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