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/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 #include <sys/resource.h>
24 
25 #include <ctype.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <string.h>
33 #include <time.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 
40 #include "bloom.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_sha1.h"
58 #include "got_lib_object_cache.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_gotconfig.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
67 
68 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
69     got_packidx_bloom_filter_cmp);
70 
71 const char *
72 got_repo_get_path(struct got_repository *repo)
73 {
74 	return repo->path;
75 }
76 
77 const char *
78 got_repo_get_path_git_dir(struct got_repository *repo)
79 {
80 	return repo->path_git_dir;
81 }
82 
83 int
84 got_repo_get_fd(struct got_repository *repo)
85 {
86 	return repo->gitdir_fd;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_name(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_name;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_email(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_email;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_name;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_email;
111 }
112 
113 const char *
114 got_repo_get_gitconfig_owner(struct got_repository *repo)
115 {
116 	return repo->gitconfig_owner;
117 }
118 
119 void
120 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
121     struct got_repository *repo)
122 {
123 	*extensions = repo->extensions;
124 	*nextensions = repo->nextensions;
125 }
126 
127 int
128 got_repo_is_bare(struct got_repository *repo)
129 {
130 	return (strcmp(repo->path, repo->path_git_dir) == 0);
131 }
132 
133 static char *
134 get_path_git_child(struct got_repository *repo, const char *basename)
135 {
136 	char *path_child;
137 
138 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
139 	    basename) == -1)
140 		return NULL;
141 
142 	return path_child;
143 }
144 
145 char *
146 got_repo_get_path_objects(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
149 }
150 
151 char *
152 got_repo_get_path_objects_pack(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
155 }
156 
157 char *
158 got_repo_get_path_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_REFS_DIR);
161 }
162 
163 char *
164 got_repo_get_path_packed_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
167 }
168 
169 static char *
170 get_path_head(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_HEAD_FILE);
173 }
174 
175 char *
176 got_repo_get_path_gitconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GITCONFIG);
179 }
180 
181 char *
182 got_repo_get_path_gotconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
185 }
186 
187 const struct got_gotconfig *
188 got_repo_get_gotconfig(struct got_repository *repo)
189 {
190 	return repo->gotconfig;
191 }
192 
193 void
194 got_repo_get_gitconfig_remotes(int *nremotes,
195     const struct got_remote_repo **remotes, struct got_repository *repo)
196 {
197 	*nremotes = repo->ngitconfig_remotes;
198 	*remotes = repo->gitconfig_remotes;
199 }
200 
201 static int
202 is_git_repo(struct got_repository *repo)
203 {
204 	const char *path_git = got_repo_get_path_git_dir(repo);
205 	char *path_objects = got_repo_get_path_objects(repo);
206 	char *path_refs = got_repo_get_path_refs(repo);
207 	char *path_head = get_path_head(repo);
208 	int ret = 0;
209 	struct stat sb;
210 	struct got_reference *head_ref;
211 
212 	if (lstat(path_git, &sb) == -1)
213 		goto done;
214 	if (!S_ISDIR(sb.st_mode))
215 		goto done;
216 
217 	if (lstat(path_objects, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_refs, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_head, &sb) == -1)
228 		goto done;
229 	if (!S_ISREG(sb.st_mode))
230 		goto done;
231 
232 	/* Check if the HEAD reference can be opened. */
233 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
234 		goto done;
235 	got_ref_close(head_ref);
236 
237 	ret = 1;
238 done:
239 	free(path_objects);
240 	free(path_refs);
241 	free(path_head);
242 	return ret;
243 
244 }
245 
246 const struct got_error *
247 got_repo_pack_fds_open(int **pack_fds)
248 {
249 	const struct got_error *err = NULL;
250 	int i, pack_fds_tmp[GOT_PACK_NUM_TEMPFILES];
251 
252 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
253 	if (*pack_fds == NULL)
254 		return got_error_from_errno("calloc");
255 
256 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
257 		pack_fds_tmp[i] = got_opentempfd();
258 		if (pack_fds_tmp[i] == -1) {
259 			err = got_error_from_errno("got_opentempfd");
260 			got_repo_pack_fds_close(pack_fds_tmp);
261 			return err;
262 		}
263 	}
264 	memcpy(*pack_fds, pack_fds_tmp, sizeof(pack_fds_tmp));
265 	return err;
266 }
267 
268 const struct got_error *
269 got_repo_pack_fds_close(int *pack_fds)
270 {
271 	const struct got_error *err = NULL;
272 	int i;
273 
274 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
275 		if (pack_fds[i] == -1)
276 			continue;
277 		if (close(pack_fds[i]) == -1) {
278 			err = got_error_from_errno("close");
279 			break;
280 		}
281 	}
282 	free(pack_fds);
283 	return err;
284 }
285 
286 const struct got_error *
287 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
288     struct got_object *obj)
289 {
290 #ifndef GOT_NO_OBJ_CACHE
291 	const struct got_error *err = NULL;
292 	err = got_object_cache_add(&repo->objcache, id, obj);
293 	if (err) {
294 		if (err->code == GOT_ERR_OBJ_EXISTS ||
295 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
296 			err = NULL;
297 		return err;
298 	}
299 	obj->refcnt++;
300 #endif
301 	return NULL;
302 }
303 
304 struct got_object *
305 got_repo_get_cached_object(struct got_repository *repo,
306     struct got_object_id *id)
307 {
308 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
309 }
310 
311 const struct got_error *
312 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
313     struct got_tree_object *tree)
314 {
315 #ifndef GOT_NO_OBJ_CACHE
316 	const struct got_error *err = NULL;
317 	err = got_object_cache_add(&repo->treecache, id, tree);
318 	if (err) {
319 		if (err->code == GOT_ERR_OBJ_EXISTS ||
320 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
321 			err = NULL;
322 		return err;
323 	}
324 	tree->refcnt++;
325 #endif
326 	return NULL;
327 }
328 
329 struct got_tree_object *
330 got_repo_get_cached_tree(struct got_repository *repo,
331     struct got_object_id *id)
332 {
333 	return (struct got_tree_object *)got_object_cache_get(
334 	    &repo->treecache, id);
335 }
336 
337 const struct got_error *
338 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
339     struct got_commit_object *commit)
340 {
341 #ifndef GOT_NO_OBJ_CACHE
342 	const struct got_error *err = NULL;
343 	err = got_object_cache_add(&repo->commitcache, id, commit);
344 	if (err) {
345 		if (err->code == GOT_ERR_OBJ_EXISTS ||
346 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
347 			err = NULL;
348 		return err;
349 	}
350 	commit->refcnt++;
351 #endif
352 	return NULL;
353 }
354 
355 struct got_commit_object *
356 got_repo_get_cached_commit(struct got_repository *repo,
357     struct got_object_id *id)
358 {
359 	return (struct got_commit_object *)got_object_cache_get(
360 	    &repo->commitcache, id);
361 }
362 
363 const struct got_error *
364 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
365     struct got_tag_object *tag)
366 {
367 #ifndef GOT_NO_OBJ_CACHE
368 	const struct got_error *err = NULL;
369 	err = got_object_cache_add(&repo->tagcache, id, tag);
370 	if (err) {
371 		if (err->code == GOT_ERR_OBJ_EXISTS ||
372 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
373 			err = NULL;
374 		return err;
375 	}
376 	tag->refcnt++;
377 #endif
378 	return NULL;
379 }
380 
381 struct got_tag_object *
382 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
383 {
384 	return (struct got_tag_object *)got_object_cache_get(
