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 (TAILQ_EMPTY(&repo->packidx_paths) ||
1149 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1150 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1151 		purge_packidx_paths(&repo->packidx_paths);
1152 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1153 		if (err)
1154 			goto done;
1155 	}
1156 done:
1157 	free(objects_pack_dir);
1158 	return err;
1159 }
1160 
1161 const struct got_error *
1162 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1163     struct got_repository *repo, struct got_object_id *id)
1164 {
1165 	const struct got_error *err;
1166 	struct got_pathlist_entry *pe;
1167 	size_t i;
1168 
1169 	/* Search pack index cache. */
1170 	for (i = 0; i < repo->pack_cache_size; i++) {
1171 		if (repo->packidx_cache[i] == NULL)
1172 			break;
1173 		if (!got_repo_check_packidx_bloom_filter(repo,
1174 		    repo->packidx_cache[i]->path_packidx, id))
1175 			continue; /* object will not be found in this index */
1176 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1177 		if (*idx != -1) {
1178 			*packidx = repo->packidx_cache[i];
1179 			/*
1180 			 * Move this cache entry to the front. Repeatedly
1181 			 * searching a wrong pack index can be expensive.
1182 			 */
1183 			if (i > 0) {
1184 				memmove(&repo->packidx_cache[1],
1185 				    &repo->packidx_cache[0],
1186 				    i * sizeof(repo->packidx_cache[0]));
1187 				repo->packidx_cache[0] = *packidx;
1188 				if (repo->pinned_packidx >= 0 &&
1189 				    repo->pinned_packidx < i)
1190 					repo->pinned_packidx++;
1191 				else if (repo->pinned_packidx == i)
1192 					repo->pinned_packidx = 0;
1193 			}
1194 			return NULL;
1195 		}
1196 	}
1197 	/* No luck. Search the filesystem. */
1198 
1199 	err = refresh_packidx_paths(repo);
1200 	if (err)
1201 		return err;
1202 
1203 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1204 		const char *path_packidx = pe->path;
1205 		int is_cached = 0;
1206 
1207 		if (!got_repo_check_packidx_bloom_filter(repo,
1208 		    pe->path, id))
1209 			continue; /* object will not be found in this index */
1210 
1211 		for (i = 0; i < repo->pack_cache_size; i++) {
1212 			if (repo->packidx_cache[i] == NULL)
1213 				break;
1214 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1215 			    path_packidx) == 0) {
1216 				is_cached = 1;
1217 				break;
1218 			}
1219 		}
1220 		if (is_cached)
1221 			continue; /* already searched */
1222 
1223 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1224 		    path_packidx, 0);
1225 		if (err)
1226 			goto done;
1227 
1228 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1229 		if (err)
1230 			goto done;
1231 
1232 		err = cache_packidx(repo, *packidx, path_packidx);
1233 		if (err)
1234 			goto done;
1235 
1236 		*idx = got_packidx_get_object_idx(*packidx, id);
1237 		if (*idx != -1) {
1238 			err = NULL; /* found the object */
1239 			goto done;
1240 		}
1241 	}
1242 
1243 	err = got_error_no_obj(id);
1244 done:
1245 	return err;
1246 }
1247 
1248 const struct got_error *
1249 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1250     struct got_repository *repo)
1251 {
1252 	const struct got_error *err = NULL;
1253 	DIR *packdir = NULL;
1254 	struct dirent *dent;
1255 	char *path_packidx = NULL;
1256 	int packdir_fd;
1257 	struct stat sb;
1258 
1259 	packdir_fd = openat(got_repo_get_fd(repo),
1260 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1261 	if (packdir_fd == -1) {
1262 		return got_error_from_errno_fmt("openat: %s/%s",
1263 		    got_repo_get_path_git_dir(repo),
1264 		    GOT_OBJECTS_PACK_DIR);
1265 	}
1266 
1267 	packdir = fdopendir(packdir_fd);
1268 	if (packdir == NULL) {
1269 		err = got_error_from_errno("fdopendir");
1270 		goto done;
1271 	}
1272 
1273 	if (fstat(packdir_fd, &sb) == -1) {
1274 		err = got_error_from_errno("fstat");
1275 		goto done;
1276 	}
1277 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1278 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1279 
1280 	while ((dent = readdir(packdir)) != NULL) {
1281 		if (!got_repo_is_packidx_filename(dent->d_name,
