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;
251 
252 	pack_fds_tmp = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(int));
253 	if (pack_fds_tmp == NULL)
254 		return got_error_from_errno("calloc");
255 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
256 	if (*pack_fds == NULL) {
257 		free(pack_fds_tmp);
258 		return got_error_from_errno("calloc");
259 	}
260 
261 	/*
262 	 * got_repo_pack_fds_close will try to close all of the
263 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
264 	 * a value from got_opentempfd(), which would result in a close(0) if
265 	 * we do not initialize to -1 here.
266 	 */
267 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
268 		pack_fds_tmp[i] = -1;
269 
270 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
271 		pack_fds_tmp[i] = got_opentempfd();
272 		if (pack_fds_tmp[i] == -1) {
273 			err = got_error_from_errno("got_opentempfd");
274 			got_repo_pack_fds_close(pack_fds_tmp);
275 			return err;
276 		}
277 	}
278 	memcpy(*pack_fds, pack_fds_tmp, GOT_PACK_NUM_TEMPFILES * sizeof(int));
279 	return err;
280 }
281 
282 const struct got_error *
283 got_repo_pack_fds_close(int *pack_fds)
284 {
285 	const struct got_error *err = NULL;
286 	int i;
287 
288 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
289 		if (pack_fds[i] == -1)
290 			continue;
291 		if (close(pack_fds[i]) == -1) {
292 			err = got_error_from_errno("close");
293 			break;
294 		}
295 	}
296 	free(pack_fds);
297 	return err;
298 }
299 
300 const struct got_error *
301 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
302     struct got_object *obj)
303 {
304 #ifndef GOT_NO_OBJ_CACHE
305 	const struct got_error *err = NULL;
306 	err = got_object_cache_add(&repo->objcache, id, obj);
307 	if (err) {
308 		if (err->code == GOT_ERR_OBJ_EXISTS ||
309 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
310 			err = NULL;
311 		return err;
312 	}
313 	obj->refcnt++;
314 #endif
315 	return NULL;
316 }
317 
318 struct got_object *
319 got_repo_get_cached_object(struct got_repository *repo,
320     struct got_object_id *id)
321 {
322 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
323 }
324 
325 const struct got_error *
326 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
327     struct got_tree_object *tree)
328 {
329 #ifndef GOT_NO_OBJ_CACHE
330 	const struct got_error *err = NULL;
331 	err = got_object_cache_add(&repo->treecache, id, tree);
332 	if (err) {
333 		if (err->code == GOT_ERR_OBJ_EXISTS ||
334 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
335 			err = NULL;
336 		return err;
337 	}
338 	tree->refcnt++;
339 #endif
340 	return NULL;
341 }
342 
343 struct got_tree_object *
344 got_repo_get_cached_tree(struct got_repository *repo,
345     struct got_object_id *id)
346 {
347 	return (struct got_tree_object *)got_object_cache_get(
348 	    &repo->treecache, id);
349 }
350 
351 const struct got_error *
352 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
353     struct got_commit_object *commit)
354 {
355 #ifndef GOT_NO_OBJ_CACHE
356 	const struct got_error *err = NULL;
357 	err = got_object_cache_add(&repo->commitcache, id, commit);
358 	if (err) {
359 		if (err->code == GOT_ERR_OBJ_EXISTS ||
360 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
361 			err = NULL;
362 		return err;
363 	}
364 	commit->refcnt++;
365 #endif
366 	return NULL;
367 }
368 
369 struct got_commit_object *
370 got_repo_get_cached_commit(struct got_repository *repo,
371     struct got_object_id *id)
372 {
373 	return (struct got_commit_object *)got_object_cache_get(
374 	    &repo->commitcache, id);
375 }
376 
377 const struct got_error *
378 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
379     struct got_tag_object *tag)
380 {
381 #ifndef GOT_NO_OBJ_CACHE
382 	const struct got_error *err = NULL;
383 	err = got_object_cache_add(&repo->tagcache, id, tag);
384 	if (err) {
385 		if (err->code == GOT_ERR_OBJ_EXISTS ||
386 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
387 			err = NULL;
388 		return err;
389 	}
390 	tag->refcnt++;
391 #endif
392 	return NULL;
393 }
394 
395 struct got_tag_object *
396 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
397 {
398 	return (struct got_tag_object *)got_object_cache_get(
399 	    &repo->tagcache, id);
400 }
401 
402 const struct got_error *
403 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
404     struct got_raw_object *raw)
405 {
406 #ifndef GOT_NO_OBJ_CACHE
407 	const struct got_error *err = NULL;
408 	err = got_object_cache_add(&repo->rawcache, id, raw);
409 	if (err) {
410 		if (err->code == GOT_ERR_OBJ_EXISTS ||
411 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
412 			err = NULL;
413 		return err;
414 	}
415 	raw->refcnt++;
416 #endif
417 	return NULL;
418 }
419 
420 
421 struct got_raw_object *
422 got_repo_get_cached_raw_object(struct got_repository *repo,
423     struct got_object_id *id)
424 {
425 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
426 }
427 
428 
429 static const struct got_error *
430 open_repo(struct got_repository *repo, const char *path)
431 {
432 	const struct got_error *err = NULL;
433 
434 	repo->gitdir_fd = -1;
435 
436 	/* bare git repository? */
437 	repo->path_git_dir = strdup(path);
438 	if (repo->path_git_dir == NULL)
439 		return got_error_from_errno("strdup");
440 	if (is_git_repo(repo)) {
441 		repo->path = strdup(repo->path_git_dir);
442 		if (repo->path == NULL) {
443 			err = got_error_from_errno("strdup");
444 			goto done;
445 		}
446 		repo->gitdir_fd = open(repo->path_git_dir,
447 		    O_DIRECTORY | O_CLOEXEC);
448 		if (repo->gitdir_fd == -1) {
449 			err = got_error_from_errno2("open",
450 			    repo->path_git_dir);
451 			goto done;
452 		}
453 		return NULL;
454 	}
455 
456 	/* git repository with working tree? */
457 	free(repo->path_git_dir);
458 	repo->path_git_dir = NULL;
459 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
460 		err = got_error_from_errno("asprintf");
461 		goto done;
462 	}
463 	if (is_git_repo(repo)) {
464 		repo->path = strdup(path);
465 		if (repo->path == NULL) {
466 			err = got_error_from_errno("strdup");
467 			goto done;
468 		}
469 		repo->gitdir_fd = open(repo->path_git_dir,
470 		    O_DIRECTORY | O_CLOEXEC);
471 		if (repo->gitdir_fd == -1) {
472 			err = got_error_from_errno2("open",
473 			    repo->path_git_dir);
474 			goto done;
475 		}
476 		return NULL;
477 	}
478 
479 	err = got_error(GOT_ERR_NOT_GIT_REPO);
480 done:
481 	if (err) {
482 		free(repo->path);
483 		repo->path = NULL;
484 		free(repo->path_git_dir);
485 		repo->path_git_dir = NULL;
486 		if (repo->gitdir_fd != -1)
487 			close(repo->gitdir_fd);
488 		repo->gitdir_fd = -1;
489 
490 	}
491 	return err;
492 }
493 
494 static const struct got_error *
495 parse_gitconfig_file(int *gitconfig_repository_format_version,
496     char **gitconfig_author_name, char **gitconfig_author_email,
497     struct got_remote_repo **remotes, int *nremotes,
498     char **gitconfig_owner, char ***extensions, int *nextensions,
499     const char *gitconfig_path)
500 {
501 	const struct got_error *err = NULL, *child_err = NULL;
502 	int fd = -1;
503 	int imsg_fds[2] = { -1, -1 };
