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/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
72 
73 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
74     got_packidx_bloom_filter_cmp);
75 
76 const char *
77 got_repo_get_path(struct got_repository *repo)
78 {
79 	return repo->path;
80 }
81 
82 const char *
83 got_repo_get_path_git_dir(struct got_repository *repo)
84 {
85 	return repo->path_git_dir;
86 }
87 
88 int
89 got_repo_get_fd(struct got_repository *repo)
90 {
91 	return repo->gitdir_fd;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_name(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_name;
98 }
99 
100 const char *
101 got_repo_get_gitconfig_author_email(struct got_repository *repo)
102 {
103 	return repo->gitconfig_author_email;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_name;
110 }
111 
112 const char *
113 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
114 {
115 	return repo->global_gitconfig_author_email;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_owner(struct got_repository *repo)
120 {
121 	return repo->gitconfig_owner;
122 }
123 
124 void
125 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
126     struct got_repository *repo)
127 {
128 	*extensions = repo->extensions;
129 	*nextensions = repo->nextensions;
130 }
131 
132 int
133 got_repo_is_bare(struct got_repository *repo)
134 {
135 	return (strcmp(repo->path, repo->path_git_dir) == 0);
136 }
137 
138 static char *
139 get_path_git_child(struct got_repository *repo, const char *basename)
140 {
141 	char *path_child;
142 
143 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
144 	    basename) == -1)
145 		return NULL;
146 
147 	return path_child;
148 }
149 
150 char *
151 got_repo_get_path_objects(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
154 }
155 
156 char *
157 got_repo_get_path_objects_pack(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
160 }
161 
162 char *
163 got_repo_get_path_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_REFS_DIR);
166 }
167 
168 char *
169 got_repo_get_path_packed_refs(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
172 }
173 
174 static char *
175 get_path_head(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_HEAD_FILE);
178 }
179 
180 char *
181 got_repo_get_path_gitconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GITCONFIG);
184 }
185 
186 char *
187 got_repo_get_path_gotconfig(struct got_repository *repo)
188 {
189 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
190 }
191 
192 const struct got_gotconfig *
193 got_repo_get_gotconfig(struct got_repository *repo)
194 {
195 	return repo->gotconfig;
196 }
197 
198 void
199 got_repo_get_gitconfig_remotes(int *nremotes,
200     const struct got_remote_repo **remotes, struct got_repository *repo)
201 {
202 	*nremotes = repo->ngitconfig_remotes;
203 	*remotes = repo->gitconfig_remotes;
204 }
205 
206 static int
207 is_git_repo(struct got_repository *repo)
208 {
209 	const char *path_git = got_repo_get_path_git_dir(repo);
210 	char *path_objects = got_repo_get_path_objects(repo);
211 	char *path_refs = got_repo_get_path_refs(repo);
212 	char *path_head = get_path_head(repo);
213 	int ret = 0;
214 	struct stat sb;
215 	struct got_reference *head_ref;
216 
217 	if (lstat(path_git, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_objects, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_refs, &sb) == -1)
228 		goto done;
229 	if (!S_ISDIR(sb.st_mode))
230 		goto done;
231 
232 	if (lstat(path_head, &sb) == -1)
233 		goto done;
234 	if (!S_ISREG(sb.st_mode))
235 		goto done;
236 
237 	/* Check if the HEAD reference can be opened. */
238 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
239 		goto done;
240 	got_ref_close(head_ref);
241 
242 	ret = 1;
243 done:
244 	free(path_objects);
245 	free(path_refs);
246 	free(path_head);
247 	return ret;
248 
249 }
250 
251 const struct got_error *
252 got_repo_pack_fds_open(int **pack_fds)
253 {
254 	const struct got_error *err = NULL;
255 	int i;
256 
257 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
258 	if (*pack_fds == NULL)
259 		return got_error_from_errno("calloc");
260 
261 	/*
262 	 * got_repo_pack_fds_close will try to close all of the
263 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
264 	 * a value from got_opentempfd(), which would result in a close(0) if
265 	 * we do not initialize to -1 here.
266 	 */
267 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
268 		(*pack_fds)[i] = -1;
269 
270 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
271 		(*pack_fds)[i] = got_opentempfd();
272 		if ((*pack_fds)[i] == -1) {
273 			err = got_error_from_errno("got_opentempfd");
274 			got_repo_pack_fds_close(*pack_fds);
275 			*pack_fds = NULL;
276 			return err;
277 		}
278 	}
279 
280 	return NULL;
281 }
282 
283 const struct got_error *
284 got_repo_pack_fds_close(int *pack_fds)
285 {
286 	const struct got_error *err = NULL;
287 	int i;
288 
289 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
290 		if (pack_fds[i] == -1)
291 			continue;
292 		if (close(pack_fds[i]) == -1) {
293 			err = got_error_from_errno("close");
294 			break;
295 		}
296 	}
297 	free(pack_fds);
298 	return err;
299 }
300 
301 const struct got_error *
302 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
303     struct got_object *obj)
304 {
305 #ifndef GOT_NO_OBJ_CACHE
306 	const struct got_error *err = NULL;
307 	err = got_object_cache_add(&repo->objcache, id, obj);
308 	if (err) {
309 		if (err->code == GOT_ERR_OBJ_EXISTS ||
310 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
311 			err = NULL;
312 		return err;
313 	}
314 	obj->refcnt++;
315 #endif
316 	return NULL;
317 }
318 
319 struct got_object *
320 got_repo_get_cached_object(struct got_repository *repo,
321     struct got_object_id *id)
322 {
323 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
324 }
325 
326 const struct got_error *
327 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
328     struct got_tree_object *tree)
329 {
330 #ifndef GOT_NO_OBJ_CACHE
331 	const struct got_error *err = NULL;
332 	err = got_object_cache_add(&repo->treecache, id, tree);
333 	if (err) {
334 		if (err->code == GOT_ERR_OBJ_EXISTS ||
335 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
336 			err = NULL;
337 		return err;
338 	}
339 	tree->refcnt++;
340 #endif
341 	return NULL;
342 }
343 
344 struct got_tree_object *
345 got_repo_get_cached_tree(struct got_repository *repo,
346     struct got_object_id *id)
347 {
348 	return (struct got_tree_object *)got_object_cache_get(
349 	    &repo->treecache, id);
350 }
351 
352 const struct got_error *
353 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
354     struct got_commit_object *commit)
355 {
356 #ifndef GOT_NO_OBJ_CACHE
357 	const struct got_error *err = NULL;
358 	err = got_object_cache_add(&repo->commitcache, id, commit);
359 	if (err) {
360 		if (err->code == GOT_ERR_OBJ_EXISTS ||
361 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
362 			err = NULL;
363 		return err;
364 	}
365 	commit->refcnt++;
366 #endif
367 	return NULL;
368 }
369 
370 struct got_commit_object *
371 got_repo_get_cached_commit(struct got_repository *repo,
372     struct got_object_id *id)
373 {
374 	return (struct got_commit_object *)got_object_cache_get(
