1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
72 
73 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
74     got_packidx_bloom_filter_cmp);
75 
76 const char *
77 got_repo_get_path(struct got_repository *repo)
78 {
79 	return repo->path;
80 }
81 
82 const char *
83 got_repo_get_path_git_dir(struct got_repository *repo)
84 {
85 	return repo->path_git_dir;
86 }
87 
88 int
89 got_repo_get_fd(struct got_repository *repo)
90 {
91 	return repo->gitdir_fd;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_name(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_name;
98 }
99 
100 const char *
101 got_repo_get_gitconfig_author_email(struct got_repository *repo)
102 {
103 	return repo->gitconfig_author_email;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_name;
110 }
111 
112 const char *
113 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
114 {
115 	return repo->global_gitconfig_author_email;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_owner(struct got_repository *repo)
120 {
121 	return repo->gitconfig_owner;
122 }
123 
124 void
125 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
126     struct got_repository *repo)
127 {
128 	*extensions = repo->extensions;
129 	*nextensions = repo->nextensions;
130 }
131 
132 int
133 got_repo_is_bare(struct got_repository *repo)
134 {
135 	return (strcmp(repo->path, repo->path_git_dir) == 0);
136 }
137 
138 static char *
139 get_path_git_child(struct got_repository *repo, const char *basename)
140 {
141 	char *path_child;
142 
143 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
144 	    basename) == -1)
145 		return NULL;
146 
147 	return path_child;
148 }
149 
150 char *
151 got_repo_get_path_objects(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
154 }
155 
156 char *
157 got_repo_get_path_objects_pack(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
160 }
161 
162 char *
163 got_repo_get_path_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_REFS_DIR);
166 }
167 
168 char *
169 got_repo_get_path_packed_refs(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
172 }
173 
174 static char *
175 get_path_head(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_HEAD_FILE);
178 }
179 
180 char *
181 got_repo_get_path_gitconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GITCONFIG);
184 }
185 
186 char *
187 got_repo_get_path_gotconfig(struct got_repository *repo)
188 {
189 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
190 }
191 
192 const struct got_gotconfig *
193 got_repo_get_gotconfig(struct got_repository *repo)
194 {
195 	return repo->gotconfig;
196 }
197 
198 void
199 got_repo_get_gitconfig_remotes(int *nremotes,
200     const struct got_remote_repo **remotes, struct got_repository *repo)
201 {
202 	*nremotes = repo->ngitconfig_remotes;
203 	*remotes = repo->gitconfig_remotes;
204 }
205 
206 static int
207 is_git_repo(struct got_repository *repo)
208 {
209 	const char *path_git = got_repo_get_path_git_dir(repo);
210 	char *path_objects = got_repo_get_path_objects(repo);
211 	char *path_refs = got_repo_get_path_refs(repo);
212 	char *path_head = get_path_head(repo);
213 	int ret = 0;
214 	struct stat sb;
215 	struct got_reference *head_ref;
216 
217 	if (lstat(path_git, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_objects, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_refs, &sb) == -1)
228 		goto done;
229 	if (!S_ISDIR(sb.st_mode))
230 		goto done;
231 
232 	if (lstat(path_head, &sb) == -1)
233 		goto done;
234 	if (!S_ISREG(sb.st_mode))
235 		goto done;
236 
237 	/* Check if the HEAD reference can be opened. */
238 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
239 		goto done;
240 	got_ref_close(head_ref);
241 
242 	ret = 1;
243 done:
244 	free(path_objects);
245 	free(path_refs);
246 	free(path_head);
247 	return ret;
248 
249 }
250 
251 const struct got_error *
252 got_repo_pack_fds_open(int **pack_fds)
253 {
254 	const struct got_error *err = NULL;
255 	int i, *pack_fds_tmp;
256 
257 	pack_fds_tmp = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(int));
258 	if (pack_fds_tmp == NULL)
259 		return got_error_from_errno("calloc");
260 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
261 	if (*pack_fds == NULL) {
262 		free(pack_fds_tmp);
263 		return got_error_from_errno("calloc");
264 	}
265 
266 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
267 		pack_fds_tmp[i] = got_opentempfd();
268 		if (pack_fds_tmp[i] == -1) {
269 			err = got_error_from_errno("got_opentempfd");
270 			got_repo_pack_fds_close(pack_fds_tmp);
271 			return err;
272 		}
273 	}
274 	memcpy(*pack_fds, pack_fds_tmp, GOT_PACK_NUM_TEMPFILES * sizeof(int));
275 	return err;
276 }
277 
278 const struct got_error *
279 got_repo_pack_fds_close(int *pack_fds)
280 {
281 	const struct got_error *err = NULL;
282 	int i;
283 
284 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
285 		if (pack_fds[i] == -1)
286 			continue;
287 		if (close(pack_fds[i]) == -1) {
288 			err = got_error_from_errno("close");
289 			break;
290 		}
291 	}
292 	free(pack_fds);
293 	return err;
294 }
295 
296 const struct got_error *
297 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
298     struct got_object *obj)
299 {
300 #ifndef GOT_NO_OBJ_CACHE
301 	const struct got_error *err = NULL;
302 	err = got_object_cache_add(&repo->objcache, id, obj);
303 	if (err) {
304 		if (err->code == GOT_ERR_OBJ_EXISTS ||
305 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
306 			err = NULL;
307 		return err;
308 	}
309 	obj->refcnt++;
310 #endif
311 	return NULL;
312 }
313 
314 struct got_object *
315 got_repo_get_cached_object(struct got_repository *repo,
316     struct got_object_id *id)
317 {
318 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
319 }
320 
321 const struct got_error *
322 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
323     struct got_tree_object *tree)
324 {
325 #ifndef GOT_NO_OBJ_CACHE
326 	const struct got_error *err = NULL;
327 	err = got_object_cache_add(&repo->treecache, id, tree);
328 	if (err) {
329 		if (err->code == GOT_ERR_OBJ_EXISTS ||
330 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
331 			err = NULL;
332 		return err;
333 	}
334 	tree->refcnt++;
335 #endif
336 	return NULL;
337 }
338 
339 struct got_tree_object *
340 got_repo_get_cached_tree(struct got_repository *repo,
341     struct got_object_id *id)
342 {
343 	return (struct got_tree_object *)got_object_cache_get(
344 	    &repo->treecache, id);
345 }
346 
347 const struct got_error *
348 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
349     struct got_commit_object *commit)
350 {
351 #ifndef GOT_NO_OBJ_CACHE
352 	const struct got_error *err = NULL;
353 	err = got_object_cache_add(&repo->commitcache, id, commit);
354 	if (err) {
355 		if (err->code == GOT_ERR_OBJ_EXISTS ||
356 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
357 			err = NULL;
358 		return err;
359 	}
360 	commit->refcnt++;
361 #endif
362 	return NULL;
363 }
364 
365 struct got_commit_object *
366 got_repo_get_cached_commit(struct got_repository *repo,
367     struct got_object_id *id)
368 {
369 	return (struct got_commit_object *)got_object_cache_get(
370 	    &repo->commitcache, id);
371 }
372 
373 const struct got_error *
374 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
375     struct got_tag_object *tag)
376 {
377 #ifndef GOT_NO_OBJ_CACHE
378 	const struct got_error *err = NULL;
379 	err = got_object_cache_add(&repo->tagcache, id, tag);
380 	if (err) {
381 		if (err->code == GOT_ERR_OBJ_EXISTS ||
382 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
383 			err = NULL;
384 		return err;
385 	}
386 	tag->refcnt++;
387 #endif
388 	return NULL;
389 }
390 
391 struct got_tag_object *
392 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
393 {
394 	return (struct got_tag_object *)got_object_cache_get(
395 	    &repo->tagcache, id);
396 }
397 
398 const struct got_error *
399 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
