ids[i];
1024 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1025 	}
1026 
1027 	RB_INSERT(got_packidx_bloom_filter_tree,
1028 	    &repo->packidx_bloom_filters, bf);
1029 	return NULL;
1030 }
1031 
1032 const struct got_error *
1033 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1034     struct got_repository *repo, struct got_object_id *id)
1035 {
1036 	const struct got_error *err;
1037 	struct got_pathlist_entry *pe;
1038 	size_t i;
1039 
1040 	/* Search pack index cache. */
1041 	for (i = 0; i < repo->pack_cache_size; i++) {
1042 		if (repo->packidx_cache[i] == NULL)
1043 			break;
1044 		if (!got_repo_check_packidx_bloom_filter(repo,
1045 		    repo->packidx_cache[i]->path_packidx, id))
1046 			continue; /* object will not be found in this index */
1047 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1048 		if (*idx != -1) {
1049 			*packidx = repo->packidx_cache[i];
1050 			/*
1051 			 * Move this cache entry to the front. Repeatedly
1052 			 * searching a wrong pack index can be expensive.
1053 			 */
1054 			if (i > 0) {
1055 				memmove(&repo->packidx_cache[1],
1056 				    &repo->packidx_cache[0],
1057 				    i * sizeof(repo->packidx_cache[0]));
1058 				repo->packidx_cache[0] = *packidx;
1059 				if (repo->pinned_packidx >= 0 &&
1060 				    repo->pinned_packidx < i)
1061 					repo->pinned_packidx++;
1062 				else if (repo->pinned_packidx == i)
1063 					repo->pinned_packidx = 0;
1064 			}
1065 			return NULL;
1066 		}
1067 	}
1068 	/* No luck. Search the filesystem. */
1069 
1070 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1071 		const char *path_packidx = pe->path;
1072 		int is_cached = 0;
1073 
1074 		if (!got_repo_check_packidx_bloom_filter(repo,
1075 		    pe->path, id))
1076 			continue; /* object will not be found in this index */
1077 
1078 		for (i = 0; i < repo->pack_cache_size; i++) {
1079 			if (repo->packidx_cache[i] == NULL)
1080 				break;
1081 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1082 			    path_packidx) == 0) {
1083 				is_cached = 1;
1084 				break;
1085 			}
1086 		}
1087 		if (is_cached)
1088 			continue; /* already searched */
1089 
1090 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1091 		    path_packidx, 0);
1092 		if (err)
1093 			goto done;
1094 
1095 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1096 		if (err)
1097 			goto done;
1098 
1099 		err = cache_packidx(repo, *packidx, path_packidx);
1100 		if (err)
1101 			goto done;
1102 
1103 		*idx = got_packidx_get_object_idx(*packidx, id);
1104 		if (*idx != -1) {
1105 			err = NULL; /* found the object */
1106 			goto done;
1107 		}
1108 	}
1109 
1110 	err = got_error_no_obj(id);
1111 done:
1112 	return err;
1113 }
1114 
1115 const struct got_error *
1116 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1117     struct got_repository *repo)
1118 {
1119 	const struct got_error *err = NULL;
1120 	DIR *packdir = NULL;
1121 	struct dirent *dent;
1122 	char *path_packidx = NULL;
1123 	int packdir_fd;
1124 	struct stat sb;
1125 
1126 	packdir_fd = openat(got_repo_get_fd(repo),
1127 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1128 	if (packdir_fd == -1) {
1129 		return got_error_from_errno_fmt("openat: %s/%s",
1130 		    got_repo_get_path_git_dir(repo),
1131 		    GOT_OBJECTS_PACK_DIR);
1132 	}
1133 
1134 	packdir = fdopendir(packdir_fd);
1135 	if (packdir == NULL) {
1136 		err = got_error_from_errno("fdopendir");
1137 		goto done;
1138 	}
1139 
1140 	if (fstat(packdir_fd, &sb) == -1) {
1141 		err = got_error_from_errno("fstat");
1142 		goto done;
1143 	}
1144 	repo->pack_path_mtime = sb.st_mtime;
1145 
1146 	while ((dent = readdir(packdir)) != NULL) {
1147 		if (!got_repo_is_packidx_filename(dent->d_name,
1148 		    strlen(dent->d_name)))
1149 			continue;
1150 
1151 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1152 		    dent->d_name) == -1) {
1153 			err = got_error_from_errno("asprintf");
1154 			path_packidx = NULL;
1155 			break;
1156 		}
1157 
1158 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1159 		if (err)
1160 			break;
1161 	}
1162 done:
1163 	if (err)
1164 		free(path_packidx);
1165 	if (packdir && closedir(packdir) != 0 && err == NULL)
