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 	size_t i;
789 
790 	for (i = 0; i < repo->pack_cache_size; i++) {
791 		if (repo->packidx_cache[i] == NULL)
792 			break;
793 		got_packidx_close(repo->packidx_cache[i]);
794 	}
795 
796 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
797 	    &repo->packidx_bloom_filters))) {
798 		RB_REMOVE(got_packidx_bloom_filter_tree,
799 		    &repo->packidx_bloom_filters, bf);
800 		bloom_free(bf->bloom);
801 		free(bf->bloom);
802 		free(bf);
803 	}
804 
805 	for (i = 0; i < repo->pack_cache_size; i++)
806 		if (repo->packs[i].path_packfile)
807 			if (repo->packs[i].path_packfile)
808 				got_pack_close(&repo->packs[i]);
809 
810 	free(repo->path);
811 	free(repo->path_git_dir);
812 
813 	got_object_cache_close(&repo->objcache);
814 	got_object_cache_close(&repo->treecache);
815 	got_object_cache_close(&repo->commitcache);
816 	got_object_cache_close(&repo->tagcache);
817 	got_object_cache_close(&repo->rawcache);
818 
819 	for (i = 0; i < nitems(repo->privsep_children); i++) {
820 		if (repo->privsep_children[i].imsg_fd == -1)
821 			continue;
822 		imsg_clear(repo->privsep_children[i].ibuf);
823 		free(repo->privsep_children[i].ibuf);
824 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
825 		child_err = got_privsep_wait_for_child(
826 		    repo->privsep_children[i].pid);
827 		if (child_err && err == NULL)
828 			err = child_err;
829 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
830 		    err == NULL)
831 			err = got_error_from_errno("close");
832 	}
833 
834 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
835 	    err == NULL)
836 		err = got_error_from_errno("close");
837 
838 	if (repo->gotconfig)
839 		got_gotconfig_free(repo->gotconfig);
840 	free(repo->gitconfig_author_name);
841 	free(repo->gitconfig_author_email);
842 	for (i = 0; i < repo->ngitconfig_remotes; i++)
843 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
844 	free(repo->gitconfig_remotes);
845 	for (i = 0; i < repo->nextensions; i++)
846 		free(repo->extensions[i]);
847 	free(repo->extensions);
848 
849 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
850 	free(repo);
851 
852 	return err;
853 }
854 
855 void
856 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
857 {
858 	int i;
859 
860 	free(repo->name);
861 	repo->name = NULL;
862 	free(repo->fetch_url);
863 	repo->fetch_url = NULL;
864 	free(repo->send_url);
865 	repo->send_url = NULL;
866 	for (i = 0; i < repo->nfetch_branches; i++)
867 		free(repo->fetch_branches[i]);
868 	free(repo->fetch_branches);
869 	repo->fetch_branches = NULL;
870 	repo->nfetch_branches = 0;
871 	for (i = 0; i < repo->nsend_branches; i++)
872 		free(repo->send_branches[i]);
873 	free(repo->send_branches);
874 	repo->send_branches = NULL;
875 	repo->nsend_branches = 0;
876 }
877 
878 const struct got_error *
879 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
880     const char *input_path)
881 {
882 	const struct got_error *err = NULL;
883 	const char *repo_abspath = NULL;
884 	size_t repolen, len;
885 	char *canonpath, *path = NULL;
886 
887 	*in_repo_path = NULL;
888 
889 	canonpath = strdup(input_path);
890 	if (canonpath == NULL) {
891 		err = got_error_from_errno("strdup");
892 		goto done;
893 	}
894 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
895 	if (err)
896 		goto done;
897 
898 	repo_abspath = got_repo_get_path(repo);
899 
900 	if (canonpath[0] == '\0') {
901 		path = strdup(canonpath);
902 		if (path == NULL) {
903 			err = got_error_from_errno("strdup");
904 			goto done;
905 		}
906 	} else {
907 		path = realpath(canonpath, NULL);
908 		if (path == NULL) {
909 			if (errno != ENOENT) {
910 				err = got_error_from_errno2("realpath",
911 				    canonpath);
912 				goto done;
913 			}
914 			/*
915 			 * Path is not on disk.
916 			 * Assume it is already relative to repository root.
917 			 */
918 			path = strdup(canonpath);
919 			if (path == NULL) {
920 				err = got_error_from_errno("strdup");
921 				goto done;
922 			}
923 		}
924 
925 		repolen = strlen(repo_abspath);
926 		len = strlen(path);
927 
928 
929 		if (strcmp(path, repo_abspath) == 0) {
930 			free(path);
931 			path = strdup("");
932 			if (path == NULL) {
933 				err = got_error_from_errno("strdup");
934 				goto done;
935 			}
936 		} else if (len > repolen &&
937 		    got_path_is_child(path, repo_abspath, repolen)) {
938 			/* Matched an on-disk path inside repository. */
939 			if (got_repo_is_bare(repo)) {
940 				/*
941 				 * Matched an on-disk path inside repository
942 				 * database. Treat input as repository-relative.
943 				 */
944 				free(path);
945 				path = canonpath;
946 				canonpath = NULL;
947 			} else {
948 				char *child;
949 				/* Strip common prefix with repository path. */
950 				err = got_path_skip_common_ancestor(&child,
951 				    repo_abspath, path);
952 				if (err)
953 					goto done;
954 				free(path);
955 				path = child;
956 			}
957 		} else {
958 			/*
959 			 * Matched unrelated on-disk path.
960 			 * Treat input as repository-relative.