385 	    &repo->tagcache, id);
386 }
387 
388 const struct got_error *
389 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
390     struct got_raw_object *raw)
391 {
392 #ifndef GOT_NO_OBJ_CACHE
393 	const struct got_error *err = NULL;
394 	err = got_object_cache_add(&repo->rawcache, id, raw);
395 	if (err) {
396 		if (err->code == GOT_ERR_OBJ_EXISTS ||
397 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
398 			err = NULL;
399 		return err;
400 	}
401 	raw->refcnt++;
402 #endif
403 	return NULL;
404 }
405 
406 
407 struct got_raw_object *
408 got_repo_get_cached_raw_object(struct got_repository *repo,
409     struct got_object_id *id)
410 {
411 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
412 }
413 
414 
415 static const struct got_error *
416 open_repo(struct got_repository *repo, const char *path)
417 {
418 	const struct got_error *err = NULL;
419 
420 	repo->gitdir_fd = -1;
421 
422 	/* bare git repository? */
423 	repo->path_git_dir = strdup(path);
424 	if (repo->path_git_dir == NULL)
425 		return got_error_from_errno("strdup");
426 	if (is_git_repo(repo)) {
427 		repo->path = strdup(repo->path_git_dir);
428 		if (repo->path == NULL) {
429 			err = got_error_from_errno("strdup");
430 			goto done;
431 		}
432 		repo->gitdir_fd = open(repo->path_git_dir,
433 		    O_DIRECTORY | O_CLOEXEC);
434 		if (repo->gitdir_fd == -1) {
435 			err = got_error_from_errno2("open",
436 			    repo->path_git_dir);
437 			goto done;
438 		}
439 		return NULL;
440 	}
441 
442 	/* git repository with working tree? */
443 	free(repo->path_git_dir);
444 	repo->path_git_dir = NULL;
445 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
446 		err = got_error_from_errno("asprintf");
447 		goto done;
448 	}
449 	if (is_git_repo(repo)) {
450 		repo->path = strdup(path);
451 		if (repo->path == NULL) {
452 			err = got_error_from_errno("strdup");
453 			goto done;
454 		}
455 		repo->gitdir_fd = open(repo->path_git_dir,
456 		    O_DIRECTORY | O_CLOEXEC);
457 		if (repo->gitdir_fd == -1) {
458 			err = got_error_from_errno2("open",
459 			    repo->path_git_dir);
460 			goto done;
461 		}
462 		return NULL;
463 	}
464 
465 	err = got_error(GOT_ERR_NOT_GIT_REPO);
466 done:
467 	if (err) {
468 		free(repo->path);
469 		repo->path = NULL;
470 		free(repo->path_git_dir);
471 		repo->path_git_dir = NULL;
472 		if (repo->gitdir_fd != -1)
473 			close(repo->gitdir_fd);
474 		repo->gitdir_fd = -1;
475 
476 	}
477 	return err;
478 }
479 
480 static const struct got_error *
481 parse_gitconfig_file(int *gitconfig_repository_format_version,
482     char **gitconfig_author_name, char **gitconfig_author_email,
483     struct got_remote_repo **remotes, int *nremotes,
484     char **gitconfig_owner, char ***extensions, int *nextensions,
485     const char *gitconfig_path)
486 {
487 	const struct got_error *err = NULL, *child_err = NULL;
488 	int fd = -1;
489 	int imsg_fds[2] = { -1, -1 };
490 	pid_t pid;
491 	struct imsgbuf *ibuf;
492 
493 	*gitconfig_repository_format_version = 0;
494 	if (extensions)
495 		*extensions = NULL;
496 	if (nextensions)
497 		*nextensions = 0;
498 	*gitconfig_author_name = NULL;
499 	*gitconfig_author_email = NULL;
500 	if (remotes)
501 		*remotes = NULL;
502 	if (nremotes)
503 		*nremotes = 0;
504 	if (gitconfig_owner)
505 		*gitconfig_owner = NULL;
506 
507 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
508 	if (fd == -1) {
509 		if (errno == ENOENT)
510 			return NULL;
511 		return got_error_from_errno2("open", gitconfig_path);
512 	}
513 
514 	ibuf = calloc(1, sizeof(*ibuf));
515 	if (ibuf == NULL) {
516 		err = got_error_from_errno("calloc");
517 		goto done;
518 	}
519 
520 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
521 		err = got_error_from_errno("socketpair");
522 		goto done;
523 	}
524 
525 	pid = fork();
526 	if (pid == -1) {
527 		err = got_error_from_errno("fork");
528 		goto done;
529 	} else if (pid == 0) {
530 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
531 		    gitconfig_path);
532 		/* not reached */
533 	}
534 
535 	if (close(imsg_fds[1]) == -1) {
536 		err = got_error_from_errno("close");
537 		goto done;
538 	}
539 	imsg_fds[1] = -1;
540 	imsg_init(ibuf, imsg_fds[0]);
541 
542 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
543 	if (err)
544 		goto done;
545 	fd = -1;
546 
547 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
548 	if (err)
549 		goto done;
550 
551 	err = got_privsep_recv_gitconfig_int(
552 	    gitconfig_repository_format_version, ibuf);
553 	if (err)
554 		goto done;
555 
556 	if (extensions && nextensions) {
557 		err = got_privsep_send_gitconfig_repository_extensions_req(
558 		    ibuf);
559 		if (err)
560 			goto done;
561 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
562 		if (err)
563 			goto done;
564 		if (*nextensions > 0) {
565 			int i;
566 			*extensions = calloc(*nextensions, sizeof(char *));
567 			if (*extensions == NULL) {
568 				err = got_error_from_errno("calloc");
569 				goto done;
570 			}
571 			for (i = 0; i < *nextensions; i++) {
572 				char *ext;
573 				err = got_privsep_recv_gitconfig_str(&ext,
574 				    ibuf);
575 				if (err)
576 					goto done;
577 				(*extensions)[i] = ext;
578 			}
579 		}
580 	}
581 
582 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
583 	if (err)
584 		goto done;
585 
586 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
587 	if (err)
588 		goto done;
589 
590 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
591 	if (err)
592 		goto done;
593 
594 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
595 	if (err)
596 		goto done;
597 
598 	if (remotes && nremotes) {
599 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
600 		if (err)
601 			goto done;
602 
603 		err = got_privsep_recv_gitconfig_remotes(remotes,
604 		    nremotes, ibuf);
605 		if (err)
606 			goto done;
607 	}
608 
609 	if (gitconfig_owner) {
610 		err = got_privsep_send_gitconfig_owner_req(ibuf);
611 		if (err)
612 			goto done;
613 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
614 		if (err)
615 			goto done;
616 	}
617 
618 	err = got_privsep_send_stop(imsg_fds[0]);
619 	child_err = got_privsep_wait_for_child(pid);
620 	if (child_err && err == NULL)
621 		err = child_err;
622 done:
623 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
624 		err = got_error_from_errno("close");
625 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
626 		err = got_error_from_errno("close");
627 	if (fd != -1 && close(fd) == -1 && err == NULL)
628 		err = got_error_from_errno2("close", gitconfig_path);
629 	free(ibuf);
630 	return err;
631 }
632 
633 static const struct got_error *
634 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
635 {
636 	const struct got_error *err = NULL;
637 	char *repo_gitconfig_path = NULL;
638 
639 	if (global_gitconfig_path) {
640 		/* Read settings from ~/.gitconfig. */
641 		int dummy_repo_version;
642 		err = parse_gitconfig_file(&dummy_repo_version,
643 		    &repo->global_gitconfig_author_name,
644 		    &repo->global_gitconfig_author_email,
645 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
646 		if (err)
647 			return err;
648 	}
649 
650 	/* Read repository's .git/config file. */
651 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
652 	if (repo_gitconfig_path == NULL)
653 		return got_error_from_errno("got_repo_get_path_gitconfig");
654 
655 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
656 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
657 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
658 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
659 	    repo_gitconfig_path);
660 	if (err)
661 		goto done;
662 done:
663 	free(repo_gitconfig_path);
664 	return err;
665 }
666 
667 static const struct got_error *
668 read_gotconfig(struct got_repository *repo)
669 {
670 	const struct got_error *err = NULL;
671 	char *gotconfig_path;
672 
673 	gotconfig_path = got_repo_get_path_gotconfig(repo);
674 	if (gotconfig_path == NULL)