1166 		err = got_error_from_errno("closedir");
1167 	return err;
1168 }
1169 
1170 const struct got_error *
1171 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1172     struct got_repository *repo)
1173 {
1174 	const struct got_error *err;
1175 	size_t i;
1176 
1177 	*packidx = NULL;
1178 
1179 	/* Search pack index cache. */
1180 	for (i = 0; i < repo->pack_cache_size; i++) {
1181 		if (repo->packidx_cache[i] == NULL)
1182 			break;
1183 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1184 		    path_packidx) == 0) {
1185 			*packidx = repo->packidx_cache[i];
1186 			return NULL;
1187 		}
1188 	}
1189 	/* No luck. Search the filesystem. */
1190 
1191 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1192 	    path_packidx, 0);
1193 	if (err)
1194 		return err;
1195 
1196 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1197 	if (err)
1198 		goto done;
1199 
1200 	err = cache_packidx(repo, *packidx, path_packidx);
1201 done:
1202 	if (err) {
1203 		got_packidx_close(*packidx);
1204 		*packidx = NULL;
1205 	}
1206 	return err;
1207 }
1208 
1209 static const struct got_error *
1210 read_packfile_hdr(int fd, struct got_packidx *packidx)
1211 {
1212 	const struct got_error *err = NULL;
1213 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1214 	struct got_packfile_hdr hdr;
1215 	ssize_t n;
1216 
1217 	n = read(fd, &hdr, sizeof(hdr));
1218 	if (n < 0)
1219 		return got_error_from_errno("read");
1220 	if (n != sizeof(hdr))
1221 		return got_error(GOT_ERR_BAD_PACKFILE);
1222 
1223 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1224 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1225 	    be32toh(hdr.nobjects) != totobj)
1226 		err = got_error(GOT_ERR_BAD_PACKFILE);
1227 
1228 	return err;
1229 }
1230 
1231 static const struct got_error *
1232 open_packfile(int *fd, struct got_repository *repo,
1233     const char *relpath, struct got_packidx *packidx)
1234 {
1235 	const struct got_error *err = NULL;
1236 
1237 	*fd = openat(got_repo_get_fd(repo), relpath,
1238 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1239 	if (*fd == -1)
1240 		return got_error_from_errno_fmt("openat: %s/%s",
1241 		    got_repo_get_path_git_dir(repo), relpath);
1242 
1243 	if (packidx) {
1244 		err = read_packfile_hdr(*fd, packidx);
1245 		if (err) {
1246 			close(*fd);
1247 			*fd = -1;
1248 		}
1249 	}
1250 
1251 	return err;
1252 }
1253 
1254 const struct got_error *
1255 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1256     const char *path_packfile, struct got_packidx *packidx)
1257 {
1258 	const struct got_error *err = NULL;
1259 	struct got_pack *pack = NULL;
1260 	struct stat sb;
1261 	size_t i;
1262 
1263 	if (packp)
1264 		*packp = NULL;
1265 
1266 	for (i = 0; i < repo->pack_cache_size; i++) {
1267 		pack = &repo->packs[i];
1268 		if (pack->path_packfile == NULL)
1269 			break;
1270 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1271 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1272 	}
1273 
1274 	if (i == repo->pack_cache_size) {
1275 		struct got_pack tmp;
1276 		do {
1277 			i--;
1278 		} while (i > 0 && repo->pinned_pack >= 0 &&
1279 		    i == repo->pinned_pack);
1280 		err = got_pack_close(&repo->packs[i]);
1281 		if (err)
1282 			return err;
1283 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1284 			return got_error_from_errno("ftruncate");
1285 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1286 			return got_error_from_errno("ftruncate");
1287 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1288 		memcpy(&repo->packs[i], &repo->packs[0],
1289 		    sizeof(repo->packs[i]));
1290 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1291 		if (repo->pinned_pack == 0)
1292 			repo->pinned_pack = i;
1293 		else if (repo->pinned_pack == i)
1294 			repo->pinned_pack = 0;
1295 		i = 0;
1296 	}
1297 
1298 	pack = &repo->packs[i];
1299 
1300 	pack->path_packfile = strdup(path_packfile);
1301 	if (pack->path_packfile == NULL) {
1302 		err = got_error_from_errno("strdup");
1303 		goto done;
1304 	}
1305 
1306 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1307 	if (err)
1308 		goto done;
1309 
1310 	if (fstat(pack->fd, &sb) != 0) {
