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->ngitconfig_remotes = 0;
691 
692 		free(repo->gitconfig_author_name);
693 		repo->gitconfig_author_name = NULL;
694 		free(repo->gitconfig_author_email);
695 		repo->gitconfig_author_email = NULL;
696 	}
697 
698 done:
699 	free(repo_gitconfig_path);
700 	return err;
701 }
702 
703 static const struct got_error *
704 read_gotconfig(struct got_repository *repo)
705 {
706 	const struct got_error *err = NULL;
707 	char *gotconfig_path;
708 
709 	gotconfig_path = got_repo_get_path_gotconfig(repo);
710 	if (gotconfig_path == NULL)
711 		return got_error_from_errno("got_repo_get_path_gotconfig");
712 
713 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
714 	free(gotconfig_path);
715 	return err;
716 }
717 
718 /* Supported repository format extensions. */
719 static const char *const repo_extensions[] = {
720 	"noop",			/* Got supports repository format version 1. */
721 	"preciousObjects",	/* Supported by gotadmin cleanup. */
722 	"worktreeConfig",	/* Got does not care about Git work trees. */
723 };
724 
725 const struct got_error *
726 got_repo_open(struct got_repository **repop, const char *path,
727     const char *global_gitconfig_path, int *pack_fds)
728 {
729 	struct got_repository *repo = NULL;
730 	const struct got_error *err = NULL;
731 	char *repo_path = NULL;
732 	size_t i, j = 0;
733 	struct rlimit rl;
734 
735 	*repop = NULL;
736 
737 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
738 		return got_error_from_errno("getrlimit");
739 
740 	repo = calloc(1, sizeof(*repo));
741 	if (repo == NULL)
742 		return got_error_from_errno("calloc");
743 
744 	RB_INIT(&repo->packidx_bloom_filters);
745 	TAILQ_INIT(&repo->packidx_paths);
746 
747 	for (i = 0; i < nitems(repo->privsep_children); i++) {
748 		memset(&repo->privsep_children[i], 0,
749 		    sizeof(repo->privsep_children[0]));
750 		repo->privsep_children[i].imsg_fd = -1;
751 	}
752 
753 	err = got_object_cache_init(&repo->objcache,
754 	    GOT_OBJECT_CACHE_TYPE_OBJ);
755 	if (err)
756 		goto done;
757 	err = got_object_cache_init(&repo->treecache,
758 	    GOT_OBJECT_CACHE_TYPE_TREE);
759 	if (err)
760 		goto done;
761 	err = got_object_cache_init(&repo->commitcache,
762 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
763 	if (err)
764 		goto done;
765 	err = got_object_cache_init(&repo->tagcache,
766 	    GOT_OBJECT_CACHE_TYPE_TAG);
767 	if (err)
768 		goto done;
769 	err = got_object_cache_init(&repo->rawcache,
770 	    GOT_OBJECT_CACHE_TYPE_RAW);
771 	if (err)
772 		goto done;
773 
774 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
775 	if (repo->pack_cache_size > rl.rlim_cur / 8)
776 		repo->pack_cache_size = rl.rlim_cur / 8;
777 	for (i = 0; i < nitems(repo->packs); i++) {
778 		if (i < repo->pack_cache_size) {
779 			repo->packs[i].basefd = pack_fds[j++];
780 			repo->packs[i].accumfd = pack_fds[j++];
781 		} else {
782 			repo->packs[i].basefd = -1;
783 			repo->packs[i].accumfd = -1;
784 		}
785 	}
786 	repo->pinned_pack = -1;
787 	repo->pinned_packidx = -1;
788 	repo->pinned_pid = 0;
789 
790 	repo_path = realpath(path, NULL);
791 	if (repo_path == NULL) {
792 		err = got_error_from_errno2("realpath", path);
793 		goto done;
794 	}
795 
796 	for (;;) {
797 		char *parent_path;
798 
799 		err = open_repo(repo, repo_path);
800 		if (err == NULL)
801 			break;
802 		if (err->code != GOT_ERR_NOT_GIT_REPO)
803 			goto done;
804 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
805 			err = got_error(GOT_ERR_NOT_GIT_REPO);
806 			goto done;
807 		}
808 		err = got_path_dirname(&parent_path, repo_path);
809 		if (err)
810 			goto done;
811 		free(repo_path);
812 		repo_path = parent_path;
813 	}
814 
815 	err = read_gotconfig(repo);
816 	if (err)
817 		goto done;
818 
819 	err = read_gitconfig(repo, global_gitconfig_path);
820 	if (err)
821 		goto done;
822 	if (repo->gitconfig_repository_format_version != 0) {
823 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
824 		goto done;
825 	}
826 	for (i = 0; i < repo->nextensions; i++) {
827 		char *ext = repo->extensions[i];
828 		int j, supported = 0;
829 		for (j = 0; j < nitems(repo_extensions); j++) {
830 			if (strcmp(ext, repo_extensions[j]) == 0) {
831 				supported = 1;
832 				break;
833 			}
834 		}
835 		if (!supported) {
836 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
837 			goto done;
838 		}
839 	}
840 
841 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
842 done:
843 	if (err)
844 		got_repo_close(repo);
845 	else
846 		*repop = repo;
847 	free(repo_path);
848 	return err;
849 }
850 
851 const struct got_error *
852 got_repo_close(struct got_repository *repo)
853 {
854 	const struct got_error *err = NULL, *child_err;
855 	struct got_packidx_bloom_filter *bf;
856 	struct got_pathlist_entry *pe;
857 	size_t i;
858 
859 	for (i = 0; i < repo->pack_cache_size; i++) {
860 		if (repo->packidx_cache[i] == NULL)
861 			break;
862 		got_packidx_close(repo->packidx_cache[i]);
863 	}
864 
865 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
866 	    &repo->packidx_bloom_filters))) {
867 		RB_REMOVE(got_packidx_bloom_filter_tree,
868 		    &repo->packidx_bloom_filters, bf);
869 		free(bf->bloom);
870 		free(bf);
871 	}
872 
873 	for (i = 0; i < repo->pack_cache_size; i++)
874 		if (repo->packs[i].path_packfile)
875 			if (repo->packs[i].path_packfile)
876 				got_pack_close(&repo->packs[i]);
877 
878 	free(repo->path);
879 	free(repo->path_git_dir);
880 
881 	got_object_cache_close(&repo->objcache);
882 	got_object_cache_close(&repo->treecache);
883 	got_object_cache_close(&repo->commitcache);
884 	got_object_cache_close(&repo->tagcache);
885 	got_object_cache_close(&repo->rawcache);
886 
887 	for (i = 0; i < nitems(repo->privsep_children); i++) {
888 		if (repo->privsep_children[i].imsg_fd == -1)
889 			continue;
890 		imsg_clear(repo->privsep_children[i].ibuf);
891 		free(repo->privsep_children[i].ibuf);
892 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
893 		child_err = got_privsep_wait_for_child(
894 		    repo->privsep_children[i].pid);
895 		if (child_err && err == NULL)
896 			err = child_err;
897 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
898 		    err == NULL)
899 			err = got_error_from_errno("close");
900 	}
901 
902 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
903 	    err == NULL)
904 		err = got_error_from_errno("close");
905 
906 	if (repo->gotconfig)
