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