961 			 */
962 			free(path);
963 			path = canonpath;
964 			canonpath = NULL;
965 		}
966 	}
967 
968 	/* Make in-repository path absolute */
969 	if (path[0] != '/') {
970 		char *abspath;
971 		if (asprintf(&abspath, "/%s", path) == -1) {
972 			err = got_error_from_errno("asprintf");
973 			goto done;
974 		}
975 		free(path);
976 		path = abspath;
977 	}
978 
979 done:
980 	free(canonpath);
981 	if (err)
982 		free(path);
983 	else
984 		*in_repo_path = path;
985 	return err;
986 }
987 
988 static const struct got_error *
989 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
990     const char *path_packidx)
991 {
992 	const struct got_error *err = NULL;
993 	size_t i;
994 
995 	for (i = 0; i < repo->pack_cache_size; i++) {
996 		if (repo->packidx_cache[i] == NULL)
997 			break;
998 		if (strcmp(repo->packidx_cache[i]->path_packidx,
999 		    path_packidx) == 0) {
1000 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1001 		}
1002 	}
1003 	if (i == repo->pack_cache_size) {
1004 		do {
1005 			i--;
1006 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1007 		    i == repo->pinned_packidx);
1008 		err = got_packidx_close(repo->packidx_cache[i]);
1009 		if (err)
1010 			return err;
1011 	}
1012 
1013 	repo->packidx_cache[i] = packidx;
1014 
1015 	return NULL;
1016 }
1017 
1018 int
1019 got_repo_is_packidx_filename(const char *name, size_t len)
1020 {
1021 	if (len != GOT_PACKIDX_NAMELEN)
1022 		return 0;
1023 
1024 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1025 		return 0;
1026 
1027 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1028 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1029 		return 0;
1030 
1031 	return 1;
1032 }
1033 
1034 static struct got_packidx_bloom_filter *
1035 get_packidx_bloom_filter(struct got_repository *repo,
1036     const char *path, size_t path_len)
1037 {
1038 	struct got_packidx_bloom_filter key;
1039 
1040 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1041 		return NULL; /* XXX */
1042 	key.path_len = path_len;
1043 
1044 	return RB_FIND(got_packidx_bloom_filter_tree,
1045 	    &repo->packidx_bloom_filters, &key);
1046 }
1047 
1048 int
1049 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1050     const char *path_packidx, struct got_object_id *id)
1051 {
1052 	struct got_packidx_bloom_filter *bf;
1053 
1054 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1055 	if (bf)
1056 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1057 
1058 	/* No bloom filter means this pack index must be searched. */
1059 	return 1;
1060 }
1061 
1062 static const struct got_error *
1063 add_packidx_bloom_filter(struct got_repository *repo,
1064     struct got_packidx *packidx, const char *path_packidx)
1065 {
1066 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1067 	struct got_packidx_bloom_filter *bf;
1068 	size_t len;
1069 
1070 	/*
1071 	 * Don't use bloom filters for very large pack index files.
1072 	 * Large pack files will contain a relatively large fraction
1073 	 * of our objects so we will likely need to visit them anyway.
1074 	 * The more objects a pack file contains the higher the probability
1075 	 * of a false-positive match from the bloom filter. And reading
1076 	 * all object IDs from a large pack index file can be expensive.
1077 	 */
1078 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1079 		return NULL;
1080 
1081 	/* Do we already have a filter for this pack index? */
1082 	if (get_packidx_bloom_filter(repo, path_packidx,
1083 	    strlen(path_packidx)) != NULL)
1084 		return NULL;
1085 
1086 	bf = calloc(1, sizeof(*bf));
1087 	if (bf == NULL)
1088 		return got_error_from_errno("calloc");
1089 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1090 	if (bf->bloom == NULL) {
1091 		free(bf);
1092 		return got_error_from_errno("calloc");
1093 	}
1094 
1095 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1096 	if (len >= sizeof(bf->path)) {
1097 		free(bf->bloom);
1098 		free(bf);
1099 		return got_error(GOT_ERR_NO_SPACE);
1100 	}
1101 	bf->path_len = len;
1102 
1103 	/* Minimum size supported by our bloom filter is 1000 entries. */
1104 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1105 	for (i = 0; i < nobjects; i++) {
1106 		struct got_packidx_object_id *id;
1107 		id = &packidx->hdr.sorted_ids[i];
1108 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1109 	}
1110 
1111 	RB_INSERT(got_packidx_bloom_filter_tree,
1112 	    &repo->packidx_bloom_filters, bf);
1113 	return NULL;
1114 }
1115 
1116 static void
1117 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1118 {
1119 	struct got_pathlist_entry *pe;
1120 
1121 	while (!TAILQ_EMPTY(packidx_paths)) {
1122 		pe = TAILQ_FIRST(packidx_paths);
1123 		TAILQ_REMOVE(packidx_paths, pe, entry);
1124 		free((char *)pe->path);
1125 		free(pe);
1126 	}
1127 }
1128 
1129 static const struct got_error *
1130 refresh_packidx_paths(struct got_repository *repo)
1131 {
1132 	const struct got_error *err = NULL;
1133 	char *objects_pack_dir = NULL;
1134 	struct stat sb;
1135 
1136 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1137 	if (objects_pack_dir == NULL)
1138 		return got_error_from_errno("got_repo_get_path_objects_pack");
1139 
1140 	if (stat(objects_pack_dir, &sb) == -1) {
1141 		if (errno != ENOENT) {
1142 			err = got_error_from_errno2("stat", objects_pack_dir);
1143 			goto done;
1144 		}
1145 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1146 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1147 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1148 		purge_packidx_paths(&repo->packidx_paths);
1149 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1150 		if (err)
1151 			goto done;
1152 	}
1153 done:
1154 	free(objects_pack_dir);
1155 	return err;
1156 }
1157 
1158 const struct got_error *
1159 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1160     struct got_repository *repo, struct got_object_id *id)
1161 {
1162 	const struct got_error *err;
1163 	struct got_pathlist_entry *pe;
1164 	size_t i;
1165 
1166 	/* Search pack index cache. */
1167 	for (i = 0; i < repo->pack_cache_size; i++) {
1168 		if (repo->packidx_cache[i] == NULL)
1169 			break;
1170 		if (!got_repo_check_packidx_bloom_filter(repo,
1171 		    repo->packidx_cache[i]->path_packidx, id))
1172 			continue; /* object will not be found in this index */
1173 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1174 		if (*idx != -1) {
1175 			*packidx = repo->packidx_cache[i];
1176 			/*
1177 			 * Move this cache entry to the front. Repeatedly
1178 			 * searching a wrong pack index can be expensive.
