1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
72 
73 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
74     got_packidx_bloom_filter_cmp);
75 
76 const char *
77 got_repo_get_path(struct got_repository *repo)
78 {
79 	return repo->path;
80 }
81 
82 const char *
83 got_repo_get_path_git_dir(struct got_repository *repo)
84 {
85 	return repo->path_git_dir;
86 }
87 
88 int
89 got_repo_get_fd(struct got_repository *repo)
90 {
91 	return repo->gitdir_fd;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_name(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_name;
98 }
99 
100 const char *
101 got_repo_get_gitconfig_author_email(struct got_repository *repo)
102 {
103 	return repo->gitconfig_author_email;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_name;
110 }
111 
112 const char *
113 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
114 {
115 	return repo->global_gitconfig_author_email;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_owner(struct got_repository *repo)
120 {
121 	return repo->gitconfig_owner;
122 }
123 
124 void
125 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
126     struct got_repository *repo)
127 {
128 	*extensions = repo->extensions;
129 	*nextensions = repo->nextensions;
130 }
131 
132 int
133 got_repo_is_bare(struct got_repository *repo)
134 {
135 	return (strcmp(repo->path, repo->path_git_dir) == 0);
136 }
137 
138 static char *
139 get_path_git_child(struct got_repository *repo, const char *basename)
140 {
141 	char *path_child;
142 
143 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
144 	    basename) == -1)
145 		return NULL;
146 
147 	return path_child;
148 }
149 
150 char *
151 got_repo_get_path_objects(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
154 }
155 
156 char *
157 got_repo_get_path_objects_pack(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
160 }
161 
162 char *
163 got_repo_get_path_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_REFS_DIR);
166 }
167 
168 char *
169 got_repo_get_path_packed_refs(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
172 }
173 
174 static char *
175 get_path_head(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_HEAD_FILE);
178 }
179 
180 char *
181 got_repo_get_path_gitconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GITCONFIG);
184 }
185 
186 char *
187 got_repo_get_path_gotconfig(struct got_repository *repo)
188 {
189 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
190 }
191 
192 const struct got_gotconfig *
193 got_repo_get_gotconfig(struct got_repository *repo)
194 {
195 	return repo->gotconfig;
196 }
197 
198 void
199 got_repo_get_gitconfig_remotes(int *nremotes,
200     const struct got_remote_repo **remotes, struct got_repository *repo)
201 {
202 	*nremotes = repo->ngitconfig_remotes;
203 	*remotes = repo->gitconfig_remotes;
204 }
205 
206 static int
207 is_git_repo(struct got_repository *repo)
208 {
209 	const char *path_git = got_repo_get_path_git_dir(repo);
210 	char *path_objects = got_repo_get_path_objects(repo);
211 	char *path_refs = got_repo_get_path_refs(repo);
212 	char *path_head = get_path_head(repo);
213 	int ret = 0;
214 	struct stat sb;
215 	struct got_reference *head_ref;
216 
217 	if (lstat(path_git, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_objects, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_refs, &sb) == -1)
228 		goto done;
229 	if (!S_ISDIR(sb.st_mode))
230 		goto done;
231 
232 	if (lstat(path_head, &sb) == -1)
233 		goto done;
234 	if (!S_ISREG(sb.st_mode))
235 		goto done;
236 
237 	/* Check if the HEAD reference can be opened. */
238 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
239 		goto done;
240 	got_ref_close(head_ref);
241 
242 	ret = 1;
243 done:
244 	free(path_objects);
245 	free(path_refs);
246 	free(path_head);
247 	return ret;
248 
249 }
250 
251 const struct got_error *
252 got_repo_pack_fds_open(int **pack_fds)
253 {
254 	const struct got_error *err = NULL;
255 	int i, *pack_fds_tmp;
256 
257 	pack_fds_tmp = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(int));
258 	if (pack_fds_tmp == NULL)
259 		return got_error_from_errno("calloc");
260 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
261 	if (*pack_fds == NULL) {
262 		free(pack_fds_tmp);
263 		return got_error_from_errno("calloc");
264 	}
265 
266 	/*
267 	 * got_repo_pack_fds_close will try to close all of the
268 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
269 	 * a value from got_opentempfd(), which would result in a close(0) if
270 	 * we do not initialize to -1 here.
271 	 */
272 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
273 		pack_fds_tmp[i] = -1;
274 
275 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
276 		pack_fds_tmp[i] = got_opentempfd();
277 		if (pack_fds_tmp[i] == -1) {
278 			err = got_error_from_errno("got_opentempfd");
279 			got_repo_pack_fds_close(pack_fds_tmp);
280 			return err;
281 		}
282 	}
283 	memcpy(*pack_fds, pack_fds_tmp, GOT_PACK_NUM_TEMPFILES * sizeof(int));
284 	return err;
285 }
286 
287 const struct got_error *
288 got_repo_pack_fds_close(int *pack_fds)
289 {
290 	const struct got_error *err = NULL;
291 	int i;
292 
293 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
294 		if (pack_fds[i] == -1)
295 			continue;
296 		if (close(pack_fds[i]) == -1) {
297 			err = got_error_from_errno("close");
298 			break;
299 		}
300 	}
301 	free(pack_fds);
302 	return err;
303 }
304 
305 const struct got_error *
306 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
307     struct got_object *obj)
308 {
309 #ifndef GOT_NO_OBJ_CACHE
310 	const struct got_error *err = NULL;
311 	err = got_object_cache_add(&repo->objcache, id, obj);
312 	if (err) {
313 		if (err->code == GOT_ERR_OBJ_EXISTS ||
314 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
315 			err = NULL;
316 		return err;
317 	}
318 	obj->refcnt++;
319 #endif
320 	return NULL;
321 }
322 
323 struct got_object *
324 got_repo_get_cached_object(struct got_repository *repo,
325     struct got_object_id *id)
326 {
327 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
328 }
329 
330 const struct got_error *
331 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
332     struct got_tree_object *tree)
333 {
334 #ifndef GOT_NO_OBJ_CACHE
335 	const struct got_error *err = NULL;
336 	err = got_object_cache_add(&repo->treecache, id, tree);
337 	if (err) {
338 		if (err->code == GOT_ERR_OBJ_EXISTS ||
339 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
340 			err = NULL;
341 		return err;
342 	}
343 	tree->refcnt++;
344 #endif
345 	return NULL;
346 }
347 
348 struct got_tree_object *
349 got_repo_get_cached_tree(struct got_repository *repo,
350     struct got_object_id *id)
351 {
352 	return (struct got_tree_object *)got_object_cache_get(
353 	    &repo->treecache, id);
354 }
355 
356 const struct got_error *
357 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
358     struct got_commit_object *commit)
359 {
360 #ifndef GOT_NO_OBJ_CACHE
361 	const struct got_error *err = NULL;
362 	err = got_object_cache_add(&repo->commitcache, id, commit);
363 	if (err) {
364 		if (err->code == GOT_ERR_OBJ_EXISTS ||
365 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
366 			err = NULL;
367 		return err;
368 	}
369 	commit->refcnt++;
370 #endif
371 	return NULL;
372 }
373 
374 struct got_commit_object *
375 got_repo_get_cached_commit(struct got_repository *repo,
376     struct got_object_id *id)
377 {
378 	return (struct got_commit_object *)got_object_cache_get(
379 	    &repo->commitcache, id);
380 }
381 
382 const struct got_error *
383 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
384     struct got_tag_object *tag)
385 {
386 #ifndef GOT_NO_OBJ_CACHE
387 	const struct got_error *err = NULL;
388 	err = got_object_cache_add(&repo->tagcache, id, tag);
389 	if (err) {
390 		if (err->code == GOT_ERR_OBJ_EXISTS ||
391 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
392 			err = NULL;
393 		return err;
394 	}
395 	tag->refcnt++;
396 #endif
397 	return NULL;
398 }
399 
400 struct got_tag_object *
401 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
402 {
403 	return (struct got_tag_object *)got_object_cache_get(
404 	    &repo->tagcache, id);
405 }
406 
407 const struct got_error *
408 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