1282 		    strlen(dent->d_name)))
1283 			continue;
1284 
1285 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1286 		    dent->d_name) == -1) {
1287 			err = got_error_from_errno("asprintf");
1288 			path_packidx = NULL;
1289 			break;
1290 		}
1291 
1292 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1293 		if (err)
1294 			break;
1295 	}
1296 done:
1297 	if (err)
1298 		free(path_packidx);
1299 	if (packdir && closedir(packdir) != 0 && err == NULL)
1300 		err = got_error_from_errno("closedir");
1301 	return err;
1302 }
1303 
1304 const struct got_error *
1305 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1306     struct got_repository *repo)
1307 {
1308 	const struct got_error *err;
1309 	size_t i;
1310 
1311 	*packidx = NULL;
1312 
1313 	/* Search pack index cache. */
1314 	for (i = 0; i < repo->pack_cache_size; i++) {
1315 		if (repo->packidx_cache[i] == NULL)
1316 			break;
1317 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1318 		    path_packidx) == 0) {
1319 			*packidx = repo->packidx_cache[i];
1320 			return NULL;
1321 		}
1322 	}
1323 	/* No luck. Search the filesystem. */
1324 
1325 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1326 	    path_packidx, 0);
1327 	if (err)
1328 		return err;
1329 
1330 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1331 	if (err)
1332 		goto done;
1333 
1334 	err = cache_packidx(repo, *packidx, path_packidx);
1335 done:
1336 	if (err) {
1337 		got_packidx_close(*packidx);
1338 		*packidx = NULL;
1339 	}
1340 	return err;
1341 }
1342 
1343 static const struct got_error *
1344 read_packfile_hdr(int fd, struct got_packidx *packidx)
1345 {
1346 	const struct got_error *err = NULL;
1347 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1348 	struct got_packfile_hdr hdr;
1349 	ssize_t n;
1350 
1351 	n = read(fd, &hdr, sizeof(hdr));
1352 	if (n < 0)
1353 		return got_error_from_errno("read");
1354 	if (n != sizeof(hdr))
1355 		return got_error(GOT_ERR_BAD_PACKFILE);
1356 
1357 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1358 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1359 	    be32toh(hdr.nobjects) != totobj)
1360 		err = got_error(GOT_ERR_BAD_PACKFILE);
1361 
1362 	return err;
1363 }
1364 
1365 static const struct got_error *
1366 open_packfile(int *fd, struct got_repository *repo,
1367     const char *relpath, struct got_packidx *packidx)
1368 {
1369 	const struct got_error *err = NULL;
1370 
1371 	*fd = openat(got_repo_get_fd(repo), relpath,
1372 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1373 	if (*fd == -1)
1374 		return got_error_from_errno_fmt("openat: %s/%s",
1375 		    got_repo_get_path_git_dir(repo), relpath);
1376 
1377 	if (packidx) {
1378 		err = read_packfile_hdr(*fd, packidx);
1379 		if (err) {
1380 			close(*fd);
1381 			*fd = -1;
1382 		}
1383 	}
1384 
1385 	return err;
1386 }
1387 
1388 const struct got_error *
1389 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1390     const char *path_packfile, struct got_packidx *packidx)
1391 {
1392 	const struct got_error *err = NULL;
1393 	struct got_pack *pack = NULL;
1394 	struct stat sb;
1395 	size_t i;
1396 
1397 	if (packp)
1398 		*packp = NULL;
1399 
1400 	for (i = 0; i < repo->pack_cache_size; i++) {
1401 		pack = &repo->packs[i];
1402 		if (pack->path_packfile == NULL)
1403 			break;
1404 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1405 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1406 	}
1407 
1408 	if (i == repo->pack_cache_size) {
1409 		struct got_pack tmp;
1410 		do {
1411 			i--;
1412 		} while (i > 0 && repo->pinned_pack >= 0 &&
1413 		    i == repo->pinned_pack);
1414 		err = got_pack_close(&repo->packs[i]);
1415 		if (err)
1416 			return err;
1417 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1418 			return got_error_from_errno("ftruncate");
1419 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1420 			return got_error_from_errno("ftruncate");
1421 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1422 		memcpy(&repo->packs[i], &repo->packs[0],
1423 		    sizeof(repo->packs[i]));