1311 		err = got_error_from_errno("fstat");
1312 		goto done;
1313 	}
1314 	pack->filesize = sb.st_size;
1315 
1316 	pack->privsep_child = NULL;
1317 
1318 #ifndef GOT_PACK_NO_MMAP
1319 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1320 	    pack->fd, 0);
1321 	if (pack->map == MAP_FAILED) {
1322 		if (errno != ENOMEM) {
1323 			err = got_error_from_errno("mmap");
1324 			goto done;
1325 		}
1326 		pack->map = NULL; /* fall back to read(2) */
1327 	}
1328 #endif
1329 done:
1330 	if (err) {
1331 		if (pack) {
1332 			free(pack->path_packfile);
1333 			memset(pack, 0, sizeof(*pack));
1334 		}
1335 	} else if (packp)
1336 		*packp = pack;
1337 	return err;
1338 }
1339 
1340 struct got_pack *
1341 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1342 {
1343 	struct got_pack *pack = NULL;
1344 	size_t i;
1345 
1346 	for (i = 0; i < repo->pack_cache_size; i++) {
1347 		pack = &repo->packs[i];
1348 		if (pack->path_packfile == NULL)
1349 			break;
1350 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1351 			return pack;
1352 	}
1353 
1354 	return NULL;
1355 }
1356 
1357 const struct got_error *
1358 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1359     struct got_pack *pack)
1360 {
1361 	size_t i;
1362 	int pinned_pack = -1, pinned_packidx = -1;
1363 
1364 	for (i = 0; i < repo->pack_cache_size; i++) {
1365 		if (repo->packidx_cache[i] &&
1366 		    strcmp(repo->packidx_cache[i]->path_packidx,
1367 		    packidx->path_packidx) == 0)
1368 			pinned_packidx = i;
1369 		if (repo->packs[i].path_packfile &&
1370 		    strcmp(repo->packs[i].path_packfile,
1371 		    pack->path_packfile) == 0)
1372 			pinned_pack = i;
1373 	}
1374 
1375 	if (pinned_packidx == -1 || pinned_pack == -1)
1376 		return got_error(GOT_ERR_PIN_PACK);
1377 
1378 	repo->pinned_pack = pinned_pack;
1379 	repo->pinned_packidx = pinned_packidx;
1380 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1381 	return NULL;
1382 }
1383 
1384 struct got_pack *
1385 got_repo_get_pinned_pack(struct got_repository *repo)
1386 {
1387 	if (repo->pinned_pack >= 0 &&
1388 	    repo->pinned_pack < repo->pack_cache_size)
1389 		return &repo->packs[repo->pinned_pack];
1390 
1391 	return NULL;
1392 }
1393 
1394 void
1395 got_repo_unpin_pack(struct got_repository *repo)
1396 {
1397 	repo->pinned_packidx = -1;
1398 	repo->pinned_pack = -1;
1399 	repo->pinned_pid = 0;
1400 }
1401 
1402 const struct got_error *
1403 got_repo_init(const char *repo_path, const char *head_name)
1404 {
1405 	const struct got_error *err = NULL;
1406 	const char *dirnames[] = {
1407 		GOT_OBJECTS_DIR,
1408 		GOT_OBJECTS_PACK_DIR,
1409 		GOT_REFS_DIR,
1410 	};
1411 	const char *description_str = "Unnamed repository; "
1412 	    "edit this file 'description' to name the repository.";
1413 	const char *headref = "ref: refs/heads/";
1414 	const char *gitconfig_str = "[core]\n"
1415 	    "\trepositoryformatversion = 0\n"
1416 	    "\tfilemode = true\n"
1417 	    "\tbare = true\n";
1418 	char *headref_str, *path;
1419 	size_t i;
1420 
1421 	if (!got_path_dir_is_empty(repo_path))
1422 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1423 
1424 	for (i = 0; i < nitems(dirnames); i++) {
1425 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1426 			return got_error_from_errno("asprintf");
1427 		}
1428 		err = got_path_mkdir(path);
1429 		free(path);
1430 		if (err)
1431 			return err;
1432 	}
1433 
1434 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1435 		return got_error_from_errno("asprintf");
1436 	err = got_path_create_file(path, description_str);
1437 	free(path);
1438 	if (err)
1439 		return err;
1440 
1441 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1442 		return got_error_from_errno("asprintf");
1443 	if (asprintf(&headref_str, "%s%s", headref,
1444 	    head_name ? head_name : "main") == -1) {
1445 		free(path);
1446 		return got_error_from_errno("asprintf");
1447 	}
1448 	err = got_path_create_file(path, headref_str);
1449 	free(headref_str);
1450 	free(path);
1451 	if (err)
1452 		return err;
1453 
1454 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1455 		return got_error_from_errno("asprintf");
1456 	err = got_path_create_file(path, gitconfig_str);
