1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_delta_cache.h"
57 #include "got_lib_inflate.h"
58 #include "got_lib_object.h"
59 #include "got_lib_object_parse.h"
60 #include "got_lib_object_create.h"
61 #include "got_lib_pack.h"
62 #include "got_lib_privsep.h"
63 #include "got_lib_sha1.h"
64 #include "got_lib_object_cache.h"
65 #include "got_lib_repository.h"
66 #include "got_lib_gotconfig.h"
67 
68 #ifndef nitems
69 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
70 #endif
71 
72 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
73 
74 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
75     got_packidx_bloom_filter_cmp);
76 
77 const char *
78 got_repo_get_path(struct got_repository *repo)
79 {
80 	return repo->path;
81 }
82 
83 const char *
84 got_repo_get_path_git_dir(struct got_repository *repo)
85 {
86 	return repo->path_git_dir;
87 }
88 
89 int
90 got_repo_get_fd(struct got_repository *repo)
91 {
92 	return repo->gitdir_fd;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_name(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_name;
99 }
100 
101 const char *
102 got_repo_get_gitconfig_author_email(struct got_repository *repo)
103 {
104 	return repo->gitconfig_author_email;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_name;
111 }
112 
113 const char *
114 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
115 {
116 	return repo->global_gitconfig_author_email;
117 }
118 
119 const char *
120 got_repo_get_gitconfig_owner(struct got_repository *repo)
121 {
122 	return repo->gitconfig_owner;
123 }
124 
125 void
126 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
127     struct got_repository *repo)
128 {
129 	*extensions = repo->extensions;
130 	*nextensions = repo->nextensions;
131 }
132 
133 int
134 got_repo_is_bare(struct got_repository *repo)
135 {
136 	return (strcmp(repo->path, repo->path_git_dir) == 0);
137 }
138 
139 static char *
140 get_path_git_child(struct got_repository *repo, const char *basename)
141 {
142 	char *path_child;
143 
144 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
145 	    basename) == -1)
146 		return NULL;
147 
148 	return path_child;
149 }
150 
151 char *
152 got_repo_get_path_objects(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
155 }
156 
157 char *
158 got_repo_get_path_objects_pack(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
161 }
162 
163 char *
164 got_repo_get_path_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_REFS_DIR);
167 }
168 
169 char *
170 got_repo_get_path_packed_refs(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
173 }
174 
175 static char *
176 get_path_head(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_HEAD_FILE);
179 }
180 
181 char *
182 got_repo_get_path_gitconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GITCONFIG);
185 }
186 
187 char *
188 got_repo_get_path_gotconfig(struct got_repository *repo)
189 {
190 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
191 }
192 
193 const struct got_gotconfig *
194 got_repo_get_gotconfig(struct got_repository *repo)
195 {
196 	return repo->gotconfig;
197 }
198 
199 void
200 got_repo_get_gitconfig_remotes(int *nremotes,
201     const struct got_remote_repo **remotes, struct got_repository *repo)
202 {
203 	*nremotes = repo->ngitconfig_remotes;
204 	*remotes = repo->gitconfig_remotes;
205 }
206 
207 static int
208 is_git_repo(struct got_repository *repo)
209 {
210 	const char *path_git = got_repo_get_path_git_dir(repo);
211 	char *path_objects = got_repo_get_path_objects(repo);
212 	char *path_refs = got_repo_get_path_refs(repo);
213 	char *path_head = get_path_head(repo);
214 	int ret = 0;
215 	struct stat sb;
216 	struct got_reference *head_ref;
217 
218 	if (lstat(path_git, &sb) == -1)
219 		goto done;
220 	if (!S_ISDIR(sb.st_mode))
221 		goto done;
222 
223 	if (lstat(path_objects, &sb) == -1)
224 		goto done;
225 	if (!S_ISDIR(sb.st_mode))
226 		goto done;
227 
228 	if (lstat(path_refs, &sb) == -1)
229 		goto done;
230 	if (!S_ISDIR(sb.st_mode))
231 		goto done;
232 
233 	if (lstat(path_head, &sb) == -1)
234 		goto done;
235 	if (!S_ISREG(sb.st_mode))
236 		goto done;
237 
238 	/* Check if the HEAD reference can be opened. */
239 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
240 		goto done;
241 	got_ref_close(head_ref);
242 
243 	ret = 1;
244 done:
245 	free(path_objects);
246 	free(path_refs);
247 	free(path_head);
248 	return ret;
249 
250 }
251 
252 static const struct got_error *
253 close_tempfiles(int *fds, size_t nfds)
254 {
255 	const struct got_error *err = NULL;
256 	int i;
257 
258 	for (i = 0; i < nfds; i++) {
259 		if (fds[i] == -1)
260 			continue;
261 		if (close(fds[i]) == -1) {
262 			err = got_error_from_errno("close");
263 			break;
264 		}
265 	}
266 	free(fds);
267 	return err;
268 }
269 
270 static const struct got_error *
271 open_tempfiles(int **fds, size_t array_size, size_t nfds)
272 {
273 	const struct got_error *err = NULL;
274 	int i;
275 
276 	*fds = calloc(array_size, sizeof(**fds));
277 	if (*fds == NULL)
278 		return got_error_from_errno("calloc");
279 
280 	for (i = 0; i < array_size; i++)
281 		(*fds)[i] = -1;
282 
283 	for (i = 0; i < nfds; i++) {
284 		(*fds)[i] = got_opentempfd();
285 		if ((*fds)[i] == -1) {
286 			err = got_error_from_errno("got_opentempfd");
287 			close_tempfiles(*fds, nfds);
288 			*fds = NULL;
289 			return err;
290 		}
291 	}
292 
293 	return NULL;
294 }
295 
296 static const struct got_error *
297 get_pack_cache_size(int *pack_cache_size)
298 {
299 	struct rlimit rl;
300 
301 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
302 		return got_error_from_errno("getrlimit");
303 
304 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
305 	if (*pack_cache_size > rl.rlim_cur / 8)
306 		*pack_cache_size = rl.rlim_cur / 8;
307 
308 	return NULL;
309 }
310 
311 const struct got_error *
312 got_repo_pack_fds_open(int **pack_fds)
313 {
314 	const struct got_error *err;
315 	int nfds;
316 
317 	err = get_pack_cache_size(&nfds);
318 	if (err)
319 		return err;
320 
321 	/*
322 	 * We need one basefd and one accumfd per cached pack.
323 	 * Our constants should be set up in a way such that
324 	 * this error never triggers.