400     struct got_raw_object *raw)
401 {
402 #ifndef GOT_NO_OBJ_CACHE
403 	const struct got_error *err = NULL;
404 	err = got_object_cache_add(&repo->rawcache, id, raw);
405 	if (err) {
406 		if (err->code == GOT_ERR_OBJ_EXISTS ||
407 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
408 			err = NULL;
409 		return err;
410 	}
411 	raw->refcnt++;
412 #endif
413 	return NULL;
414 }
415 
416 
417 struct got_raw_object *
418 got_repo_get_cached_raw_object(struct got_repository *repo,
419     struct got_object_id *id)
420 {
421 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
422 }
423 
424 
425 static const struct got_error *
426 open_repo(struct got_repository *repo, const char *path)
427 {
428 	const struct got_error *err = NULL;
429 
430 	repo->gitdir_fd = -1;
431 
432 	/* bare git repository? */
433 	repo->path_git_dir = strdup(path);
434 	if (repo->path_git_dir == NULL)
435 		return got_error_from_errno("strdup");
436 	if (is_git_repo(repo)) {
437 		repo->path = strdup(repo->path_git_dir);
438 		if (repo->path == NULL) {
439 			err = got_error_from_errno("strdup");
440 			goto done;
441 		}
442 		repo->gitdir_fd = open(repo->path_git_dir,
443 		    O_DIRECTORY | O_CLOEXEC);
444 		if (repo->gitdir_fd == -1) {
445 			err = got_error_from_errno2("open",
446 			    repo->path_git_dir);
447 			goto done;
448 		}
449 		return NULL;
450 	}
451 
452 	/* git repository with working tree? */
453 	free(repo->path_git_dir);
454 	repo->path_git_dir = NULL;
455 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
456 		err = got_error_from_errno("asprintf");
457 		goto done;
458 	}
459 	if (is_git_repo(repo)) {
460 		repo->path = strdup(path);
461 		if (repo->path == NULL) {
462 			err = got_error_from_errno("strdup");
463 			goto done;
464 		}
465 		repo->gitdir_fd = open(repo->path_git_dir,
466 		    O_DIRECTORY | O_CLOEXEC);
467 		if (repo->gitdir_fd == -1) {
468 			err = got_error_from_errno2("open",
469 			    repo->path_git_dir);
470 			goto done;
471 		}
472 		return NULL;
473 	}
474 
475 	err = got_error(GOT_ERR_NOT_GIT_REPO);
476 done:
477 	if (err) {
478 		free(repo->path);
479 		repo->path = NULL;
480 		free(repo->path_git_dir);
481 		repo->path_git_dir = NULL;
482 		if (repo->gitdir_fd != -1)
483 			close(repo->gitdir_fd);
484 		repo->gitdir_fd = -1;
485 
486 	}
487 	return err;
488 }
489 
490 static const struct got_error *
491 parse_gitconfig_file(int *gitconfig_repository_format_version,
492     char **gitconfig_author_name, char **gitconfig_author_email,
493     struct got_remote_repo **remotes, int *nremotes,
494     char **gitconfig_owner, char ***extensions, int *nextensions,
495     const char *gitconfig_path)
496 {
497 	const struct got_error *err = NULL, *child_err = NULL;
498 	int fd = -1;
499 	int imsg_fds[2] = { -1, -1 };
500 	pid_t pid;
501 	struct imsgbuf *ibuf;
502 
503 	*gitconfig_repository_format_version = 0;
504 	if (extensions)
505 		*extensions = NULL;
506 	if (nextensions)
507 		*nextensions = 0;
508 	*gitconfig_author_name = NULL;
509 	*gitconfig_author_email = NULL;
510 	if (remotes)
511 		*remotes = NULL;
512 	if (nremotes)
513 		*nremotes = 0;
514 	if (gitconfig_owner)
515 		*gitconfig_owner = NULL;
516 
517 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
518 	if (fd == -1) {
519 		if (errno == ENOENT)
520 			return NULL;
521 		return got_error_from_errno2("open", gitconfig_path);
522 	}
523 
524 	ibuf = calloc(1, sizeof(*ibuf));
525 	if (ibuf == NULL) {
526 		err = got_error_from_errno("calloc");
527 		goto done;
528 	}
529 
530 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
531 		err = got_error_from_errno("socketpair");
532 		goto done;
533 	}
534 
535 	pid = fork();
536 	if (pid == -1) {
537 		err = got_error_from_errno("fork");
538 		goto done;
539 	} else if (pid == 0) {
540 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
541 		    gitconfig_path);
542 		/* not reached */
543 	}
544 
545 	if (close(imsg_fds[1]) == -1) {
546 		err = got_error_from_errno("close");
547 		goto done;
548 	}
549 	imsg_fds[1] = -1;
550 	imsg_init(ibuf, imsg_fds[0]);
551 
552 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
553 	if (err)
554 		goto done;
555 	fd = -1;
556 
557 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
558 	if (err)
559 		goto done;
560 
561 	err = got_privsep_recv_gitconfig_int(
562 	    gitconfig_repository_format_version, ibuf);
563 	if (err)
564 		goto done;
565 
566 	if (extensions && nextensions) {
567 		err = got_privsep_send_gitconfig_repository_extensions_req(
568 		    ibuf);
569 		if (err)
570 			goto done;
571 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
572 		if (err)
573 			goto done;
574 		if (*nextensions > 0) {
575 			int i;
576 			*extensions = calloc(*nextensions, sizeof(char *));
577 			if (*extensions == NULL) {
578 				err = got_error_from_errno("calloc");
579 				goto done;
580 			}
581 			for (i = 0; i < *nextensions; i++) {
582 				char *ext;
583 				err = got_privsep_recv_gitconfig_str(&ext,
584 				    ibuf);
585 				if (err)
586 					goto done;
587 				(*extensions)[i] = ext;
588 			}
589 		}
590 	}
591 
592 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
593 	if (err)
594 		goto done;
595 
596 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
597 	if (err)
598 		goto done;
599 
600 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
601 	if (err)
602 		goto done;
603 
604 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
605 	if (err)
606 		goto done;
607 
608 	if (remotes && nremotes) {
609 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
610 		if (err)
611 			goto done;
612 
613 		err = got_privsep_recv_gitconfig_remotes(remotes,
614 		    nremotes, ibuf);
615 		if (err)
616 			goto done;
617 	}
618 
619 	if (gitconfig_owner) {
620 		err = got_privsep_send_gitconfig_owner_req(ibuf);
621 		if (err)
622 			goto done;
623 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
624 		if (err)
625 			goto done;
626 	}
627 
628 	err = got_privsep_send_stop(imsg_fds[0]);
629 	child_err = got_privsep_wait_for_child(pid);
630 	if (child_err && err == NULL)
631 		err = child_err;
632 done:
633 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
634 		err = got_error_from_errno("close");
635 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
636 		err = got_error_from_errno("close");
637 	if (fd != -1 && close(fd) == -1 && err == NULL)
638 		err = got_error_from_errno2("close", gitconfig_path);
639 	free(ibuf);
640 	return err;
641 }
642 
643 static const struct got_error *
644 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
645 {
646 	const struct got_error *err = NULL;
647 	char *repo_gitconfig_path = NULL;
648 
649 	if (global_gitconfig_path) {
650 		/* Read settings from ~/.gitconfig. */
651 		int dummy_repo_version;
652 		err = parse_gitconfig_file(&dummy_repo_version,
653 		    &repo->global_gitconfig_author_name,
654 		    &repo->global_gitconfig_author_email,
655 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
656 		if (err)
657 			return err;
658 	}
659 
660 	/* Read repository's .git/config file. */
661 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
662 	if (repo_gitconfig_path == NULL)
663 		return got_error_from_errno("got_repo_get_path_gitconfig");
664 
665 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
666 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
667 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
668 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
669 	    repo_gitconfig_path);
670 	if (err)
671 		goto done;
672 done:
673 	free(repo_gitconfig_path);
674 	return err;
675 }
676 
677 static const struct got_error *
678 read_gotconfig(struct got_repository *repo)
679 {
680 	const struct got_error *err = NULL;
681 	char *gotconfig_path;
682 
683 	gotconfig_path = got_repo_get_path_gotconfig(repo);
684 	if (gotconfig_path == NULL)
685 		return got_error_from_errno("got_repo_get_path_gotconfig");
686 
687 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