675 		return got_error_from_errno("got_repo_get_path_gotconfig");
676 
677 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
678 	free(gotconfig_path);
679 	return err;
680 }
681 
682 /* Supported repository format extensions. */
683 static const char *const repo_extensions[] = {
684 	"noop",			/* Got supports repository format version 1. */
685 	"preciousObjects",	/* Supported by gotadmin cleanup. */
686 	"worktreeConfig",	/* Got does not care about Git work trees. */
687 };
688 
689 const struct got_error *
690 got_repo_open(struct got_repository **repop, const char *path,
691     const char *global_gitconfig_path, int *pack_fds)
692 {
693 	struct got_repository *repo = NULL;
694 	const struct got_error *err = NULL;
695 	char *repo_path = NULL;
696 	size_t i, j = 0;
697 	struct rlimit rl;
698 
699 	*repop = NULL;
700 
701 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
702 		return got_error_from_errno("getrlimit");
703 
704 	repo = calloc(1, sizeof(*repo));
705 	if (repo == NULL)
706 		return got_error_from_errno("calloc");
707 
708 	RB_INIT(&repo->packidx_bloom_filters);
709 	TAILQ_INIT(&repo->packidx_paths);
710 
711 	for (i = 0; i < nitems(repo->privsep_children); i++) {
712 		memset(&repo->privsep_children[i], 0,
713 		    sizeof(repo->privsep_children[0]));
714 		repo->privsep_children[i].imsg_fd = -1;
715 	}
716 
717 	err = got_object_cache_init(&repo->objcache,
718 	    GOT_OBJECT_CACHE_TYPE_OBJ);
719 	if (err)
720 		goto done;
721 	err = got_object_cache_init(&repo->treecache,
722 	    GOT_OBJECT_CACHE_TYPE_TREE);
723 	if (err)
724 		goto done;
725 	err = got_object_cache_init(&repo->commitcache,
726 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
727 	if (err)
728 		goto done;
729 	err = got_object_cache_init(&repo->tagcache,
730 	    GOT_OBJECT_CACHE_TYPE_TAG);
731 	if (err)
732 		goto done;
733 	err = got_object_cache_init(&repo->rawcache,
734 	    GOT_OBJECT_CACHE_TYPE_RAW);
735 	if (err)
736 		goto done;
737 
738 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
739 	if (repo->pack_cache_size > rl.rlim_cur / 8)
740 		repo->pack_cache_size = rl.rlim_cur / 8;
741 	for (i = 0; i < nitems(repo->packs); i++) {
742 		if (i < repo->pack_cache_size) {
743 			repo->packs[i].basefd = pack_fds[j++];
744 			repo->packs[i].accumfd = pack_fds[j++];
745 		} else {
746 			repo->packs[i].basefd = -1;
747 			repo->packs[i].accumfd = -1;
748 		}
749 	}
750 	repo->pinned_pack = -1;
751 	repo->pinned_packidx = -1;
752 	repo->pinned_pid = 0;
753 
754 	repo_path = realpath(path, NULL);
755 	if (repo_path == NULL) {
756 		err = got_error_from_errno2("realpath", path);
757 		goto done;
758 	}
759 
760 	for (;;) {
761 		char *parent_path;
762 
763 		err = open_repo(repo, repo_path);
764 		if (err == NULL)
765 			break;
766 		if (err->code != GOT_ERR_NOT_GIT_REPO)
767 			goto done;
768 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
769 			err = got_error(GOT_ERR_NOT_GIT_REPO);
770 			goto done;
771 		}
772 		err = got_path_dirname(&parent_path, repo_path);
773 		if (err)
774 			goto done;
775 		free(repo_path);
776 		repo_path = parent_path;
777 	}
778 
779 	err = read_gotconfig(repo);
780 	if (err)
781 		goto done;
782 
783 	err = read_gitconfig(repo, global_gitconfig_path);
784 	if (err)
785 		goto done;
786 	if (repo->gitconfig_repository_format_version != 0)
787 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
788 	for (i = 0; i < repo->nextensions; i++) {
789 		char *ext = repo->extensions[i];
790 		int j, supported = 0;
791 		for (j = 0; j < nitems(repo_extensions); j++) {
792 			if (strcmp(ext, repo_extensions[j]) == 0) {
793 				supported = 1;
794 				break;
795 			}
796 		}
797 		if (!supported) {
798 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
799 			goto done;
800 		}
801 	}
802 
803 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
804 done:
805 	if (err)
806 		got_repo_close(repo);
807 	else
808 		*repop = repo;
809 	free(repo_path);
810 	return err;
811 }
812 
813 const struct got_error *
814 got_repo_close(struct got_repository *repo)
815 {
816 	const struct got_error *err = NULL, *child_err;
817 	struct got_packidx_bloom_filter *bf;
818 	struct got_pathlist_entry *pe;
819 	size_t i;
820 
821 	for (i = 0; i < repo->pack_cache_size; i++) {
822 		if (repo->packidx_cache[i] == NULL)
823 			break;
824 		got_packidx_close(repo->packidx_cache[i]);
825 	}
826 
827 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
828 	    &repo->packidx_bloom_filters))) {
829 		RB_REMOVE(got_packidx_bloom_filter_tree,
830 		    &repo->packidx_bloom_filters, bf);
831 		free(bf->bloom);
832 		free(bf);
833 	}
834 
835 	for (i = 0; i < repo->pack_cache_size; i++)
836 		if (repo->packs[i].path_packfile)
837 			if (repo->packs[i].path_packfile)
838 				got_pack_close(&repo->packs[i]);
839 
840 	free(repo->path);
841 	free(repo->path_git_dir);
842 
843 	got_object_cache_close(&repo->objcache);
844 	got_object_cache_close(&repo->treecache);
845 	got_object_cache_close(&repo->commitcache);
846 	got_object_cache_close(&repo->tagcache);
847 	got_object_cache_close(&repo->rawcache);
848 
849 	for (i = 0; i < nitems(repo->privsep_children); i++) {
850 		if (repo->privsep_children[i].imsg_fd == -1)
851 			continue;
852 		imsg_clear(repo->privsep_children[i].ibuf);
853 		free(repo->privsep_children[i].ibuf);
854 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
855 		child_err = got_privsep_wait_for_child(
856 		    repo->privsep_children[i].pid);
857 		if (child_err && err == NULL)
858 			err = child_err;
859 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
860 		    err == NULL)
861 			err = got_error_from_errno("close");
862 	}
863 
864 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
865 	    err == NULL)
866 		err = got_error_from_errno("close");
867 
868 	if (repo->gotconfig)
869 		got_gotconfig_free(repo->gotconfig);
870 	free(repo->gitconfig_author_name);
871 	free(repo->gitconfig_author_email);
872 	for (i = 0; i < repo->ngitconfig_remotes; i++)
873 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
874 	free(repo->gitconfig_remotes);
875 	for (i = 0; i < repo->nextensions; i++)
876 		free(repo->extensions[i]);
877 	free(repo->extensions);
878 
879 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
880 		free((void *)pe->path);
881 	got_pathlist_free(&repo->packidx_paths);
882 	free(repo);
883 
884 	return err;
885 }
886 
887 void
888 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
889 {
890 	int i;
891 
892 	free(repo->name);
893 	repo->name = NULL;
894 	free(repo->fetch_url);
895 	repo->fetch_url = NULL;
896 	free(repo->send_url);
897 	repo->send_url = NULL;
898 	for (i = 0; i < repo->nfetch_branches; i++)
899 		free(repo->fetch_branches[i]);
900 	free(repo->fetch_branches);
901 	repo->fetch_branches = NULL;
902 	repo->nfetch_branches = 0;
903 	for (i = 0; i < repo->nsend_branches; i++)
904 		free(repo->send_branches[i]);
905 	free(repo->send_branches);
906 	repo->send_branches = NULL;
907 	repo->nsend_branches = 0;
908 }
909 
910 const struct got_error *
911 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
912     const char *input_path)
913 {
914 	const struct got_error *err = NULL;
915 	const char *repo_abspath = NULL;
916 	size_t repolen, len;
917 	char *canonpath, *path = NULL;
918 
919 	*in_repo_path = NULL;
920 
921 	canonpath = strdup(input_path);
922 	if (canonpath == NULL) {
923 		err = got_error_from_errno("strdup");
924 		goto done;
925 	}
926 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
927 	if (err)
928 		goto done;
929 
930 	repo_abspath = got_repo_get_path(repo);
931 
932 	if (canonpath[0] == '\0') {
933 		path = strdup(canonpath);
934 		if (path == NULL) {
935 			err = got_error_from_errno("strdup");
936 			goto done;
937 		}
938 	} else {
939 		path = realpath(canonpath, NULL);
940 		if (path == NULL) {
941 			if (errno != ENOENT) {
942 				err = got_error_from_errno2("realpath",
943 				    canonpath);
944 				goto done;
945 			}
946 			/*
947 			 * Path is not on disk.
