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 <sha2.h>
36 #include <string.h>
37 #include <time.h>
38 #include <unistd.h>
39 #include <zlib.h>
40 #include <errno.h>
41 #include <libgen.h>
42 #include <stdint.h>
43 #include <imsg.h>
44 #include <uuid.h>
45 
46 #include "bloom.h"
47 
48 #include "got_error.h"
49 #include "got_reference.h"
50 #include "got_repository.h"
51 #include "got_path.h"
52 #include "got_cancel.h"
53 #include "got_object.h"
54 #include "got_opentemp.h"
55 
56 #include "got_lib_delta.h"
57 #include "got_lib_delta_cache.h"
58 #include "got_lib_inflate.h"
59 #include "got_lib_object.h"
60 #include "got_lib_object_parse.h"
61 #include "got_lib_object_create.h"
62 #include "got_lib_pack.h"
63 #include "got_lib_privsep.h"
64 #include "got_lib_hash.h"
65 #include "got_lib_object_cache.h"
66 #include "got_lib_repository.h"
67 #include "got_lib_gotconfig.h"
68 
69 #ifndef nitems
70 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
71 #endif
72 
73 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
74 
75 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
76     got_packidx_bloom_filter_cmp);
77 
78 const char *
79 got_repo_get_path(struct got_repository *repo)
80 {
81 	return repo->path;
82 }
83 
84 const char *
85 got_repo_get_path_git_dir(struct got_repository *repo)
86 {
87 	return repo->path_git_dir;
88 }
89 
90 int
91 got_repo_get_fd(struct got_repository *repo)
92 {
93 	return repo->gitdir_fd;
94 }
95 
96 enum got_hash_algorithm
97 got_repo_get_hash_algorithm(struct got_repository *repo)
98 {
99 	return repo->algo;
100 }
101 
102 const char *
103 got_repo_get_gitconfig_author_name(struct got_repository *repo)
104 {
105 	return repo->gitconfig_author_name;
106 }
107 
108 const char *
109 got_repo_get_gitconfig_author_email(struct got_repository *repo)
110 {
111 	return repo->gitconfig_author_email;
112 }
113 
114 const char *
115 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
116 {
117 	return repo->global_gitconfig_author_name;
118 }
119 
120 const char *
121 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
122 {
123 	return repo->global_gitconfig_author_email;
124 }
125 
126 const char *
127 got_repo_get_gitconfig_owner(struct got_repository *repo)
128 {
129 	return repo->gitconfig_owner;
130 }
131 
132 void
133 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
134     struct got_repository *repo)
135 {
136 	*extensions = repo->extensions;
137 	*nextensions = repo->nextensions;
138 }
139 
140 int
141 got_repo_is_bare(struct got_repository *repo)
142 {
143 	return (strcmp(repo->path, repo->path_git_dir) == 0);
144 }
145 
146 static char *
147 get_path_git_child(struct got_repository *repo, const char *basename)
148 {
149 	char *path_child;
150 
151 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
152 	    basename) == -1)
153 		return NULL;
154 
155 	return path_child;
156 }
157 
158 char *
159 got_repo_get_path_objects(struct got_repository *repo)
160 {
161 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
162 }
163 
164 char *
165 got_repo_get_path_objects_pack(struct got_repository *repo)
166 {
167 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
168 }
169 
170 char *
171 got_repo_get_path_refs(struct got_repository *repo)
172 {
173 	return get_path_git_child(repo, GOT_REFS_DIR);
174 }
175 
176 char *
177 got_repo_get_path_packed_refs(struct got_repository *repo)
178 {
179 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
180 }
181 
182 static char *
183 get_path_head(struct got_repository *repo)
184 {
185 	return get_path_git_child(repo, GOT_HEAD_FILE);
186 }
187 
188 char *
189 got_repo_get_path_gitconfig(struct got_repository *repo)
190 {
191 	return get_path_git_child(repo, GOT_GITCONFIG);
192 }
193 
194 char *
195 got_repo_get_path_gotconfig(struct got_repository *repo)
196 {
197 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
198 }
199 
200 const struct got_gotconfig *
201 got_repo_get_gotconfig(struct got_repository *repo)
202 {
203 	return repo->gotconfig;
204 }
205 
206 void
207 got_repo_get_gitconfig_remotes(int *nremotes,
208     const struct got_remote_repo **remotes, struct got_repository *repo)
209 {
210 	*nremotes = repo->ngitconfig_remotes;
211 	*remotes = repo->gitconfig_remotes;
212 }
213 
214 static int
215 is_git_repo(struct got_repository *repo)
216 {
217 	const char *path_git = got_repo_get_path_git_dir(repo);
218 	char *path_objects = got_repo_get_path_objects(repo);
219 	char *path_refs = got_repo_get_path_refs(repo);
220 	char *path_head = get_path_head(repo);
221 	int ret = 0;
222 	struct stat sb;
223 	struct got_reference *head_ref;
224 
225 	if (lstat(path_git, &sb) == -1)
226 		goto done;
227 	if (!S_ISDIR(sb.st_mode))
228 		goto done;
229 
230 	if (lstat(path_objects, &sb) == -1)
231 		goto done;
232 	if (!S_ISDIR(sb.st_mode))
233 		goto done;
234 
235 	if (lstat(path_refs, &sb) == -1)
236 		goto done;
237 	if (!S_ISDIR(sb.st_mode))
238 		goto done;
239 
240 	if (lstat(path_head, &sb) == -1)
241 		goto done;
242 	if (!S_ISREG(sb.st_mode))
243 		goto done;
244 
245 	/* Check if the HEAD reference can be opened. */
246 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
247 		goto done;
248 	got_ref_close(head_ref);
249 
250 	ret = 1;
251 done:
252 	free(path_objects);
253 	free(path_refs);
254 	free(path_head);
255 	return ret;
256 
257 }
258 
259 static const struct got_error *
260 close_tempfiles(int *fds, size_t nfds)
261 {
262 	const struct got_error *err = NULL;
263 	int i;
264 
265 	for (i = 0; i < nfds; i++) {
266 		if (fds[i] == -1)
267 			continue;
268 		if (close(fds[i]) == -1) {
269 			err = got_error_from_errno("close");
270 			break;
271 		}
272 	}
273 	free(fds);
274 	return err;
275 }
276 
277 static const struct got_error *
278 open_tempfiles(int **fds, size_t array_size, size_t nfds)
279 {
280 	const struct got_error *err = NULL;
281 	int i;
282 
283 	*fds = calloc(array_size, sizeof(**fds));
284 	if (*fds == NULL)
285 		return got_error_from_errno("calloc");
286 
287 	for (i = 0; i < array_size; i++)
288 		(*fds)[i] = -1;
289 
290 	for (i = 0; i < nfds; i++) {
291 		(*fds)[i] = got_opentempfd();
292 		if ((*fds)[i] == -1) {
293 			err = got_error_from_errno("got_opentempfd");
294 			close_tempfiles(*fds, nfds);
295 			*fds = NULL;
296 			return err;
297 		}
298 	}
299 
300 	return NULL;
301 }
302 
303 static const struct got_error *
304 get_pack_cache_size(int *pack_cache_size)
305 {
306 	struct rlimit rl;
307 
308 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
309 		return got_error_from_errno("getrlimit");
310 
311 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
312 	if (*pack_cache_size > rl.rlim_cur / 8)
313 		*pack_cache_size = rl.rlim_cur / 8;
314 
315 	return NULL;
316 }
317 
318 const struct got_error *
319 got_repo_pack_fds_open(int **pack_fds)
320 {
321 	const struct got_error *err;
322 	int nfds;
323 
324 	err = get_pack_cache_size(&nfds);
325 	if (err)
326 		return err;
327 
328 	/*
329 	 * We need one basefd and one accumfd per cached pack.
330 	 * Our constants should be set up in a way such that
331 	 * this error never triggers.