1179 			 */
1180 			if (i > 0) {
1181 				memmove(&repo->packidx_cache[1],
1182 				    &repo->packidx_cache[0],
1183 				    i * sizeof(repo->packidx_cache[0]));
1184 				repo->packidx_cache[0] = *packidx;
1185 				if (repo->pinned_packidx >= 0 &&
1186 				    repo->pinned_packidx < i)
1187 					repo->pinned_packidx++;
1188 				else if (repo->pinned_packidx == i)
1189 					repo->pinned_packidx = 0;
1190 			}
1191 			return NULL;
1192 		}
1193 	}
1194 	/* No luck. Search the filesystem. */
1195 
1196 	err = refresh_packidx_paths(repo);
1197 	if (err)
1198 		return err;
1199 
1200 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1201 		const char *path_packidx = pe->path;
1202 		int is_cached = 0;
1203 
1204 		if (!got_repo_check_packidx_bloom_filter(repo,
1205 		    pe->path, id))
1206 			continue; /* object will not be found in this index */
1207 
1208 		for (i = 0; i < repo->pack_cache_size; i++) {
1209 			if (repo->packidx_cache[i] == NULL)
1210 				break;
1211 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1212 			    path_packidx) == 0) {
1213 				is_cached = 1;
1214 				break;
1215 			}
1216 		}
1217 		if (is_cached)
1218 			continue; /* already searched */
1219 
1220 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1221 		    path_packidx, 0);
1222 		if (err)
1223 			goto done;
1224 
1225 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1226 		if (err)
1227 			goto done;
1228 
1229 		err = cache_packidx(repo, *packidx, path_packidx);
1230 		if (err)
1231 			goto done;
1232 
1233 		*idx = got_packidx_get_object_idx(*packidx, id);
1234 		if (*idx != -1) {
1235 			err = NULL; /* found the object */
1236 			goto done;
1237 		}
1238 	}
1239 
1240 	err = got_error_no_obj(id);
1241 done:
1242 	return err;
1243 }
1244 
1245 const struct got_error *
1246 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1247     struct got_repository *repo)
1248 {
1249 	const struct got_error *err = NULL;
1250 	DIR *packdir = NULL;
1251 	struct dirent *dent;
1252 	char *path_packidx = NULL;
1253 	int packdir_fd;
1254 	struct stat sb;
1255 
1256 	packdir_fd = openat(got_repo_get_fd(repo),
1257 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1258 	if (packdir_fd == -1) {
1259 		return got_error_from_errno_fmt("openat: %s/%s",
1260 		    got_repo_get_path_git_dir(repo),
1261 		    GOT_OBJECTS_PACK_DIR);
1262 	}
1263 
1264 	packdir = fdopendir(packdir_fd);
1265 	if (packdir == NULL) {
1266 		err = got_error_from_errno("fdopendir");
1267 		goto done;
1268 	}
1269 
1270 	if (fstat(packdir_fd, &sb) == -1) {
1271 		err = got_error_from_errno("fstat");
1272 		goto done;
1273 	}
1274 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1275 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1276 
1277 	while ((dent = readdir(packdir)) != NULL) {
1278 		if (!got_repo_is_packidx_filename(dent->d_name,
1279 		    strlen(dent->d_name)))
1280 			continue;
1281 
1282 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1283 		    dent->d_name) == -1) {
1284 			err = got_error_from_errno("asprintf");
1285 			path_packidx = NULL;
1286 			break;
1287 		}
1288 
1289 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1290 		if (err)
1291 			break;
1292 	}
1293 done:
1294 	if (err)
1295 		free(path_packidx);
1296 	if (packdir && closedir(packdir) != 0 && err == NULL)
1297 		err = got_error_from_errno("closedir");
1298 	return err;
1299 }
1300 
1301 const struct got_error *
1302 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1303     struct got_repository *repo)
1304 {
1305 	const struct got_error *err;
1306 	size_t i;
1307 
1308 	*packidx = NULL;
1309 
1310 	/* Search pack index cache. */
1311 	for (i = 0; i < repo->pack_cache_size; i++) {
1312 		if (repo->packidx_cache[i] == NULL)
1313 			break;
1314 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1315 		    path_packidx) == 0) {
1316 			*packidx = repo->packidx_cache[i];
1317 			return NULL;
1318 		}
1319 	}
1320 	/* No luck. Search the filesystem. */
1321 
1322 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1323 	    path_packidx, 0);
1324 	if (err)
1325 		return err;
1326 
1327 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1328 	if (err)
1329 		goto done;
1330 
1331 	err = cache_packidx(repo, *packidx, path_packidx);
1332 done:
1333 	if (err) {
1334 		got_packidx_close(*packidx);
1335 		*packidx = NULL;
1336 	}
1337 	return err;
1338 }
1339 
1340 static const struct got_error *
1341 read_packfile_hdr(int fd, struct got_packidx *packidx)
1342 {
1343 	const struct got_error *err = NULL;
1344 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1345 	struct got_packfile_hdr hdr;
1346 	ssize_t n;
1347 
1348 	n = read(fd, &hdr, sizeof(hdr));
1349 	if (n < 0)
1350 		return got_error_from_errno("read");
1351 	if (n != sizeof(hdr))
1352 		return got_error(GOT_ERR_BAD_PACKFILE);
1353 
1354 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1355 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1356 	    be32toh(hdr.nobjects) != totobj)
1357 		err = got_error(GOT_ERR_BAD_PACKFILE);
1358 