375 	    &repo->commitcache, id);
376 }
377 
378 const struct got_error *
379 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
380     struct got_tag_object *tag)
381 {
382 #ifndef GOT_NO_OBJ_CACHE
383 	const struct got_error *err = NULL;
384 	err = got_object_cache_add(&repo->tagcache, id, tag);
385 	if (err) {
386 		if (err->code == GOT_ERR_OBJ_EXISTS ||
387 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
388 			err = NULL;
389 		return err;
390 	}
391 	tag->refcnt++;
392 #endif
393 	return NULL;
394 }
395 
396 struct got_tag_object *
397 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
398 {
399 	return (struct got_tag_object *)got_object_cache_get(
400 	    &repo->tagcache, id);
401 }
402 
403 const struct got_error *
404 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
405     struct got_raw_object *raw)
406 {
407 #ifndef GOT_NO_OBJ_CACHE
408 	const struct got_error *err = NULL;
409 	err = got_object_cache_add(&repo->rawcache, id, raw);
410 	if (err) {
411 		if (err->code == GOT_ERR_OBJ_EXISTS ||
412 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
413 			err = NULL;
414 		return err;
415 	}
416 	raw->refcnt++;
417 #endif
418 	return NULL;
419 }
420 
421 
422 struct got_raw_object *
423 got_repo_get_cached_raw_object(struct got_repository *repo,
424     struct got_object_id *id)
425 {
426 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
427 }
428 
429 
430 static const struct got_error *
431 open_repo(struct got_repository *repo, const char *path)
432 {
433 	const struct got_error *err = NULL;
434 
435 	repo->gitdir_fd = -1;
436 
437 	/* bare git repository? */
438 	repo->path_git_dir = strdup(path);
439 	if (repo->path_git_dir == NULL)
440 		return got_error_from_errno("strdup");
441 	if (is_git_repo(repo)) {
442 		repo->path = strdup(repo->path_git_dir);
443 		if (repo->path == NULL) {
444 			err = got_error_from_errno("strdup");
445 			goto done;
446 		}
447 		repo->gitdir_fd = open(repo->path_git_dir,
448 		    O_DIRECTORY | O_CLOEXEC);
449 		if (repo->gitdir_fd == -1) {
450 			err = got_error_from_errno2("open",
451 			    repo->path_git_dir);
452 			goto done;
453 		}
454 		return NULL;
455 	}
456 
457 	/* git repository with working tree? */
458 	free(repo->path_git_dir);
459 	repo->path_git_dir = NULL;
460 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
461 		err = got_error_from_errno("asprintf");
462 		goto done;
463 	}
464 	if (is_git_repo(repo)) {
465 		repo->path = strdup(path);
466 		if (repo->path == NULL) {
467 			err = got_error_from_errno("strdup");
468 			goto done;
469 		}
470 		repo->gitdir_fd = open(repo->path_git_dir,
471 		    O_DIRECTORY | O_CLOEXEC);
472 		if (repo->gitdir_fd == -1) {
473 			err = got_error_from_errno2("open",
474 			    repo->path_git_dir);
475 			goto done;
476 		}
477 		return NULL;
478 	}
479 
480 	err = got_error(GOT_ERR_NOT_GIT_REPO);
481 done:
482 	if (err) {
483 		free(repo->path);
484 		repo->path = NULL;
485 		free(repo->path_git_dir);
486 		repo->path_git_dir = NULL;
487 		if (repo->gitdir_fd != -1)
488 			close(repo->gitdir_fd);
489 		repo->gitdir_fd = -1;
490 
491 	}
492 	return err;
493 }
494 
495 static const struct got_error *
496 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
497 {
498 	const struct got_error *err = NULL;
499 	char *repo_gitconfig_path = NULL;
500 
501 	if (global_gitconfig_path) {
502 		/* Read settings from ~/.gitconfig. */
503 		int dummy_repo_version;
504 		err = got_repo_read_gitconfig(&dummy_repo_version,
505 		    &repo->global_gitconfig_author_name,
506 		    &repo->global_gitconfig_author_email,
507 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
508 		if (err)
509 			return err;
510 	}
511 
512 	/* Read repository's .git/config file. */
513 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
514 	if (repo_gitconfig_path == NULL)
515 		return got_error_from_errno("got_repo_get_path_gitconfig");
516 
517 	err = got_repo_read_gitconfig(
518 	    &repo->gitconfig_repository_format_version,
519 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
520 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
521 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
522 	    repo_gitconfig_path);
523 	if (err)
524 		goto done;
525 
526 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
527 		int i;
528 
529 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
530 			got_repo_free_remote_repo_data(
531 			    &repo->gitconfig_remotes[i]);
532 		}
533 		free(repo->gitconfig_remotes);
534 		repo->gitconfig_remotes = NULL;
535 		repo->ngitconfig_remotes = 0;
536 
537 		free(repo->gitconfig_author_name);
538 		repo->gitconfig_author_name = NULL;
539 		free(repo->gitconfig_author_email);
540 		repo->gitconfig_author_email = NULL;
541 
542 		free(repo->global_gitconfig_author_name);
543 		repo->global_gitconfig_author_name = NULL;
544 		free(repo->global_gitconfig_author_email);
545 		repo->global_gitconfig_author_email = NULL;
546 	}
547 
548 done:
549 	free(repo_gitconfig_path);
550 	return err;
551 }
552 
553 static const struct got_error *
554 read_gotconfig(struct got_repository *repo)
555 {
556 	const struct got_error *err = NULL;
557 	char *gotconfig_path;
558 
559 	gotconfig_path = got_repo_get_path_gotconfig(repo);
560 	if (gotconfig_path == NULL)
561 		return got_error_from_errno("got_repo_get_path_gotconfig");
562 
563 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
564 	free(gotconfig_path);
565 	return err;
566 }
567 
568 /* Supported repository format extensions. */
569 static const char *const repo_extensions[] = {
570 	"noop",			/* Got supports repository format version 1. */
571 	"preciousObjects",	/* Supported by gotadmin cleanup. */
572 	"worktreeConfig",	/* Got does not care about Git work trees. */
573 };
574 
575 const struct got_error *
576 got_repo_open(struct got_repository **repop, const char *path,
577     const char *global_gitconfig_path, int *pack_fds)
578 {
579 	struct got_repository *repo = NULL;
580 	const struct got_error *err = NULL;
581 	char *repo_path = NULL;
582 	size_t i, j = 0;
583 	struct rlimit rl;
584 
585 	*repop = NULL;
586 
587 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
588 		return got_error_from_errno("getrlimit");
589 
590 	repo = calloc(1, sizeof(*repo));
591 	if (repo == NULL)
592 		return got_error_from_errno("calloc");
593 
594 	RB_INIT(&repo->packidx_bloom_filters);
595 	TAILQ_INIT(&repo->packidx_paths);
596 
597 	for (i = 0; i < nitems(repo->privsep_children); i++) {
598 		memset(&repo->privsep_children[i], 0,
599 		    sizeof(repo->privsep_children[0]));
600 		repo->privsep_children[i].imsg_fd = -1;
601 	}
602 
603 	err = got_object_cache_init(&repo->objcache,
604 	    GOT_OBJECT_CACHE_TYPE_OBJ);
605 	if (err)
606 		goto done;
607 	err = got_object_cache_init(&repo->treecache,
608 	    GOT_OBJECT_CACHE_TYPE_TREE);
609 	if (err)
610 		goto done;
611 	err = got_object_cache_init(&repo->commitcache,
612 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
613 	if (err)
614 		goto done;
615 	err = got_object_cache_init(&repo->tagcache,
616 	    GOT_OBJECT_CACHE_TYPE_TAG);
617 	if (err)
618 		goto done;
619 	err = got_object_cache_init(&repo->rawcache,
620 	    GOT_OBJECT_CACHE_TYPE_RAW);
621 	if (err)
622 		goto done;
623 
624 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