332 	 */
333 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
334 		return got_error(GOT_ERR_NO_SPACE);
335 
336 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
337 }
338 
339 const struct got_error *
340 got_repo_pack_fds_close(int *pack_fds)
341 {
342 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
343 }
344 
345 const struct got_error *
346 got_repo_temp_fds_open(int **temp_fds)
347 {
348 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
349 	    GOT_REPO_NUM_TEMPFILES);
350 }
351 
352 void
353 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
354 {
355 	int i;
356 
357 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
358 		repo->tempfiles[i] = temp_fds[i];
359 }
360 
361 const struct got_error *
362 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
363 {
364 	int i;
365 
366 	*fd = -1;
367 	*idx = -1;
368 
369 	for (i = 0; i < nitems(repo->tempfiles); i++) {
370 		if (repo->tempfile_use_mask & (1 << i))
371 			continue;
372 		if (repo->tempfiles[i] != -1) {
373 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
374 				return got_error_from_errno("ftruncate");
375 			*fd = repo->tempfiles[i];
376 			*idx = i;
377 			repo->tempfile_use_mask |= (1 << i);
378 			return NULL;
379 		}
380 	}
381 
382 	return got_error(GOT_ERR_REPO_TEMPFILE);
383 }
384 
385 void
386 got_repo_temp_fds_put(int idx, struct got_repository *repo)
387 {
388 	repo->tempfile_use_mask &= ~(1 << idx);
389 }
390 
391 const struct got_error *
392 got_repo_temp_fds_close(int *temp_fds)
393 {
394 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
395 }
396 
397 const struct got_error *
398 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
399     struct got_object *obj)
400 {
401 #ifndef GOT_NO_OBJ_CACHE
402 	const struct got_error *err = NULL;
403 	err = got_object_cache_add(&repo->objcache, id, obj);
404 	if (err) {
405 		if (err->code == GOT_ERR_OBJ_EXISTS ||
406 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
407 			err = NULL;
408 		return err;
409 	}
410 	obj->refcnt++;
411 #endif
412 	return NULL;
413 }
414 
415 struct got_object *
416 got_repo_get_cached_object(struct got_repository *repo,
417     struct got_object_id *id)
418 {
419 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
420 }
421 
422 const struct got_error *
423 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
424     struct got_tree_object *tree)
425 {
426 #ifndef GOT_NO_OBJ_CACHE
427 	const struct got_error *err = NULL;
428 	err = got_object_cache_add(&repo->treecache, id, tree);
429 	if (err) {
430 		if (err->code == GOT_ERR_OBJ_EXISTS ||
431 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
432 			err = NULL;
433 		return err;
434 	}
435 	tree->refcnt++;
436 #endif
437 	return NULL;
438 }
439 
440 struct got_tree_object *
441 got_repo_get_cached_tree(struct got_repository *repo,
442     struct got_object_id *id)
443 {
444 	return (struct got_tree_object *)got_object_cache_get(
445 	    &repo->treecache, id);
446 }
447 
448 const struct got_error *
449 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
450     struct got_commit_object *commit)
451 {
452 #ifndef GOT_NO_OBJ_CACHE
453 	const struct got_error *err = NULL;
454 	err = got_object_cache_add(&repo->commitcache, id, commit);
455 	if (err) {
456 		if (err->code == GOT_ERR_OBJ_EXISTS ||
457 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
458 			err = NULL;
459 		return err;
460 	}
461 	commit->refcnt++;
462 #endif
463 	return NULL;
464 }
465 
466 struct got_commit_object *
467 got_repo_get_cached_commit(struct got_repository *repo,
468     struct got_object_id *id)
469 {
470 	return (struct got_commit_object *)got_object_cache_get(
471 	    &repo->commitcache, id);
472 }
473 
474 const struct got_error *
475 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
476     struct got_tag_object *tag)
477 {
478 #ifndef GOT_NO_OBJ_CACHE
479 	const struct got_error *err = NULL;
480 	err = got_object_cache_add(&repo->tagcache, id, tag);
481 	if (err) {
482 		if (err->code == GOT_ERR_OBJ_EXISTS ||
483 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
484 			err = NULL;
485 		return err;
486 	}
487 	tag->refcnt++;
488 #endif
489 	return NULL;
490 }
491 
492 struct got_tag_object *
493 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
494 {
495 	return (struct got_tag_object *)got_object_cache_get(
496 	    &repo->tagcache, id);
497 }
498 
499 const struct got_error *
500 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
501     struct got_raw_object *raw)
502 {
503 #ifndef GOT_NO_OBJ_CACHE
504 	const struct got_error *err = NULL;
505 	err = got_object_cache_add(&repo->rawcache, id, raw);
506 	if (err) {
507 		if (err->code == GOT_ERR_OBJ_EXISTS ||
508 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
509 			err = NULL;
510 		return err;
511 	}
512 	raw->refcnt++;
513 #endif
514 	return NULL;
515 }
516 
517 
518 struct got_raw_object *
519 got_repo_get_cached_raw_object(struct got_repository *repo,
520     struct got_object_id *id)
521 {
522 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
523 }
524 
525 
526 static const struct got_error *
527 open_repo(struct got_repository *repo, const char *path)
528 {
529 	const struct got_error *err = NULL;
530 
531 	repo->gitdir_fd = -1;
532 
533 	/* bare git repository? */
534 	repo->path_git_dir = strdup(path);
535 	if (repo->path_git_dir == NULL)
536 		return got_error_from_errno("strdup");
537 	if (is_git_repo(repo)) {
538 		repo->path = strdup(repo->path_git_dir);
539 		if (repo->path == NULL) {
540 			err = got_error_from_errno("strdup");
541 			goto done;
542 		}
543 		repo->gitdir_fd = open(repo->path_git_dir,
544 		    O_DIRECTORY | O_CLOEXEC);
545 		if (repo->gitdir_fd == -1) {
546 			err = got_error_from_errno2("open",
547 			    repo->path_git_dir);
548 			goto done;
549 		}
550 		return NULL;
551 	}
552 
553 	/* git repository with working tree? */
554 	free(repo->path_git_dir);
555 	repo->path_git_dir = NULL;
556 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
557 		err = got_error_from_errno("asprintf");
558 		goto done;
559 	}
560 	if (is_git_repo(repo)) {
561 		repo->path = strdup(path);
562 		if (repo->path == NULL) {
563 			err = got_error_from_errno("strdup");
564 			goto done;
565 		}
566 		repo->gitdir_fd = open(repo->path_git_dir,
567 		    O_DIRECTORY | O_CLOEXEC);
568 		if (repo->gitdir_fd == -1) {
569 			err = got_error_from_errno2("open",
570 			    repo->path_git_dir);
571 			goto done;
572 		}
573 		return NULL;
574 	}
575 
576 	err = got_error(GOT_ERR_NOT_GIT_REPO);
577 done:
578 	if (err) {
579 		free(repo->path);
580 		repo->path = NULL;
581 		free(repo->path_git_dir);
582 		repo->path_git_dir = NULL;
583 		if (repo->gitdir_fd != -1)
584 			close(repo->gitdir_fd);
585 		repo->gitdir_fd = -1;
586 
587 	}
588 	return err;
589 }
590 
591 static const struct got_error *
592 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
593 {
594 	const struct got_error *err = NULL;
595 	char *repo_gitconfig_path = NULL;
596 	char *object_format = NULL;
597 
598 	if (global_gitconfig_path) {
599 		/* Read settings from ~/.gitconfig. */
600 		int dummy_repo_version;
601 		err = got_repo_read_gitconfig(&dummy_repo_version,
602 		    &repo->global_gitconfig_author_name,
603 		    &repo->global_gitconfig_author_email,
604 		    NULL, NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
605 		if (err)
606 			return err;
607 	}
608 
609 	/* Read repository's .git/config file. */
610 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
611 	if (repo_gitconfig_path == NULL)
612 		return got_error_from_errno("got_repo_get_path_gitconfig");
613 
614 	err = got_repo_read_gitconfig(
615 	    &repo->gitconfig_repository_format_version,
616 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
617 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
618 	    &repo->gitconfig_owner, &object_format,
619 	    &repo->extensions, &repo->nextensions,
620 	    repo_gitconfig_path);
621 	if (err)
622 		goto done;
623 
624 	if (!object_format)
625 		repo->algo = GOT_HASH_SHA1;
626 	else {
627 		if (!strcmp(object_format, "sha256"))
628 			repo->algo = GOT_HASH_SHA256;
629 		else {
630 			err = got_error_msg(GOT_ERR_OBJECT_FORMAT,
631 			    object_format);
632 			goto done;
633 		}
634 	}
635 
636 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
637 		int i;
638 
639 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
640 			got_repo_free_remote_repo_data(
641 			    &repo->gitconfig_remotes[i]);
642 		}
643 		free(repo->gitconfig_remotes);
644 		repo->gitconfig_remotes = NULL;
645 		repo->ngitconfig_remotes = 0;
646 
647 		free(repo->gitconfig_author_name);
648 		repo->gitconfig_author_name = NULL;