325 	 */
326 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
327 		return got_error(GOT_ERR_NO_SPACE);
328 
329 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
330 }
331 
332 const struct got_error *
333 got_repo_pack_fds_close(int *pack_fds)
334 {
335 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
336 }
337 
338 const struct got_error *
339 got_repo_temp_fds_open(int **temp_fds) 
340 {
341 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
342 	    GOT_REPO_NUM_TEMPFILES);
343 }
344 
345 void
346 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds) 
347 {
348 	int i;
349 
350 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
351 		repo->tempfiles[i] = temp_fds[i];
352 }
353 
354 const struct got_error *
355 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
356 {
357 	int i;
358 
359 	*fd = -1;
360 	*idx = -1;
361 
362 	for (i = 0; i < nitems(repo->tempfiles); i++) {
363 		if (repo->tempfile_use_mask & (1 << i))
364 			continue;
365 		if (repo->tempfiles[i] != -1) {
366 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
367 				return got_error_from_errno("ftruncate");
368 			*fd = repo->tempfiles[i];
369 			*idx = i;
370 			repo->tempfile_use_mask |= (1 << i);
371 			return NULL;
372 		}
373 	}
374 
375 	return got_error(GOT_ERR_REPO_TEMPFILE);
376 }
377 
378 void
379 got_repo_temp_fds_put(int idx, struct got_repository *repo)
380 {
381 	repo->tempfile_use_mask &= ~(1 << idx);
382 }
383 
384 const struct got_error *
385 got_repo_temp_fds_close(int *temp_fds)
386 {
387 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
388 }
389 
390 const struct got_error *
391 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
392     struct got_object *obj)
393 {
394 #ifndef GOT_NO_OBJ_CACHE
395 	const struct got_error *err = NULL;
396 	err = got_object_cache_add(&repo->objcache, id, obj);
397 	if (err) {
398 		if (err->code == GOT_ERR_OBJ_EXISTS ||
399 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
400 			err = NULL;
401 		return err;
402 	}
403 	obj->refcnt++;
404 #endif
405 	return NULL;
406 }
407 
408 struct got_object *
409 got_repo_get_cached_object(struct got_repository *repo,
410     struct got_object_id *id)
411 {
412 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
413 }
414 
415 const struct got_error *
416 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
417     struct got_tree_object *tree)
418 {
419 #ifndef GOT_NO_OBJ_CACHE
420 	const struct got_error *err = NULL;
421 	err = got_object_cache_add(&repo->treecache, id, tree);
422 	if (err) {
423 		if (err->code == GOT_ERR_OBJ_EXISTS ||
424 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
425 			err = NULL;
426 		return err;
427 	}
428 	tree->refcnt++;
429 #endif
430 	return NULL;
431 }
432 
433 struct got_tree_object *
434 got_repo_get_cached_tree(struct got_repository *repo,
435     struct got_object_id *id)
436 {
437 	return (struct got_tree_object *)got_object_cache_get(
438 	    &repo->treecache, id);
439 }
440 
441 const struct got_error *
442 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
443     struct got_commit_object *commit)
444 {
445 #ifndef GOT_NO_OBJ_CACHE
446 	const struct got_error *err = NULL;
447 	err = got_object_cache_add(&repo->commitcache, id, commit);
448 	if (err) {
449 		if (err->code == GOT_ERR_OBJ_EXISTS ||
450 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
451 			err = NULL;
452 		return err;
453 	}
454 	commit->refcnt++;
455 #endif
456 	return NULL;
457 }
458 
459 struct got_commit_object *
460 got_repo_get_cached_commit(struct got_repository *repo,
461     struct got_object_id *id)
462 {
463 	return (struct got_commit_object *)got_object_cache_get(
464 	    &repo->commitcache, id);
465 }
466 
467 const struct got_error *
468 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
469     struct got_tag_object *tag)
470 {
471 #ifndef GOT_NO_OBJ_CACHE
472 	const struct got_error *err = NULL;
473 	err = got_object_cache_add(&repo->tagcache, id, tag);
474 	if (err) {
475 		if (err->code == GOT_ERR_OBJ_EXISTS ||
476 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
477 			err = NULL;
478 		return err;
479 	}
480 	tag->refcnt++;
481 #endif
482 	return NULL;
483 }
484 
485 struct got_tag_object *
486 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
487 {
488 	return (struct got_tag_object *)got_object_cache_get(
489 	    &repo->tagcache, id);
490 }
491 
492 const struct got_error *
493 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
494     struct got_raw_object *raw)
495 {
496 #ifndef GOT_NO_OBJ_CACHE
497 	const struct got_error *err = NULL;
498 	err = got_object_cache_add(&repo->rawcache, id, raw);
499 	if (err) {
500 		if (err->code == GOT_ERR_OBJ_EXISTS ||
501 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
502 			err = NULL;
503 		return err;
504 	}
505 	raw->refcnt++;
506 #endif
507 	return NULL;
508 }
509 
510 
511 struct got_raw_object *
512 got_repo_get_cached_raw_object(struct got_repository *repo,
513     struct got_object_id *id)
514 {
515 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
516 }
517 
518 
519 static const struct got_error *
520 open_repo(struct got_repository *repo, const char *path)
521 {
522 	const struct got_error *err = NULL;
523 
524 	repo->gitdir_fd = -1;
525 
526 	/* bare git repository? */
527 	repo->path_git_dir = strdup(path);
528 	if (repo->path_git_dir == NULL)
529 		return got_error_from_errno("strdup");
530 	if (is_git_repo(repo)) {
531 		repo->path = strdup(repo->path_git_dir);
532 		if (repo->path == NULL) {
533 			err = got_error_from_errno("strdup");
534 			goto done;
535 		}
536 		repo->gitdir_fd = open(repo->path_git_dir,
537 		    O_DIRECTORY | O_CLOEXEC);
538 		if (repo->gitdir_fd == -1) {
539 			err = got_error_from_errno2("open",
540 			    repo->path_git_dir);
541 			goto done;
542 		}
543 		return NULL;
544 	}
545 
546 	/* git repository with working tree? */
547 	free(repo->path_git_dir);
548 	repo->path_git_dir = NULL;
549 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
550 		err = got_error_from_errno("asprintf");
551 		goto done;
552 	}
553 	if (is_git_repo(repo)) {
554 		repo->path = strdup(path);
555 		if (repo->path == NULL) {
556 			err = got_error_from_errno("strdup");
557 			goto done;
558 		}
559 		repo->gitdir_fd = open(repo->path_git_dir,
560 		    O_DIRECTORY | O_CLOEXEC);
561 		if (repo->gitdir_fd == -1) {
562 			err = got_error_from_errno2("open",
563 			    repo->path_git_dir);
564 			goto done;
565 		}
566 		return NULL;
567 	}
568 
569 	err = got_error(GOT_ERR_NOT_GIT_REPO);
570 done:
571 	if (err) {
572 		free(repo->path);
573 		repo->path = NULL;
574 		free(repo->path_git_dir);
575 		repo->path_git_dir = NULL;
576 		if (repo->gitdir_fd != -1)
577 			close(repo->gitdir_fd);
578 		repo->gitdir_fd = -1;
579 
580 	}
581 	return err;
582 }
583 
584 static const struct got_error *
585 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
586 {
587 	const struct got_error *err = NULL;
588 	char *repo_gitconfig_path = NULL;
589 
590 	if (global_gitconfig_path) {
591 		/* Read settings from ~/.gitconfig. */
592 		int dummy_repo_version;
593 		err = got_repo_read_gitconfig(&dummy_repo_version,
594 		    &repo->global_gitconfig_author_name,
595 		    &repo->global_gitconfig_author_email,
596 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
597 		if (err)
598 			return err;
599 	}
600 
601 	/* Read repository's .git/config file. */
602 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
603 	if (repo_gitconfig_path == NULL)
604 		return got_error_from_errno("got_repo_get_path_gitconfig");
605 
606 	err = got_repo_read_gitconfig(
607 	    &repo->gitconfig_repository_format_version,
608 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
609 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
610 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
611 	    repo_gitconfig_path);
612 	if (err)
613 		goto done;
614 
615 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
616 		int i;
617 
618 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
619 			got_repo_free_remote_repo_data(
620 			    &repo->gitconfig_remotes[i]);
621 		}
622 		free(repo->gitconfig_remotes);
623 		repo->gitconfig_remotes = NULL;
624 		repo->ngitconfig_remotes = 0;
625 
626 		free(repo->gitconfig_author_name);
627 		repo->gitconfig_author_name = NULL;
628 		free(repo->gitconfig_author_email);
629 		repo->gitconfig_author_email = NULL;
630 
631 		free(repo->global_gitconfig_author_name);
632 		repo->global_gitconfig_author_name = NULL;
633 		free(repo->global_gitconfig_author_email);
634 		repo->global_gitconfig_author_email = NULL;
635 	}
636 
637 done:
638 	free(repo_gitconfig_path);
639 	return err;
640 }
641 
642 static const struct got_error *
643 read_gotconfig(struct got_repository *repo)
644 {
645 	const struct got_error *err = NULL;
646 	char *gotconfig_path;