1424 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1425 		if (repo->pinned_pack == 0)
1426 			repo->pinned_pack = i;
1427 		else if (repo->pinned_pack == i)
1428 			repo->pinned_pack = 0;
1429 		i = 0;
1430 	}
1431 
1432 	pack = &repo->packs[i];
1433 
1434 	pack->path_packfile = strdup(path_packfile);
1435 	if (pack->path_packfile == NULL) {
1436 		err = got_error_from_errno("strdup");
1437 		goto done;
1438 	}
1439 
1440 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1441 	if (err)
1442 		goto done;
1443 
1444 	if (fstat(pack->fd, &sb) != 0) {
1445 		err = got_error_from_errno("fstat");
1446 		goto done;
1447 	}
1448 	pack->filesize = sb.st_size;
1449 
1450 	pack->privsep_child = NULL;
1451 
1452 	err = got_delta_cache_alloc(&pack->delta_cache);
1453 	if (err)
1454 		goto done;
1455 
1456 #ifndef GOT_PACK_NO_MMAP
1457 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1458 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1459 		    pack->fd, 0);
1460 		if (pack->map == MAP_FAILED) {
1461 			if (errno != ENOMEM) {
1462 				err = got_error_from_errno("mmap");
1463 				goto done;
1464 			}
1465 			pack->map = NULL; /* fall back to read(2) */
1466 		}
1467 	}
1468 #endif
1469 done:
1470 	if (err) {
1471 		if (pack)
1472 			got_pack_close(pack);
1473 	} else if (packp)
1474 		*packp = pack;
1475 	return err;
1476 }
1477 
1478 struct got_pack *
1479 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1480 {
1481 	struct got_pack *pack = NULL;
1482 	size_t i;
1483 
1484 	for (i = 0; i < repo->pack_cache_size; i++) {
1485 		pack = &repo->packs[i];
1486 		if (pack->path_packfile == NULL)
1487 			break;
1488 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1489 			return pack;
1490 	}
1491 
1492 	return NULL;
1493 }
1494 
1495 const struct got_error *
1496 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1497     struct got_pack *pack)
1498 {
1499 	size_t i;
1500 	int pinned_pack = -1, pinned_packidx = -1;
1501 
1502 	for (i = 0; i < repo->pack_cache_size; i++) {
1503 		if (repo->packidx_cache[i] &&
1504 		    strcmp(repo->packidx_cache[i]->path_packidx,
1505 		    packidx->path_packidx) == 0)
1506 			pinned_packidx = i;
1507 		if (repo->packs[i].path_packfile &&
1508 		    strcmp(repo->packs[i].path_packfile,
1509 		    pack->path_packfile) == 0)
1510 			pinned_pack = i;
1511 	}
1512 
1513 	if (pinned_packidx == -1 || pinned_pack == -1)
1514 		return got_error(GOT_ERR_PIN_PACK);
1515 
1516 	repo->pinned_pack = pinned_pack;
1517 	repo->pinned_packidx = pinned_packidx;
1518 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1519 	return NULL;
1520 }
1521 
1522 struct got_pack *
1523 got_repo_get_pinned_pack(struct got_repository *repo)
1524 {
1525 	if (repo->pinned_pack >= 0 &&
1526 	    repo->pinned_pack < repo->pack_cache_size)
1527 		return &repo->packs[repo->pinned_pack];
1528 
1529 	return NULL;
1530 }
1531 
1532 void
1533 got_repo_unpin_pack(struct got_repository *repo)
1534 {
1535 	repo->pinned_packidx = -1;
1536 	repo->pinned_pack = -1;
1537 	repo->pinned_pid = 0;
1538 }
1539 
1540 const struct got_error *
1541 got_repo_init(const char *repo_path, const char *head_name)
1542 {
1543 	const struct got_error *err = NULL;
1544 	const char *dirnames[] = {
1545 		GOT_OBJECTS_DIR,
1546 		GOT_OBJECTS_PACK_DIR,
1547 		GOT_REFS_DIR,
1548 	};
1549 	const char *description_str = "Unnamed repository; "
1550 	    "edit this file 'description' to name the repository.";
1551 	const char *headref = "ref: refs/heads/";
1552 	const char *gitconfig_str = "[core]\n"
1553 	    "\trepositoryformatversion = 0\n"
1554 	    "\tfilemode = true\n"
1555 	    "\tbare = true\n";
1556 	char *headref_str, *path;
1557 	size_t i;
1558 
1559 	if (!got_path_dir_is_empty(repo_path))
1560 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1561 
1562 	for (i = 0; i < nitems(dirnames); i++) {
1563 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1564 			return got_error_from_errno("asprintf");
1565 		}
1566 		err = got_path_mkdir(path);
1567 		free(path);
1568 		if (err)
1569 			return err;
1570 	}
1571 