1359 	return err;
1360 }
1361 
1362 static const struct got_error *
1363 open_packfile(int *fd, struct got_repository *repo,
1364     const char *relpath, struct got_packidx *packidx)
1365 {
1366 	const struct got_error *err = NULL;
1367 
1368 	*fd = openat(got_repo_get_fd(repo), relpath,
1369 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1370 	if (*fd == -1)
1371 		return got_error_from_errno_fmt("openat: %s/%s",
1372 		    got_repo_get_path_git_dir(repo), relpath);
1373 
1374 	if (packidx) {
1375 		err = read_packfile_hdr(*fd, packidx);
1376 		if (err) {
1377 			close(*fd);
1378 			*fd = -1;
1379 		}
1380 	}
1381 
1382 	return err;
1383 }
1384 
1385 const struct got_error *
1386 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1387     const char *path_packfile, struct got_packidx *packidx)
1388 {
1389 	const struct got_error *err = NULL;
1390 	struct got_pack *pack = NULL;
1391 	struct stat sb;
1392 	size_t i;
1393 
1394 	if (packp)
1395 		*packp = NULL;
1396 
1397 	for (i = 0; i < repo->pack_cache_size; i++) {
1398 		pack = &repo->packs[i];
1399 		if (pack->path_packfile == NULL)
1400 			break;
1401 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1402 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1403 	}
1404 
1405 	if (i == repo->pack_cache_size) {
1406 		struct got_pack tmp;
1407 		do {
1408 			i--;
1409 		} while (i > 0 && repo->pinned_pack >= 0 &&
1410 		    i == repo->pinned_pack);
1411 		err = got_pack_close(&repo->packs[i]);
1412 		if (err)
1413 			return err;
1414 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1415 			return got_error_from_errno("ftruncate");
1416 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1417 			return got_error_from_errno("ftruncate");
1418 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1419 		memcpy(&repo->packs[i], &repo->packs[0],
1420 		    sizeof(repo->packs[i]));
1421 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1422 		if (repo->pinned_pack == 0)
1423 			repo->pinned_pack = i;
1424 		else if (repo->pinned_pack == i)
1425 			repo->pinned_pack = 0;
1426 		i = 0;
1427 	}
1428 
1429 	pack = &repo->packs[i];
1430 
1431 	pack->path_packfile = strdup(path_packfile);
1432 	if (pack->path_packfile == NULL) {
1433 		err = got_error_from_errno("strdup");
1434 		goto done;
1435 	}
1436 
1437 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1438 	if (err)
1439 		goto done;
1440 
1441 	if (fstat(pack->fd, &sb) != 0) {
1442 		err = got_error_from_errno("fstat");
1443 		goto done;
1444 	}
1445 	pack->filesize = sb.st_size;
1446 
1447 	pack->privsep_child = NULL;
1448 
1449 	err = got_delta_cache_alloc(&pack->delta_cache);
1450 	if (err)
1451 		goto done;
1452 
1453 #ifndef GOT_PACK_NO_MMAP
1454 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1455 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1456 		    pack->fd, 0);
1457 		if (pack->map == MAP_FAILED) {
1458 			if (errno != ENOMEM) {
1459 				err = got_error_from_errno("mmap");
1460 				goto done;
1461 			}
1462 			pack->map = NULL; /* fall back to read(2) */
1463 		}
1464 	}
1465 #endif
1466 done:
1467 	if (err) {
1468 		if (pack)
1469 			got_pack_close(pack);
1470 	} else if (packp)
1471 		*packp = pack;
1472 	return err;
1473 }
1474 
1475 struct got_pack *
1476 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1477 {
1478 	struct got_pack *pack = NULL;
1479 	size_t i;
1480 
1481 	for (i = 0; i < repo->pack_cache_size; i++) {
1482 		pack = &repo->packs[i];
1483 		if (pack->path_packfile == NULL)
1484 			break;
1485 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1486 			return pack;
1487 	}
1488 
1489 	return NULL;
1490 }
1491 
1492 const struct got_error *
1493 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1494     struct got_pack *pack)
1495 {
1496 	size_t i;
1497 	int pinned_pack = -1, pinned_packidx = -1;
1498 
1499 	for (i = 0; i < repo->pack_cache_size; i++) {
1500 		if (repo->packidx_cache[i] &&
1501 		    strcmp(repo->packidx_cache[i]->path_packidx,
1502 		    packidx->path_packidx) == 0)
1503 			pinned_packidx = i;
1504 		if (repo->packs[i].path_packfile &&
1505 		    strcmp(repo->packs[i].path_packfile,
1506 		    pack->path_packfile) == 0)
1507 			pinned_pack = i;
1508 	}
1509 
1510 	if (pinned_packidx == -1 || pinned_pack == -1)
1511 		return got_error(GOT_ERR_PIN_PACK);
1512 
1513 	repo->pinned_pack = pinned_pack;
1514 	repo->pinned_packidx = pinned_packidx;
1515 	if (repo->packs[pinned_pack].privsep_child)
1516 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1517 	return NULL;
1518 }
1519 
1520 struct got_pack *
1521 got_repo_get_pinned_pack(struct got_repository *repo)
1522 {
1523 	if (repo->pinned_pack >= 0 &&
1524 	    repo->pinned_pack < repo->pack_cache_size)
1525 		return &repo->packs[repo->pinned_pack];
1526 
1527 	return NULL;
1528 }
1529 
1530 void
1531 got_repo_unpin_pack(struct got_repository *repo)
1532 {
1533 	repo->pinned_packidx = -1;
1534 	repo->pinned_pack = -1;
1535 	repo->pinned_pid = 0;
1536 }
1537 
1538 const struct got_error *