1457 	free(path);
1458 	if (err)
1459 		return err;
1460 
1461 	return NULL;
1462 }
1463 
1464 static void
1465 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1466 {
1467 	struct got_pathlist_entry *pe;
1468 
1469 	while (!TAILQ_EMPTY(packidx_paths)) {
1470 		pe = TAILQ_FIRST(packidx_paths);
1471 		TAILQ_REMOVE(packidx_paths, pe, entry);
1472 		free((char *)pe->path);
1473 		free(pe);
1474 	}
1475 }
1476 
1477 static const struct got_error *
1478 match_packed_object(struct got_object_id **unique_id,
1479     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1480 {
1481 	const struct got_error *err = NULL;
1482 	char *objects_pack_dir = NULL;
1483 	struct got_object_id_queue matched_ids;
1484 	struct got_pathlist_entry *pe;
1485 	struct stat sb;
1486 
1487 	STAILQ_INIT(&matched_ids);
1488 
1489 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1490 	if (objects_pack_dir == NULL)
1491 		return got_error_from_errno("got_repo_get_path_objects_pack");
1492 
1493 	if (stat(objects_pack_dir, &sb) == -1) {
1494 		if (errno != ENOENT) {
1495 			err = got_error_from_errno2("stat", objects_pack_dir);
1496 			goto done;
1497 		}
1498 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1499 		purge_packidx_paths(&repo->packidx_paths);
1500 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1501 		if (err)
1502 			goto done;
1503 	}
1504 
1505 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1506 		const char *path_packidx = pe->path;
1507 		struct got_packidx *packidx;
1508 		struct got_object_qid *qid;
1509 
1510 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1511 		    path_packidx, 0);
1512 		if (err)
1513 			break;
1514 
1515 		err = got_packidx_match_id_str_prefix(&matched_ids,
1516 		    packidx, id_str_prefix);
1517 		if (err) {
1518 			got_packidx_close(packidx);
1519 			break;
1520 		}
1521 		err = got_packidx_close(packidx);
1522 		if (err)
1523 			break;
1524 
1525 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1526 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1527 				int matched_type;
1528 				err = got_object_get_type(&matched_type, repo,
1529 				    &qid->id);
1530 				if (err)
1531 					goto done;
1532 				if (matched_type != obj_type)
1533 					continue;
1534 			}
1535 			if (*unique_id == NULL) {
1536 				*unique_id = got_object_id_dup(&qid->id);
1537 				if (*unique_id == NULL) {
1538 					err = got_error_from_errno("malloc");
1539 					goto done;
1540 				}
1541 			} else {
1542 				if (got_object_id_cmp(*unique_id,
1543 				    &qid->id) == 0)
1544 					continue; /* packed multiple times */
1545 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1546 				goto done;
1547 			}
1548 		}
1549 	}
1550 done:
1551 	free(objects_pack_dir);
1552 	got_object_id_queue_free(&matched_ids);
1553 	if (err) {
1554 		free(*unique_id);
1555 		*unique_id = NULL;
1556 	}
1557 	return err;
1558 }
1559 
1560 static const struct got_error *
1561 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1562     const char *object_dir, const char *id_str_prefix, int obj_type,
1563     struct got_repository *repo)
1564 {
1565 	const struct got_error *err = NULL;
1566 	char *path, *id_str = NULL;
1567 	DIR *dir = NULL;
1568 	struct dirent *dent;
1569 	struct got_object_id id;
1570 
1571 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1572 		err = got_error_from_errno("asprintf");
1573 		goto done;
1574 	}
1575 
1576 	dir = opendir(path);
1577 	if (dir == NULL) {
1578 		if (errno == ENOENT) {
1579 			err = NULL;
1580 			goto done;
1581 		}
1582 		err = got_error_from_errno2("opendir", path);
1583 		goto done;
1584 	}
1585 	while ((dent = readdir(dir)) != NULL) {
1586 		int cmp;
1587 
1588 		free(id_str);
1589 		id_str = NULL;
1590 
1591 		if (strcmp(dent->d_name, ".") == 0 ||
1592 		    strcmp(dent->d_name, "..") == 0)
1593 			continue;
1594 
1595 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1596 			err = got_error_from_errno("asprintf");
1597 			goto done;
1598 		}
1599 
1600 		if (!got_parse_sha1_digest(id.sha1, id_str))
1601 			continue;
1602 
1603 		/*
1604 		 * Directory entries do not necessarily appear in
1605 		 * sorted order, so we must iterate over all of them.