1572 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1573 		return got_error_from_errno("asprintf");
1574 	err = got_path_create_file(path, description_str);
1575 	free(path);
1576 	if (err)
1577 		return err;
1578 
1579 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1580 		return got_error_from_errno("asprintf");
1581 	if (asprintf(&headref_str, "%s%s", headref,
1582 	    head_name ? head_name : "main") == -1) {
1583 		free(path);
1584 		return got_error_from_errno("asprintf");
1585 	}
1586 	err = got_path_create_file(path, headref_str);
1587 	free(headref_str);
1588 	free(path);
1589 	if (err)
1590 		return err;
1591 
1592 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1593 		return got_error_from_errno("asprintf");
1594 	err = got_path_create_file(path, gitconfig_str);
1595 	free(path);
1596 	if (err)
1597 		return err;
1598 
1599 	return NULL;
1600 }
1601 
1602 static const struct got_error *
1603 match_packed_object(struct got_object_id **unique_id,
1604     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1605 {
1606 	const struct got_error *err = NULL;
1607 	struct got_object_id_queue matched_ids;
1608 	struct got_pathlist_entry *pe;
1609 
1610 	STAILQ_INIT(&matched_ids);
1611 
1612 	err = refresh_packidx_paths(repo);
1613 	if (err)
1614 		return err;
1615 
1616 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1617 		const char *path_packidx = pe->path;
1618 		struct got_packidx *packidx;
1619 		struct got_object_qid *qid;
1620 
1621 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1622 		    path_packidx, 0);
1623 		if (err)
1624 			break;
1625 
1626 		err = got_packidx_match_id_str_prefix(&matched_ids,
1627 		    packidx, id_str_prefix);
1628 		if (err) {
1629 			got_packidx_close(packidx);
1630 			break;
1631 		}
1632 		err = got_packidx_close(packidx);
1633 		if (err)
1634 			break;
1635 
1636 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1637 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1638 				int matched_type;
1639 				err = got_object_get_type(&matched_type, repo,
1640 				    &qid->id);
1641 				if (err)
1642 					goto done;
1643 				if (matched_type != obj_type)
1644 					continue;
1645 			}
1646 			if (*unique_id == NULL) {
1647 				*unique_id = got_object_id_dup(&qid->id);
1648 				if (*unique_id == NULL) {
1649 					err = got_error_from_errno("malloc");
1650 					goto done;
1651 				}
1652 			} else {
1653 				if (got_object_id_cmp(*unique_id,
1654 				    &qid->id) == 0)
1655 					continue; /* packed multiple times */
1656 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1657 				goto done;
1658 			}
1659 		}
1660 	}
1661 done:
1662 	got_object_id_queue_free(&matched_ids);
1663 	if (err) {
1664 		free(*unique_id);
1665 		*unique_id = NULL;
1666 	}
1667 	return err;
1668 }
1669 
1670 static const struct got_error *
1671 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1672     const char *object_dir, const char *id_str_prefix, int obj_type,
1673     struct got_repository *repo)
1674 {
1675 	const struct got_error *err = NULL;
1676 	char *path, *id_str = NULL;
1677 	DIR *dir = NULL;
1678 	struct dirent *dent;
1679 	struct got_object_id id;
1680 
1681 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1682 		err = got_error_from_errno("asprintf");
1683 		goto done;
1684 	}
1685 
1686 	dir = opendir(path);
1687 	if (dir == NULL) {
1688 		if (errno == ENOENT) {
1689 			err = NULL;
1690 			goto done;
1691 		}
1692 		err = got_error_from_errno2("opendir", path);
1693 		goto done;
1694 	}
1695 	while ((dent = readdir(dir)) != NULL) {
1696 		int cmp;
1697 
1698 		free(id_str);
1699 		id_str = NULL;
1700 
1701 		if (strcmp(dent->d_name, ".") == 0 ||
1702 		    strcmp(dent->d_name, "..") == 0)
1703 			continue;
1704 
1705 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1706 			err = got_error_from_errno("asprintf");
1707 			goto done;
1708 		}
1709 
1710 		if (!got_parse_sha1_digest(id.sha1, id_str))
1711 			continue;
1712 
1713 		/*
1714 		 * Directory entries do not necessarily appear in
1715 		 * sorted order, so we must iterate over all of them.