625 	if (repo->pack_cache_size > rl.rlim_cur / 8)
626 		repo->pack_cache_size = rl.rlim_cur / 8;
627 	for (i = 0; i < nitems(repo->packs); i++) {
628 		if (i < repo->pack_cache_size) {
629 			repo->packs[i].basefd = pack_fds[j++];
630 			repo->packs[i].accumfd = pack_fds[j++];
631 		} else {
632 			repo->packs[i].basefd = -1;
633 			repo->packs[i].accumfd = -1;
634 		}
635 	}
636 	repo->pinned_pack = -1;
637 	repo->pinned_packidx = -1;
638 	repo->pinned_pid = 0;
639 
640 	repo_path = realpath(path, NULL);
641 	if (repo_path == NULL) {
642 		err = got_error_from_errno2("realpath", path);
643 		goto done;
644 	}
645 
646 	for (;;) {
647 		char *parent_path;
648 
649 		err = open_repo(repo, repo_path);
650 		if (err == NULL)
651 			break;
652 		if (err->code != GOT_ERR_NOT_GIT_REPO)
653 			goto done;
654 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
655 			err = got_error(GOT_ERR_NOT_GIT_REPO);
656 			goto done;
657 		}
658 		err = got_path_dirname(&parent_path, repo_path);
659 		if (err)
660 			goto done;
661 		free(repo_path);
662 		repo_path = parent_path;
663 	}
664 
665 	err = read_gotconfig(repo);
666 	if (err)
667 		goto done;
668 
669 	err = read_gitconfig(repo, global_gitconfig_path);
670 	if (err)
671 		goto done;
672 	if (repo->gitconfig_repository_format_version != 0) {
673 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
674 		goto done;
675 	}
676 	for (i = 0; i < repo->nextensions; i++) {
677 		char *ext = repo->extensions[i];
678 		int j, supported = 0;
679 		for (j = 0; j < nitems(repo_extensions); j++) {
680 			if (strcmp(ext, repo_extensions[j]) == 0) {
681 				supported = 1;
682 				break;
683 			}
684 		}
685 		if (!supported) {
686 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
687 			goto done;
688 		}
689 	}
690 
691 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
692 done:
693 	if (err)
694 		got_repo_close(repo);
695 	else
696 		*repop = repo;
697 	free(repo_path);
698 	return err;
699 }
700 
701 const struct got_error *
702 got_repo_close(struct got_repository *repo)
703 {
704 	const struct got_error *err = NULL, *child_err;
705 	struct got_packidx_bloom_filter *bf;
706 	struct got_pathlist_entry *pe;
707 	size_t i;
708 
709 	for (i = 0; i < repo->pack_cache_size; i++) {
710 		if (repo->packidx_cache[i] == NULL)
711 			break;
712 		got_packidx_close(repo->packidx_cache[i]);
713 	}
714 
715 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
716 	    &repo->packidx_bloom_filters))) {
717 		RB_REMOVE(got_packidx_bloom_filter_tree,
718 		    &repo->packidx_bloom_filters, bf);
719 		bloom_free(bf->bloom);
720 		free(bf->bloom);
721 		free(bf);
722 	}
723 
724 	for (i = 0; i < repo->pack_cache_size; i++)
725 		if (repo->packs[i].path_packfile)
726 			if (repo->packs[i].path_packfile)
727 				got_pack_close(&repo->packs[i]);
728 
729 	free(repo->path);
730 	free(repo->path_git_dir);
731 
732 	got_object_cache_close(&repo->objcache);
733 	got_object_cache_close(&repo->treecache);
734 	got_object_cache_close(&repo->commitcache);
735 	got_object_cache_close(&repo->tagcache);
736 	got_object_cache_close(&repo->rawcache);
737 
738 	for (i = 0; i < nitems(repo->privsep_children); i++) {
739 		if (repo->privsep_children[i].imsg_fd == -1)
740 			continue;
741 		imsg_clear(repo->privsep_children[i].ibuf);
742 		free(repo->privsep_children[i].ibuf);
743 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
744 		child_err = got_privsep_wait_for_child(
745 		    repo->privsep_children[i].pid);
746 		if (child_err && err == NULL)
747 			err = child_err;
748 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
749 		    err == NULL)
750 			err = got_error_from_errno("close");
751 	}
752 
753 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
754 	    err == NULL)
755 		err = got_error_from_errno("close");
756 
757 	if (repo->gotconfig)
758 		got_gotconfig_free(repo->gotconfig);
759 	free(repo->gitconfig_author_name);
760 	free(repo->gitconfig_author_email);
761 	for (i = 0; i < repo->ngitconfig_remotes; i++)
762 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
763 	free(repo->gitconfig_remotes);
764 	for (i = 0; i < repo->nextensions; i++)
765 		free(repo->extensions[i]);
766 	free(repo->extensions);
767 
768 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
769 		free((void *)pe->path);
770 	got_pathlist_free(&repo->packidx_paths);
771 	free(repo);
772 
773 	return err;
774 }
775 
776 void
777 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
778 {
779 	int i;
780 
781 	free(repo->name);
782 	repo->name = NULL;
783 	free(repo->fetch_url);
784 	repo->fetch_url = NULL;
785 	free(repo->send_url);
786 	repo->send_url = NULL;
787 	for (i = 0; i < repo->nfetch_branches; i++)
788 		free(repo->fetch_branches[i]);
789 	free(repo->fetch_branches);
790 	repo->fetch_branches = NULL;
791 	repo->nfetch_branches = 0;
792 	for (i = 0; i < repo->nsend_branches; i++)
793 		free(repo->send_branches[i]);
794 	free(repo->send_branches);
795 	repo->send_branches = NULL;
796 	repo->nsend_branches = 0;
797 }
798 
799 const struct got_error *
800 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
801     const char *input_path)
802 {
803 	const struct got_error *err = NULL;
804 	const char *repo_abspath = NULL;
805 	size_t repolen, len;
806 	char *canonpath, *path = NULL;
807 
808 	*in_repo_path = NULL;
809 
810 	canonpath = strdup(input_path);
811 	if (canonpath == NULL) {
812 		err = got_error_from_errno("strdup");
813 		goto done;
814 	}
815 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
816 	if (err)
817 		goto done;
818 
819 	repo_abspath = got_repo_get_path(repo);
820 
821 	if (canonpath[0] == '\0') {
822 		path = strdup(canonpath);
823 		if (path == NULL) {
824 			err = got_error_from_errno("strdup");
825 			goto done;
826 		}
827 	} else {
828 		path = realpath(canonpath, NULL);
829 		if (path == NULL) {
830 			if (errno != ENOENT) {
831 				err = got_error_from_errno2("realpath",
832 				    canonpath);
833 				goto done;
834 			}
835 			/*
836 			 * Path is not on disk.
837 			 * Assume it is already relative to repository root.
838 			 */
839 			path = strdup(canonpath);
840 			if (path == NULL) {
841 				err = got_error_from_errno("strdup");
842 				goto done;
843 			}
844 		}
845 
846 		repolen = strlen(repo_abspath);
847 		len = strlen(path);
848 
849 
850 		if (strcmp(path, repo_abspath) == 0) {
851 			free(path);
852 			path = strdup("");
853 			if (path == NULL) {
854 				err = got_error_from_errno("strdup");
855 				goto done;
856 			}
857 		} else if (len > repolen &&
858 		    got_path_is_child(path, repo_abspath, repolen)) {
859 			/* Matched an on-disk path inside repository. */
860 			if (got_repo_is_bare(repo)) {
861 				/*
862 				 * Matched an on-disk path inside repository
863 				 * database. Treat input as repository-relative.
864 				 */
865 				free(path);
866 				path = canonpath;
867 				canonpath = NULL;
868 			} else {
869 				char *child;
870 				/* Strip common prefix with repository path. */
871 				err = got_path_skip_common_ancestor(&child,
872 				    repo_abspath, path);
873 				if (err)
874 					goto done;
875 				free(path);
876 				path = child;
877 			}
878 		} else {
879 			/*
880 			 * Matched unrelated on-disk path.
881 			 * Treat input as repository-relative.