409     struct got_raw_object *raw)
410 {
411 #ifndef GOT_NO_OBJ_CACHE
412 	const struct got_error *err = NULL;
413 	err = got_object_cache_add(&repo->rawcache, id, raw);
414 	if (err) {
415 		if (err->code == GOT_ERR_OBJ_EXISTS ||
416 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
417 			err = NULL;
418 		return err;
419 	}
420 	raw->refcnt++;
421 #endif
422 	return NULL;
423 }
424 
425 
426 struct got_raw_object *
427 got_repo_get_cached_raw_object(struct got_repository *repo,
428     struct got_object_id *id)
429 {
430 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
431 }
432 
433 
434 static const struct got_error *
435 open_repo(struct got_repository *repo, const char *path)
436 {
437 	const struct got_error *err = NULL;
438 
439 	repo->gitdir_fd = -1;
440 
441 	/* bare git repository? */
442 	repo->path_git_dir = strdup(path);
443 	if (repo->path_git_dir == NULL)
444 		return got_error_from_errno("strdup");
445 	if (is_git_repo(repo)) {
446 		repo->path = strdup(repo->path_git_dir);
447 		if (repo->path == NULL) {
448 			err = got_error_from_errno("strdup");
449 			goto done;
450 		}
451 		repo->gitdir_fd = open(repo->path_git_dir,
452 		    O_DIRECTORY | O_CLOEXEC);
453 		if (repo->gitdir_fd == -1) {
454 			err = got_error_from_errno2("open",
455 			    repo->path_git_dir);
456 			goto done;
457 		}
458 		return NULL;
459 	}
460 
461 	/* git repository with working tree? */
462 	free(repo->path_git_dir);
463 	repo->path_git_dir = NULL;
464 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
465 		err = got_error_from_errno("asprintf");
466 		goto done;
467 	}
468 	if (is_git_repo(repo)) {
469 		repo->path = strdup(path);
470 		if (repo->path == NULL) {
471 			err = got_error_from_errno("strdup");
472 			goto done;
473 		}
474 		repo->gitdir_fd = open(repo->path_git_dir,
475 		    O_DIRECTORY | O_CLOEXEC);
476 		if (repo->gitdir_fd == -1) {
477 			err = got_error_from_errno2("open",
478 			    repo->path_git_dir);
479 			goto done;
480 		}
481 		return NULL;
482 	}
483 
484 	err = got_error(GOT_ERR_NOT_GIT_REPO);
485 done:
486 	if (err) {
487 		free(repo->path);
488 		repo->path = NULL;
489 		free(repo->path_git_dir);
490 		repo->path_git_dir = NULL;
491 		if (repo->gitdir_fd != -1)
492 			close(repo->gitdir_fd);
493 		repo->gitdir_fd = -1;
494 
495 	}
496 	return err;
497 }
498 
499 static const struct got_error *
500 parse_gitconfig_file(int *gitconfig_repository_format_version,
501     char **gitconfig_author_name, char **gitconfig_author_email,
502     struct got_remote_repo **remotes, int *nremotes,
503     char **gitconfig_owner, char ***extensions, int *nextensions,
504     const char *gitconfig_path)
505 {
506 	const struct got_error *err = NULL, *child_err = NULL;
507 	int fd = -1;
508 	int imsg_fds[2] = { -1, -1 };
509 	pid_t pid;
510 	struct imsgbuf *ibuf;
511 
512 	*gitconfig_repository_format_version = 0;
513 	if (extensions)
514 		*extensions = NULL;
515 	if (nextensions)
516 		*nextensions = 0;
517 	*gitconfig_author_name = NULL;
518 	*gitconfig_author_email = NULL;
519 	if (remotes)
520 		*remotes = NULL;
521 	if (nremotes)
522 		*nremotes = 0;
523 	if (gitconfig_owner)
524 		*gitconfig_owner = NULL;
525 
526 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
527 	if (fd == -1) {
528 		if (errno == ENOENT)
529 			return NULL;
530 		return got_error_from_errno2("open", gitconfig_path);
531 	}
532 
533 	ibuf = calloc(1, sizeof(*ibuf));
534 	if (ibuf == NULL) {
535 		err = got_error_from_errno("calloc");
536 		goto done;
537 	}
538 
539 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
540 		err = got_error_from_errno("socketpair");
541 		goto done;
542 	}
543 
544 	pid = fork();
545 	if (pid == -1) {
546 		err = got_error_from_errno("fork");
547 		goto done;
548 	} else if (pid == 0) {
549 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
550 		    gitconfig_path);
551 		/* not reached */
552 	}
553 
554 	if (close(imsg_fds[1]) == -1) {
555 		err = got_error_from_errno("close");
556 		goto done;
557 	}
558 	imsg_fds[1] = -1;
559 	imsg_init(ibuf, imsg_fds[0]);
560 
561 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
562 	if (err)
563 		goto done;
564 	fd = -1;
565 
566 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
567 	if (err)
568 		goto done;
569 
570 	err = got_privsep_recv_gitconfig_int(
571 	    gitconfig_repository_format_version, ibuf);
572 	if (err)
573 		goto done;
574 
575 	if (extensions && nextensions) {
576 		err = got_privsep_send_gitconfig_repository_extensions_req(
577 		    ibuf);
578 		if (err)
579 			goto done;
580 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
581 		if (err)
582 			goto done;
583 		if (*nextensions > 0) {
584 			int i;
585 			*extensions = calloc(*nextensions, sizeof(char *));
586 			if (*extensions == NULL) {
587 				err = got_error_from_errno("calloc");
588 				goto done;
589 			}
590 			for (i = 0; i < *nextensions; i++) {
591 				char *ext;
592 				err = got_privsep_recv_gitconfig_str(&ext,
593 				    ibuf);
594 				if (err)
595 					goto done;
596 				(*extensions)[i] = ext;
597 			}
598 		}
599 	}
600 
601 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
602 	if (err)
603 		goto done;
604 
605 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
606 	if (err)
607 		goto done;
608 
609 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
610 	if (err)
611 		goto done;
612 
613 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
614 	if (err)
615 		goto done;
616 
617 	if (remotes && nremotes) {
618 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
619 		if (err)
620 			goto done;
621 
622 		err = got_privsep_recv_gitconfig_remotes(remotes,
623 		    nremotes, ibuf);
624 		if (err)
625 			goto done;
626 	}
627 
628 	if (gitconfig_owner) {
629 		err = got_privsep_send_gitconfig_owner_req(ibuf);
630 		if (err)
631 			goto done;
632 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
633 		if (err)
634 			goto done;
635 	}
636 
637 	err = got_privsep_send_stop(imsg_fds[0]);
638 	child_err = got_privsep_wait_for_child(pid);
639 	if (child_err && err == NULL)
640 		err = child_err;
641 done:
642 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
643 		err = got_error_from_errno("close");
644 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
645 		err = got_error_from_errno("close");
646 	if (fd != -1 && close(fd) == -1 && err == NULL)
647 		err = got_error_from_errno2("close", gitconfig_path);
648 	free(ibuf);
649 	return err;
650 }
651 
652 static const struct got_error *
653 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
654 {
655 	const struct got_error *err = NULL;
656 	char *repo_gitconfig_path = NULL;
657 
658 	if (global_gitconfig_path) {
659 		/* Read settings from ~/.gitconfig. */
660 		int dummy_repo_version;
661 		err = parse_gitconfig_file(&dummy_repo_version,
662 		    &repo->global_gitconfig_author_name,
663 		    &repo->global_gitconfig_author_email,
664 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
665 		if (err)
666 			return err;
667 	}
668 
669 	/* Read repository's .git/config file. */
670 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
671 	if (repo_gitconfig_path == NULL)
672 		return got_error_from_errno("got_repo_get_path_gitconfig");
673 
674 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
675 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
676 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
677 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
678 	    repo_gitconfig_path);
679 	if (err)
680 		goto done;
681 
682 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
683 		int i;
684 
685 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
686 			got_repo_free_remote_repo_data(
687 			    &repo->gitconfig_remotes[i]);
688 		}
689 		free(repo->gitconfig_remotes);
690 		repo->gitconfig_remotes = NULL;
691 		repo->ngitconfig_remotes = 0;
692 
693 		free(repo->gitconfig_author_name);
694 		repo->gitconfig_author_name = NULL;
695 		free(repo->gitconfig_author_email);
696 		repo->gitconfig_author_email = NULL;
697 
698 		free(repo->global_gitconfig_author_name);
699 		repo->global_gitconfig_author_name = NULL;
700 		free(repo->global_gitconfig_author_email);
701 		repo->global_gitconfig_author_email = NULL;
702 	}
703 
704 done:
705 	free(repo_gitconfig_path);
706 	return err;
707 }