1539 got_repo_init(const char *repo_path, const char *head_name)
1540 {
1541 	const struct got_error *err = NULL;
1542 	const char *dirnames[] = {
1543 		GOT_OBJECTS_DIR,
1544 		GOT_OBJECTS_PACK_DIR,
1545 		GOT_REFS_DIR,
1546 	};
1547 	const char *description_str = "Unnamed repository; "
1548 	    "edit this file 'description' to name the repository.";
1549 	const char *headref = "ref: refs/heads/";
1550 	const char *gitconfig_str = "[core]\n"
1551 	    "\trepositoryformatversion = 0\n"
1552 	    "\tfilemode = true\n"
1553 	    "\tbare = true\n";
1554 	char *headref_str, *path;
1555 	size_t i;
1556 
1557 	if (!got_path_dir_is_empty(repo_path))
1558 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1559 
1560 	for (i = 0; i < nitems(dirnames); i++) {
1561 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1562 			return got_error_from_errno("asprintf");
1563 		}
1564 		err = got_path_mkdir(path);
1565 		free(path);
1566 		if (err)
1567 			return err;
1568 	}
1569 
1570 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1571 		return got_error_from_errno("asprintf");
1572 	err = got_path_create_file(path, description_str);
1573 	free(path);
1574 	if (err)
1575 		return err;
1576 
1577 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1578 		return got_error_from_errno("asprintf");
1579 	if (asprintf(&headref_str, "%s%s", headref,
1580 	    head_name ? head_name : "main") == -1) {
1581 		free(path);
1582 		return got_error_from_errno("asprintf");
1583 	}
1584 	err = got_path_create_file(path, headref_str);
1585 	free(headref_str);
1586 	free(path);
1587 	if (err)
1588 		return err;
1589 
1590 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1591 		return got_error_from_errno("asprintf");
1592 	err = got_path_create_file(path, gitconfig_str);
1593 	free(path);
1594 	if (err)
1595 		return err;
1596 
1597 	return NULL;
1598 }
1599 
1600 static const struct got_error *
1601 match_packed_object(struct got_object_id **unique_id,
1602     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1603 {
1604 	const struct got_error *err = NULL;
1605 	struct got_object_id_queue matched_ids;
1606 	struct got_pathlist_entry *pe;
1607 
1608 	STAILQ_INIT(&matched_ids);
1609 
1610 	err = refresh_packidx_paths(repo);
1611 	if (err)
1612 		return err;
1613 
1614 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1615 		const char *path_packidx = pe->path;
1616 		struct got_packidx *packidx;
1617 		struct got_object_qid *qid;
1618 
1619 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1620 		    path_packidx, 0);
1621 		if (err)
1622 			break;
1623 
1624 		err = got_packidx_match_id_str_prefix(&matched_ids,
1625 		    packidx, id_str_prefix);
1626 		if (err) {
1627 			got_packidx_close(packidx);
1628 			break;
1629 		}
1630 		err = got_packidx_close(packidx);
1631 		if (err)
1632 			break;
1633 
1634 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1635 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1636 				int matched_type;
1637 				err = got_object_get_type(&matched_type, repo,
1638 				    &qid->id);
1639 				if (err)
1640 					goto done;
1641 				if (matched_type != obj_type)
1642 					continue;
1643 			}
1644 			if (*unique_id == NULL) {
1645 				*unique_id = got_object_id_dup(&qid->id);
1646 				if (*unique_id == NULL) {
1647 					err = got_error_from_errno("malloc");
1648 					goto done;
1649 				}
1650 			} else {
1651 				if (got_object_id_cmp(*unique_id,
1652 				    &qid->id) == 0)
1653 					continue; /* packed multiple times */
1654 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1655 				goto done;
1656 			}
1657 		}
1658 	}
1659 done:
1660 	got_object_id_queue_free(&matched_ids);
1661 	if (err) {
1662 		free(*unique_id);
1663 		*unique_id = NULL;
1664 	}
1665 	return err;
1666 }
1667 
1668 static const struct got_error *
1669 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1670     const char *object_dir, const char *id_str_prefix, int obj_type,
1671     struct got_repository *repo)
1672 {
1673 	const struct got_error *err = NULL;
1674 	char *path, *id_str = NULL;
1675 	DIR *dir = NULL;
1676 	struct dirent *dent;
1677 	struct got_object_id id;
1678 
1679 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1680 		err = got_error_from_errno("asprintf");
1681 		goto done;
1682 	}
1683 
1684 	dir = opendir(path);
1685 	if (dir == NULL) {
1686 		if (errno == ENOENT) {
1687 			err = NULL;
1688 			goto done;
1689 		}
1690 		err = got_error_from_errno2("opendir", path);
1691 		goto done;
1692 	}
1693 	while ((dent = readdir(dir)) != NULL) {
1694 		int cmp;
1695 
1696 		free(id_str);
1697 		id_str = NULL;
1698 
1699 		if (strcmp(dent->d_name, ".") == 0 ||
1700 		    strcmp(dent->d_name, "..") == 0)
1701 			continue;
1702 
1703 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1704 			err = got_error_from_errno("asprintf");
1705 			goto done;
1706 		}
1707 
1708 		if (!got_parse_sha1_digest(id.sha1, id_str))
1709 			continue;
1710 
1711 		/*
1712 		 * Directory entries do not necessarily appear in
1713 		 * sorted order, so we must iterate over all of them.