1716 		 */
1717 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1718 		if (cmp != 0)
1719 			continue;
1720 
1721 		if (*unique_id == NULL) {
1722 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1723 				int matched_type;
1724 				err = got_object_get_type(&matched_type, repo,
1725 				    &id);
1726 				if (err)
1727 					goto done;
1728 				if (matched_type != obj_type)
1729 					continue;
1730 			}
1731 			*unique_id = got_object_id_dup(&id);
1732 			if (*unique_id == NULL) {
1733 				err = got_error_from_errno("got_object_id_dup");
1734 				goto done;
1735 			}
1736 		} else {
1737 			if (got_object_id_cmp(*unique_id, &id) == 0)
1738 				continue; /* both packed and loose */
1739 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1740 			goto done;
1741 		}
1742 	}
1743 done:
1744 	if (dir && closedir(dir) != 0 && err == NULL)
1745 		err = got_error_from_errno("closedir");
1746 	if (err) {
1747 		free(*unique_id);
1748 		*unique_id = NULL;
1749 	}
1750 	free(id_str);
1751 	free(path);
1752 	return err;
1753 }
1754 
1755 const struct got_error *
1756 got_repo_match_object_id_prefix(struct got_object_id **id,
1757     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1758 {
1759 	const struct got_error *err = NULL;
1760 	char *path_objects = NULL, *object_dir = NULL;
1761 	size_t len;
1762 	int i;
1763 
1764 	*id = NULL;
1765 
1766 	path_objects = got_repo_get_path_objects(repo);
1767 
1768 	len = strlen(id_str_prefix);
1769 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1770 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1771 		goto done;
1772 	}
1773 
1774 	for (i = 0; i < len; i++) {
1775 		if (isxdigit((unsigned char)id_str_prefix[i]))
1776 			continue;
1777 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1778 		goto done;
1779 	}
1780 
1781 	if (len >= 2) {
1782 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1783 		if (err)
1784 			goto done;
1785 		object_dir = strndup(id_str_prefix, 2);
1786 		if (object_dir == NULL) {
1787 			err = got_error_from_errno("strdup");
1788 			goto done;
1789 		}
1790 		err = match_loose_object(id, path_objects, object_dir,
1791 		    id_str_prefix, obj_type, repo);
1792 	} else if (len == 1) {
1793 		int i;
1794 		for (i = 0; i < 0xf; i++) {
1795 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1796 			    == -1) {
1797 				err = got_error_from_errno("asprintf");
1798 				goto done;
1799 			}
1800 			err = match_packed_object(id, repo, object_dir,
1801 			    obj_type);
1802 			if (err)
1803 				goto done;
1804 			err = match_loose_object(id, path_objects, object_dir,
1805 			    id_str_prefix, obj_type, repo);
1806 			if (err)
1807 				goto done;
1808 		}
1809 	} else {
1810 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1811 		goto done;
1812 	}
1813 done:
1814 	free(path_objects);
1815 	free(object_dir);
1816 	if (err) {
1817 		free(*id);
1818 		*id = NULL;
1819 	} else if (*id == NULL) {
1820 		switch (obj_type) {
1821 		case GOT_OBJ_TYPE_BLOB:
1822 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1823 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1824 			break;
1825 		case GOT_OBJ_TYPE_TREE:
1826 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1827 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1828 			break;
1829 		case GOT_OBJ_TYPE_COMMIT:
1830 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1831 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1832 			break;
1833 		case GOT_OBJ_TYPE_TAG:
1834 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1835 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1836 			break;
1837 		default:
1838 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1839 			break;
1840 		}
1841 	}
1842 
1843 	return err;
1844 }
1845 
1846 const struct got_error *
1847 got_repo_match_object_id(struct got_object_id **id, char **label,
1848     const char *id_str, int obj_type, struct got_reflist_head *refs,
1849     struct got_repository *repo)
1850 {
1851 	const struct got_error *err;
1852 	struct got_tag_object *tag;
1853 	struct got_reference *ref = NULL;
1854 
1855 	*id = NULL;
1856 	if (label)
1857 		*label = NULL;
1858 
1859 	if (refs) {
1860 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1861 		    refs, repo);
1862 		if (err == NULL) {
1863 			*id = got_object_id_dup(
1864 			    got_object_tag_get_object_id(tag));
1865 			if (*id == NULL)
1866 				err = got_error_from_errno("got_object_id_dup");
1867 			else if (label && asprintf(label, "refs/tags/%s",
1868 			    got_object_tag_get_name(tag)) == -1) {
1869 				err = got_error_from_errno("asprintf");
1870 				free(*id);
1871 				*id = NULL;
1872 			}
