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_delta_cache.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 #include "got_lib_gotconfig.h"
62 
63 #ifndef nitems
64 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
65 #endif
66 
67 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
68 
69 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
70     got_packidx_bloom_filter_cmp);
71 
72 const char *
73 got_repo_get_path(struct got_repository *repo)
74 {
75 	return repo->path;
76 }
77 
78 const char *
79 got_repo_get_path_git_dir(struct got_repository *repo)
80 {
81 	return repo->path_git_dir;
82 }
83 
84 int
85 got_repo_get_fd(struct got_repository *repo)
86 {
87 	return repo->gitdir_fd;
88 }
89 
90 const char *
91 got_repo_get_gitconfig_author_name(struct got_repository *repo)
92 {
93 	return repo->gitconfig_author_name;
94 }
95 
96 const char *
97 got_repo_get_gitconfig_author_email(struct got_repository *repo)
98 {
99 	return repo->gitconfig_author_email;
100 }
101 
102 const char *
103 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
104 {
105 	return repo->global_gitconfig_author_name;
106 }
107 
108 const char *
109 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
110 {
111 	return repo->global_gitconfig_author_email;
112 }
113 
114 const char *
115 got_repo_get_gitconfig_owner(struct got_repository *repo)
116 {
117 	return repo->gitconfig_owner;
118 }
119 
120 void
121 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
122     struct got_repository *repo)
123 {
124 	*extensions = repo->extensions;
125 	*nextensions = repo->nextensions;
126 }
127 
128 int
129 got_repo_is_bare(struct got_repository *repo)
130 {
131 	return (strcmp(repo->path, repo->path_git_dir) == 0);
132 }
133 
134 static char *
135 get_path_git_child(struct got_repository *repo, const char *basename)
136 {
137 	char *path_child;
138 
139 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
140 	    basename) == -1)
141 		return NULL;
142 
143 	return path_child;
144 }
145 
146 char *
147 got_repo_get_path_objects(struct got_repository *repo)
148 {
149 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
150 }
151 
152 char *
153 got_repo_get_path_objects_pack(struct got_repository *repo)
154 {
155 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
156 }
157 
158 char *
159 got_repo_get_path_refs(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_REFS_DIR);
162 }
163 
164 char *
165 got_repo_get_path_packed_refs(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
168 }
169 
170 static char *
171 get_path_head(struct got_repository *repo)
172 {
173 	return get_path_git_child(repo, GOT_HEAD_FILE);
174 }
175 
176 char *
177 got_repo_get_path_gitconfig(struct got_repository *repo)
178 {
179 	return get_path_git_child(repo, GOT_GITCONFIG);
180 }
181 
182 char *
183 got_repo_get_path_gotconfig(struct got_repository *repo)
184 {
185 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
186 }
187 
188 const struct got_gotconfig *
189 got_repo_get_gotconfig(struct got_repository *repo)
190 {
191 	return repo->gotconfig;
192 }
193 
194 void
195 got_repo_get_gitconfig_remotes(int *nremotes,
196     const struct got_remote_repo **remotes, struct got_repository *repo)
197 {
198 	*nremotes = repo->ngitconfig_remotes;
199 	*remotes = repo->gitconfig_remotes;
200 }
201 
202 static int
203 is_git_repo(struct got_repository *repo)
204 {
205 	const char *path_git = got_repo_get_path_git_dir(repo);
206 	char *path_objects = got_repo_get_path_objects(repo);
207 	char *path_refs = got_repo_get_path_refs(repo);
208 	char *path_head = get_path_head(repo);
209 	int ret = 0;
210 	struct stat sb;
211 	struct got_reference *head_ref;
212 
213 	if (lstat(path_git, &sb) == -1)
214 		goto done;
215 	if (!S_ISDIR(sb.st_mode))
216 		goto done;
217 
218 	if (lstat(path_objects, &sb) == -1)
219 		goto done;
220 	if (!S_ISDIR(sb.st_mode))
221 		goto done;
222 
223 	if (lstat(path_refs, &sb) == -1)
224 		goto done;
225 	if (!S_ISDIR(sb.st_mode))
226 		goto done;
227 
228 	if (lstat(path_head, &sb) == -1)
229 		goto done;
230 	if (!S_ISREG(sb.st_mode))
231 		goto done;
232 
233 	/* Check if the HEAD reference can be opened. */
234 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
235 		goto done;
236 	got_ref_close(head_ref);
237 
238 	ret = 1;
239 done:
240 	free(path_objects);
241 	free(path_refs);
242 	free(path_head);
243 	return ret;
244 
245 }
246 
247 static const struct got_error *
248 close_tempfiles(int *fds, size_t nfds)
249 {
250 	const struct got_error *err = NULL;
251 	int i;
252 
253 	for (i = 0; i < nfds; i++) {
254 		if (fds[i] == -1)
255 			continue;
256 		if (close(fds[i]) == -1) {
257 			err = got_error_from_errno("close");
258 			break;
259 		}
260 	}
261 	free(fds);
262 	return err;
263 }
264 
265 static const struct got_error *
266 open_tempfiles(int **fds, size_t array_size, size_t nfds)
267 {
268 	const struct got_error *err = NULL;
269 	int i;
270 
271 	*fds = calloc(array_size, sizeof(**fds));
272 	if (*fds == NULL)
273 		return got_error_from_errno("calloc");
274 
275 	for (i = 0; i < array_size; i++)
276 		(*fds)[i] = -1;
277 
278 	for (i = 0; i < nfds; i++) {
279 		(*fds)[i] = got_opentempfd();
280 		if ((*fds)[i] == -1) {
281 			err = got_error_from_errno("got_opentempfd");
282 			close_tempfiles(*fds, nfds);
283 			*fds = NULL;
284 			return err;
285 		}
286 	}
287 
288 	return NULL;
289 }
290 
291 static const struct got_error *
292 get_pack_cache_size(int *pack_cache_size)
293 {
294 	struct rlimit rl;
295 
296 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
297 		return got_error_from_errno("getrlimit");
298 
299 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
300 	if (*pack_cache_size > rl.rlim_cur / 8)
301 		*pack_cache_size = rl.rlim_cur / 8;
302 
303 	return NULL;
304 }
305 
306 const struct got_error *
307 got_repo_pack_fds_open(int **pack_fds)
308 {
309 	const struct got_error *err;
310 	int nfds;
311 
312 	err = get_pack_cache_size(&nfds);
313 	if (err)
314 		return err;
315 
316 	/*
317 	 * We need one basefd and one accumfd per cached pack.
318 	 * Our constants should be set up in a way such that
319 	 * this error never triggers.