1714 		 */
1715 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1716 		if (cmp != 0)
1717 			continue;
1718 
1719 		if (*unique_id == NULL) {
1720 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1721 				int matched_type;
1722 				err = got_object_get_type(&matched_type, repo,
1723 				    &id);
1724 				if (err)
1725 					goto done;
1726 				if (matched_type != obj_type)
1727 					continue;
1728 			}
1729 			*unique_id = got_object_id_dup(&id);
1730 			if (*unique_id == NULL) {
1731 				err = got_error_from_errno("got_object_id_dup");
1732 				goto done;
1733 			}
1734 		} else {
1735 			if (got_object_id_cmp(*unique_id, &id) == 0)
1736 				continue; /* both packed and loose */
1737 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1738 			goto done;
1739 		}
1740 	}
1741 done:
1742 	if (dir && closedir(dir) != 0 && err == NULL)
1743 		err = got_error_from_errno("closedir");
1744 	if (err) {
1745 		free(*unique_id);
1746 		*unique_id = NULL;
1747 	}
1748 	free(id_str);
1749 	free(path);
1750 	return err;
1751 }
1752 
1753 const struct got_error *
1754 got_repo_match_object_id_prefix(struct got_object_id **id,
1755     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1756 {
1757 	const struct got_error *err = NULL;
1758 	char *path_objects = NULL, *object_dir = NULL;
1759 	size_t len;
1760 	int i;
1761 
1762 	*id = NULL;
1763 
1764 	path_objects = got_repo_get_path_objects(repo);
1765 
1766 	len = strlen(id_str_prefix);
1767 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1768 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1769 		goto done;
1770 	}
1771 
1772 	for (i = 0; i < len; i++) {
1773 		if (isxdigit((unsigned char)id_str_prefix[i]))
1774 			continue;
1775 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1776 		goto done;
1777 	}
1778 
1779 	if (len >= 2) {
1780 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1781 		if (err)
1782 			goto done;
1783 		object_dir = strndup(id_str_prefix, 2);
1784 		if (object_dir == NULL) {
1785 			err = got_error_from_errno("strdup");
1786 			goto done;
1787 		}
1788 		err = match_loose_object(id, path_objects, object_dir,
1789 		    id_str_prefix, obj_type, repo);
1790 	} else if (len == 1) {
1791 		int i;
1792 		for (i = 0; i < 0xf; i++) {
1793 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1794 			    == -1) {
1795 				err = got_error_from_errno("asprintf");
1796 				goto done;
1797 			}
1798 			err = match_packed_object(id, repo, object_dir,
1799 			    obj_type);
1800 			if (err)
1801 				goto done;
1802 			err = match_loose_object(id, path_objects, object_dir,
1803 			    id_str_prefix, obj_type, repo);
1804 			if (err)
1805 				goto done;
1806 		}
1807 	} else {
1808 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1809 		goto done;
1810 	}
1811 done:
1812 	free(path_objects);
1813 	free(object_dir);
1814 	if (err) {
1815 		free(*id);
1816 		*id = NULL;
1817 	} else if (*id == NULL) {
1818 		switch (obj_type) {
1819 		case GOT_OBJ_TYPE_BLOB:
1820 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1821 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1822 			break;
1823 		case GOT_OBJ_TYPE_TREE:
1824 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1825 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1826 			break;
1827 		case GOT_OBJ_TYPE_COMMIT:
1828 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1829 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1830 			break;
1831 		case GOT_OBJ_TYPE_TAG:
1832 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1833 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1834 			break;
1835 		default:
1836 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1837 			break;
1838 		}
1839 	}
1840 
1841 	return err;
1842 }
1843 
1844 const struct got_error *
1845 got_repo_match_object_id(struct got_object_id **id, char **label,
1846     const char *id_str, int obj_type, struct got_reflist_head *refs,
1847     struct got_repository *repo)
1848 {
1849 	const struct got_error *err;
1850 	struct got_tag_object *tag;
1851 	struct got_reference *ref = NULL;
1852 
1853 	*id = NULL;
1854 	if (label)
1855 		*label = NULL;
1856 
1857 	if (refs) {
1858 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1859 		    refs, repo);
1860 		if (err == NULL) {
1861 			*id = got_object_id_dup(
1862 			    got_object_tag_get_object_id(tag));
1863 			if (*id == NULL)
1864 				err = got_error_from_errno("got_object_id_dup");
1865 			else if (label && asprintf(label, "refs/tags/%s",
1866 			    got_object_tag_get_name(tag)) == -1) {
1867 				err = got_error_from_errno("asprintf");
1868 				free(*id);
1869 				*id = NULL;
1870 			}
1871 			got_object_tag_close(tag);
1872 			return err;
1873 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1874 		    err->code != GOT_ERR_NO_OBJ)
1875 			return err;
1876 	}
1877 
1878 	err = got_ref_open(&ref, repo, id_str, 0);
1879 	if (err == NULL) {
1880 		err = got_ref_resolve(id, repo, ref);
1881 		if (err)
1882 			goto done;
1883 		if (label) {
1884 			*label = strdup(got_ref_get_name(ref));
1885 			if (*label == NULL) {
1886 				err = got_error_from_errno("strdup");
1887 				goto done;
1888 			}
1889 		}
1890 	} else {
1891 		if (err->code != GOT_ERR_NOT_REF &&
1892 		    err->code != GOT_ERR_BAD_REF_NAME)
1893 			goto done;
1894 		err = got_repo_match_object_id_prefix(id, id_str,
1895 		    obj_type, repo);
1896 		if (err) {
1897 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1898 				err = got_error_not_ref(id_str);
1899 			goto done;
1900 		}
1901 		if (label) {
1902 			err = got_object_id_str(label, *id);
1903 			if (*label == NULL) {
1904 				err = got_error_from_errno("strdup");
1905 				goto done;
1906 			}
1907 		}
1908 	}
1909 done:
1910 	if (ref)
1911 		got_ref_close(ref);
1912 	return err;
1913 }
1914 
1915 const struct got_error *
1916 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1917     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1918 {
1919 	const struct got_error *err = NULL;
1920 	struct got_reflist_entry *re;
1921 	struct got_object_id *tag_id;
1922 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1923 
1924 	*tag = NULL;
1925 
1926 	TAILQ_FOREACH(re, refs, entry) {