1606 		 */
1607 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1608 		if (cmp != 0)
1609 			continue;
1610 
1611 		if (*unique_id == NULL) {
1612 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1613 				int matched_type;
1614 				err = got_object_get_type(&matched_type, repo,
1615 				    &id);
1616 				if (err)
1617 					goto done;
1618 				if (matched_type != obj_type)
1619 					continue;
1620 			}
1621 			*unique_id = got_object_id_dup(&id);
1622 			if (*unique_id == NULL) {
1623 				err = got_error_from_errno("got_object_id_dup");
1624 				goto done;
1625 			}
1626 		} else {
1627 			if (got_object_id_cmp(*unique_id, &id) == 0)
1628 				continue; /* both packed and loose */
1629 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1630 			goto done;
1631 		}
1632 	}
1633 done:
1634 	if (dir && closedir(dir) != 0 && err == NULL)
1635 		err = got_error_from_errno("closedir");
1636 	if (err) {
1637 		free(*unique_id);
1638 		*unique_id = NULL;
1639 	}
1640 	free(id_str);
1641 	free(path);
1642 	return err;
1643 }
1644 
1645 const struct got_error *
1646 got_repo_match_object_id_prefix(struct got_object_id **id,
1647     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1648 {
1649 	const struct got_error *err = NULL;
1650 	char *path_objects = NULL, *object_dir = NULL;
1651 	size_t len;
1652 	int i;
1653 
1654 	*id = NULL;
1655 
1656 	path_objects = got_repo_get_path_objects(repo);
1657 
1658 	len = strlen(id_str_prefix);
1659 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1660 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1661 		goto done;
1662 	}
1663 
1664 	for (i = 0; i < len; i++) {
1665 		if (isxdigit((unsigned char)id_str_prefix[i]))
1666 			continue;
1667 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1668 		goto done;
1669 	}
1670 
1671 	if (len >= 2) {
1672 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1673 		if (err)
1674 			goto done;
1675 		object_dir = strndup(id_str_prefix, 2);
1676 		if (object_dir == NULL) {
1677 			err = got_error_from_errno("strdup");
1678 			goto done;
1679 		}
1680 		err = match_loose_object(id, path_objects, object_dir,
1681 		    id_str_prefix, obj_type, repo);
1682 	} else if (len == 1) {
1683 		int i;
1684 		for (i = 0; i < 0xf; i++) {
1685 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1686 			    == -1) {
1687 				err = got_error_from_errno("asprintf");
1688 				goto done;
1689 			}
1690 			err = match_packed_object(id, repo, object_dir,
1691 			    obj_type);
1692 			if (err)
1693 				goto done;
1694 			err = match_loose_object(id, path_objects, object_dir,
1695 			    id_str_prefix, obj_type, repo);
1696 			if (err)
1697 				goto done;
1698 		}
1699 	} else {
1700 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1701 		goto done;
1702 	}
1703 done:
1704 	free(path_objects);
1705 	free(object_dir);
1706 	if (err) {
1707 		free(*id);
1708 		*id = NULL;
1709 	} else if (*id == NULL) {
1710 		switch (obj_type) {
1711 		case GOT_OBJ_TYPE_BLOB:
1712 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1713 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1714 			break;
1715 		case GOT_OBJ_TYPE_TREE:
1716 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1717 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1718 			break;
1719 		case GOT_OBJ_TYPE_COMMIT:
1720 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1721 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1722 			break;
1723 		case GOT_OBJ_TYPE_TAG:
1724 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1725 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1726 			break;
1727 		default:
1728 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1729 			break;
1730 		}
1731 	}
1732 
1733 	return err;
1734 }
1735 
1736 const struct got_error *
1737 got_repo_match_object_id(struct got_object_id **id, char **label,
1738     const char *id_str, int obj_type, struct got_reflist_head *refs,
1739     struct got_repository *repo)
1740 {
1741 	const struct got_error *err;
1742 	struct got_tag_object *tag;
1743 	struct got_reference *ref = NULL;
1744 
1745 	*id = NULL;
1746 	if (label)
1747 		*label = NULL;
1748 
1749 	if (refs) {
1750 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1751 		    refs, repo);
1752 		if (err == NULL) {
1753 			*id = got_object_id_dup(
1754 			    got_object_tag_get_object_id(tag));
1755 			if (*id == NULL)
1756 				err = got_error_from_errno("got_object_id_dup");
1757 			else if (label && asprintf(label, "refs/tags/%s",
1758 			    got_object_tag_get_name(tag)) == -1) {
1759 				err = got_error_from_errno("asprintf");
1760 				free(*id);
1761 				*id = NULL;
1762 			}