647 
648 	gotconfig_path = got_repo_get_path_gotconfig(repo);
649 	if (gotconfig_path == NULL)
650 		return got_error_from_errno("got_repo_get_path_gotconfig");
651 
652 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
653 	free(gotconfig_path);
654 	return err;
655 }
656 
657 /* Supported repository format extensions. */
658 static const char *const repo_extensions[] = {
659 	"noop",			/* Got supports repository format version 1. */
660 	"preciousObjects",	/* Supported by gotadmin cleanup. */
661 	"worktreeConfig",	/* Got does not care about Git work trees. */
662 };
663 
664 const struct got_error *
665 got_repo_open(struct got_repository **repop, const char *path,
666     const char *global_gitconfig_path, int *pack_fds)
667 {
668 	struct got_repository *repo = NULL;
669 	const struct got_error *err = NULL;
670 	char *repo_path = NULL;
671 	size_t i, j = 0;
672 
673 	*repop = NULL;
674 
675 	repo = calloc(1, sizeof(*repo));
676 	if (repo == NULL)
677 		return got_error_from_errno("calloc");
678 
679 	RB_INIT(&repo->packidx_bloom_filters);
680 	TAILQ_INIT(&repo->packidx_paths);
681 
682 	for (i = 0; i < nitems(repo->privsep_children); i++) {
683 		memset(&repo->privsep_children[i], 0,
684 		    sizeof(repo->privsep_children[0]));
685 		repo->privsep_children[i].imsg_fd = -1;
686 	}
687 
688 	err = got_object_cache_init(&repo->objcache,
689 	    GOT_OBJECT_CACHE_TYPE_OBJ);
690 	if (err)
691 		goto done;
692 	err = got_object_cache_init(&repo->treecache,
693 	    GOT_OBJECT_CACHE_TYPE_TREE);
694 	if (err)
695 		goto done;
696 	err = got_object_cache_init(&repo->commitcache,
697 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
698 	if (err)
699 		goto done;
700 	err = got_object_cache_init(&repo->tagcache,
701 	    GOT_OBJECT_CACHE_TYPE_TAG);
702 	if (err)
703 		goto done;
704 	err = got_object_cache_init(&repo->rawcache,
705 	    GOT_OBJECT_CACHE_TYPE_RAW);
706 	if (err)
707 		goto done;
708 
709 	err = get_pack_cache_size(&repo->pack_cache_size);
710 	if (err)
711 		goto done;
712 	for (i = 0; i < nitems(repo->packs); i++) {
713 		if (pack_fds != NULL && i < repo->pack_cache_size) {
714 			repo->packs[i].basefd = pack_fds[j++];
715 			repo->packs[i].accumfd = pack_fds[j++];
716 		} else {
717 			repo->packs[i].basefd = -1;
718 			repo->packs[i].accumfd = -1;
719 		}
720 	}
721 	for (i = 0; i < nitems(repo->tempfiles); i++)
722 		repo->tempfiles[i] = -1;
723 	repo->pinned_pack = -1;
724 	repo->pinned_packidx = -1;
725 	repo->pinned_pid = 0;
726 
727 	repo_path = realpath(path, NULL);
728 	if (repo_path == NULL) {
729 		err = got_error_from_errno2("realpath", path);
730 		goto done;
731 	}
732 
733 	for (;;) {
734 		char *parent_path;
735 
736 		err = open_repo(repo, repo_path);
737 		if (err == NULL)
738 			break;
739 		if (err->code != GOT_ERR_NOT_GIT_REPO)
740 			goto done;
741 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
742 			err = got_error(GOT_ERR_NOT_GIT_REPO);
743 			goto done;
744 		}
745 		err = got_path_dirname(&parent_path, repo_path);
746 		if (err)
747 			goto done;
748 		free(repo_path);
749 		repo_path = parent_path;
750 	}
751 
752 	err = read_gotconfig(repo);
753 	if (err)
754 		goto done;
755 
756 	err = read_gitconfig(repo, global_gitconfig_path);
757 	if (err)
758 		goto done;
759 	if (repo->gitconfig_repository_format_version != 0) {
760 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
761 		goto done;
762 	}
763 	for (i = 0; i < repo->nextensions; i++) {
764 		char *ext = repo->extensions[i];
765 		int j, supported = 0;
766 		for (j = 0; j < nitems(repo_extensions); j++) {
767 			if (strcmp(ext, repo_extensions[j]) == 0) {
768 				supported = 1;
769 				break;
770 			}
771 		}
772 		if (!supported) {
773 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
774 			goto done;
775 		}
776 	}
777 
778 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
779 done:
780 	if (err)
781 		got_repo_close(repo);
782 	else
783 		*repop = repo;
784 	free(repo_path);
785 	return err;
786 }
787 
788 const struct got_error *
789 got_repo_close(struct got_repository *repo)
790 {
791 	const struct got_error *err = NULL, *child_err;
792 	struct got_packidx_bloom_filter *bf;
793 	struct got_pathlist_entry *pe;
794 	size_t i;
795 
796 	for (i = 0; i < repo->pack_cache_size; i++) {
797 		if (repo->packidx_cache[i] == NULL)
798 			break;
799 		got_packidx_close(repo->packidx_cache[i]);
800 	}
801 
802 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
803 	    &repo->packidx_bloom_filters))) {
804 		RB_REMOVE(got_packidx_bloom_filter_tree,
805 		    &repo->packidx_bloom_filters, bf);
806 		bloom_free(bf->bloom);
807 		free(bf->bloom);
808 		free(bf);
809 	}
810 
811 	for (i = 0; i < repo->pack_cache_size; i++)
812 		if (repo->packs[i].path_packfile)
813 			if (repo->packs[i].path_packfile)
814 				got_pack_close(&repo->packs[i]);
815 
816 	free(repo->path);
817 	free(repo->path_git_dir);
818 
819 	got_object_cache_close(&repo->objcache);
820 	got_object_cache_close(&repo->treecache);
821 	got_object_cache_close(&repo->commitcache);
822 	got_object_cache_close(&repo->tagcache);
823 	got_object_cache_close(&repo->rawcache);
824 
825 	for (i = 0; i < nitems(repo->privsep_children); i++) {
826 		if (repo->privsep_children[i].imsg_fd == -1)
827 			continue;
828 		imsg_clear(repo->privsep_children[i].ibuf);
829 		free(repo->privsep_children[i].ibuf);
830 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
831 		child_err = got_privsep_wait_for_child(
832 		    repo->privsep_children[i].pid);
833 		if (child_err && err == NULL)
834 			err = child_err;
835 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
836 		    err == NULL)
837 			err = got_error_from_errno("close");
838 	}
839 
840 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
841 	    err == NULL)
842 		err = got_error_from_errno("close");
843 
844 	if (repo->gotconfig)
845 		got_gotconfig_free(repo->gotconfig);
846 	free(repo->gitconfig_author_name);
847 	free(repo->gitconfig_author_email);
848 	for (i = 0; i < repo->ngitconfig_remotes; i++)
849 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
850 	free(repo->gitconfig_remotes);
851 	for (i = 0; i < repo->nextensions; i++)
852 		free(repo->extensions[i]);
853 	free(repo->extensions);
854 
855 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
856 		free((void *)pe->path);
857 	got_pathlist_free(&repo->packidx_paths);
858 	free(repo);
859 
860 	return err;
861 }
862 
863 void
864 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
865 {
866 	int i;
867 
868 	free(repo->name);
869 	repo->name = NULL;
870 	free(repo->fetch_url);
871 	repo->fetch_url = NULL;
872 	free(repo->send_url);
873 	repo->send_url = NULL;
874 	for (i = 0; i < repo->nfetch_branches; i++)
875 		free(repo->fetch_branches[i]);
876 	free(repo->fetch_branches);
877 	repo->fetch_branches = NULL;
878 	repo->nfetch_branches = 0;
879 	for (i = 0; i < repo->nsend_branches; i++)
880 		free(repo->send_branches[i]);
881 	free(repo->send_branches);
882 	repo->send_branches = NULL;
883 	repo->nsend_branches = 0;
884 }
885 
886 const struct got_error *
887 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
888     const char *input_path)
889 {
890 	const struct got_error *err = NULL;
891 	const char *repo_abspath = NULL;
892 	size_t repolen, len;
893 	char *canonpath, *path = NULL;
894 
895 	*in_repo_path = NULL;
896 
897 	canonpath = strdup(input_path);
898 	if (canonpath == NULL) {
899 		err = got_error_from_errno("strdup");
900 		goto done;
901 	}
902 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
903 	if (err)
904 		goto done;
905 
906 	repo_abspath = got_repo_get_path(repo);
907 
908 	if (canonpath[0] == '\0') {
909 		path = strdup(canonpath);
910 		if (path == NULL) {
911 			err = got_error_from_errno("strdup");
912 			goto done;
913 		}
914 	} else {
915 		path = realpath(canonpath, NULL);
916 		if (path == NULL) {
917 			if (errno != ENOENT) {
918 				err = got_error_from_errno2("realpath",
919 				    canonpath);
920 				goto done;
921 			}
922 			/*
923 			 * Path is not on disk.
924 			 * Assume it is already relative to repository root.
925 			 */
926 			path = strdup(canonpath);
927 			if (path == NULL) {
928 				err = got_error_from_errno("strdup");
929 				goto done;
930 			}
931 		}
932 
933 		repolen = strlen(repo_abspath);
934 		len = strlen(path);
935 
936 
937 		if (strcmp(path, repo_abspath) == 0) {
938 			free(path);
939 			path = strdup("");
940 			if (path == NULL) {
941 				err = got_error_from_errno("strdup");
942 				goto done;
943 			}
944 		} else if (len > repolen &&
945 		    got_path_is_child(path, repo_abspath, repolen)) {
946 			/* Matched an on-disk path inside repository. */
947 			if (got_repo_is_bare(repo)) {
948 				/*
949 				 * Matched an on-disk path inside repository
950 				 * database. Treat input as repository-relative.