907 		got_gotconfig_free(repo->gotconfig);
908 	free(repo->gitconfig_author_name);
909 	free(repo->gitconfig_author_email);
910 	for (i = 0; i < repo->ngitconfig_remotes; i++)
911 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
912 	free(repo->gitconfig_remotes);
913 	for (i = 0; i < repo->nextensions; i++)
914 		free(repo->extensions[i]);
915 	free(repo->extensions);
916 
917 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
918 		free((void *)pe->path);
919 	got_pathlist_free(&repo->packidx_paths);
920 	free(repo);
921 
922 	return err;
923 }
924 
925 void
926 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
927 {
928 	int i;
929 
930 	free(repo->name);
931 	repo->name = NULL;
932 	free(repo->fetch_url);
933 	repo->fetch_url = NULL;
934 	free(repo->send_url);
935 	repo->send_url = NULL;
936 	for (i = 0; i < repo->nfetch_branches; i++)
937 		free(repo->fetch_branches[i]);
938 	free(repo->fetch_branches);
939 	repo->fetch_branches = NULL;
940 	repo->nfetch_branches = 0;
941 	for (i = 0; i < repo->nsend_branches; i++)
942 		free(repo->send_branches[i]);
943 	free(repo->send_branches);
944 	repo->send_branches = NULL;
945 	repo->nsend_branches = 0;
946 }
947 
948 const struct got_error *
949 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
950     const char *input_path)
951 {
952 	const struct got_error *err = NULL;
953 	const char *repo_abspath = NULL;
954 	size_t repolen, len;
955 	char *canonpath, *path = NULL;
956 
957 	*in_repo_path = NULL;
958 
959 	canonpath = strdup(input_path);
960 	if (canonpath == NULL) {
961 		err = got_error_from_errno("strdup");
962 		goto done;
963 	}
964 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
965 	if (err)
966 		goto done;
967 
968 	repo_abspath = got_repo_get_path(repo);
969 
970 	if (canonpath[0] == '\0') {
971 		path = strdup(canonpath);
972 		if (path == NULL) {
973 			err = got_error_from_errno("strdup");
974 			goto done;
975 		}
976 	} else {
977 		path = realpath(canonpath, NULL);
978 		if (path == NULL) {
979 			if (errno != ENOENT) {
980 				err = got_error_from_errno2("realpath",
981 				    canonpath);
982 				goto done;
983 			}
984 			/*
985 			 * Path is not on disk.
986 			 * Assume it is already relative to repository root.
987 			 */
988 			path = strdup(canonpath);
989 			if (path == NULL) {
990 				err = got_error_from_errno("strdup");
991 				goto done;
992 			}
993 		}
994 
995 		repolen = strlen(repo_abspath);
996 		len = strlen(path);
997 
998 
999 		if (strcmp(path, repo_abspath) == 0) {
1000 			free(path);
1001 			path = strdup("");
1002 			if (path == NULL) {
1003 				err = got_error_from_errno("strdup");
1004 				goto done;
1005 			}
1006 		} else if (len > repolen &&
1007 		    got_path_is_child(path, repo_abspath, repolen)) {
1008 			/* Matched an on-disk path inside repository. */
1009 			if (got_repo_is_bare(repo)) {
1010 				/*
1011 				 * Matched an on-disk path inside repository
1012 				 * database. Treat input as repository-relative.
1013 				 */
1014 				free(path);
1015 				path = canonpath;
1016 				canonpath = NULL;
1017 			} else {
1018 				char *child;
1019 				/* Strip common prefix with repository path. */
1020 				err = got_path_skip_common_ancestor(&child,
1021 				    repo_abspath, path);
1022 				if (err)
1023 					goto done;
1024 				free(path);
1025 				path = child;
1026 			}
1027 		} else {
1028 			/*
1029 			 * Matched unrelated on-disk path.
1030 			 * Treat input as repository-relative.
1031 			 */
1032 			free(path);
1033 			path = canonpath;
1034 			canonpath = NULL;
1035 		}
1036 	}
1037 
1038 	/* Make in-repository path absolute */
1039 	if (path[0] != '/') {
1040 		char *abspath;
1041 		if (asprintf(&abspath, "/%s", path) == -1) {
1042 			err = got_error_from_errno("asprintf");
1043 			goto done;
1044 		}
1045 		free(path);
1046 		path = abspath;
1047 	}
1048 
1049 done:
1050 	free(canonpath);
1051 	if (err)
1052 		free(path);
1053 	else
1054 		*in_repo_path = path;
1055 	return err;
1056 }
1057 
1058 static const struct got_error *
1059 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1060     const char *path_packidx)
1061 {
1062 	const struct got_error *err = NULL;
1063 	size_t i;
1064 
1065 	for (i = 0; i < repo->pack_cache_size; i++) {
1066 		if (repo->packidx_cache[i] == NULL)
1067 			break;
1068 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1069 		    path_packidx) == 0) {
1070 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1071 		}
1072 	}
1073 	if (i == repo->pack_cache_size) {
1074 		do {
1075 			i--;
1076 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1077 		    i == repo->pinned_packidx);
1078 		err = got_packidx_close(repo->packidx_cache[i]);
1079 		if (err)
1080 			return err;
1081 	}
1082 
1083 	repo->packidx_cache[i] = packidx;
1084 
1085 	return NULL;
1086 }
1087 
1088 int
1089 got_repo_is_packidx_filename(const char *name, size_t len)
1090 {
1091 	if (len != GOT_PACKIDX_NAMELEN)
1092 		return 0;
1093 
1094 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1095 		return 0;
1096 
1097 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1098 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1099 		return 0;
1100 
1101 	return 1;
1102 }
1103 
1104 static struct got_packidx_bloom_filter *
1105 get_packidx_bloom_filter(struct got_repository *repo,
1106     const char *path, size_t path_len)
1107 {
1108 	struct got_packidx_bloom_filter key;
1109 
1110 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1111 		return NULL; /* XXX */
1112 	key.path_len = path_len;
1113 
1114 	return RB_FIND(got_packidx_bloom_filter_tree,
1115 	    &repo->packidx_bloom_filters, &key);
1116 }
1117 
1118 int
1119 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1120     const char *path_packidx, struct got_object_id *id)
1121 {
1122 	struct got_packidx_bloom_filter *bf;
1123 
1124 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1125 	if (bf)
1126 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1127 
1128 	/* No bloom filter means this pack index must be searched. */
1129 	return 1;
1130 }
1131 
1132 static const struct got_error *
1133 add_packidx_bloom_filter(struct got_repository *repo,
1134     struct got_packidx *packidx, const char *path_packidx)
1135 {
1136 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1137 	struct got_packidx_bloom_filter *bf;
1138 	size_t len;
1139 
1140 	/*
1141 	 * Don't use bloom filters for very large pack index files.