882 			 */
883 			free(path);
884 			path = canonpath;
885 			canonpath = NULL;
886 		}
887 	}
888 
889 	/* Make in-repository path absolute */
890 	if (path[0] != '/') {
891 		char *abspath;
892 		if (asprintf(&abspath, "/%s", path) == -1) {
893 			err = got_error_from_errno("asprintf");
894 			goto done;
895 		}
896 		free(path);
897 		path = abspath;
898 	}
899 
900 done:
901 	free(canonpath);
902 	if (err)
903 		free(path);
904 	else
905 		*in_repo_path = path;
906 	return err;
907 }
908 
909 static const struct got_error *
910 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
911     const char *path_packidx)
912 {
913 	const struct got_error *err = NULL;
914 	size_t i;
915 
916 	for (i = 0; i < repo->pack_cache_size; i++) {
917 		if (repo->packidx_cache[i] == NULL)
918 			break;
919 		if (strcmp(repo->packidx_cache[i]->path_packidx,
920 		    path_packidx) == 0) {
921 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
922 		}
923 	}
924 	if (i == repo->pack_cache_size) {
925 		do {
926 			i--;
927 		} while (i > 0 && repo->pinned_packidx >= 0 &&
928 		    i == repo->pinned_packidx);
929 		err = got_packidx_close(repo->packidx_cache[i]);
930 		if (err)
931 			return err;
932 	}
933 
934 	repo->packidx_cache[i] = packidx;
935 
936 	return NULL;
937 }
938 
939 int
940 got_repo_is_packidx_filename(const char *name, size_t len)
941 {
942 	if (len != GOT_PACKIDX_NAMELEN)
943 		return 0;
944 
945 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
946 		return 0;
947 
948 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
949 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
950 		return 0;
951 
952 	return 1;
953 }
954 
955 static struct got_packidx_bloom_filter *
956 get_packidx_bloom_filter(struct got_repository *repo,
957     const char *path, size_t path_len)
958 {
959 	struct got_packidx_bloom_filter key;
960 
961 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
962 		return NULL; /* XXX */
963 	key.path_len = path_len;
964 
965 	return RB_FIND(got_packidx_bloom_filter_tree,
966 	    &repo->packidx_bloom_filters, &key);
967 }
968 
969 int
970 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
971     const char *path_packidx, struct got_object_id *id)
972 {
973 	struct got_packidx_bloom_filter *bf;
974 
975 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
976 	if (bf)
977 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
978 
979 	/* No bloom filter means this pack index must be searched. */
980 	return 1;
981 }
982 
983 static const struct got_error *
984 add_packidx_bloom_filter(struct got_repository *repo,
985     struct got_packidx *packidx, const char *path_packidx)
986 {
987 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
988 	struct got_packidx_bloom_filter *bf;
989 	size_t len;
990 
991 	/*
992 	 * Don't use bloom filters for very large pack index files.
993 	 * Large pack files will contain a relatively large fraction
994 	 * of our objects so we will likely need to visit them anyway.
995 	 * The more objects a pack file contains the higher the probability
996 	 * of a false-positive match from the bloom filter. And reading
997 	 * all object IDs from a large pack index file can be expensive.
998 	 */
999 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1000 		return NULL;
1001 
1002 	/* Do we already have a filter for this pack index? */
1003 	if (get_packidx_bloom_filter(repo, path_packidx,
1004 	    strlen(path_packidx)) != NULL)
1005 		return NULL;
1006 
1007 	bf = calloc(1, sizeof(*bf));
1008 	if (bf == NULL)
1009 		return got_error_from_errno("calloc");
1010 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1011 	if (bf->bloom == NULL) {
1012 		free(bf);
1013 		return got_error_from_errno("calloc");
1014 	}
1015 
1016 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1017 	if (len >= sizeof(bf->path)) {
1018 		free(bf->bloom);
1019 		free(bf);
1020 		return got_error(GOT_ERR_NO_SPACE);
1021 	}
1022 	bf->path_len = len;
1023 
1024 	/* Minimum size supported by our bloom filter is 1000 entries. */
1025 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1026 	for (i = 0; i < nobjects; i++) {
1027 		struct got_packidx_object_id *id;
1028 		id = &packidx->hdr.sorted_ids[i];
1029 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1030 	}
1031 
1032 	RB_INSERT(got_packidx_bloom_filter_tree,
1033 	    &repo->packidx_bloom_filters, bf);
1034 	return NULL;
1035 }
1036 
1037 const struct got_error *
1038 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1039     struct got_repository *repo, struct got_object_id *id)
1040 {
1041 	const struct got_error *err;
1042 	struct got_pathlist_entry *pe;
1043 	size_t i;
1044 
1045 	/* Search pack index cache. */
1046 	for (i = 0; i < repo->pack_cache_size; i++) {
1047 		if (repo->packidx_cache[i] == NULL)
1048 			break;
1049 		if (!got_repo_check_packidx_bloom_filter(repo,
1050 		    repo->packidx_cache[i]->path_packidx, id))
1051 			continue; /* object will not be found in this index */
1052 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1053 		if (*idx != -1) {
1054 			*packidx = repo->packidx_cache[i];
1055 			/*
1056 			 * Move this cache entry to the front. Repeatedly
1057 			 * searching a wrong pack index can be expensive.
1058 			 */
1059 			if (i > 0) {
1060 				memmove(&repo->packidx_cache[1],
1061 				    &repo->packidx_cache[0],
1062 				    i * sizeof(repo->packidx_cache[0]));
1063 				repo->packidx_cache[0] = *packidx;
1064 				if (repo->pinned_packidx >= 0 &&
1065 				    repo->pinned_packidx < i)
1066 					repo->pinned_packidx++;
1067 				else if (repo->pinned_packidx == i)
1068 					repo->pinned_packidx = 0;
1069 			}
1070 			return NULL;
1071 		}
1072 	}
1073 	/* No luck. Search the filesystem. */
1074 
1075 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1076 		const char *path_packidx = pe->path;
1077 		int is_cached = 0;
1078 
1079 		if (!got_repo_check_packidx_bloom_filter(repo,
1080 		    pe->path, id))
1081 			continue; /* object will not be found in this index */
1082 
1083 		for (i = 0; i < repo->pack_cache_size; i++) {
1084 			if (repo->packidx_cache[i] == NULL)
1085 				break;
1086 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1087 			    path_packidx) == 0) {
1088 				is_cached = 1;
1089 				break;
1090 			}
1091 		}
1092 		if (is_cached)
1093 			continue; /* already searched */
1094 
1095 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1096 		    path_packidx, 0);
1097 		if (err)
1098 			goto done;
1099 
1100 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1101 		if (err)
1102 			goto done;
1103 
1104 		err = cache_packidx(repo, *packidx, path_packidx);
1105 		if (err)
1106 			goto done;
1107 
1108 		*idx = got_packidx_get_object_idx(*packidx, id);
1109 		if (*idx != -1) {
1110 			err = NULL; /* found the object */
1111 			goto done;
1112 		}
1113 	}
1114 
1115 	err = got_error_no_obj(id);
1116 done:
1117 	return err;
1118 }
1119 
1120 const struct got_error *
1121 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1122     struct got_repository *repo)
1123 {
1124 	const struct got_error *err = NULL;
1125 	DIR *packdir = NULL;
1126 	struct dirent *dent;
1127 	char *path_packidx = NULL;
1128 	int packdir_fd;
1129 	struct stat sb;
1130 
1131 	packdir_fd = openat(got_repo_get_fd(repo),
1132 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1133 	if (packdir_fd == -1) {
1134 		return got_error_from_errno_fmt("openat: %s/%s",
1135 		    got_repo_get_path_git_dir(repo),
1136 		    GOT_OBJECTS_PACK_DIR);
1137 	}
1138 
1139 	packdir = fdopendir(packdir_fd);
1140 	if (packdir == NULL) {
1141 		err = got_error_from_errno("fdopendir");
1142 		goto done;
1143 	}
1144 
1145 	if (fstat(packdir_fd, &sb) == -1) {
1146 		err = got_error_from_errno("fstat");
1147 		goto done;
1148 	}
1149 	repo->pack_path_mtime = sb.st_mtime;
1150 
1151 	while ((dent = readdir(packdir)) != NULL) {