688 	free(gotconfig_path);
689 	return err;
690 }
691 
692 /* Supported repository format extensions. */
693 static const char *const repo_extensions[] = {
694 	"noop",			/* Got supports repository format version 1. */
695 	"preciousObjects",	/* Supported by gotadmin cleanup. */
696 	"worktreeConfig",	/* Got does not care about Git work trees. */
697 };
698 
699 const struct got_error *
700 got_repo_open(struct got_repository **repop, const char *path,
701     const char *global_gitconfig_path, int *pack_fds)
702 {
703 	struct got_repository *repo = NULL;
704 	const struct got_error *err = NULL;
705 	char *repo_path = NULL;
706 	size_t i, j = 0;
707 	struct rlimit rl;
708 
709 	*repop = NULL;
710 
711 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
712 		return got_error_from_errno("getrlimit");
713 
714 	repo = calloc(1, sizeof(*repo));
715 	if (repo == NULL)
716 		return got_error_from_errno("calloc");
717 
718 	RB_INIT(&repo->packidx_bloom_filters);
719 	TAILQ_INIT(&repo->packidx_paths);
720 
721 	for (i = 0; i < nitems(repo->privsep_children); i++) {
722 		memset(&repo->privsep_children[i], 0,
723 		    sizeof(repo->privsep_children[0]));
724 		repo->privsep_children[i].imsg_fd = -1;
725 	}
726 
727 	err = got_object_cache_init(&repo->objcache,
728 	    GOT_OBJECT_CACHE_TYPE_OBJ);
729 	if (err)
730 		goto done;
731 	err = got_object_cache_init(&repo->treecache,
732 	    GOT_OBJECT_CACHE_TYPE_TREE);
733 	if (err)
734 		goto done;
735 	err = got_object_cache_init(&repo->commitcache,
736 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
737 	if (err)
738 		goto done;
739 	err = got_object_cache_init(&repo->tagcache,
740 	    GOT_OBJECT_CACHE_TYPE_TAG);
741 	if (err)
742 		goto done;
743 	err = got_object_cache_init(&repo->rawcache,
744 	    GOT_OBJECT_CACHE_TYPE_RAW);
745 	if (err)
746 		goto done;
747 
748 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
749 	if (repo->pack_cache_size > rl.rlim_cur / 8)
750 		repo->pack_cache_size = rl.rlim_cur / 8;
751 	for (i = 0; i < nitems(repo->packs); i++) {
752 		if (i < repo->pack_cache_size) {
753 			repo->packs[i].basefd = pack_fds[j++];
754 			repo->packs[i].accumfd = pack_fds[j++];
755 		} else {
756 			repo->packs[i].basefd = -1;
757 			repo->packs[i].accumfd = -1;
758 		}
759 	}
760 	repo->pinned_pack = -1;
761 	repo->pinned_packidx = -1;
762 	repo->pinned_pid = 0;
763 
764 	repo_path = realpath(path, NULL);
765 	if (repo_path == NULL) {
766 		err = got_error_from_errno2("realpath", path);
767 		goto done;
768 	}
769 
770 	for (;;) {
771 		char *parent_path;
772 
773 		err = open_repo(repo, repo_path);
774 		if (err == NULL)
775 			break;
776 		if (err->code != GOT_ERR_NOT_GIT_REPO)
777 			goto done;
778 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
779 			err = got_error(GOT_ERR_NOT_GIT_REPO);
780 			goto done;
781 		}
782 		err = got_path_dirname(&parent_path, repo_path);
783 		if (err)
784 			goto done;
785 		free(repo_path);
786 		repo_path = parent_path;
787 	}
788 
789 	err = read_gotconfig(repo);
790 	if (err)
791 		goto done;
792 
793 	err = read_gitconfig(repo, global_gitconfig_path);
794 	if (err)
795 		goto done;
796 	if (repo->gitconfig_repository_format_version != 0) {
797 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
798 		goto done;
799 	}
800 	for (i = 0; i < repo->nextensions; i++) {
801 		char *ext = repo->extensions[i];
802 		int j, supported = 0;
803 		for (j = 0; j < nitems(repo_extensions); j++) {
804 			if (strcmp(ext, repo_extensions[j]) == 0) {
805 				supported = 1;
806 				break;
807 			}
808 		}
809 		if (!supported) {
810 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
811 			goto done;
812 		}
813 	}
814 
815 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
816 done:
817 	if (err)
818 		got_repo_close(repo);
819 	else
820 		*repop = repo;
821 	free(repo_path);
822 	return err;
823 }
824 
825 const struct got_error *
826 got_repo_close(struct got_repository *repo)
827 {
828 	const struct got_error *err = NULL, *child_err;
829 	struct got_packidx_bloom_filter *bf;
830 	struct got_pathlist_entry *pe;
831 	size_t i;
832 
833 	for (i = 0; i < repo->pack_cache_size; i++) {
834 		if (repo->packidx_cache[i] == NULL)
835 			break;
836 		got_packidx_close(repo->packidx_cache[i]);
837 	}
838 
839 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
840 	    &repo->packidx_bloom_filters))) {
841 		RB_REMOVE(got_packidx_bloom_filter_tree,
842 		    &repo->packidx_bloom_filters, bf);
843 		free(bf->bloom);
844 		free(bf);
845 	}
846 
847 	for (i = 0; i < repo->pack_cache_size; i++)
848 		if (repo->packs[i].path_packfile)
849 			if (repo->packs[i].path_packfile)
850 				got_pack_close(&repo->packs[i]);
851 
852 	free(repo->path);
853 	free(repo->path_git_dir);
854 
855 	got_object_cache_close(&repo->objcache);
856 	got_object_cache_close(&repo->treecache);
857 	got_object_cache_close(&repo->commitcache);
858 	got_object_cache_close(&repo->tagcache);
859 	got_object_cache_close(&repo->rawcache);
860 
861 	for (i = 0; i < nitems(repo->privsep_children); i++) {
862 		if (repo->privsep_children[i].imsg_fd == -1)
863 			continue;
864 		imsg_clear(repo->privsep_children[i].ibuf);
865 		free(repo->privsep_children[i].ibuf);
866 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
867 		child_err = got_privsep_wait_for_child(
868 		    repo->privsep_children[i].pid);
869 		if (child_err && err == NULL)
870 			err = child_err;
871 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
872 		    err == NULL)
873 			err = got_error_from_errno("close");
874 	}
875 
876 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
877 	    err == NULL)
878 		err = got_error_from_errno("close");
879 
880 	if (repo->gotconfig)
881 		got_gotconfig_free(repo->gotconfig);
882 	free(repo->gitconfig_author_name);
883 	free(repo->gitconfig_author_email);
884 	for (i = 0; i < repo->ngitconfig_remotes; i++)
885 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
886 	free(repo->gitconfig_remotes);
887 	for (i = 0; i < repo->nextensions; i++)
888 		free(repo->extensions[i]);
889 	free(repo->extensions);
890 
891 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
892 		free((void *)pe->path);
893 	got_pathlist_free(&repo->packidx_paths);
894 	free(repo);
895 
896 	return err;
897 }
898 
899 void
900 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
901 {
902 	int i;
903 
904 	free(repo->name);
905 	repo->name = NULL;
906 	free(repo->fetch_url);
907 	repo->fetch_url = NULL;
908 	free(repo->send_url);
909 	repo->send_url = NULL;
910 	for (i = 0; i < repo->nfetch_branches; i++)
911 		free(repo->fetch_branches[i]);
912 	free(repo->fetch_branches);
913 	repo->fetch_branches = NULL;
914 	repo->nfetch_branches = 0;
915 	for (i = 0; i < repo->nsend_branches; i++)
916 		free(repo->send_branches[i]);
917 	free(repo->send_branches);
918 	repo->send_branches = NULL;
919 	repo->nsend_branches = 0;
920 }
921 
922 const struct got_error *
923 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
924     const char *input_path)
925 {
926 	const struct got_error *err = NULL;
927 	const char *repo_abspath = NULL;
928 	size_t repolen, len;
929 	char *canonpath, *path = NULL;
930 
931 	*in_repo_path = NULL;
932 
933 	canonpath = strdup(input_path);
934 	if (canonpath == NULL) {
935 		err = got_error_from_errno("strdup");
936 		goto done;
937 	}
938 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
939 	if (err)
940 		goto done;
941 
942 	repo_abspath = got_repo_get_path(repo);
943 
944 	if (canonpath[0] == '\0') {
945 		path = strdup(canonpath);
946 		if (path == NULL) {
947 			err = got_error_from_errno("strdup");
948 			goto done;
949 		}
950 	} else {
951 		path = realpath(canonpath, NULL);
952 		if (path == NULL) {
953 			if (errno != ENOENT) {
954 				err = got_error_from_errno2("realpath",
955 				    canonpath);
956 				goto done;
957 			}
958 			/*
959 			 * Path is not on disk.
960 			 * Assume it is already relative to repository root.