1873 			got_object_tag_close(tag);
1874 			return err;
1875 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1876 		    err->code != GOT_ERR_NO_OBJ)
1877 			return err;
1878 	}
1879 
1880 	err = got_ref_open(&ref, repo, id_str, 0);
1881 	if (err == NULL) {
1882 		err = got_ref_resolve(id, repo, ref);
1883 		if (err)
1884 			goto done;
1885 		if (label) {
1886 			*label = strdup(got_ref_get_name(ref));
1887 			if (*label == NULL) {
1888 				err = got_error_from_errno("strdup");
1889 				goto done;
1890 			}
1891 		}
1892 	} else {
1893 		if (err->code != GOT_ERR_NOT_REF &&
1894 		    err->code != GOT_ERR_BAD_REF_NAME)
1895 			goto done;
1896 		err = got_repo_match_object_id_prefix(id, id_str,
1897 		    obj_type, repo);
1898 		if (err) {
1899 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1900 				err = got_error_not_ref(id_str);
1901 			goto done;
1902 		}
1903 		if (label) {
1904 			err = got_object_id_str(label, *id);
1905 			if (*label == NULL) {
1906 				err = got_error_from_errno("strdup");
1907 				goto done;
1908 			}
1909 		}
1910 	}
1911 done:
1912 	if (ref)
1913 		got_ref_close(ref);
1914 	return err;
1915 }
1916 
1917 const struct got_error *
1918 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1919     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1920 {
1921 	const struct got_error *err = NULL;
1922 	struct got_reflist_entry *re;
1923 	struct got_object_id *tag_id;
1924 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1925 
1926 	*tag = NULL;
1927 
1928 	TAILQ_FOREACH(re, refs, entry) {
1929 		const char *refname;
1930 		refname = got_ref_get_name(re->ref);
1931 		if (got_ref_is_symbolic(re->ref))
1932 			continue;
1933 		if (strncmp(refname, "refs/tags/", 10) != 0)
1934 			continue;
1935 		if (!name_is_absolute)
1936 			refname += strlen("refs/tags/");
1937 		if (strcmp(refname, name) != 0)
1938 			continue;
1939 		err = got_ref_resolve(&tag_id, repo, re->ref);
1940 		if (err)
1941 			break;
1942 		err = got_object_open_as_tag(tag, repo, tag_id);
1943 		free(tag_id);
1944 		if (err)
1945 			break;
1946 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1947 		    got_object_tag_get_object_type(*tag) == obj_type)
1948 			break;
1949 		got_object_tag_close(*tag);
1950 		*tag = NULL;
1951 	}
1952 
1953 	if (err == NULL && *tag == NULL)
1954 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1955 		    GOT_OBJ_LABEL_TAG, name);
1956 	return err;
1957 }
1958 
1959 static const struct got_error *
1960 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1961     const char *name, mode_t mode, struct got_object_id *blob_id)
1962 {
1963 	const struct got_error *err = NULL;
1964 
1965 	 *new_te = NULL;
1966 
1967 	*new_te = calloc(1, sizeof(**new_te));
1968 	if (*new_te == NULL)
1969 		return got_error_from_errno("calloc");
1970 
1971 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1972 	    sizeof((*new_te)->name)) {
1973 		err = got_error(GOT_ERR_NO_SPACE);
1974 		goto done;
1975 	}
1976 
1977 	if (S_ISLNK(mode)) {
1978 		(*new_te)->mode = S_IFLNK;
1979 	} else {
1980 		(*new_te)->mode = S_IFREG;
1981 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1982 	}
1983 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1984 done:
1985 	if (err && *new_te) {
1986 		free(*new_te);
1987 		*new_te = NULL;
1988 	}
1989 	return err;
1990 }
1991 
1992 static const struct got_error *
1993 import_file(struct got_tree_entry **new_te, struct dirent *de,
1994     const char *path, struct got_repository *repo)
1995 {
1996 	const struct got_error *err;
1997 	struct got_object_id *blob_id = NULL;
1998 	char *filepath;
1999 	struct stat sb;
2000 
2001 	if (asprintf(&filepath, "%s%s%s", path,
2002 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2003 		return got_error_from_errno("asprintf");
2004 
2005 	if (lstat(filepath, &sb) != 0) {
2006 		err = got_error_from_errno2("lstat", path);
2007 		goto done;
2008 	}
2009 
2010 	err = got_object_blob_create(&blob_id, filepath, repo);
2011 	if (err)
2012 		goto done;
2013 
2014 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2015 	    blob_id);
2016 done:
2017 	free(filepath);
2018 	if (err)
2019 		free(blob_id);
2020 	return err;
2021 }
2022 
2023 static const struct got_error *
2024 insert_tree_entry(struct got_tree_entry *new_te,
2025     struct got_pathlist_head *paths)
2026 {
2027 	const struct got_error *err = NULL;
2028 	struct got_pathlist_entry *new_pe;
2029 
2030 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2031 	if (err)
2032 		return err;
2033 	if (new_pe == NULL)
2034 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2035 	return NULL;
2036 }
2037 