1152 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1153 			continue;
1154 
1155 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1156 		    dent->d_name) == -1) {
1157 			err = got_error_from_errno("asprintf");
1158 			path_packidx = NULL;
1159 			break;
1160 		}
1161 
1162 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1163 		if (err)
1164 			break;
1165 	}
1166 done:
1167 	if (err)
1168 		free(path_packidx);
1169 	if (packdir && closedir(packdir) != 0 && err == NULL)
1170 		err = got_error_from_errno("closedir");
1171 	return err;
1172 }
1173 
1174 const struct got_error *
1175 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1176     struct got_repository *repo)
1177 {
1178 	const struct got_error *err;
1179 	size_t i;
1180 
1181 	*packidx = NULL;
1182 
1183 	/* Search pack index cache. */
1184 	for (i = 0; i < repo->pack_cache_size; i++) {
1185 		if (repo->packidx_cache[i] == NULL)
1186 			break;
1187 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1188 		    path_packidx) == 0) {
1189 			*packidx = repo->packidx_cache[i];
1190 			return NULL;
1191 		}
1192 	}
1193 	/* No luck. Search the filesystem. */
1194 
1195 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1196 	    path_packidx, 0);
1197 	if (err)
1198 		return err;
1199 
1200 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1201 	if (err)
1202 		goto done;
1203 
1204 	err = cache_packidx(repo, *packidx, path_packidx);
1205 done:
1206 	if (err) {
1207 		got_packidx_close(*packidx);
1208 		*packidx = NULL;
1209 	}
1210 	return err;
1211 }
1212 
1213 static const struct got_error *
1214 read_packfile_hdr(int fd, struct got_packidx *packidx)
1215 {
1216 	const struct got_error *err = NULL;
1217 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1218 	struct got_packfile_hdr hdr;
1219 	ssize_t n;
1220 
1221 	n = read(fd, &hdr, sizeof(hdr));
1222 	if (n < 0)
1223 		return got_error_from_errno("read");
1224 	if (n != sizeof(hdr))
1225 		return got_error(GOT_ERR_BAD_PACKFILE);
1226 
1227 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1228 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1229 	    be32toh(hdr.nobjects) != totobj)
1230 		err = got_error(GOT_ERR_BAD_PACKFILE);
1231 
1232 	return err;
1233 }
1234 
1235 static const struct got_error *
1236 open_packfile(int *fd, struct got_repository *repo,
1237     const char *relpath, struct got_packidx *packidx)
1238 {
1239 	const struct got_error *err = NULL;
1240 
1241 	*fd = openat(got_repo_get_fd(repo), relpath,
1242 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1243 	if (*fd == -1)
1244 		return got_error_from_errno_fmt("openat: %s/%s",
1245 		    got_repo_get_path_git_dir(repo), relpath);
1246 
1247 	if (packidx) {
1248 		err = read_packfile_hdr(*fd, packidx);
1249 		if (err) {
1250 			close(*fd);
1251 			*fd = -1;
1252 		}
1253 	}
1254 
1255 	return err;
1256 }
1257 
1258 const struct got_error *
1259 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1260     const char *path_packfile, struct got_packidx *packidx)
1261 {
1262 	const struct got_error *err = NULL;
1263 	struct got_pack *pack = NULL;
1264 	struct stat sb;
1265 	size_t i;
1266 
1267 	if (packp)
1268 		*packp = NULL;
1269 
1270 	for (i = 0; i < repo->pack_cache_size; i++) {
1271 		pack = &repo->packs[i];
1272 		if (pack->path_packfile == NULL)
1273 			break;
1274 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1275 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1276 	}
1277 
1278 	if (i == repo->pack_cache_size) {
1279 		struct got_pack tmp;
1280 		do {
1281 			i--;
1282 		} while (i > 0 && repo->pinned_pack >= 0 &&
1283 		    i == repo->pinned_pack);
1284 		err = got_pack_close(&repo->packs[i]);
1285 		if (err)
1286 			return err;
1287 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1288 			return got_error_from_errno("ftruncate");
1289 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1290 			return got_error_from_errno("ftruncate");
1291 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1292 		memcpy(&repo->packs[i], &repo->packs[0],
1293 		    sizeof(repo->packs[i]));
1294 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1295 		if (repo->pinned_pack == 0)
1296 			repo->pinned_pack = i;
1297 		else if (repo->pinned_pack == i)
1298 			repo->pinned_pack = 0;
1299 		i = 0;
1300 	}
1301 
1302 	pack = &repo->packs[i];
1303 
1304 	pack->path_packfile = strdup(path_packfile);
1305 	if (pack->path_packfile == NULL) {
1306 		err = got_error_from_errno("strdup");
1307 		goto done;
1308 	}
1309 
1310 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1311 	if (err)
1312 		goto done;
1313 
1314 	if (fstat(pack->fd, &sb) != 0) {
1315 		err = got_error_from_errno("fstat");
1316 		goto done;
1317 	}
1318 	pack->filesize = sb.st_size;
1319 
1320 	pack->privsep_child = NULL;
1321 
1322 #ifndef GOT_PACK_NO_MMAP
1323 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1324 	    pack->fd, 0);
1325 	if (pack->map == MAP_FAILED) {
1326 		if (errno != ENOMEM) {
1327 			err = got_error_from_errno("mmap");
1328 			goto done;
1329 		}
1330 		pack->map = NULL; /* fall back to read(2) */
1331 	}
1332 #endif
1333 done:
1334 	if (err) {
1335 		if (pack) {
1336 			free(pack->path_packfile);
1337 			memset(pack, 0, sizeof(*pack));
1338 		}
1339 	} else if (packp)
1340 		*packp = pack;
1341 	return err;
1342 }
1343 
1344 struct got_pack *
1345 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1346 {
1347 	struct got_pack *pack = NULL;
1348 	size_t i;
1349 
1350 	for (i = 0; i < repo->pack_cache_size; i++) {
1351 		pack = &repo->packs[i];
1352 		if (pack->path_packfile == NULL)
1353 			break;
1354 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1355 			return pack;
1356 	}
1357 
1358 	return NULL;
1359 }
1360 
1361 const struct got_error *
1362 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1363     struct got_pack *pack)
1364 {
1365 	size_t i;
1366 	int pinned_pack = -1, pinned_packidx = -1;
1367 
1368 	for (i = 0; i < repo->pack_cache_size; i++) {
1369 		if (repo->packidx_cache[i] &&
1370 		    strcmp(repo->packidx_cache[i]->path_packidx,
1371 		    packidx->path_packidx) == 0)
1372 			pinned_packidx = i;
1373 		if (repo->packs[i].path_packfile &&
1374 		    strcmp(repo->packs[i].path_packfile,
1375 		    pack->path_packfile) == 0)
1376 			pinned_pack = i;
1377 	}
1378 
1379 	if (pinned_packidx == -1 || pinned_pack == -1)
1380 		return got_error(GOT_ERR_PIN_PACK);
1381 
1382 	repo->pinned_pack = pinned_pack;
1383 	repo->pinned_packidx = pinned_packidx;
1384 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1385 	return NULL;
1386 }
1387 
1388 struct got_pack *
1389 got_repo_get_pinned_pack(struct got_repository *repo)
1390 {
1391 	if (repo->pinned_pack >= 0 &&
1392 	    repo->pinned_pack < repo->pack_cache_size)
1393 		return &repo->packs[repo->pinned_pack];
1394 
1395 	return NULL;
1396 }
1397 
1398 void
1399 got_repo_unpin_pack(struct got_repository *repo)
1400 {
1401 	repo->pinned_packidx = -1;
1402 	repo->pinned_pack = -1;
1403 	repo->pinned_pid = 0;
1404 }
1405 
1406 const struct got_error *
1407 got_repo_init(const char *repo_path, const char *head_name)
1408 {
1409 	const struct got_error *err = NULL;
1410 	const char *dirnames[] = {
1411 		GOT_OBJECTS_DIR,
1412 		GOT_OBJECTS_PACK_DIR,
1413 		GOT_REFS_DIR,
1414 	};
1415 	const char *description_str = "Unnamed repository; "
1416 	    "edit this file 'description' to name the repository.";
1417 	const char *headref = "ref: refs/heads/";
1418 	const char *gitconfig_str = "[core]\n"
1419 	    "\trepositoryformatversion = 0\n"
1420 	    "\tfilemode = true\n"
1421 	    "\tbare = true\n";
1422 	char *headref_str, *path;
1423 	size_t i;
1424 
1425 	if (!got_path_dir_is_empty(repo_path))
1426 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1427 
1428 	for (i = 0; i < nitems(dirnames); i++) {
1429 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1430 			return got_error_from_errno("asprintf");