504 	pid_t pid;
505 	struct imsgbuf *ibuf;
506 
507 	*gitconfig_repository_format_version = 0;
508 	if (extensions)
509 		*extensions = NULL;
510 	if (nextensions)
511 		*nextensions = 0;
512 	*gitconfig_author_name = NULL;
513 	*gitconfig_author_email = NULL;
514 	if (remotes)
515 		*remotes = NULL;
516 	if (nremotes)
517 		*nremotes = 0;
518 	if (gitconfig_owner)
519 		*gitconfig_owner = NULL;
520 
521 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
522 	if (fd == -1) {
523 		if (errno == ENOENT)
524 			return NULL;
525 		return got_error_from_errno2("open", gitconfig_path);
526 	}
527 
528 	ibuf = calloc(1, sizeof(*ibuf));
529 	if (ibuf == NULL) {
530 		err = got_error_from_errno("calloc");
531 		goto done;
532 	}
533 
534 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
535 		err = got_error_from_errno("socketpair");
536 		goto done;
537 	}
538 
539 	pid = fork();
540 	if (pid == -1) {
541 		err = got_error_from_errno("fork");
542 		goto done;
543 	} else if (pid == 0) {
544 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
545 		    gitconfig_path);
546 		/* not reached */
547 	}
548 
549 	if (close(imsg_fds[1]) == -1) {
550 		err = got_error_from_errno("close");
551 		goto done;
552 	}
553 	imsg_fds[1] = -1;
554 	imsg_init(ibuf, imsg_fds[0]);
555 
556 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
557 	if (err)
558 		goto done;
559 	fd = -1;
560 
561 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
562 	if (err)
563 		goto done;
564 
565 	err = got_privsep_recv_gitconfig_int(
566 	    gitconfig_repository_format_version, ibuf);
567 	if (err)
568 		goto done;
569 
570 	if (extensions && nextensions) {
571 		err = got_privsep_send_gitconfig_repository_extensions_req(
572 		    ibuf);
573 		if (err)
574 			goto done;
575 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
576 		if (err)
577 			goto done;
578 		if (*nextensions > 0) {
579 			int i;
580 			*extensions = calloc(*nextensions, sizeof(char *));
581 			if (*extensions == NULL) {
582 				err = got_error_from_errno("calloc");
583 				goto done;
584 			}
585 			for (i = 0; i < *nextensions; i++) {
586 				char *ext;
587 				err = got_privsep_recv_gitconfig_str(&ext,
588 				    ibuf);
589 				if (err)
590 					goto done;
591 				(*extensions)[i] = ext;
592 			}
593 		}
594 	}
595 
596 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
597 	if (err)
598 		goto done;
599 
600 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
601 	if (err)
602 		goto done;
603 
604 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
605 	if (err)
606 		goto done;
607 
608 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
609 	if (err)
610 		goto done;
611 
612 	if (remotes && nremotes) {
613 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
614 		if (err)
615 			goto done;
616 
617 		err = got_privsep_recv_gitconfig_remotes(remotes,
618 		    nremotes, ibuf);
619 		if (err)
620 			goto done;
621 	}
622 
623 	if (gitconfig_owner) {
624 		err = got_privsep_send_gitconfig_owner_req(ibuf);
625 		if (err)
626 			goto done;
627 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
628 		if (err)
629 			goto done;
630 	}
631 
632 	err = got_privsep_send_stop(imsg_fds[0]);
633 	child_err = got_privsep_wait_for_child(pid);
634 	if (child_err && err == NULL)
635 		err = child_err;
636 done:
637 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
638 		err = got_error_from_errno("close");
639 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
640 		err = got_error_from_errno("close");
641 	if (fd != -1 && close(fd) == -1 && err == NULL)
642 		err = got_error_from_errno2("close", gitconfig_path);
643 	free(ibuf);
644 	return err;
645 }
646 
647 static const struct got_error *
648 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
649 {
650 	const struct got_error *err = NULL;
651 	char *repo_gitconfig_path = NULL;
652 
653 	if (global_gitconfig_path) {
654 		/* Read settings from ~/.gitconfig. */
655 		int dummy_repo_version;
656 		err = parse_gitconfig_file(&dummy_repo_version,
657 		    &repo->global_gitconfig_author_name,
658 		    &repo->global_gitconfig_author_email,
659 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
660 		if (err)
661 			return err;
662 	}
663 
664 	/* Read repository's .git/config file. */
665 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
666 	if (repo_gitconfig_path == NULL)
667 		return got_error_from_errno("got_repo_get_path_gitconfig");
668 
669 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
670 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
671 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
672 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
673 	    repo_gitconfig_path);
674 	if (err)
675 		goto done;
676 done:
677 	free(repo_gitconfig_path);
678 	return err;
679 }
680 
681 static const struct got_error *
682 read_gotconfig(struct got_repository *repo)
683 {
684 	const struct got_error *err = NULL;
685 	char *gotconfig_path;
686 
687 	gotconfig_path = got_repo_get_path_gotconfig(repo);
688 	if (gotconfig_path == NULL)
689 		return got_error_from_errno("got_repo_get_path_gotconfig");
690 
691 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
692 	free(gotconfig_path);
693 	return err;
694 }
695 
696 /* Supported repository format extensions. */
697 static const char *const repo_extensions[] = {
698 	"noop",			/* Got supports repository format version 1. */
699 	"preciousObjects",	/* Supported by gotadmin cleanup. */
700 	"worktreeConfig",	/* Got does not care about Git work trees. */
701 };
702 
703 const struct got_error *
704 got_repo_open(struct got_repository **repop, const char *path,
705     const char *global_gitconfig_path, int *pack_fds)
706 {
707 	struct got_repository *repo = NULL;
708 	const struct got_error *err = NULL;
709 	char *repo_path = NULL;
710 	size_t i, j = 0;
711 	struct rlimit rl;
712 
713 	*repop = NULL;
714 
715 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
716 		return got_error_from_errno("getrlimit");
717 
718 	repo = calloc(1, sizeof(*repo));
719 	if (repo == NULL)
720 		return got_error_from_errno("calloc");
721 
722 	RB_INIT(&repo->packidx_bloom_filters);
723 	TAILQ_INIT(&repo->packidx_paths);
724 
725 	for (i = 0; i < nitems(repo->privsep_children); i++) {
726 		memset(&repo->privsep_children[i], 0,
727 		    sizeof(repo->privsep_children[0]));
728 		repo->privsep_children[i].imsg_fd = -1;
729 	}
730 
731 	err = got_object_cache_init(&repo->objcache,
732 	    GOT_OBJECT_CACHE_TYPE_OBJ);
733 	if (err)
734 		goto done;
735 	err = got_object_cache_init(&repo->treecache,
736 	    GOT_OBJECT_CACHE_TYPE_TREE);
737 	if (err)
738 		goto done;
739 	err = got_object_cache_init(&repo->commitcache,
740 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
741 	if (err)
742 		goto done;
743 	err = got_object_cache_init(&repo->tagcache,
744 	    GOT_OBJECT_CACHE_TYPE_TAG);
745 	if (err)
746 		goto done;
747 	err = got_object_cache_init(&repo->rawcache,
748 	    GOT_OBJECT_CACHE_TYPE_RAW);
749 	if (err)
750 		goto done;
751 
752 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
753 	if (repo->pack_cache_size > rl.rlim_cur / 8)
754 		repo->pack_cache_size = rl.rlim_cur / 8;