948 			 * Assume it is already relative to repository root.
949 			 */
950 			path = strdup(canonpath);
951 			if (path == NULL) {
952 				err = got_error_from_errno("strdup");
953 				goto done;
954 			}
955 		}
956 
957 		repolen = strlen(repo_abspath);
958 		len = strlen(path);
959 
960 
961 		if (strcmp(path, repo_abspath) == 0) {
962 			free(path);
963 			path = strdup("");
964 			if (path == NULL) {
965 				err = got_error_from_errno("strdup");
966 				goto done;
967 			}
968 		} else if (len > repolen &&
969 		    got_path_is_child(path, repo_abspath, repolen)) {
970 			/* Matched an on-disk path inside repository. */
971 			if (got_repo_is_bare(repo)) {
972 				/*
973 				 * Matched an on-disk path inside repository
974 				 * database. Treat input as repository-relative.
975 				 */
976 				free(path);
977 				path = canonpath;
978 				canonpath = NULL;
979 			} else {
980 				char *child;
981 				/* Strip common prefix with repository path. */
982 				err = got_path_skip_common_ancestor(&child,
983 				    repo_abspath, path);
984 				if (err)
985 					goto done;
986 				free(path);
987 				path = child;
988 			}
989 		} else {
990 			/*
991 			 * Matched unrelated on-disk path.
992 			 * Treat input as repository-relative.
993 			 */
994 			free(path);
995 			path = canonpath;
996 			canonpath = NULL;
997 		}
998 	}
999 
1000 	/* Make in-repository path absolute */
1001 	if (path[0] != '/') {
1002 		char *abspath;
1003 		if (asprintf(&abspath, "/%s", path) == -1) {
1004 			err = got_error_from_errno("asprintf");
1005 			goto done;
1006 		}
1007 		free(path);
1008 		path = abspath;
1009 	}
1010 
1011 done:
1012 	free(canonpath);
1013 	if (err)
1014 		free(path);
1015 	else
1016 		*in_repo_path = path;
1017 	return err;
1018 }
1019 
1020 static const struct got_error *
1021 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1022     const char *path_packidx)
1023 {
1024 	const struct got_error *err = NULL;
1025 	size_t i;
1026 
1027 	for (i = 0; i < repo->pack_cache_size; i++) {
1028 		if (repo->packidx_cache[i] == NULL)
1029 			break;
1030 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1031 		    path_packidx) == 0) {
1032 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1033 		}
1034 	}
1035 	if (i == repo->pack_cache_size) {
1036 		do {
1037 			i--;
1038 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1039 		    i == repo->pinned_packidx);
1040 		err = got_packidx_close(repo->packidx_cache[i]);
1041 		if (err)
1042 			return err;
1043 	}
1044 
1045 	repo->packidx_cache[i] = packidx;
1046 
1047 	return NULL;
1048 }
1049 
1050 int
1051 got_repo_is_packidx_filename(const char *name, size_t len)
1052 {
1053 	if (len != GOT_PACKIDX_NAMELEN)
1054 		return 0;
1055 
1056 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1057 		return 0;
1058 
1059 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1060 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1061 		return 0;
1062 
1063 	return 1;
1064 }
1065 
1066 static struct got_packidx_bloom_filter *
1067 get_packidx_bloom_filter(struct got_repository *repo,
1068     const char *path, size_t path_len)
1069 {
1070 	struct got_packidx_bloom_filter key;
1071 
1072 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1073 		return NULL; /* XXX */
1074 	key.path_len = path_len;
1075 
1076 	return RB_FIND(got_packidx_bloom_filter_tree,
1077 	    &repo->packidx_bloom_filters, &key);
1078 }
1079 
1080 int
1081 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1082     const char *path_packidx, struct got_object_id *id)
1083 {
1084 	struct got_packidx_bloom_filter *bf;
1085 
1086 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1087 	if (bf)
1088 		return bloom_check(bf->bloom, id->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 				if (repo->pinned_packidx >= 0 &&
1176 				    repo->pinned_packidx < i)
1177 					repo->pinned_packidx++;
1178 				else if (repo->pinned_packidx == i)
1179 					repo->pinned_packidx = 0;
1180 			}
1181 			return NULL;
1182 		}
1183 	}
1184 	/* No luck. Search the filesystem. */
1185 
1186 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1187 		const char *path_packidx = pe->path;
1188 		int is_cached = 0;
1189 
1190 		if (!got_repo_check_packidx_bloom_filter(repo,
1191 		    pe->path, id))
1192 			continue; /* object will not be found in this index */
1193 
1194 		for (i = 0; i < repo->pack_cache_size; i++) {
1195 			if (repo->packidx_cache[i] == NULL)
1196 				break;
1197 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1198 			    path_packidx) == 0) {
1199 				is_cached = 1;
1200 				break;
1201 			}
1202 		}
1203 		if (is_cached)
1204 			continue; /* already searched */
1205 
1206 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1207 		    path_packidx, 0);
1208 		if (err)
1209 			goto done;
1210 
1211 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1212 		if (err)
1213 			goto done;
1214 
1215 		err = cache_packidx(repo, *packidx, path_packidx);
1216 		if (err)
1217 			goto done;
1218 
1219 		*idx = got_packidx_get_object_idx(*packidx, id);
1220 		if (*idx != -1) {
1221 			err = NULL; /* found the object */
1222 			goto done;
1223 		}
1224 	}
1225 
1226 	err = got_error_no_obj(id);
1227 done:
1228 	return err;
1229 }
1230 
1231 const struct got_error *
1232 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1233     struct got_repository *repo)
1234 {
1235 	const struct got_error *err = NULL;
1236 	DIR *packdir = NULL;
1237 	struct dirent *dent;
1238 	char *path_packidx = NULL;
1239 	int packdir_fd;
1240 
1241 	packdir_fd = openat(got_repo_get_fd(repo),
1242 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1243 	if (packdir_fd == -1) {
1244 		return got_error_from_errno_fmt("openat: %s/%s",
1245 		    got_repo_get_path_git_dir(repo),
1246 		    GOT_OBJECTS_PACK_DIR);
1247 	}
1248 
1249 	packdir = fdopendir(packdir_fd);
1250 	if (packdir == NULL) {
1251 		err = got_error_from_errno("fdopendir");
1252 		goto done;
1253 	}
1254 
1255 	while ((dent = readdir(packdir)) != NULL) {
1256 		if (!got_repo_is_packidx_filename(dent->d_name,
1257 		    strlen(dent->d_name)))
1258 			continue;
1259 
1260 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1261 		    dent->d_name) == -1) {
1262 			err = got_error_from_errno("asprintf");
1263 			path_packidx = NULL;
1264 			break;
1265 		}
1266 
1267 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1268 		if (err)
1269 			break;
1270 	}
1271 done:
1272 	if (err)
1273 		free(path_packidx);
1274 	if (packdir && closedir(packdir) != 0 && err == NULL)
1275 		err = got_error_from_errno("closedir");
1276 	return err;
1277 }
1278 
1279 const struct got_error *
1280 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1281     struct got_repository *repo)
1282 {
1283 	const struct got_error *err;
1284 	size_t i;
1285 
1286 	*packidx = NULL;
1287 
1288 	/* Search pack index cache. */
1289 	for (i = 0; i < repo->pack_cache_size; i++) {
1290 		if (repo->packidx_cache[i] == NULL)
1291 			break;
1292 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1293 		    path_packidx) == 0) {
1294 			*packidx = repo->packidx_cache[i];
1295 			return NULL;
1296 		}
1297 	}
1298 	/* No luck. Search the filesystem. */
1299 
1300 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1301 	    path_packidx, 0);
