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