320 	 */
321 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
322 		return got_error(GOT_ERR_NO_SPACE);
323 
324 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
325 }
326 
327 const struct got_error *
328 got_repo_pack_fds_close(int *pack_fds)
329 {
330 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
331 }
332 
333 const struct got_error *
334 got_repo_temp_fds_open(int **temp_fds)
335 {
336 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
337 	    GOT_REPO_NUM_TEMPFILES);
338 }
339 
340 void
341 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
342 {
343 	int i;
344 
345 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
346 		repo->tempfiles[i] = temp_fds[i];
347 }
348 
349 const struct got_error *
350 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
351 {
352 	int i;
353 
354 	*fd = -1;
355 	*idx = -1;
356 
357 	for (i = 0; i < nitems(repo->tempfiles); i++) {
358 		if (repo->tempfile_use_mask & (1 << i))
359 			continue;
360 		if (repo->tempfiles[i] != -1) {
361 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
362 				return got_error_from_errno("ftruncate");
363 			*fd = repo->tempfiles[i];
364 			*idx = i;
365 			repo->tempfile_use_mask |= (1 << i);
366 			return NULL;
367 		}
368 	}
369 
370 	return got_error(GOT_ERR_REPO_TEMPFILE);
371 }
372 
373 void
374 got_repo_temp_fds_put(int idx, struct got_repository *repo)
375 {
376 	repo->tempfile_use_mask &= ~(1 << idx);
377 }
378 
379 const struct got_error *
380 got_repo_temp_fds_close(int *temp_fds)
381 {
382 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
383 }
384 
385 const struct got_error *
386 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
387     struct got_object *obj)
388 {
389 #ifndef GOT_NO_OBJ_CACHE
390 	const struct got_error *err = NULL;
391 	err = got_object_cache_add(&repo->objcache, id, obj);
392 	if (err) {
393 		if (err->code == GOT_ERR_OBJ_EXISTS ||
394 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
395 			err = NULL;
396 		return err;
397 	}
398 	obj->refcnt++;
399 #endif
400 	return NULL;
401 }
402 
403 struct got_object *
404 got_repo_get_cached_object(struct got_repository *repo,
405     struct got_object_id *id)
406 {
407 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
408 }
409 
410 const struct got_error *
411 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
412     struct got_tree_object *tree)
413 {
414 #ifndef GOT_NO_OBJ_CACHE
415 	const struct got_error *err = NULL;
416 	err = got_object_cache_add(&repo->treecache, id, tree);
417 	if (err) {
418 		if (err->code == GOT_ERR_OBJ_EXISTS ||
419 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
420 			err = NULL;
421 		return err;
422 	}
423 	tree->refcnt++;
424 #endif
425 	return NULL;
426 }
427 
428 struct got_tree_object *
429 got_repo_get_cached_tree(struct got_repository *repo,
430     struct got_object_id *id)
431 {
432 	return (struct got_tree_object *)got_object_cache_get(
433 	    &repo->treecache, id);
434 }
435 
436 const struct got_error *
437 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
438     struct got_commit_object *commit)
439 {
440 #ifndef GOT_NO_OBJ_CACHE
441 	const struct got_error *err = NULL;
442 	err = got_object_cache_add(&repo->commitcache, id, commit);
443 	if (err) {
444 		if (err->code == GOT_ERR_OBJ_EXISTS ||
445 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
446 			err = NULL;
447 		return err;
448 	}
449 	commit->refcnt++;
450 #endif
451 	return NULL;
452 }
453 
454 struct got_commit_object *
455 got_repo_get_cached_commit(struct got_repository *repo,
456     struct got_object_id *id)
457 {
458 	return (struct got_commit_object *)got_object_cache_get(
459 	    &repo->commitcache, id);
460 }
461 
462 const struct got_error *
463 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
464     struct got_tag_object *tag)
465 {
466 #ifndef GOT_NO_OBJ_CACHE
467 	const struct got_error *err = NULL;
468 	err = got_object_cache_add(&repo->tagcache, id, tag);
469 	if (err) {
470 		if (err->code == GOT_ERR_OBJ_EXISTS ||
471 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
472 			err = NULL;
473 		return err;
474 	}
475 	tag->refcnt++;
476 #endif
477 	return NULL;
478 }
479 
480 struct got_tag_object *
481 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
482 {
483 	return (struct got_tag_object *)got_object_cache_get(
484 	    &repo->tagcache, id);
485 }
486 
487 const struct got_error *
488 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
489     struct got_raw_object *raw)
490 {
491 #ifndef GOT_NO_OBJ_CACHE
492 	const struct got_error *err = NULL;
493 	err = got_object_cache_add(&repo->rawcache, id, raw);
494 	if (err) {
495 		if (err->code == GOT_ERR_OBJ_EXISTS ||
496 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
497 			err = NULL;
498 		return err;
499 	}
500 	raw->refcnt++;
501 #endif
502 	return NULL;
503 }
504 
505 
506 struct got_raw_object *
507 got_repo_get_cached_raw_object(struct got_repository *repo,
508     struct got_object_id *id)
509 {
510 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
511 }
512 
513 
514 static const struct got_error *
515 open_repo(struct got_repository *repo, const char *path)
516 {
517 	const struct got_error *err = NULL;
518 
519 	repo->gitdir_fd = -1;
520 
521 	/* bare git repository? */
522 	repo->path_git_dir = strdup(path);
523 	if (repo->path_git_dir == NULL)
524 		return got_error_from_errno("strdup");
525 	if (is_git_repo(repo)) {
526 		repo->path = strdup(repo->path_git_dir);
527 		if (repo->path == NULL) {
528 			err = got_error_from_errno("strdup");
529 			goto done;
530 		}
531 		repo->gitdir_fd = open(repo->path_git_dir,
532 		    O_DIRECTORY | O_CLOEXEC);
533 		if (repo->gitdir_fd == -1) {
534 			err = got_error_from_errno2("open",
535 			    repo->path_git_dir);
536 			goto done;
537 		}
538 		return NULL;
539 	}
540 
541 	/* git repository with working tree? */
542 	free(repo->path_git_dir);
543 	repo->path_git_dir = NULL;
544 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
545 		err = got_error_from_errno("asprintf");
546 		goto done;
547 	}
548 	if (is_git_repo(repo)) {
549 		repo->path = strdup(path);
550 		if (repo->path == NULL) {
551 			err = got_error_from_errno("strdup");
552 			goto done;
553 		}
554 		repo->gitdir_fd = open(repo->path_git_dir,
555 		    O_DIRECTORY | O_CLOEXEC);
556 		if (repo->gitdir_fd == -1) {
557 			err = got_error_from_errno2("open",
558 			    repo->path_git_dir);
559 			goto done;
560 		}
561 		return NULL;
562 	}
563 
564 	err = got_error(GOT_ERR_NOT_GIT_REPO);
565 done:
566 	if (err) {
567 		free(repo->path);
568 		repo->path = NULL;
569 		free(repo->path_git_dir);
570 		repo->path_git_dir = NULL;
571 		if (repo->gitdir_fd != -1)
572 			close(repo->gitdir_fd);
573 		repo->gitdir_fd = -1;
574 
575 	}
576 	return err;
577 }
578 
579 static const struct got_error *
580 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
581 {
582 	const struct got_error *err = NULL;
583 	char *repo_gitconfig_path = NULL;
584 
585 	if (global_gitconfig_path) {
586 		/* Read settings from ~/.gitconfig. */
587 		int dummy_repo_version;
588 		err = got_repo_read_gitconfig(&dummy_repo_version,
589 		    &repo->global_gitconfig_author_name,
590 		    &repo->global_gitconfig_author_email,
591 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
592 		if (err)
593 			return err;
594 	}
595 
596 	/* Read repository's .git/config file. */
597 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
598 	if (repo_gitconfig_path == NULL)
599 		return got_error_from_errno("got_repo_get_path_gitconfig");
600 
601 	err = got_repo_read_gitconfig(
602 	    &repo->gitconfig_repository_format_version,
603 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
604 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
605 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
606 	    repo_gitconfig_path);
607 	if (err)
608 		goto done;
609 
610 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
611 		int i;
612 
613 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
614 			got_repo_free_remote_repo_data(
615 			    &repo->gitconfig_remotes[i]);
616 		}
617 		free(repo->gitconfig_remotes);
618 		repo->gitconfig_remotes = NULL;
619 		repo->ngitconfig_remotes = 0;
620 
621 		free(repo->gitconfig_author_name);
622 		repo->gitconfig_author_name = NULL;
623 		free(repo->gitconfig_author_email);
624 		repo->gitconfig_author_email = NULL;
625 
626 		free(repo->global_gitconfig_author_name);
627 		repo->global_gitconfig_author_name = NULL;
628 		free(repo->global_gitconfig_author_email);
629 		repo->global_gitconfig_author_email = NULL;
630 	}
631 
632 done:
633 	free(repo_gitconfig_path);
634 	return err;
635 }
636 
637 static const struct got_error *
638 read_gotconfig(struct got_repository *repo)
639 {
640 	const struct got_error *err = NULL;
641 	char *gotconfig_path;
