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