1302 	if (err)
1303 		return err;
1304 
1305 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1306 	if (err)
1307 		goto done;
1308 
1309 	err = cache_packidx(repo, *packidx, path_packidx);
1310 done:
1311 	if (err) {
1312 		got_packidx_close(*packidx);
1313 		*packidx = NULL;
1314 	}
1315 	return err;
1316 }
1317 
1318 static const struct got_error *
1319 read_packfile_hdr(int fd, struct got_packidx *packidx)
1320 {
1321 	const struct got_error *err = NULL;
1322 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1323 	struct got_packfile_hdr hdr;
1324 	ssize_t n;
1325 
1326 	n = read(fd, &hdr, sizeof(hdr));
1327 	if (n < 0)
1328 		return got_error_from_errno("read");
1329 	if (n != sizeof(hdr))
1330 		return got_error(GOT_ERR_BAD_PACKFILE);
1331 
1332 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1333 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1334 	    be32toh(hdr.nobjects) != totobj)
1335 		err = got_error(GOT_ERR_BAD_PACKFILE);
1336 
1337 	return err;
1338 }
1339 
1340 static const struct got_error *
1341 open_packfile(int *fd, struct got_repository *repo,
1342     const char *relpath, struct got_packidx *packidx)
1343 {
1344 	const struct got_error *err = NULL;
1345 
1346 	*fd = openat(got_repo_get_fd(repo), relpath,
1347 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1348 	if (*fd == -1)
1349 		return got_error_from_errno_fmt("openat: %s/%s",
1350 		    got_repo_get_path_git_dir(repo), relpath);
1351 
1352 	if (packidx) {
1353 		err = read_packfile_hdr(*fd, packidx);
1354 		if (err) {
1355 			close(*fd);
1356 			*fd = -1;
1357 		}
1358 	}
1359 
1360 	return err;
1361 }
1362 
1363 const struct got_error *
1364 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1365     const char *path_packfile, struct got_packidx *packidx)
1366 {
1367 	const struct got_error *err = NULL;
1368 	struct got_pack *pack = NULL;
1369 	struct stat sb;
1370 	size_t i;
1371 
1372 	if (packp)
1373 		*packp = NULL;
1374 
1375 	for (i = 0; i < repo->pack_cache_size; i++) {
1376 		pack = &repo->packs[i];
1377 		if (pack->path_packfile == NULL)
1378 			break;
1379 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1380 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1381 	}
1382 
1383 	if (i == repo->pack_cache_size) {
1384 		struct got_pack tmp;
1385 		do {
1386 			i--;
1387 		} while (i > 0 && repo->pinned_pack >= 0 &&
1388 		    i == repo->pinned_pack);
1389 		err = got_pack_close(&repo->packs[i]);
1390 		if (err)
1391 			return err;
1392 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1393 			return got_error_from_errno("ftruncate");
1394 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1395 			return got_error_from_errno("ftruncate");
1396 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1397 		memcpy(&repo->packs[i], &repo->packs[0],
1398 		    sizeof(repo->packs[i]));
1399 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1400 		if (repo->pinned_pack == 0)
1401 			repo->pinned_pack = i;
1402 		else if (repo->pinned_pack == i)
1403 			repo->pinned_pack = 0;
1404 		i = 0;
1405 	}
1406 
1407 	pack = &repo->packs[i];
1408 
1409 	pack->path_packfile = strdup(path_packfile);
1410 	if (pack->path_packfile == NULL) {
1411 		err = got_error_from_errno("strdup");
1412 		goto done;
1413 	}
1414 
1415 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1416 	if (err)
1417 		goto done;
1418 
1419 	if (fstat(pack->fd, &sb) != 0) {
1420 		err = got_error_from_errno("fstat");
1421 		goto done;
1422 	}
1423 	pack->filesize = sb.st_size;
1424 
1425 	pack->privsep_child = NULL;
1426 
1427 #ifndef GOT_PACK_NO_MMAP
1428 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1429 	    pack->fd, 0);
1430 	if (pack->map == MAP_FAILED) {
1431 		if (errno != ENOMEM) {
1432 			err = got_error_from_errno("mmap");
1433 			goto done;
1434 		}
1435 		pack->map = NULL; /* fall back to read(2) */
1436 	}
1437 #endif
1438 done:
1439 	if (err) {
1440 		if (pack) {
1441 			free(pack->path_packfile);
1442 			memset(pack, 0, sizeof(*pack));
1443 		}
1444 	} else if (packp)
1445 		*packp = pack;
1446 	return err;
1447 }
1448 
1449 struct got_pack *
1450 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1451 {
1452 	struct got_pack *pack = NULL;
1453 	size_t i;
1454 
1455 	for (i = 0; i < repo->pack_cache_size; i++) {
1456 		pack = &repo->packs[i];
1457 		if (pack->path_packfile == NULL)
1458 			break;
1459 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1460 			return pack;
1461 	}
1462 
1463 	return NULL;
1464 }
1465 
1466 const struct got_error *
1467 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1468     struct got_pack *pack)
1469 {
1470 	size_t i;
1471 	int pinned_pack = -1, pinned_packidx = -1;
1472 
1473 	for (i = 0; i < repo->pack_cache_size; i++) {
1474 		if (repo->packidx_cache[i] &&
1475 		    strcmp(repo->packidx_cache[i]->path_packidx,
1476 		    packidx->path_packidx) == 0)
1477 			pinned_packidx = i;
1478 		if (repo->packs[i].path_packfile &&
1479 		    strcmp(repo->packs[i].path_packfile,
1480 		    pack->path_packfile) == 0)
1481 			pinned_pack = i;
1482 	}
1483 
1484 	if (pinned_packidx == -1 || pinned_pack == -1)
1485 		return got_error(GOT_ERR_PIN_PACK);
1486 
1487 	repo->pinned_pack = pinned_pack;
1488 	repo->pinned_packidx = pinned_packidx;
1489 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1490 	return NULL;
1491 }
1492 
1493 struct got_pack *
1494 got_repo_get_pinned_pack(struct got_repository *repo)
1495 {
1496 	if (repo->pinned_pack >= 0 &&
1497 	    repo->pinned_pack < repo->pack_cache_size)
1498 		return &repo->packs[repo->pinned_pack];
1499 
1500 	return NULL;
1501 }
1502 
1503 void
1504 got_repo_unpin_pack(struct got_repository *repo)
1505 {
1506 	repo->pinned_packidx = -1;
1507 	repo->pinned_pack = -1;
1508 	repo->pinned_pid = 0;
1509 }
1510 
1511 const struct got_error *
1512 got_repo_init(const char *repo_path)
1513 {
1514 	const struct got_error *err = NULL;
1515 	const char *dirnames[] = {
1516 		GOT_OBJECTS_DIR,
1517 		GOT_OBJECTS_PACK_DIR,
1518 		GOT_REFS_DIR,
1519 	};
1520 	const char *description_str = "Unnamed repository; "
1521 	    "edit this file 'description' to name the repository.";
1522 	const char *headref_str = "ref: refs/heads/main";
1523 	const char *gitconfig_str = "[core]\n"
1524 	    "\trepositoryformatversion = 0\n"
1525 	    "\tfilemode = true\n"
1526 	    "\tbare = true\n";
1527 	char *path;
1528 	size_t i;
1529 
1530 	if (!got_path_dir_is_empty(repo_path))
1531 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1532 
1533 	for (i = 0; i < nitems(dirnames); i++) {
1534 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1535 			return got_error_from_errno("asprintf");
1536 		}
1537 		err = got_path_mkdir(path);
1538 		free(path);
1539 		if (err)
1540 			return err;
1541 	}
1542 
1543 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1544 		return got_error_from_errno("asprintf");
1545 	err = got_path_create_file(path, description_str);
1546 	free(path);
1547 	if (err)
1548 		return err;
1549 
1550 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1551 		return got_error_from_errno("asprintf");
1552 	err = got_path_create_file(path, headref_str);
1553 	free(path);
1554 	if (err)
1555 		return err;
1556 