642 
643 	gotconfig_path = got_repo_get_path_gotconfig(repo);
644 	if (gotconfig_path == NULL)
645 		return got_error_from_errno("got_repo_get_path_gotconfig");
646 
647 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
648 	free(gotconfig_path);
649 	return err;
650 }
651 
652 /* Supported repository format extensions. */
653 static const char *const repo_extensions[] = {
654 	"noop",			/* Got supports repository format version 1. */
655 	"preciousObjects",	/* Supported by gotadmin cleanup. */
656 	"worktreeConfig",	/* Got does not care about Git work trees. */
657 };
658 
659 const struct got_error *
660 got_repo_open(struct got_repository **repop, const char *path,
661     const char *global_gitconfig_path, int *pack_fds)
662 {
663 	struct got_repository *repo = NULL;
664 	const struct got_error *err = NULL;
665 	char *repo_path = NULL;
666 	size_t i, j = 0;
667 
668 	*repop = NULL;
669 
670 	repo = calloc(1, sizeof(*repo));
671 	if (repo == NULL)
672 		return got_error_from_errno("calloc");
673 
674 	RB_INIT(&repo->packidx_bloom_filters);
675 	TAILQ_INIT(&repo->packidx_paths);
676 
677 	for (i = 0; i < nitems(repo->privsep_children); i++) {
678 		memset(&repo->privsep_children[i], 0,
679 		    sizeof(repo->privsep_children[0]));
680 		repo->privsep_children[i].imsg_fd = -1;
681 	}
682 
683 	err = got_object_cache_init(&repo->objcache,
684 	    GOT_OBJECT_CACHE_TYPE_OBJ);
685 	if (err)
686 		goto done;
687 	err = got_object_cache_init(&repo->treecache,
688 	    GOT_OBJECT_CACHE_TYPE_TREE);
689 	if (err)
690 		goto done;
691 	err = got_object_cache_init(&repo->commitcache,
692 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
693 	if (err)
694 		goto done;
695 	err = got_object_cache_init(&repo->tagcache,
696 	    GOT_OBJECT_CACHE_TYPE_TAG);
697 	if (err)
698 		goto done;
699 	err = got_object_cache_init(&repo->rawcache,
700 	    GOT_OBJECT_CACHE_TYPE_RAW);
701 	if (err)
702 		goto done;
703 
704 	err = get_pack_cache_size(&repo->pack_cache_size);
705 	if (err)
706 		goto done;
707 	for (i = 0; i < nitems(repo->packs); i++) {
708 		if (pack_fds != NULL && i < repo->pack_cache_size) {
709 			repo->packs[i].basefd = pack_fds[j++];
710 			repo->packs[i].accumfd = pack_fds[j++];
711 		} else {
712 			repo->packs[i].basefd = -1;
713 			repo->packs[i].accumfd = -1;
714 		}
715 	}
716 	for (i = 0; i < nitems(repo->tempfiles); i++)
717 		repo->tempfiles[i] = -1;
718 	repo->pinned_pack = -1;
719 	repo->pinned_packidx = -1;
720 	repo->pinned_pid = 0;
721 
722 	repo_path = realpath(path, NULL);
723 	if (repo_path == NULL) {
724 		err = got_error_from_errno2("realpath", path);
725 		goto done;
726 	}
727 
728 	for (;;) {
729 		char *parent_path;
730 
731 		err = open_repo(repo, repo_path);
732 		if (err == NULL)
733 			break;
734 		if (err->code != GOT_ERR_NOT_GIT_REPO)
735 			goto done;
736 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
737 			err = got_error(GOT_ERR_NOT_GIT_REPO);
738 			goto done;
739 		}
740 		err = got_path_dirname(&parent_path, repo_path);
741 		if (err)
742 			goto done;
743 		free(repo_path);
744 		repo_path = parent_path;
745 	}
746 
747 	err = read_gotconfig(repo);
748 	if (err)
749 		goto done;
750 
751 	err = read_gitconfig(repo, global_gitconfig_path);
752 	if (err)
753 		goto done;
754 	if (repo->gitconfig_repository_format_version != 0) {
755 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
756 		goto done;
757 	}
758 	for (i = 0; i < repo->nextensions; i++) {
759 		char *ext = repo->extensions[i];
760 		int j, supported = 0;
761 		for (j = 0; j < nitems(repo_extensions); j++) {
762 			if (strcmp(ext, repo_extensions[j]) == 0) {
763 				supported = 1;
764 				break;
765 			}
766 		}
767 		if (!supported) {
768 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
769 			goto done;
770 		}
771 	}
772 
773 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
774 done:
775 	if (err)
776 		got_repo_close(repo);
777 	else
778 		*repop = repo;
779 	free(repo_path);
780 	return err;
781 }
782 
783 const struct got_error *
784 got_repo_close(struct got_repository *repo)
785 {
786 	const struct got_error *err = NULL, *child_err;
787 	struct got_packidx_bloom_filter *bf;
788 	struct got_pathlist_entry *pe;
789 	size_t i;
790 
791 	for (i = 0; i < repo->pack_cache_size; i++) {
792 		if (repo->packidx_cache[i] == NULL)
793 			break;
794 		got_packidx_close(repo->packidx_cache[i]);
795 	}
796 
797 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
798 	    &repo->packidx_bloom_filters))) {
799 		RB_REMOVE(got_packidx_bloom_filter_tree,
800 		    &repo->packidx_bloom_filters, bf);
801 		bloom_free(bf->bloom);
802 		free(bf->bloom);
803 		free(bf);
804 	}
805 
806 	for (i = 0; i < repo->pack_cache_size; i++)
807 		if (repo->packs[i].path_packfile)
808 			if (repo->packs[i].path_packfile)
809 				got_pack_close(&repo->packs[i]);
810 
811 	free(repo->path);
812 	free(repo->path_git_dir);
813 
814 	got_object_cache_close(&repo->objcache);
815 	got_object_cache_close(&repo->treecache);
816 	got_object_cache_close(&repo->commitcache);
817 	got_object_cache_close(&repo->tagcache);
818 	got_object_cache_close(&repo->rawcache);
819 
820 	for (i = 0; i < nitems(repo->privsep_children); i++) {
821 		if (repo->privsep_children[i].imsg_fd == -1)
822 			continue;
823 		imsg_clear(repo->privsep_children[i].ibuf);
824 		free(repo->privsep_children[i].ibuf);
825 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
826 		child_err = got_privsep_wait_for_child(
827 		    repo->privsep_children[i].pid);
828 		if (child_err && err == NULL)
829 			err = child_err;
830 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
831 		    err == NULL)
832 			err = got_error_from_errno("close");
833 	}
834 
835 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
836 	    err == NULL)
837 		err = got_error_from_errno("close");
838 
839 	if (repo->gotconfig)
840 		got_gotconfig_free(repo->gotconfig);
841 	free(repo->gitconfig_author_name);
842 	free(repo->gitconfig_author_email);
843 	for (i = 0; i < repo->ngitconfig_remotes; i++)
844 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
845 	free(repo->gitconfig_remotes);
846 	for (i = 0; i < repo->nextensions; i++)
847 		free(repo->extensions[i]);
848 	free(repo->extensions);
849 
850 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
851 		free((void *)pe->path);
852 	got_pathlist_free(&repo->packidx_paths);
853 	free(repo);
854 
855 	return err;
856 }
857 
858 void
859 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
860 {
861 	int i;
862 
863 	free(repo->name);
864 	repo->name = NULL;
865 	free(repo->fetch_url);
866 	repo->fetch_url = NULL;
867 	free(repo->send_url);
868 	repo->send_url = NULL;
869 	for (i = 0; i < repo->nfetch_branches; i++)
870 		free(repo->fetch_branches[i]);
871 	free(repo->fetch_branches);
872 	repo->fetch_branches = NULL;
873 	repo->nfetch_branches = 0;
874 	for (i = 0; i < repo->nsend_branches; i++)
875 		free(repo->send_branches[i]);
876 	free(repo->send_branches);
877 	repo->send_branches = NULL;
878 	repo->nsend_branches = 0;
879 }
880 
881 const struct got_error *
882 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
883     const char *input_path)
884 {
885 	const struct got_error *err = NULL;
886 	const char *repo_abspath = NULL;
887 	size_t repolen, len;
888 	char *canonpath, *path = NULL;
889 
890 	*in_repo_path = NULL;
891 
892 	canonpath = strdup(input_path);
893 	if (canonpath == NULL) {
894 		err = got_error_from_errno("strdup");
895 		goto done;
896 	}
897 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
898 	if (err)
899 		goto done;
900 
901 	repo_abspath = got_repo_get_path(repo);
902 
903 	if (canonpath[0] == '\0') {
904 		path = strdup(canonpath);
905 		if (path == NULL) {
906 			err = got_error_from_errno("strdup");
907 			goto done;
908 		}
909 	} else {
910 		path = realpath(canonpath, NULL);
911 		if (path == NULL) {
912 			if (errno != ENOENT) {
913 				err = got_error_from_errno2("realpath",
914 				    canonpath);
915 				goto done;
916 			}
917 			/*
918 			 * Path is not on disk.
919 			 * Assume it is already relative to repository root.
920 			 */
921 			path = strdup(canonpath);
922 			if (path == NULL) {
923 				err = got_error_from_errno("strdup");
924 				goto done;
925 			}
926 		}
927 
928 		repolen = strlen(repo_abspath);
929 		len = strlen(path);
930 
931 
932 		if (strcmp(path, repo_abspath) == 0) {
933 			free(path);
934 			path = strdup("");
935 			if (path == NULL) {
936 				err = got_error_from_errno("strdup");
937 				goto done;
938 			}
939 		} else if (len > repolen &&
940 		    got_path_is_child(path, repo_abspath, repolen)) {
941 			/* Matched an on-disk path inside repository. */
942 			if (got_repo_is_bare(repo)) {
943 				/*
944 				 * Matched an on-disk path inside repository
945 				 * database. Treat input as repository-relative.