755 	for (i = 0; i < nitems(repo->packs); i++) {
756 		if (i < repo->pack_cache_size) {
757 			repo->packs[i].basefd = pack_fds[j++];
758 			repo->packs[i].accumfd = pack_fds[j++];
759 		} else {
760 			repo->packs[i].basefd = -1;
761 			repo->packs[i].accumfd = -1;
762 		}
763 	}
764 	repo->pinned_pack = -1;
765 	repo->pinned_packidx = -1;
766 	repo->pinned_pid = 0;
767 
768 	repo_path = realpath(path, NULL);
769 	if (repo_path == NULL) {
770 		err = got_error_from_errno2("realpath", path);
771 		goto done;
772 	}
773 
774 	for (;;) {
775 		char *parent_path;
776 
777 		err = open_repo(repo, repo_path);
778 		if (err == NULL)
779 			break;
780 		if (err->code != GOT_ERR_NOT_GIT_REPO)
781 			goto done;
782 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
783 			err = got_error(GOT_ERR_NOT_GIT_REPO);
784 			goto done;
785 		}
786 		err = got_path_dirname(&parent_path, repo_path);
787 		if (err)
788 			goto done;
789 		free(repo_path);
790 		repo_path = parent_path;
791 	}
792 
793 	err = read_gotconfig(repo);
794 	if (err)
795 		goto done;
796 
797 	err = read_gitconfig(repo, global_gitconfig_path);
798 	if (err)
799 		goto done;
800 	if (repo->gitconfig_repository_format_version != 0) {
801 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
802 		goto done;
803 	}
804 	for (i = 0; i < repo->nextensions; i++) {
805 		char *ext = repo->extensions[i];
806 		int j, supported = 0;
807 		for (j = 0; j < nitems(repo_extensions); j++) {
808 			if (strcmp(ext, repo_extensions[j]) == 0) {
809 				supported = 1;
810 				break;
811 			}
812 		}
813 		if (!supported) {
814 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
815 			goto done;
816 		}
817 	}
818 
819 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
820 done:
821 	if (err)
822 		got_repo_close(repo);
823 	else
824 		*repop = repo;
825 	free(repo_path);
826 	return err;
827 }
828 
829 const struct got_error *
830 got_repo_close(struct got_repository *repo)
831 {
832 	const struct got_error *err = NULL, *child_err;
833 	struct got_packidx_bloom_filter *bf;
834 	struct got_pathlist_entry *pe;
835 	size_t i;
836 
837 	for (i = 0; i < repo->pack_cache_size; i++) {
838 		if (repo->packidx_cache[i] == NULL)
839 			break;
840 		got_packidx_close(repo->packidx_cache[i]);
841 	}
842 
843 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
844 	    &repo->packidx_bloom_filters))) {
845 		RB_REMOVE(got_packidx_bloom_filter_tree,
846 		    &repo->packidx_bloom_filters, bf);
847 		free(bf->bloom);
848 		free(bf);
849 	}
850 
851 	for (i = 0; i < repo->pack_cache_size; i++)
852 		if (repo->packs[i].path_packfile)
853 			if (repo->packs[i].path_packfile)
854 				got_pack_close(&repo->packs[i]);
855 
856 	free(repo->path);
857 	free(repo->path_git_dir);
858 
859 	got_object_cache_close(&repo->objcache);
860 	got_object_cache_close(&repo->treecache);
861 	got_object_cache_close(&repo->commitcache);
862 	got_object_cache_close(&repo->tagcache);
863 	got_object_cache_close(&repo->rawcache);
864 
865 	for (i = 0; i < nitems(repo->privsep_children); i++) {
866 		if (repo->privsep_children[i].imsg_fd == -1)
867 			continue;
868 		imsg_clear(repo->privsep_children[i].ibuf);
869 		free(repo->privsep_children[i].ibuf);
870 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
871 		child_err = got_privsep_wait_for_child(
872 		    repo->privsep_children[i].pid);
873 		if (child_err && err == NULL)
874 			err = child_err;
875 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
876 		    err == NULL)
877 			err = got_error_from_errno("close");
878 	}
879 
880 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
881 	    err == NULL)
882 		err = got_error_from_errno("close");
883 
884 	if (repo->gotconfig)
885 		got_gotconfig_free(repo->gotconfig);
886 	free(repo->gitconfig_author_name);
887 	free(repo->gitconfig_author_email);
888 	for (i = 0; i < repo->ngitconfig_remotes; i++)
889 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
890 	free(repo->gitconfig_remotes);
891 	for (i = 0; i < repo->nextensions; i++)
892 		free(repo->extensions[i]);
893 	free(repo->extensions);
894 
895 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
896 		free((void *)pe->path);
897 	got_pathlist_free(&repo->packidx_paths);
898 	free(repo);
899 
900 	return err;
901 }
902 
903 void
904 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
905 {
906 	int i;
907 
908 	free(repo->name);
909 	repo->name = NULL;
910 	free(repo->fetch_url);
911 	repo->fetch_url = NULL;
912 	free(repo->send_url);
913 	repo->send_url = NULL;
914 	for (i = 0; i < repo->nfetch_branches; i++)
915 		free(repo->fetch_branches[i]);
916 	free(repo->fetch_branches);
917 	repo->fetch_branches = NULL;
918 	repo->nfetch_branches = 0;
919 	for (i = 0; i < repo->nsend_branches; i++)
920 		free(repo->send_branches[i]);
921 	free(repo->send_branches);
922 	repo->send_branches = NULL;
923 	repo->nsend_branches = 0;
924 }
925 
926 const struct got_error *
927 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
928     const char *input_path)
929 {
930 	const struct got_error *err = NULL;
931 	const char *repo_abspath = NULL;
932 	size_t repolen, len;
933 	char *canonpath, *path = NULL;
934 
935 	*in_repo_path = NULL;
936 
937 	canonpath = strdup(input_path);
938 	if (canonpath == NULL) {
939 		err = got_error_from_errno("strdup");
940 		goto done;
941 	}
942 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
943 	if (err)
944 		goto done;
945 
946 	repo_abspath = got_repo_get_path(repo);
947 
948 	if (canonpath[0] == '\0') {
949 		path = strdup(canonpath);
950 		if (path == NULL) {
951 			err = got_error_from_errno("strdup");
952 			goto done;
953 		}
954 	} else {
955 		path = realpath(canonpath, NULL);
956 		if (path == NULL) {
957 			if (errno != ENOENT) {
958 				err = got_error_from_errno2("realpath",
959 				    canonpath);
960 				goto done;
961 			}
962 			/*
963 			 * Path is not on disk.
964 			 * Assume it is already relative to repository root.
965 			 */
966 			path = strdup(canonpath);
967 			if (path == NULL) {
968 				err = got_error_from_errno("strdup");
969 				goto done;
970 			}
971 		}
972 
973 		repolen = strlen(repo_abspath);
974 		len = strlen(path);
975 
976 
977 		if (strcmp(path, repo_abspath) == 0) {
978 			free(path);
979 			path = strdup("");
980 			if (path == NULL) {
981 				err = got_error_from_errno("strdup");
982 				goto done;
983 			}
984 		} else if (len > repolen &&
985 		    got_path_is_child(path, repo_abspath, repolen)) {
986 			/* Matched an on-disk path inside repository. */
987 			if (got_repo_is_bare(repo)) {
988 				/*
989 				 * Matched an on-disk path inside repository
990 				 * database. Treat input as repository-relative.
991 				 */
992 				free(path);
993 				path = canonpath;
994 				canonpath = NULL;
995 			} else {
996 				char *child;
997 				/* Strip common prefix with repository path. */
998 				err = got_path_skip_common_ancestor(&child,
999 				    repo_abspath, path);
1000 				if (err)
1001 					goto done;
1002 				free(path);
1003 				path = child;
1004 			}
1005 		} else {
1006 			/*
1007 			 * Matched unrelated on-disk path.
1008 			 * Treat input as repository-relative.