649 		free(repo->gitconfig_author_email);
650 		repo->gitconfig_author_email = NULL;
651 
652 		free(repo->global_gitconfig_author_name);
653 		repo->global_gitconfig_author_name = NULL;
654 		free(repo->global_gitconfig_author_email);
655 		repo->global_gitconfig_author_email = NULL;
656 	}
657 
658 done:
659 	free(object_format);
660 	free(repo_gitconfig_path);
661 	return err;
662 }
663 
664 static const struct got_error *
665 read_gotconfig(struct got_repository *repo)
666 {
667 	const struct got_error *err = NULL;
668 	char *gotconfig_path;
669 
670 	gotconfig_path = got_repo_get_path_gotconfig(repo);
671 	if (gotconfig_path == NULL)
672 		return got_error_from_errno("got_repo_get_path_gotconfig");
673 
674 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
675 	free(gotconfig_path);
676 	return err;
677 }
678 
679 /* Supported repository format extensions. */
680 static const char *const repo_extensions[] = {
681 	"noop",			/* Got supports repository format version 1. */
682 	"preciousObjects",	/* Supported by gotadmin cleanup. */
683 	"worktreeConfig",	/* Got does not care about Git work trees. */
684 };
685 
686 const struct got_error *
687 got_repo_open(struct got_repository **repop, const char *path,
688     const char *global_gitconfig_path, int *pack_fds)
689 {
690 	struct got_repository *repo = NULL;
691 	const struct got_error *err = NULL;
692 	char *repo_path = NULL;
693 	size_t i, j = 0;
694 
695 	*repop = NULL;
696 
697 	repo = calloc(1, sizeof(*repo));
698 	if (repo == NULL)
699 		return got_error_from_errno("calloc");
700 
701 	RB_INIT(&repo->packidx_bloom_filters);
702 	TAILQ_INIT(&repo->packidx_paths);
703 
704 	for (i = 0; i < nitems(repo->privsep_children); i++) {
705 		memset(&repo->privsep_children[i], 0,
706 		    sizeof(repo->privsep_children[0]));
707 		repo->privsep_children[i].imsg_fd = -1;
708 	}
709 
710 	err = got_object_cache_init(&repo->objcache,
711 	    GOT_OBJECT_CACHE_TYPE_OBJ);
712 	if (err)
713 		goto done;
714 	err = got_object_cache_init(&repo->treecache,
715 	    GOT_OBJECT_CACHE_TYPE_TREE);
716 	if (err)
717 		goto done;
718 	err = got_object_cache_init(&repo->commitcache,
719 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
720 	if (err)
721 		goto done;
722 	err = got_object_cache_init(&repo->tagcache,
723 	    GOT_OBJECT_CACHE_TYPE_TAG);
724 	if (err)
725 		goto done;
726 	err = got_object_cache_init(&repo->rawcache,
727 	    GOT_OBJECT_CACHE_TYPE_RAW);
728 	if (err)
729 		goto done;
730 
731 	err = get_pack_cache_size(&repo->pack_cache_size);
732 	if (err)
733 		goto done;
734 	for (i = 0; i < nitems(repo->packs); i++) {
735 		if (pack_fds != NULL && i < repo->pack_cache_size) {
736 			repo->packs[i].basefd = pack_fds[j++];
737 			repo->packs[i].accumfd = pack_fds[j++];
738 		} else {
739 			repo->packs[i].basefd = -1;
740 			repo->packs[i].accumfd = -1;
741 		}
742 	}
743 	for (i = 0; i < nitems(repo->tempfiles); i++)
744 		repo->tempfiles[i] = -1;
745 	repo->pinned_pack = -1;
746 	repo->pinned_packidx = -1;
747 	repo->pinned_pid = 0;
748 
749 	repo_path = realpath(path, NULL);
750 	if (repo_path == NULL) {
751 		err = got_error_from_errno2("realpath", path);
752 		goto done;
753 	}
754 
755 	for (;;) {
756 		char *parent_path;
757 
758 		err = open_repo(repo, repo_path);
759 		if (err == NULL)
760 			break;
761 		if (err->code != GOT_ERR_NOT_GIT_REPO)
762 			goto done;
763 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
764 			err = got_error(GOT_ERR_NOT_GIT_REPO);
765 			goto done;
766 		}
767 		err = got_path_dirname(&parent_path, repo_path);
768 		if (err)
769 			goto done;
770 		free(repo_path);
771 		repo_path = parent_path;
772 	}
773 
774 	err = read_gotconfig(repo);
775 	if (err)
776 		goto done;
777 
778 	err = read_gitconfig(repo, global_gitconfig_path);
779 	if (err)
780 		goto done;
781 	if (repo->gitconfig_repository_format_version > 1) {
782 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
783 		goto done;
784 	}
785 	if (repo->gitconfig_repository_format_version == 0 &&
786 	    repo->algo != GOT_HASH_SHA1) {
787 		err = got_error(GOT_ERR_OBJECT_FORMAT);
788 		goto done;
789 	}
790 	for (i = 0; i < repo->nextensions; i++) {
791 		char *ext = repo->extensions[i];
792 		int j, supported = 0;
793 		for (j = 0; j < nitems(repo_extensions); j++) {
794 			if (strcmp(ext, repo_extensions[j]) == 0) {
795 				supported = 1;
796 				break;
797 			}
798 		}
799 		if (!supported) {
800 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
801 			goto done;
802 		}
803 	}
804 
805 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
806 done:
807 	if (err)
808 		got_repo_close(repo);
809 	else
810 		*repop = repo;
811 	free(repo_path);
812 	return err;
813 }
814 
815 const struct got_error *
816 got_repo_close(struct got_repository *repo)
817 {
818 	const struct got_error *err = NULL, *child_err;
819 	struct got_packidx_bloom_filter *bf;
820 	size_t i;
821 
822 	for (i = 0; i < repo->pack_cache_size; i++) {
823 		if (repo->packidx_cache[i] == NULL)
824 			break;
825 		got_packidx_close(repo->packidx_cache[i]);
826 	}
827 
828 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
829 	    &repo->packidx_bloom_filters))) {
830 		RB_REMOVE(got_packidx_bloom_filter_tree,
831 		    &repo->packidx_bloom_filters, bf);
832 		bloom_free(bf->bloom);
833 		free(bf->bloom);
834 		free(bf);
835 	}
836 
837 	for (i = 0; i < repo->pack_cache_size; i++)
838 		if (repo->packs[i].path_packfile)
839 			if (repo->packs[i].path_packfile)
840 				got_pack_close(&repo->packs[i]);
841 
842 	free(repo->path);
843 	free(repo->path_git_dir);
844 
845 	got_object_cache_close(&repo->objcache);
846 	got_object_cache_close(&repo->treecache);
847 	got_object_cache_close(&repo->commitcache);
848 	got_object_cache_close(&repo->tagcache);
849 	got_object_cache_close(&repo->rawcache);
850 
851 	for (i = 0; i < nitems(repo->privsep_children); i++) {
852 		if (repo->privsep_children[i].imsg_fd == -1)
853 			continue;
854 		imsg_clear(repo->privsep_children[i].ibuf);
855 		free(repo->privsep_children[i].ibuf);
856 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
857 		child_err = got_privsep_wait_for_child(
858 		    repo->privsep_children[i].pid);
859 		if (child_err && err == NULL)
860 			err = child_err;
861 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
862 		    err == NULL)
863 			err = got_error_from_errno("close");
864 	}
865 
866 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
867 	    err == NULL)
868 		err = got_error_from_errno("close");
869 
870 	if (repo->gotconfig)
871 		got_gotconfig_free(repo->gotconfig);
872 	free(repo->gitconfig_author_name);
873 	free(repo->gitconfig_author_email);
874 	for (i = 0; i < repo->ngitconfig_remotes; i++)
875 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
876 	free(repo->gitconfig_remotes);
877 	for (i = 0; i < repo->nextensions; i++)
878 		free(repo->extensions[i]);
879 	free(repo->extensions);
880 
881 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
882 	free(repo);
883 
884 	return err;
885 }
886 
887 void
888 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
889 {
890 	int i;
891 
892 	free(repo->name);
893 	repo->name = NULL;
894 	free(repo->fetch_url);
895 	repo->fetch_url = NULL;
896 	free(repo->send_url);
897 	repo->send_url = NULL;
898 	for (i = 0; i < repo->nfetch_branches; i++)
899 		free(repo->fetch_branches[i]);
900 	free(repo->fetch_branches);
901 	repo->fetch_branches = NULL;
902 	repo->nfetch_branches = 0;
903 	for (i = 0; i < repo->nsend_branches; i++)
904 		free(repo->send_branches[i]);
905 	free(repo->send_branches);
906 	repo->send_branches = NULL;
907 	repo->nsend_branches = 0;
908 }
909 
910 const struct got_error *
911 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
912     const char *input_path)
913 {
914 	const struct got_error *err = NULL;
915 	const char *repo_abspath = NULL;
916 	size_t repolen, len;
917 	char *canonpath, *path = NULL;
918 
919 	*in_repo_path = NULL;
920 
921 	canonpath = strdup(input_path);
922 	if (canonpath == NULL) {
923 		err = got_error_from_errno("strdup");
924 		goto done;
925 	}
926 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
927 	if (err)
928 		goto done;
929 
930 	repo_abspath = got_repo_get_path(repo);
931 
932 	if (canonpath[0] == '\0') {
933 		path = strdup(canonpath);
934 		if (path == NULL) {
935 			err = got_error_from_errno("strdup");
936 			goto done;
937 		}
938 	} else {
939 		path = realpath(canonpath, NULL);
940 		if (path == NULL) {
941 			if (errno != ENOENT) {
942 				err = got_error_from_errno2("realpath",
943 				    canonpath);
944 				goto done;
945 			}
946 			/*
947 			 * Path is not on disk.