946 				 */
947 				free(path);
948 				path = canonpath;
949 				canonpath = NULL;
950 			} else {
951 				char *child;
952 				/* Strip common prefix with repository path. */
953 				err = got_path_skip_common_ancestor(&child,
954 				    repo_abspath, path);
955 				if (err)
956 					goto done;
957 				free(path);
958 				path = child;
959 			}
960 		} else {
961 			/*
962 			 * Matched unrelated on-disk path.
963 			 * Treat input as repository-relative.
964 			 */
965 			free(path);
966 			path = canonpath;
967 			canonpath = NULL;
968 		}
969 	}
970 
971 	/* Make in-repository path absolute */
972 	if (path[0] != '/') {
973 		char *abspath;
974 		if (asprintf(&abspath, "/%s", path) == -1) {
975 			err = got_error_from_errno("asprintf");
976 			goto done;
977 		}
978 		free(path);
979 		path = abspath;
980 	}
981 
982 done:
983 	free(canonpath);
984 	if (err)
985 		free(path);
986 	else
987 		*in_repo_path = path;
988 	return err;
989 }
990 
991 static const struct got_error *
992 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
993     const char *path_packidx)
994 {
995 	const struct got_error *err = NULL;
996 	size_t i;
997 
998 	for (i = 0; i < repo->pack_cache_size; i++) {
999 		if (repo->packidx_cache[i] == NULL)
1000 			break;
1001 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1002 		    path_packidx) == 0) {
1003 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1004 		}
1005 	}
1006 	if (i == repo->pack_cache_size) {
1007 		do {
1008 			i--;
1009 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1010 		    i == repo->pinned_packidx);
1011 		err = got_packidx_close(repo->packidx_cache[i]);
1012 		if (err)
1013 			return err;
1014 	}
1015 
1016 	repo->packidx_cache[i] = packidx;
1017 
1018 	return NULL;
1019 }
1020 
1021 int
1022 got_repo_is_packidx_filename(const char *name, size_t len)
1023 {
1024 	if (len != GOT_PACKIDX_NAMELEN)
1025 		return 0;
1026 
1027 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1028 		return 0;
1029 
1030 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1031 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1032 		return 0;
1033 
1034 	return 1;
1035 }
1036 
1037 static struct got_packidx_bloom_filter *
1038 get_packidx_bloom_filter(struct got_repository *repo,
1039     const char *path, size_t path_len)
1040 {
1041 	struct got_packidx_bloom_filter key;
1042 
1043 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1044 		return NULL; /* XXX */
1045 	key.path_len = path_len;
1046 
1047 	return RB_FIND(got_packidx_bloom_filter_tree,
1048 	    &repo->packidx_bloom_filters, &key);
1049 }
1050 
1051 int
1052 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1053     const char *path_packidx, struct got_object_id *id)
1054 {
1055 	struct got_packidx_bloom_filter *bf;
1056 
1057 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1058 	if (bf)
1059 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1060 
1061 	/* No bloom filter means this pack index must be searched. */
1062 	return 1;
1063 }
1064 
1065 static const struct got_error *
1066 add_packidx_bloom_filter(struct got_repository *repo,
1067     struct got_packidx *packidx, const char *path_packidx)
1068 {
1069 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1070 	struct got_packidx_bloom_filter *bf;
1071 	size_t len;
1072 
1073 	/*
1074 	 * Don't use bloom filters for very large pack index files.
1075 	 * Large pack files will contain a relatively large fraction
1076 	 * of our objects so we will likely need to visit them anyway.
1077 	 * The more objects a pack file contains the higher the probability
1078 	 * of a false-positive match from the bloom filter. And reading
1079 	 * all object IDs from a large pack index file can be expensive.
1080 	 */
1081 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1082 		return NULL;
1083 
1084 	/* Do we already have a filter for this pack index? */
1085 	if (get_packidx_bloom_filter(repo, path_packidx,
1086 	    strlen(path_packidx)) != NULL)
1087 		return NULL;
1088 
1089 	bf = calloc(1, sizeof(*bf));
1090 	if (bf == NULL)
1091 		return got_error_from_errno("calloc");
1092 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1093 	if (bf->bloom == NULL) {
1094 		free(bf);
1095 		return got_error_from_errno("calloc");
1096 	}
1097 
1098 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1099 	if (len >= sizeof(bf->path)) {
1100 		free(bf->bloom);
1101 		free(bf);
1102 		return got_error(GOT_ERR_NO_SPACE);
1103 	}
1104 	bf->path_len = len;
1105 
1106 	/* Minimum size supported by our bloom filter is 1000 entries. */
1107 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1108 	for (i = 0; i < nobjects; i++) {
1109 		struct got_packidx_object_id *id;
1110 		id = &packidx->hdr.sorted_ids[i];
1111 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1112 	}
1113 
1114 	RB_INSERT(got_packidx_bloom_filter_tree,
1115 	    &repo->packidx_bloom_filters, bf);
1116 	return NULL;
1117 }
1118 
1119 static void
1120 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1121 {
1122 	struct got_pathlist_entry *pe;
1123 
1124 	while (!TAILQ_EMPTY(packidx_paths)) {
1125 		pe = TAILQ_FIRST(packidx_paths);
1126 		TAILQ_REMOVE(packidx_paths, pe, entry);
1127 		free((char *)pe->path);
1128 		free(pe);
1129 	}
1130 }
1131 
1132 static const struct got_error *
1133 refresh_packidx_paths(struct got_repository *repo)
1134 {
1135 	const struct got_error *err = NULL;
1136 	char *objects_pack_dir = NULL;
1137 	struct stat sb;
1138 
1139 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1140 	if (objects_pack_dir == NULL)
1141 		return got_error_from_errno("got_repo_get_path_objects_pack");
1142 
1143 	if (stat(objects_pack_dir, &sb) == -1) {
1144 		if (errno != ENOENT) {
1145 			err = got_error_from_errno2("stat", objects_pack_dir);
1146 			goto done;
1147 		}
1148 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1149 		purge_packidx_paths(&repo->packidx_paths);
1150 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1151 		if (err)
1152 			goto done;
1153 	}
1154 done:
1155 	free(objects_pack_dir);
1156 	return err;
1157 }
1158 
1159 const struct got_error *
1160 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1161     struct got_repository *repo, struct got_object_id *id)
1162 {
1163 	const struct got_error *err;
1164 	struct got_pathlist_entry *pe;
1165 	size_t i;
1166 
1167 	/* Search pack index cache. */
1168 	for (i = 0; i < repo->pack_cache_size; i++) {
1169 		if (repo->packidx_cache[i] == NULL)
1170 			break;
1171 		if (!got_repo_check_packidx_bloom_filter(repo,
1172 		    repo->packidx_cache[i]->path_packidx, id))
1173 			continue; /* object will not be found in this index */
1174 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1175 		if (*idx != -1) {
1176 			*packidx = repo->packidx_cache[i];
1177 			/*
1178 			 * Move this cache entry to the front. Repeatedly
1179 			 * searching a wrong pack index can be expensive.