708 
709 static const struct got_error *
710 read_gotconfig(struct got_repository *repo)
711 {
712 	const struct got_error *err = NULL;
713 	char *gotconfig_path;
714 
715 	gotconfig_path = got_repo_get_path_gotconfig(repo);
716 	if (gotconfig_path == NULL)
717 		return got_error_from_errno("got_repo_get_path_gotconfig");
718 
719 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
720 	free(gotconfig_path);
721 	return err;
722 }
723 
724 /* Supported repository format extensions. */
725 static const char *const repo_extensions[] = {
726 	"noop",			/* Got supports repository format version 1. */
727 	"preciousObjects",	/* Supported by gotadmin cleanup. */
728 	"worktreeConfig",	/* Got does not care about Git work trees. */
729 };
730 
731 const struct got_error *
732 got_repo_open(struct got_repository **repop, const char *path,
733     const char *global_gitconfig_path, int *pack_fds)
734 {
735 	struct got_repository *repo = NULL;
736 	const struct got_error *err = NULL;
737 	char *repo_path = NULL;
738 	size_t i, j = 0;
739 	struct rlimit rl;
740 
741 	*repop = NULL;
742 
743 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
744 		return got_error_from_errno("getrlimit");
745 
746 	repo = calloc(1, sizeof(*repo));
747 	if (repo == NULL)
748 		return got_error_from_errno("calloc");
749 
750 	RB_INIT(&repo->packidx_bloom_filters);
751 	TAILQ_INIT(&repo->packidx_paths);
752 
753 	for (i = 0; i < nitems(repo->privsep_children); i++) {
754 		memset(&repo->privsep_children[i], 0,
755 		    sizeof(repo->privsep_children[0]));
756 		repo->privsep_children[i].imsg_fd = -1;
757 	}
758 
759 	err = got_object_cache_init(&repo->objcache,
760 	    GOT_OBJECT_CACHE_TYPE_OBJ);
761 	if (err)
762 		goto done;
763 	err = got_object_cache_init(&repo->treecache,
764 	    GOT_OBJECT_CACHE_TYPE_TREE);
765 	if (err)
766 		goto done;
767 	err = got_object_cache_init(&repo->commitcache,
768 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
769 	if (err)
770 		goto done;
771 	err = got_object_cache_init(&repo->tagcache,
772 	    GOT_OBJECT_CACHE_TYPE_TAG);
773 	if (err)
774 		goto done;
775 	err = got_object_cache_init(&repo->rawcache,
776 	    GOT_OBJECT_CACHE_TYPE_RAW);
777 	if (err)
778 		goto done;
779 
780 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
781 	if (repo->pack_cache_size > rl.rlim_cur / 8)
782 		repo->pack_cache_size = rl.rlim_cur / 8;
783 	for (i = 0; i < nitems(repo->packs); i++) {
784 		if (i < repo->pack_cache_size) {
785 			repo->packs[i].basefd = pack_fds[j++];
786 			repo->packs[i].accumfd = pack_fds[j++];
787 		} else {
788 			repo->packs[i].basefd = -1;
789 			repo->packs[i].accumfd = -1;
790 		}
791 	}
792 	repo->pinned_pack = -1;
793 	repo->pinned_packidx = -1;
794 	repo->pinned_pid = 0;
795 
796 	repo_path = realpath(path, NULL);
797 	if (repo_path == NULL) {
798 		err = got_error_from_errno2("realpath", path);
799 		goto done;
800 	}
801 
802 	for (;;) {
803 		char *parent_path;
804 
805 		err = open_repo(repo, repo_path);
806 		if (err == NULL)
807 			break;
808 		if (err->code != GOT_ERR_NOT_GIT_REPO)
809 			goto done;
810 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
811 			err = got_error(GOT_ERR_NOT_GIT_REPO);
812 			goto done;
813 		}
814 		err = got_path_dirname(&parent_path, repo_path);
815 		if (err)
816 			goto done;
817 		free(repo_path);
818 		repo_path = parent_path;
819 	}
820 
821 	err = read_gotconfig(repo);
822 	if (err)
823 		goto done;
824 
825 	err = read_gitconfig(repo, global_gitconfig_path);
826 	if (err)
827 		goto done;
828 	if (repo->gitconfig_repository_format_version != 0) {
829 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
830 		goto done;
831 	}
832 	for (i = 0; i < repo->nextensions; i++) {
833 		char *ext = repo->extensions[i];
834 		int j, supported = 0;
835 		for (j = 0; j < nitems(repo_extensions); j++) {
836 			if (strcmp(ext, repo_extensions[j]) == 0) {
837 				supported = 1;
838 				break;
839 			}
840 		}
841 		if (!supported) {
842 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
843 			goto done;
844 		}
845 	}
846 
847 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
848 done:
849 	if (err)
850 		got_repo_close(repo);
851 	else
852 		*repop = repo;
853 	free(repo_path);
854 	return err;
855 }
856 
857 const struct got_error *
858 got_repo_close(struct got_repository *repo)
859 {
860 	const struct got_error *err = NULL, *child_err;
861 	struct got_packidx_bloom_filter *bf;
862 	struct got_pathlist_entry *pe;
863 	size_t i;
864 
865 	for (i = 0; i < repo->pack_cache_size; i++) {
866 		if (repo->packidx_cache[i] == NULL)
867 			break;
868 		got_packidx_close(repo->packidx_cache[i]);
869 	}
870 
871 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
872 	    &repo->packidx_bloom_filters))) {
873 		RB_REMOVE(got_packidx_bloom_filter_tree,
874 		    &repo->packidx_bloom_filters, bf);
875 		bloom_free(bf->bloom);
876 		free(bf->bloom);
877 		free(bf);
878 	}
879 
880 	for (i = 0; i < repo->pack_cache_size; i++)
881 		if (repo->packs[i].path_packfile)
882 			if (repo->packs[i].path_packfile)
883 				got_pack_close(&repo->packs[i]);
884 
885 	free(repo->path);
886 	free(repo->path_git_dir);
887 
888 	got_object_cache_close(&repo->objcache);
889 	got_object_cache_close(&repo->treecache);
890 	got_object_cache_close(&repo->commitcache);
891 	got_object_cache_close(&repo->tagcache);
892 	got_object_cache_close(&repo->rawcache);
893 
894 	for (i = 0; i < nitems(repo->privsep_children); i++) {
895 		if (repo->privsep_children[i].imsg_fd == -1)
896 			continue;
897 		imsg_clear(repo->privsep_children[i].ibuf);
898 		free(repo->privsep_children[i].ibuf);
899 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
900 		child_err = got_privsep_wait_for_child(
901 		    repo->privsep_children[i].pid);
902 		if (child_err && err == NULL)
903 			err = child_err;
904 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
905 		    err == NULL)
906 			err = got_error_from_errno("close");
907 	}
908 
909 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
910 	    err == NULL)
911 		err = got_error_from_errno("close");
912 
913 	if (repo->gotconfig)
914 		got_gotconfig_free(repo->gotconfig);
915 	free(repo->gitconfig_author_name);
916 	free(repo->gitconfig_author_email);
917 	for (i = 0; i < repo->ngitconfig_remotes; i++)
918 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
919 	free(repo->gitconfig_remotes);
920 	for (i = 0; i < repo->nextensions; i++)
921 		free(repo->extensions[i]);
922 	free(repo->extensions);
923 
924 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
925 		free((void *)pe->path);
926 	got_pathlist_free(&repo->packidx_paths);
927 	free(repo);
928 
929 	return err;
930 }
931 
932 void
933 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
934 {
935 	int i;
936 
937 	free(repo->name);
938 	repo->name = NULL;
939 	free(repo->fetch_url);
940 	repo->fetch_url = NULL;
941 	free(repo->send_url);
942 	repo->send_url = NULL;
943 	for (i = 0; i < repo->nfetch_branches; i++)
944 		free(repo->fetch_branches[i]);
945 	free(repo->fetch_branches);
946 	repo->fetch_branches = NULL;
947 	repo->nfetch_branches = 0;
948 	for (i = 0; i < repo->nsend_branches; i++)
949 		free(repo->send_branches[i]);
950 	free(repo->send_branches);
951 	repo->send_branches = NULL;
952 	repo->nsend_branches = 0;
953 }
954 
955 const struct got_error *
956 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
957     const char *input_path)
958 {
959 	const struct got_error *err = NULL;
960 	const char *repo_abspath = NULL;
961 	size_t repolen, len;
962 	char *canonpath, *path = NULL;
963 
964 	*in_repo_path = NULL;
965 
966 	canonpath = strdup(input_path);
967 	if (canonpath == NULL) {
968 		err = got_error_from_errno("strdup");
969 		goto done;
970 	}
971 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
972 	if (err)
973 		goto done;
974 
975 	repo_abspath = got_repo_get_path(repo);
976 
977 	if (canonpath[0] == '\0') {
978 		path = strdup(canonpath);
979 		if (path == NULL) {
980 			err = got_error_from_errno("strdup");
981 			goto done;
982 		}
983 	} else {
984 		path = realpath(canonpath, NULL);
985 		if (path == NULL) {
986 			if (errno != ENOENT) {
987 				err = got_error_from_errno2("realpath",
988 				    canonpath);
989 				goto done;
990 			}
991 			/*
992 			 * Path is not on disk.
