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