1557 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1558 		return got_error_from_errno("asprintf");
1559 	err = got_path_create_file(path, gitconfig_str);
1560 	free(path);
1561 	if (err)
1562 		return err;
1563 
1564 	return NULL;
1565 }
1566 
1567 static const struct got_error *
1568 match_packed_object(struct got_object_id **unique_id,
1569     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1570 {
1571 	const struct got_error *err = NULL;
1572 	struct got_object_id_queue matched_ids;
1573 	struct got_pathlist_entry *pe;
1574 
1575 	STAILQ_INIT(&matched_ids);
1576 
1577 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1578 		const char *path_packidx = pe->path;
1579 		struct got_packidx *packidx;
1580 		struct got_object_qid *qid;
1581 
1582 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1583 		    path_packidx, 0);
1584 		if (err)
1585 			break;
1586 
1587 		err = got_packidx_match_id_str_prefix(&matched_ids,
1588 		    packidx, id_str_prefix);
1589 		if (err) {
1590 			got_packidx_close(packidx);
1591 			break;
1592 		}
1593 		err = got_packidx_close(packidx);
1594 		if (err)
1595 			break;
1596 
1597 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1598 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1599 				int matched_type;
1600 				err = got_object_get_type(&matched_type, repo,
1601 				    &qid->id);
1602 				if (err)
1603 					goto done;
1604 				if (matched_type != obj_type)
1605 					continue;
1606 			}
1607 			if (*unique_id == NULL) {
1608 				*unique_id = got_object_id_dup(&qid->id);
1609 				if (*unique_id == NULL) {
1610 					err = got_error_from_errno("malloc");
1611 					goto done;
1612 				}
1613 			} else {
1614 				if (got_object_id_cmp(*unique_id,
1615 				    &qid->id) == 0)
1616 					continue; /* packed multiple times */
1617 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1618 				goto done;
1619 			}
1620 		}
1621 	}
1622 done:
1623 	got_object_id_queue_free(&matched_ids);
1624 	if (err) {
1625 		free(*unique_id);
1626 		*unique_id = NULL;
1627 	}
1628 	return err;
1629 }
1630 
1631 static const struct got_error *
1632 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1633     const char *object_dir, const char *id_str_prefix, int obj_type,
1634     struct got_repository *repo)
1635 {
1636 	const struct got_error *err = NULL;
1637 	char *path;
1638 	DIR *dir = NULL;
1639 	struct dirent *dent;
1640 	struct got_object_id id;
1641 
1642 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1643 		err = got_error_from_errno("asprintf");
1644 		goto done;
1645 	}
1646 
1647 	dir = opendir(path);
1648 	if (dir == NULL) {
1649 		if (errno == ENOENT) {
1650 			err = NULL;
1651 			goto done;
1652 		}
1653 		err = got_error_from_errno2("opendir", path);
1654 		goto done;
1655 	}
1656 	while ((dent = readdir(dir)) != NULL) {
1657 		char *id_str;
1658 		int cmp;
1659 
1660 		if (strcmp(dent->d_name, ".") == 0 ||
1661 		    strcmp(dent->d_name, "..") == 0)
1662 			continue;
1663 
1664 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1665 			err = got_error_from_errno("asprintf");
1666 			goto done;
1667 		}
1668 
1669 		if (!got_parse_sha1_digest(id.sha1, id_str))
1670 			continue;
1671 
1672 		/*
1673 		 * Directory entries do not necessarily appear in
1674 		 * sorted order, so we must iterate over all of them.
1675 		 */
1676 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1677 		if (cmp != 0) {
1678 			free(id_str);
1679 			continue;
1680 		}
1681 
1682 		if (*unique_id == NULL) {
1683 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1684 				int matched_type;
1685 				err = got_object_get_type(&matched_type, repo,
1686 				    &id);
1687 				if (err)
1688 					goto done;
1689 				if (matched_type != obj_type)
1690 					continue;
1691 			}
1692 			*unique_id = got_object_id_dup(&id);
1693 			if (*unique_id == NULL) {
1694 				err = got_error_from_errno("got_object_id_dup");
1695 				free(id_str);
1696 				goto done;
1697 			}
1698 		} else {
1699 			if (got_object_id_cmp(*unique_id, &id) == 0)
1700 				continue; /* both packed and loose */
1701 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1702 			free(id_str);
1703 			goto done;
1704 		}
1705 	}
1706 done:
1707 	if (dir && closedir(dir) != 0 && err == NULL)
1708 		err = got_error_from_errno("closedir");
1709 	if (err) {
1710 		free(*unique_id);
1711 		*unique_id = NULL;
1712 	}
1713 	free(path);
1714 	return err;
1715 }
1716 
1717 const struct got_error *
1718 got_repo_match_object_id_prefix(struct got_object_id **id,
1719     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1720 {
1721 	const struct got_error *err = NULL;
1722 	char *path_objects = got_repo_get_path_objects(repo);
1723 	char *object_dir = NULL;
1724 	size_t len;
1725 	int i;
1726 
1727 	*id = NULL;
1728 
1729 	len = strlen(id_str_prefix);
1730 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1731 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1732 
1733 	for (i = 0; i < len; i++) {
1734 		if (isxdigit((unsigned char)id_str_prefix[i]))
1735 			continue;
1736 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1737 	}
1738 
1739 	if (len >= 2) {
1740 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1741 		if (err)
1742 			goto done;
1743 		object_dir = strndup(id_str_prefix, 2);
1744 		if (object_dir == NULL) {
1745 			err = got_error_from_errno("strdup");
1746 			goto done;
1747 		}
1748 		err = match_loose_object(id, path_objects, object_dir,
1749 		    id_str_prefix, obj_type, repo);
1750 	} else if (len == 1) {
1751 		int i;
1752 		for (i = 0; i < 0xf; i++) {
1753 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1754 			    == -1) {
1755 				err = got_error_from_errno("asprintf");
1756 				goto done;
1757 			}
1758 			err = match_packed_object(id, repo, object_dir,
1759 			    obj_type);
1760 			if (err)
1761 				goto done;
1762 			err = match_loose_object(id, path_objects, object_dir,
1763 			    id_str_prefix, obj_type, repo);
1764 			if (err)
1765 				goto done;
1766 		}
1767 	} else {
1768 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1769 		goto done;
1770 	}
1771 done:
1772 	free(object_dir);
1773 	if (err) {
1774 		free(*id);
1775 		*id = NULL;
1776 	} else if (*id == NULL) {
1777 		switch (obj_type) {
1778 		case GOT_OBJ_TYPE_BLOB:
1779 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1780 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1781 			break;
1782 		case GOT_OBJ_TYPE_TREE:
1783 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1784 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1785 			break;
1786 		case GOT_OBJ_TYPE_COMMIT:
1787 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1788 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1789 			break;
1790 		case GOT_OBJ_TYPE_TAG:
1791 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1792 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1793 			break;
1794 		default:
1795 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1796 			break;
1797 		}
1798 	}
1799 
1800 	return err;
1801 }
1802 
1803 const struct got_error *
1804 got_repo_match_object_id(struct got_object_id **id, char **label,
1805     const char *id_str, int obj_type, struct got_reflist_head *refs,
1806     struct got_repository *repo)
1807 {
1808 	const struct got_error *err;
1809 	struct got_tag_object *tag;