951 				 */
952 				free(path);
953 				path = canonpath;
954 				canonpath = NULL;
955 			} else {
956 				char *child;
957 				/* Strip common prefix with repository path. */
958 				err = got_path_skip_common_ancestor(&child,
959 				    repo_abspath, path);
960 				if (err)
961 					goto done;
962 				free(path);
963 				path = child;
964 			}
965 		} else {
966 			/*
967 			 * Matched unrelated on-disk path.
968 			 * Treat input as repository-relative.
969 			 */
970 			free(path);
971 			path = canonpath;
972 			canonpath = NULL;
973 		}
974 	}
975 
976 	/* Make in-repository path absolute */
977 	if (path[0] != '/') {
978 		char *abspath;
979 		if (asprintf(&abspath, "/%s", path) == -1) {
980 			err = got_error_from_errno("asprintf");
981 			goto done;
982 		}
983 		free(path);
984 		path = abspath;
985 	}
986 
987 done:
988 	free(canonpath);
989 	if (err)
990 		free(path);
991 	else
992 		*in_repo_path = path;
993 	return err;
994 }
995 
996 static const struct got_error *
997 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
998     const char *path_packidx)
999 {
1000 	const struct got_error *err = NULL;
1001 	size_t i;
1002 
1003 	for (i = 0; i < repo->pack_cache_size; i++) {
1004 		if (repo->packidx_cache[i] == NULL)
1005 			break;
1006 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1007 		    path_packidx) == 0) {
1008 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1009 		}
1010 	}
1011 	if (i == repo->pack_cache_size) {
1012 		do {
1013 			i--;
1014 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1015 		    i == repo->pinned_packidx);
1016 		err = got_packidx_close(repo->packidx_cache[i]);
1017 		if (err)
1018 			return err;
1019 	}
1020 
1021 	repo->packidx_cache[i] = packidx;
1022 
1023 	return NULL;
1024 }
1025 
1026 int
1027 got_repo_is_packidx_filename(const char *name, size_t len)
1028 {
1029 	if (len != GOT_PACKIDX_NAMELEN)
1030 		return 0;
1031 
1032 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1033 		return 0;
1034 
1035 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1036 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1037 		return 0;
1038 
1039 	return 1;
1040 }
1041 
1042 static struct got_packidx_bloom_filter *
1043 get_packidx_bloom_filter(struct got_repository *repo,
1044     const char *path, size_t path_len)
1045 {
1046 	struct got_packidx_bloom_filter key;
1047 
1048 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1049 		return NULL; /* XXX */
1050 	key.path_len = path_len;
1051 
1052 	return RB_FIND(got_packidx_bloom_filter_tree,
1053 	    &repo->packidx_bloom_filters, &key);
1054 }
1055 
1056 int
1057 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1058     const char *path_packidx, struct got_object_id *id)
1059 {
1060 	struct got_packidx_bloom_filter *bf;
1061 
1062 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1063 	if (bf)
1064 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1065 
1066 	/* No bloom filter means this pack index must be searched. */
1067 	return 1;
1068 }
1069 
1070 static const struct got_error *
1071 add_packidx_bloom_filter(struct got_repository *repo,
1072     struct got_packidx *packidx, const char *path_packidx)
1073 {
1074 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1075 	struct got_packidx_bloom_filter *bf;
1076 	size_t len;
1077 
1078 	/*
1079 	 * Don't use bloom filters for very large pack index files.
1080 	 * Large pack files will contain a relatively large fraction
1081 	 * of our objects so we will likely need to visit them anyway.
1082 	 * The more objects a pack file contains the higher the probability
1083 	 * of a false-positive match from the bloom filter. And reading
1084 	 * all object IDs from a large pack index file can be expensive.
1085 	 */
1086 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1087 		return NULL;
1088 
1089 	/* Do we already have a filter for this pack index? */
1090 	if (get_packidx_bloom_filter(repo, path_packidx,
1091 	    strlen(path_packidx)) != NULL)
1092 		return NULL;
1093 
1094 	bf = calloc(1, sizeof(*bf));
1095 	if (bf == NULL)
1096 		return got_error_from_errno("calloc");
1097 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1098 	if (bf->bloom == NULL) {
1099 		free(bf);
1100 		return got_error_from_errno("calloc");
1101 	}
1102 
1103 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1104 	if (len >= sizeof(bf->path)) {
1105 		free(bf->bloom);
1106 		free(bf);
1107 		return got_error(GOT_ERR_NO_SPACE);
1108 	}
1109 	bf->path_len = len;
1110 
1111 	/* Minimum size supported by our bloom filter is 1000 entries. */
1112 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1113 	for (i = 0; i < nobjects; i++) {
1114 		struct got_packidx_object_id *id;
1115 		id = &packidx->hdr.sorted_ids[i];
1116 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1117 	}
1118 
1119 	RB_INSERT(got_packidx_bloom_filter_tree,
1120 	    &repo->packidx_bloom_filters, bf);
1121 	return NULL;
1122 }
1123 
1124 static void
1125 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1126 {
1127 	struct got_pathlist_entry *pe;
1128 
1129 	while (!TAILQ_EMPTY(packidx_paths)) {
1130 		pe = TAILQ_FIRST(packidx_paths);
1131 		TAILQ_REMOVE(packidx_paths, pe, entry);
1132 		free((char *)pe->path);
1133 		free(pe);
1134 	}
1135 }
1136 
1137 static const struct got_error *
1138 refresh_packidx_paths(struct got_repository *repo)
1139 {
1140 	const struct got_error *err = NULL;
1141 	char *objects_pack_dir = NULL;
1142 	struct stat sb;
1143 
1144 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1145 	if (objects_pack_dir == NULL)
1146 		return got_error_from_errno("got_repo_get_path_objects_pack");
1147 
1148 	if (stat(objects_pack_dir, &sb) == -1) {
1149 		if (errno != ENOENT) {
1150 			err = got_error_from_errno2("stat", objects_pack_dir);
1151 			goto done;
1152 		}
1153 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1154 		purge_packidx_paths(&repo->packidx_paths);
1155 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1156 		if (err)
1157 			goto done;
1158 	}
1159 done:
1160 	free(objects_pack_dir);
1161 	return err;
1162 }
1163 
1164 const struct got_error *
1165 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1166     struct got_repository *repo, struct got_object_id *id)
1167 {
1168 	const struct got_error *err;
1169 	struct got_pathlist_entry *pe;
1170 	size_t i;
1171 
1172 	/* Search pack index cache. */
1173 	for (i = 0; i < repo->pack_cache_size; i++) {
1174 		if (repo->packidx_cache[i] == NULL)
1175 			break;
1176 		if (!got_repo_check_packidx_bloom_filter(repo,
1177 		    repo->packidx_cache[i]->path_packidx, id))
1178 			continue; /* object will not be found in this index */
1179 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1180 		if (*idx != -1) {
1181 			*packidx = repo->packidx_cache[i];
1182 			/*
1183 			 * Move this cache entry to the front. Repeatedly
1184 			 * searching a wrong pack index can be expensive.