1431 		}
1432 		err = got_path_mkdir(path);
1433 		free(path);
1434 		if (err)
1435 			return err;
1436 	}
1437 
1438 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1439 		return got_error_from_errno("asprintf");
1440 	err = got_path_create_file(path, description_str);
1441 	free(path);
1442 	if (err)
1443 		return err;
1444 
1445 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1446 		return got_error_from_errno("asprintf");
1447 	if (asprintf(&headref_str, "%s%s", headref,
1448 	    head_name ? head_name : "main") == -1) {
1449 		free(path);
1450 		return got_error_from_errno("asprintf");
1451 	}
1452 	err = got_path_create_file(path, headref_str);
1453 	free(headref_str);
1454 	free(path);
1455 	if (err)
1456 		return err;
1457 
1458 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1459 		return got_error_from_errno("asprintf");
1460 	err = got_path_create_file(path, gitconfig_str);
1461 	free(path);
1462 	if (err)
1463 		return err;
1464 
1465 	return NULL;
1466 }
1467 
1468 static void
1469 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1470 {
1471 	struct got_pathlist_entry *pe;
1472 
1473 	while (!TAILQ_EMPTY(packidx_paths)) {
1474 		pe = TAILQ_FIRST(packidx_paths);
1475 		TAILQ_REMOVE(packidx_paths, pe, entry);
1476 		free((char *)pe->path);
1477 		free(pe);
1478 	}
1479 }
1480 
1481 static const struct got_error *
1482 match_packed_object(struct got_object_id **unique_id,
1483     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1484 {
1485 	const struct got_error *err = NULL;
1486 	char *objects_pack_dir = NULL;
1487 	struct got_object_id_queue matched_ids;
1488 	struct got_pathlist_entry *pe;
1489 	struct stat sb;
1490 
1491 	STAILQ_INIT(&matched_ids);
1492 
1493 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1494 	if (objects_pack_dir == NULL)
1495 		return got_error_from_errno("got_repo_get_path_objects_pack");
1496 
1497 	if (stat(objects_pack_dir, &sb) == -1) {
1498 		if (errno != ENOENT) {
1499 			err = got_error_from_errno2("stat", objects_pack_dir);
1500 			goto done;
1501 		}
1502 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1503 		purge_packidx_paths(&repo->packidx_paths);
1504 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1505 		if (err)
1506 			goto done;
1507 	}
1508 
1509 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1510 		const char *path_packidx = pe->path;
1511 		struct got_packidx *packidx;
1512 		struct got_object_qid *qid;
1513 
1514 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1515 		    path_packidx, 0);
1516 		if (err)
1517 			break;
1518 
1519 		err = got_packidx_match_id_str_prefix(&matched_ids,
1520 		    packidx, id_str_prefix);
1521 		if (err) {
1522 			got_packidx_close(packidx);
1523 			break;
1524 		}
1525 		err = got_packidx_close(packidx);
1526 		if (err)
1527 			break;
1528 
1529 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1530 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1531 				int matched_type;
1532 				err = got_object_get_type(&matched_type, repo,
1533 				    &qid->id);
1534 				if (err)
1535 					goto done;
1536 				if (matched_type != obj_type)
1537 					continue;
1538 			}
1539 			if (*unique_id == NULL) {
1540 				*unique_id = got_object_id_dup(&qid->id);
1541 				if (*unique_id == NULL) {
1542 					err = got_error_from_errno("malloc");
1543 					goto done;
1544 				}
1545 			} else {
1546 				if (got_object_id_cmp(*unique_id,
1547 				    &qid->id) == 0)
1548 					continue; /* packed multiple times */
1549 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1550 				goto done;
1551 			}
1552 		}
1553 	}
1554 done:
1555 	free(objects_pack_dir);
1556 	got_object_id_queue_free(&matched_ids);
1557 	if (err) {
1558 		free(*unique_id);
1559 		*unique_id = NULL;
1560 	}
1561 	return err;
1562 }
1563 
1564 static const struct got_error *
1565 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1566     const char *object_dir, const char *id_str_prefix, int obj_type,
1567     struct got_repository *repo)
1568 {
1569 	const struct got_error *err = NULL;
1570 	char *path, *id_str = NULL;
1571 	DIR *dir = NULL;
1572 	struct dirent *dent;
1573 	struct got_object_id id;
1574 
1575 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1576 		err = got_error_from_errno("asprintf");
1577 		goto done;
1578 	}
1579 
1580 	dir = opendir(path);
1581 	if (dir == NULL) {
1582 		if (errno == ENOENT) {
1583 			err = NULL;
1584 			goto done;
1585 		}
1586 		err = got_error_from_errno2("opendir", path);
1587 		goto done;
1588 	}
1589 	while ((dent = readdir(dir)) != NULL) {
1590 		int cmp;
1591 
1592 		free(id_str);
1593 		id_str = NULL;
1594 
1595 		if (strcmp(dent->d_name, ".") == 0 ||
1596 		    strcmp(dent->d_name, "..") == 0)
1597 			continue;
1598 
1599 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1600 			err = got_error_from_errno("asprintf");
1601 			goto done;
1602 		}
1603 
1604 		if (!got_parse_sha1_digest(id.sha1, id_str))
1605 			continue;
1606 
1607 		/*
1608 		 * Directory entries do not necessarily appear in
1609 		 * sorted order, so we must iterate over all of them.
1610 		 */
1611 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1612 		if (cmp != 0)
1613 			continue;
1614 
1615 		if (*unique_id == NULL) {
1616 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1617 				int matched_type;
1618 				err = got_object_get_type(&matched_type, repo,
1619 				    &id);
1620 				if (err)
1621 					goto done;
1622 				if (matched_type != obj_type)
1623 					continue;
1624 			}
1625 			*unique_id = got_object_id_dup(&id);
1626 			if (*unique_id == NULL) {
1627 				err = got_error_from_errno("got_object_id_dup");
1628 				goto done;
1629 			}
1630 		} else {
1631 			if (got_object_id_cmp(*unique_id, &id) == 0)
1632 				continue; /* both packed and loose */
1633 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1634 			goto done;
1635 		}
1636 	}
1637 done:
1638 	if (dir && closedir(dir) != 0 && err == NULL)
1639 		err = got_error_from_errno("closedir");
1640 	if (err) {
1641 		free(*unique_id);
1642 		*unique_id = NULL;
1643 	}
1644 	free(id_str);
1645 	free(path);
1646 	return err;
1647 }
1648 
1649 const struct got_error *
1650 got_repo_match_object_id_prefix(struct got_object_id **id,
1651     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1652 {
1653 	const struct got_error *err = NULL;
1654 	char *path_objects = NULL, *object_dir = NULL;
1655 	size_t len;
1656 	int i;
1657 
1658 	*id = NULL;
1659 
1660 	path_objects = got_repo_get_path_objects(repo);
1661 
1662 	len = strlen(id_str_prefix);
1663 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1664 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1665 		goto done;
1666 	}
1667 
1668 	for (i = 0; i < len; i++) {
1669 		if (isxdigit((unsigned char)id_str_prefix[i]))
1670 			continue;
1671 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1672 		goto done;
1673 	}
1674 
1675 	if (len >= 2) {
1676 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1677 		if (err)
1678 			goto done;
1679 		object_dir = strndup(id_str_prefix, 2);
1680 		if (object_dir == NULL) {
1681 			err = got_error_from_errno("strdup");
1682 			goto done;
1683 		}
1684 		err = match_loose_object(id, path_objects, object_dir,
1685 		    id_str_prefix, obj_type, repo);
1686 	} else if (len == 1) {
1687 		int i;
1688 		for (i = 0; i < 0xf; i++) {
1689 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1690 			    == -1) {
1691 				err = got_error_from_errno("asprintf");
1692 				goto done;
1693 			}
1694 			err = match_packed_object(id, repo, object_dir,
1695 			    obj_type);
1696 			if (err)
1697 				goto done;
1698 			err = match_loose_object(id, path_objects, object_dir,
1699 			    id_str_prefix, obj_type, repo);
1700 			if (err)
1701 				goto done;
1702 		}
1703 	} else {
1704 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1705 		goto done;