1180 			 */
1181 			if (i > 0) {
1182 				memmove(&repo->packidx_cache[1],
1183 				    &repo->packidx_cache[0],
1184 				    i * sizeof(repo->packidx_cache[0]));
1185 				repo->packidx_cache[0] = *packidx;
1186 				if (repo->pinned_packidx >= 0 &&
1187 				    repo->pinned_packidx < i)
1188 					repo->pinned_packidx++;
1189 				else if (repo->pinned_packidx == i)
1190 					repo->pinned_packidx = 0;
1191 			}
1192 			return NULL;
1193 		}
1194 	}
1195 	/* No luck. Search the filesystem. */
1196 
1197 	err = refresh_packidx_paths(repo);
1198 	if (err)
1199 		return err;
1200 
1201 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1202 		const char *path_packidx = pe->path;
1203 		int is_cached = 0;
1204 
1205 		if (!got_repo_check_packidx_bloom_filter(repo,
1206 		    pe->path, id))
1207 			continue; /* object will not be found in this index */
1208 
1209 		for (i = 0; i < repo->pack_cache_size; i++) {
1210 			if (repo->packidx_cache[i] == NULL)
1211 				break;
1212 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1213 			    path_packidx) == 0) {
1214 				is_cached = 1;
1215 				break;
1216 			}
1217 		}
1218 		if (is_cached)
1219 			continue; /* already searched */
1220 
1221 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1222 		    path_packidx, 0);
1223 		if (err)
1224 			goto done;
1225 
1226 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1227 		if (err)
1228 			goto done;
1229 
1230 		err = cache_packidx(repo, *packidx, path_packidx);
1231 		if (err)
1232 			goto done;
1233 
1234 		*idx = got_packidx_get_object_idx(*packidx, id);
1235 		if (*idx != -1) {
1236 			err = NULL; /* found the object */
1237 			goto done;
1238 		}
1239 	}
1240 
1241 	err = got_error_no_obj(id);
1242 done:
1243 	return err;
1244 }
1245 
1246 const struct got_error *
1247 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1248     struct got_repository *repo)
1249 {
1250 	const struct got_error *err = NULL;
1251 	DIR *packdir = NULL;
1252 	struct dirent *dent;
1253 	char *path_packidx = NULL;
1254 	int packdir_fd;
1255 	struct stat sb;
1256 
1257 	packdir_fd = openat(got_repo_get_fd(repo),
1258 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1259 	if (packdir_fd == -1) {
1260 		return got_error_from_errno_fmt("openat: %s/%s",
1261 		    got_repo_get_path_git_dir(repo),
1262 		    GOT_OBJECTS_PACK_DIR);
1263 	}
1264 
1265 	packdir = fdopendir(packdir_fd);
1266 	if (packdir == NULL) {
1267 		err = got_error_from_errno("fdopendir");
1268 		goto done;
1269 	}
1270 
1271 	if (fstat(packdir_fd, &sb) == -1) {
1272 		err = got_error_from_errno("fstat");
1273 		goto done;
1274 	}
1275 	repo->pack_path_mtime = sb.st_mtime;
1276 
1277 	while ((dent = readdir(packdir)) != NULL) {
1278 		if (!got_repo_is_packidx_filename(dent->d_name,
1279 		    strlen(dent->d_name)))
1280 			continue;
1281 
1282 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1283 		    dent->d_name) == -1) {
1284 			err = got_error_from_errno("asprintf");
1285 			path_packidx = NULL;
1286 			break;
1287 		}
1288 
1289 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1290 		if (err)
1291 			break;
1292 	}
1293 done:
1294 	if (err)
1295 		free(path_packidx);
1296 	if (packdir && closedir(packdir) != 0 && err == NULL)
1297 		err = got_error_from_errno("closedir");
1298 	return err;
1299 }
1300 
1301 const struct got_error *
1302 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1303     struct got_repository *repo)
1304 {
1305 	const struct got_error *err;
1306 	size_t i;
1307 
1308 	*packidx = NULL;
1309 
1310 	/* Search pack index cache. */
1311 	for (i = 0; i < repo->pack_cache_size; i++) {
1312 		if (repo->packidx_cache[i] == NULL)
1313 			break;
1314 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1315 		    path_packidx) == 0) {
1316 			*packidx = repo->packidx_cache[i];
1317 			return NULL;
1318 		}
1319 	}
1320 	/* No luck. Search the filesystem. */
1321 
1322 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1323 	    path_packidx, 0);
1324 	if (err)
1325 		return err;
1326 
1327 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1328 	if (err)
1329 		goto done;
1330 
1331 	err = cache_packidx(repo, *packidx, path_packidx);
1332 done:
1333 	if (err) {
1334 		got_packidx_close(*packidx);
1335 		*packidx = NULL;
1336 	}
1337 	return err;
1338 }
1339 
1340 static const struct got_error *
1341 read_packfile_hdr(int fd, struct got_packidx *packidx)
1342 {
1343 	const struct got_error *err = NULL;
1344 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1345 	struct got_packfile_hdr hdr;
1346 	ssize_t n;
1347 
1348 	n = read(fd, &hdr, sizeof(hdr));
1349 	if (n < 0)
1350 		return got_error_from_errno("read");
1351 	if (n != sizeof(hdr))
1352 		return got_error(GOT_ERR_BAD_PACKFILE);
1353 
1354 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1355 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1356 	    be32toh(hdr.nobjects) != totobj)
1357 		err = got_error(GOT_ERR_BAD_PACKFILE);
1358 
1359 	return err;
1360 }
1361 
1362 static const struct got_error *
1363 open_packfile(int *fd, struct got_repository *repo,
1364     const char *relpath, struct got_packidx *packidx)
1365 {
1366 	const struct got_error *err = NULL;
1367 
1368 	*fd = openat(got_repo_get_fd(repo), relpath,
1369 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1370 	if (*fd == -1)
1371 		return got_error_from_errno_fmt("openat: %s/%s",
1372 		    got_repo_get_path_git_dir(repo), relpath);
1373 
1374 	if (packidx) {
1375 		err = read_packfile_hdr(*fd, packidx);
1376 		if (err) {
1377 			close(*fd);
1378 			*fd = -1;
1379 		}
1380 	}
1381 
1382 	return err;
1383 }
1384 
1385 const struct got_error *
1386 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1387     const char *path_packfile, struct got_packidx *packidx)
1388 {
1389 	const struct got_error *err = NULL;
1390 	struct got_pack *pack = NULL;
1391 	struct stat sb;
1392 	size_t i;
1393 
1394 	if (packp)
1395 		*packp = NULL;
1396 
1397 	for (i = 0; i < repo->pack_cache_size; i++) {
1398 		pack = &repo->packs[i];
1399 		if (pack->path_packfile == NULL)
1400 			break;
1401 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1402 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1403 	}
1404 
1405 	if (i == repo->pack_cache_size) {
1406 		struct got_pack tmp;
1407 		do {
1408 			i--;
1409 		} while (i > 0 && repo->pinned_pack >= 0 &&
1410 		    i == repo->pinned_pack);
1411 		err = got_pack_close(&repo->packs[i]);
1412 		if (err)
1413 			return err;
1414 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1415 			return got_error_from_errno("ftruncate");
1416 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1417 			return got_error_from_errno("ftruncate");
1418 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1419 		memcpy(&repo->packs[i], &repo->packs[0],
1420 		    sizeof(repo->packs[i]));
1421 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1422 		if (repo->pinned_pack == 0)
1423 			repo->pinned_pack = i;
1424 		else if (repo->pinned_pack == i)
1425 			repo->pinned_pack = 0;
1426 		i = 0;
1427 	}
1428 
1429 	pack = &repo->packs[i];
1430 
1431 	pack->path_packfile = strdup(path_packfile);
1432 	if (pack->path_packfile == NULL) {
1433 		err = got_error_from_errno("strdup");
1434 		goto done;
1435 	}
1436 
1437 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1438 	if (err)
1439 		goto done;
1440 
1441 	if (fstat(pack->fd, &sb) != 0) {
1442 		err = got_error_from_errno("fstat");
1443 		goto done;
1444 	}
1445 	pack->filesize = sb.st_size;
1446 
1447 	pack->privsep_child = NULL;
1448 
1449 	err = got_delta_cache_alloc(&pack->delta_cache);
1450 	if (err)
1451 		goto done;
1452 
1453 #ifndef GOT_PACK_NO_MMAP
1454 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1455 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1456 		    pack->fd, 0);
1457 		if (pack->map == MAP_FAILED) {
1458 			if (errno != ENOMEM) {
1459 				err = got_error_from_errno("mmap");
1460 				goto done;
1461 			}
1462 			pack->map = NULL; /* fall back to read(2) */
1463 		}
1464 	}
1465 #endif
1466 done:
1467 	if (err) {
1468 		if (pack)
1469 			got_pack_close(pack);
1470 	} else if (packp)
1471 		*packp = pack;
1472 	return err;
1473 }
1474 
1475 struct got_pack *
1476 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1477 {
1478 	struct got_pack *pack = NULL;
1479 	size_t i;
1480 
1481 	for (i = 0; i < repo->pack_cache_size; i++) {
