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