1763 			got_object_tag_close(tag);
1764 			return err;
1765 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1766 		    err->code != GOT_ERR_NO_OBJ)
1767 			return err;
1768 	}
1769 
1770 	err = got_ref_open(&ref, repo, id_str, 0);
1771 	if (err == NULL) {
1772 		err = got_ref_resolve(id, repo, ref);
1773 		if (err)
1774 			goto done;
1775 		if (label) {
1776 			*label = strdup(got_ref_get_name(ref));
1777 			if (*label == NULL) {
1778 				err = got_error_from_errno("strdup");
1779 				goto done;
1780 			}
1781 		}
1782 	} else {
1783 		if (err->code != GOT_ERR_NOT_REF &&
1784 		    err->code != GOT_ERR_BAD_REF_NAME)
1785 			goto done;
1786 		err = got_repo_match_object_id_prefix(id, id_str,
1787 		    obj_type, repo);
1788 		if (err) {
1789 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1790 				err = got_error_not_ref(id_str);
1791 			goto done;
1792 		}
1793 		if (label) {
1794 			err = got_object_id_str(label, *id);
1795 			if (*label == NULL) {
1796 				err = got_error_from_errno("strdup");
1797 				goto done;
1798 			}
1799 		}
1800 	}
1801 done:
1802 	if (ref)
1803 		got_ref_close(ref);
1804 	return err;
1805 }
1806 
1807 const struct got_error *
1808 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1809     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1810 {
1811 	const struct got_error *err = NULL;
1812 	struct got_reflist_entry *re;
1813 	struct got_object_id *tag_id;
1814 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1815 
1816 	*tag = NULL;
1817 
1818 	TAILQ_FOREACH(re, refs, entry) {
1819 		const char *refname;
1820 		refname = got_ref_get_name(re->ref);
1821 		if (got_ref_is_symbolic(re->ref))
1822 			continue;
1823 		if (strncmp(refname, "refs/tags/", 10) != 0)
1824 			continue;
1825 		if (!name_is_absolute)
1826 			refname += strlen("refs/tags/");
1827 		if (strcmp(refname, name) != 0)
1828 			continue;
1829 		err = got_ref_resolve(&tag_id, repo, re->ref);
1830 		if (err)
1831 			break;
1832 		err = got_object_open_as_tag(tag, repo, tag_id);
1833 		free(tag_id);
1834 		if (err)
1835 			break;
1836 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1837 		    got_object_tag_get_object_type(*tag) == obj_type)
1838 			break;
1839 		got_object_tag_close(*tag);
1840 		*tag = NULL;
1841 	}
1842 
1843 	if (err == NULL && *tag == NULL)
1844 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1845 		    GOT_OBJ_LABEL_TAG, name);
1846 	return err;
1847 }
1848 
1849 static const struct got_error *
1850 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1851     const char *name, mode_t mode, struct got_object_id *blob_id)
1852 {
1853 	const struct got_error *err = NULL;
1854 
1855 	 *new_te = NULL;
1856 
1857 	*new_te = calloc(1, sizeof(**new_te));
1858 	if (*new_te == NULL)
1859 		return got_error_from_errno("calloc");
1860 
1861 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1862 	    sizeof((*new_te)->name)) {
1863 		err = got_error(GOT_ERR_NO_SPACE);
1864 		goto done;
1865 	}
1866 
1867 	if (S_ISLNK(mode)) {
1868 		(*new_te)->mode = S_IFLNK;
1869 	} else {
1870 		(*new_te)->mode = S_IFREG;
1871 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1872 	}
1873 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1874 done:
1875 	if (err && *new_te) {
1876 		free(*new_te);
1877 		*new_te = NULL;
1878 	}
1879 	return err;
1880 }
1881 
1882 static const struct got_error *
1883 import_file(struct got_tree_entry **new_te, struct dirent *de,
1884     const char *path, struct got_repository *repo)
1885 {
1886 	const struct got_error *err;
1887 	struct got_object_id *blob_id = NULL;
1888 	char *filepath;
1889 	struct stat sb;
1890 
1891 	if (asprintf(&filepath, "%s%s%s", path,
1892 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1893 		return got_error_from_errno("asprintf");
1894 
1895 	if (lstat(filepath, &sb) != 0) {
1896 		err = got_error_from_errno2("lstat", path);
1897 		goto done;
1898 	}
1899 
1900 	err = got_object_blob_create(&blob_id, filepath, repo);
1901 	if (err)
1902 		goto done;
1903 
1904 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1905 	    blob_id);
1906 done:
1907 	free(filepath);
1908 	if (err)
1909 		free(blob_id);
1910 	return err;
1911 }
1912 
1913 static const struct got_error *
1914 insert_tree_entry(struct got_tree_entry *new_te,
1915     struct got_pathlist_head *paths)
1916 {
1917 	const struct got_error *err = NULL;
1918 	struct got_pathlist_entry *new_pe;
1919 
1920 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1921 	if (err)
1922 		return err;
1923 	if (new_pe == NULL)
1924 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1925 	return NULL;
1926 }
1927 