993 			 * Assume it is already relative to repository root.
994 			 */
995 			path = strdup(canonpath);
996 			if (path == NULL) {
997 				err = got_error_from_errno("strdup");
998 				goto done;
999 			}
1000 		}
1001 
1002 		repolen = strlen(repo_abspath);
1003 		len = strlen(path);
1004 
1005 
1006 		if (strcmp(path, repo_abspath) == 0) {
1007 			free(path);
1008 			path = strdup("");
1009 			if (path == NULL) {
1010 				err = got_error_from_errno("strdup");
1011 				goto done;
1012 			}
1013 		} else if (len > repolen &&
1014 		    got_path_is_child(path, repo_abspath, repolen)) {
1015 			/* Matched an on-disk path inside repository. */
1016 			if (got_repo_is_bare(repo)) {
1017 				/*
1018 				 * Matched an on-disk path inside repository
1019 				 * database. Treat input as repository-relative.
1020 				 */
1021 				free(path);
1022 				path = canonpath;
1023 				canonpath = NULL;
1024 			} else {
1025 				char *child;
1026 				/* Strip common prefix with repository path. */
1027 				err = got_path_skip_common_ancestor(&child,
1028 				    repo_abspath, path);
1029 				if (err)
1030 					goto done;
1031 				free(path);
1032 				path = child;
1033 			}
1034 		} else {
1035 			/*
1036 			 * Matched unrelated on-disk path.
1037 			 * Treat input as repository-relative.
1038 			 */
1039 			free(path);
1040 			path = canonpath;
1041 			canonpath = NULL;
1042 		}
1043 	}
1044 
1045 	/* Make in-repository path absolute */
1046 	if (path[0] != '/') {
1047 		char *abspath;
1048 		if (asprintf(&abspath, "/%s", path) == -1) {
1049 			err = got_error_from_errno("asprintf");
1050 			goto done;
1051 		}
1052 		free(path);
1053 		path = abspath;
1054 	}
1055 
1056 done:
1057 	free(canonpath);
1058 	if (err)
1059 		free(path);
1060 	else
1061 		*in_repo_path = path;
1062 	return err;
1063 }
1064 
1065 static const struct got_error *
1066 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1067     const char *path_packidx)
1068 {
1069 	const struct got_error *err = NULL;
1070 	size_t i;
1071 
1072 	for (i = 0; i < repo->pack_cache_size; i++) {
1073 		if (repo->packidx_cache[i] == NULL)
1074 			break;
1075 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1076 		    path_packidx) == 0) {
1077 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1078 		}
1079 	}
1080 	if (i == repo->pack_cache_size) {
1081 		do {
1082 			i--;
1083 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1084 		    i == repo->pinned_packidx);
1085 		err = got_packidx_close(repo->packidx_cache[i]);
1086 		if (err)
1087 			return err;
1088 	}
1089 
1090 	repo->packidx_cache[i] = packidx;
1091 
1092 	return NULL;
1093 }
1094 
1095 int
1096 got_repo_is_packidx_filename(const char *name, size_t len)
1097 {
1098 	if (len != GOT_PACKIDX_NAMELEN)
1099 		return 0;
1100 
1101 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1102 		return 0;
1103 
1104 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1105 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1106 		return 0;
1107 
1108 	return 1;
1109 }
1110 
1111 static struct got_packidx_bloom_filter *
1112 get_packidx_bloom_filter(struct got_repository *repo,
1113     const char *path, size_t path_len)
1114 {
1115 	struct got_packidx_bloom_filter key;
1116 
1117 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1118 		return NULL; /* XXX */
1119 	key.path_len = path_len;
1120 
1121 	return RB_FIND(got_packidx_bloom_filter_tree,
1122 	    &repo->packidx_bloom_filters, &key);
1123 }
1124 
1125 int
1126 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1127     const char *path_packidx, struct got_object_id *id)
1128 {
1129 	struct got_packidx_bloom_filter *bf;
1130 
1131 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1132 	if (bf)
1133 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1134 
1135 	/* No bloom filter means this pack index must be searched. */
1136 	return 1;
1137 }
1138 
1139 static const struct got_error *
1140 add_packidx_bloom_filter(struct got_repository *repo,
1141     struct got_packidx *packidx, const char *path_packidx)
1142 {
1143 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1144 	struct got_packidx_bloom_filter *bf;
1145 	size_t len;
1146 
1147 	/*
1148 	 * Don't use bloom filters for very large pack index files.
1149 	 * Large pack files will contain a relatively large fraction
1150 	 * of our objects so we will likely need to visit them anyway.
1151 	 * The more objects a pack file contains the higher the probability
1152 	 * of a false-positive match from the bloom filter. And reading
1153 	 * all object IDs from a large pack index file can be expensive.
1154 	 */
1155 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1156 		return NULL;
1157 
1158 	/* Do we already have a filter for this pack index? */
1159 	if (get_packidx_bloom_filter(repo, path_packidx,
1160 	    strlen(path_packidx)) != NULL)
1161 		return NULL;
1162 
1163 	bf = calloc(1, sizeof(*bf));
1164 	if (bf == NULL)
1165 		return got_error_from_errno("calloc");
1166 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1167 	if (bf->bloom == NULL) {
1168 		free(bf);
1169 		return got_error_from_errno("calloc");
1170 	}
1171 
1172 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1173 	if (len >= sizeof(bf->path)) {
1174 		free(bf->bloom);
1175 		free(bf);
1176 		return got_error(GOT_ERR_NO_SPACE);
1177 	}
1178 	bf->path_len = len;
1179 
1180 	/* Minimum size supported by our bloom filter is 1000 entries. */
1181 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1182 	for (i = 0; i < nobjects; i++) {
1183 		struct got_packidx_object_id *id;
1184 		id = &packidx->hdr.sorted_ids[i];
1185 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1186 	}
1187 
1188 	RB_INSERT(got_packidx_bloom_filter_tree,
1189 	    &repo->packidx_bloom_filters, bf);
1190 	return NULL;
1191 }
1192 
1193 const struct got_error *
1194 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1195     struct got_repository *repo, struct got_object_id *id)
1196 {
1197 	const struct got_error *err;
1198 	struct got_pathlist_entry *pe;
1199 	size_t i;
1200 
1201 	/* Search pack index cache. */
1202 	for (i = 0; i < repo->pack_cache_size; i++) {
1203 		if (repo->packidx_cache[i] == NULL)
1204 			break;
1205 		if (!got_repo_check_packidx_bloom_filter(repo,
1206 		    repo->packidx_cache[i]->path_packidx, id))
1207 			continue; /* object will not be found in this index */
1208 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1209 		if (*idx != -1) {
1210 			*packidx = repo->packidx_cache[i];
1211 			/*
1212 			 * Move this cache entry to the front. Repeatedly
1213 			 * searching a wrong pack index can be expensive.
