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 		free(bf->bloom);
876 		free(bf);
877 	}
878 
879 	for (i = 0; i < repo->pack_cache_size; i++)
880 		if (repo->packs[i].path_packfile)
881 			if (repo->packs[i].path_packfile)
882 				got_pack_close(&repo->packs[i]);
883 
884 	free(repo->path);
885 	free(repo->path_git_dir);
886 
887 	got_object_cache_close(&repo->objcache);
888 	got_object_cache_close(&repo->treecache);
889 	got_object_cache_close(&repo->commitcache);
890 	got_object_cache_close(&repo->tagcache);
891 	got_object_cache_close(&repo->rawcache);
892 
893 	for (i = 0; i < nitems(repo->privsep_children); i++) {
894 		if (repo->privsep_children[i].imsg_fd == -1)
895 			continue;
896 		imsg_clear(repo->privsep_children[i].ibuf);
897 		free(repo->privsep_children[i].ibuf);
898 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
899 		child_err = got_privsep_wait_for_child(
900 		    repo->privsep_children[i].pid);
901 		if (child_err && err == NULL)
902 			err = child_err;
903 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
904 		    err == NULL)
905 			err = got_error_from_errno("close");
906 	}
907 
908 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
909 	    err == NULL)
910 		err = got_error_from_errno("close");
911 
912 	if (repo->gotconfig)
913 		got_gotconfig_free(repo->gotconfig);
914 	free(repo->gitconfig_author_name);
915 	free(repo->gitconfig_author_email);
916 	for (i = 0; i < repo->ngitconfig_remotes; i++)
917 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
918 	free(repo->gitconfig_remotes);
919 	for (i = 0; i < repo->nextensions; i++)
920 		free(repo->extensions[i]);
921 	free(repo->extensions);
922 
923 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
924 		free((void *)pe->path);
925 	got_pathlist_free(&repo->packidx_paths);
926 	free(repo);
927 
928 	return err;
929 }
930 
931 void
932 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
933 {
934 	int i;
935 
936 	free(repo->name);
937 	repo->name = NULL;
938 	free(repo->fetch_url);
939 	repo->fetch_url = NULL;
940 	free(repo->send_url);
941 	repo->send_url = NULL;
942 	for (i = 0; i < repo->nfetch_branches; i++)
943 		free(repo->fetch_branches[i]);
944 	free(repo->fetch_branches);
945 	repo->fetch_branches = NULL;
946 	repo->nfetch_branches = 0;
947 	for (i = 0; i < repo->nsend_branches; i++)
948 		free(repo->send_branches[i]);
949 	free(repo->send_branches);
950 	repo->send_branches = NULL;
951 	repo->nsend_branches = 0;
952 }
953 
954 const struct got_error *
955 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
956     const char *input_path)
957 {
958 	const struct got_error *err = NULL;
959 	const char *repo_abspath = NULL;
960 	size_t repolen, len;
961 	char *canonpath, *path = NULL;
962 
963 	*in_repo_path = NULL;
964 
965 	canonpath = strdup(input_path);
966 	if (canonpath == NULL) {
967 		err = got_error_from_errno("strdup");
968 		goto done;
969 	}
970 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
971 	if (err)
972 		goto done;
973 
974 	repo_abspath = got_repo_get_path(repo);
975 
976 	if (canonpath[0] == '\0') {
977 		path = strdup(canonpath);
978 		if (path == NULL) {
979 			err = got_error_from_errno("strdup");
980 			goto done;
981 		}
982 	} else {
983 		path = realpath(canonpath, NULL);
984 		if (path == NULL) {
985 			if (errno != ENOENT) {
986 				err = got_error_from_errno2("realpath",
987 				    canonpath);
988 				goto done;
989 			}
990 			/*
991 			 * Path is not on disk.
992 			 * Assume it is already relative to repository root.
993 			 */
994 			path = strdup(canonpath);
995 			if (path == NULL) {
996 				err = got_error_from_errno("strdup");
997 				goto done;
998 			}
999 		}
1000 
1001 		repolen = strlen(repo_abspath);
1002 		len = strlen(path);
1003 
1004 
1005 		if (strcmp(path, repo_abspath) == 0) {
1006 			free(path);
1007 			path = strdup("");
1008 			if (path == NULL) {
1009 				err = got_error_from_errno("strdup");
1010 				goto done;
1011 			}
1012 		} else if (len > repolen &&
1013 		    got_path_is_child(path, repo_abspath, repolen)) {
1014 			/* Matched an on-disk path inside repository. */
1015 			if (got_repo_is_bare(repo)) {
1016 				/*
1017 				 * Matched an on-disk path inside repository
1018 				 * database. Treat input as repository-relative.
1019 				 */
1020 				free(path);
1021 				path = canonpath;
1022 				canonpath = NULL;
1023 			} else {
1024 				char *child;
1025 				/* Strip common prefix with repository path. */
1026 				err = got_path_skip_common_ancestor(&child,
1027 				    repo_abspath, path);
1028 				if (err)
1029 					goto done;
1030 				free(path);
1031 				path = child;
1032 			}
1033 		} else {
1034 			/*
1035 			 * Matched unrelated on-disk path.
1036 			 * Treat input as repository-relative.