961 			 */
962 			path = strdup(canonpath);
963 			if (path == NULL) {
964 				err = got_error_from_errno("strdup");
965 				goto done;
966 			}
967 		}
968 
969 		repolen = strlen(repo_abspath);
970 		len = strlen(path);
971 
972 
973 		if (strcmp(path, repo_abspath) == 0) {
974 			free(path);
975 			path = strdup("");
976 			if (path == NULL) {
977 				err = got_error_from_errno("strdup");
978 				goto done;
979 			}
980 		} else if (len > repolen &&
981 		    got_path_is_child(path, repo_abspath, repolen)) {
982 			/* Matched an on-disk path inside repository. */
983 			if (got_repo_is_bare(repo)) {
984 				/*
985 				 * Matched an on-disk path inside repository
986 				 * database. Treat input as repository-relative.
987 				 */
988 				free(path);
989 				path = canonpath;
990 				canonpath = NULL;
991 			} else {
992 				char *child;
993 				/* Strip common prefix with repository path. */
994 				err = got_path_skip_common_ancestor(&child,
995 				    repo_abspath, path);
996 				if (err)
997 					goto done;
998 				free(path);
999 				path = child;
1000 			}
1001 		} else {
1002 			/*
1003 			 * Matched unrelated on-disk path.
1004 			 * Treat input as repository-relative.
1005 			 */
1006 			free(path);
1007 			path = canonpath;
1008 			canonpath = NULL;
1009 		}
1010 	}
1011 
1012 	/* Make in-repository path absolute */
1013 	if (path[0] != '/') {
1014 		char *abspath;
1015 		if (asprintf(&abspath, "/%s", path) == -1) {
1016 			err = got_error_from_errno("asprintf");
1017 			goto done;
1018 		}
1019 		free(path);
1020 		path = abspath;
1021 	}
1022 
1023 done:
1024 	free(canonpath);
1025 	if (err)
1026 		free(path);
1027 	else
1028 		*in_repo_path = path;
1029 	return err;
1030 }
1031 
1032 static const struct got_error *
1033 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1034     const char *path_packidx)
1035 {
1036 	const struct got_error *err = NULL;
1037 	size_t i;
1038 
1039 	for (i = 0; i < repo->pack_cache_size; i++) {
1040 		if (repo->packidx_cache[i] == NULL)
1041 			break;
1042 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1043 		    path_packidx) == 0) {
1044 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1045 		}
1046 	}
1047 	if (i == repo->pack_cache_size) {
1048 		do {
1049 			i--;
1050 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1051 		    i == repo->pinned_packidx);
1052 		err = got_packidx_close(repo->packidx_cache[i]);
1053 		if (err)
1054 			return err;
1055 	}
1056 
1057 	repo->packidx_cache[i] = packidx;
1058 
1059 	return NULL;
1060 }
1061 
1062 int
1063 got_repo_is_packidx_filename(const char *name, size_t len)
1064 {
1065 	if (len != GOT_PACKIDX_NAMELEN)
1066 		return 0;
1067 
1068 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1069 		return 0;
1070 
1071 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1072 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1073 		return 0;
1074 
1075 	return 1;
1076 }
1077 
1078 static struct got_packidx_bloom_filter *
1079 get_packidx_bloom_filter(struct got_repository *repo,
1080     const char *path, size_t path_len)
1081 {
1082 	struct got_packidx_bloom_filter key;
1083 
1084 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1085 		return NULL; /* XXX */
1086 	key.path_len = path_len;
1087 
1088 	return RB_FIND(got_packidx_bloom_filter_tree,
1089 	    &repo->packidx_bloom_filters, &key);
1090 }
1091 
1092 int
1093 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1094     const char *path_packidx, struct got_object_id *id)
1095 {
1096 	struct got_packidx_bloom_filter *bf;
1097 
1098 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1099 	if (bf)
1100 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1101 
1102 	/* No bloom filter means this pack index must be searched. */
1103 	return 1;
1104 }
1105 
1106 static const struct got_error *
1107 add_packidx_bloom_filter(struct got_repository *repo,
1108     struct got_packidx *packidx, const char *path_packidx)
1109 {
1110 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1111 	struct got_packidx_bloom_filter *bf;
1112 	size_t len;
1113 
1114 	/*
1115 	 * Don't use bloom filters for very large pack index files.
1116 	 * Large pack files will contain a relatively large fraction
1117 	 * of our objects so we will likely need to visit them anyway.
1118 	 * The more objects a pack file contains the higher the probability
1119 	 * of a false-positive match from the bloom filter. And reading
1120 	 * all object IDs from a large pack index file can be expensive.
1121 	 */
1122 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1123 		return NULL;
1124 
1125 	/* Do we already have a filter for this pack index? */
1126 	if (get_packidx_bloom_filter(repo, path_packidx,
1127 	    strlen(path_packidx)) != NULL)
1128 		return NULL;
1129 
1130 	bf = calloc(1, sizeof(*bf));
1131 	if (bf == NULL)
1132 		return got_error_from_errno("calloc");
1133 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1134 	if (bf->bloom == NULL) {
1135 		free(bf);
1136 		return got_error_from_errno("calloc");
1137 	}
1138 	
1139 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1140 	if (len >= sizeof(bf->path)) {
1141 		free(bf->bloom);
1142 		free(bf);
1143 		return got_error(GOT_ERR_NO_SPACE);
1144 	}
1145 	bf->path_len = len;
1146 
1147 	/* Minimum size supported by our bloom filter is 1000 entries. */
1148 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1149 	for (i = 0; i < nobjects; i++) {
1150 		struct got_packidx_object_id *id;
1151 		id = &packidx->hdr.sorted_ids[i];
1152 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1153 	}
1154 
1155 	RB_INSERT(got_packidx_bloom_filter_tree,
1156 	    &repo->packidx_bloom_filters, bf);
1157 	return NULL;
1158 }
1159 
1160 const struct got_error *
1161 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1162     struct got_repository *repo, struct got_object_id *id)
1163 {
1164 	const struct got_error *err;
1165 	struct got_pathlist_entry *pe;
1166 	size_t i;
1167 
1168 	/* Search pack index cache. */
1169 	for (i = 0; i < repo->pack_cache_size; i++) {
1170 		if (repo->packidx_cache[i] == NULL)
1171 			break;
1172 		if (!got_repo_check_packidx_bloom_filter(repo,
1173 		    repo->packidx_cache[i]->path_packidx, id))
1174 			continue; /* object will not be found in this index */
1175 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1176 		if (*idx != -1) {
1177 			*packidx = repo->packidx_cache[i];
1178 			/*
1179 			 * Move this cache entry to the front. Repeatedly
1180 			 * searching a wrong pack index can be expensive.