1214 			 */
1215 			if (i > 0) {
1216 				memmove(&repo->packidx_cache[1],
1217 				    &repo->packidx_cache[0],
1218 				    i * sizeof(repo->packidx_cache[0]));
1219 				repo->packidx_cache[0] = *packidx;
1220 				if (repo->pinned_packidx >= 0 &&
1221 				    repo->pinned_packidx < i)
1222 					repo->pinned_packidx++;
1223 				else if (repo->pinned_packidx == i)
1224 					repo->pinned_packidx = 0;
1225 			}
1226 			return NULL;
1227 		}
1228 	}
1229 	/* No luck. Search the filesystem. */
1230 
1231 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1232 		const char *path_packidx = pe->path;
1233 		int is_cached = 0;
1234 
1235 		if (!got_repo_check_packidx_bloom_filter(repo,
1236 		    pe->path, id))
1237 			continue; /* object will not be found in this index */
1238 
1239 		for (i = 0; i < repo->pack_cache_size; i++) {
1240 			if (repo->packidx_cache[i] == NULL)
1241 				break;
1242 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1243 			    path_packidx) == 0) {
1244 				is_cached = 1;
1245 				break;
1246 			}
1247 		}
1248 		if (is_cached)
1249 			continue; /* already searched */
1250 
1251 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1252 		    path_packidx, 0);
1253 		if (err)
1254 			goto done;
1255 
1256 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1257 		if (err)
1258 			goto done;
1259 
1260 		err = cache_packidx(repo, *packidx, path_packidx);
1261 		if (err)
1262 			goto done;
1263 
1264 		*idx = got_packidx_get_object_idx(*packidx, id);
1265 		if (*idx != -1) {
1266 			err = NULL; /* found the object */
1267 			goto done;
1268 		}
1269 	}
1270 
1271 	err = got_error_no_obj(id);
1272 done:
1273 	return err;
1274 }
1275 
1276 const struct got_error *
1277 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1278     struct got_repository *repo)
1279 {
1280 	const struct got_error *err = NULL;
1281 	DIR *packdir = NULL;
1282 	struct dirent *dent;
1283 	char *path_packidx = NULL;
1284 	int packdir_fd;
1285 
1286 	packdir_fd = openat(got_repo_get_fd(repo),
1287 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1288 	if (packdir_fd == -1) {
1289 		return got_error_from_errno_fmt("openat: %s/%s",
1290 		    got_repo_get_path_git_dir(repo),
1291 		    GOT_OBJECTS_PACK_DIR);
1292 	}
1293 
1294 	packdir = fdopendir(packdir_fd);
1295 	if (packdir == NULL) {
1296 		err = got_error_from_errno("fdopendir");
1297 		goto done;
1298 	}
1299 
1300 	while ((dent = readdir(packdir)) != NULL) {
1301 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1302 			continue;
1303 
1304 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1305 		    dent->d_name) == -1) {
1306 			err = got_error_from_errno("asprintf");
1307 			path_packidx = NULL;
1308 			break;
1309 		}
1310 
1311 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1312 		if (err)
1313 			break;
1314 	}
1315 done:
1316 	if (err)
1317 		free(path_packidx);
1318 	if (packdir && closedir(packdir) != 0 && err == NULL)
1319 		err = got_error_from_errno("closedir");
1320 	return err;
1321 }
1322 
1323 const struct got_error *
1324 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1325     struct got_repository *repo)
1326 {
1327 	const struct got_error *err;
1328 	size_t i;
1329 
1330 	*packidx = NULL;
1331 
1332 	/* Search pack index cache. */
1333 	for (i = 0; i < repo->pack_cache_size; i++) {
1334 		if (repo->packidx_cache[i] == NULL)
1335 			break;
1336 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1337 		    path_packidx) == 0) {
1338 			*packidx = repo->packidx_cache[i];
1339 			return NULL;
1340 		}
1341 	}
1342 	/* No luck. Search the filesystem. */
1343 
1344 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1345 	    path_packidx, 0);
1346 	if (err)
1347 		return err;
1348 
1349 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1350 	if (err)
1351 		goto done;
1352 
1353 	err = cache_packidx(repo, *packidx, path_packidx);
1354 done:
1355 	if (err) {
1356 		got_packidx_close(*packidx);
1357 		*packidx = NULL;
1358 	}
1359 	return err;
1360 }
1361 
1362 static const struct got_error *
1363 read_packfile_hdr(int fd, struct got_packidx *packidx)
1364 {
1365 	const struct got_error *err = NULL;
1366 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1367 	struct got_packfile_hdr hdr;
1368 	ssize_t n;
1369 
1370 	n = read(fd, &hdr, sizeof(hdr));
1371 	if (n < 0)
1372 		return got_error_from_errno("read");
1373 	if (n != sizeof(hdr))
1374 		return got_error(GOT_ERR_BAD_PACKFILE);
1375 
1376 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1377 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1378 	    be32toh(hdr.nobjects) != totobj)
1379 		err = got_error(GOT_ERR_BAD_PACKFILE);
1380 
1381 	return err;
1382 }
1383 
1384 static const struct got_error *
1385 open_packfile(int *fd, struct got_repository *repo,
1386     const char *relpath, struct got_packidx *packidx)
1387 {
1388 	const struct got_error *err = NULL;
1389 
1390 	*fd = openat(got_repo_get_fd(repo), relpath,
1391 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1392 	if (*fd == -1)
1393 		return got_error_from_errno_fmt("openat: %s/%s",
1394 		    got_repo_get_path_git_dir(repo), relpath);
1395 
1396 	if (packidx) {
1397 		err = read_packfile_hdr(*fd, packidx);
1398 		if (err) {
1399 			close(*fd);
1400 			*fd = -1;
1401 		}
1402 	}
1403 
1404 	return err;
1405 }
1406 
1407 const struct got_error *
1408 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1409     const char *path_packfile, struct got_packidx *packidx)
1410 {
1411 	const struct got_error *err = NULL;
1412 	struct got_pack *pack = NULL;
1413 	struct stat sb;
1414 	size_t i;
1415 
1416 	if (packp)
1417 		*packp = NULL;
1418 
1419 	for (i = 0; i < repo->pack_cache_size; i++) {
1420 		pack = &repo->packs[i];
1421 		if (pack->path_packfile == NULL)
1422 			break;
1423 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1424 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1425 	}
1426 
1427 	if (i == repo->pack_cache_size) {
1428 		struct got_pack tmp;
1429 		do {
1430 			i--;
1431 		} while (i > 0 && repo->pinned_pack >= 0 &&
1432 		    i == repo->pinned_pack);
1433 		err = got_pack_close(&repo->packs[i]);
1434 		if (err)
1435 			return err;
1436 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1437 			return got_error_from_errno("ftruncate");
1438 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1439 			return got_error_from_errno("ftruncate");
1440 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1441 		memcpy(&repo->packs[i], &repo->packs[0],
1442 		    sizeof(repo->packs[i]));
1443 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1444 		if (repo->pinned_pack == 0)
1445 			repo->pinned_pack = i;
1446 		else if (repo->pinned_pack == i)
1447 			repo->pinned_pack = 0;
1448 		i = 0;
1449 	}
1450 
1451 	pack = &repo->packs[i];
1452 
1453 	pack->path_packfile = strdup(path_packfile);
1454 	if (pack->path_packfile == NULL) {
1455 		err = got_error_from_errno("strdup");
1456 		goto done;
1457 	}
1458 
1459 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1460 	if (err)
1461 		goto done;
1462 
1463 	if (fstat(pack->fd, &sb) != 0) {
1464 		err = got_error_from_errno("fstat");
1465 		goto done;
1466 	}
1467 	pack->filesize = sb.st_size;
1468 
1469 	pack->privsep_child = NULL;
1470 
1471 #ifndef GOT_PACK_NO_MMAP
1472 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1473 	    pack->fd, 0);
1474 	if (pack->map == MAP_FAILED) {
1475 		if (errno != ENOMEM) {
1476 			err = got_error_from_errno("mmap");
1477 			goto done;
1478 		}
1479 		pack->map = NULL; /* fall back to read(2) */
1480 	}
1481 #endif
1482 done:
1483 	if (err) {
1484 		if (pack) {
1485 			free(pack->path_packfile);
1486 			memset(pack, 0, sizeof(*pack));
1487 		}
1488 	} else if (packp)
1489 		*packp = pack;
1490 	return err;
1491 }
1492 
1493 struct got_pack *
1494 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1495 {
1496 	struct got_pack *pack = NULL;
1497 	size_t i;
1498 
1499 	for (i = 0; i < repo->pack_cache_size; i++) {
1500 		pack = &repo->packs[i];
1501 		if (pack->path_packfile == NULL)
1502 			break;
1503 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1504 			return pack;
1505 	}
1506 
1507 	return NULL;