1927 		const char *refname;
1928 		refname = got_ref_get_name(re->ref);
1929 		if (got_ref_is_symbolic(re->ref))
1930 			continue;
1931 		if (strncmp(refname, "refs/tags/", 10) != 0)
1932 			continue;
1933 		if (!name_is_absolute)
1934 			refname += strlen("refs/tags/");
1935 		if (strcmp(refname, name) != 0)
1936 			continue;
1937 		err = got_ref_resolve(&tag_id, repo, re->ref);
1938 		if (err)
1939 			break;
1940 		err = got_object_open_as_tag(tag, repo, tag_id);
1941 		free(tag_id);
1942 		if (err)
1943 			break;
1944 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1945 		    got_object_tag_get_object_type(*tag) == obj_type)
1946 			break;
1947 		got_object_tag_close(*tag);
1948 		*tag = NULL;
1949 	}
1950 
1951 	if (err == NULL && *tag == NULL)
1952 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1953 		    GOT_OBJ_LABEL_TAG, name);
1954 	return err;
1955 }
1956 
1957 static const struct got_error *
1958 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1959     const char *name, mode_t mode, struct got_object_id *blob_id)
1960 {
1961 	const struct got_error *err = NULL;
1962 
1963 	 *new_te = NULL;
1964 
1965 	*new_te = calloc(1, sizeof(**new_te));
1966 	if (*new_te == NULL)
1967 		return got_error_from_errno("calloc");
1968 
1969 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1970 	    sizeof((*new_te)->name)) {
1971 		err = got_error(GOT_ERR_NO_SPACE);
1972 		goto done;
1973 	}
1974 
1975 	if (S_ISLNK(mode)) {
1976 		(*new_te)->mode = S_IFLNK;
1977 	} else {
1978 		(*new_te)->mode = S_IFREG;
1979 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1980 	}
1981 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1982 done:
1983 	if (err && *new_te) {
1984 		free(*new_te);
1985 		*new_te = NULL;
1986 	}
1987 	return err;
1988 }
1989 
1990 static const struct got_error *
1991 import_file(struct got_tree_entry **new_te, struct dirent *de,
1992     const char *path, struct got_repository *repo)
1993 {
1994 	const struct got_error *err;
1995 	struct got_object_id *blob_id = NULL;
1996 	char *filepath;
1997 	struct stat sb;
1998 
1999 	if (asprintf(&filepath, "%s%s%s", path,
2000 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2001 		return got_error_from_errno("asprintf");
2002 
2003 	if (lstat(filepath, &sb) != 0) {
2004 		err = got_error_from_errno2("lstat", path);
2005 		goto done;
2006 	}
2007 
2008 	err = got_object_blob_create(&blob_id, filepath, repo);
2009 	if (err)
2010 		goto done;
2011 
2012 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2013 	    blob_id);
2014 done:
2015 	free(filepath);
2016 	if (err)
2017 		free(blob_id);
2018 	return err;
2019 }
2020 
2021 static const struct got_error *
2022 insert_tree_entry(struct got_tree_entry *new_te,
2023     struct got_pathlist_head *paths)
2024 {
2025 	const struct got_error *err = NULL;
2026 	struct got_pathlist_entry *new_pe;
2027 
2028 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2029 	if (err)
2030 		return err;
2031 	if (new_pe == NULL)
2032 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2033 	return NULL;
2034 }
2035 
2036 static const struct got_error *write_tree(struct got_object_id **,
2037     const char *, struct got_pathlist_head *, struct got_repository *,
2038     got_repo_import_cb progress_cb, void *progress_arg);
2039 
2040 static const struct got_error *
2041 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2042     const char *path, struct got_pathlist_head *ignores,
2043     struct got_repository *repo,
2044     got_repo_import_cb progress_cb, void *progress_arg)
2045 {
2046 	const struct got_error *err;
2047 	struct got_object_id *id = NULL;
2048 	char *subdirpath;
2049 
2050 	if (asprintf(&subdirpath, "%s%s%s", path,
2051 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2052 		return got_error_from_errno("asprintf");
2053 
2054 	(*new_te) = calloc(1, sizeof(**new_te));
2055 	if (*new_te == NULL)
2056 		return got_error_from_errno("calloc");
2057 	(*new_te)->mode = S_IFDIR;
2058 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2059 	    sizeof((*new_te)->name)) {
2060 		err = got_error(GOT_ERR_NO_SPACE);
2061 		goto done;
2062 	}
2063 	err = write_tree(&id, subdirpath, ignores,  repo,
2064 	    progress_cb, progress_arg);
2065 	if (err)
2066 		goto done;
2067 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2068 
2069 done:
2070 	free(id);
2071 	free(subdirpath);
2072 	if (err) {
2073 		free(*new_te);
2074 		*new_te = NULL;
2075 	}
2076 	return err;
2077 }
2078 
2079 static const struct got_error *
2080 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2081     struct got_pathlist_head *ignores, struct got_repository *repo,
2082     got_repo_import_cb progress_cb, void *progress_arg)
2083 {
2084 	const struct got_error *err = NULL;
2085 	DIR *dir;
2086 	struct dirent *de;
2087 	int nentries;
2088 	struct got_tree_entry *new_te = NULL;
2089 	struct got_pathlist_head paths;
2090 	struct got_pathlist_entry *pe;
2091 
2092 	*new_tree_id = NULL;
2093 
2094 	TAILQ_INIT(&paths);
2095 
2096 	dir = opendir(path_dir);
2097 	if (dir == NULL) {
2098 		err = got_error_from_errno2("opendir", path_dir);
2099 		goto done;
2100 	}
2101 
2102 	nentries = 0;
2103 	while ((de = readdir(dir)) != NULL) {
2104 		int ignore = 0;
2105 		int type;
2106 
2107 		if (strcmp(de->d_name, ".") == 0 ||
2108 		    strcmp(de->d_name, "..") == 0)
2109 			continue;
2110 
2111 		TAILQ_FOREACH(pe, ignores, entry) {
2112 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2113 				ignore = 1;
2114 				break;
2115 			}
2116 		}
2117 		if (ignore)
2118 			continue;
2119 
2120 		err = got_path_dirent_type(&type, path_dir, de);
2121 		if (err)
2122 			goto done;
2123 
2124 		if (type == DT_DIR) {
2125 			err = import_subdir(&new_te, de, path_dir,
2126 			    ignores, repo, progress_cb, progress_arg);
2127 			if (err) {
2128 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2129 					goto done;
2130 				err = NULL;
2131 				continue;
2132 			}
2133 		} else if (type == DT_REG || type == DT_LNK) {
2134 			err = import_file(&new_te, de, path_dir, repo);
2135 			if (err)
2136 				goto done;
2137 		} else
2138 			continue;
2139 
2140 		err = insert_tree_entry(new_te, &paths);
2141 		if (err)
2142 			goto done;