1181 			 */
1182 			if (i > 0) {
1183 				memmove(&repo->packidx_cache[1],
1184 				    &repo->packidx_cache[0],
1185 				    i * sizeof(repo->packidx_cache[0]));
1186 				repo->packidx_cache[0] = *packidx;
1187 				if (repo->pinned_packidx >= 0 &&
1188 				    repo->pinned_packidx < i)
1189 					repo->pinned_packidx++;
1190 				else if (repo->pinned_packidx == i)
1191 					repo->pinned_packidx = 0;
1192 			}
1193 			return NULL;
1194 		}
1195 	}
1196 	/* No luck. Search the filesystem. */
1197 
1198 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1199 		const char *path_packidx = pe->path;
1200 		int is_cached = 0;
1201 
1202 		if (!got_repo_check_packidx_bloom_filter(repo,
1203 		    pe->path, id))
1204 			continue; /* object will not be found in this index */
1205 
1206 		for (i = 0; i < repo->pack_cache_size; i++) {
1207 			if (repo->packidx_cache[i] == NULL)
1208 				break;
1209 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1210 			    path_packidx) == 0) {
1211 				is_cached = 1;
1212 				break;
1213 			}
1214 		}
1215 		if (is_cached)
1216 			continue; /* already searched */
1217 
1218 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1219 		    path_packidx, 0);
1220 		if (err)
1221 			goto done;
1222 
1223 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1224 		if (err)
1225 			goto done;
1226 
1227 		err = cache_packidx(repo, *packidx, path_packidx);
1228 		if (err)
1229 			goto done;
1230 
1231 		*idx = got_packidx_get_object_idx(*packidx, id);
1232 		if (*idx != -1) {
1233 			err = NULL; /* found the object */
1234 			goto done;
1235 		}
1236 	}
1237 
1238 	err = got_error_no_obj(id);
1239 done:
1240 	return err;
1241 }
1242 
1243 const struct got_error *
1244 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1245     struct got_repository *repo)
1246 {
1247 	const struct got_error *err = NULL;
1248 	DIR *packdir = NULL;
1249 	struct dirent *dent;
1250 	char *path_packidx = NULL;
1251 	int packdir_fd;
1252 
1253 	packdir_fd = openat(got_repo_get_fd(repo),
1254 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1255 	if (packdir_fd == -1) {
1256 		return got_error_from_errno_fmt("openat: %s/%s",
1257 		    got_repo_get_path_git_dir(repo),
1258 		    GOT_OBJECTS_PACK_DIR);
1259 	}
1260 
1261 	packdir = fdopendir(packdir_fd);
1262 	if (packdir == NULL) {
1263 		err = got_error_from_errno("fdopendir");
1264 		goto done;
1265 	}
1266 
1267 	while ((dent = readdir(packdir)) != NULL) {
1268 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1269 			continue;
1270 
1271 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1272 		    dent->d_name) == -1) {
1273 			err = got_error_from_errno("asprintf");
1274 			path_packidx = NULL;
1275 			break;
1276 		}
1277 
1278 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1279 		if (err)
1280 			break;
1281 	}
1282 done:
1283 	if (err)
1284 		free(path_packidx);
1285 	if (packdir && closedir(packdir) != 0 && err == NULL)
1286 		err = got_error_from_errno("closedir");
1287 	return err;
1288 }
1289 
1290 const struct got_error *
1291 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1292     struct got_repository *repo)
1293 {
1294 	const struct got_error *err;
1295 	size_t i;
1296 
1297 	*packidx = NULL;
1298 
1299 	/* Search pack index cache. */
1300 	for (i = 0; i < repo->pack_cache_size; i++) {
1301 		if (repo->packidx_cache[i] == NULL)
1302 			break;
1303 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1304 		    path_packidx) == 0) {
1305 			*packidx = repo->packidx_cache[i];
1306 			return NULL;
1307 		}
1308 	}
1309 	/* No luck. Search the filesystem. */
1310 
1311 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1312 	    path_packidx, 0);
1313 	if (err)
1314 		return err;
1315 
1316 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1317 	if (err)
1318 		goto done;
1319 
1320 	err = cache_packidx(repo, *packidx, path_packidx);
1321 done:
1322 	if (err) {
1323 		got_packidx_close(*packidx);
1324 		*packidx = NULL;
1325 	}
1326 	return err;
1327 }
1328 
1329 static const struct got_error *
1330 read_packfile_hdr(int fd, struct got_packidx *packidx)
1331 {
1332 	const struct got_error *err = NULL;
1333 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1334 	struct got_packfile_hdr hdr;
1335 	ssize_t n;
1336 
1337 	n = read(fd, &hdr, sizeof(hdr));
1338 	if (n < 0)
1339 		return got_error_from_errno("read");
1340 	if (n != sizeof(hdr))
1341 		return got_error(GOT_ERR_BAD_PACKFILE);
1342 
1343 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1344 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1345 	    be32toh(hdr.nobjects) != totobj)
1346 		err = got_error(GOT_ERR_BAD_PACKFILE);
1347 
1348 	return err;
1349 }
1350 
1351 static const struct got_error *
1352 open_packfile(int *fd, struct got_repository *repo,
1353     const char *relpath, struct got_packidx *packidx)
1354 {
1355 	const struct got_error *err = NULL;
1356 
1357 	*fd = openat(got_repo_get_fd(repo), relpath,
1358 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1359 	if (*fd == -1)
1360 		return got_error_from_errno_fmt("openat: %s/%s",
1361 		    got_repo_get_path_git_dir(repo), relpath);
1362 
1363 	if (packidx) {
1364 		err = read_packfile_hdr(*fd, packidx);
1365 		if (err) {
1366 			close(*fd);
1367 			*fd = -1;
1368 		}
1369 	}
1370 
1371 	return err;
1372 }
1373 
1374 const struct got_error *
1375 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1376     const char *path_packfile, struct got_packidx *packidx)
1377 {
1378 	const struct got_error *err = NULL;
1379 	struct got_pack *pack = NULL;
1380 	struct stat sb;
1381 	size_t i;
1382 
1383 	if (packp)
1384 		*packp = NULL;
1385 
1386 	for (i = 0; i < repo->pack_cache_size; i++) {
1387 		pack = &repo->packs[i];
1388 		if (pack->path_packfile == NULL)
1389 			break;
1390 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1391 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1392 	}
1393 
1394 	if (i == repo->pack_cache_size) {
1395 		struct got_pack tmp;
1396 		do {
1397 			i--;
1398 		} while (i > 0 && repo->pinned_pack >= 0 &&
1399 		    i == repo->pinned_pack);
1400 		err = got_pack_close(&repo->packs[i]);
1401 		if (err)
1402 			return err;
1403 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1404 			return got_error_from_errno("ftruncate");
1405 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1406 			return got_error_from_errno("ftruncate");
1407 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1408 		memcpy(&repo->packs[i], &repo->packs[0],
1409 		    sizeof(repo->packs[i]));
1410 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1411 		if (repo->pinned_pack == 0)
1412 			repo->pinned_pack = i;
1413 		else if (repo->pinned_pack == i)
1414 			repo->pinned_pack = 0;
1415 		i = 0;
1416 	}
1417 
1418 	pack = &repo->packs[i];
1419 
1420 	pack->path_packfile = strdup(path_packfile);
1421 	if (pack->path_packfile == NULL) {
1422 		err = got_error_from_errno("strdup");
1423 		goto done;
1424 	}
1425 
1426 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1427 	if (err)
1428 		goto done;
1429 
1430 	if (fstat(pack->fd, &sb) != 0) {
1431 		err = got_error_from_errno("fstat");
1432 		goto done;
1433 	}
1434 	pack->filesize = sb.st_size;
1435 
1436 	pack->privsep_child = NULL;
1437 
1438 #ifndef GOT_PACK_NO_MMAP
1439 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1440 	    pack->fd, 0);
1441 	if (pack->map == MAP_FAILED) {
1442 		if (errno != ENOMEM) {
1443 			err = got_error_from_errno("mmap");
1444 			goto done;
1445 		}
1446 		pack->map = NULL; /* fall back to read(2) */
1447 	}
1448 #endif
1449 done:
1450 	if (err) {
1451 		if (pack) {
1452 			free(pack->path_packfile);
1453 			memset(pack, 0, sizeof(*pack));
1454 		}
1455 	} else if (packp)
1456 		*packp = pack;
1457 	return err;
1458 }
1459 
1460 struct got_pack *
1461 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1462 {
1463 	struct got_pack *pack = NULL;
1464 	size_t i;
1465 
1466 	for (i = 0; i < repo->pack_cache_size; i++) {
1467 		pack = &repo->packs[i];
1468 		if (pack->path_packfile == NULL)
1469 			break;
1470 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1471 			return pack;
1472 	}
1473 
1474 	return NULL;