1508 }
1509 
1510 const struct got_error *
1511 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1512     struct got_pack *pack)
1513 {
1514 	size_t i;
1515 	int pinned_pack = -1, pinned_packidx = -1;
1516 
1517 	for (i = 0; i < repo->pack_cache_size; i++) {
1518 		if (repo->packidx_cache[i] &&
1519 		    strcmp(repo->packidx_cache[i]->path_packidx,
1520 		    packidx->path_packidx) == 0)
1521 			pinned_packidx = i;
1522 		if (repo->packs[i].path_packfile &&
1523 		    strcmp(repo->packs[i].path_packfile,
1524 		    pack->path_packfile) == 0)
1525 			pinned_pack = i;
1526 	}
1527 
1528 	if (pinned_packidx == -1 || pinned_pack == -1)
1529 		return got_error(GOT_ERR_PIN_PACK);
1530 
1531 	repo->pinned_pack = pinned_pack;
1532 	repo->pinned_packidx = pinned_packidx;
1533 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1534 	return NULL;
1535 }
1536 
1537 struct got_pack *
1538 got_repo_get_pinned_pack(struct got_repository *repo)
1539 {
1540 	if (repo->pinned_pack >= 0 &&
1541 	    repo->pinned_pack < repo->pack_cache_size)
1542 		return &repo->packs[repo->pinned_pack];
1543 
1544 	return NULL;
1545 }
1546 
1547 void
1548 got_repo_unpin_pack(struct got_repository *repo)
1549 {
1550 	repo->pinned_packidx = -1;
1551 	repo->pinned_pack = -1;
1552 	repo->pinned_pid = 0;
1553 }
1554 
1555 const struct got_error *
1556 got_repo_init(const char *repo_path)
1557 {
1558 	const struct got_error *err = NULL;
1559 	const char *dirnames[] = {
1560 		GOT_OBJECTS_DIR,
1561 		GOT_OBJECTS_PACK_DIR,
1562 		GOT_REFS_DIR,
1563 	};
1564 	const char *description_str = "Unnamed repository; "
1565 	    "edit this file 'description' to name the repository.";
1566 	const char *headref_str = "ref: refs/heads/main";
1567 	const char *gitconfig_str = "[core]\n"
1568 	    "\trepositoryformatversion = 0\n"
1569 	    "\tfilemode = true\n"
1570 	    "\tbare = true\n";
1571 	char *path;
1572 	size_t i;
1573 
1574 	if (!got_path_dir_is_empty(repo_path))
1575 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1576 
1577 	for (i = 0; i < nitems(dirnames); i++) {
1578 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1579 			return got_error_from_errno("asprintf");
1580 		}
1581 		err = got_path_mkdir(path);
1582 		free(path);
1583 		if (err)
1584 			return err;
1585 	}
1586 
1587 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1588 		return got_error_from_errno("asprintf");
1589 	err = got_path_create_file(path, description_str);
1590 	free(path);
1591 	if (err)
1592 		return err;
1593 
1594 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1595 		return got_error_from_errno("asprintf");
1596 	err = got_path_create_file(path, headref_str);
1597 	free(path);
1598 	if (err)
1599 		return err;
1600 
1601 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1602 		return got_error_from_errno("asprintf");
1603 	err = got_path_create_file(path, gitconfig_str);
1604 	free(path);
1605 	if (err)
1606 		return err;
1607 
1608 	return NULL;
1609 }
1610 
1611 static const struct got_error *
1612 match_packed_object(struct got_object_id **unique_id,
1613     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1614 {
1615 	const struct got_error *err = NULL;
1616 	struct got_object_id_queue matched_ids;
1617 	struct got_pathlist_entry *pe;
1618 
1619 	STAILQ_INIT(&matched_ids);
1620 
1621 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1622 		const char *path_packidx = pe->path;
1623 		struct got_packidx *packidx;
1624 		struct got_object_qid *qid;
1625 
1626 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1627 		    path_packidx, 0);
1628 		if (err)
1629 			break;
1630 
1631 		err = got_packidx_match_id_str_prefix(&matched_ids,
1632 		    packidx, id_str_prefix);
1633 		if (err) {
1634 			got_packidx_close(packidx);
1635 			break;
1636 		}
1637 		err = got_packidx_close(packidx);
1638 		if (err)
1639 			break;
1640 
1641 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1642 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1643 				int matched_type;
1644 				err = got_object_get_type(&matched_type, repo,
1645 				    &qid->id);
1646 				if (err)
1647 					goto done;
1648 				if (matched_type != obj_type)
1649 					continue;
1650 			}
1651 			if (*unique_id == NULL) {
1652 				*unique_id = got_object_id_dup(&qid->id);
1653 				if (*unique_id == NULL) {
1654 					err = got_error_from_errno("malloc");
1655 					goto done;
1656 				}
1657 			} else {
1658 				if (got_object_id_cmp(*unique_id,
1659 				    &qid->id) == 0)
1660 					continue; /* packed multiple times */
1661 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1662 				goto done;
1663 			}
1664 		}
1665 	}
1666 done:
1667 	got_object_id_queue_free(&matched_ids);
1668 	if (err) {
1669 		free(*unique_id);
1670 		*unique_id = NULL;
1671 	}
1672 	return err;
1673 }
1674 
1675 static const struct got_error *
1676 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1677     const char *object_dir, const char *id_str_prefix, int obj_type,
1678     struct got_repository *repo)
1679 {
1680 	const struct got_error *err = NULL;
1681 	char *path;
1682 	DIR *dir = NULL;
1683 	struct dirent *dent;
1684 	struct got_object_id id;
1685 
1686 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1687 		err = got_error_from_errno("asprintf");
1688 		goto done;
1689 	}
1690 
1691 	dir = opendir(path);
1692 	if (dir == NULL) {
1693 		if (errno == ENOENT) {
1694 			err = NULL;
1695 			goto done;
1696 		}
1697 		err = got_error_from_errno2("opendir", path);
1698 		goto done;
1699 	}
1700 	while ((dent = readdir(dir)) != NULL) {
1701 		char *id_str;
1702 		int cmp;
1703 
1704 		if (strcmp(dent->d_name, ".") == 0 ||
1705 		    strcmp(dent->d_name, "..") == 0)
1706 			continue;
1707 
1708 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1709 			err = got_error_from_errno("asprintf");
1710 			goto done;
1711 		}
1712 
1713 		if (!got_parse_sha1_digest(id.sha1, id_str))
1714 			continue;
1715 
1716 		/*
1717 		 * Directory entries do not necessarily appear in
1718 		 * sorted order, so we must iterate over all of them.
1719 		 */
1720 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1721 		if (cmp != 0) {
1722 			free(id_str);
1723 			continue;
1724 		}
1725 
1726 		if (*unique_id == NULL) {
1727 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1728 				int matched_type;
1729 				err = got_object_get_type(&matched_type, repo,
1730 				    &id);
1731 				if (err)
1732 					goto done;
1733 				if (matched_type != obj_type)
1734 					continue;
1735 			}
1736 			*unique_id = got_object_id_dup(&id);
1737 			if (*unique_id == NULL) {
1738 				err = got_error_from_errno("got_object_id_dup");
1739 				free(id_str);
1740 				goto done;
1741 			}
1742 		} else {
1743 			if (got_object_id_cmp(*unique_id, &id) == 0)
1744 				continue; /* both packed and loose */
1745 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1746 			free(id_str);
1747 			goto done;
1748 		}
1749 	}
1750 done:
1751 	if (dir && closedir(dir) != 0 && err == NULL)
1752 		err = got_error_from_errno("closedir");
1753 	if (err) {
1754 		free(*unique_id);
1755 		*unique_id = NULL;
1756 	}
1757 	free(path);
1758 	return err;
1759 }
1760 
1761 const struct got_error *
1762 got_repo_match_object_id_prefix(struct got_object_id **id,
1763     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1764 {
1765 	const struct got_error *err = NULL;
1766 	char *path_objects = got_repo_get_path_objects(repo);
1767 	char *object_dir = NULL;
1768 	size_t len;
1769 	int i;
1770 
1771 	*id = NULL;
1772 
1773 	len = strlen(id_str_prefix);
1774 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1775 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1776 
1777 	for (i = 0; i < len; i++) {
1778 		if (isxdigit((unsigned char)id_str_prefix[i]))
1779 			continue;
1780 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1781 	}
1782 
1783 	if (len >= 2) {
1784 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1785 		if (err)
1786 			goto done;
1787 		object_dir = strndup(id_str_prefix, 2);
1788 		if (object_dir == NULL) {
1789 			err = got_error_from_errno("strdup");
1790 			goto done;
1791 		}
1792 		err = match_loose_object(id, path_objects, object_dir,