1142 	 * Large pack files will contain a relatively large fraction
1143 	 * of our objects so we will likely need to visit them anyway.
1144 	 * The more objects a pack file contains the higher the probability
1145 	 * of a false-positive match from the bloom filter. And reading
1146 	 * all object IDs from a large pack index file can be expensive.
1147 	 */
1148 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1149 		return NULL;
1150 
1151 	/* Do we already have a filter for this pack index? */
1152 	if (get_packidx_bloom_filter(repo, path_packidx,
1153 	    strlen(path_packidx)) != NULL)
1154 		return NULL;
1155 
1156 	bf = calloc(1, sizeof(*bf));
1157 	if (bf == NULL)
1158 		return got_error_from_errno("calloc");
1159 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1160 	if (bf->bloom == NULL) {
1161 		free(bf);
1162 		return got_error_from_errno("calloc");
1163 	}
1164 	
1165 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1166 	if (len >= sizeof(bf->path)) {
1167 		free(bf->bloom);
1168 		free(bf);
1169 		return got_error(GOT_ERR_NO_SPACE);
1170 	}
1171 	bf->path_len = len;
1172 
1173 	/* Minimum size supported by our bloom filter is 1000 entries. */
1174 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1175 	for (i = 0; i < nobjects; i++) {
1176 		struct got_packidx_object_id *id;
1177 		id = &packidx->hdr.sorted_ids[i];
1178 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1179 	}
1180 
1181 	RB_INSERT(got_packidx_bloom_filter_tree,
1182 	    &repo->packidx_bloom_filters, bf);
1183 	return NULL;
1184 }
1185 
1186 const struct got_error *
1187 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1188     struct got_repository *repo, struct got_object_id *id)
1189 {
1190 	const struct got_error *err;
1191 	struct got_pathlist_entry *pe;
1192 	size_t i;
1193 
1194 	/* Search pack index cache. */
1195 	for (i = 0; i < repo->pack_cache_size; i++) {
1196 		if (repo->packidx_cache[i] == NULL)
1197 			break;
1198 		if (!got_repo_check_packidx_bloom_filter(repo,
1199 		    repo->packidx_cache[i]->path_packidx, id))
1200 			continue; /* object will not be found in this index */
1201 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1202 		if (*idx != -1) {
1203 			*packidx = repo->packidx_cache[i];
1204 			/*
1205 			 * Move this cache entry to the front. Repeatedly
1206 			 * searching a wrong pack index can be expensive.
1207 			 */
1208 			if (i > 0) {
1209 				memmove(&repo->packidx_cache[1],
1210 				    &repo->packidx_cache[0],
1211 				    i * sizeof(repo->packidx_cache[0]));
1212 				repo->packidx_cache[0] = *packidx;
1213 				if (repo->pinned_packidx >= 0 &&
1214 				    repo->pinned_packidx < i)
1215 					repo->pinned_packidx++;
1216 				else if (repo->pinned_packidx == i)
1217 					repo->pinned_packidx = 0;
1218 			}
1219 			return NULL;
1220 		}
1221 	}
1222 	/* No luck. Search the filesystem. */
1223 
1224 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1225 		const char *path_packidx = pe->path;
1226 		int is_cached = 0;
1227 
1228 		if (!got_repo_check_packidx_bloom_filter(repo,
1229 		    pe->path, id))
1230 			continue; /* object will not be found in this index */
1231 
1232 		for (i = 0; i < repo->pack_cache_size; i++) {
1233 			if (repo->packidx_cache[i] == NULL)
1234 				break;
1235 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1236 			    path_packidx) == 0) {
1237 				is_cached = 1;
1238 				break;
1239 			}
1240 		}
1241 		if (is_cached)
1242 			continue; /* already searched */
1243 
1244 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1245 		    path_packidx, 0);
1246 		if (err)
1247 			goto done;
1248 
1249 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1250 		if (err)
1251 			goto done;
1252 
1253 		err = cache_packidx(repo, *packidx, path_packidx);
1254 		if (err)
1255 			goto done;
1256 
1257 		*idx = got_packidx_get_object_idx(*packidx, id);
1258 		if (*idx != -1) {
1259 			err = NULL; /* found the object */
1260 			goto done;
1261 		}
1262 	}
1263 
1264 	err = got_error_no_obj(id);
1265 done:
1266 	return err;
1267 }
1268 
1269 const struct got_error *
1270 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1271     struct got_repository *repo)
1272 {
1273 	const struct got_error *err = NULL;
1274 	DIR *packdir = NULL;
1275 	struct dirent *dent;
1276 	char *path_packidx = NULL;
1277 	int packdir_fd;
1278 
1279 	packdir_fd = openat(got_repo_get_fd(repo),
1280 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1281 	if (packdir_fd == -1) {
1282 		return got_error_from_errno_fmt("openat: %s/%s",
1283 		    got_repo_get_path_git_dir(repo),
1284 		    GOT_OBJECTS_PACK_DIR);
1285 	}
1286 
1287 	packdir = fdopendir(packdir_fd);
1288 	if (packdir == NULL) {
1289 		err = got_error_from_errno("fdopendir");
1290 		goto done;
1291 	}
1292 
1293 	while ((dent = readdir(packdir)) != NULL) {
1294 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1295 			continue;
1296 
1297 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1298 		    dent->d_name) == -1) {
1299 			err = got_error_from_errno("asprintf");
1300 			path_packidx = NULL;
1301 			break;
1302 		}
1303 
1304 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1305 		if (err)
1306 			break;
1307 	}
1308 done:
1309 	if (err)
1310 		free(path_packidx);
1311 	if (packdir && closedir(packdir) != 0 && err == NULL)
1312 		err = got_error_from_errno("closedir");
1313 	return err;
1314 }
1315 
1316 const struct got_error *
1317 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1318     struct got_repository *repo)
1319 {