1475 }
1476 
1477 const struct got_error *
1478 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1479     struct got_pack *pack)
1480 {
1481 	size_t i;
1482 	int pinned_pack = -1, pinned_packidx = -1;
1483 
1484 	for (i = 0; i < repo->pack_cache_size; i++) {
1485 		if (repo->packidx_cache[i] &&
1486 		    strcmp(repo->packidx_cache[i]->path_packidx,
1487 		    packidx->path_packidx) == 0)
1488 			pinned_packidx = i;
1489 		if (repo->packs[i].path_packfile &&
1490 		    strcmp(repo->packs[i].path_packfile,
1491 		    pack->path_packfile) == 0)
1492 			pinned_pack = i;
1493 	}
1494 
1495 	if (pinned_packidx == -1 || pinned_pack == -1)
1496 		return got_error(GOT_ERR_PIN_PACK);
1497 
1498 	repo->pinned_pack = pinned_pack;
1499 	repo->pinned_packidx = pinned_packidx;
1500 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1501 	return NULL;
1502 }
1503 
1504 struct got_pack *
1505 got_repo_get_pinned_pack(struct got_repository *repo)
1506 {
1507 	if (repo->pinned_pack >= 0 &&
1508 	    repo->pinned_pack < repo->pack_cache_size)
1509 		return &repo->packs[repo->pinned_pack];
1510 
1511 	return NULL;
1512 }
1513 
1514 void
1515 got_repo_unpin_pack(struct got_repository *repo)
1516 {
1517 	repo->pinned_packidx = -1;
1518 	repo->pinned_pack = -1;
1519 	repo->pinned_pid = 0;
1520 }
1521 
1522 const struct got_error *
1523 got_repo_init(const char *repo_path)
1524 {
1525 	const struct got_error *err = NULL;
1526 	const char *dirnames[] = {
1527 		GOT_OBJECTS_DIR,
1528 		GOT_OBJECTS_PACK_DIR,
1529 		GOT_REFS_DIR,
1530 	};
1531 	const char *description_str = "Unnamed repository; "
1532 	    "edit this file 'description' to name the repository.";
1533 	const char *headref_str = "ref: refs/heads/main";
1534 	const char *gitconfig_str = "[core]\n"
1535 	    "\trepositoryformatversion = 0\n"
1536 	    "\tfilemode = true\n"
1537 	    "\tbare = true\n";
1538 	char *path;
1539 	size_t i;
1540 
1541 	if (!got_path_dir_is_empty(repo_path))
1542 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1543 
1544 	for (i = 0; i < nitems(dirnames); i++) {
1545 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1546 			return got_error_from_errno("asprintf");
1547 		}
1548 		err = got_path_mkdir(path);
1549 		free(path);
1550 		if (err)
1551 			return err;
1552 	}
1553 
1554 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1555 		return got_error_from_errno("asprintf");
1556 	err = got_path_create_file(path, description_str);
1557 	free(path);
1558 	if (err)
1559 		return err;
1560 
1561 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1562 		return got_error_from_errno("asprintf");
1563 	err = got_path_create_file(path, headref_str);
1564 	free(path);
1565 	if (err)
1566 		return err;
1567 
1568 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1569 		return got_error_from_errno("asprintf");
1570 	err = got_path_create_file(path, gitconfig_str);
1571 	free(path);
1572 	if (err)
1573 		return err;
1574 
1575 	return NULL;
1576 }
1577 
1578 static const struct got_error *
1579 match_packed_object(struct got_object_id **unique_id,
1580     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1581 {
1582 	const struct got_error *err = NULL;
1583 	struct got_object_id_queue matched_ids;
1584 	struct got_pathlist_entry *pe;
1585 
1586 	STAILQ_INIT(&matched_ids);
1587 
1588 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1589 		const char *path_packidx = pe->path;
1590 		struct got_packidx *packidx;
1591 		struct got_object_qid *qid;
1592 
1593 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1594 		    path_packidx, 0);
1595 		if (err)
1596 			break;
1597 
1598 		err = got_packidx_match_id_str_prefix(&matched_ids,
1599 		    packidx, id_str_prefix);
1600 		if (err) {
1601 			got_packidx_close(packidx);
1602 			break;
1603 		}
1604 		err = got_packidx_close(packidx);
1605 		if (err)
1606 			break;
1607 
1608 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1609 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1610 				int matched_type;
1611 				err = got_object_get_type(&matched_type, repo,
1612 				    &qid->id);
1613 				if (err)
1614 					goto done;
1615 				if (matched_type != obj_type)
1616 					continue;
1617 			}
1618 			if (*unique_id == NULL) {
1619 				*unique_id = got_object_id_dup(&qid->id);
1620 				if (*unique_id == NULL) {
1621 					err = got_error_from_errno("malloc");
1622 					goto done;
1623 				}
1624 			} else {
1625 				if (got_object_id_cmp(*unique_id,
1626 				    &qid->id) == 0)
1627 					continue; /* packed multiple times */
1628 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1629 				goto done;
1630 			}
1631 		}
1632 	}
1633 done:
1634 	got_object_id_queue_free(&matched_ids);
1635 	if (err) {
1636 		free(*unique_id);
1637 		*unique_id = NULL;
1638 	}
1639 	return err;
1640 }
1641 
1642 static const struct got_error *
1643 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1644     const char *object_dir, const char *id_str_prefix, int obj_type,
1645     struct got_repository *repo)
1646 {
1647 	const struct got_error *err = NULL;
1648 	char *path;
1649 	DIR *dir = NULL;
1650 	struct dirent *dent;
1651 	struct got_object_id id;
1652 
1653 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1654 		err = got_error_from_errno("asprintf");
1655 		goto done;
1656 	}
1657 
1658 	dir = opendir(path);
1659 	if (dir == NULL) {
1660 		if (errno == ENOENT) {
1661 			err = NULL;
1662 			goto done;
1663 		}
1664 		err = got_error_from_errno2("opendir", path);
1665 		goto done;
1666 	}
1667 	while ((dent = readdir(dir)) != NULL) {
1668 		char *id_str;
1669 		int cmp;
1670 
1671 		if (strcmp(dent->d_name, ".") == 0 ||
1672 		    strcmp(dent->d_name, "..") == 0)
1673 			continue;
1674 
1675 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1676 			err = got_error_from_errno("asprintf");
1677 			goto done;
1678 		}
1679 
1680 		if (!got_parse_sha1_digest(id.sha1, id_str))
1681 			continue;
1682 
1683 		/*
1684 		 * Directory entries do not necessarily appear in
1685 		 * sorted order, so we must iterate over all of them.
1686 		 */
1687 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1688 		if (cmp != 0) {
1689 			free(id_str);
1690 			continue;
1691 		}
1692 
1693 		if (*unique_id == NULL) {
1694 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1695 				int matched_type;
1696 				err = got_object_get_type(&matched_type, repo,
1697 				    &id);
1698 				if (err)
1699 					goto done;
1700 				if (matched_type != obj_type)
1701 					continue;
1702 			}
1703 			*unique_id = got_object_id_dup(&id);
1704 			if (*unique_id == NULL) {
1705 				err = got_error_from_errno("got_object_id_dup");
1706 				free(id_str);
1707 				goto done;
1708 			}
1709 		} else {
1710 			if (got_object_id_cmp(*unique_id, &id) == 0)
1711 				continue; /* both packed and loose */
1712 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1713 			free(id_str);
1714 			goto done;
1715 		}
1716 	}
1717 done:
1718 	if (dir && closedir(dir) != 0 && err == NULL)
1719 		err = got_error_from_errno("closedir");
1720 	if (err) {
1721 		free(*unique_id);
1722 		*unique_id = NULL;
1723 	}
1724 	free(path);
1725 	return err;
1726 }
1727 
1728 const struct got_error *
1729 got_repo_match_object_id_prefix(struct got_object_id **id,
1730     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1731 {
1732 	const struct got_error *err = NULL;
1733 	char *path_objects = got_repo_get_path_objects(repo);
1734 	char *object_dir = NULL;
1735 	size_t len;
1736 	int i;
1737 
1738 	*id = NULL;
1739 
1740 	len = strlen(id_str_prefix);
1741 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1742 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1743 
1744 	for (i = 0; i < len; i++) {
1745 		if (isxdigit((unsigned char)id_str_prefix[i]))
1746 			continue;
1747 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1748 	}
1749 
1750 	if (len >= 2) {
1751 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1752 		if (err)
1753 			goto done;
1754 		object_dir = strndup(id_str_prefix, 2);
1755 		if (object_dir == NULL) {
1756 			err = got_error_from_errno("strdup");
1757 			goto done;
1758 		}
1759 		err = match_loose_object(id, path_objects, object_dir,