1009 			 */
1010 			free(path);
1011 			path = canonpath;
1012 			canonpath = NULL;
1013 		}
1014 	}
1015 
1016 	/* Make in-repository path absolute */
1017 	if (path[0] != '/') {
1018 		char *abspath;
1019 		if (asprintf(&abspath, "/%s", path) == -1) {
1020 			err = got_error_from_errno("asprintf");
1021 			goto done;
1022 		}
1023 		free(path);
1024 		path = abspath;
1025 	}
1026 
1027 done:
1028 	free(canonpath);
1029 	if (err)
1030 		free(path);
1031 	else
1032 		*in_repo_path = path;
1033 	return err;
1034 }
1035 
1036 static const struct got_error *
1037 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1038     const char *path_packidx)
1039 {
1040 	const struct got_error *err = NULL;
1041 	size_t i;
1042 
1043 	for (i = 0; i < repo->pack_cache_size; i++) {
1044 		if (repo->packidx_cache[i] == NULL)
1045 			break;
1046 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1047 		    path_packidx) == 0) {
1048 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1049 		}
1050 	}
1051 	if (i == repo->pack_cache_size) {
1052 		do {
1053 			i--;
1054 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1055 		    i == repo->pinned_packidx);
1056 		err = got_packidx_close(repo->packidx_cache[i]);
1057 		if (err)
1058 			return err;
1059 	}
1060 
1061 	repo->packidx_cache[i] = packidx;
1062 
1063 	return NULL;
1064 }
1065 
1066 int
1067 got_repo_is_packidx_filename(const char *name, size_t len)
1068 {
1069 	if (len != GOT_PACKIDX_NAMELEN)
1070 		return 0;
1071 
1072 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1073 		return 0;
1074 
1075 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1076 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1077 		return 0;
1078 
1079 	return 1;
1080 }
1081 
1082 static struct got_packidx_bloom_filter *
1083 get_packidx_bloom_filter(struct got_repository *repo,
1084     const char *path, size_t path_len)
1085 {
1086 	struct got_packidx_bloom_filter key;
1087 
1088 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1089 		return NULL; /* XXX */
1090 	key.path_len = path_len;
1091 
1092 	return RB_FIND(got_packidx_bloom_filter_tree,
1093 	    &repo->packidx_bloom_filters, &key);
1094 }
1095 
1096 int
1097 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1098     const char *path_packidx, struct got_object_id *id)
1099 {
1100 	struct got_packidx_bloom_filter *bf;
1101 
1102 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1103 	if (bf)
1104 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1105 
1106 	/* No bloom filter means this pack index must be searched. */
1107 	return 1;
1108 }
1109 
1110 static const struct got_error *
1111 add_packidx_bloom_filter(struct got_repository *repo,
1112     struct got_packidx *packidx, const char *path_packidx)
1113 {
1114 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1115 	struct got_packidx_bloom_filter *bf;
1116 	size_t len;
1117 
1118 	/*
1119 	 * Don't use bloom filters for very large pack index files.
1120 	 * Large pack files will contain a relatively large fraction
1121 	 * of our objects so we will likely need to visit them anyway.
1122 	 * The more objects a pack file contains the higher the probability
1123 	 * of a false-positive match from the bloom filter. And reading
1124 	 * all object IDs from a large pack index file can be expensive.
1125 	 */
1126 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1127 		return NULL;
1128 
1129 	/* Do we already have a filter for this pack index? */
1130 	if (get_packidx_bloom_filter(repo, path_packidx,
1131 	    strlen(path_packidx)) != NULL)
1132 		return NULL;
1133 
1134 	bf = calloc(1, sizeof(*bf));
1135 	if (bf == NULL)
1136 		return got_error_from_errno("calloc");
1137 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1138 	if (bf->bloom == NULL) {
1139 		free(bf);
1140 		return got_error_from_errno("calloc");
1141 	}
1142 	
1143 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1144 	if (len >= sizeof(bf->path)) {
1145 		free(bf->bloom);
1146 		free(bf);
1147 		return got_error(GOT_ERR_NO_SPACE);
1148 	}
1149 	bf->path_len = len;
1150 
1151 	/* Minimum size supported by our bloom filter is 1000 entries. */
1152 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1153 	for (i = 0; i < nobjects; i++) {
1154 		struct got_packidx_object_id *id;
1155 		id = &packidx->hdr.sorted_ids[i];
1156 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1157 	}
1158 
1159 	RB_INSERT(got_packidx_bloom_filter_tree,
1160 	    &repo->packidx_bloom_filters, bf);
1161 	return NULL;
1162 }
1163 
1164 const struct got_error *
1165 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1166     struct got_repository *repo, struct got_object_id *id)
1167 {
1168 	const struct got_error *err;
1169 	struct got_pathlist_entry *pe;
1170 	size_t i;
1171 
1172 	/* Search pack index cache. */
1173 	for (i = 0; i < repo->pack_cache_size; i++) {
1174 		if (repo->packidx_cache[i] == NULL)
1175 			break;
1176 		if (!got_repo_check_packidx_bloom_filter(repo,
1177 		    repo->packidx_cache[i]->path_packidx, id))
1178 			continue; /* object will not be found in this index */
1179 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1180 		if (*idx != -1) {
1181 			*packidx = repo->packidx_cache[i];
1182 			/*
1183 			 * Move this cache entry to the front. Repeatedly
1184 			 * searching a wrong pack index can be expensive.
1185 			 */
1186 			if (i > 0) {
1187 				memmove(&repo->packidx_cache[1],
1188 				    &repo->packidx_cache[0],
1189 				    i * sizeof(repo->packidx_cache[0]));
1190 				repo->packidx_cache[0] = *packidx;
1191 				if (repo->pinned_packidx >= 0 &&
1192 				    repo->pinned_packidx < i)
1193 					repo->pinned_packidx++;
1194 				else if (repo->pinned_packidx == i)
1195 					repo->pinned_packidx = 0;
1196 			}
1197 			return NULL;
1198 		}
1199 	}
1200 	/* No luck. Search the filesystem. */
1201 
1202 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1203 		const char *path_packidx = pe->path;
1204 		int is_cached = 0;
1205 
1206 		if (!got_repo_check_packidx_bloom_filter(repo,
1207 		    pe->path, id))
1208 			continue; /* object will not be found in this index */
1209 
1210 		for (i = 0; i < repo->pack_cache_size; i++) {
1211 			if (repo->packidx_cache[i] == NULL)
1212 				break;
1213 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1214 			    path_packidx) == 0) {
1215 				is_cached = 1;
1216 				break;
1217 			}
1218 		}
1219 		if (is_cached)
1220 			continue; /* already searched */
1221 
1222 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1223 		    path_packidx, 0);
1224 		if (err)
1225 			goto done;
1226 
1227 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1228 		if (err)
1229 			goto done;
1230 
1231 		err = cache_packidx(repo, *packidx, path_packidx);
1232 		if (err)
1233 			goto done;
1234 
1235 		*idx = got_packidx_get_object_idx(*packidx, id);
1236 		if (*idx != -1) {
1237 			err = NULL; /* found the object */
1238 			goto done;
1239 		}
1240 	}
1241 
1242 	err = got_error_no_obj(id);
1243 done:
1244 	return err;
1245 }
1246 
1247 const struct got_error *
1248 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1249     struct got_repository *repo)
1250 {
1251 	const struct got_error *err = NULL;
1252 	DIR *packdir = NULL;
1253 	struct dirent *dent;
1254 	char *path_packidx = NULL;
1255 	int packdir_fd;
1256 
1257 	packdir_fd = openat(got_repo_get_fd(repo),
1258 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1259 	if (packdir_fd == -1) {
1260 		return got_error_from_errno_fmt("openat: %s/%s",
1261 		    got_repo_get_path_git_dir(repo),
1262 		    GOT_OBJECTS_PACK_DIR);
1263 	}
1264 
1265 	packdir = fdopendir(packdir_fd);