1320 	const struct got_error *err;
1321 	size_t i;
1322 
1323 	*packidx = NULL;
1324 
1325 	/* Search pack index cache. */
1326 	for (i = 0; i < repo->pack_cache_size; i++) {
1327 		if (repo->packidx_cache[i] == NULL)
1328 			break;
1329 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1330 		    path_packidx) == 0) {
1331 			*packidx = repo->packidx_cache[i];
1332 			return NULL;
1333 		}
1334 	}
1335 	/* No luck. Search the filesystem. */
1336 
1337 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1338 	    path_packidx, 0);
1339 	if (err)
1340 		return err;
1341 
1342 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1343 	if (err)
1344 		goto done;
1345 
1346 	err = cache_packidx(repo, *packidx, path_packidx);
1347 done:
1348 	if (err) {
1349 		got_packidx_close(*packidx);
1350 		*packidx = NULL;
1351 	}
1352 	return err;
1353 }
1354 
1355 static const struct got_error *
1356 read_packfile_hdr(int fd, struct got_packidx *packidx)
1357 {
1358 	const struct got_error *err = NULL;
1359 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1360 	struct got_packfile_hdr hdr;
1361 	ssize_t n;
1362 
1363 	n = read(fd, &hdr, sizeof(hdr));
1364 	if (n < 0)
1365 		return got_error_from_errno("read");
1366 	if (n != sizeof(hdr))
1367 		return got_error(GOT_ERR_BAD_PACKFILE);
1368 
1369 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1370 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1371 	    be32toh(hdr.nobjects) != totobj)
1372 		err = got_error(GOT_ERR_BAD_PACKFILE);
1373 
1374 	return err;
1375 }
1376 
1377 static const struct got_error *
1378 open_packfile(int *fd, struct got_repository *repo,
1379     const char *relpath, struct got_packidx *packidx)
1380 {
1381 	const struct got_error *err = NULL;
1382 
1383 	*fd = openat(got_repo_get_fd(repo), relpath,
1384 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1385 	if (*fd == -1)
1386 		return got_error_from_errno_fmt("openat: %s/%s",
1387 		    got_repo_get_path_git_dir(repo), relpath);
1388 
1389 	if (packidx) {
1390 		err = read_packfile_hdr(*fd, packidx);
1391 		if (err) {
1392 			close(*fd);
1393 			*fd = -1;
1394 		}
1395 	}
1396 
1397 	return err;
1398 }
1399 
1400 const struct got_error *
1401 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1402     const char *path_packfile, struct got_packidx *packidx)
1403 {
1404 	const struct got_error *err = NULL;
1405 	struct got_pack *pack = NULL;
1406 	struct stat sb;
1407 	size_t i;
1408 
1409 	if (packp)
1410 		*packp = NULL;
1411 
1412 	for (i = 0; i < repo->pack_cache_size; i++) {
1413 		pack = &repo->packs[i];
1414 		if (pack->path_packfile == NULL)
1415 			break;
1416 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1417 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1418 	}
1419 
1420 	if (i == repo->pack_cache_size) {
1421 		struct got_pack tmp;
1422 		do {
1423 			i--;
1424 		} while (i > 0 && repo->pinned_pack >= 0 &&
1425 		    i == repo->pinned_pack);
1426 		err = got_pack_close(&repo->packs[i]);
1427 		if (err)
1428 			return err;
1429 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1430 			return got_error_from_errno("ftruncate");
1431 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1432 			return got_error_from_errno("ftruncate");
1433 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1434 		memcpy(&repo->packs[i], &repo->packs[0],
1435 		    sizeof(repo->packs[i]));
1436 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1437 		if (repo->pinned_pack == 0)
1438 			repo->pinned_pack = i;
1439 		else if (repo->pinned_pack == i)
1440 			repo->pinned_pack = 0;
1441 		i = 0;
1442 	}
1443 
1444 	pack = &repo->packs[i];
1445 
1446 	pack->path_packfile = strdup(path_packfile);
1447 	if (pack->path_packfile == NULL) {
1448 		err = got_error_from_errno("strdup");
1449 		goto done;
1450 	}
1451 
1452 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1453 	if (err)
1454 		goto done;
1455 
1456 	if (fstat(pack->fd, &sb) != 0) {
1457 		err = got_error_from_errno("fstat");
1458 		goto done;
1459 	}
1460 	pack->filesize = sb.st_size;
1461 
1462 	pack->privsep_child = NULL;
1463 
1464 #ifndef GOT_PACK_NO_MMAP
1465 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1466 	    pack->fd, 0);
1467 	if (pack->map == MAP_FAILED) {
1468 		if (errno != ENOMEM) {
1469 			err = got_error_from_errno("mmap");
1470 			goto done;
1471 		}
1472 		pack->map = NULL; /* fall back to read(2) */
1473 	}
1474 #endif
1475 done:
1476 	if (err) {
1477 		if (pack) {
1478 			free(pack->path_packfile);
1479 			memset(pack, 0, sizeof(*pack));
1480 		}
1481 	} else if (packp)
1482 		*packp = pack;
1483 	return err;
1484 }
1485 
1486 struct got_pack *
1487 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1488 {
1489 	struct got_pack *pack = NULL;
1490 	size_t i;
1491 
1492 	for (i = 0; i < repo->pack_cache_size; i++) {
1493 		pack = &repo->packs[i];
1494 		if (pack->path_packfile == NULL)
1495 			break;
1496 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1497 			return pack;
1498 	}
1499 
1500 	return NULL;
1501 }
1502 
1503 const struct got_error *
1504 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1505     struct got_pack *pack)
1506 {
1507 	size_t i;
1508 	int pinned_pack = -1, pinned_packidx = -1;
1509 
1510 	for (i = 0; i < repo->pack_cache_size; i++) {
1511 		if (repo->packidx_cache[i] &&
1512 		    strcmp(repo->packidx_cache[i]->path_packidx,
1513 		    packidx->path_packidx) == 0)
1514 			pinned_packidx = i;
1515 		if (repo->packs[i].path_packfile &&
1516 		    strcmp(repo->packs[i].path_packfile,