1482 		pack = &repo->packs[i];
1483 		if (pack->path_packfile == NULL)
1484 			break;
1485 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1486 			return pack;
1487 	}
1488 
1489 	return NULL;
1490 }
1491 
1492 const struct got_error *
1493 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1494     struct got_pack *pack)
1495 {
1496 	size_t i;
1497 	int pinned_pack = -1, pinned_packidx = -1;
1498 
1499 	for (i = 0; i < repo->pack_cache_size; i++) {
1500 		if (repo->packidx_cache[i] &&
1501 		    strcmp(repo->packidx_cache[i]->path_packidx,
1502 		    packidx->path_packidx) == 0)
1503 			pinned_packidx = i;
1504 		if (repo->packs[i].path_packfile &&
1505 		    strcmp(repo->packs[i].path_packfile,
1506 		    pack->path_packfile) == 0)
1507 			pinned_pack = i;
1508 	}
1509 
1510 	if (pinned_packidx == -1 || pinned_pack == -1)
1511 		return got_error(GOT_ERR_PIN_PACK);
1512 
1513 	repo->pinned_pack = pinned_pack;
1514 	repo->pinned_packidx = pinned_packidx;
1515 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1516 	return NULL;
1517 }
1518 
1519 struct got_pack *
1520 got_repo_get_pinned_pack(struct got_repository *repo)
1521 {
1522 	if (repo->pinned_pack >= 0 &&
1523 	    repo->pinned_pack < repo->pack_cache_size)
1524 		return &repo->packs[repo->pinned_pack];
1525 
1526 	return NULL;
1527 }
1528 
1529 void
1530 got_repo_unpin_pack(struct got_repository *repo)
1531 {
1532 	repo->pinned_packidx = -1;
1533 	repo->pinned_pack = -1;
1534 	repo->pinned_pid = 0;
1535 }
1536 
1537 const struct got_error *
1538 got_repo_init(const char *repo_path, const char *head_name)
1539 {
1540 	const struct got_error *err = NULL;
1541 	const char *dirnames[] = {
1542 		GOT_OBJECTS_DIR,
1543 		GOT_OBJECTS_PACK_DIR,
1544 		GOT_REFS_DIR,
1545 	};
1546 	const char *description_str = "Unnamed repository; "
1547 	    "edit this file 'description' to name the repository.";
1548 	const char *headref = "ref: refs/heads/";
1549 	const char *gitconfig_str = "[core]\n"
1550 	    "\trepositoryformatversion = 0\n"
1551 	    "\tfilemode = true\n"
1552 	    "\tbare = true\n";
1553 	char *headref_str, *path;
1554 	size_t i;
1555 
1556 	if (!got_path_dir_is_empty(repo_path))
1557 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1558 
1559 	for (i = 0; i < nitems(dirnames); i++) {
1560 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1561 			return got_error_from_errno("asprintf");
1562 		}
1563 		err = got_path_mkdir(path);
1564 		free(path);
1565 		if (err)
1566 			return err;
1567 	}
1568 
1569 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1570 		return got_error_from_errno("asprintf");
1571 	err = got_path_create_file(path, description_str);
1572 	free(path);
1573 	if (err)
1574 		return err;
1575 
1576 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1577 		return got_error_from_errno("asprintf");
1578 	if (asprintf(&headref_str, "%s%s", headref,
1579 	    head_name ? head_name : "main") == -1) {
1580 		free(path);
1581 		return got_error_from_errno("asprintf");
1582 	}
1583 	err = got_path_create_file(path, headref_str);
1584 	free(headref_str);
1585 	free(path);
1586 	if (err)
1587 		return err;
1588 
1589 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1590 		return got_error_from_errno("asprintf");
1591 	err = got_path_create_file(path, gitconfig_str);
1592 	free(path);
1593 	if (err)
1594 		return err;
1595 
1596 	return NULL;
1597 }
1598 
1599 static const struct got_error *
1600 match_packed_object(struct got_object_id **unique_id,
1601     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1602 {
1603 	const struct got_error *err = NULL;
1604 	struct got_object_id_queue matched_ids;
1605 	struct got_pathlist_entry *pe;
1606 
1607 	STAILQ_INIT(&matched_ids);
1608 
1609 	err = refresh_packidx_paths(repo);
1610 	if (err)
1611 		return err;
1612 
1613 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1614 		const char *path_packidx = pe->path;
1615 		struct got_packidx *packidx;
1616 		struct got_object_qid *qid;
1617 
1618 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1619 		    path_packidx, 0);
1620 		if (err)
1621 			break;
1622 
1623 		err = got_packidx_match_id_str_prefix(&matched_ids,
1624 		    packidx, id_str_prefix);
1625 		if (err) {
1626 			got_packidx_close(packidx);
1627 			break;
1628 		}
1629 		err = got_packidx_close(packidx);
1630 		if (err)
1631 			break;
1632 
1633 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1634 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1635 				int matched_type;
1636 				err = got_object_get_type(&matched_type, repo,
1637 				    &qid->id);
1638 				if (err)
1639 					goto done;
1640 				if (matched_type != obj_type)
1641 					continue;
1642 			}
1643 			if (*unique_id == NULL) {
1644 				*unique_id = got_object_id_dup(&qid->id);
1645 				if (*unique_id == NULL) {
1646 					err = got_error_from_errno("malloc");
1647 					goto done;
1648 				}
1649 			} else {
1650 				if (got_object_id_cmp(*unique_id,
1651 				    &qid->id) == 0)
1652 					continue; /* packed multiple times */
1653 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1654 				goto done;
1655 			}
1656 		}
1657 	}
1658 done:
1659 	got_object_id_queue_free(&matched_ids);
1660 	if (err) {
1661 		free(*unique_id);
1662 		*unique_id = NULL;
1663 	}
1664 	return err;
1665 }
1666 
1667 static const struct got_error *
1668 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1669     const char *object_dir, const char *id_str_prefix, int obj_type,
1670     struct got_repository *repo)
1671 {
1672 	const struct got_error *err = NULL;
1673 	char *path, *id_str = NULL;
1674 	DIR *dir = NULL;
1675 	struct dirent *dent;
1676 	struct got_object_id id;
1677 
1678 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1679 		err = got_error_from_errno("asprintf");
1680 		goto done;
1681 	}
1682 
1683 	dir = opendir(path);
1684 	if (dir == NULL) {
1685 		if (errno == ENOENT) {
1686 			err = NULL;
1687 			goto done;
1688 		}
1689 		err = got_error_from_errno2("opendir", path);
1690 		goto done;
1691 	}
1692 	while ((dent = readdir(dir)) != NULL) {
1693 		int cmp;
1694 
1695 		free(id_str);
1696 		id_str = NULL;
1697 
1698 		if (strcmp(dent->d_name, ".") == 0 ||
1699 		    strcmp(dent->d_name, "..") == 0)
1700 			continue;
1701 
1702 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1703 			err = got_error_from_errno("asprintf");
1704 			goto done;
1705 		}
1706 
1707 		if (!got_parse_sha1_digest(id.sha1, id_str))
1708 			continue;
1709 
1710 		/*
1711 		 * Directory entries do not necessarily appear in
1712 		 * sorted order, so we must iterate over all of them.
1713 		 */
1714 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1715 		if (cmp != 0)
1716 			continue;
1717 
1718 		if (*unique_id == NULL) {
1719 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1720 				int matched_type;
1721 				err = got_object_get_type(&matched_type, repo,
1722 				    &id);
1723 				if (err)
1724 					goto done;
1725 				if (matched_type != obj_type)
1726 					continue;
1727 			}
1728 			*unique_id = got_object_id_dup(&id);
1729 			if (*unique_id == NULL) {
1730 				err = got_error_from_errno("got_object_id_dup");
1731 				goto done;
1732 			}
1733 		} else {
1734 			if (got_object_id_cmp(*unique_id, &id) == 0)
1735 				continue; /* both packed and loose */
1736 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1737 			goto done;
1738 		}
1739 	}
1740 done:
1741 	if (dir && closedir(dir) != 0 && err == NULL)
1742 		err = got_error_from_errno("closedir");
1743 	if (err) {
1744 		free(*unique_id);
1745 		*unique_id = NULL;
1746 	}
1747 	free(id_str);
1748 	free(path);
1749 	return err;
1750 }
1751 
1752 const struct got_error *
1753 got_repo_match_object_id_prefix(struct got_object_id **id,
1754     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1755 {
1756 	const struct got_error *err = NULL;
1757 	char *path_objects = NULL, *object_dir = NULL;
1758 	size_t len;
1759 	int i;
1760 
1761 	*id = NULL;
1762 
1763 	path_objects = got_repo_get_path_objects(repo);
1764 
1765 	len = strlen(id_str_prefix);
1766 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1767 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1768 		goto done;
1769 	}
1770 
1771 	for (i = 0; i < len; i++) {
1772 		if (isxdigit((unsigned char)id_str_prefix[i]))
1773 			continue;
1774 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1775 		goto done;