1185 			 */
1186 			if (i > 0) {
1187 				memmove(&repo->packidx_cache[1],
1188 				    &repo->packidx_cache[0],
1189 				    i * sizeof(repo->packidx_cache[0]));
1190 				repo->packidx_cache[0] = *packidx;
1191 				if (repo->pinned_packidx >= 0 &&
1192 				    repo->pinned_packidx < i)
1193 					repo->pinned_packidx++;
1194 				else if (repo->pinned_packidx == i)
1195 					repo->pinned_packidx = 0;
1196 			}
1197 			return NULL;
1198 		}
1199 	}
1200 	/* No luck. Search the filesystem. */
1201 
1202 	err = refresh_packidx_paths(repo);
1203 	if (err)
1204 		return err;
1205 
1206 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1207 		const char *path_packidx = pe->path;
1208 		int is_cached = 0;
1209 
1210 		if (!got_repo_check_packidx_bloom_filter(repo,
1211 		    pe->path, id))
1212 			continue; /* object will not be found in this index */
1213 
1214 		for (i = 0; i < repo->pack_cache_size; i++) {
1215 			if (repo->packidx_cache[i] == NULL)
1216 				break;
1217 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1218 			    path_packidx) == 0) {
1219 				is_cached = 1;
1220 				break;
1221 			}
1222 		}
1223 		if (is_cached)
1224 			continue; /* already searched */
1225 
1226 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1227 		    path_packidx, 0);
1228 		if (err)
1229 			goto done;
1230 
1231 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1232 		if (err)
1233 			goto done;
1234 
1235 		err = cache_packidx(repo, *packidx, path_packidx);
1236 		if (err)
1237 			goto done;
1238 
1239 		*idx = got_packidx_get_object_idx(*packidx, id);
1240 		if (*idx != -1) {
1241 			err = NULL; /* found the object */
1242 			goto done;
1243 		}
1244 	}
1245 
1246 	err = got_error_no_obj(id);
1247 done:
1248 	return err;
1249 }
1250 
1251 const struct got_error *
1252 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1253     struct got_repository *repo)
1254 {
1255 	const struct got_error *err = NULL;
1256 	DIR *packdir = NULL;
1257 	struct dirent *dent;
1258 	char *path_packidx = NULL;
1259 	int packdir_fd;
1260 	struct stat sb;
1261 
1262 	packdir_fd = openat(got_repo_get_fd(repo),
1263 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1264 	if (packdir_fd == -1) {
1265 		return got_error_from_errno_fmt("openat: %s/%s",
1266 		    got_repo_get_path_git_dir(repo),
1267 		    GOT_OBJECTS_PACK_DIR);
1268 	}
1269 
1270 	packdir = fdopendir(packdir_fd);
1271 	if (packdir == NULL) {
1272 		err = got_error_from_errno("fdopendir");
1273 		goto done;
1274 	}
1275 
1276 	if (fstat(packdir_fd, &sb) == -1) {
1277 		err = got_error_from_errno("fstat");
1278 		goto done;
1279 	}
1280 	repo->pack_path_mtime = sb.st_mtime;
1281 
1282 	while ((dent = readdir(packdir)) != NULL) {
1283 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1284 			continue;
1285 
1286 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1287 		    dent->d_name) == -1) {
1288 			err = got_error_from_errno("asprintf");
1289 			path_packidx = NULL;
1290 			break;
1291 		}
1292 
1293 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1294 		if (err)
1295 			break;
1296 	}
1297 done:
1298 	if (err)
1299 		free(path_packidx);
1300 	if (packdir && closedir(packdir) != 0 && err == NULL)
1301 		err = got_error_from_errno("closedir");
1302 	return err;
1303 }
1304 
1305 const struct got_error *
1306 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1307     struct got_repository *repo)
1308 {
1309 	const struct got_error *err;
1310 	size_t i;
1311 
1312 	*packidx = NULL;
1313 
1314 	/* Search pack index cache. */
1315 	for (i = 0; i < repo->pack_cache_size; i++) {
1316 		if (repo->packidx_cache[i] == NULL)
1317 			break;
1318 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1319 		    path_packidx) == 0) {
1320 			*packidx = repo->packidx_cache[i];
1321 			return NULL;
1322 		}
1323 	}
1324 	/* No luck. Search the filesystem. */
1325 
1326 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1327 	    path_packidx, 0);
1328 	if (err)
1329 		return err;
1330 
1331 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1332 	if (err)
1333 		goto done;
1334 
1335 	err = cache_packidx(repo, *packidx, path_packidx);
1336 done:
1337 	if (err) {
1338 		got_packidx_close(*packidx);
1339 		*packidx = NULL;
1340 	}
1341 	return err;
1342 }
1343 
1344 static const struct got_error *
1345 read_packfile_hdr(int fd, struct got_packidx *packidx)
1346 {
1347 	const struct got_error *err = NULL;
1348 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1349 	struct got_packfile_hdr hdr;
1350 	ssize_t n;
1351 
1352 	n = read(fd, &hdr, sizeof(hdr));
1353 	if (n < 0)
1354 		return got_error_from_errno("read");
1355 	if (n != sizeof(hdr))
1356 		return got_error(GOT_ERR_BAD_PACKFILE);
1357 
1358 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1359 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1360 	    be32toh(hdr.nobjects) != totobj)
1361 		err = got_error(GOT_ERR_BAD_PACKFILE);
1362 
1363 	return err;
1364 }
1365 
1366 static const struct got_error *
1367 open_packfile(int *fd, struct got_repository *repo,
1368     const char *relpath, struct got_packidx *packidx)
1369 {
1370 	const struct got_error *err = NULL;
1371 
1372 	*fd = openat(got_repo_get_fd(repo), relpath,
1373 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1374 	if (*fd == -1)
1375 		return got_error_from_errno_fmt("openat: %s/%s",
1376 		    got_repo_get_path_git_dir(repo), relpath);
1377 
1378 	if (packidx) {
1379 		err = read_packfile_hdr(*fd, packidx);
1380 		if (err) {
1381 			close(*fd);
1382 			*fd = -1;
1383 		}
1384 	}
1385 
1386 	return err;
1387 }
1388 
1389 const struct got_error *
1390 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1391     const char *path_packfile, struct got_packidx *packidx)
1392 {
1393 	const struct got_error *err = NULL;
1394 	struct got_pack *pack = NULL;
1395 	struct stat sb;
1396 	size_t i;
1397 
1398 	if (packp)
1399 		*packp = NULL;
1400 
1401 	for (i = 0; i < repo->pack_cache_size; i++) {
1402 		pack = &repo->packs[i];
1403 		if (pack->path_packfile == NULL)
1404 			break;
1405 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1406 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1407 	}
1408 
1409 	if (i == repo->pack_cache_size) {
1410 		struct got_pack tmp;
1411 		do {
1412 			i--;
1413 		} while (i > 0 && repo->pinned_pack >= 0 &&
1414 		    i == repo->pinned_pack);
1415 		err = got_pack_close(&repo->packs[i]);
1416 		if (err)
1417 			return err;
1418 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1419 			return got_error_from_errno("ftruncate");
1420 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1421 			return got_error_from_errno("ftruncate");
1422 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1423 		memcpy(&repo->packs[i], &repo->packs[0],
1424 		    sizeof(repo->packs[i]));
1425 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1426 		if (repo->pinned_pack == 0)
1427 			repo->pinned_pack = i;
1428 		else if (repo->pinned_pack == i)
1429 			repo->pinned_pack = 0;
1430 		i = 0;
1431 	}
1432 
1433 	pack = &repo->packs[i];
1434 
1435 	pack->path_packfile = strdup(path_packfile);
1436 	if (pack->path_packfile == NULL) {
1437 		err = got_error_from_errno("strdup");
1438 		goto done;
1439 	}
1440 
1441 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1442 	if (err)
1443 		goto done;
1444 
1445 	if (fstat(pack->fd, &sb) != 0) {
1446 		err = got_error_from_errno("fstat");
1447 		goto done;
1448 	}
1449 	pack->filesize = sb.st_size;
1450 