2143 		nentries++;
2144 	}
2145 
2146 	if (TAILQ_EMPTY(&paths)) {
2147 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2148 		    "cannot create tree without any entries");
2149 		goto done;
2150 	}
2151 
2152 	TAILQ_FOREACH(pe, &paths, entry) {
2153 		struct got_tree_entry *te = pe->data;
2154 		char *path;
2155 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2156 			continue;
2157 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2158 			err = got_error_from_errno("asprintf");
2159 			goto done;
2160 		}
2161 		err = (*progress_cb)(progress_arg, path);
2162 		free(path);
2163 		if (err)
2164 			goto done;
2165 	}
2166 
2167 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2168 done:
2169 	if (dir)
2170 		closedir(dir);
2171 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2172 	return err;
2173 }
2174 
2175 const struct got_error *
2176 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2177     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2178     struct got_repository *repo, got_repo_import_cb progress_cb,
2179     void *progress_arg)
2180 {
2181 	const struct got_error *err;
2182 	struct got_object_id *new_tree_id;
2183 
2184 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2185 	    progress_cb, progress_arg);
2186 	if (err)
2187 		return err;
2188 
2189 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2190 	    author, time(NULL), author, time(NULL), logmsg, repo);
2191 	free(new_tree_id);
2192 	return err;
2193 }
2194 
2195 const struct got_error *
2196 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2197     struct got_repository *repo)
2198 {
2199 	const struct got_error *err = NULL;
2200 	char *path_objects = NULL, *path = NULL;
2201 	DIR *dir = NULL;
2202 	struct got_object_id id;
2203 	int i;
2204 
2205 	*nobjects = 0;
2206 	*ondisk_size = 0;
2207 
2208 	path_objects = got_repo_get_path_objects(repo);
2209 	if (path_objects == NULL)
2210 		return got_error_from_errno("got_repo_get_path_objects");
2211 
2212 	for (i = 0; i <= 0xff; i++) {
2213 		struct dirent *dent;
2214 
2215 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2216 			err = got_error_from_errno("asprintf");
2217 			break;
2218 		}
2219 
2220 		dir = opendir(path);
2221 		if (dir == NULL) {
2222 			if (errno == ENOENT) {
2223 				err = NULL;
2224 				continue;
2225 			}
2226 			err = got_error_from_errno2("opendir", path);
2227 			break;
2228 		}
2229 
2230 		while ((dent = readdir(dir)) != NULL) {
2231 			char *id_str;
2232 			int fd;
2233 			struct stat sb;
2234 
2235 			if (strcmp(dent->d_name, ".") == 0 ||
2236 			    strcmp(dent->d_name, "..") == 0)
2237 				continue;
2238 
2239 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2240 				err = got_error_from_errno("asprintf");
2241 				goto done;
2242 			}
2243 
2244 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2245 				free(id_str);
2246 				continue;
2247 			}
2248 			free(id_str);
2249 
2250 			err = got_object_open_loose_fd(&fd, &id, repo);
2251 			if (err)
2252 				goto done;
2253 
2254 			if (fstat(fd, &sb) == -1) {
2255 				err = got_error_from_errno("fstat");
2256 				close(fd);
2257 				goto done;
2258 			}
2259 			(*nobjects)++;
2260 			(*ondisk_size) += sb.st_size;
2261 
2262 			if (close(fd) == -1) {
2263 				err = got_error_from_errno("close");
2264 				goto done;
2265 			}
2266 		}
2267 
2268 		if (closedir(dir) != 0) {
2269 			err = got_error_from_errno("closedir");
2270 			goto done;
2271 		}
2272 		dir = NULL;
2273 
2274 		free(path);
2275 		path = NULL;
2276 	}
2277 done:
2278 	if (dir && closedir(dir) != 0 && err == NULL)
2279 		err = got_error_from_errno("closedir");
2280 
2281 	if (err) {
2282 		*nobjects = 0;
2283 		*ondisk_size = 0;
2284 	}
2285 	free(path_objects);
2286 	free(path);
2287 	return err;
2288 }
2289 
2290 const struct got_error *
2291 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2292     off_t *total_packsize, struct got_repository *repo)
2293 {
2294 	const struct got_error *err = NULL;
2295 	DIR *packdir = NULL;
2296 	struct dirent *dent;
2297 	struct got_packidx *packidx = NULL;
2298 	char *path_packidx;
2299 	char *path_packfile;
2300 	int packdir_fd;
2301 	struct stat sb;
2302 
2303 	*npackfiles = 0;
2304 	*nobjects = 0;
2305 	*total_packsize = 0;
2306 
2307 	packdir_fd = openat(got_repo_get_fd(repo),
2308 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2309 	if (packdir_fd == -1) {
2310 		return got_error_from_errno_fmt("openat: %s/%s",
2311 		    got_repo_get_path_git_dir(repo),
2312 		    GOT_OBJECTS_PACK_DIR);
2313 	}
2314 
2315 	packdir = fdopendir(packdir_fd);
2316 	if (packdir == NULL) {
2317 		err = got_error_from_errno("fdopendir");
2318 		goto done;
2319 	}
2320 
2321 	while ((dent = readdir(packdir)) != NULL) {
2322 		if (!got_repo_is_packidx_filename(dent->d_name,
2323 		    strlen(dent->d_name)))
2324 			continue;
2325 
2326 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2327 		    dent->d_name) == -1) {
2328 			err = got_error_from_errno("asprintf");
2329 			goto done;
2330 		}
2331 
2332 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2333 		    path_packidx, 0);
2334 		free(path_packidx);
2335 		if (err)
2336 			goto done;
2337 
2338 		if (fstat(packidx->fd, &sb) == -1)
2339 			goto done;
2340 		*total_packsize += sb.st_size;
2341 
2342 		err = got_packidx_get_packfile_path(&path_packfile,
2343 		    packidx->path_packidx);
2344 		if (err)
2345 			goto done;
2346 
2347 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2348 		    0) == -1) {
2349 			free(path_packfile);
2350 			goto done;
2351 		}
2352 		free(path_packfile);
2353 		*total_packsize += sb.st_size;
2354 
2355 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2356 
2357 		(*npackfiles)++;
2358 
2359 		got_packidx_close(packidx);
2360 		packidx = NULL;
2361 	}
2362 done:
2363 	if (packidx)
2364 		got_packidx_close(packidx);
2365 	if (packdir && closedir(packdir) != 0 && err == NULL)
2366 		err = got_error_from_errno("closedir");
2367 	if (err) {
2368 		*npackfiles = 0;
2369 		*nobjects = 0;
2370 		*total_packsize = 0;
2371 	}
2372 	return err;
2373 }
2374 
2375 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2376     got_packidx_bloom_filter_cmp);