1793 		    id_str_prefix, obj_type, repo);
1794 	} else if (len == 1) {
1795 		int i;
1796 		for (i = 0; i < 0xf; i++) {
1797 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1798 			    == -1) {
1799 				err = got_error_from_errno("asprintf");
1800 				goto done;
1801 			}
1802 			err = match_packed_object(id, repo, object_dir,
1803 			    obj_type);
1804 			if (err)
1805 				goto done;
1806 			err = match_loose_object(id, path_objects, object_dir,
1807 			    id_str_prefix, obj_type, repo);
1808 			if (err)
1809 				goto done;
1810 		}
1811 	} else {
1812 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1813 		goto done;
1814 	}
1815 done:
1816 	free(object_dir);
1817 	if (err) {
1818 		free(*id);
1819 		*id = NULL;
1820 	} else if (*id == NULL) {
1821 		switch (obj_type) {
1822 		case GOT_OBJ_TYPE_BLOB:
1823 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1824 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1825 			break;
1826 		case GOT_OBJ_TYPE_TREE:
1827 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1828 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1829 			break;
1830 		case GOT_OBJ_TYPE_COMMIT:
1831 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1832 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1833 			break;
1834 		case GOT_OBJ_TYPE_TAG:
1835 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1836 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1837 			break;
1838 		default:
1839 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1840 			break;
1841 		}
1842 	}
1843 
1844 	return err;
1845 }
1846 
1847 const struct got_error *
1848 got_repo_match_object_id(struct got_object_id **id, char **label,
1849     const char *id_str, int obj_type, struct got_reflist_head *refs,
1850     struct got_repository *repo)
1851 {
1852 	const struct got_error *err;
1853 	struct got_tag_object *tag;
1854 	struct got_reference *ref = NULL;
1855 
1856 	*id = NULL;
1857 	if (label)
1858 		*label = NULL;
1859 
1860 	if (refs) {
1861 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1862 		    refs, repo);
1863 		if (err == NULL) {
1864 			*id = got_object_id_dup(
1865 			    got_object_tag_get_object_id(tag));
1866 			if (*id == NULL)
1867 				err = got_error_from_errno("got_object_id_dup");
1868 			else if (label && asprintf(label, "refs/tags/%s",
1869 			    got_object_tag_get_name(tag)) == -1) {
1870 				err = got_error_from_errno("asprintf");
1871 				free(*id);
1872 				*id = NULL;
1873 			}
1874 			got_object_tag_close(tag);
1875 			return err;
1876 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1877 		    err->code != GOT_ERR_NO_OBJ)
1878 			return err;
1879 	}
1880 
1881 	err = got_ref_open(&ref, repo, id_str, 0);
1882 	if (err == NULL) {
1883 		err = got_ref_resolve(id, repo, ref);
1884 		if (err)
1885 			goto done;
1886 		if (label) {
1887 			*label = strdup(got_ref_get_name(ref));
1888 			if (*label == NULL) {
1889 				err = got_error_from_errno("strdup");
1890 				goto done;
1891 			}
1892 		}
1893 	} else {
1894 		if (err->code != GOT_ERR_NOT_REF &&
1895 		    err->code != GOT_ERR_BAD_REF_NAME)
1896 			goto done;
1897 		err = got_repo_match_object_id_prefix(id, id_str,
1898 		    obj_type, repo);
1899 		if (err) {
1900 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1901 				err = got_error_not_ref(id_str);
1902 			goto done;
1903 		}
1904 		if (label) {
1905 			err = got_object_id_str(label, *id);
1906 			if (*label == NULL) {
1907 				err = got_error_from_errno("strdup");
1908 				goto done;
1909 			}
1910 		}
1911 	}
1912 done:
1913 	if (ref)
1914 		got_ref_close(ref);
1915 	return err;
1916 }
1917 
1918 const struct got_error *
1919 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1920     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1921 {
1922 	const struct got_error *err = NULL;
1923 	struct got_reflist_entry *re;
1924 	struct got_object_id *tag_id;
1925 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1926 
1927 	*tag = NULL;
1928 
1929 	TAILQ_FOREACH(re, refs, entry) {
1930 		const char *refname;
1931 		refname = got_ref_get_name(re->ref);
1932 		if (got_ref_is_symbolic(re->ref))
1933 			continue;
1934 		if (strncmp(refname, "refs/tags/", 10) != 0)
1935 			continue;
1936 		if (!name_is_absolute)
1937 			refname += strlen("refs/tags/");
1938 		if (strcmp(refname, name) != 0)
1939 			continue;
1940 		err = got_ref_resolve(&tag_id, repo, re->ref);
1941 		if (err)
1942 			break;
1943 		err = got_object_open_as_tag(tag, repo, tag_id);
1944 		free(tag_id);
1945 		if (err)
1946 			break;
1947 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1948 		    got_object_tag_get_object_type(*tag) == obj_type)
1949 			break;
1950 		got_object_tag_close(*tag);
1951 		*tag = NULL;
1952 	}
1953 
1954 	if (err == NULL && *tag == NULL)
1955 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1956 		    GOT_OBJ_LABEL_TAG, name);
1957 	return err;
1958 }
1959 
1960 static const struct got_error *
1961 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1962     const char *name, mode_t mode, struct got_object_id *blob_id)
1963 {
1964 	const struct got_error *err = NULL;
1965 
1966 	 *new_te = NULL;
1967 
1968 	*new_te = calloc(1, sizeof(**new_te));
1969 	if (*new_te == NULL)
1970 		return got_error_from_errno("calloc");
1971 
1972 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1973 	    sizeof((*new_te)->name)) {
1974 		err = got_error(GOT_ERR_NO_SPACE);
1975 		goto done;
1976 	}
1977 
1978 	if (S_ISLNK(mode)) {
1979 		(*new_te)->mode = S_IFLNK;
1980 	} else {
1981 		(*new_te)->mode = S_IFREG;
1982 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1983 	}
1984 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1985 done:
1986 	if (err && *new_te) {
1987 		free(*new_te);
1988 		*new_te = NULL;
1989 	}
1990 	return err;
1991 }
1992 
1993 static const struct got_error *
1994 import_file(struct got_tree_entry **new_te, struct dirent *de,
1995     const char *path, struct got_repository *repo)
1996 {
1997 	const struct got_error *err;
1998 	struct got_object_id *blob_id = NULL;
1999 	char *filepath;
2000 	struct stat sb;
2001 
2002 	if (asprintf(&filepath, "%s%s%s", path,
2003 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2004 		return got_error_from_errno("asprintf");
2005 
2006 	if (lstat(filepath, &sb) != 0) {
2007 		err = got_error_from_errno2("lstat", path);
2008 		goto done;
2009 	}
2010 
2011 	err = got_object_blob_create(&blob_id, filepath, repo);
2012 	if (err)
2013 		goto done;
2014 
2015 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2016 	    blob_id);
2017 done:
2018 	free(filepath);
2019 	if (err)
2020 		free(blob_id);
2021 	return err;
2022 }
2023 
2024 static const struct got_error *
2025 insert_tree_entry(struct got_tree_entry *new_te,
2026     struct got_pathlist_head *paths)
2027 {
2028 	const struct got_error *err = NULL;
2029 	struct got_pathlist_entry *new_pe;
2030 
2031 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2032 	if (err)
2033 		return err;
2034 	if (new_pe == NULL)
2035 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2036 	return NULL;
2037 }
2038 
2039 static const struct got_error *write_tree(struct got_object_id **,
2040     const char *, struct got_pathlist_head *, struct got_repository *,
2041     got_repo_import_cb progress_cb, void *progress_arg);
2042 
2043 static const struct got_error *
2044 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2045     const char *path, struct got_pathlist_head *ignores,
2046     struct got_repository *repo,
2047     got_repo_import_cb progress_cb, void *progress_arg)
2048 {
2049 	const struct got_error *err;
2050 	struct got_object_id *id = NULL;
2051 	char *subdirpath;
2052 
2053 	if (asprintf(&subdirpath, "%s%s%s", path,
2054 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2055 		return got_error_from_errno("asprintf");
2056 
2057 	(*new_te) = calloc(1, sizeof(**new_te));
2058 	if (*new_te == NULL)
2059 		return got_error_from_errno("calloc");
2060 	(*new_te)->mode = S_IFDIR;
2061 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2062 	    sizeof((*new_te)->name)) {
2063 		err = got_error(GOT_ERR_NO_SPACE);
2064 		goto done;
2065 	}
2066 	err = write_tree(&id, subdirpath, ignores,  repo,