1776 	}
1777 
1778 	if (len >= 2) {
1779 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1780 		if (err)
1781 			goto done;
1782 		object_dir = strndup(id_str_prefix, 2);
1783 		if (object_dir == NULL) {
1784 			err = got_error_from_errno("strdup");
1785 			goto done;
1786 		}
1787 		err = match_loose_object(id, path_objects, object_dir,
1788 		    id_str_prefix, obj_type, repo);
1789 	} else if (len == 1) {
1790 		int i;
1791 		for (i = 0; i < 0xf; i++) {
1792 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1793 			    == -1) {
1794 				err = got_error_from_errno("asprintf");
1795 				goto done;
1796 			}
1797 			err = match_packed_object(id, repo, object_dir,
1798 			    obj_type);
1799 			if (err)
1800 				goto done;
1801 			err = match_loose_object(id, path_objects, object_dir,
1802 			    id_str_prefix, obj_type, repo);
1803 			if (err)
1804 				goto done;
1805 		}
1806 	} else {
1807 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1808 		goto done;
1809 	}
1810 done:
1811 	free(path_objects);
1812 	free(object_dir);
1813 	if (err) {
1814 		free(*id);
1815 		*id = NULL;
1816 	} else if (*id == NULL) {
1817 		switch (obj_type) {
1818 		case GOT_OBJ_TYPE_BLOB:
1819 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1820 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1821 			break;
1822 		case GOT_OBJ_TYPE_TREE:
1823 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1824 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1825 			break;
1826 		case GOT_OBJ_TYPE_COMMIT:
1827 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1828 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1829 			break;
1830 		case GOT_OBJ_TYPE_TAG:
1831 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1832 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1833 			break;
1834 		default:
1835 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1836 			break;
1837 		}
1838 	}
1839 
1840 	return err;
1841 }
1842 
1843 const struct got_error *
1844 got_repo_match_object_id(struct got_object_id **id, char **label,
1845     const char *id_str, int obj_type, struct got_reflist_head *refs,
1846     struct got_repository *repo)
1847 {
1848 	const struct got_error *err;
1849 	struct got_tag_object *tag;
1850 	struct got_reference *ref = NULL;
1851 
1852 	*id = NULL;
1853 	if (label)
1854 		*label = NULL;
1855 
1856 	if (refs) {
1857 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1858 		    refs, repo);
1859 		if (err == NULL) {
1860 			*id = got_object_id_dup(
1861 			    got_object_tag_get_object_id(tag));
1862 			if (*id == NULL)
1863 				err = got_error_from_errno("got_object_id_dup");
1864 			else if (label && asprintf(label, "refs/tags/%s",
1865 			    got_object_tag_get_name(tag)) == -1) {
1866 				err = got_error_from_errno("asprintf");
1867 				free(*id);
1868 				*id = NULL;
1869 			}
1870 			got_object_tag_close(tag);
1871 			return err;
1872 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1873 		    err->code != GOT_ERR_NO_OBJ)
1874 			return err;
1875 	}
1876 
1877 	err = got_ref_open(&ref, repo, id_str, 0);
1878 	if (err == NULL) {
1879 		err = got_ref_resolve(id, repo, ref);
1880 		if (err)
1881 			goto done;
1882 		if (label) {
1883 			*label = strdup(got_ref_get_name(ref));
1884 			if (*label == NULL) {
1885 				err = got_error_from_errno("strdup");
1886 				goto done;
1887 			}
1888 		}
1889 	} else {
1890 		if (err->code != GOT_ERR_NOT_REF &&
1891 		    err->code != GOT_ERR_BAD_REF_NAME)
1892 			goto done;
1893 		err = got_repo_match_object_id_prefix(id, id_str,
1894 		    obj_type, repo);
1895 		if (err) {
1896 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1897 				err = got_error_not_ref(id_str);
1898 			goto done;
1899 		}
1900 		if (label) {
1901 			err = got_object_id_str(label, *id);
1902 			if (*label == NULL) {
1903 				err = got_error_from_errno("strdup");
1904 				goto done;
1905 			}
1906 		}
1907 	}
1908 done:
1909 	if (ref)
1910 		got_ref_close(ref);
1911 	return err;
1912 }
1913 
1914 const struct got_error *
1915 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1916     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1917 {
1918 	const struct got_error *err = NULL;
1919 	struct got_reflist_entry *re;
1920 	struct got_object_id *tag_id;
1921 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1922 
1923 	*tag = NULL;
1924 
1925 	TAILQ_FOREACH(re, refs, entry) {
1926 		const char *refname;
1927 		refname = got_ref_get_name(re->ref);
1928 		if (got_ref_is_symbolic(re->ref))
1929 			continue;
1930 		if (strncmp(refname, "refs/tags/", 10) != 0)
1931 			continue;
1932 		if (!name_is_absolute)
1933 			refname += strlen("refs/tags/");
1934 		if (strcmp(refname, name) != 0)
1935 			continue;
1936 		err = got_ref_resolve(&tag_id, repo, re->ref);
1937 		if (err)
1938 			break;
1939 		err = got_object_open_as_tag(tag, repo, tag_id);
1940 		free(tag_id);
1941 		if (err)
1942 			break;
1943 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1944 		    got_object_tag_get_object_type(*tag) == obj_type)
1945 			break;
1946 		got_object_tag_close(*tag);
1947 		*tag = NULL;
1948 	}
1949 
1950 	if (err == NULL && *tag == NULL)
1951 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1952 		    GOT_OBJ_LABEL_TAG, name);
1953 	return err;
1954 }
1955 
1956 static const struct got_error *
1957 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1958     const char *name, mode_t mode, struct got_object_id *blob_id)
1959 {
1960 	const struct got_error *err = NULL;
1961 
1962 	 *new_te = NULL;
1963 
1964 	*new_te = calloc(1, sizeof(**new_te));
1965 	if (*new_te == NULL)
1966 		return got_error_from_errno("calloc");
1967 
1968 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1969 	    sizeof((*new_te)->name)) {
1970 		err = got_error(GOT_ERR_NO_SPACE);
1971 		goto done;
1972 	}
1973 
1974 	if (S_ISLNK(mode)) {
1975 		(*new_te)->mode = S_IFLNK;
1976 	} else {
1977 		(*new_te)->mode = S_IFREG;
1978 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1979 	}
1980 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1981 done:
1982 	if (err && *new_te) {
1983 		free(*new_te);
1984 		*new_te = NULL;
1985 	}
1986 	return err;
1987 }
1988 
1989 static const struct got_error *
1990 import_file(struct got_tree_entry **new_te, struct dirent *de,
1991     const char *path, struct got_repository *repo)
1992 {
1993 	const struct got_error *err;
1994 	struct got_object_id *blob_id = NULL;
1995 	char *filepath;
1996 	struct stat sb;
1997 
1998 	if (asprintf(&filepath, "%s%s%s", path,
1999 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2000 		return got_error_from_errno("asprintf");
2001 
2002 	if (lstat(filepath, &sb) != 0) {
2003 		err = got_error_from_errno2("lstat", path);
2004 		goto done;
2005 	}
2006 
2007 	err = got_object_blob_create(&blob_id, filepath, repo);
2008 	if (err)
2009 		goto done;
2010 
2011 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2012 	    blob_id);
2013 done:
2014 	free(filepath);
2015 	if (err)
2016 		free(blob_id);
2017 	return err;
2018 }
2019 
2020 static const struct got_error *
2021 insert_tree_entry(struct got_tree_entry *new_te,
2022     struct got_pathlist_head *paths)
2023 {
2024 	const struct got_error *err = NULL;
2025 	struct got_pathlist_entry *new_pe;
2026 
2027 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2028 	if (err)
2029 		return err;
2030 	if (new_pe == NULL)
2031 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2032 	return NULL;
2033 }
2034 
2035 static const struct got_error *write_tree(struct got_object_id **,
2036     const char *, struct got_pathlist_head *, struct got_repository *,
2037     got_repo_import_cb progress_cb, void *progress_arg);
2038 
2039 static const struct got_error *
2040 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2041     const char *path, struct got_pathlist_head *ignores,
2042     struct got_repository *repo,
2043     got_repo_import_cb progress_cb, void *progress_arg)
2044 {
2045 	const struct got_error *err;
2046 	struct got_object_id *id = NULL;
2047 	char *subdirpath;
2048 
2049 	if (asprintf(&subdirpath, "%s%s%s", path,
2050 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2051 		return got_error_from_errno("asprintf");
2052 
2053 	(*new_te) = calloc(1, sizeof(**new_te));
2054 	if (*new_te == NULL)
2055 		return got_error_from_errno("calloc");
2056 	(*new_te)->mode = S_IFDIR;