2038 static const struct got_error *write_tree(struct got_object_id **,
2039     const char *, struct got_pathlist_head *, struct got_repository *,
2040     got_repo_import_cb progress_cb, void *progress_arg);
2041 
2042 static const struct got_error *
2043 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2044     const char *path, struct got_pathlist_head *ignores,
2045     struct got_repository *repo,
2046     got_repo_import_cb progress_cb, void *progress_arg)
2047 {
2048 	const struct got_error *err;
2049 	struct got_object_id *id = NULL;
2050 	char *subdirpath;
2051 
2052 	if (asprintf(&subdirpath, "%s%s%s", path,
2053 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2054 		return got_error_from_errno("asprintf");
2055 
2056 	(*new_te) = calloc(1, sizeof(**new_te));
2057 	if (*new_te == NULL)
2058 		return got_error_from_errno("calloc");
2059 	(*new_te)->mode = S_IFDIR;
2060 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2061 	    sizeof((*new_te)->name)) {
2062 		err = got_error(GOT_ERR_NO_SPACE);
2063 		goto done;
2064 	}
2065 	err = write_tree(&id, subdirpath, ignores,  repo,
2066 	    progress_cb, progress_arg);
2067 	if (err)
2068 		goto done;
2069 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2070 
2071 done:
2072 	free(id);
2073 	free(subdirpath);
2074 	if (err) {
2075 		free(*new_te);
2076 		*new_te = NULL;
2077 	}
2078 	return err;
2079 }
2080 
2081 static const struct got_error *
2082 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2083     struct got_pathlist_head *ignores, struct got_repository *repo,
2084     got_repo_import_cb progress_cb, void *progress_arg)
2085 {
2086 	const struct got_error *err = NULL;
2087 	DIR *dir;
2088 	struct dirent *de;
2089 	int nentries;
2090 	struct got_tree_entry *new_te = NULL;
2091 	struct got_pathlist_head paths;
2092 	struct got_pathlist_entry *pe;
2093 
2094 	*new_tree_id = NULL;
2095 
2096 	TAILQ_INIT(&paths);
2097 
2098 	dir = opendir(path_dir);
2099 	if (dir == NULL) {
2100 		err = got_error_from_errno2("opendir", path_dir);
2101 		goto done;
2102 	}
2103 
2104 	nentries = 0;
2105 	while ((de = readdir(dir)) != NULL) {
2106 		int ignore = 0;
2107 		int type;
2108 
2109 		if (strcmp(de->d_name, ".") == 0 ||
2110 		    strcmp(de->d_name, "..") == 0)
2111 			continue;
2112 
2113 		TAILQ_FOREACH(pe, ignores, entry) {
2114 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2115 				ignore = 1;
2116 				break;
2117 			}
2118 		}
2119 		if (ignore)
2120 			continue;
2121 
2122 		err = got_path_dirent_type(&type, path_dir, de);
2123 		if (err)
2124 			goto done;
2125 
2126 		if (type == DT_DIR) {
2127 			err = import_subdir(&new_te, de, path_dir,
2128 			    ignores, repo, progress_cb, progress_arg);
2129 			if (err) {
2130 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2131 					goto done;
2132 				err = NULL;
2133 				continue;
2134 			}
2135 		} else if (type == DT_REG || type == DT_LNK) {
2136 			err = import_file(&new_te, de, path_dir, repo);
2137 			if (err)
2138 				goto done;
2139 		} else
2140 			continue;
2141 
2142 		err = insert_tree_entry(new_te, &paths);
2143 		if (err)
2144 			goto done;
2145 		nentries++;
2146 	}
2147 
2148 	if (TAILQ_EMPTY(&paths)) {
2149 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2150 		    "cannot create tree without any entries");
2151 		goto done;
2152 	}
2153 
2154 	TAILQ_FOREACH(pe, &paths, entry) {
2155 		struct got_tree_entry *te = pe->data;
2156 		char *path;
2157 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2158 			continue;
2159 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2160 			err = got_error_from_errno("asprintf");
2161 			goto done;
2162 		}
2163 		err = (*progress_cb)(progress_arg, path);
2164 		free(path);
2165 		if (err)
2166 			goto done;
2167 	}
2168 
2169 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2170 done:
2171 	if (dir)
2172 		closedir(dir);
2173 	got_pathlist_free(&paths);
2174 	return err;
2175 }
2176 
2177 const struct got_error *
2178 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2179     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2180     struct got_repository *repo, got_repo_import_cb progress_cb,
2181     void *progress_arg)
2182 {
2183 	const struct got_error *err;
2184 	struct got_object_id *new_tree_id;
2185 
2186 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2187 	    progress_cb, progress_arg);
2188 	if (err)
2189 		return err;
2190 
2191 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2192 	    author, time(NULL), author, time(NULL), logmsg, repo);
2193 	free(new_tree_id);
2194 	return err;
2195 }
2196 
2197 const struct got_error *