1706 	}
1707 done:
1708 	free(path_objects);
1709 	free(object_dir);
1710 	if (err) {
1711 		free(*id);
1712 		*id = NULL;
1713 	} else if (*id == NULL) {
1714 		switch (obj_type) {
1715 		case GOT_OBJ_TYPE_BLOB:
1716 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1717 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1718 			break;
1719 		case GOT_OBJ_TYPE_TREE:
1720 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1721 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1722 			break;
1723 		case GOT_OBJ_TYPE_COMMIT:
1724 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1725 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1726 			break;
1727 		case GOT_OBJ_TYPE_TAG:
1728 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1729 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1730 			break;
1731 		default:
1732 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1733 			break;
1734 		}
1735 	}
1736 
1737 	return err;
1738 }
1739 
1740 const struct got_error *
1741 got_repo_match_object_id(struct got_object_id **id, char **label,
1742     const char *id_str, int obj_type, struct got_reflist_head *refs,
1743     struct got_repository *repo)
1744 {
1745 	const struct got_error *err;
1746 	struct got_tag_object *tag;
1747 	struct got_reference *ref = NULL;
1748 
1749 	*id = NULL;
1750 	if (label)
1751 		*label = NULL;
1752 
1753 	if (refs) {
1754 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1755 		    refs, repo);
1756 		if (err == NULL) {
1757 			*id = got_object_id_dup(
1758 			    got_object_tag_get_object_id(tag));
1759 			if (*id == NULL)
1760 				err = got_error_from_errno("got_object_id_dup");
1761 			else if (label && asprintf(label, "refs/tags/%s",
1762 			    got_object_tag_get_name(tag)) == -1) {
1763 				err = got_error_from_errno("asprintf");
1764 				free(*id);
1765 				*id = NULL;
1766 			}
1767 			got_object_tag_close(tag);
1768 			return err;
1769 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1770 		    err->code != GOT_ERR_NO_OBJ)
1771 			return err;
1772 	}
1773 
1774 	err = got_ref_open(&ref, repo, id_str, 0);
1775 	if (err == NULL) {
1776 		err = got_ref_resolve(id, repo, ref);
1777 		if (err)
1778 			goto done;
1779 		if (label) {
1780 			*label = strdup(got_ref_get_name(ref));
1781 			if (*label == NULL) {
1782 				err = got_error_from_errno("strdup");
1783 				goto done;
1784 			}
1785 		}
1786 	} else {
1787 		if (err->code != GOT_ERR_NOT_REF &&
1788 		    err->code != GOT_ERR_BAD_REF_NAME)
1789 			goto done;
1790 		err = got_repo_match_object_id_prefix(id, id_str,
1791 		    obj_type, repo);
1792 		if (err) {
1793 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1794 				err = got_error_not_ref(id_str);
1795 			goto done;
1796 		}
1797 		if (label) {
1798 			err = got_object_id_str(label, *id);
1799 			if (*label == NULL) {
1800 				err = got_error_from_errno("strdup");
1801 				goto done;
1802 			}
1803 		}
1804 	}
1805 done:
1806 	if (ref)
1807 		got_ref_close(ref);
1808 	return err;
1809 }
1810 
1811 const struct got_error *
1812 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1813     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1814 {
1815 	const struct got_error *err = NULL;
1816 	struct got_reflist_entry *re;
1817 	struct got_object_id *tag_id;
1818 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1819 
1820 	*tag = NULL;
1821 
1822 	TAILQ_FOREACH(re, refs, entry) {
1823 		const char *refname;
1824 		refname = got_ref_get_name(re->ref);
1825 		if (got_ref_is_symbolic(re->ref))
1826 			continue;
1827 		if (strncmp(refname, "refs/tags/", 10) != 0)
1828 			continue;
1829 		if (!name_is_absolute)
1830 			refname += strlen("refs/tags/");
1831 		if (strcmp(refname, name) != 0)
1832 			continue;
1833 		err = got_ref_resolve(&tag_id, repo, re->ref);
1834 		if (err)
1835 			break;
1836 		err = got_object_open_as_tag(tag, repo, tag_id);
1837 		free(tag_id);
1838 		if (err)
1839 			break;
1840 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1841 		    got_object_tag_get_object_type(*tag) == obj_type)
1842 			break;
1843 		got_object_tag_close(*tag);
1844 		*tag = NULL;
1845 	}
1846 
1847 	if (err == NULL && *tag == NULL)
1848 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1849 		    GOT_OBJ_LABEL_TAG, name);
1850 	return err;
1851 }
1852 
1853 static const struct got_error *
1854 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1855     const char *name, mode_t mode, struct got_object_id *blob_id)
1856 {
1857 	const struct got_error *err = NULL;
1858 
1859 	 *new_te = NULL;
1860 
1861 	*new_te = calloc(1, sizeof(**new_te));
1862 	if (*new_te == NULL)
1863 		return got_error_from_errno("calloc");
1864 
1865 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1866 	    sizeof((*new_te)->name)) {
1867 		err = got_error(GOT_ERR_NO_SPACE);
1868 		goto done;
1869 	}
1870 
1871 	if (S_ISLNK(mode)) {
1872 		(*new_te)->mode = S_IFLNK;
1873 	} else {
1874 		(*new_te)->mode = S_IFREG;
1875 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1876 	}
1877 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1878 done:
1879 	if (err && *new_te) {
1880 		free(*new_te);
1881 		*new_te = NULL;
1882 	}
1883 	return err;
1884 }
1885 
1886 static const struct got_error *
1887 import_file(struct got_tree_entry **new_te, struct dirent *de,
1888     const char *path, struct got_repository *repo)
1889 {
1890 	const struct got_error *err;
1891 	struct got_object_id *blob_id = NULL;
1892 	char *filepath;
1893 	struct stat sb;
1894 
1895 	if (asprintf(&filepath, "%s%s%s", path,
1896 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1897 		return got_error_from_errno("asprintf");
1898 
1899 	if (lstat(filepath, &sb) != 0) {
1900 		err = got_error_from_errno2("lstat", path);
1901 		goto done;
1902 	}
1903 
1904 	err = got_object_blob_create(&blob_id, filepath, repo);
1905 	if (err)
1906 		goto done;
1907 
1908 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1909 	    blob_id);
1910 done:
1911 	free(filepath);
1912 	if (err)
1913 		free(blob_id);
1914 	return err;
1915 }
1916 
1917 static const struct got_error *
1918 insert_tree_entry(struct got_tree_entry *new_te,
1919     struct got_pathlist_head *paths)
1920 {
1921 	const struct got_error *err = NULL;
1922 	struct got_pathlist_entry *new_pe;
1923 
1924 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1925 	if (err)
1926 		return err;
1927 	if (new_pe == NULL)
1928 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1929 	return NULL;
1930 }
1931 
1932 static const struct got_error *write_tree(struct got_object_id **,
1933     const char *, struct got_pathlist_head *, struct got_repository *,
1934     got_repo_import_cb progress_cb, void *progress_arg);
1935 
1936 static const struct got_error *
1937 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1938     const char *path, struct got_pathlist_head *ignores,
1939     struct got_repository *repo,
1940     got_repo_import_cb progress_cb, void *progress_arg)
1941 {
1942 	const struct got_error *err;
1943 	struct got_object_id *id = NULL;
1944 	char *subdirpath;
1945 
1946 	if (asprintf(&subdirpath, "%s%s%s", path,
1947 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1948 		return got_error_from_errno("asprintf");
1949 
1950 	(*new_te) = calloc(1, sizeof(**new_te));
1951 	if (*new_te == NULL)
1952 		return got_error_from_errno("calloc");
1953 	(*new_te)->mode = S_IFDIR;
1954 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1955 	    sizeof((*new_te)->name)) {
1956 		err = got_error(GOT_ERR_NO_SPACE);
1957 		goto done;
1958 	}
1959 	err = write_tree(&id, subdirpath, ignores,  repo,
1960 	    progress_cb, progress_arg);
1961 	if (err)
1962 		goto done;
1963 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1964 
1965 done:
1966 	free(id);
1967 	free(subdirpath);
1968 	if (err) {
1969 		free(*new_te);
1970 		*new_te = NULL;
1971 	}
1972 	return err;