948 			 * Assume it is already relative to repository root.
949 			 */
950 			path = strdup(canonpath);
951 			if (path == NULL) {
952 				err = got_error_from_errno("strdup");
953 				goto done;
954 			}
955 		}
956 
957 		repolen = strlen(repo_abspath);
958 		len = strlen(path);
959 
960 
961 		if (strcmp(path, repo_abspath) == 0) {
962 			free(path);
963 			path = strdup("");
964 			if (path == NULL) {
965 				err = got_error_from_errno("strdup");
966 				goto done;
967 			}
968 		} else if (len > repolen &&
969 		    got_path_is_child(path, repo_abspath, repolen)) {
970 			/* Matched an on-disk path inside repository. */
971 			if (got_repo_is_bare(repo)) {
972 				/*
973 				 * Matched an on-disk path inside repository
974 				 * database. Treat input as repository-relative.
975 				 */
976 				free(path);
977 				path = canonpath;
978 				canonpath = NULL;
979 			} else {
980 				char *child;
981 				/* Strip common prefix with repository path. */
982 				err = got_path_skip_common_ancestor(&child,
983 				    repo_abspath, path);
984 				if (err)
985 					goto done;
986 				free(path);
987 				path = child;
988 			}
989 		} else {
990 			/*
991 			 * Matched unrelated on-disk path.
992 			 * Treat input as repository-relative.
993 			 */
994 			free(path);
995 			path = canonpath;
996 			canonpath = NULL;
997 		}
998 	}
999 
1000 	/* Make in-repository path absolute */
1001 	if (path[0] != '/') {
1002 		char *abspath;
1003 		if (asprintf(&abspath, "/%s", path) == -1) {
1004 			err = got_error_from_errno("asprintf");
1005 			goto done;
1006 		}
1007 		free(path);
1008 		path = abspath;
1009 	}
1010 
1011 done:
1012 	free(canonpath);
1013 	if (err)
1014 		free(path);
1015 	else
1016 		*in_repo_path = path;
1017 	return err;
1018 }
1019 
1020 static const struct got_error *
1021 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1022     const char *path_packidx)
1023 {
1024 	const struct got_error *err = NULL;
1025 	size_t i;
1026 
1027 	for (i = 0; i < repo->pack_cache_size; i++) {
1028 		if (repo->packidx_cache[i] == NULL)
1029 			break;
1030 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1031 		    path_packidx) == 0) {
1032 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1033 		}
1034 	}
1035 	if (i == repo->pack_cache_size) {
1036 		do {
1037 			i--;
1038 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1039 		    i == repo->pinned_packidx);
1040 		err = got_packidx_close(repo->packidx_cache[i]);
1041 		if (err)
1042 			return err;
1043 	}
1044 
1045 	repo->packidx_cache[i] = packidx;
1046 
1047 	return NULL;
1048 }
1049 
1050 int
1051 got_repo_is_packidx_filename(const char *name, size_t len)
1052 {
1053 	if (len != GOT_PACKIDX_NAMELEN)
1054 		return 0;
1055 
1056 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1057 		return 0;
1058 
1059 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1060 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1061 		return 0;
1062 
1063 	return 1;
1064 }
1065 
1066 static struct got_packidx_bloom_filter *
1067 get_packidx_bloom_filter(struct got_repository *repo,
1068     const char *path, size_t path_len)
1069 {
1070 	struct got_packidx_bloom_filter key;
1071 
1072 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1073 		return NULL; /* XXX */
1074 	key.path_len = path_len;
1075 
1076 	return RB_FIND(got_packidx_bloom_filter_tree,
1077 	    &repo->packidx_bloom_filters, &key);
1078 }
1079 
1080 int
1081 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1082     const char *path_packidx, struct got_object_id *id)
1083 {
1084 	struct got_packidx_bloom_filter *bf;
1085 
1086 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1087 	if (bf)
1088 		return bloom_check(bf->bloom, id->hash, sizeof(id->hash));
1089 
1090 	/* No bloom filter means this pack index must be searched. */
1091 	return 1;
1092 }
1093 
1094 static const struct got_error *
1095 add_packidx_bloom_filter(struct got_repository *repo,
1096     struct got_packidx *packidx, const char *path_packidx)
1097 {
1098 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1099 	struct got_packidx_bloom_filter *bf;
1100 	size_t len;
1101 
1102 	/*
1103 	 * Don't use bloom filters for very large pack index files.
1104 	 * Large pack files will contain a relatively large fraction
1105 	 * of our objects so we will likely need to visit them anyway.
1106 	 * The more objects a pack file contains the higher the probability
1107 	 * of a false-positive match from the bloom filter. And reading
1108 	 * all object IDs from a large pack index file can be expensive.
1109 	 */
1110 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1111 		return NULL;
1112 
1113 	/* Do we already have a filter for this pack index? */
1114 	if (get_packidx_bloom_filter(repo, path_packidx,
1115 	    strlen(path_packidx)) != NULL)
1116 		return NULL;
1117 
1118 	bf = calloc(1, sizeof(*bf));
1119 	if (bf == NULL)
1120 		return got_error_from_errno("calloc");
1121 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1122 	if (bf->bloom == NULL) {
1123 		free(bf);
1124 		return got_error_from_errno("calloc");
1125 	}
1126 
1127 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1128 	if (len >= sizeof(bf->path)) {
1129 		free(bf->bloom);
1130 		free(bf);
1131 		return got_error(GOT_ERR_NO_SPACE);
1132 	}
1133 	bf->path_len = len;
1134 
1135 	/* Minimum size supported by our bloom filter is 1000 entries. */
1136 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1137 	for (i = 0; i < nobjects; i++) {
1138 		struct got_packidx_object_id *id;
1139 		id = &packidx->hdr.sorted_ids[i];
1140 		bloom_add(bf->bloom, id->hash, sizeof(id->hash));
1141 	}
1142 
1143 	RB_INSERT(got_packidx_bloom_filter_tree,
1144 	    &repo->packidx_bloom_filters, bf);
1145 	return NULL;
1146 }
1147 
1148 static void
1149 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1150 {
1151 	struct got_pathlist_entry *pe;
1152 
1153 	while (!TAILQ_EMPTY(packidx_paths)) {
1154 		pe = TAILQ_FIRST(packidx_paths);
1155 		TAILQ_REMOVE(packidx_paths, pe, entry);
1156 		free((char *)pe->path);
1157 		free(pe);
1158 	}
1159 }
1160 
1161 static const struct got_error *
1162 refresh_packidx_paths(struct got_repository *repo)
1163 {
1164 	const struct got_error *err = NULL;
1165 	char *objects_pack_dir = NULL;
1166 	struct stat sb;
1167 
1168 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1169 	if (objects_pack_dir == NULL)
1170 		return got_error_from_errno("got_repo_get_path_objects_pack");
1171 
1172 	if (stat(objects_pack_dir, &sb) == -1) {
1173 		if (errno != ENOENT) {
1174 			err = got_error_from_errno2("stat", objects_pack_dir);
1175 			goto done;
1176 		}
1177 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1178 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1179 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1180 		purge_packidx_paths(&repo->packidx_paths);
1181 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1182 		if (err)
1183 			goto done;
1184 	}
1185 done:
1186 	free(objects_pack_dir);
1187 	return err;
1188 }
1189 
1190 const struct got_error *
1191 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1192     struct got_repository *repo, struct got_object_id *id)
1193 {
1194 	const struct got_error *err;
1195 	struct got_pathlist_entry *pe;
1196 	size_t i;
1197 
1198 	/* Search pack index cache. */
1199 	for (i = 0; i < repo->pack_cache_size; i++) {
1200 		if (repo->packidx_cache[i] == NULL)
1201 			break;
1202 		if (!got_repo_check_packidx_bloom_filter(repo,
1203 		    repo->packidx_cache[i]->path_packidx, id))
1204 			continue; /* object will not be found in this index */
1205 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1206 		if (*idx != -1) {
1207 			*packidx = repo->packidx_cache[i];
1208 			/*
1209 			 * Move this cache entry to the front. Repeatedly
1210 			 * searching a wrong pack index can be expensive.