1517 		    pack->path_packfile) == 0)
1518 			pinned_pack = i;
1519 	}
1520 
1521 	if (pinned_packidx == -1 || pinned_pack == -1)
1522 		return got_error(GOT_ERR_PIN_PACK);
1523 
1524 	repo->pinned_pack = pinned_pack;
1525 	repo->pinned_packidx = pinned_packidx;
1526 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1527 	return NULL;
1528 }
1529 
1530 struct got_pack *
1531 got_repo_get_pinned_pack(struct got_repository *repo)
1532 {
1533 	if (repo->pinned_pack >= 0 &&
1534 	    repo->pinned_pack < repo->pack_cache_size)
1535 		return &repo->packs[repo->pinned_pack];
1536 
1537 	return NULL;
1538 }
1539 
1540 void
1541 got_repo_unpin_pack(struct got_repository *repo)
1542 {
1543 	repo->pinned_packidx = -1;
1544 	repo->pinned_pack = -1;
1545 	repo->pinned_pid = 0;
1546 }
1547 
1548 const struct got_error *
1549 got_repo_init(const char *repo_path)
1550 {
1551 	const struct got_error *err = NULL;
1552 	const char *dirnames[] = {
1553 		GOT_OBJECTS_DIR,
1554 		GOT_OBJECTS_PACK_DIR,
1555 		GOT_REFS_DIR,
1556 	};
1557 	const char *description_str = "Unnamed repository; "
1558 	    "edit this file 'description' to name the repository.";
1559 	const char *headref_str = "ref: refs/heads/main";
1560 	const char *gitconfig_str = "[core]\n"
1561 	    "\trepositoryformatversion = 0\n"
1562 	    "\tfilemode = true\n"
1563 	    "\tbare = true\n";
1564 	char *path;
1565 	size_t i;
1566 
1567 	if (!got_path_dir_is_empty(repo_path))
1568 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1569 
1570 	for (i = 0; i < nitems(dirnames); i++) {
1571 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1572 			return got_error_from_errno("asprintf");
1573 		}
1574 		err = got_path_mkdir(path);
1575 		free(path);
1576 		if (err)
1577 			return err;
1578 	}
1579 
1580 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1581 		return got_error_from_errno("asprintf");
1582 	err = got_path_create_file(path, description_str);
1583 	free(path);
1584 	if (err)
1585 		return err;
1586 
1587 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1588 		return got_error_from_errno("asprintf");
1589 	err = got_path_create_file(path, headref_str);
1590 	free(path);
1591 	if (err)
1592 		return err;
1593 
1594 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1595 		return got_error_from_errno("asprintf");
1596 	err = got_path_create_file(path, gitconfig_str);
1597 	free(path);
1598 	if (err)
1599 		return err;
1600 
1601 	return NULL;
1602 }
1603 
1604 static const struct got_error *
1605 match_packed_object(struct got_object_id **unique_id,
1606     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1607 {
1608 	const struct got_error *err = NULL;
1609 	struct got_object_id_queue matched_ids;
1610 	struct got_pathlist_entry *pe;
1611 
1612 	STAILQ_INIT(&matched_ids);
1613 
1614 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1615 		const char *path_packidx = pe->path;
1616 		struct got_packidx *packidx;
1617 		struct got_object_qid *qid;
1618 
1619 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1620 		    path_packidx, 0);
1621 		if (err)
1622 			break;
1623 
1624 		err = got_packidx_match_id_str_prefix(&matched_ids,
1625 		    packidx, id_str_prefix);
1626 		if (err) {
1627 			got_packidx_close(packidx);
1628 			break;
1629 		}
1630 		err = got_packidx_close(packidx);
1631 		if (err)
1632 			break;
1633 
1634 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1635 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1636 				int matched_type;
1637 				err = got_object_get_type(&matched_type, repo,
1638 				    &qid->id);
1639 				if (err)
1640 					goto done;
1641 				if (matched_type != obj_type)
1642 					continue;
1643 			}
1644 			if (*unique_id == NULL) {
1645 				*unique_id = got_object_id_dup(&qid->id);
1646 				if (*unique_id == NULL) {
1647 					err = got_error_from_errno("malloc");
1648 					goto done;
1649 				}
1650 			} else {
1651 				if (got_object_id_cmp(*unique_id,
1652 				    &qid->id) == 0)
1653 					continue; /* packed multiple times */
1654 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1655 				goto done;
1656 			}
1657 		}
1658 	}
1659 done:
1660 	got_object_id_queue_free(&matched_ids);
1661 	if (err) {
1662 		free(*unique_id);
1663 		*unique_id = NULL;
1664 	}
1665 	return err;
1666 }
1667 
1668 static const struct got_error *
1669 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1670     const char *object_dir, const char *id_str_prefix, int obj_type,
1671     struct got_repository *repo)
1672 {
1673 	const struct got_error *err = NULL;
1674 	char *path;
1675 	DIR *dir = NULL;
1676 	struct dirent *dent;
1677 	struct got_object_id id;
1678 
1679 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1680 		err = got_error_from_errno("asprintf");
1681 		goto done;
1682 	}
1683 
1684 	dir = opendir(path);
1685 	if (dir == NULL) {
1686 		if (errno == ENOENT) {
1687 			err = NULL;
1688 			goto done;
1689 		}
1690 		err = got_error_from_errno2("opendir", path);
1691 		goto done;
1692 	}
1693 	while ((dent = readdir(dir)) != NULL) {
1694 		char *id_str;
1695 		int cmp;
1696 
1697 		if (strcmp(dent->d_name, ".") == 0 ||
1698 		    strcmp(dent->d_name, "..") == 0)
1699 			continue;
1700 
1701 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1702 			err = got_error_from_errno("asprintf");
1703 			goto done;
1704 		}
1705 
1706 		if (!got_parse_sha1_digest(id.sha1, id_str))
1707 			continue;
1708 
1709 		/*
1710 		 * Directory entries do not necessarily appear in
1711 		 * sorted order, so we must iterate over all of them.