1810 	struct got_reference *ref = NULL;
1811 
1812 	*id = NULL;
1813 	if (label)
1814 		*label = NULL;
1815 
1816 	if (refs) {
1817 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1818 		    refs, repo);
1819 		if (err == NULL) {
1820 			*id = got_object_id_dup(
1821 			    got_object_tag_get_object_id(tag));
1822 			if (*id == NULL)
1823 				err = got_error_from_errno("got_object_id_dup");
1824 			else if (label && asprintf(label, "refs/tags/%s",
1825 			    got_object_tag_get_name(tag)) == -1) {
1826 				err = got_error_from_errno("asprintf");
1827 				free(*id);
1828 				*id = NULL;
1829 			}
1830 			got_object_tag_close(tag);
1831 			return err;
1832 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1833 		    err->code != GOT_ERR_NO_OBJ)
1834 			return err;
1835 	}
1836 
1837 	err = got_ref_open(&ref, repo, id_str, 0);
1838 	if (err == NULL) {
1839 		err = got_ref_resolve(id, repo, ref);
1840 		if (err)
1841 			goto done;
1842 		if (label) {
1843 			*label = strdup(got_ref_get_name(ref));
1844 			if (*label == NULL) {
1845 				err = got_error_from_errno("strdup");
1846 				goto done;
1847 			}
1848 		}
1849 	} else {
1850 		if (err->code != GOT_ERR_NOT_REF &&
1851 		    err->code != GOT_ERR_BAD_REF_NAME)
1852 			goto done;
1853 		err = got_repo_match_object_id_prefix(id, id_str,
1854 		    obj_type, repo);
1855 		if (err) {
1856 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1857 				err = got_error_not_ref(id_str);
1858 			goto done;
1859 		}
1860 		if (label) {
1861 			err = got_object_id_str(label, *id);
1862 			if (*label == NULL) {
1863 				err = got_error_from_errno("strdup");
1864 				goto done;
1865 			}
1866 		}
1867 	}
1868 done:
1869 	if (ref)
1870 		got_ref_close(ref);
1871 	return err;
1872 }
1873 
1874 const struct got_error *
1875 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1876     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1877 {
1878 	const struct got_error *err = NULL;
1879 	struct got_reflist_entry *re;
1880 	struct got_object_id *tag_id;
1881 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1882 
1883 	*tag = NULL;
1884 
1885 	TAILQ_FOREACH(re, refs, entry) {
1886 		const char *refname;
1887 		refname = got_ref_get_name(re->ref);
1888 		if (got_ref_is_symbolic(re->ref))
1889 			continue;
1890 		if (strncmp(refname, "refs/tags/", 10) != 0)
1891 			continue;
1892 		if (!name_is_absolute)
1893 			refname += strlen("refs/tags/");
1894 		if (strcmp(refname, name) != 0)
1895 			continue;
1896 		err = got_ref_resolve(&tag_id, repo, re->ref);
1897 		if (err)
1898 			break;
1899 		err = got_object_open_as_tag(tag, repo, tag_id);
1900 		free(tag_id);
1901 		if (err)
1902 			break;
1903 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1904 		    got_object_tag_get_object_type(*tag) == obj_type)
1905 			break;
1906 		got_object_tag_close(*tag);
1907 		*tag = NULL;
1908 	}
1909 
1910 	if (err == NULL && *tag == NULL)
1911 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1912 		    GOT_OBJ_LABEL_TAG, name);
1913 	return err;
1914 }
1915 
1916 static const struct got_error *
1917 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1918     const char *name, mode_t mode, struct got_object_id *blob_id)
1919 {
1920 	const struct got_error *err = NULL;
1921 
1922 	 *new_te = NULL;
1923 
1924 	*new_te = calloc(1, sizeof(**new_te));
1925 	if (*new_te == NULL)
1926 		return got_error_from_errno("calloc");
1927 
1928 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1929 	    sizeof((*new_te)->name)) {
1930 		err = got_error(GOT_ERR_NO_SPACE);
1931 		goto done;
1932 	}
1933 
1934 	if (S_ISLNK(mode)) {
1935 		(*new_te)->mode = S_IFLNK;
1936 	} else {
1937 		(*new_te)->mode = S_IFREG;
1938 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1939 	}
1940 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1941 done:
1942 	if (err && *new_te) {
1943 		free(*new_te);
1944 		*new_te = NULL;
1945 	}
1946 	return err;
1947 }
1948 
1949 static const struct got_error *
1950 import_file(struct got_tree_entry **new_te, struct dirent *de,
1951     const char *path, struct got_repository *repo)
1952 {
1953 	const struct got_error *err;
1954 	struct got_object_id *blob_id = NULL;
1955 	char *filepath;
1956 	struct stat sb;
1957 
1958 	if (asprintf(&filepath, "%s%s%s", path,
1959 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1960 		return got_error_from_errno("asprintf");
1961 
1962 	if (lstat(filepath, &sb) != 0) {
1963 		err = got_error_from_errno2("lstat", path);
1964 		goto done;
1965 	}
1966 
1967 	err = got_object_blob_create(&blob_id, filepath, repo);
1968 	if (err)
1969 		goto done;
1970 
1971 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1972 	    blob_id);
1973 done:
1974 	free(filepath);
1975 	if (err)
1976 		free(blob_id);
1977 	return err;
1978 }
1979 
1980 static const struct got_error *
1981 insert_tree_entry(struct got_tree_entry *new_te,
1982     struct got_pathlist_head *paths)
1983 {
1984 	const struct got_error *err = NULL;
1985 	struct got_pathlist_entry *new_pe;
1986 
1987 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1988 	if (err)
1989 		return err;
1990 	if (new_pe == NULL)
1991 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1992 	return NULL;
1993 }
1994 
1995 static const struct got_error *write_tree(struct got_object_id **,
1996     const char *, struct got_pathlist_head *, struct got_repository *,
1997     got_repo_import_cb progress_cb, void *progress_arg);
1998 
1999 static const struct got_error *
2000 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2001     const char *path, struct got_pathlist_head *ignores,
2002     struct got_repository *repo,
2003     got_repo_import_cb progress_cb, void *progress_arg)
2004 {
2005 	const struct got_error *err;
2006 	struct got_object_id *id = NULL;
2007 	char *subdirpath;
2008 
2009 	if (asprintf(&subdirpath, "%s%s%s", path,
2010 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2011 		return got_error_from_errno("asprintf");
2012 
2013 	(*new_te) = calloc(1, sizeof(**new_te));
2014 	if (*new_te == NULL)
2015 		return got_error_from_errno("calloc");
2016 	(*new_te)->mode = S_IFDIR;
2017 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2018 	    sizeof((*new_te)->name)) {
2019 		err = got_error(GOT_ERR_NO_SPACE);
2020 		goto done;
2021 	}
2022 	err = write_tree(&id, subdirpath, ignores,  repo,
2023 	    progress_cb, progress_arg);
2024 	if (err)
2025 		goto done;
2026 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2027 	
2028 done:
2029 	free(id);
2030 	free(subdirpath);
2031 	if (err) {
2032 		free(*new_te);
2033 		*new_te = NULL;
2034 	}
2035 	return err;
2036 }
2037 
2038 static const struct got_error *
2039 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2040     struct got_pathlist_head *ignores, struct got_repository *repo,
2041     got_repo_import_cb progress_cb, void *progress_arg)
2042 {
2043 	const struct got_error *err = NULL;
2044 	DIR *dir;
2045 	struct dirent *de;
2046 	int nentries;
2047 	struct got_tree_entry *new_te = NULL;
2048 	struct got_pathlist_head paths;
2049 	struct got_pathlist_entry *pe;
2050 
2051 	*new_tree_id = NULL;
2052 
2053 	TAILQ_INIT(&paths);
2054 
2055 	dir = opendir(path_dir);
2056 	if (dir == NULL) {
2057 		err = got_error_from_errno2("opendir", path_dir);