1211 			 */
1212 			if (i > 0) {
1213 				memmove(&repo->packidx_cache[1],
1214 				    &repo->packidx_cache[0],
1215 				    i * sizeof(repo->packidx_cache[0]));
1216 				repo->packidx_cache[0] = *packidx;
1217 				if (repo->pinned_packidx >= 0 &&
1218 				    repo->pinned_packidx < i)
1219 					repo->pinned_packidx++;
1220 				else if (repo->pinned_packidx == i)
1221 					repo->pinned_packidx = 0;
1222 			}
1223 			return NULL;
1224 		}
1225 	}
1226 	/* No luck. Search the filesystem. */
1227 
1228 	err = refresh_packidx_paths(repo);
1229 	if (err)
1230 		return err;
1231 
1232 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1233 		const char *path_packidx = pe->path;
1234 		int is_cached = 0;
1235 
1236 		if (!got_repo_check_packidx_bloom_filter(repo,
1237 		    pe->path, id))
1238 			continue; /* object will not be found in this index */
1239 
1240 		for (i = 0; i < repo->pack_cache_size; i++) {
1241 			if (repo->packidx_cache[i] == NULL)
1242 				break;
1243 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1244 			    path_packidx) == 0) {
1245 				is_cached = 1;
1246 				break;
1247 			}
1248 		}
1249 		if (is_cached)
1250 			continue; /* already searched */
1251 
1252 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1253 		    path_packidx, 0);
1254 		if (err)
1255 			goto done;
1256 
1257 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1258 		if (err)
1259 			goto done;
1260 
1261 		err = cache_packidx(repo, *packidx, path_packidx);
1262 		if (err)
1263 			goto done;
1264 
1265 		*idx = got_packidx_get_object_idx(*packidx, id);
1266 		if (*idx != -1) {
1267 			err = NULL; /* found the object */
1268 			goto done;
1269 		}
1270 	}
1271 
1272 	err = got_error_no_obj(id);
1273 done:
1274 	return err;
1275 }
1276 
1277 const struct got_error *
1278 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1279     struct got_repository *repo)
1280 {
1281 	const struct got_error *err = NULL;
1282 	DIR *packdir = NULL;
1283 	struct dirent *dent;
1284 	char *path_packidx = NULL;
1285 	int packdir_fd;
1286 	struct stat sb;
1287 
1288 	packdir_fd = openat(got_repo_get_fd(repo),
1289 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1290 	if (packdir_fd == -1) {
1291 		return got_error_from_errno_fmt("openat: %s/%s",
1292 		    got_repo_get_path_git_dir(repo),
1293 		    GOT_OBJECTS_PACK_DIR);
1294 	}
1295 
1296 	packdir = fdopendir(packdir_fd);
1297 	if (packdir == NULL) {
1298 		err = got_error_from_errno("fdopendir");
1299 		goto done;
1300 	}
1301 
1302 	if (fstat(packdir_fd, &sb) == -1) {
1303 		err = got_error_from_errno("fstat");
1304 		goto done;
1305 	}
1306 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1307 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1308 
1309 	while ((dent = readdir(packdir)) != NULL) {
1310 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1311 			continue;
1312 
1313 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1314 		    dent->d_name) == -1) {
1315 			err = got_error_from_errno("asprintf");
1316 			path_packidx = NULL;
1317 			break;
1318 		}
1319 
1320 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1321 		if (err)
1322 			break;
1323 	}
1324 done:
1325 	if (err)
1326 		free(path_packidx);
1327 	if (packdir && closedir(packdir) != 0 && err == NULL)
1328 		err = got_error_from_errno("closedir");
1329 	return err;
1330 }
1331 
1332 const struct got_error *
1333 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1334     struct got_repository *repo)
1335 {
1336 	const struct got_error *err;
1337 	size_t i;
1338 
1339 	*packidx = NULL;
1340 
1341 	/* Search pack index cache. */
1342 	for (i = 0; i < repo->pack_cache_size; i++) {
1343 		if (repo->packidx_cache[i] == NULL)
1344 			break;
1345 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1346 		    path_packidx) == 0) {
1347 			*packidx = repo->packidx_cache[i];
1348 			return NULL;
1349 		}
1350 	}
1351 	/* No luck. Search the filesystem. */
1352 
1353 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1354 	    path_packidx, 0);
1355 	if (err)
1356 		return err;
1357 
1358 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1359 	if (err)
1360 		goto done;
1361 
1362 	err = cache_packidx(repo, *packidx, path_packidx);
1363 done:
1364 	if (err) {
1365 		got_packidx_close(*packidx);
1366 		*packidx = NULL;
1367 	}
1368 	return err;
1369 }
1370 
1371 static const struct got_error *
1372 read_packfile_hdr(int fd, struct got_packidx *packidx)
1373 {
1374 	const struct got_error *err = NULL;
1375 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1376 	struct got_packfile_hdr hdr;
1377 	ssize_t n;
1378 
1379 	n = read(fd, &hdr, sizeof(hdr));
1380 	if (n < 0)
1381 		return got_error_from_errno("read");
1382 	if (n != sizeof(hdr))
1383 		return got_error(GOT_ERR_BAD_PACKFILE);
1384 
1385 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1386 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1387 	    be32toh(hdr.nobjects) != totobj)
1388 		err = got_error(GOT_ERR_BAD_PACKFILE);
1389 
1390 	return err;
1391 }
1392 
1393 static const struct got_error *
1394 open_packfile(int *fd, struct got_repository *repo,
1395     const char *relpath, struct got_packidx *packidx)
1396 {
1397 	const struct got_error *err = NULL;
1398 
1399 	*fd = openat(got_repo_get_fd(repo), relpath,
1400 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1401 	if (*fd == -1)
1402 		return got_error_from_errno_fmt("openat: %s/%s",
1403 		    got_repo_get_path_git_dir(repo), relpath);
1404 
1405 	if (packidx) {
1406 		err = read_packfile_hdr(*fd, packidx);
1407 		if (err) {
1408 			close(*fd);
1409 			*fd = -1;
1410 		}
1411 	}
1412 
1413 	return err;
1414 }
1415 
1416 const struct got_error *
1417 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1418     const char *path_packfile, struct got_packidx *packidx)
1419 {
1420 	const struct got_error *err = NULL;
1421 	struct got_pack *pack = NULL;
1422 	struct stat sb;
1423 	size_t i;
1424 
1425 	if (packp)
1426 		*packp = NULL;
1427 
1428 	for (i = 0; i < repo->pack_cache_size; i++) {
1429 		pack = &repo->packs[i];
1430 		if (pack->path_packfile == NULL)
1431 			break;
1432 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1433 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1434 	}
1435 
1436 	if (i == repo->pack_cache_size) {
1437 		struct got_pack tmp;
1438 		do {
1439 			i--;
1440 		} while (i > 0 && repo->pinned_pack >= 0 &&
1441 		    i == repo->pinned_pack);
1442 		err = got_pack_close(&repo->packs[i]);
1443 		if (err)
1444 			return err;
1445 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1446 			return got_error_from_errno("ftruncate");
1447 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1448 			return got_error_from_errno("ftruncate");
1449 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1450 		memcpy(&repo->packs[i], &repo->packs[0],
1451 		    sizeof(repo->packs[i]));
1452 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1453 		if (repo->pinned_pack == 0)
1454 			repo->pinned_pack = i;
1455 		else if (repo->pinned_pack == i)
1456 			repo->pinned_pack = 0;
1457 		i = 0;
1458 	}
1459 
1460 	pack = &repo->packs[i];
1461 
1462 	pack->path_packfile = strdup(path_packfile);
1463 	if (pack->path_packfile == NULL) {
1464 		err = got_error_from_errno("strdup");
1465 		goto done;
1466 	}
1467 
1468 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1469 	if (err)
1470 		goto done;
1471 
1472 	if (fstat(pack->fd, &sb) != 0) {
1473 		err = got_error_from_errno("fstat");
1474 		goto done;
1475 	}
1476 	pack->filesize = sb.st_size;