1928 static const struct got_error *write_tree(struct got_object_id **,
1929     const char *, struct got_pathlist_head *, struct got_repository *,
1930     got_repo_import_cb progress_cb, void *progress_arg);
1931 
1932 static const struct got_error *
1933 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1934     const char *path, struct got_pathlist_head *ignores,
1935     struct got_repository *repo,
1936     got_repo_import_cb progress_cb, void *progress_arg)
1937 {
1938 	const struct got_error *err;
1939 	struct got_object_id *id = NULL;
1940 	char *subdirpath;
1941 
1942 	if (asprintf(&subdirpath, "%s%s%s", path,
1943 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1944 		return got_error_from_errno("asprintf");
1945 
1946 	(*new_te) = calloc(1, sizeof(**new_te));
1947 	if (*new_te == NULL)
1948 		return got_error_from_errno("calloc");
1949 	(*new_te)->mode = S_IFDIR;
1950 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1951 	    sizeof((*new_te)->name)) {
1952 		err = got_error(GOT_ERR_NO_SPACE);
1953 		goto done;
1954 	}
1955 	err = write_tree(&id, subdirpath, ignores,  repo,
1956 	    progress_cb, progress_arg);
1957 	if (err)
1958 		goto done;
1959 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1960 
1961 done:
1962 	free(id);
1963 	free(subdirpath);
1964 	if (err) {
1965 		free(*new_te);
1966 		*new_te = NULL;
1967 	}
1968 	return err;
1969 }
1970 
1971 static const struct got_error *
1972 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1973     struct got_pathlist_head *ignores, struct got_repository *repo,
1974     got_repo_import_cb progress_cb, void *progress_arg)
1975 {
1976 	const struct got_error *err = NULL;
1977 	DIR *dir;
1978 	struct dirent *de;
1979 	int nentries;
1980 	struct got_tree_entry *new_te = NULL;
1981 	struct got_pathlist_head paths;
1982 	struct got_pathlist_entry *pe;
1983 
1984 	*new_tree_id = NULL;
1985 
1986 	TAILQ_INIT(&paths);
1987 
1988 	dir = opendir(path_dir);
1989 	if (dir == NULL) {
1990 		err = got_error_from_errno2("opendir", path_dir);
1991 		goto done;
1992 	}
1993 
1994 	nentries = 0;
1995 	while ((de = readdir(dir)) != NULL) {
1996 		int ignore = 0;
1997 		int type;
1998 
1999 		if (strcmp(de->d_name, ".") == 0 ||
2000 		    strcmp(de->d_name, "..") == 0)
2001 			continue;
2002 
2003 		TAILQ_FOREACH(pe, ignores, entry) {
2004 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2005 				ignore = 1;
2006 				break;
2007 			}
2008 		}
2009 		if (ignore)
2010 			continue;
2011 
2012 		err = got_path_dirent_type(&type, path_dir, de);
2013 		if (err)
2014 			goto done;
2015 
2016 		if (type == DT_DIR) {
2017 			err = import_subdir(&new_te, de, path_dir,
2018 			    ignores, repo, progress_cb, progress_arg);
2019 			if (err) {
2020 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2021 					goto done;
2022 				err = NULL;
2023 				continue;
2024 			}
2025 		} else if (type == DT_REG || type == DT_LNK) {
2026 			err = import_file(&new_te, de, path_dir, repo);
2027 			if (err)
2028 				goto done;
2029 		} else
2030 			continue;
2031 
2032 		err = insert_tree_entry(new_te, &paths);
2033 		if (err)
2034 			goto done;
2035 		nentries++;
2036 	}
2037 
2038 	if (TAILQ_EMPTY(&paths)) {
2039 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2040 		    "cannot create tree without any entries");
2041 		goto done;
2042 	}
2043 
2044 	TAILQ_FOREACH(pe, &paths, entry) {
2045 		struct got_tree_entry *te = pe->data;
2046 		char *path;
2047 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2048 			continue;
2049 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2050 			err = got_error_from_errno("asprintf");
2051 			goto done;
2052 		}
2053 		err = (*progress_cb)(progress_arg, path);
2054 		free(path);
2055 		if (err)
2056 			goto done;
2057 	}
2058 
2059 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2060 done:
2061 	if (dir)
2062 		closedir(dir);
2063 	got_pathlist_free(&paths);
2064 	return err;
2065 }
2066 
2067 const struct got_error *
2068 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2069     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2070     struct got_repository *repo, got_repo_import_cb progress_cb,
2071     void *progress_arg)
2072 {
2073 	const struct got_error *err;
2074 	struct got_object_id *new_tree_id;
2075 
2076 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2077 	    progress_cb, progress_arg);
2078 	if (err)
2079 		return err;
2080 
2081 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2082 	    author, time(NULL), author, time(NULL), logmsg, repo);
2083 	free(new_tree_id);
2084 	return err;
2085 }
2086 
2087 const struct got_error *