1451 	pack->privsep_child = NULL;
1452 
1453 	err = got_delta_cache_alloc(&pack->delta_cache);
1454 	if (err)
1455 		goto done;
1456 
1457 #ifndef GOT_PACK_NO_MMAP
1458 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1459 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1460 		    pack->fd, 0);
1461 		if (pack->map == MAP_FAILED) {
1462 			if (errno != ENOMEM) {
1463 				err = got_error_from_errno("mmap");
1464 				goto done;
1465 			}
1466 			pack->map = NULL; /* fall back to read(2) */
1467 		}
1468 	}
1469 #endif
1470 done:
1471 	if (err) {
1472 		if (pack)
1473 			got_pack_close(pack);
1474 	} else if (packp)
1475 		*packp = pack;
1476 	return err;
1477 }
1478 
1479 struct got_pack *
1480 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1481 {
1482 	struct got_pack *pack = NULL;
1483 	size_t i;
1484 
1485 	for (i = 0; i < repo->pack_cache_size; i++) {
1486 		pack = &repo->packs[i];
1487 		if (pack->path_packfile == NULL)
1488 			break;
1489 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1490 			return pack;
1491 	}
1492 
1493 	return NULL;
1494 }
1495 
1496 const struct got_error *
1497 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1498     struct got_pack *pack)
1499 {
1500 	size_t i;
1501 	int pinned_pack = -1, pinned_packidx = -1;
1502 
1503 	for (i = 0; i < repo->pack_cache_size; i++) {
1504 		if (repo->packidx_cache[i] &&
1505 		    strcmp(repo->packidx_cache[i]->path_packidx,
1506 		    packidx->path_packidx) == 0)
1507 			pinned_packidx = i;
1508 		if (repo->packs[i].path_packfile &&
1509 		    strcmp(repo->packs[i].path_packfile,
1510 		    pack->path_packfile) == 0)
1511 			pinned_pack = i;
1512 	}
1513 
1514 	if (pinned_packidx == -1 || pinned_pack == -1)
1515 		return got_error(GOT_ERR_PIN_PACK);
1516 
1517 	repo->pinned_pack = pinned_pack;
1518 	repo->pinned_packidx = pinned_packidx;
1519 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1520 	return NULL;
1521 }
1522 
1523 struct got_pack *
1524 got_repo_get_pinned_pack(struct got_repository *repo)
1525 {
1526 	if (repo->pinned_pack >= 0 &&
1527 	    repo->pinned_pack < repo->pack_cache_size)
1528 		return &repo->packs[repo->pinned_pack];
1529 
1530 	return NULL;
1531 }
1532 
1533 void
1534 got_repo_unpin_pack(struct got_repository *repo)
1535 {
1536 	repo->pinned_packidx = -1;
1537 	repo->pinned_pack = -1;
1538 	repo->pinned_pid = 0;
1539 }
1540 
1541 const struct got_error *
1542 got_repo_init(const char *repo_path, const char *head_name)
1543 {
1544 	const struct got_error *err = NULL;
1545 	const char *dirnames[] = {
1546 		GOT_OBJECTS_DIR,
1547 		GOT_OBJECTS_PACK_DIR,
1548 		GOT_REFS_DIR,
1549 	};
1550 	const char *description_str = "Unnamed repository; "
1551 	    "edit this file 'description' to name the repository.";
1552 	const char *headref = "ref: refs/heads/";
1553 	const char *gitconfig_str = "[core]\n"
1554 	    "\trepositoryformatversion = 0\n"
1555 	    "\tfilemode = true\n"
1556 	    "\tbare = true\n";
1557 	char *headref_str, *path;
1558 	size_t i;
1559 
1560 	if (!got_path_dir_is_empty(repo_path))
1561 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1562 
1563 	for (i = 0; i < nitems(dirnames); i++) {
1564 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1565 			return got_error_from_errno("asprintf");
1566 		}
1567 		err = got_path_mkdir(path);
1568 		free(path);
1569 		if (err)
1570 			return err;
1571 	}
1572 
1573 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1574 		return got_error_from_errno("asprintf");
1575 	err = got_path_create_file(path, description_str);
1576 	free(path);
1577 	if (err)
1578 		return err;
1579 
1580 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1581 		return got_error_from_errno("asprintf");
1582 	if (asprintf(&headref_str, "%s%s", headref,
1583 	    head_name ? head_name : "main") == -1) {
1584 		free(path);
1585 		return got_error_from_errno("asprintf");
1586 	}
1587 	err = got_path_create_file(path, headref_str);
1588 	free(headref_str);
1589 	free(path);
1590 	if (err)
1591 		return err;
1592 
1593 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1594 		return got_error_from_errno("asprintf");
1595 	err = got_path_create_file(path, gitconfig_str);
1596 	free(path);
1597 	if (err)
1598 		return err;
1599 
1600 	return NULL;
1601 }
1602 
1603 static const struct got_error *
1604 match_packed_object(struct got_object_id **unique_id,
1605     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1606 {
1607 	const struct got_error *err = NULL;
1608 	struct got_object_id_queue matched_ids;
1609 	struct got_pathlist_entry *pe;
1610 
1611 	STAILQ_INIT(&matched_ids);
1612 
1613 	err = refresh_packidx_paths(repo);
1614 	if (err)
1615 		return err;
1616 
1617 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1618 		const char *path_packidx = pe->path;
1619 		struct got_packidx *packidx;
1620 		struct got_object_qid *qid;
1621 
1622 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1623 		    path_packidx, 0);
1624 		if (err)
1625 			break;
1626 
1627 		err = got_packidx_match_id_str_prefix(&matched_ids,
1628 		    packidx, id_str_prefix);
1629 		if (err) {
1630 			got_packidx_close(packidx);
1631 			break;
1632 		}
1633 		err = got_packidx_close(packidx);
1634 		if (err)
1635 			break;
1636 
1637 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1638 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1639 				int matched_type;
1640 				err = got_object_get_type(&matched_type, repo,
1641 				    &qid->id);
1642 				if (err)
1643 					goto done;
1644 				if (matched_type != obj_type)
1645 					continue;
1646 			}
1647 			if (*unique_id == NULL) {
1648 				*unique_id = got_object_id_dup(&qid->id);
1649 				if (*unique_id == NULL) {
1650 					err = got_error_from_errno("malloc");
1651 					goto done;
1652 				}
1653 			} else {
1654 				if (got_object_id_cmp(*unique_id,
1655 				    &qid->id) == 0)
1656 					continue; /* packed multiple times */
1657 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1658 				goto done;
1659 			}
1660 		}
1661 	}
1662 done:
1663 	got_object_id_queue_free(&matched_ids);
1664 	if (err) {
1665 		free(*unique_id);
1666 		*unique_id = NULL;
1667 	}
1668 	return err;
1669 }
1670 
1671 static const struct got_error *
1672 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1673     const char *object_dir, const char *id_str_prefix, int obj_type,
1674     struct got_repository *repo)
1675 {
1676 	const struct got_error *err = NULL;
1677 	char *path, *id_str = NULL;
1678 	DIR *dir = NULL;
1679 	struct dirent *dent;
1680 	struct got_object_id id;
1681 
1682 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1683 		err = got_error_from_errno("asprintf");
1684 		goto done;
1685 	}
1686 
1687 	dir = opendir(path);
1688 	if (dir == NULL) {
1689 		if (errno == ENOENT) {
1690 			err = NULL;
1691 			goto done;
1692 		}
1693 		err = got_error_from_errno2("opendir", path);
1694 		goto done;
1695 	}
1696 	while ((dent = readdir(dir)) != NULL) {
1697 		int cmp;
1698 
1699 		free(id_str);
1700 		id_str = NULL;
1701 
1702 		if (strcmp(dent->d_name, ".") == 0 ||
1703 		    strcmp(dent->d_name, "..") == 0)
1704 			continue;
1705 
1706 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1707 			err = got_error_from_errno("asprintf");
1708 			goto done;
1709 		}
1710 
1711 		if (!got_parse_sha1_digest(id.sha1, id_str))
1712 			continue;
1713 
1714 		/*
1715 		 * Directory entries do not necessarily appear in
1716 		 * sorted order, so we must iterate over all of them.