1712 		 */
1713 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1714 		if (cmp != 0) {
1715 			free(id_str);
1716 			continue;
1717 		}
1718 
1719 		if (*unique_id == NULL) {
1720 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1721 				int matched_type;
1722 				err = got_object_get_type(&matched_type, repo,
1723 				    &id);
1724 				if (err)
1725 					goto done;
1726 				if (matched_type != obj_type)
1727 					continue;
1728 			}
1729 			*unique_id = got_object_id_dup(&id);
1730 			if (*unique_id == NULL) {
1731 				err = got_error_from_errno("got_object_id_dup");
1732 				free(id_str);
1733 				goto done;
1734 			}
1735 		} else {
1736 			if (got_object_id_cmp(*unique_id, &id) == 0)
1737 				continue; /* both packed and loose */
1738 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1739 			free(id_str);
1740 			goto done;
1741 		}
1742 	}
1743 done:
1744 	if (dir && closedir(dir) != 0 && err == NULL)
1745 		err = got_error_from_errno("closedir");
1746 	if (err) {
1747 		free(*unique_id);
1748 		*unique_id = NULL;
1749 	}
1750 	free(path);
1751 	return err;
1752 }
1753 
1754 const struct got_error *
1755 got_repo_match_object_id_prefix(struct got_object_id **id,
1756     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1757 {
1758 	const struct got_error *err = NULL;
1759 	char *path_objects = got_repo_get_path_objects(repo);
1760 	char *object_dir = NULL;
1761 	size_t len;
1762 	int i;
1763 
1764 	*id = NULL;
1765 
1766 	len = strlen(id_str_prefix);
1767 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1768 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1769 
1770 	for (i = 0; i < len; i++) {
1771 		if (isxdigit((unsigned char)id_str_prefix[i]))
1772 			continue;
1773 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1774 	}
1775 
1776 	if (len >= 2) {
1777 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1778 		if (err)
1779 			goto done;
1780 		object_dir = strndup(id_str_prefix, 2);
1781 		if (object_dir == NULL) {
1782 			err = got_error_from_errno("strdup");
1783 			goto done;
1784 		}
1785 		err = match_loose_object(id, path_objects, object_dir,
1786 		    id_str_prefix, obj_type, repo);
1787 	} else if (len == 1) {
1788 		int i;
1789 		for (i = 0; i < 0xf; i++) {
1790 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1791 			    == -1) {
1792 				err = got_error_from_errno("asprintf");
1793 				goto done;
1794 			}
1795 			err = match_packed_object(id, repo, object_dir,
1796 			    obj_type);
1797 			if (err)
1798 				goto done;
1799 			err = match_loose_object(id, path_objects, object_dir,
1800 			    id_str_prefix, obj_type, repo);
1801 			if (err)
1802 				goto done;
1803 		}
1804 	} else {
1805 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1806 		goto done;
1807 	}
1808 done:
1809 	free(object_dir);
1810 	if (err) {
1811 		free(*id);
1812 		*id = NULL;
1813 	} else if (*id == NULL) {
1814 		switch (obj_type) {
1815 		case GOT_OBJ_TYPE_BLOB:
1816 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1817 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1818 			break;
1819 		case GOT_OBJ_TYPE_TREE:
1820 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1821 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1822 			break;
1823 		case GOT_OBJ_TYPE_COMMIT:
1824 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1825 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1826 			break;
1827 		case GOT_OBJ_TYPE_TAG:
1828 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1829 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1830 			break;
1831 		default:
1832 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1833 			break;
1834 		}
1835 	}
1836 
1837 	return err;
1838 }
1839 
1840 const struct got_error *
1841 got_repo_match_object_id(struct got_object_id **id, char **label,
1842     const char *id_str, int obj_type, struct got_reflist_head *refs,
1843     struct got_repository *repo)
1844 {
1845 	const struct got_error *err;
1846 	struct got_tag_object *tag;
1847 	struct got_reference *ref = NULL;
1848 
1849 	*id = NULL;
1850 	if (label)
1851 		*label = NULL;
1852 
1853 	if (refs) {
1854 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1855 		    refs, repo);
1856 		if (err == NULL) {
1857 			*id = got_object_id_dup(
1858 			    got_object_tag_get_object_id(tag));
1859 			if (*id == NULL)
1860 				err = got_error_from_errno("got_object_id_dup");
1861 			else if (label && asprintf(label, "refs/tags/%s",
1862 			    got_object_tag_get_name(tag)) == -1) {
1863 				err = got_error_from_errno("asprintf");
1864 				free(*id);
1865 				*id = NULL;
1866 			}
1867 			got_object_tag_close(tag);
1868 			return err;
1869 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1870 		    err->code != GOT_ERR_NO_OBJ)
1871 			return err;
1872 	}
1873 
1874 	err = got_ref_open(&ref, repo, id_str, 0);
1875 	if (err == NULL) {
1876 		err = got_ref_resolve(id, repo, ref);
1877 		if (err)
1878 			goto done;
1879 		if (label) {
1880 			*label = strdup(got_ref_get_name(ref));
1881 			if (*label == NULL) {
1882 				err = got_error_from_errno("strdup");
1883 				goto done;
1884 			}
1885 		}
1886 	} else {
1887 		if (err->code != GOT_ERR_NOT_REF &&
1888 		    err->code != GOT_ERR_BAD_REF_NAME)
1889 			goto done;
1890 		err = got_repo_match_object_id_prefix(id, id_str,
1891 		    obj_type, repo);
1892 		if (err) {
1893 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1894 				err = got_error_not_ref(id_str);
1895 			goto done;
1896 		}
1897 		if (label) {
1898 			err = got_object_id_str(label, *id);
1899 			if (*label == NULL) {
1900 				err = got_error_from_errno("strdup");
1901 				goto done;
1902 			}
1903 		}
1904 	}
1905 done:
1906 	if (ref)
1907 		got_ref_close(ref);
1908 	return err;
1909 }
1910 
1911 const struct got_error *
1912 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1913     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1914 {
1915 	const struct got_error *err = NULL;
1916 	struct got_reflist_entry *re;
1917 	struct got_object_id *tag_id;
1918 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1919 
1920 	*tag = NULL;
1921 
1922 	TAILQ_FOREACH(re, refs, entry) {
1923 		const char *refname;