2058 		goto done;
2059 	}
2060 
2061 	nentries = 0;
2062 	while ((de = readdir(dir)) != NULL) {
2063 		int ignore = 0;
2064 		int type;
2065 
2066 		if (strcmp(de->d_name, ".") == 0 ||
2067 		    strcmp(de->d_name, "..") == 0)
2068 			continue;
2069 
2070 		TAILQ_FOREACH(pe, ignores, entry) {
2071 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2072 				ignore = 1;
2073 				break;
2074 			}
2075 		}
2076 		if (ignore)
2077 			continue;
2078 
2079 		err = got_path_dirent_type(&type, path_dir, de);
2080 		if (err)
2081 			goto done;
2082 
2083 		if (type == DT_DIR) {
2084 			err = import_subdir(&new_te, de, path_dir,
2085 			    ignores, repo, progress_cb, progress_arg);
2086 			if (err) {
2087 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2088 					goto done;
2089 				err = NULL;
2090 				continue;
2091 			}
2092 		} else if (type == DT_REG || type == DT_LNK) {
2093 			err = import_file(&new_te, de, path_dir, repo);
2094 			if (err)
2095 				goto done;
2096 		} else
2097 			continue;
2098 
2099 		err = insert_tree_entry(new_te, &paths);
2100 		if (err)
2101 			goto done;
2102 		nentries++;
2103 	}
2104 
2105 	if (TAILQ_EMPTY(&paths)) {
2106 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2107 		    "cannot create tree without any entries");
2108 		goto done;
2109 	}
2110 
2111 	TAILQ_FOREACH(pe, &paths, entry) {
2112 		struct got_tree_entry *te = pe->data;
2113 		char *path;
2114 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2115 			continue;
2116 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2117 			err = got_error_from_errno("asprintf");
2118 			goto done;
2119 		}
2120 		err = (*progress_cb)(progress_arg, path);
2121 		free(path);
2122 		if (err)
2123 			goto done;
2124 	}
2125 
2126 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2127 done:
2128 	if (dir)
2129 		closedir(dir);
2130 	got_pathlist_free(&paths);
2131 	return err;
2132 }
2133 
2134 const struct got_error *
2135 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2136     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2137     struct got_repository *repo, got_repo_import_cb progress_cb,
2138     void *progress_arg)
2139 {
2140 	const struct got_error *err;
2141 	struct got_object_id *new_tree_id;
2142 
2143 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2144 	    progress_cb, progress_arg);
2145 	if (err)
2146 		return err;
2147 
2148 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2149 	    author, time(NULL), author, time(NULL), logmsg, repo);
2150 	free(new_tree_id);
2151 	return err;
2152 }
2153 
2154 const struct got_error *
2155 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2156     struct got_repository *repo)
2157 {
2158 	const struct got_error *err = NULL;
2159 	char *path_objects = NULL, *path = NULL;
2160 	DIR *dir = NULL;
2161 	struct got_object_id id;
2162 	int i;
2163 
2164 	*nobjects = 0;
2165 	*ondisk_size = 0;
2166 
2167 	path_objects = got_repo_get_path_objects(repo);
2168 	if (path_objects == NULL)
2169 		return got_error_from_errno("got_repo_get_path_objects");
2170 
2171 	for (i = 0; i <= 0xff; i++) {
2172 		struct dirent *dent;
2173 
2174 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2175 			err = got_error_from_errno("asprintf");
2176 			break;
2177 		}
2178 
2179 		dir = opendir(path);
2180 		if (dir == NULL) {
2181 			if (errno == ENOENT) {
2182 				err = NULL;
2183 				continue;
2184 			}
2185 			err = got_error_from_errno2("opendir", path);
2186 			break;
2187 		}
2188 
2189 		while ((dent = readdir(dir)) != NULL) {
2190 			char *id_str;
2191 			int fd;
2192 			struct stat sb;
2193 
2194 			if (strcmp(dent->d_name, ".") == 0 ||
2195 			    strcmp(dent->d_name, "..") == 0)
2196 				continue;
2197 
2198 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2199 				err = got_error_from_errno("asprintf");
2200 				goto done;
2201 			}
2202 
2203 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2204 				free(id_str);
2205 				continue;
2206 			}
2207 			free(id_str);
2208 
2209 			err = got_object_open_loose_fd(&fd, &id, repo);
2210 			if (err)
2211 				goto done;
2212 
2213 			if (fstat(fd, &sb) == -1) {
2214 				err = got_error_from_errno("fstat");
2215 				close(fd);
2216 				goto done;
2217 			}
2218 			(*nobjects)++;
2219 			(*ondisk_size) += sb.st_size;
2220 
2221 			if (close(fd) == -1) {
2222 				err = got_error_from_errno("close");
2223 				goto done;
2224 			}
2225 		}
2226 
2227 		if (closedir(dir) != 0) {
2228 			err = got_error_from_errno("closedir");
2229 			goto done;
2230 		}
2231 		dir = NULL;
2232 
2233 		free(path);
2234 		path = NULL;
2235 	}
2236 done:
2237 	if (dir && closedir(dir) != 0 && err == NULL)
2238 		err = got_error_from_errno("closedir");
2239 
2240 	if (err) {
2241 		*nobjects = 0;
2242 		*ondisk_size = 0;
2243 	}
2244 	free(path_objects);
2245 	free(path);
2246 	return err;
2247 }
2248 
2249 const struct got_error *
2250 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2251     off_t *total_packsize, struct got_repository *repo)
2252 {
2253 	const struct got_error *err = NULL;
2254 	DIR *packdir = NULL;
2255 	struct dirent *dent;
2256 	struct got_packidx *packidx = NULL;
2257 	char *path_packidx;
2258 	char *path_packfile;
2259 	int packdir_fd;
2260 	struct stat sb;
2261 
2262 	*npackfiles = 0;
2263 	*nobjects = 0;
2264 	*total_packsize = 0;
2265 
2266 	packdir_fd = openat(got_repo_get_fd(repo),
2267 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2268 	if (packdir_fd == -1) {
2269 		return got_error_from_errno_fmt("openat: %s/%s",
2270 		    got_repo_get_path_git_dir(repo),
2271 		    GOT_OBJECTS_PACK_DIR);
2272 	}
2273 
2274 	packdir = fdopendir(packdir_fd);
2275 	if (packdir == NULL) {
2276 		err = got_error_from_errno("fdopendir");
2277 		goto done;
2278 	}
2279 
2280 	while ((dent = readdir(packdir)) != NULL) {
2281 		if (!got_repo_is_packidx_filename(dent->d_name,
2282 		    strlen(dent->d_name)))
2283 			continue;
2284 
2285 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2286 		    dent->d_name) == -1) {
2287 			err = got_error_from_errno("asprintf");
2288 			goto done;
2289 		}
2290 
2291 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2292 		    path_packidx, 0);
2293 		free(path_packidx);
2294 		if (err)
2295 			goto done;
2296 
2297 		if (fstat(packidx->fd, &sb) == -1)
2298 			goto done;
2299 		*total_packsize += sb.st_size;
2300 
2301 		err = got_packidx_get_packfile_path(&path_packfile,
2302 		    packidx->path_packidx);
2303 		if (err)
2304 			goto done;
2305 
2306 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2307 		    0) == -1) {
2308 			free(path_packfile);
2309 			goto done;
2310 		}
2311 		free(path_packfile);
2312 		*total_packsize += sb.st_size;
2313 
2314 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2315 
2316 		(*npackfiles)++;
2317 
2318 		got_packidx_close(packidx);
2319 		packidx = NULL;
2320 	}
2321 done:
2322 	if (packidx)
2323 		got_packidx_close(packidx);
2324 	if (packdir && closedir(packdir) != 0 && err == NULL)
2325 		err = got_error_from_errno("closedir");
2326 	if (err) {
2327 		*npackfiles = 0;
2328 		*nobjects = 0;
2329 		*total_packsize = 0;
2330 	}
2331 	return err;
2332 }
2333 
2334 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2335     got_packidx_bloom_filter_cmp);