1973 }
1974 
1975 static const struct got_error *
1976 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1977     struct got_pathlist_head *ignores, struct got_repository *repo,
1978     got_repo_import_cb progress_cb, void *progress_arg)
1979 {
1980 	const struct got_error *err = NULL;
1981 	DIR *dir;
1982 	struct dirent *de;
1983 	int nentries;
1984 	struct got_tree_entry *new_te = NULL;
1985 	struct got_pathlist_head paths;
1986 	struct got_pathlist_entry *pe;
1987 
1988 	*new_tree_id = NULL;
1989 
1990 	TAILQ_INIT(&paths);
1991 
1992 	dir = opendir(path_dir);
1993 	if (dir == NULL) {
1994 		err = got_error_from_errno2("opendir", path_dir);
1995 		goto done;
1996 	}
1997 
1998 	nentries = 0;
1999 	while ((de = readdir(dir)) != NULL) {
2000 		int ignore = 0;
2001 		int type;
2002 
2003 		if (strcmp(de->d_name, ".") == 0 ||
2004 		    strcmp(de->d_name, "..") == 0)
2005 			continue;
2006 
2007 		TAILQ_FOREACH(pe, ignores, entry) {
2008 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2009 				ignore = 1;
2010 				break;
2011 			}
2012 		}
2013 		if (ignore)
2014 			continue;
2015 
2016 		err = got_path_dirent_type(&type, path_dir, de);
2017 		if (err)
2018 			goto done;
2019 
2020 		if (type == DT_DIR) {
2021 			err = import_subdir(&new_te, de, path_dir,
2022 			    ignores, repo, progress_cb, progress_arg);
2023 			if (err) {
2024 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2025 					goto done;
2026 				err = NULL;
2027 				continue;
2028 			}
2029 		} else if (type == DT_REG || type == DT_LNK) {
2030 			err = import_file(&new_te, de, path_dir, repo);
2031 			if (err)
2032 				goto done;
2033 		} else
2034 			continue;
2035 
2036 		err = insert_tree_entry(new_te, &paths);
2037 		if (err)
2038 			goto done;
2039 		nentries++;
2040 	}
2041 
2042 	if (TAILQ_EMPTY(&paths)) {
2043 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2044 		    "cannot create tree without any entries");
2045 		goto done;
2046 	}
2047 
2048 	TAILQ_FOREACH(pe, &paths, entry) {
2049 		struct got_tree_entry *te = pe->data;
2050 		char *path;
2051 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2052 			continue;
2053 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2054 			err = got_error_from_errno("asprintf");
2055 			goto done;
2056 		}
2057 		err = (*progress_cb)(progress_arg, path);
2058 		free(path);
2059 		if (err)
2060 			goto done;
2061 	}
2062 
2063 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2064 done:
2065 	if (dir)
2066 		closedir(dir);
2067 	got_pathlist_free(&paths);
2068 	return err;
2069 }
2070 
2071 const struct got_error *
2072 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2073     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2074     struct got_repository *repo, got_repo_import_cb progress_cb,
2075     void *progress_arg)
2076 {
2077 	const struct got_error *err;
2078 	struct got_object_id *new_tree_id;
2079 
2080 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2081 	    progress_cb, progress_arg);
2082 	if (err)
2083 		return err;
2084 
2085 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2086 	    author, time(NULL), author, time(NULL), logmsg, repo);
2087 	free(new_tree_id);
2088 	return err;
2089 }
2090 
2091 const struct got_error *
2092 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2093     struct got_repository *repo)
2094 {
2095 	const struct got_error *err = NULL;
2096 	char *path_objects = NULL, *path = NULL;
2097 	DIR *dir = NULL;
2098 	struct got_object_id id;
2099 	int i;
2100 
2101 	*nobjects = 0;
2102 	*ondisk_size = 0;
2103 
2104 	path_objects = got_repo_get_path_objects(repo);
2105 	if (path_objects == NULL)
2106 		return got_error_from_errno("got_repo_get_path_objects");
2107 
2108 	for (i = 0; i <= 0xff; i++) {
2109 		struct dirent *dent;
2110 
2111 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2112 			err = got_error_from_errno("asprintf");
2113 			break;
2114 		}
2115 
2116 		dir = opendir(path);
2117 		if (dir == NULL) {
2118 			if (errno == ENOENT) {
2119 				err = NULL;
2120 				continue;
2121 			}
2122 			err = got_error_from_errno2("opendir", path);
2123 			break;
2124 		}
2125 
2126 		while ((dent = readdir(dir)) != NULL) {
2127 			char *id_str;
2128 			int fd;
2129 			struct stat sb;
2130 
2131 			if (strcmp(dent->d_name, ".") == 0 ||
2132 			    strcmp(dent->d_name, "..") == 0)
2133 				continue;
2134 
2135 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2136 				err = got_error_from_errno("asprintf");
2137 				goto done;
2138 			}
2139 
2140 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2141 				free(id_str);
2142 				continue;
2143 			}
2144 			free(id_str);
2145 
2146 			err = got_object_open_loose_fd(&fd, &id, repo);
2147 			if (err)
2148 				goto done;
2149 
2150 			if (fstat(fd, &sb) == -1) {
2151 				err = got_error_from_errno("fstat");
2152 				close(fd);
2153 				goto done;
2154 			}
2155 			(*nobjects)++;
2156 			(*ondisk_size) += sb.st_size;
2157 
2158 			if (close(fd) == -1) {
2159 				err = got_error_from_errno("close");
2160 				goto done;
2161 			}
2162 		}
2163 
2164 		if (closedir(dir) != 0) {
2165 			err = got_error_from_errno("closedir");
2166 			goto done;
2167 		}
2168 		dir = NULL;
2169 
2170 		free(path);
2171 		path = NULL;
2172 	}
2173 done:
2174 	if (dir && closedir(dir) != 0 && err == NULL)
2175 		err = got_error_from_errno("closedir");
2176 
2177 	if (err) {
2178 		*nobjects = 0;
2179 		*ondisk_size = 0;
2180 	}
2181 	free(path_objects);
2182 	free(path);
2183 	return err;
2184 }
2185 
2186 const struct got_error *
2187 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2188     off_t *total_packsize, struct got_repository *repo)
2189 {
2190 	const struct got_error *err = NULL;
2191 	DIR *packdir = NULL;
2192 	struct dirent *dent;
2193 	struct got_packidx *packidx = NULL;
2194 	char *path_packidx;
2195 	char *path_packfile;
2196 	int packdir_fd;
2197 	struct stat sb;
2198 
2199 	*npackfiles = 0;
2200 	*nobjects = 0;
2201 	*total_packsize = 0;
2202 
2203 	packdir_fd = openat(got_repo_get_fd(repo),
2204 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2205 	if (packdir_fd == -1) {
2206 		return got_error_from_errno_fmt("openat: %s/%s",
2207 		    got_repo_get_path_git_dir(repo),
2208 		    GOT_OBJECTS_PACK_DIR);
2209 	}
2210 
2211 	packdir = fdopendir(packdir_fd);
2212 	if (packdir == NULL) {
2213 		err = got_error_from_errno("fdopendir");
2214 		goto done;
2215 	}
2216 
2217 	while ((dent = readdir(packdir)) != NULL) {
2218 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2219 			continue;
2220 
2221 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2222 		    dent->d_name) == -1) {
2223 			err = got_error_from_errno("asprintf");
2224 			goto done;
2225 		}
2226 
2227 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2228 		    path_packidx, 0);
2229 		free(path_packidx);
2230 		if (err)
2231 			goto done;
2232 
2233 		if (fstat(packidx->fd, &sb) == -1)
2234 			goto done;
2235 		*total_packsize += sb.st_size;
2236 
2237 		err = got_packidx_get_packfile_path(&path_packfile,
2238 		    packidx->path_packidx);
2239 		if (err)
2240 			goto done;
2241 
2242 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2243 		    0) == -1) {
2244 			free(path_packfile);
2245 			goto done;
2246 		}
2247 		free(path_packfile);
2248 		*total_packsize += sb.st_size;
2249 
2250 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2251 
2252 		(*npackfiles)++;
2253 
2254 		got_packidx_close(packidx);
2255 		packidx = NULL;
2256 	}
2257 done:
2258 	if (packidx)
2259 		got_packidx_close(packidx);
2260 	if (packdir && closedir(packdir) != 0 && err == NULL)
2261 		err = got_error_from_errno("closedir");
2262 	if (err) {
2263 		*npackfiles = 0;
2264 		*nobjects = 0;
2265 		*total_packsize = 0;
2266 	}
2267 	return err;
2268 }
2269 
2270 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2271     got_packidx_bloom_filter_cmp);