1717 		 */
1718 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1719 		if (cmp != 0)
1720 			continue;
1721 
1722 		if (*unique_id == NULL) {
1723 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1724 				int matched_type;
1725 				err = got_object_get_type(&matched_type, repo,
1726 				    &id);
1727 				if (err)
1728 					goto done;
1729 				if (matched_type != obj_type)
1730 					continue;
1731 			}
1732 			*unique_id = got_object_id_dup(&id);
1733 			if (*unique_id == NULL) {
1734 				err = got_error_from_errno("got_object_id_dup");
1735 				goto done;
1736 			}
1737 		} else {
1738 			if (got_object_id_cmp(*unique_id, &id) == 0)
1739 				continue; /* both packed and loose */
1740 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1741 			goto done;
1742 		}
1743 	}
1744 done:
1745 	if (dir && closedir(dir) != 0 && err == NULL)
1746 		err = got_error_from_errno("closedir");
1747 	if (err) {
1748 		free(*unique_id);
1749 		*unique_id = NULL;
1750 	}
1751 	free(id_str);
1752 	free(path);
1753 	return err;
1754 }
1755 
1756 const struct got_error *
1757 got_repo_match_object_id_prefix(struct got_object_id **id,
1758     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1759 {
1760 	const struct got_error *err = NULL;
1761 	char *path_objects = NULL, *object_dir = NULL;
1762 	size_t len;
1763 	int i;
1764 
1765 	*id = NULL;
1766 
1767 	path_objects = got_repo_get_path_objects(repo);
1768 
1769 	len = strlen(id_str_prefix);
1770 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1771 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1772 		goto done;
1773 	}
1774 
1775 	for (i = 0; i < len; i++) {
1776 		if (isxdigit((unsigned char)id_str_prefix[i]))
1777 			continue;
1778 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1779 		goto done;
1780 	}
1781 
1782 	if (len >= 2) {
1783 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1784 		if (err)
1785 			goto done;
1786 		object_dir = strndup(id_str_prefix, 2);
1787 		if (object_dir == NULL) {
1788 			err = got_error_from_errno("strdup");
1789 			goto done;
1790 		}
1791 		err = match_loose_object(id, path_objects, object_dir,
1792 		    id_str_prefix, obj_type, repo);
1793 	} else if (len == 1) {
1794 		int i;
1795 		for (i = 0; i < 0xf; i++) {
1796 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1797 			    == -1) {
1798 				err = got_error_from_errno("asprintf");
1799 				goto done;
1800 			}
1801 			err = match_packed_object(id, repo, object_dir,
1802 			    obj_type);
1803 			if (err)
1804 				goto done;
1805 			err = match_loose_object(id, path_objects, object_dir,
1806 			    id_str_prefix, obj_type, repo);
1807 			if (err)
1808 				goto done;
1809 		}
1810 	} else {
1811 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1812 		goto done;
1813 	}
1814 done:
1815 	free(path_objects);
1816 	free(object_dir);
1817 	if (err) {
1818 		free(*id);
1819 		*id = NULL;
1820 	} else if (*id == NULL) {
1821 		switch (obj_type) {
1822 		case GOT_OBJ_TYPE_BLOB:
1823 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1824 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1825 			break;
1826 		case GOT_OBJ_TYPE_TREE:
1827 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1828 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1829 			break;
1830 		case GOT_OBJ_TYPE_COMMIT:
1831 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1832 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1833 			break;
1834 		case GOT_OBJ_TYPE_TAG:
1835 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1836 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1837 			break;
1838 		default:
1839 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1840 			break;
1841 		}
1842 	}
1843 
1844 	return err;
1845 }
1846 
1847 const struct got_error *
1848 got_repo_match_object_id(struct got_object_id **id, char **label,
1849     const char *id_str, int obj_type, struct got_reflist_head *refs,
1850     struct got_repository *repo)
1851 {
1852 	const struct got_error *err;
1853 	struct got_tag_object *tag;
1854 	struct got_reference *ref = NULL;
1855 
1856 	*id = NULL;
1857 	if (label)
1858 		*label = NULL;
1859 
1860 	if (refs) {
1861 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1862 		    refs, repo);
1863 		if (err == NULL) {
1864 			*id = got_object_id_dup(
1865 			    got_object_tag_get_object_id(tag));
1866 			if (*id == NULL)
1867 				err = got_error_from_errno("got_object_id_dup");
1868 			else if (label && asprintf(label, "refs/tags/%s",
1869 			    got_object_tag_get_name(tag)) == -1) {
1870 				err = got_error_from_errno("asprintf");
1871 				free(*id);
1872 				*id = NULL;
1873 			}
1874 			got_object_tag_close(tag);
1875 			return err;
1876 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1877 		    err->code != GOT_ERR_NO_OBJ)
1878 			return err;
1879 	}
1880 
1881 	err = got_ref_open(&ref, repo, id_str, 0);
1882 	if (err == NULL) {
1883 		err = got_ref_resolve(id, repo, ref);
1884 		if (err)
1885 			goto done;
1886 		if (label) {
1887 			*label = strdup(got_ref_get_name(ref));
1888 			if (*label == NULL) {
1889 				err = got_error_from_errno("strdup");
1890 				goto done;
1891 			}
1892 		}
1893 	} else {
1894 		if (err->code != GOT_ERR_NOT_REF &&
1895 		    err->code != GOT_ERR_BAD_REF_NAME)
1896 			goto done;
1897 		err = got_repo_match_object_id_prefix(id, id_str,
1898 		    obj_type, repo);
1899 		if (err) {
1900 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1901 				err = got_error_not_ref(id_str);
1902 			goto done;
1903 		}
1904 		if (label) {
1905 			err = got_object_id_str(label, *id);
1906 			if (*label == NULL) {
1907 				err = got_error_from_errno("strdup");
1908 				goto done;
1909 			}
1910 		}
1911 	}
1912 done:
1913 	if (ref)
1914 		got_ref_close(ref);
1915 	return err;
1916 }
1917 
1918 const struct got_error *
1919 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1920     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1921 {
1922 	const struct got_error *err = NULL;
1923 	struct got_reflist_entry *re;
1924 	struct got_object_id *tag_id;
1925 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1926 
1927 	*tag = NULL;
1928 
1929 	TAILQ_FOREACH(re, refs, entry) {
1930 		const char *refname;
1931 		refname = got_ref_get_name(re->ref);
1932 		if (got_ref_is_symbolic(re->ref))
1933 			continue;
1934 		if (strncmp(refname, "refs/tags/", 10) != 0)
1935 			continue;
1936 		if (!name_is_absolute)
1937 			refname += strlen("refs/tags/");
1938 		if (strcmp(refname, name) != 0)
1939 			continue;
1940 		err = got_ref_resolve(&tag_id, repo, re->ref);
1941 		if (err)
1942 			break;
1943 		err = got_object_open_as_tag(tag, repo, tag_id);
1944 		free(tag_id);
1945 		if (err)
1946 			break;
1947 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1948 		    got_object_tag_get_object_type(*tag) == obj_type)
1949 			break;
1950 		got_object_tag_close(*tag);
1951 		*tag = NULL;
1952 	}
1953 
1954 	if (err == NULL && *tag == NULL)
1955 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1956 		    GOT_OBJ_LABEL_TAG, name);
1957 	return err;
1958 }
1959 
1960 static const struct got_error *
1961 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1962     const char *name, mode_t mode, struct got_object_id *blob_id)
1963 {
1964 	const struct got_error *err = NULL;
1965 
1966 	 *new_te = NULL;
1967 
1968 	*new_te = calloc(1, sizeof(**new_te));
1969 	if (*new_te == NULL)
1970 		return got_error_from_errno("calloc");
1971 
1972 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1973 	    sizeof((*new_te)->name)) {
1974 		err = got_error(GOT_ERR_NO_SPACE);
1975 		goto done;
1976 	}
1977 
1978 	if (S_ISLNK(mode)) {
1979 		(*new_te)->mode = S_IFLNK;
1980 	} else {
1981 		(*new_te)->mode = S_IFREG;
1982 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1983 	}
1984 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1985 done:
1986 	if (err && *new_te) {
1987 		free(*new_te);
1988 		*new_te = NULL;
1989 	}
1990 	return err;
1991 }
1992 
1993 static const struct got_error *
1994 import_file(struct got_tree_entry **new_te, struct dirent *de,
1995     const char *path, struct got_repository *repo)
1996 {
1997 	const struct got_error *err;
1998 	struct got_object_id *blob_id = NULL;
1999 	char *filepath;
2000 	struct stat sb;
2001 
2002 	if (asprintf(&filepath, "%s%s%s", path,
2003 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2004 		return got_error_from_errno("asprintf");
2005 
2006 	if (lstat(filepath, &sb) != 0) {
2007 		err = got_error_from_errno2("lstat", path);
2008 		goto done;
2009 	}
2010 
2011 	err = got_object_blob_create(&blob_id, filepath, repo);
2012 	if (err)
2013 		goto done;
2014 
2015 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2016 	    blob_id);
2017 done:
2018 	free(filepath);
2019 	if (err)
2020 		free(blob_id);
2021 	return err;
2022 }
2023 
2024 static const struct got_error *
2025 insert_tree_entry(struct got_tree_entry *new_te,
2026     struct got_pathlist_head *paths)
2027 {
2028 	const struct got_error *err = NULL;
2029 	struct got_pathlist_entry *new_pe;
2030 
2031 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2032 	if (err)
2033 		return err;
2034 	if (new_pe == NULL)
2035 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2036 	return NULL;
2037 }
2038 