2057 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2058 	    sizeof((*new_te)->name)) {
2059 		err = got_error(GOT_ERR_NO_SPACE);
2060 		goto done;
2061 	}
2062 	err = write_tree(&id, subdirpath, ignores,  repo,
2063 	    progress_cb, progress_arg);
2064 	if (err)
2065 		goto done;
2066 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2067 
2068 done:
2069 	free(id);
2070 	free(subdirpath);
2071 	if (err) {
2072 		free(*new_te);
2073 		*new_te = NULL;
2074 	}
2075 	return err;
2076 }
2077 
2078 static const struct got_error *
2079 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2080     struct got_pathlist_head *ignores, struct got_repository *repo,
2081     got_repo_import_cb progress_cb, void *progress_arg)
2082 {
2083 	const struct got_error *err = NULL;
2084 	DIR *dir;
2085 	struct dirent *de;
2086 	int nentries;
2087 	struct got_tree_entry *new_te = NULL;
2088 	struct got_pathlist_head paths;
2089 	struct got_pathlist_entry *pe;
2090 
2091 	*new_tree_id = NULL;
2092 
2093 	TAILQ_INIT(&paths);
2094 
2095 	dir = opendir(path_dir);
2096 	if (dir == NULL) {
2097 		err = got_error_from_errno2("opendir", path_dir);
2098 		goto done;
2099 	}
2100 
2101 	nentries = 0;
2102 	while ((de = readdir(dir)) != NULL) {
2103 		int ignore = 0;
2104 		int type;
2105 
2106 		if (strcmp(de->d_name, ".") == 0 ||
2107 		    strcmp(de->d_name, "..") == 0)
2108 			continue;
2109 
2110 		TAILQ_FOREACH(pe, ignores, entry) {
2111 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2112 				ignore = 1;
2113 				break;
2114 			}
2115 		}
2116 		if (ignore)
2117 			continue;
2118 
2119 		err = got_path_dirent_type(&type, path_dir, de);
2120 		if (err)
2121 			goto done;
2122 
2123 		if (type == DT_DIR) {
2124 			err = import_subdir(&new_te, de, path_dir,
2125 			    ignores, repo, progress_cb, progress_arg);
2126 			if (err) {
2127 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2128 					goto done;
2129 				err = NULL;
2130 				continue;
2131 			}
2132 		} else if (type == DT_REG || type == DT_LNK) {
2133 			err = import_file(&new_te, de, path_dir, repo);
2134 			if (err)
2135 				goto done;
2136 		} else
2137 			continue;
2138 
2139 		err = insert_tree_entry(new_te, &paths);
2140 		if (err)
2141 			goto done;
2142 		nentries++;
2143 	}
2144 
2145 	if (TAILQ_EMPTY(&paths)) {
2146 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2147 		    "cannot create tree without any entries");
2148 		goto done;
2149 	}
2150 
2151 	TAILQ_FOREACH(pe, &paths, entry) {
2152 		struct got_tree_entry *te = pe->data;
2153 		char *path;
2154 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2155 			continue;
2156 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2157 			err = got_error_from_errno("asprintf");
2158 			goto done;
2159 		}
2160 		err = (*progress_cb)(progress_arg, path);
2161 		free(path);
2162 		if (err)
2163 			goto done;
2164 	}
2165 
2166 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2167 done:
2168 	if (dir)
2169 		closedir(dir);
2170 	got_pathlist_free(&paths);
2171 	return err;
2172 }
2173 
2174 const struct got_error *
2175 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2176     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2177     struct got_repository *repo, got_repo_import_cb progress_cb,
2178     void *progress_arg)
2179 {
2180 	const struct got_error *err;
2181 	struct got_object_id *new_tree_id;
2182 
2183 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2184 	    progress_cb, progress_arg);
2185 	if (err)
2186 		return err;
2187 
2188 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2189 	    author, time(NULL), author, time(NULL), logmsg, repo);
2190 	free(new_tree_id);
2191 	return err;
2192 }
2193 
2194 const struct got_error *
2195 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2196     struct got_repository *repo)
2197 {
2198 	const struct got_error *err = NULL;
2199 	char *path_objects = NULL, *path = NULL;
2200 	DIR *dir = NULL;
2201 	struct got_object_id id;
2202 	int i;
2203 
2204 	*nobjects = 0;
2205 	*ondisk_size = 0;
2206 
2207 	path_objects = got_repo_get_path_objects(repo);
2208 	if (path_objects == NULL)
2209 		return got_error_from_errno("got_repo_get_path_objects");
2210 
2211 	for (i = 0; i <= 0xff; i++) {
2212 		struct dirent *dent;
2213 
2214 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2215 			err = got_error_from_errno("asprintf");
2216 			break;
2217 		}
2218 
2219 		dir = opendir(path);
2220 		if (dir == NULL) {
2221 			if (errno == ENOENT) {
2222 				err = NULL;
2223 				continue;
2224 			}
2225 			err = got_error_from_errno2("opendir", path);
2226 			break;
2227 		}
2228 
2229 		while ((dent = readdir(dir)) != NULL) {
2230 			char *id_str;
2231 			int fd;
2232 			struct stat sb;
2233 
2234 			if (strcmp(dent->d_name, ".") == 0 ||
2235 			    strcmp(dent->d_name, "..") == 0)
2236 				continue;
2237 
2238 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2239 				err = got_error_from_errno("asprintf");
2240 				goto done;
2241 			}
2242 
2243 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2244 				free(id_str);
2245 				continue;
2246 			}
2247 			free(id_str);
2248 
2249 			err = got_object_open_loose_fd(&fd, &id, repo);
2250 			if (err)
2251 				goto done;
2252 
2253 			if (fstat(fd, &sb) == -1) {
2254 				err = got_error_from_errno("fstat");
2255 				close(fd);
2256 				goto done;
2257 			}
2258 			(*nobjects)++;
2259 			(*ondisk_size) += sb.st_size;
2260 
2261 			if (close(fd) == -1) {
2262 				err = got_error_from_errno("close");
2263 				goto done;
2264 			}
2265 		}
2266 
2267 		if (closedir(dir) != 0) {
2268 			err = got_error_from_errno("closedir");
2269 			goto done;
2270 		}
2271 		dir = NULL;
2272 
2273 		free(path);
2274 		path = NULL;
2275 	}
2276 done:
2277 	if (dir && closedir(dir) != 0 && err == NULL)
2278 		err = got_error_from_errno("closedir");
2279 
2280 	if (err) {
2281 		*nobjects = 0;
2282 		*ondisk_size = 0;
2283 	}
2284 	free(path_objects);
2285 	free(path);
2286 	return err;
2287 }
2288 
2289 const struct got_error *
2290 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2291     off_t *total_packsize, struct got_repository *repo)
2292 {
2293 	const struct got_error *err = NULL;
2294 	DIR *packdir = NULL;
2295 	struct dirent *dent;
2296 	struct got_packidx *packidx = NULL;
2297 	char *path_packidx;
2298 	char *path_packfile;
2299 	int packdir_fd;
2300 	struct stat sb;
2301 
2302 	*npackfiles = 0;
2303 	*nobjects = 0;
2304 	*total_packsize = 0;
2305 
2306 	packdir_fd = openat(got_repo_get_fd(repo),
2307 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2308 	if (packdir_fd == -1) {
2309 		return got_error_from_errno_fmt("openat: %s/%s",
2310 		    got_repo_get_path_git_dir(repo),
2311 		    GOT_OBJECTS_PACK_DIR);
2312 	}
2313 
2314 	packdir = fdopendir(packdir_fd);
2315 	if (packdir == NULL) {
2316 		err = got_error_from_errno("fdopendir");
2317 		goto done;
2318 	}
2319 
2320 	while ((dent = readdir(packdir)) != NULL) {
2321 		if (!got_repo_is_packidx_filename(dent->d_name,
2322 		    strlen(dent->d_name)))
2323 			continue;
2324 
2325 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2326 		    dent->d_name) == -1) {
2327 			err = got_error_from_errno("asprintf");
2328 			goto done;
2329 		}
2330 
2331 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2332 		    path_packidx, 0);
2333 		free(path_packidx);
2334 		if (err)
2335 			goto done;
2336 
2337 		if (fstat(packidx->fd, &sb) == -1)
2338 			goto done;
2339 		*total_packsize += sb.st_size;
2340 
2341 		err = got_packidx_get_packfile_path(&path_packfile,
2342 		    packidx->path_packidx);
2343 		if (err)
2344 			goto done;
2345 
2346 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2347 		    0) == -1) {
2348 			free(path_packfile);
2349 			goto done;
2350 		}
2351 		free(path_packfile);
2352 		*total_packsize += sb.st_size;
2353 
2354 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2355 
2356 		(*npackfiles)++;
2357 
2358 		got_packidx_close(packidx);
2359 		packidx = NULL;
2360 	}
2361 done:
2362 	if (packidx)
2363 		got_packidx_close(packidx);
2364 	if (packdir && closedir(packdir) != 0 && err == NULL)
2365 		err = got_error_from_errno("closedir");
2366 	if (err) {
2367 		*npackfiles = 0;
2368 		*nobjects = 0;
2369 		*total_packsize = 0;
2370 	}
2371 	return err;
2372 }
2373 
2374 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2375     got_packidx_bloom_filter_cmp);