1760 		    id_str_prefix, obj_type, repo);
1761 	} else if (len == 1) {
1762 		int i;
1763 		for (i = 0; i < 0xf; i++) {
1764 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1765 			    == -1) {
1766 				err = got_error_from_errno("asprintf");
1767 				goto done;
1768 			}
1769 			err = match_packed_object(id, repo, object_dir,
1770 			    obj_type);
1771 			if (err)
1772 				goto done;
1773 			err = match_loose_object(id, path_objects, object_dir,
1774 			    id_str_prefix, obj_type, repo);
1775 			if (err)
1776 				goto done;
1777 		}
1778 	} else {
1779 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1780 		goto done;
1781 	}
1782 done:
1783 	free(object_dir);
1784 	if (err) {
1785 		free(*id);
1786 		*id = NULL;
1787 	} else if (*id == NULL) {
1788 		switch (obj_type) {
1789 		case GOT_OBJ_TYPE_BLOB:
1790 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1791 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1792 			break;
1793 		case GOT_OBJ_TYPE_TREE:
1794 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1795 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1796 			break;
1797 		case GOT_OBJ_TYPE_COMMIT:
1798 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1799 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1800 			break;
1801 		case GOT_OBJ_TYPE_TAG:
1802 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1803 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1804 			break;
1805 		default:
1806 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1807 			break;
1808 		}
1809 	}
1810 
1811 	return err;
1812 }
1813 
1814 const struct got_error *
1815 got_repo_match_object_id(struct got_object_id **id, char **label,
1816     const char *id_str, int obj_type, struct got_reflist_head *refs,
1817     struct got_repository *repo)
1818 {
1819 	const struct got_error *err;
1820 	struct got_tag_object *tag;
1821 	struct got_reference *ref = NULL;
1822 
1823 	*id = NULL;
1824 	if (label)
1825 		*label = NULL;
1826 
1827 	if (refs) {
1828 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1829 		    refs, repo);
1830 		if (err == NULL) {
1831 			*id = got_object_id_dup(
1832 			    got_object_tag_get_object_id(tag));
1833 			if (*id == NULL)
1834 				err = got_error_from_errno("got_object_id_dup");
1835 			else if (label && asprintf(label, "refs/tags/%s",
1836 			    got_object_tag_get_name(tag)) == -1) {
1837 				err = got_error_from_errno("asprintf");
1838 				free(*id);
1839 				*id = NULL;
1840 			}
1841 			got_object_tag_close(tag);
1842 			return err;
1843 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1844 		    err->code != GOT_ERR_NO_OBJ)
1845 			return err;
1846 	}
1847 
1848 	err = got_ref_open(&ref, repo, id_str, 0);
1849 	if (err == NULL) {
1850 		err = got_ref_resolve(id, repo, ref);
1851 		if (err)
1852 			goto done;
1853 		if (label) {
1854 			*label = strdup(got_ref_get_name(ref));
1855 			if (*label == NULL) {
1856 				err = got_error_from_errno("strdup");
1857 				goto done;
1858 			}
1859 		}
1860 	} else {
1861 		if (err->code != GOT_ERR_NOT_REF &&
1862 		    err->code != GOT_ERR_BAD_REF_NAME)
1863 			goto done;
1864 		err = got_repo_match_object_id_prefix(id, id_str,
1865 		    obj_type, repo);
1866 		if (err) {
1867 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1868 				err = got_error_not_ref(id_str);
1869 			goto done;
1870 		}
1871 		if (label) {
1872 			err = got_object_id_str(label, *id);
1873 			if (*label == NULL) {
1874 				err = got_error_from_errno("strdup");
1875 				goto done;
1876 			}
1877 		}
1878 	}
1879 done:
1880 	if (ref)
1881 		got_ref_close(ref);
1882 	return err;
1883 }
1884 
1885 const struct got_error *
1886 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1887     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1888 {
1889 	const struct got_error *err = NULL;
1890 	struct got_reflist_entry *re;
1891 	struct got_object_id *tag_id;
1892 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1893 
1894 	*tag = NULL;
1895 
1896 	TAILQ_FOREACH(re, refs, entry) {
1897 		const char *refname;
1898 		refname = got_ref_get_name(re->ref);
1899 		if (got_ref_is_symbolic(re->ref))
1900 			continue;
1901 		if (strncmp(refname, "refs/tags/", 10) != 0)
1902 			continue;
1903 		if (!name_is_absolute)
1904 			refname += strlen("refs/tags/");
1905 		if (strcmp(refname, name) != 0)
1906 			continue;
1907 		err = got_ref_resolve(&tag_id, repo, re->ref);
1908 		if (err)
1909 			break;
1910 		err = got_object_open_as_tag(tag, repo, tag_id);
1911 		free(tag_id);
1912 		if (err)
1913 			break;
1914 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1915 		    got_object_tag_get_object_type(*tag) == obj_type)
1916 			break;
1917 		got_object_tag_close(*tag);
1918 		*tag = NULL;
1919 	}
1920 
1921 	if (err == NULL && *tag == NULL)
1922 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1923 		    GOT_OBJ_LABEL_TAG, name);
1924 	return err;
1925 }
1926 
1927 static const struct got_error *
1928 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1929     const char *name, mode_t mode, struct got_object_id *blob_id)
1930 {
1931 	const struct got_error *err = NULL;
1932 
1933 	 *new_te = NULL;
1934 
1935 	*new_te = calloc(1, sizeof(**new_te));
1936 	if (*new_te == NULL)
1937 		return got_error_from_errno("calloc");
1938 
1939 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1940 	    sizeof((*new_te)->name)) {
1941 		err = got_error(GOT_ERR_NO_SPACE);
1942 		goto done;
1943 	}
1944 
1945 	if (S_ISLNK(mode)) {
1946 		(*new_te)->mode = S_IFLNK;
1947 	} else {
1948 		(*new_te)->mode = S_IFREG;
1949 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1950 	}
1951 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1952 done:
1953 	if (err && *new_te) {
1954 		free(*new_te);
1955 		*new_te = NULL;
1956 	}
1957 	return err;
1958 }
1959 
1960 static const struct got_error *
1961 import_file(struct got_tree_entry **new_te, struct dirent *de,
1962     const char *path, struct got_repository *repo)
1963 {
1964 	const struct got_error *err;
1965 	struct got_object_id *blob_id = NULL;
1966 	char *filepath;
1967 	struct stat sb;
1968 
1969 	if (asprintf(&filepath, "%s%s%s", path,
1970 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1971 		return got_error_from_errno("asprintf");
1972 
1973 	if (lstat(filepath, &sb) != 0) {
1974 		err = got_error_from_errno2("lstat", path);
1975 		goto done;
1976 	}
1977 
1978 	err = got_object_blob_create(&blob_id, filepath, repo);
1979 	if (err)
1980 		goto done;
1981 
1982 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1983 	    blob_id);
1984 done:
1985 	free(filepath);
1986 	if (err)
1987 		free(blob_id);
1988 	return err;
1989 }
1990 
1991 static const struct got_error *
1992 insert_tree_entry(struct got_tree_entry *new_te,
1993     struct got_pathlist_head *paths)
1994 {
1995 	const struct got_error *err = NULL;
1996 	struct got_pathlist_entry *new_pe;
1997 
1998 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1999 	if (err)
2000 		return err;
2001 	if (new_pe == NULL)
2002 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2003 	return NULL;
2004 }
2005 
2006 static const struct got_error *write_tree(struct got_object_id **,
2007     const char *, struct got_pathlist_head *, struct got_repository *,
2008     got_repo_import_cb progress_cb, void *progress_arg);
2009 
2010 static const struct got_error *
2011 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2012     const char *path, struct got_pathlist_head *ignores,
2013     struct got_repository *repo,
2014     got_repo_import_cb progress_cb, void *progress_arg)
2015 {
2016 	const struct got_error *err;
2017 	struct got_object_id *id = NULL;
2018 	char *subdirpath;
2019 
2020 	if (asprintf(&subdirpath, "%s%s%s", path,
2021 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2022 		return got_error_from_errno("asprintf");
2023 
2024 	(*new_te) = calloc(1, sizeof(**new_te));
2025 	if (*new_te == NULL)
2026 		return got_error_from_errno("calloc");
2027 	(*new_te)->mode = S_IFDIR;
2028 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2029 	    sizeof((*new_te)->name)) {
2030 		err = got_error(GOT_ERR_NO_SPACE);
2031 		goto done;
2032 	}
2033 	err = write_tree(&id, subdirpath, ignores,  repo,