1924 		refname = got_ref_get_name(re->ref);
1925 		if (got_ref_is_symbolic(re->ref))
1926 			continue;
1927 		if (strncmp(refname, "refs/tags/", 10) != 0)
1928 			continue;
1929 		if (!name_is_absolute)
1930 			refname += strlen("refs/tags/");
1931 		if (strcmp(refname, name) != 0)
1932 			continue;
1933 		err = got_ref_resolve(&tag_id, repo, re->ref);
1934 		if (err)
1935 			break;
1936 		err = got_object_open_as_tag(tag, repo, tag_id);
1937 		free(tag_id);
1938 		if (err)
1939 			break;
1940 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1941 		    got_object_tag_get_object_type(*tag) == obj_type)
1942 			break;
1943 		got_object_tag_close(*tag);
1944 		*tag = NULL;
1945 	}
1946 
1947 	if (err == NULL && *tag == NULL)
1948 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1949 		    GOT_OBJ_LABEL_TAG, name);
1950 	return err;
1951 }
1952 
1953 static const struct got_error *
1954 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1955     const char *name, mode_t mode, struct got_object_id *blob_id)
1956 {
1957 	const struct got_error *err = NULL;
1958 
1959 	 *new_te = NULL;
1960 
1961 	*new_te = calloc(1, sizeof(**new_te));
1962 	if (*new_te == NULL)
1963 		return got_error_from_errno("calloc");
1964 
1965 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1966 	    sizeof((*new_te)->name)) {
1967 		err = got_error(GOT_ERR_NO_SPACE);
1968 		goto done;
1969 	}
1970 
1971 	if (S_ISLNK(mode)) {
1972 		(*new_te)->mode = S_IFLNK;
1973 	} else {
1974 		(*new_te)->mode = S_IFREG;
1975 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1976 	}
1977 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1978 done:
1979 	if (err && *new_te) {
1980 		free(*new_te);
1981 		*new_te = NULL;
1982 	}
1983 	return err;
1984 }
1985 
1986 static const struct got_error *
1987 import_file(struct got_tree_entry **new_te, struct dirent *de,
1988     const char *path, struct got_repository *repo)
1989 {
1990 	const struct got_error *err;
1991 	struct got_object_id *blob_id = NULL;
1992 	char *filepath;
1993 	struct stat sb;
1994 
1995 	if (asprintf(&filepath, "%s%s%s", path,
1996 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1997 		return got_error_from_errno("asprintf");
1998 
1999 	if (lstat(filepath, &sb) != 0) {
2000 		err = got_error_from_errno2("lstat", path);
2001 		goto done;
2002 	}
2003 
2004 	err = got_object_blob_create(&blob_id, filepath, repo);
2005 	if (err)
2006 		goto done;
2007 
2008 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2009 	    blob_id);
2010 done:
2011 	free(filepath);
2012 	if (err)
2013 		free(blob_id);
2014 	return err;
2015 }
2016 
2017 static const struct got_error *
2018 insert_tree_entry(struct got_tree_entry *new_te,
2019     struct got_pathlist_head *paths)
2020 {
2021 	const struct got_error *err = NULL;
2022 	struct got_pathlist_entry *new_pe;
2023 
2024 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2025 	if (err)
2026 		return err;
2027 	if (new_pe == NULL)
2028 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2029 	return NULL;
2030 }
2031 
2032 static const struct got_error *write_tree(struct got_object_id **,
2033     const char *, struct got_pathlist_head *, struct got_repository *,
2034     got_repo_import_cb progress_cb, void *progress_arg);
2035 
2036 static const struct got_error *
2037 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2038     const char *path, struct got_pathlist_head *ignores,
2039     struct got_repository *repo,
2040     got_repo_import_cb progress_cb, void *progress_arg)
2041 {
2042 	const struct got_error *err;
2043 	struct got_object_id *id = NULL;
2044 	char *subdirpath;
2045 
2046 	if (asprintf(&subdirpath, "%s%s%s", path,
2047 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2048 		return got_error_from_errno("asprintf");
2049 
2050 	(*new_te) = calloc(1, sizeof(**new_te));
2051 	if (*new_te == NULL)
2052 		return got_error_from_errno("calloc");
2053 	(*new_te)->mode = S_IFDIR;
2054 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2055 	    sizeof((*new_te)->name)) {
2056 		err = got_error(GOT_ERR_NO_SPACE);
2057 		goto done;
2058 	}
2059 	err = write_tree(&id, subdirpath, ignores,  repo,
2060 	    progress_cb, progress_arg);
2061 	if (err)
2062 		goto done;
2063 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2064 	
2065 done:
2066 	free(id);
2067 	free(subdirpath);
2068 	if (err) {
2069 		free(*new_te);
2070 		*new_te = NULL;
2071 	}
2072 	return err;
2073 }
2074 
2075 static const struct got_error *
2076 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2077     struct got_pathlist_head *ignores, struct got_repository *repo,
2078     got_repo_import_cb progress_cb, void *progress_arg)
2079 {
2080 	const struct got_error *err = NULL;
2081 	DIR *dir;
2082 	struct dirent *de;
2083 	int nentries;
2084 	struct got_tree_entry *new_te = NULL;
2085 	struct got_pathlist_head paths;
2086 	struct got_pathlist_entry *pe;
2087 
2088 	*new_tree_id = NULL;
2089 
2090 	TAILQ_INIT(&paths);
2091 
2092 	dir = opendir(path_dir);
2093 	if (dir == NULL) {
2094 		err = got_error_from_errno2("opendir", path_dir);
2095 		goto done;
2096 	}
2097 
2098 	nentries = 0;
2099 	while ((de = readdir(dir)) != NULL) {
2100 		int ignore = 0;
2101 		int type;
2102 
2103 		if (strcmp(de->d_name, ".") == 0 ||
2104 		    strcmp(de->d_name, "..") == 0)
2105 			continue;
2106 
2107 		TAILQ_FOREACH(pe, ignores, entry) {
2108 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2109 				ignore = 1;
2110 				break;
2111 			}
2112 		}
2113 		if (ignore)
2114 			continue;
2115 
2116 		err = got_path_dirent_type(&type, path_dir, de);
2117 		if (err)
2118 			goto done;
2119 
2120 		if (type == DT_DIR) {
2121 			err = import_subdir(&new_te, de, path_dir,
2122 			    ignores, repo, progress_cb, progress_arg);
2123 			if (err) {
2124 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2125 					goto done;
2126 				err = NULL;
2127 				continue;
2128 			}
2129 		} else if (type == DT_REG || type == DT_LNK) {
2130 			err = import_file(&new_te, de, path_dir, repo);
2131 			if (err)
2132 				goto done;
2133 		} else
2134 			continue;
2135 
2136 		err = insert_tree_entry(new_te, &paths);