1477 
1478 	pack->privsep_child = NULL;
1479 
1480 	err = got_delta_cache_alloc(&pack->delta_cache);
1481 	if (err)
1482 		goto done;
1483 
1484 #ifndef GOT_PACK_NO_MMAP
1485 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1486 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1487 		    pack->fd, 0);
1488 		if (pack->map == MAP_FAILED) {
1489 			if (errno != ENOMEM) {
1490 				err = got_error_from_errno("mmap");
1491 				goto done;
1492 			}
1493 			pack->map = NULL; /* fall back to read(2) */
1494 		}
1495 	}
1496 #endif
1497 done:
1498 	if (err) {
1499 		if (pack)
1500 			got_pack_close(pack);
1501 	} else if (packp)
1502 		*packp = pack;
1503 	return err;
1504 }
1505 
1506 struct got_pack *
1507 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1508 {
1509 	struct got_pack *pack = NULL;
1510 	size_t i;
1511 
1512 	for (i = 0; i < repo->pack_cache_size; i++) {
1513 		pack = &repo->packs[i];
1514 		if (pack->path_packfile == NULL)
1515 			break;
1516 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1517 			return pack;
1518 	}
1519 
1520 	return NULL;
1521 }
1522 
1523 const struct got_error *
1524 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1525     struct got_pack *pack)
1526 {
1527 	size_t i;
1528 	int pinned_pack = -1, pinned_packidx = -1;
1529 
1530 	for (i = 0; i < repo->pack_cache_size; i++) {
1531 		if (repo->packidx_cache[i] &&
1532 		    strcmp(repo->packidx_cache[i]->path_packidx,
1533 		    packidx->path_packidx) == 0)
1534 			pinned_packidx = i;
1535 		if (repo->packs[i].path_packfile &&
1536 		    strcmp(repo->packs[i].path_packfile,
1537 		    pack->path_packfile) == 0)
1538 			pinned_pack = i;
1539 	}
1540 
1541 	if (pinned_packidx == -1 || pinned_pack == -1)
1542 		return got_error(GOT_ERR_PIN_PACK);
1543 
1544 	repo->pinned_pack = pinned_pack;
1545 	repo->pinned_packidx = pinned_packidx;
1546 	if (repo->packs[pinned_pack].privsep_child)
1547 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1548 	return NULL;
1549 }
1550 
1551 struct got_pack *
1552 got_repo_get_pinned_pack(struct got_repository *repo)
1553 {
1554 	if (repo->pinned_pack >= 0 &&
1555 	    repo->pinned_pack < repo->pack_cache_size)
1556 		return &repo->packs[repo->pinned_pack];
1557 
1558 	return NULL;
1559 }
1560 
1561 void
1562 got_repo_unpin_pack(struct got_repository *repo)
1563 {
1564 	repo->pinned_packidx = -1;
1565 	repo->pinned_pack = -1;
1566 	repo->pinned_pid = 0;
1567 }
1568 
1569 const struct got_error *
1570 got_repo_init(const char *repo_path, const char *head_name)
1571 {
1572 	const struct got_error *err = NULL;
1573 	const char *dirnames[] = {
1574 		GOT_OBJECTS_DIR,
1575 		GOT_OBJECTS_PACK_DIR,
1576 		GOT_REFS_DIR,
1577 	};
1578 	const char *description_str = "Unnamed repository; "
1579 	    "edit this file 'description' to name the repository.";
1580 	const char *headref = "ref: refs/heads/";
1581 	const char *gitconfig_str = "[core]\n"
1582 	    "\trepositoryformatversion = 0\n"
1583 	    "\tfilemode = true\n"
1584 	    "\tbare = true\n";
1585 	char *headref_str, *path;
1586 	size_t i;
1587 
1588 	if (!got_path_dir_is_empty(repo_path))
1589 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1590 
1591 	for (i = 0; i < nitems(dirnames); i++) {
1592 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1593 			return got_error_from_errno("asprintf");
1594 		}
1595 		err = got_path_mkdir(path);
1596 		free(path);
1597 		if (err)
1598 			return err;
1599 	}
1600 
1601 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1602 		return got_error_from_errno("asprintf");
1603 	err = got_path_create_file(path, description_str);
1604 	free(path);
1605 	if (err)
1606 		return err;
1607 
1608 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1609 		return got_error_from_errno("asprintf");
1610 	if (asprintf(&headref_str, "%s%s", headref,
1611 	    head_name ? head_name : "main") == -1) {
1612 		free(path);
1613 		return got_error_from_errno("asprintf");
1614 	}
1615 	err = got_path_create_file(path, headref_str);
1616 	free(headref_str);
1617 	free(path);
1618 	if (err)
1619 		return err;
1620 
1621 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1622 		return got_error_from_errno("asprintf");
1623 	err = got_path_create_file(path, gitconfig_str);
1624 	free(path);
1625 	if (err)
1626 		return err;
1627 
1628 	return NULL;
1629 }
1630 
1631 static const struct got_error *
1632 match_packed_object(struct got_object_id **unique_id,
1633     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1634 {
1635 	const struct got_error *err = NULL;
1636 	struct got_object_id_queue matched_ids;
1637 	struct got_pathlist_entry *pe;
1638 
1639 	STAILQ_INIT(&matched_ids);
1640 
1641 	err = refresh_packidx_paths(repo);
1642 	if (err)
1643 		return err;
1644 
1645 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1646 		const char *path_packidx = pe->path;
1647 		struct got_packidx *packidx;
1648 		struct got_object_qid *qid;
1649 
1650 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1651 		    path_packidx, 0);
1652 		if (err)
1653 			break;
1654 
1655 		err = got_packidx_match_id_str_prefix(&matched_ids,
1656 		    packidx, id_str_prefix);
1657 		if (err) {
1658 			got_packidx_close(packidx);
1659 			break;
1660 		}
1661 		err = got_packidx_close(packidx);
1662 		if (err)
1663 			break;
1664 
1665 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1666 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1667 				int matched_type;
1668 				err = got_object_get_type(&matched_type, repo,
1669 				    &qid->id);
1670 				if (err)
1671 					goto done;
1672 				if (matched_type != obj_type)
1673 					continue;
1674 			}
1675 			if (*unique_id == NULL) {
1676 				*unique_id = got_object_id_dup(&qid->id);
1677 				if (*unique_id == NULL) {
1678 					err = got_error_from_errno("malloc");
1679 					goto done;
1680 				}
1681 			} else {
1682 				if (got_object_id_cmp(*unique_id,
1683 				    &qid->id) == 0)
1684 					continue; /* packed multiple times */
1685 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1686 				goto done;
1687 			}
1688 		}
1689 	}
1690 done:
1691 	got_object_id_queue_free(&matched_ids);
1692 	if (err) {
1693 		free(*unique_id);
1694 		*unique_id = NULL;
1695 	}
1696 	return err;
1697 }
1698 
1699 static const struct got_error *
1700 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1701     const char *object_dir, const char *id_str_prefix, int obj_type,
1702     struct got_repository *repo)
1703 {
1704 	const struct got_error *err = NULL;
1705 	char *path, *id_str = NULL;
1706 	DIR *dir = NULL;
1707 	struct dirent *dent;
1708 	struct got_object_id id;
1709 
1710 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1711 		err = got_error_from_errno("asprintf");
1712 		goto done;
1713 	}
1714 
1715 	dir = opendir(path);
1716 	if (dir == NULL) {
1717 		if (errno == ENOENT) {
1718 			err = NULL;
1719 			goto done;
1720 		}
1721 		err = got_error_from_errno2("opendir", path);
1722 		goto done;
1723 	}
1724 	while ((dent = readdir(dir)) != NULL) {
1725 		int cmp;
1726 
1727 		free(id_str);
1728 		id_str = NULL;
1729 
1730 		if (strcmp(dent->d_name, ".") == 0 ||
1731 		    strcmp(dent->d_name, "..") == 0)
1732 			continue;
1733 
1734 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1735 			err = got_error_from_errno("asprintf");
1736 			goto done;
1737 		}
1738 
1739 		if (!got_parse_sha1_digest(id.hash, id_str))
1740 			continue;
1741 
1742 		/*
1743 		 * Directory entries do not necessarily appear in
1744 		 * sorted order, so we must iterate over all of them.