2034 	    progress_cb, progress_arg);
2035 	if (err)
2036 		goto done;
2037 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2038 	
2039 done:
2040 	free(id);
2041 	free(subdirpath);
2042 	if (err) {
2043 		free(*new_te);
2044 		*new_te = NULL;
2045 	}
2046 	return err;
2047 }
2048 
2049 static const struct got_error *
2050 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2051     struct got_pathlist_head *ignores, struct got_repository *repo,
2052     got_repo_import_cb progress_cb, void *progress_arg)
2053 {
2054 	const struct got_error *err = NULL;
2055 	DIR *dir;
2056 	struct dirent *de;
2057 	int nentries;
2058 	struct got_tree_entry *new_te = NULL;
2059 	struct got_pathlist_head paths;
2060 	struct got_pathlist_entry *pe;
2061 
2062 	*new_tree_id = NULL;
2063 
2064 	TAILQ_INIT(&paths);
2065 
2066 	dir = opendir(path_dir);
2067 	if (dir == NULL) {
2068 		err = got_error_from_errno2("opendir", path_dir);
2069 		goto done;
2070 	}
2071 
2072 	nentries = 0;
2073 	while ((de = readdir(dir)) != NULL) {
2074 		int ignore = 0;
2075 		int type;
2076 
2077 		if (strcmp(de->d_name, ".") == 0 ||
2078 		    strcmp(de->d_name, "..") == 0)
2079 			continue;
2080 
2081 		TAILQ_FOREACH(pe, ignores, entry) {
2082 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2083 				ignore = 1;
2084 				break;
2085 			}
2086 		}
2087 		if (ignore)
2088 			continue;
2089 
2090 		err = got_path_dirent_type(&type, path_dir, de);
2091 		if (err)
2092 			goto done;
2093 
2094 		if (type == DT_DIR) {
2095 			err = import_subdir(&new_te, de, path_dir,
2096 			    ignores, repo, progress_cb, progress_arg);
2097 			if (err) {
2098 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2099 					goto done;
2100 				err = NULL;
2101 				continue;
2102 			}
2103 		} else if (type == DT_REG || type == DT_LNK) {
2104 			err = import_file(&new_te, de, path_dir, repo);
2105 			if (err)
2106 				goto done;
2107 		} else
2108 			continue;
2109 
2110 		err = insert_tree_entry(new_te, &paths);
2111 		if (err)
2112 			goto done;
2113 		nentries++;
2114 	}
2115 
2116 	if (TAILQ_EMPTY(&paths)) {
2117 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2118 		    "cannot create tree without any entries");
2119 		goto done;
2120 	}
2121 
2122 	TAILQ_FOREACH(pe, &paths, entry) {
2123 		struct got_tree_entry *te = pe->data;
2124 		char *path;
2125 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2126 			continue;
2127 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2128 			err = got_error_from_errno("asprintf");
2129 			goto done;
2130 		}
2131 		err = (*progress_cb)(progress_arg, path);
2132 		free(path);
2133 		if (err)
2134 			goto done;
2135 	}
2136 
2137 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2138 done:
2139 	if (dir)
2140 		closedir(dir);
2141 	got_pathlist_free(&paths);
2142 	return err;
2143 }
2144 
2145 const struct got_error *
2146 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2147     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2148     struct got_repository *repo, got_repo_import_cb progress_cb,
2149     void *progress_arg)
2150 {
2151 	const struct got_error *err;
2152 	struct got_object_id *new_tree_id;
2153 
2154 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2155 	    progress_cb, progress_arg);
2156 	if (err)
2157 		return err;
2158 
2159 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2160 	    author, time(NULL), author, time(NULL), logmsg, repo);
2161 	free(new_tree_id);
2162 	return err;
2163 }
2164 
2165 const struct got_error *
2166 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2167     struct got_repository *repo)
2168 {
2169 	const struct got_error *err = NULL;
2170 	char *path_objects = NULL, *path = NULL;
2171 	DIR *dir = NULL;
2172 	struct got_object_id id;
2173 	int i;
2174 
2175 	*nobjects = 0;
2176 	*ondisk_size = 0;
2177 
2178 	path_objects = got_repo_get_path_objects(repo);
2179 	if (path_objects == NULL)
2180 		return got_error_from_errno("got_repo_get_path_objects");
2181 
2182 	for (i = 0; i <= 0xff; i++) {
2183 		struct dirent *dent;
2184 
2185 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2186 			err = got_error_from_errno("asprintf");
2187 			break;
2188 		}
2189 
2190 		dir = opendir(path);
2191 		if (dir == NULL) {
2192 			if (errno == ENOENT) {
2193 				err = NULL;
2194 				continue;
2195 			}
2196 			err = got_error_from_errno2("opendir", path);
2197 			break;
2198 		}
2199 
2200 		while ((dent = readdir(dir)) != NULL) {
2201 			char *id_str;
2202 			int fd;
2203 			struct stat sb;
2204 
2205 			if (strcmp(dent->d_name, ".") == 0 ||
2206 			    strcmp(dent->d_name, "..") == 0)
2207 				continue;
2208 
2209 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2210 				err = got_error_from_errno("asprintf");
2211 				goto done;
2212 			}
2213 
2214 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2215 				free(id_str);
2216 				continue;
2217 			}
2218 			free(id_str);
2219 
2220 			err = got_object_open_loose_fd(&fd, &id, repo);
2221 			if (err)
2222 				goto done;
2223 
2224 			if (fstat(fd, &sb) == -1) {
2225 				err = got_error_from_errno("fstat");
2226 				close(fd);
2227 				goto done;
2228 			}
2229 			(*nobjects)++;
2230 			(*ondisk_size) += sb.st_size;
2231 
2232 			if (close(fd) == -1) {
2233 				err = got_error_from_errno("close");
2234 				goto done;
2235 			}
2236 		}
2237 
2238 		if (closedir(dir) != 0) {
2239 			err = got_error_from_errno("closedir");
2240 			goto done;
2241 		}
2242 		dir = NULL;
2243 
2244 		free(path);
2245 		path = NULL;
2246 	}
2247 done:
2248 	if (dir && closedir(dir) != 0 && err == NULL)
2249 		err = got_error_from_errno("closedir");
2250 
2251 	if (err) {
2252 		*nobjects = 0;
2253 		*ondisk_size = 0;
2254 	}
2255 	free(path_objects);
2256 	free(path);
2257 	return err;
2258 }
2259 
2260 const struct got_error *
2261 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2262     off_t *total_packsize, struct got_repository *repo)
2263 {
2264 	const struct got_error *err = NULL;
2265 	DIR *packdir = NULL;
2266 	struct dirent *dent;
2267 	struct got_packidx *packidx = NULL;
2268 	char *path_packidx;
2269 	char *path_packfile;
2270 	int packdir_fd;
2271 	struct stat sb;
2272 
2273 	*npackfiles = 0;
2274 	*nobjects = 0;
2275 	*total_packsize = 0;
2276 
2277 	packdir_fd = openat(got_repo_get_fd(repo),
2278 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2279 	if (packdir_fd == -1) {
2280 		return got_error_from_errno_fmt("openat: %s/%s",
2281 		    got_repo_get_path_git_dir(repo),
2282 		    GOT_OBJECTS_PACK_DIR);
2283 	}
2284 
2285 	packdir = fdopendir(packdir_fd);
2286 	if (packdir == NULL) {
2287 		err = got_error_from_errno("fdopendir");
2288 		goto done;
2289 	}
2290 
2291 	while ((dent = readdir(packdir)) != NULL) {
2292 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2293 			continue;
2294 
2295 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2296 		    dent->d_name) == -1) {
2297 			err = got_error_from_errno("asprintf");
2298 			goto done;
2299 		}
2300 
2301 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2302 		    path_packidx, 0);
2303 		free(path_packidx);
2304 		if (err)
2305 			goto done;
2306 
2307 		if (fstat(packidx->fd, &sb) == -1)
2308 			goto done;
2309 		*total_packsize += sb.st_size;
2310 
2311 		err = got_packidx_get_packfile_path(&path_packfile,
2312 		    packidx->path_packidx);
2313 		if (err)
2314 			goto done;
2315 
2316 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2317 		    0) == -1) {
2318 			free(path_packfile);
2319 			goto done;
2320 		}
2321 		free(path_packfile);
2322 		*total_packsize += sb.st_size;
2323 
2324 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2325 
2326 		(*npackfiles)++;
2327 
2328 		got_packidx_close(packidx);
2329 		packidx = NULL;
2330 	}
2331 done:
2332 	if (packidx)
2333 		got_packidx_close(packidx);
2334 	if (packdir && closedir(packdir) != 0 && err == NULL)
2335 		err = got_error_from_errno("closedir");
2336 	if (err) {
2337 		*npackfiles = 0;
2338 		*nobjects = 0;
2339 		*total_packsize = 0;
2340 	}
2341 	return err;
2342 }
2343 
2344 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2345     got_packidx_bloom_filter_cmp);