2067 	    progress_cb, progress_arg);
2068 	if (err)
2069 		goto done;
2070 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2071 
2072 done:
2073 	free(id);
2074 	free(subdirpath);
2075 	if (err) {
2076 		free(*new_te);
2077 		*new_te = NULL;
2078 	}
2079 	return err;
2080 }
2081 
2082 static const struct got_error *
2083 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2084     struct got_pathlist_head *ignores, struct got_repository *repo,
2085     got_repo_import_cb progress_cb, void *progress_arg)
2086 {
2087 	const struct got_error *err = NULL;
2088 	DIR *dir;
2089 	struct dirent *de;
2090 	int nentries;
2091 	struct got_tree_entry *new_te = NULL;
2092 	struct got_pathlist_head paths;
2093 	struct got_pathlist_entry *pe;
2094 
2095 	*new_tree_id = NULL;
2096 
2097 	TAILQ_INIT(&paths);
2098 
2099 	dir = opendir(path_dir);
2100 	if (dir == NULL) {
2101 		err = got_error_from_errno2("opendir", path_dir);
2102 		goto done;
2103 	}
2104 
2105 	nentries = 0;
2106 	while ((de = readdir(dir)) != NULL) {
2107 		int ignore = 0;
2108 		int type;
2109 
2110 		if (strcmp(de->d_name, ".") == 0 ||
2111 		    strcmp(de->d_name, "..") == 0)
2112 			continue;
2113 
2114 		TAILQ_FOREACH(pe, ignores, entry) {
2115 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2116 				ignore = 1;
2117 				break;
2118 			}
2119 		}
2120 		if (ignore)
2121 			continue;
2122 
2123 		err = got_path_dirent_type(&type, path_dir, de);
2124 		if (err)
2125 			goto done;
2126 
2127 		if (type == DT_DIR) {
2128 			err = import_subdir(&new_te, de, path_dir,
2129 			    ignores, repo, progress_cb, progress_arg);
2130 			if (err) {
2131 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2132 					goto done;
2133 				err = NULL;
2134 				continue;
2135 			}
2136 		} else if (type == DT_REG || type == DT_LNK) {
2137 			err = import_file(&new_te, de, path_dir, repo);
2138 			if (err)
2139 				goto done;
2140 		} else
2141 			continue;
2142 
2143 		err = insert_tree_entry(new_te, &paths);
2144 		if (err)
2145 			goto done;
2146 		nentries++;
2147 	}
2148 
2149 	if (TAILQ_EMPTY(&paths)) {
2150 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2151 		    "cannot create tree without any entries");
2152 		goto done;
2153 	}
2154 
2155 	TAILQ_FOREACH(pe, &paths, entry) {
2156 		struct got_tree_entry *te = pe->data;
2157 		char *path;
2158 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2159 			continue;
2160 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2161 			err = got_error_from_errno("asprintf");
2162 			goto done;
2163 		}
2164 		err = (*progress_cb)(progress_arg, path);
2165 		free(path);
2166 		if (err)
2167 			goto done;
2168 	}
2169 
2170 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2171 done:
2172 	if (dir)
2173 		closedir(dir);
2174 	got_pathlist_free(&paths);
2175 	return err;
2176 }
2177 
2178 const struct got_error *
2179 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2180     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2181     struct got_repository *repo, got_repo_import_cb progress_cb,
2182     void *progress_arg)
2183 {
2184 	const struct got_error *err;
2185 	struct got_object_id *new_tree_id;
2186 
2187 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2188 	    progress_cb, progress_arg);
2189 	if (err)
2190 		return err;
2191 
2192 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2193 	    author, time(NULL), author, time(NULL), logmsg, repo);
2194 	free(new_tree_id);
2195 	return err;
2196 }
2197 
2198 const struct got_error *
2199 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2200     struct got_repository *repo)
2201 {
2202 	const struct got_error *err = NULL;
2203 	char *path_objects = NULL, *path = NULL;
2204 	DIR *dir = NULL;
2205 	struct got_object_id id;
2206 	int i;
2207 
2208 	*nobjects = 0;
2209 	*ondisk_size = 0;
2210 
2211 	path_objects = got_repo_get_path_objects(repo);
2212 	if (path_objects == NULL)
2213 		return got_error_from_errno("got_repo_get_path_objects");
2214 
2215 	for (i = 0; i <= 0xff; i++) {
2216 		struct dirent *dent;
2217 
2218 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2219 			err = got_error_from_errno("asprintf");
2220 			break;
2221 		}
2222 
2223 		dir = opendir(path);
2224 		if (dir == NULL) {
2225 			if (errno == ENOENT) {
2226 				err = NULL;
2227 				continue;
2228 			}
2229 			err = got_error_from_errno2("opendir", path);
2230 			break;
2231 		}
2232 
2233 		while ((dent = readdir(dir)) != NULL) {
2234 			char *id_str;
2235 			int fd;
2236 			struct stat sb;
2237 
2238 			if (strcmp(dent->d_name, ".") == 0 ||
2239 			    strcmp(dent->d_name, "..") == 0)
2240 				continue;
2241 
2242 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2243 				err = got_error_from_errno("asprintf");
2244 				goto done;
2245 			}
2246 
2247 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2248 				free(id_str);
2249 				continue;
2250 			}
2251 			free(id_str);
2252 
2253 			err = got_object_open_loose_fd(&fd, &id, repo);
2254 			if (err)
2255 				goto done;
2256 
2257 			if (fstat(fd, &sb) == -1) {
2258 				err = got_error_from_errno("fstat");
2259 				close(fd);
2260 				goto done;
2261 			}
2262 			(*nobjects)++;
2263 			(*ondisk_size) += sb.st_size;
2264 
2265 			if (close(fd) == -1) {
2266 				err = got_error_from_errno("close");
2267 				goto done;
2268 			}
2269 		}
2270 
2271 		if (closedir(dir) != 0) {
2272 			err = got_error_from_errno("closedir");
2273 			goto done;
2274 		}
2275 		dir = NULL;
2276 
2277 		free(path);
2278 		path = NULL;
2279 	}
2280 done:
2281 	if (dir && closedir(dir) != 0 && err == NULL)
2282 		err = got_error_from_errno("closedir");
2283 
2284 	if (err) {
2285 		*nobjects = 0;
2286 		*ondisk_size = 0;
2287 	}
2288 	free(path_objects);
2289 	free(path);
2290 	return err;
2291 }
2292 
2293 const struct got_error *
2294 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2295     off_t *total_packsize, struct got_repository *repo)
2296 {
2297 	const struct got_error *err = NULL;
2298 	DIR *packdir = NULL;
2299 	struct dirent *dent;
2300 	struct got_packidx *packidx = NULL;
2301 	char *path_packidx;
2302 	char *path_packfile;
2303 	int packdir_fd;
2304 	struct stat sb;
2305 
2306 	*npackfiles = 0;
2307 	*nobjects = 0;
2308 	*total_packsize = 0;
2309 
2310 	packdir_fd = openat(got_repo_get_fd(repo),
2311 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2312 	if (packdir_fd == -1) {
2313 		return got_error_from_errno_fmt("openat: %s/%s",
2314 		    got_repo_get_path_git_dir(repo),
2315 		    GOT_OBJECTS_PACK_DIR);
2316 	}
2317 
2318 	packdir = fdopendir(packdir_fd);
2319 	if (packdir == NULL) {
2320 		err = got_error_from_errno("fdopendir");
2321 		goto done;
2322 	}
2323 
2324 	while ((dent = readdir(packdir)) != NULL) {
2325 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2326 			continue;
2327 
2328 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2329 		    dent->d_name) == -1) {
2330 			err = got_error_from_errno("asprintf");
2331 			goto done;
2332 		}
2333 
2334 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2335 		    path_packidx, 0);
2336 		free(path_packidx);
2337 		if (err)
2338 			goto done;
2339 
2340 		if (fstat(packidx->fd, &sb) == -1)
2341 			goto done;
2342 		*total_packsize += sb.st_size;
2343 
2344 		err = got_packidx_get_packfile_path(&path_packfile,
2345 		    packidx->path_packidx);
2346 		if (err)
2347 			goto done;
2348 
2349 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2350 		    0) == -1) {
2351 			free(path_packfile);
2352 			goto done;
2353 		}
2354 		free(path_packfile);
2355 		*total_packsize += sb.st_size;
2356 
2357 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2358 
2359 		(*npackfiles)++;
2360 
2361 		got_packidx_close(packidx);
2362 		packidx = NULL;
2363 	}
2364 done:
2365 	if (packidx)
2366 		got_packidx_close(packidx);
2367 	if (packdir && closedir(packdir) != 0 && err == NULL)
2368 		err = got_error_from_errno("closedir");
2369 	if (err) {
2370 		*npackfiles = 0;
2371 		*nobjects = 0;
2372 		*total_packsize = 0;
2373 	}
2374 	return err;
2375 }
2376 
2377 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2378     got_packidx_bloom_filter_cmp);