2088 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2089     struct got_repository *repo)
2090 {
2091 	const struct got_error *err = NULL;
2092 	char *path_objects = NULL, *path = NULL;
2093 	DIR *dir = NULL;
2094 	struct got_object_id id;
2095 	int i;
2096 
2097 	*nobjects = 0;
2098 	*ondisk_size = 0;
2099 
2100 	path_objects = got_repo_get_path_objects(repo);
2101 	if (path_objects == NULL)
2102 		return got_error_from_errno("got_repo_get_path_objects");
2103 
2104 	for (i = 0; i <= 0xff; i++) {
2105 		struct dirent *dent;
2106 
2107 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2108 			err = got_error_from_errno("asprintf");
2109 			break;
2110 		}
2111 
2112 		dir = opendir(path);
2113 		if (dir == NULL) {
2114 			if (errno == ENOENT) {
2115 				err = NULL;
2116 				continue;
2117 			}
2118 			err = got_error_from_errno2("opendir", path);
2119 			break;
2120 		}
2121 
2122 		while ((dent = readdir(dir)) != NULL) {
2123 			char *id_str;
2124 			int fd;
2125 			struct stat sb;
2126 
2127 			if (strcmp(dent->d_name, ".") == 0 ||
2128 			    strcmp(dent->d_name, "..") == 0)
2129 				continue;
2130 
2131 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2132 				err = got_error_from_errno("asprintf");
2133 				goto done;
2134 			}
2135 
2136 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2137 				free(id_str);
2138 				continue;
2139 			}
2140 			free(id_str);
2141 
2142 			err = got_object_open_loose_fd(&fd, &id, repo);
2143 			if (err)
2144 				goto done;
2145 
2146 			if (fstat(fd, &sb) == -1) {
2147 				err = got_error_from_errno("fstat");
2148 				close(fd);
2149 				goto done;
2150 			}
2151 			(*nobjects)++;
2152 			(*ondisk_size) += sb.st_size;
2153 
2154 			if (close(fd) == -1) {
2155 				err = got_error_from_errno("close");
2156 				goto done;
2157 			}
2158 		}
2159 
2160 		if (closedir(dir) != 0) {
2161 			err = got_error_from_errno("closedir");
2162 			goto done;
2163 		}
2164 		dir = NULL;
2165 
2166 		free(path);
2167 		path = NULL;
2168 	}
2169 done:
2170 	if (dir && closedir(dir) != 0 && err == NULL)
2171 		err = got_error_from_errno("closedir");
2172 
2173 	if (err) {
2174 		*nobjects = 0;
2175 		*ondisk_size = 0;
2176 	}
2177 	free(path_objects);
2178 	free(path);
2179 	return err;
2180 }
2181 
2182 const struct got_error *
2183 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2184     off_t *total_packsize, struct got_repository *repo)
2185 {
2186 	const struct got_error *err = NULL;
2187 	DIR *packdir = NULL;
2188 	struct dirent *dent;
2189 	struct got_packidx *packidx = NULL;
2190 	char *path_packidx;
2191 	char *path_packfile;
2192 	int packdir_fd;
2193 	struct stat sb;
2194 
2195 	*npackfiles = 0;
2196 	*nobjects = 0;
2197 	*total_packsize = 0;
2198 
2199 	packdir_fd = openat(got_repo_get_fd(repo),
2200 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2201 	if (packdir_fd == -1) {
2202 		return got_error_from_errno_fmt("openat: %s/%s",
2203 		    got_repo_get_path_git_dir(repo),
2204 		    GOT_OBJECTS_PACK_DIR);
2205 	}
2206 
2207 	packdir = fdopendir(packdir_fd);
2208 	if (packdir == NULL) {
2209 		err = got_error_from_errno("fdopendir");
2210 		goto done;
2211 	}
2212 
2213 	while ((dent = readdir(packdir)) != NULL) {
2214 		if (!got_repo_is_packidx_filename(dent->d_name,
2215 		    strlen(dent->d_name)))
2216 			continue;
2217 
2218 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2219 		    dent->d_name) == -1) {
2220 			err = got_error_from_errno("asprintf");
2221 			goto done;
2222 		}
2223 
2224 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2225 		    path_packidx, 0);
2226 		free(path_packidx);
2227 		if (err)
2228 			goto done;
2229 
2230 		if (fstat(packidx->fd, &sb) == -1)
2231 			goto done;
2232 		*total_packsize += sb.st_size;
2233 
2234 		err = got_packidx_get_packfile_path(&path_packfile,
2235 		    packidx->path_packidx);
2236 		if (err)
2237 			goto done;
2238 
2239 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2240 		    0) == -1) {
2241 			free(path_packfile);
2242 			goto done;
2243 		}
2244 		free(path_packfile);
2245 		*total_packsize += sb.st_size;
2246 
2247 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2248 
2249 		(*npackfiles)++;
2250 
2251 		got_packidx_close(packidx);
2252 		packidx = NULL;
2253 	}
2254 done:
2255 	if (packidx)
2256 		got_packidx_close(packidx);
2257 	if (packdir && closedir(packdir) != 0 && err == NULL)
2258 		err = got_error_from_errno("closedir");
2259 	if (err) {
2260 		*npackfiles = 0;
2261 		*nobjects = 0;
2262 		*total_packsize = 0;
2263 	}
2264 	return err;
2265 }
2266 
2267 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2268     got_packidx_bloom_filter_cmp);