2137 		if (err)
2138 			goto done;
2139 		nentries++;
2140 	}
2141 
2142 	if (TAILQ_EMPTY(&paths)) {
2143 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2144 		    "cannot create tree without any entries");
2145 		goto done;
2146 	}
2147 
2148 	TAILQ_FOREACH(pe, &paths, entry) {
2149 		struct got_tree_entry *te = pe->data;
2150 		char *path;
2151 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2152 			continue;
2153 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2154 			err = got_error_from_errno("asprintf");
2155 			goto done;
2156 		}
2157 		err = (*progress_cb)(progress_arg, path);
2158 		free(path);
2159 		if (err)
2160 			goto done;
2161 	}
2162 
2163 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2164 done:
2165 	if (dir)
2166 		closedir(dir);
2167 	got_pathlist_free(&paths);
2168 	return err;
2169 }
2170 
2171 const struct got_error *
2172 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2173     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2174     struct got_repository *repo, got_repo_import_cb progress_cb,
2175     void *progress_arg)
2176 {
2177 	const struct got_error *err;
2178 	struct got_object_id *new_tree_id;
2179 
2180 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2181 	    progress_cb, progress_arg);
2182 	if (err)
2183 		return err;
2184 
2185 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2186 	    author, time(NULL), author, time(NULL), logmsg, repo);
2187 	free(new_tree_id);
2188 	return err;
2189 }
2190 
2191 const struct got_error *
2192 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2193     struct got_repository *repo)
2194 {
2195 	const struct got_error *err = NULL;
2196 	char *path_objects = NULL, *path = NULL;
2197 	DIR *dir = NULL;
2198 	struct got_object_id id;
2199 	int i;
2200 
2201 	*nobjects = 0;
2202 	*ondisk_size = 0;
2203 
2204 	path_objects = got_repo_get_path_objects(repo);
2205 	if (path_objects == NULL)
2206 		return got_error_from_errno("got_repo_get_path_objects");
2207 
2208 	for (i = 0; i <= 0xff; i++) {
2209 		struct dirent *dent;
2210 
2211 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2212 			err = got_error_from_errno("asprintf");
2213 			break;
2214 		}
2215 
2216 		dir = opendir(path);
2217 		if (dir == NULL) {
2218 			if (errno == ENOENT) {
2219 				err = NULL;
2220 				continue;
2221 			}
2222 			err = got_error_from_errno2("opendir", path);
2223 			break;
2224 		}
2225 
2226 		while ((dent = readdir(dir)) != NULL) {
2227 			char *id_str;
2228 			int fd;
2229 			struct stat sb;
2230 
2231 			if (strcmp(dent->d_name, ".") == 0 ||
2232 			    strcmp(dent->d_name, "..") == 0)
2233 				continue;
2234 
2235 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2236 				err = got_error_from_errno("asprintf");
2237 				goto done;
2238 			}
2239 
2240 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2241 				free(id_str);
2242 				continue;
2243 			}
2244 			free(id_str);
2245 
2246 			err = got_object_open_loose_fd(&fd, &id, repo);
2247 			if (err)
2248 				goto done;
2249 
2250 			if (fstat(fd, &sb) == -1) {
2251 				err = got_error_from_errno("fstat");
2252 				close(fd);
2253 				goto done;
2254 			}
2255 			(*nobjects)++;
2256 			(*ondisk_size) += sb.st_size;
2257 
2258 			if (close(fd) == -1) {
2259 				err = got_error_from_errno("close");
2260 				goto done;
2261 			}
2262 		}
2263 
2264 		if (closedir(dir) != 0) {
2265 			err = got_error_from_errno("closedir");
2266 			goto done;
2267 		}
2268 		dir = NULL;
2269 
2270 		free(path);
2271 		path = NULL;
2272 	}
2273 done:
2274 	if (dir && closedir(dir) != 0 && err == NULL)
2275 		err = got_error_from_errno("closedir");
2276 
2277 	if (err) {
2278 		*nobjects = 0;
2279 		*ondisk_size = 0;
2280 	}
2281 	free(path_objects);
2282 	free(path);
2283 	return err;
2284 }
2285 
2286 const struct got_error *
2287 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2288     off_t *total_packsize, struct got_repository *repo)
2289 {
2290 	const struct got_error *err = NULL;
2291 	DIR *packdir = NULL;
2292 	struct dirent *dent;
2293 	struct got_packidx *packidx = NULL;
2294 	char *path_packidx;
2295 	char *path_packfile;
2296 	int packdir_fd;
2297 	struct stat sb;
2298 
2299 	*npackfiles = 0;
2300 	*nobjects = 0;
2301 	*total_packsize = 0;
2302 
2303 	packdir_fd = openat(got_repo_get_fd(repo),
2304 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2305 	if (packdir_fd == -1) {
2306 		return got_error_from_errno_fmt("openat: %s/%s",
2307 		    got_repo_get_path_git_dir(repo),
2308 		    GOT_OBJECTS_PACK_DIR);
2309 	}
2310 
2311 	packdir = fdopendir(packdir_fd);
2312 	if (packdir == NULL) {
2313 		err = got_error_from_errno("fdopendir");
2314 		goto done;
2315 	}
2316 
2317 	while ((dent = readdir(packdir)) != NULL) {
2318 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2319 			continue;
2320 
2321 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2322 		    dent->d_name) == -1) {
2323 			err = got_error_from_errno("asprintf");
2324 			goto done;
2325 		}
2326 
2327 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2328 		    path_packidx, 0);
2329 		free(path_packidx);
2330 		if (err)
2331 			goto done;
2332 
2333 		if (fstat(packidx->fd, &sb) == -1)
2334 			goto done;
2335 		*total_packsize += sb.st_size;
2336 
2337 		err = got_packidx_get_packfile_path(&path_packfile,
2338 		    packidx->path_packidx);
2339 		if (err)
2340 			goto done;
2341 
2342 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2343 		    0) == -1) {
2344 			free(path_packfile);
2345 			goto done;
2346 		}
2347 		free(path_packfile);
2348 		*total_packsize += sb.st_size;
2349 
2350 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2351 
2352 		(*npackfiles)++;
2353 
2354 		got_packidx_close(packidx);
2355 		packidx = NULL;
2356 	}
2357 done:
2358 	if (packidx)
2359 		got_packidx_close(packidx);
2360 	if (packdir && closedir(packdir) != 0 && err == NULL)
2361 		err = got_error_from_errno("closedir");
2362 	if (err) {
2363 		*npackfiles = 0;
2364 		*nobjects = 0;
2365 		*total_packsize = 0;
2366 	}
2367 	return err;
2368 }
2369 
2370 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2371     got_packidx_bloom_filter_cmp);