2039 static const struct got_error *write_tree(struct got_object_id **,
2040     const char *, struct got_pathlist_head *, struct got_repository *,
2041     got_repo_import_cb progress_cb, void *progress_arg);
2042 
2043 static const struct got_error *
2044 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2045     const char *path, struct got_pathlist_head *ignores,
2046     struct got_repository *repo,
2047     got_repo_import_cb progress_cb, void *progress_arg)
2048 {
2049 	const struct got_error *err;
2050 	struct got_object_id *id = NULL;
2051 	char *subdirpath;
2052 
2053 	if (asprintf(&subdirpath, "%s%s%s", path,
2054 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2055 		return got_error_from_errno("asprintf");
2056 
2057 	(*new_te) = calloc(1, sizeof(**new_te));
2058 	if (*new_te == NULL)
2059 		return got_error_from_errno("calloc");
2060 	(*new_te)->mode = S_IFDIR;
2061 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2062 	    sizeof((*new_te)->name)) {
2063 		err = got_error(GOT_ERR_NO_SPACE);
2064 		goto done;
2065 	}
2066 	err = write_tree(&id, subdirpath, ignores,  repo,
2067 	    progress_cb, progress_arg);
2068 	if (err)
2069 		goto done;
2070 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2071 
2072 done:
2073 	free(id);
2074 	free(subdirpath);
2075 	if (err) {
2076 		free(*new_te);
2077 		*new_te = NULL;
2078 	}
2079 	return err;
2080 }
2081 
2082 static const struct got_error *
2083 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2084     struct got_pathlist_head *ignores, struct got_repository *repo,
2085     got_repo_import_cb progress_cb, void *progress_arg)
2086 {
2087 	const struct got_error *err = NULL;
2088 	DIR *dir;
2089 	struct dirent *de;
2090 	int nentries;
2091 	struct got_tree_entry *new_te = NULL;
2092 	struct got_pathlist_head paths;
2093 	struct got_pathlist_entry *pe;
2094 
2095 	*new_tree_id = NULL;
2096 
2097 	TAILQ_INIT(&paths);
2098 
2099 	dir = opendir(path_dir);
2100 	if (dir == NULL) {
2101 		err = got_error_from_errno2("opendir", path_dir);
2102 		goto done;
2103 	}
2104 
2105 	nentries = 0;
2106 	while ((de = readdir(dir)) != NULL) {
2107 		int ignore = 0;
2108 		int type;
2109 
2110 		if (strcmp(de->d_name, ".") == 0 ||
2111 		    strcmp(de->d_name, "..") == 0)
2112 			continue;
2113 
2114 		TAILQ_FOREACH(pe, ignores, entry) {
2115 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2116 				ignore = 1;
2117 				break;
2118 			}
2119 		}
2120 		if (ignore)
2121 			continue;
2122 
2123 		err = got_path_dirent_type(&type, path_dir, de);
2124 		if (err)
2125 			goto done;
2126 
2127 		if (type == DT_DIR) {
2128 			err = import_subdir(&new_te, de, path_dir,
2129 			    ignores, repo, progress_cb, progress_arg);
2130 			if (err) {
2131 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2132 					goto done;
2133 				err = NULL;
2134 				continue;
2135 			}
2136 		} else if (type == DT_REG || type == DT_LNK) {
2137 			err = import_file(&new_te, de, path_dir, repo);
2138 			if (err)
2139 				goto done;
2140 		} else
2141 			continue;
2142 
2143 		err = insert_tree_entry(new_te, &paths);
2144 		if (err)
2145 			goto done;
2146 		nentries++;
2147 	}
2148 
2149 	if (TAILQ_EMPTY(&paths)) {
2150 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2151 		    "cannot create tree without any entries");
2152 		goto done;
2153 	}
2154 
2155 	TAILQ_FOREACH(pe, &paths, entry) {
2156 		struct got_tree_entry *te = pe->data;
2157 		char *path;
2158 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2159 			continue;
2160 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2161 			err = got_error_from_errno("asprintf");
2162 			goto done;
2163 		}
2164 		err = (*progress_cb)(progress_arg, path);
2165 		free(path);
2166 		if (err)
2167 			goto done;
2168 	}
2169 
2170 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2171 done:
2172 	if (dir)
2173 		closedir(dir);
2174 	got_pathlist_free(&paths);
2175 	return err;
2176 }
2177 
2178 const struct got_error *
2179 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2180     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2181     struct got_repository *repo, got_repo_import_cb progress_cb,
2182     void *progress_arg)
2183 {
2184 	const struct got_error *err;
2185 	struct got_object_id *new_tree_id;
2186 
2187 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2188 	    progress_cb, progress_arg);
2189 	if (err)
2190 		return err;
2191 
2192 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2193 	    author, time(NULL), author, time(NULL), logmsg, repo);
2194 	free(new_tree_id);
2195 	return err;
2196 }
2197 
2198 const struct got_error *
2199 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2200     struct got_repository *repo)
2201 {
2202 	const struct got_error *err = NULL;
2203 	char *path_objects = NULL, *path = NULL;
2204 	DIR *dir = NULL;
2205 	struct got_object_id id;
2206 	int i;
2207 
2208 	*nobjects = 0;
2209 	*ondisk_size = 0;
2210 
2211 	path_objects = got_repo_get_path_objects(repo);
2212 	if (path_objects == NULL)
2213 		return got_error_from_errno("got_repo_get_path_objects");
2214 
2215 	for (i = 0; i <= 0xff; i++) {
2216 		struct dirent *dent;
2217 
2218 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2219 			err = got_error_from_errno("asprintf");
2220 			break;
2221 		}
2222 
2223 		dir = opendir(path);
2224 		if (dir == NULL) {
2225 			if (errno == ENOENT) {
2226 				err = NULL;
2227 				continue;
2228 			}
2229 			err = got_error_from_errno2("opendir", path);
2230 			break;
2231 		}
2232 
2233 		while ((dent = readdir(dir)) != NULL) {
2234 			char *id_str;
2235 			int fd;
2236 			struct stat sb;
2237 
2238 			if (strcmp(dent->d_name, ".") == 0 ||
2239 			    strcmp(dent->d_name, "..") == 0)
2240 				continue;
2241 
2242 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2243 				err = got_error_from_errno("asprintf");
2244 				goto done;
2245 			}
2246 
2247 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2248 				free(id_str);
2249 				continue;
2250 			}
2251 			free(id_str);
2252 
2253 			err = got_object_open_loose_fd(&fd, &id, repo);
2254 			if (err)
2255 				goto done;
2256 
2257 			if (fstat(fd, &sb) == -1) {
2258 				err = got_error_from_errno("fstat");
2259 				close(fd);
2260 				goto done;
2261 			}
2262 			(*nobjects)++;
2263 			(*ondisk_size) += sb.st_size;
2264 
2265 			if (close(fd) == -1) {
2266 				err = got_error_from_errno("close");
2267 				goto done;
2268 			}
2269 		}
2270 
2271 		if (closedir(dir) != 0) {
2272 			err = got_error_from_errno("closedir");
2273 			goto done;
2274 		}
2275 		dir = NULL;
2276 
2277 		free(path);
2278 		path = NULL;
2279 	}
2280 done:
2281 	if (dir && closedir(dir) != 0 && err == NULL)
2282 		err = got_error_from_errno("closedir");
2283 
2284 	if (err) {
2285 		*nobjects = 0;
2286 		*ondisk_size = 0;
2287 	}
2288 	free(path_objects);
2289 	free(path);
2290 	return err;
2291 }
2292 
2293 const struct got_error *
2294 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2295     off_t *total_packsize, struct got_repository *repo)
2296 {
2297 	const struct got_error *err = NULL;
2298 	DIR *packdir = NULL;
2299 	struct dirent *dent;
2300 	struct got_packidx *packidx = NULL;
2301 	char *path_packidx;
2302 	char *path_packfile;
2303 	int packdir_fd;
2304 	struct stat sb;
2305 
2306 	*npackfiles = 0;
2307 	*nobjects = 0;
2308 	*total_packsize = 0;
2309 
2310 	packdir_fd = openat(got_repo_get_fd(repo),
2311 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2312 	if (packdir_fd == -1) {
2313 		return got_error_from_errno_fmt("openat: %s/%s",
2314 		    got_repo_get_path_git_dir(repo),
2315 		    GOT_OBJECTS_PACK_DIR);
2316 	}
2317 
2318 	packdir = fdopendir(packdir_fd);
2319 	if (packdir == NULL) {
2320 		err = got_error_from_errno("fdopendir");
2321 		goto done;
2322 	}
2323 
2324 	while ((dent = readdir(packdir)) != NULL) {
2325 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2326 			continue;
2327 
2328 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2329 		    dent->d_name) == -1) {
2330 			err = got_error_from_errno("asprintf");
2331 			goto done;
2332 		}
2333 
2334 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2335 		    path_packidx, 0);
2336 		free(path_packidx);
2337 		if (err)
2338 			goto done;
2339 
2340 		if (fstat(packidx->fd, &sb) == -1)
2341 			goto done;
2342 		*total_packsize += sb.st_size;
2343 
2344 		err = got_packidx_get_packfile_path(&path_packfile,
2345 		    packidx->path_packidx);
2346 		if (err)
2347 			goto done;
2348 
2349 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2350 		    0) == -1) {
2351 			free(path_packfile);
2352 			goto done;
2353 		}
2354 		free(path_packfile);
2355 		*total_packsize += sb.st_size;
2356 
2357 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2358 
2359 		(*npackfiles)++;
2360 
2361 		got_packidx_close(packidx);
2362 		packidx = NULL;
2363 	}
2364 done:
2365 	if (packidx)
2366 		got_packidx_close(packidx);
2367 	if (packdir && closedir(packdir) != 0 && err == NULL)
2368 		err = got_error_from_errno("closedir");
2369 	if (err) {
2370 		*npackfiles = 0;
2371 		*nobjects = 0;
2372 		*total_packsize = 0;
2373 	}
2374 	return err;
2375 }
2376 
2377 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2378     got_packidx_bloom_filter_cmp);