2198 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2199     struct got_repository *repo)
2200 {
2201 	const struct got_error *err = NULL;
2202 	char *path_objects = NULL, *path = NULL;
2203 	DIR *dir = NULL;
2204 	struct got_object_id id;
2205 	int i;
2206 
2207 	*nobjects = 0;
2208 	*ondisk_size = 0;
2209 
2210 	path_objects = got_repo_get_path_objects(repo);
2211 	if (path_objects == NULL)
2212 		return got_error_from_errno("got_repo_get_path_objects");
2213 
2214 	for (i = 0; i <= 0xff; i++) {
2215 		struct dirent *dent;
2216 
2217 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2218 			err = got_error_from_errno("asprintf");
2219 			break;
2220 		}
2221 
2222 		dir = opendir(path);
2223 		if (dir == NULL) {
2224 			if (errno == ENOENT) {
2225 				err = NULL;
2226 				continue;
2227 			}
2228 			err = got_error_from_errno2("opendir", path);
2229 			break;
2230 		}
2231 
2232 		while ((dent = readdir(dir)) != NULL) {
2233 			char *id_str;
2234 			int fd;
2235 			struct stat sb;
2236 
2237 			if (strcmp(dent->d_name, ".") == 0 ||
2238 			    strcmp(dent->d_name, "..") == 0)
2239 				continue;
2240 
2241 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2242 				err = got_error_from_errno("asprintf");
2243 				goto done;
2244 			}
2245 
2246 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2247 				free(id_str);
2248 				continue;
2249 			}
2250 			free(id_str);
2251 
2252 			err = got_object_open_loose_fd(&fd, &id, repo);
2253 			if (err)
2254 				goto done;
2255 
2256 			if (fstat(fd, &sb) == -1) {
2257 				err = got_error_from_errno("fstat");
2258 				close(fd);
2259 				goto done;
2260 			}
2261 			(*nobjects)++;
2262 			(*ondisk_size) += sb.st_size;
2263 
2264 			if (close(fd) == -1) {
2265 				err = got_error_from_errno("close");
2266 				goto done;
2267 			}
2268 		}
2269 
2270 		if (closedir(dir) != 0) {
2271 			err = got_error_from_errno("closedir");
2272 			goto done;
2273 		}
2274 		dir = NULL;
2275 
2276 		free(path);
2277 		path = NULL;
2278 	}
2279 done:
2280 	if (dir && closedir(dir) != 0 && err == NULL)
2281 		err = got_error_from_errno("closedir");
2282 
2283 	if (err) {
2284 		*nobjects = 0;
2285 		*ondisk_size = 0;
2286 	}
2287 	free(path_objects);
2288 	free(path);
2289 	return err;
2290 }
2291 
2292 const struct got_error *
2293 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2294     off_t *total_packsize, struct got_repository *repo)
2295 {
2296 	const struct got_error *err = NULL;
2297 	DIR *packdir = NULL;
2298 	struct dirent *dent;
2299 	struct got_packidx *packidx = NULL;
2300 	char *path_packidx;
2301 	char *path_packfile;
2302 	int packdir_fd;
2303 	struct stat sb;
2304 
2305 	*npackfiles = 0;
2306 	*nobjects = 0;
2307 	*total_packsize = 0;
2308 
2309 	packdir_fd = openat(got_repo_get_fd(repo),
2310 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2311 	if (packdir_fd == -1) {
2312 		return got_error_from_errno_fmt("openat: %s/%s",
2313 		    got_repo_get_path_git_dir(repo),
2314 		    GOT_OBJECTS_PACK_DIR);
2315 	}
2316 
2317 	packdir = fdopendir(packdir_fd);
2318 	if (packdir == NULL) {
2319 		err = got_error_from_errno("fdopendir");
2320 		goto done;
2321 	}
2322 
2323 	while ((dent = readdir(packdir)) != NULL) {
2324 		if (!got_repo_is_packidx_filename(dent->d_name,
2325 		    strlen(dent->d_name)))
2326 			continue;
2327 
2328 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2329 		    dent->d_name) == -1) {
2330 			err = got_error_from_errno("asprintf");
2331 			goto done;
2332 		}
2333 
2334 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2335 		    path_packidx, 0);
2336 		free(path_packidx);
2337 		if (err)
2338 			goto done;
2339 
2340 		if (fstat(packidx->fd, &sb) == -1)
2341 			goto done;
2342 		*total_packsize += sb.st_size;
2343 
2344 		err = got_packidx_get_packfile_path(&path_packfile,
2345 		    packidx->path_packidx);
2346 		if (err)
2347 			goto done;
2348 
2349 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2350 		    0) == -1) {
2351 			free(path_packfile);
2352 			goto done;
2353 		}
2354 		free(path_packfile);
2355 		*total_packsize += sb.st_size;
2356 
2357 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2358 
2359 		(*npackfiles)++;
2360 
2361 		got_packidx_close(packidx);
2362 		packidx = NULL;
2363 	}
2364 done:
2365 	if (packidx)
2366 		got_packidx_close(packidx);
2367 	if (packdir && closedir(packdir) != 0 && err == NULL)
2368 		err = got_error_from_errno("closedir");
2369 	if (err) {
2370 		*npackfiles = 0;
2371 		*nobjects = 0;
2372 		*total_packsize = 0;
2373 	}
2374 	return err;
2375 }
2376 
2377 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2378     got_packidx_bloom_filter_cmp);