1266 	if (packdir == NULL) {
1267 		err = got_error_from_errno("fdopendir");
1268 		goto done;
1269 	}
1270 
1271 	while ((dent = readdir(packdir)) != NULL) {
1272 		if (!got_repo_is_packidx_filename(dent->d_name,
1273 		    strlen(dent->d_name)))
1274 			continue;
1275 
1276 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1277 		    dent->d_name) == -1) {
1278 			err = got_error_from_errno("asprintf");
1279 			path_packidx = NULL;
1280 			break;
1281 		}
1282 
1283 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1284 		if (err)
1285 			break;
1286 	}
1287 done:
1288 	if (err)
1289 		free(path_packidx);
1290 	if (packdir && closedir(packdir) != 0 && err == NULL)
1291 		err = got_error_from_errno("closedir");
1292 	return err;
1293 }
1294 
1295 const struct got_error *
1296 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1297     struct got_repository *repo)
1298 {
1299 	const struct got_error *err;
1300 	size_t i;
1301 
1302 	*packidx = NULL;
1303 
1304 	/* Search pack index cache. */
1305 	for (i = 0; i < repo->pack_cache_size; i++) {
1306 		if (repo->packidx_cache[i] == NULL)
1307 			break;
1308 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1309 		    path_packidx) == 0) {
1310 			*packidx = repo->packidx_cache[i];
1311 			return NULL;
1312 		}
1313 	}
1314 	/* No luck. Search the filesystem. */
1315 
1316 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1317 	    path_packidx, 0);
1318 	if (err)
1319 		return err;
1320 
1321 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1322 	if (err)
1323 		goto done;
1324 
1325 	err = cache_packidx(repo, *packidx, path_packidx);
1326 done:
1327 	if (err) {
1328 		got_packidx_close(*packidx);
1329 		*packidx = NULL;
1330 	}
1331 	return err;
1332 }
1333 
1334 static const struct got_error *
1335 read_packfile_hdr(int fd, struct got_packidx *packidx)
1336 {
1337 	const struct got_error *err = NULL;
1338 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1339 	struct got_packfile_hdr hdr;
1340 	ssize_t n;
1341 
1342 	n = read(fd, &hdr, sizeof(hdr));
1343 	if (n < 0)
1344 		return got_error_from_errno("read");
1345 	if (n != sizeof(hdr))
1346 		return got_error(GOT_ERR_BAD_PACKFILE);
1347 
1348 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1349 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1350 	    be32toh(hdr.nobjects) != totobj)
1351 		err = got_error(GOT_ERR_BAD_PACKFILE);
1352 
1353 	return err;
1354 }
1355 
1356 static const struct got_error *
1357 open_packfile(int *fd, struct got_repository *repo,
1358     const char *relpath, struct got_packidx *packidx)
1359 {
1360 	const struct got_error *err = NULL;
1361 
1362 	*fd = openat(got_repo_get_fd(repo), relpath,
1363 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1364 	if (*fd == -1)
1365 		return got_error_from_errno_fmt("openat: %s/%s",
1366 		    got_repo_get_path_git_dir(repo), relpath);
1367 
1368 	if (packidx) {
1369 		err = read_packfile_hdr(*fd, packidx);
1370 		if (err) {
1371 			close(*fd);
1372 			*fd = -1;
1373 		}
1374 	}
1375 
1376 	return err;
1377 }
1378 
1379 const struct got_error *
1380 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1381     const char *path_packfile, struct got_packidx *packidx)
1382 {
1383 	const struct got_error *err = NULL;
1384 	struct got_pack *pack = NULL;
1385 	struct stat sb;
1386 	size_t i;
1387 
1388 	if (packp)
1389 		*packp = NULL;
1390 
1391 	for (i = 0; i < repo->pack_cache_size; i++) {
1392 		pack = &repo->packs[i];
1393 		if (pack->path_packfile == NULL)
1394 			break;
1395 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1396 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1397 	}
1398 
1399 	if (i == repo->pack_cache_size) {
1400 		struct got_pack tmp;
1401 		do {
1402 			i--;
1403 		} while (i > 0 && repo->pinned_pack >= 0 &&
1404 		    i == repo->pinned_pack);
1405 		err = got_pack_close(&repo->packs[i]);
1406 		if (err)
1407 			return err;
1408 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1409 			return got_error_from_errno("ftruncate");
1410 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1411 			return got_error_from_errno("ftruncate");
1412 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1413 		memcpy(&repo->packs[i], &repo->packs[0],
1414 		    sizeof(repo->packs[i]));
1415 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1416 		if (repo->pinned_pack == 0)
1417 			repo->pinned_pack = i;
1418 		else if (repo->pinned_pack == i)
1419 			repo->pinned_pack = 0;
1420 		i = 0;
1421 	}
1422 
1423 	pack = &repo->packs[i];
1424 
1425 	pack->path_packfile = strdup(path_packfile);
1426 	if (pack->path_packfile == NULL) {
1427 		err = got_error_from_errno("strdup");
1428 		goto done;
1429 	}
1430 
1431 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1432 	if (err)
1433 		goto done;
1434 
1435 	if (fstat(pack->fd, &sb) != 0) {
1436 		err = got_error_from_errno("fstat");
1437 		goto done;
1438 	}
1439 	pack->filesize = sb.st_size;
1440 
1441 	pack->privsep_child = NULL;
1442 
1443 #ifndef GOT_PACK_NO_MMAP
1444 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1445 	    pack->fd, 0);
1446 	if (pack->map == MAP_FAILED) {
1447 		if (errno != ENOMEM) {
1448 			err = got_error_from_errno("mmap");
1449 			goto done;
1450 		}
1451 		pack->map = NULL; /* fall back to read(2) */
1452 	}
1453 #endif
1454 done:
1455 	if (err) {
1456 		if (pack) {
1457 			free(pack->path_packfile);
1458 			memset(pack, 0, sizeof(*pack));
1459 		}
1460 	} else if (packp)
1461 		*packp = pack;
1462 	return err;
1463 }
1464 
1465 struct got_pack *
1466 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1467 {
1468 	struct got_pack *pack = NULL;
1469 	size_t i;
1470 
1471 	for (i = 0; i < repo->pack_cache_size; i++) {
1472 		pack = &repo->packs[i];
1473 		if (pack->path_packfile == NULL)
1474 			break;
1475 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1476 			return pack;
1477 	}
1478 
1479 	return NULL;
1480 }
1481 
1482 const struct got_error *
1483 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1484     struct got_pack *pack)
1485 {
1486 	size_t i;
1487 	int pinned_pack = -1, pinned_packidx = -1;
1488 
1489 	for (i = 0; i < repo->pack_cache_size; i++) {
1490 		if (repo->packidx_cache[i] &&
1491 		    strcmp(repo->packidx_cache[i]->path_packidx,
1492 		    packidx->path_packidx) == 0)
1493 			pinned_packidx = i;
1494 		if (repo->packs[i].path_packfile &&
1495 		    strcmp(repo->packs[i].path_packfile,
1496 		    pack->path_packfile) == 0)
1497 			pinned_pack = i;
1498 	}
1499 
1500 	if (pinned_packidx == -1 || pinned_pack == -1)
1501 		return got_error(GOT_ERR_PIN_PACK);
1502 
1503 	repo->pinned_pack = pinned_pack;
1504 	repo->pinned_packidx = pinned_packidx;
1505 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1506 	return NULL;
1507 }
1508 
1509 struct got_pack *
1510 got_repo_get_pinned_pack(struct got_repository *repo)
1511 {
1512 	if (repo->pinned_pack >= 0 &&
1513 	    repo->pinned_pack < repo->pack_cache_size)
1514 		return &repo->packs[repo->pinned_pack];
1515 
1516 	return NULL;
1517 }
1518 
1519 void
1520 got_repo_unpin_pack(struct got_repository *repo)
1521 {
1522 	repo->pinned_packidx = -1;
1523 	repo->pinned_pack = -1;
1524 	repo->pinned_pid = 0;
1525 }
1526 
1527 const struct got_error *
1528 got_repo_init(const char *repo_path)
1529 {
1530 	const struct got_error *err = NULL;
1531 	const char *dirnames[] = {
1532 		GOT_OBJECTS_DIR,
1533 		GOT_OBJECTS_PACK_DIR,
1534 		GOT_REFS_DIR,
1535 	};
1536 	const char *description_str = "Unnamed repository; "
1537 	    "edit this file 'description' to name the repository.";
1538 	const char *headref_str = "ref: refs/heads/main";