1745 		 */
1746 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1747 		if (cmp != 0)
1748 			continue;
1749 
1750 		if (*unique_id == NULL) {
1751 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1752 				int matched_type;
1753 				err = got_object_get_type(&matched_type, repo,
1754 				    &id);
1755 				if (err)
1756 					goto done;
1757 				if (matched_type != obj_type)
1758 					continue;
1759 			}
1760 			*unique_id = got_object_id_dup(&id);
1761 			if (*unique_id == NULL) {
1762 				err = got_error_from_errno("got_object_id_dup");
1763 				goto done;
1764 			}
1765 		} else {
1766 			if (got_object_id_cmp(*unique_id, &id) == 0)
1767 				continue; /* both packed and loose */
1768 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1769 			goto done;
1770 		}
1771 	}
1772 done:
1773 	if (dir && closedir(dir) != 0 && err == NULL)
1774 		err = got_error_from_errno("closedir");
1775 	if (err) {
1776 		free(*unique_id);
1777 		*unique_id = NULL;
1778 	}
1779 	free(id_str);
1780 	free(path);
1781 	return err;
1782 }
1783 
1784 const struct got_error *
1785 got_repo_match_object_id_prefix(struct got_object_id **id,
1786     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1787 {
1788 	const struct got_error *err = NULL;
1789 	char *path_objects = NULL, *object_dir = NULL;
1790 	size_t len;
1791 	int i;
1792 
1793 	*id = NULL;
1794 
1795 	path_objects = got_repo_get_path_objects(repo);
1796 
1797 	len = strlen(id_str_prefix);
1798 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1799 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1800 		goto done;
1801 	}
1802 
1803 	for (i = 0; i < len; i++) {
1804 		if (isxdigit((unsigned char)id_str_prefix[i]))
1805 			continue;
1806 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1807 		goto done;
1808 	}
1809 
1810 	if (len >= 2) {
1811 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1812 		if (err)
1813 			goto done;
1814 		object_dir = strndup(id_str_prefix, 2);
1815 		if (object_dir == NULL) {
1816 			err = got_error_from_errno("strdup");
1817 			goto done;
1818 		}
1819 		err = match_loose_object(id, path_objects, object_dir,
1820 		    id_str_prefix, obj_type, repo);
1821 	} else if (len == 1) {
1822 		int i;
1823 		for (i = 0; i < 0xf; i++) {
1824 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1825 			    == -1) {
1826 				err = got_error_from_errno("asprintf");
1827 				goto done;
1828 			}
1829 			err = match_packed_object(id, repo, object_dir,
1830 			    obj_type);
1831 			if (err)
1832 				goto done;
1833 			err = match_loose_object(id, path_objects, object_dir,
1834 			    id_str_prefix, obj_type, repo);
1835 			if (err)
1836 				goto done;
1837 		}
1838 	} else {
1839 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1840 		goto done;
1841 	}
1842 done:
1843 	free(path_objects);
1844 	free(object_dir);
1845 	if (err) {
1846 		free(*id);
1847 		*id = NULL;
1848 	} else if (*id == NULL) {
1849 		switch (obj_type) {
1850 		case GOT_OBJ_TYPE_BLOB:
1851 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1852 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1853 			break;
1854 		case GOT_OBJ_TYPE_TREE:
1855 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1856 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1857 			break;
1858 		case GOT_OBJ_TYPE_COMMIT:
1859 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1860 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1861 			break;
1862 		case GOT_OBJ_TYPE_TAG:
1863 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1864 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1865 			break;
1866 		default:
1867 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1868 			break;
1869 		}
1870 	}
1871 
1872 	return err;
1873 }
1874 
1875 const struct got_error *
1876 got_repo_match_object_id(struct got_object_id **id, char **label,
1877     const char *id_str, int obj_type, struct got_reflist_head *refs,
1878     struct got_repository *repo)
1879 {
1880 	const struct got_error *err;
1881 	struct got_tag_object *tag;
1882 	struct got_reference *ref = NULL;
1883 
1884 	*id = NULL;
1885 	if (label)
1886 		*label = NULL;
1887 
1888 	if (refs) {
1889 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1890 		    refs, repo);
1891 		if (err == NULL) {
1892 			*id = got_object_id_dup(
1893 			    got_object_tag_get_object_id(tag));
1894 			if (*id == NULL)
1895 				err = got_error_from_errno("got_object_id_dup");
1896 			else if (label && asprintf(label, "refs/tags/%s",
1897 			    got_object_tag_get_name(tag)) == -1) {
1898 				err = got_error_from_errno("asprintf");
1899 				free(*id);
1900 				*id = NULL;
1901 			}
1902 			got_object_tag_close(tag);
1903 			return err;
1904 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1905 		    err->code != GOT_ERR_NO_OBJ)
1906 			return err;
1907 	}
1908 
1909 	err = got_ref_open(&ref, repo, id_str, 0);
1910 	if (err == NULL) {
1911 		err = got_ref_resolve(id, repo, ref);
1912 		if (err)
1913 			goto done;
1914 		if (label) {
1915 			*label = strdup(got_ref_get_name(ref));
1916 			if (*label == NULL) {
1917 				err = got_error_from_errno("strdup");
1918 				goto done;
1919 			}
1920 		}
1921 	} else {
1922 		if (err->code != GOT_ERR_NOT_REF &&
1923 		    err->code != GOT_ERR_BAD_REF_NAME)
1924 			goto done;
1925 		err = got_repo_match_object_id_prefix(id, id_str,
1926 		    obj_type, repo);
1927 		if (err) {
1928 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1929 				err = got_error_not_ref(id_str);
1930 			goto done;
1931 		}
1932 		if (label) {
1933 			err = got_object_id_str(label, *id);
1934 			if (*label == NULL) {
1935 				err = got_error_from_errno("strdup");
1936 				goto done;
1937 			}
1938 		}
1939 	}
1940 done:
1941 	if (ref)
1942 		got_ref_close(ref);
1943 	return err;
1944 }
1945 
1946 const struct got_error *
1947 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1948     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1949 {
1950 	const struct got_error *err = NULL;
1951 	struct got_reflist_entry *re;
1952 	struct got_object_id *tag_id;
1953 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1954 
1955 	*tag = NULL;
1956 
1957 	TAILQ_FOREACH(re, refs, entry) {
1958 		const char *refname;
1959 		refname = got_ref_get_name(re->ref);
1960 		if (got_ref_is_symbolic(re->ref))
1961 			continue;
1962 		if (strncmp(refname, "refs/tags/", 10) != 0)
1963 			continue;
1964 		if (!name_is_absolute)
1965 			refname += strlen("refs/tags/");
1966 		if (strcmp(refname, name) != 0)
1967 			continue;
1968 		err = got_ref_resolve(&tag_id, repo, re->ref);
1969 		if (err)
1970 			break;
1971 		err = got_object_open_as_tag(tag, repo, tag_id);
1972 		free(tag_id);
1973 		if (err)
1974 			break;
1975 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1976 		    got_object_tag_get_object_type(*tag) == obj_type)
1977 			break;
1978 		got_object_tag_close(*tag);
1979 		*tag = NULL;
1980 	}
1981 
1982 	if (err == NULL && *tag == NULL)
1983 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1984 		    GOT_OBJ_LABEL_TAG, name);
1985 	return err;
1986 }
1987 
1988 static const struct got_error *
1989 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1990     const char *name, mode_t mode, struct got_object_id *blob_id)
1991 {
1992 	const struct got_error *err = NULL;
1993 
1994 	 *new_te = NULL;
1995 
1996 	*new_te = calloc(1, sizeof(**new_te));
1997 	if (*new_te == NULL)
1998 		return got_error_from_errno("calloc");
1999 
2000 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2001 	    sizeof((*new_te)->name)) {
2002 		err = got_error(GOT_ERR_NO_SPACE);
2003 		goto done;
2004 	}
2005 
2006 	if (S_ISLNK(mode)) {
2007 		(*new_te)->mode = S_IFLNK;
2008 	} else {
2009 		(*new_te)->mode = S_IFREG;
2010 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2011 	}
2012 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2013 done:
2014 	if (err && *new_te) {
2015 		free(*new_te);
2016 		*new_te = NULL;
2017 	}
2018 	return err;
2019 }
2020 
2021 static const struct got_error *
2022 import_file(struct got_tree_entry **new_te, struct dirent *de,
2023     const char *path, struct got_repository *repo)
2024 {
2025 	const struct got_error *err;
2026 	struct got_object_id *blob_id = NULL;
2027 	char *filepath;
2028 	struct stat sb;
2029 
2030 	if (asprintf(&filepath, "%s%s%s", path,
2031 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2032 		return got_error_from_errno("asprintf");
2033 
2034 	if (lstat(filepath, &sb) != 0) {
2035 		err = got_error_from_errno2("lstat", path);
2036 		goto done;
2037 	}
2038 
2039 	err = got_object_blob_create(&blob_id, filepath, repo);
2040 	if (err)
2041 		goto done;
2042 
2043 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2044 	    blob_id);
2045 done:
2046 	free(filepath);
2047 	if (err)
2048 		free(blob_id);
2049 	return err;
2050 }
2051 
2052 static const struct got_error *
2053 insert_tree_entry(struct got_tree_entry *new_te,
2054     struct got_pathlist_head *paths)
2055 {
2056 	const struct got_error *err = NULL;
2057 	struct got_pathlist_entry *new_pe;
2058 
2059 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2060 	if (err)
2061 		return err;
2062 	if (new_pe == NULL)
2063 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2064 	return NULL;
2065 }
2066 
2067 static const struct got_error *write_tree(struct got_object_id **,
2068     const char *, struct got_pathlist_head *, struct got_repository *,
2069     got_repo_import_cb progress_cb, void *progress_arg);
2070 
2071 static const struct got_error *
2072 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2073     const char *path, struct got_pathlist_head *ignores,
2074     struct got_repository *repo,
2075     got_repo_import_cb progress_cb, void *progress_arg)
2076 {
2077 	const struct got_error *err;
2078 	struct got_object_id *id = NULL;
2079 	char *subdirpath;
2080 
2081 	if (asprintf(&subdirpath, "%s%s%s", path,
2082 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2083 		return got_error_from_errno("asprintf");
2084 
2085 	(*new_te) = calloc(1, sizeof(**new_te));
2086 	if (*new_te == NULL)
2087 		return got_error_from_errno("calloc");
2088 	(*new_te)->mode = S_IFDIR;
2089 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2090 	    sizeof((*new_te)->name)) {
2091 		err = got_error(GOT_ERR_NO_SPACE);
2092 		goto done;
2093 	}
2094 	err = write_tree(&id, subdirpath, ignores,  repo,
2095 	    progress_cb, progress_arg);
2096 	if (err)
2097 		goto done;
2098 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2099 
2100 done:
2101 	free(id);
2102 	free(subdirpath);
2103 	if (err) {
2104 		free(*new_te);
2105 		*new_te = NULL;
2106 	}
2107 	return err;
2108 }
2109 
2110 static const struct got_error *
2111 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2112     struct got_pathlist_head *ignores, struct got_repository *repo,
2113     got_repo_import_cb progress_cb, void *progress_arg)
2114 {
2115 	const struct got_error *err = NULL;
2116 	DIR *dir;
2117 	struct dirent *de;
2118 	int nentries;
2119 	struct got_tree_entry *new_te = NULL;
2120 	struct got_pathlist_head paths;
2121 	struct got_pathlist_entry *pe;
2122 
2123 	*new_tree_id = NULL;
2124 
2125 	TAILQ_INIT(&paths);
2126 
2127 	dir = opendir(path_dir);
2128 	if (dir == NULL) {
2129 		err = got_error_from_errno2("opendir", path_dir);
2130 		goto done;
2131 	}
2132 
2133 	nentries = 0;
2134 	while ((de = readdir(dir)) != NULL) {
2135 		int ignore = 0;
2136 		int type;
2137 
2138 		if (strcmp(de->d_name, ".") == 0 ||
2139 		    strcmp(de->d_name, "..") == 0)
2140 			continue;
2141 
2142 		TAILQ_FOREACH(pe, ignores, entry) {
2143 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2144 				ignore = 1;
2145 				break;
2146 			}
2147 		}
2148 		if (ignore)
2149 			continue;
2150 
2151 		err = got_path_dirent_type(&type, path_dir, de);
2152 		if (err)
2153 			goto done;
2154 
2155 		if (type == DT_DIR) {
2156 			err = import_subdir(&new_te, de, path_dir,
2157 			    ignores, repo, progress_cb, progress_arg);
2158 			if (err) {
2159 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2160 					goto done;
2161 				err = NULL;
2162 				continue;
2163 			}
2164 		} else if (type == DT_REG || type == DT_LNK) {
2165 			err = import_file(&new_te, de, path_dir, repo);
2166 			if (err)
2167 				goto done;
2168 		} else
2169 			continue;
2170 
2171 		err = insert_tree_entry(new_te, &paths);
2172 		if (err)
2173 			goto done;
2174 		nentries++;
2175 	}
2176 
2177 	if (TAILQ_EMPTY(&paths)) {
2178 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2179 		    "cannot create tree without any entries");
2180 		goto done;
2181 	}
2182 
2183 	TAILQ_FOREACH(pe, &paths, entry) {
2184 		struct got_tree_entry *te = pe->data;
2185 		char *path;
2186 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2187 			continue;
2188 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2189 			err = got_error_from_errno("asprintf");
2190 			goto done;
2191 		}
2192 		err = (*progress_cb)(progress_arg, path);
2193 		free(path);
2194 		if (err)
2195 			goto done;
2196 	}
2197 
2198 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2199 done:
2200 	if (dir)
2201 		closedir(dir);
2202 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2203 	return err;
2204 }
2205 
2206 const struct got_error *
2207 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2208     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2209     struct got_repository *repo, got_repo_import_cb progress_cb,
2210     void *progress_arg)
2211 {
2212 	const struct got_error *err;
2213 	struct got_object_id *new_tree_id;
2214 
2215 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2216 	    progress_cb, progress_arg);
2217 	if (err)
2218 		return err;
2219 
2220 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2221 	    author, time(NULL), author, time(NULL), logmsg, repo);
2222 	free(new_tree_id);
2223 	return err;
2224 }
2225 
2226 const struct got_error *
2227 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2228     struct got_repository *repo)
2229 {
2230 	const struct got_error *err = NULL;
2231 	char *path_objects = NULL, *path = NULL;
2232 	DIR *dir = NULL;
2233 	struct got_object_id id;
2234 	int i;
2235 
2236 	*nobjects = 0;
2237 	*ondisk_size = 0;
2238 
2239 	path_objects = got_repo_get_path_objects(repo);
2240 	if (path_objects == NULL)
2241 		return got_error_from_errno("got_repo_get_path_objects");
2242 
2243 	for (i = 0; i <= 0xff; i++) {
2244 		struct dirent *dent;
2245 
2246 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2247 			err = got_error_from_errno("asprintf");
2248 			break;
2249 		}
2250 
2251 		dir = opendir(path);
2252 		if (dir == NULL) {
2253 			if (errno == ENOENT) {
2254 				err = NULL;
2255 				continue;
2256 			}
2257 			err = got_error_from_errno2("opendir", path);
2258 			break;
2259 		}
2260 
2261 		while ((dent = readdir(dir)) != NULL) {
2262 			char *id_str;
2263 			int fd;
2264 			struct stat sb;
2265 
2266 			if (strcmp(dent->d_name, ".") == 0 ||
2267 			    strcmp(dent->d_name, "..") == 0)
2268 				continue;
2269 
2270 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2271 				err = got_error_from_errno("asprintf");
2272 				goto done;
2273 			}
2274 
2275 			if (!got_parse_sha1_digest(id.hash, id_str)) {
2276 				free(id_str);
2277 				continue;
2278 			}
2279 			free(id_str);
2280 
2281 			err = got_object_open_loose_fd(&fd, &id, repo);
2282 			if (err)
2283 				goto done;
2284 
2285 			if (fstat(fd, &sb) == -1) {
2286 				err = got_error_from_errno("fstat");
2287 				close(fd);
2288 				goto done;
2289 			}
2290 			(*nobjects)++;
2291 			(*ondisk_size) += sb.st_size;
2292 
2293 			if (close(fd) == -1) {
2294 				err = got_error_from_errno("close");
2295 				goto done;
2296 			}
2297 		}
2298 
2299 		if (closedir(dir) != 0) {
2300 			err = got_error_from_errno("closedir");
2301 			goto done;
2302 		}
2303 		dir = NULL;
2304 
2305 		free(path);
2306 		path = NULL;
2307 	}
2308 done:
2309 	if (dir && closedir(dir) != 0 && err == NULL)
2310 		err = got_error_from_errno("closedir");
2311 
2312 	if (err) {
2313 		*nobjects = 0;
2314 		*ondisk_size = 0;
2315 	}
2316 	free(path_objects);
2317 	free(path);
2318 	return err;
2319 }
2320 
2321 const struct got_error *
2322 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2323     off_t *total_packsize, struct got_repository *repo)
2324 {
2325 	const struct got_error *err = NULL;
2326 	DIR *packdir = NULL;
2327 	struct dirent *dent;
2328 	struct got_packidx *packidx = NULL;
2329 	char *path_packidx;
2330 	char *path_packfile;
2331 	int packdir_fd;
2332 	struct stat sb;
2333 
2334 	*npackfiles = 0;
2335 	*nobjects = 0;
2336 	*total_packsize = 0;
2337 
2338 	packdir_fd = openat(got_repo_get_fd(repo),
2339 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2340 	if (packdir_fd == -1) {
2341 		return got_error_from_errno_fmt("openat: %s/%s",
2342 		    got_repo_get_path_git_dir(repo),
2343 		    GOT_OBJECTS_PACK_DIR);
2344 	}
2345 
2346 	packdir = fdopendir(packdir_fd);
2347 	if (packdir == NULL) {
2348 		err = got_error_from_errno("fdopendir");
2349 		goto done;
2350 	}
2351 
2352 	while ((dent = readdir(packdir)) != NULL) {
2353 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2354 			continue;
2355 
2356 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2357 		    dent->d_name) == -1) {
2358 			err = got_error_from_errno("asprintf");
2359 			goto done;
2360 		}
2361 
2362 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2363 		    path_packidx, 0);
2364 		free(path_packidx);
2365 		if (err)
2366 			goto done;
2367 
2368 		if (fstat(packidx->fd, &sb) == -1)
2369 			goto done;
2370 		*total_packsize += sb.st_size;
2371 
2372 		err = got_packidx_get_packfile_path(&path_packfile,
2373 		    packidx->path_packidx);
2374 		if (err)
2375 			goto done;
2376 
2377 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2378 		    0) == -1) {
2379 			free(path_packfile);
2380 			goto done;
2381 		}
2382 		free(path_packfile);
2383 		*total_packsize += sb.st_size;
2384 
2385 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2386 
2387 		(*npackfiles)++;
2388 
2389 		got_packidx_close(packidx);
2390 		packidx = NULL;
2391 	}
2392 done:
2393 	if (packidx)
2394 		got_packidx_close(packidx);
2395 	if (packdir && closedir(packdir) != 0 && err == NULL)
2396 		err = got_error_from_errno("closedir");
2397 	if (err) {
2398 		*npackfiles = 0;
2399 		*nobjects = 0;
2400 		*total_packsize = 0;
2401 	}
2402 	return err;
2403 }
2404 
2405 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2406     got_packidx_bloom_filter_cmp);