1539 	const char *gitconfig_str = "[core]\n"
1540 	    "\trepositoryformatversion = 0\n"
1541 	    "\tfilemode = true\n"
1542 	    "\tbare = true\n";
1543 	char *path;
1544 	size_t i;
1545 
1546 	if (!got_path_dir_is_empty(repo_path))
1547 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1548 
1549 	for (i = 0; i < nitems(dirnames); i++) {
1550 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1551 			return got_error_from_errno("asprintf");
1552 		}
1553 		err = got_path_mkdir(path);
1554 		free(path);
1555 		if (err)
1556 			return err;
1557 	}
1558 
1559 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1560 		return got_error_from_errno("asprintf");
1561 	err = got_path_create_file(path, description_str);
1562 	free(path);
1563 	if (err)
1564 		return err;
1565 
1566 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1567 		return got_error_from_errno("asprintf");
1568 	err = got_path_create_file(path, headref_str);
1569 	free(path);
1570 	if (err)
1571 		return err;
1572 
1573 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1574 		return got_error_from_errno("asprintf");
1575 	err = got_path_create_file(path, gitconfig_str);
1576 	free(path);
1577 	if (err)
1578 		return err;
1579 
1580 	return NULL;
1581 }
1582 
1583 static const struct got_error *
1584 match_packed_object(struct got_object_id **unique_id,
1585     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1586 {
1587 	const struct got_error *err = NULL;
1588 	struct got_object_id_queue matched_ids;
1589 	struct got_pathlist_entry *pe;
1590 
1591 	STAILQ_INIT(&matched_ids);
1592 
1593 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1594 		const char *path_packidx = pe->path;
1595 		struct got_packidx *packidx;
1596 		struct got_object_qid *qid;
1597 
1598 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1599 		    path_packidx, 0);
1600 		if (err)
1601 			break;
1602 
1603 		err = got_packidx_match_id_str_prefix(&matched_ids,
1604 		    packidx, id_str_prefix);
1605 		if (err) {
1606 			got_packidx_close(packidx);
1607 			break;
1608 		}
1609 		err = got_packidx_close(packidx);
1610 		if (err)
1611 			break;
1612 
1613 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1614 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1615 				int matched_type;
1616 				err = got_object_get_type(&matched_type, repo,
1617 				    &qid->id);
1618 				if (err)
1619 					goto done;
1620 				if (matched_type != obj_type)
1621 					continue;
1622 			}
1623 			if (*unique_id == NULL) {
1624 				*unique_id = got_object_id_dup(&qid->id);
1625 				if (*unique_id == NULL) {
1626 					err = got_error_from_errno("malloc");
1627 					goto done;
1628 				}
1629 			} else {
1630 				if (got_object_id_cmp(*unique_id,
1631 				    &qid->id) == 0)
1632 					continue; /* packed multiple times */
1633 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1634 				goto done;
1635 			}
1636 		}
1637 	}
1638 done:
1639 	got_object_id_queue_free(&matched_ids);
1640 	if (err) {
1641 		free(*unique_id);
1642 		*unique_id = NULL;
1643 	}
1644 	return err;
1645 }
1646 
1647 static const struct got_error *
1648 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1649     const char *object_dir, const char *id_str_prefix, int obj_type,
1650     struct got_repository *repo)
1651 {
1652 	const struct got_error *err = NULL;
1653 	char *path;
1654 	DIR *dir = NULL;
1655 	struct dirent *dent;
1656 	struct got_object_id id;
1657 
1658 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1659 		err = got_error_from_errno("asprintf");
1660 		goto done;
1661 	}
1662 
1663 	dir = opendir(path);
1664 	if (dir == NULL) {
1665 		if (errno == ENOENT) {
1666 			err = NULL;
1667 			goto done;
1668 		}
1669 		err = got_error_from_errno2("opendir", path);
1670 		goto done;
1671 	}
1672 	while ((dent = readdir(dir)) != NULL) {
1673 		char *id_str;
1674 		int cmp;
1675 
1676 		if (strcmp(dent->d_name, ".") == 0 ||
1677 		    strcmp(dent->d_name, "..") == 0)
1678 			continue;
1679 
1680 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1681 			err = got_error_from_errno("asprintf");
1682 			goto done;
1683 		}
1684 
1685 		if (!got_parse_sha1_digest(id.sha1, id_str))
1686 			continue;
1687 
1688 		/*
1689 		 * Directory entries do not necessarily appear in
1690 		 * sorted order, so we must iterate over all of them.
1691 		 */
1692 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1693 		if (cmp != 0) {
1694 			free(id_str);
1695 			continue;
1696 		}
1697 
1698 		if (*unique_id == NULL) {
1699 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1700 				int matched_type;
1701 				err = got_object_get_type(&matched_type, repo,
1702 				    &id);
1703 				if (err)
1704 					goto done;
1705 				if (matched_type != obj_type)
1706 					continue;
1707 			}
1708 			*unique_id = got_object_id_dup(&id);
1709 			if (*unique_id == NULL) {
1710 				err = got_error_from_errno("got_object_id_dup");
1711 				free(id_str);
1712 				goto done;
1713 			}
1714 		} else {
1715 			if (got_object_id_cmp(*unique_id, &id) == 0)
1716 				continue; /* both packed and loose */
1717 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1718 			free(id_str);
1719 			goto done;
1720 		}
1721 	}
1722 done:
1723 	if (dir && closedir(dir) != 0 && err == NULL)
1724 		err = got_error_from_errno("closedir");
1725 	if (err) {
1726 		free(*unique_id);
1727 		*unique_id = NULL;
1728 	}
1729 	free(path);
1730 	return err;
1731 }
1732 
1733 const struct got_error *
1734 got_repo_match_object_id_prefix(struct got_object_id **id,
1735     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1736 {
1737 	const struct got_error *err = NULL;
1738 	char *path_objects = got_repo_get_path_objects(repo);
1739 	char *object_dir = NULL;
1740 	size_t len;
1741 	int i;
1742 
1743 	*id = NULL;
1744 
1745 	len = strlen(id_str_prefix);
1746 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1747 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1748 
1749 	for (i = 0; i < len; i++) {
1750 		if (isxdigit((unsigned char)id_str_prefix[i]))
1751 			continue;
1752 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1753 	}
1754 
1755 	if (len >= 2) {
1756 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1757 		if (err)
1758 			goto done;
1759 		object_dir = strndup(id_str_prefix, 2);
1760 		if (object_dir == NULL) {
1761 			err = got_error_from_errno("strdup");
1762 			goto done;
1763 		}
1764 		err = match_loose_object(id, path_objects, object_dir,
1765 		    id_str_prefix, obj_type, repo);
1766 	} else if (len == 1) {
1767 		int i;
1768 		for (i = 0; i < 0xf; i++) {
1769 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1770 			    == -1) {
1771 				err = got_error_from_errno("asprintf");
1772 				goto done;
1773 			}
1774 			err = match_packed_object(id, repo, object_dir,
1775 			    obj_type);
1776 			if (err)
1777 				goto done;
1778 			err = match_loose_object(id, path_objects, object_dir,
1779 			    id_str_prefix, obj_type, repo);
1780 			if (err)
1781 				goto done;
1782 		}
1783 	} else {
1784 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1785 		goto done;
1786 	}
1787 done:
1788 	free(object_dir);
1789 	if (err) {
1790 		free(*id);
1791 		*id = NULL;
1792 	} else if (*id == NULL) {
1793 		switch (obj_type) {
1794 		case GOT_OBJ_TYPE_BLOB:
1795 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1796 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1797 			break;
1798 		case GOT_OBJ_TYPE_TREE:
1799 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1800 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1801 			break;
1802 		case GOT_OBJ_TYPE_COMMIT:
1803 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1804 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1805 			break;
1806 		case GOT_OBJ_TYPE_TAG:
1807 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1808 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1809 			break;
1810 		default:
1811 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1812 			break;
1813 		}
1814 	}
1815 
1816 	return err;
1817 }
1818 
1819 const struct got_error *
1820 got_repo_match_object_id(struct got_object_id **id, char **label,
1821     const char *id_str, int obj_type, struct got_reflist_head *refs,
1822     struct got_repository *repo)
1823 {
1824 	const struct got_error *err;
1825 	struct got_tag_object *tag;
1826 	struct got_reference *ref = NULL;
1827 
1828 	*id = NULL;
1829 	if (label)
1830 		*label = NULL;
1831 
1832 	if (refs) {
1833 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1834 		    refs, repo);
1835 		if (err == NULL) {
1836 			*id = got_object_id_dup(
1837 			    got_object_tag_get_object_id(tag));
1838 			if (*id == NULL)
1839 				err = got_error_from_errno("got_object_id_dup");
1840 			else if (label && asprintf(label, "refs/tags/%s",
1841 			    got_object_tag_get_name(tag)) == -1) {
1842 				err = got_error_from_errno("asprintf");
1843 				free(*id);
1844 				*id = NULL;
1845 			}
1846 			got_object_tag_close(tag);
1847 			return err;
1848 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1849 		    err->code != GOT_ERR_NO_OBJ)
1850 			return err;
1851 	}
1852 
1853 	err = got_ref_open(&ref, repo, id_str, 0);
1854 	if (err == NULL) {
1855 		err = got_ref_resolve(id, repo, ref);
1856 		if (err)
1857 			goto done;
1858 		if (label) {
1859 			*label = strdup(got_ref_get_name(ref));
1860 			if (*label == NULL) {
1861 				err = got_error_from_errno("strdup");
1862 				goto done;
1863 			}
1864 		}
1865 	} else {
1866 		if (err->code != GOT_ERR_NOT_REF &&
1867 		    err->code != GOT_ERR_BAD_REF_NAME)
1868 			goto done;
1869 		err = got_repo_match_object_id_prefix(id, id_str,
1870 		    obj_type, repo);
1871 		if (err) {
1872 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1873 				err = got_error_not_ref(id_str);
1874 			goto done;
1875 		}
1876 		if (label) {
1877 			err = got_object_id_str(label, *id);
1878 			if (*label == NULL) {
1879 				err = got_error_from_errno("strdup");
1880 				goto done;
1881 			}
1882 		}
1883 	}
1884 done:
1885 	if (ref)
1886 		got_ref_close(ref);
1887 	return err;
1888 }
1889 
1890 const struct got_error *
1891 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1892     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1893 {
1894 	const struct got_error *err = NULL;
1895 	struct got_reflist_entry *re;
1896 	struct got_object_id *tag_id;
1897 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1898 
1899 	*tag = NULL;
1900 
1901 	TAILQ_FOREACH(re, refs, entry) {
1902 		const char *refname;
1903 		refname = got_ref_get_name(re->ref);
1904 		if (got_ref_is_symbolic(re->ref))
1905 			continue;
1906 		if (strncmp(refname, "refs/tags/", 10) != 0)
1907 			continue;
1908 		if (!name_is_absolute)
1909 			refname += strlen("refs/tags/");
1910 		if (strcmp(refname, name) != 0)
1911 			continue;
1912 		err = got_ref_resolve(&tag_id, repo, re->ref);
1913 		if (err)
1914 			break;
1915 		err = got_object_open_as_tag(tag, repo, tag_id);
1916 		free(tag_id);
1917 		if (err)
1918 			break;
1919 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1920 		    got_object_tag_get_object_type(*tag) == obj_type)
1921 			break;
1922 		got_object_tag_close(*tag);
1923 		*tag = NULL;
1924 	}
1925 
1926 	if (err == NULL && *tag == NULL)
1927 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1928 		    GOT_OBJ_LABEL_TAG, name);
1929 	return err;
1930 }
1931 
1932 static const struct got_error *
1933 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1934     const char *name, mode_t mode, struct got_object_id *blob_id)
1935 {
1936 	const struct got_error *err = NULL;
1937 
1938 	 *new_te = NULL;
1939 
1940 	*new_te = calloc(1, sizeof(**new_te));
1941 	if (*new_te == NULL)
1942 		return got_error_from_errno("calloc");
1943 
1944 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1945 	    sizeof((*new_te)->name)) {
1946 		err = got_error(GOT_ERR_NO_SPACE);
1947 		goto done;
1948 	}
1949 
1950 	if (S_ISLNK(mode)) {
1951 		(*new_te)->mode = S_IFLNK;
1952 	} else {
1953 		(*new_te)->mode = S_IFREG;
1954 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1955 	}
1956 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1957 done:
1958 	if (err && *new_te) {
1959 		free(*new_te);
1960 		*new_te = NULL;
1961 	}
1962 	return err;
1963 }
1964 
1965 static const struct got_error *
1966 import_file(struct got_tree_entry **new_te, struct dirent *de,
1967     const char *path, struct got_repository *repo)
1968 {
1969 	const struct got_error *err;
1970 	struct got_object_id *blob_id = NULL;
1971 	char *filepath;
1972 	struct stat sb;
1973 
1974 	if (asprintf(&filepath, "%s%s%s", path,
1975 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1976 		return got_error_from_errno("asprintf");
1977 
1978 	if (lstat(filepath, &sb) != 0) {
1979 		err = got_error_from_errno2("lstat", path);
1980 		goto done;
1981 	}
1982 
1983 	err = got_object_blob_create(&blob_id, filepath, repo);
1984 	if (err)
1985 		goto done;
1986 
1987 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1988 	    blob_id);
1989 done:
1990 	free(filepath);
1991 	if (err)
1992 		free(blob_id);
1993 	return err;
1994 }
1995 
1996 static const struct got_error *
1997 insert_tree_entry(struct got_tree_entry *new_te,
1998     struct got_pathlist_head *paths)
1999 {
2000 	const struct got_error *err = NULL;
2001 	struct got_pathlist_entry *new_pe;
2002 
2003 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2004 	if (err)
2005 		return err;
2006 	if (new_pe == NULL)
2007 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2008 	return NULL;
2009 }
2010 
2011 static const struct got_error *write_tree(struct got_object_id **,
2012     const char *, struct got_pathlist_head *, struct got_repository *,
2013     got_repo_import_cb progress_cb, void *progress_arg);
2014 
2015 static const struct got_error *
2016 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2017     const char *path, struct got_pathlist_head *ignores,
2018     struct got_repository *repo,
2019     got_repo_import_cb progress_cb, void *progress_arg)
2020 {
2021 	const struct got_error *err;
2022 	struct got_object_id *id = NULL;
2023 	char *subdirpath;
2024 
2025 	if (asprintf(&subdirpath, "%s%s%s", path,
2026 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2027 		return got_error_from_errno("asprintf");
2028 
2029 	(*new_te) = calloc(1, sizeof(**new_te));
2030 	if (*new_te == NULL)
2031 		return got_error_from_errno("calloc");
2032 	(*new_te)->mode = S_IFDIR;
2033 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2034 	    sizeof((*new_te)->name)) {
2035 		err = got_error(GOT_ERR_NO_SPACE);
2036 		goto done;
2037 	}
2038 	err = write_tree(&id, subdirpath, ignores,  repo,
2039 	    progress_cb, progress_arg);
2040 	if (err)
2041 		goto done;
2042 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2043 	
2044 done:
2045 	free(id);
2046 	free(subdirpath);
2047 	if (err) {
2048 		free(*new_te);
2049 		*new_te = NULL;
2050 	}
2051 	return err;
2052 }
2053 
2054 static const struct got_error *
2055 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2056     struct got_pathlist_head *ignores, struct got_repository *repo,
2057     got_repo_import_cb progress_cb, void *progress_arg)
2058 {
2059 	const struct got_error *err = NULL;
2060 	DIR *dir;
2061 	struct dirent *de;
2062 	int nentries;
2063 	struct got_tree_entry *new_te = NULL;
2064 	struct got_pathlist_head paths;
2065 	struct got_pathlist_entry *pe;
2066 
2067 	*new_tree_id = NULL;
2068 
2069 	TAILQ_INIT(&paths);
2070 
2071 	dir = opendir(path_dir);
2072 	if (dir == NULL) {
2073 		err = got_error_from_errno2("opendir", path_dir);
2074 		goto done;
2075 	}
2076 
2077 	nentries = 0;
2078 	while ((de = readdir(dir)) != NULL) {
2079 		int ignore = 0;
2080 		int type;
2081 
2082 		if (strcmp(de->d_name, ".") == 0 ||
2083 		    strcmp(de->d_name, "..") == 0)
2084 			continue;
2085 
2086 		TAILQ_FOREACH(pe, ignores, entry) {
2087 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2088 				ignore = 1;
2089 				break;
2090 			}
2091 		}
2092 		if (ignore)
2093 			continue;
2094 
2095 		err = got_path_dirent_type(&type, path_dir, de);
2096 		if (err)
2097 			goto done;
2098 
2099 		if (type == DT_DIR) {
2100 			err = import_subdir(&new_te, de, path_dir,
2101 			    ignores, repo, progress_cb, progress_arg);
2102 			if (err) {
2103 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2104 					goto done;
2105 				err = NULL;
2106 				continue;
2107 			}
2108 		} else if (type == DT_REG || type == DT_LNK) {
2109 			err = import_file(&new_te, de, path_dir, repo);
2110 			if (err)
2111 				goto done;
2112 		} else
2113 			continue;
2114 
2115 		err = insert_tree_entry(new_te, &paths);
2116 		if (err)
2117 			goto done;
2118 		nentries++;
2119 	}
2120 
2121 	if (TAILQ_EMPTY(&paths)) {
2122 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2123 		    "cannot create tree without any entries");
2124 		goto done;
2125 	}
2126 
2127 	TAILQ_FOREACH(pe, &paths, entry) {
2128 		struct got_tree_entry *te = pe->data;
2129 		char *path;
2130 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2131 			continue;
2132 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2133 			err = got_error_from_errno("asprintf");
2134 			goto done;
2135 		}
2136 		err = (*progress_cb)(progress_arg, path);
2137 		free(path);
2138 		if (err)
2139 			goto done;
2140 	}
2141 
2142 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2143 done:
2144 	if (dir)
2145 		closedir(dir);
2146 	got_pathlist_free(&paths);
2147 	return err;
2148 }
2149 
2150 const struct got_error *
2151 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2152     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2153     struct got_repository *repo, got_repo_import_cb progress_cb,
2154     void *progress_arg)
2155 {
2156 	const struct got_error *err;
2157 	struct got_object_id *new_tree_id;
2158 
2159 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2160 	    progress_cb, progress_arg);
2161 	if (err)
2162 		return err;
2163 
2164 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2165 	    author, time(NULL), author, time(NULL), logmsg, repo);
2166 	free(new_tree_id);
2167 	return err;
2168 }
2169 
2170 const struct got_error *
2171 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2172     struct got_repository *repo)
2173 {
2174 	const struct got_error *err = NULL;
2175 	char *path_objects = NULL, *path = NULL;
2176 	DIR *dir = NULL;
2177 	struct got_object_id id;
2178 	int i;
2179 
2180 	*nobjects = 0;
2181 	*ondisk_size = 0;
2182 
2183 	path_objects = got_repo_get_path_objects(repo);
2184 	if (path_objects == NULL)
2185 		return got_error_from_errno("got_repo_get_path_objects");
2186 
2187 	for (i = 0; i <= 0xff; i++) {
2188 		struct dirent *dent;
2189 
2190 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2191 			err = got_error_from_errno("asprintf");
2192 			break;
2193 		}
2194 
2195 		dir = opendir(path);
2196 		if (dir == NULL) {
2197 			if (errno == ENOENT) {
2198 				err = NULL;
2199 				continue;
2200 			}
2201 			err = got_error_from_errno2("opendir", path);
2202 			break;
2203 		}
2204 
2205 		while ((dent = readdir(dir)) != NULL) {
2206 			char *id_str;
2207 			int fd;
2208 			struct stat sb;
2209 
2210 			if (strcmp(dent->d_name, ".") == 0 ||
2211 			    strcmp(dent->d_name, "..") == 0)
2212 				continue;
2213 
2214 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2215 				err = got_error_from_errno("asprintf");
2216 				goto done;
2217 			}
2218 
2219 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2220 				free(id_str);
2221 				continue;
2222 			}
2223 			free(id_str);
2224 
2225 			err = got_object_open_loose_fd(&fd, &id, repo);
2226 			if (err)
2227 				goto done;
2228 
2229 			if (fstat(fd, &sb) == -1) {
2230 				err = got_error_from_errno("fstat");
2231 				close(fd);
2232 				goto done;
2233 			}
2234 			(*nobjects)++;
2235 			(*ondisk_size) += sb.st_size;
2236 
2237 			if (close(fd) == -1) {
2238 				err = got_error_from_errno("close");
2239 				goto done;
2240 			}
2241 		}
2242 
2243 		if (closedir(dir) != 0) {
2244 			err = got_error_from_errno("closedir");
2245 			goto done;
2246 		}
2247 		dir = NULL;
2248 
2249 		free(path);
2250 		path = NULL;
2251 	}
2252 done:
2253 	if (dir && closedir(dir) != 0 && err == NULL)
2254 		err = got_error_from_errno("closedir");
2255 
2256 	if (err) {
2257 		*nobjects = 0;
2258 		*ondisk_size = 0;
2259 	}
2260 	free(path_objects);
2261 	free(path);
2262 	return err;
2263 }
2264 
2265 const struct got_error *
2266 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2267     off_t *total_packsize, struct got_repository *repo)
2268 {
2269 	const struct got_error *err = NULL;
2270 	DIR *packdir = NULL;
2271 	struct dirent *dent;
2272 	struct got_packidx *packidx = NULL;
2273 	char *path_packidx;
2274 	char *path_packfile;
2275 	int packdir_fd;
2276 	struct stat sb;
2277 
2278 	*npackfiles = 0;
2279 	*nobjects = 0;
2280 	*total_packsize = 0;
2281 
2282 	packdir_fd = openat(got_repo_get_fd(repo),
2283 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2284 	if (packdir_fd == -1) {
2285 		return got_error_from_errno_fmt("openat: %s/%s",
2286 		    got_repo_get_path_git_dir(repo),
2287 		    GOT_OBJECTS_PACK_DIR);
2288 	}
2289 
2290 	packdir = fdopendir(packdir_fd);
2291 	if (packdir == NULL) {
2292 		err = got_error_from_errno("fdopendir");
2293 		goto done;
2294 	}
2295 
2296 	while ((dent = readdir(packdir)) != NULL) {
2297 		if (!got_repo_is_packidx_filename(dent->d_name,
2298 		    strlen(dent->d_name)))
2299 			continue;
2300 
2301 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2302 		    dent->d_name) == -1) {
2303 			err = got_error_from_errno("asprintf");
2304 			goto done;
2305 		}
2306 
2307 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2308 		    path_packidx, 0);
2309 		free(path_packidx);
2310 		if (err)
2311 			goto done;
2312 
2313 		if (fstat(packidx->fd, &sb) == -1)
2314 			goto done;
2315 		*total_packsize += sb.st_size;
2316 
2317 		err = got_packidx_get_packfile_path(&path_packfile,
2318 		    packidx->path_packidx);
2319 		if (err)
2320 			goto done;
2321 
2322 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2323 		    0) == -1) {
2324 			free(path_packfile);
2325 			goto done;
2326 		}
2327 		free(path_packfile);
2328 		*total_packsize += sb.st_size;
2329 
2330 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2331 
2332 		(*npackfiles)++;
2333 
2334 		got_packidx_close(packidx);
2335 		packidx = NULL;
2336 	}
2337 done:
2338 	if (packidx)
2339 		got_packidx_close(packidx);
2340 	if (packdir && closedir(packdir) != 0 && err == NULL)
2341 		err = got_error_from_errno("closedir");
2342 	if (err) {
2343 		*npackfiles = 0;
2344 		*nobjects = 0;
2345 		*total_packsize = 0;
2346 	}
2347 	return err;
2348 }
2349 
2350 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2351     got_packidx_bloom_filter_cmp);