1037 			 */
1038 			free(path);
1039 			path = canonpath;
1040 			canonpath = NULL;
1041 		}
1042 	}
1043 
1044 	/* Make in-repository path absolute */
1045 	if (path[0] != '/') {
1046 		char *abspath;
1047 		if (asprintf(&abspath, "/%s", path) == -1) {
1048 			err = got_error_from_errno("asprintf");
1049 			goto done;
1050 		}
1051 		free(path);
1052 		path = abspath;
1053 	}
1054 
1055 done:
1056 	free(canonpath);
1057 	if (err)
1058 		free(path);
1059 	else
1060 		*in_repo_path = path;
1061 	return err;
1062 }
1063 
1064 static const struct got_error *
1065 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1066     const char *path_packidx)
1067 {
1068 	const struct got_error *err = NULL;
1069 	size_t i;
1070 
1071 	for (i = 0; i < repo->pack_cache_size; i++) {
1072 		if (repo->packidx_cache[i] == NULL)
1073 			break;
1074 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1075 		    path_packidx) == 0) {
1076 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1077 		}
1078 	}
1079 	if (i == repo->pack_cache_size) {
1080 		do {
1081 			i--;
1082 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1083 		    i == repo->pinned_packidx);
1084 		err = got_packidx_close(repo->packidx_cache[i]);
1085 		if (err)
1086 			return err;
1087 	}
1088 
1089 	repo->packidx_cache[i] = packidx;
1090 
1091 	return NULL;
1092 }
1093 
1094 int
1095 got_repo_is_packidx_filename(const char *name, size_t len)
1096 {
1097 	if (len != GOT_PACKIDX_NAMELEN)
1098 		return 0;
1099 
1100 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1101 		return 0;
1102 
1103 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1104 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1105 		return 0;
1106 
1107 	return 1;
1108 }
1109 
1110 static struct got_packidx_bloom_filter *
1111 get_packidx_bloom_filter(struct got_repository *repo,
1112     const char *path, size_t path_len)
1113 {
1114 	struct got_packidx_bloom_filter key;
1115 
1116 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1117 		return NULL; /* XXX */
1118 	key.path_len = path_len;
1119 
1120 	return RB_FIND(got_packidx_bloom_filter_tree,
1121 	    &repo->packidx_bloom_filters, &key);
1122 }
1123 
1124 int
1125 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1126     const char *path_packidx, struct got_object_id *id)
1127 {
1128 	struct got_packidx_bloom_filter *bf;
1129 
1130 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1131 	if (bf)
1132 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1133 
1134 	/* No bloom filter means this pack index must be searched. */
1135 	return 1;
1136 }
1137 
1138 static const struct got_error *
1139 add_packidx_bloom_filter(struct got_repository *repo,
1140     struct got_packidx *packidx, const char *path_packidx)
1141 {
1142 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1143 	struct got_packidx_bloom_filter *bf;
1144 	size_t len;
1145 
1146 	/*
1147 	 * Don't use bloom filters for very large pack index files.
1148 	 * Large pack files will contain a relatively large fraction
1149 	 * of our objects so we will likely need to visit them anyway.
1150 	 * The more objects a pack file contains the higher the probability
1151 	 * of a false-positive match from the bloom filter. And reading
1152 	 * all object IDs from a large pack index file can be expensive.
1153 	 */
1154 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1155 		return NULL;
1156 
1157 	/* Do we already have a filter for this pack index? */
1158 	if (get_packidx_bloom_filter(repo, path_packidx,
1159 	    strlen(path_packidx)) != NULL)
1160 		return NULL;
1161 
1162 	bf = calloc(1, sizeof(*bf));
1163 	if (bf == NULL)
1164 		return got_error_from_errno("calloc");
1165 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1166 	if (bf->bloom == NULL) {
1167 		free(bf);
1168 		return got_error_from_errno("calloc");
1169 	}
1170 
1171 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1172 	if (len >= sizeof(bf->path)) {
1173 		free(bf->bloom);
1174 		free(bf);
1175 		return got_error(GOT_ERR_NO_SPACE);
1176 	}
1177 	bf->path_len = len;
1178 
1179 	/* Minimum size supported by our bloom filter is 1000 entries. */
1180 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1181 	for (i = 0; i < nobjects; i++) {
1182 		struct got_packidx_object_id *id;
1183 		id = &packidx->hdr.sorted_ids[i];
1184 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1185 	}
1186 
1187 	RB_INSERT(got_packidx_bloom_filter_tree,
1188 	    &repo->packidx_bloom_filters, bf);
1189 	return NULL;
1190 }
1191 
1192 const struct got_error *
1193 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1194     struct got_repository *repo, struct got_object_id *id)
1195 {
1196 	const struct got_error *err;
1197 	struct got_pathlist_entry *pe;
1198 	size_t i;
1199 
1200 	/* Search pack index cache. */
1201 	for (i = 0; i < repo->pack_cache_size; i++) {
1202 		if (repo->packidx_cache[i] == NULL)
1203 			break;
1204 		if (!got_repo_check_packidx_bloom_filter(repo,
1205 		    repo->packidx_cache[i]->path_packidx, id))
1206 			continue; /* object will not be found in this index */
1207 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1208 		if (*idx != -1) {
1209 			*packidx = repo->packidx_cache[i];
1210 			/*
1211 			 * Move this cache entry to the front. Repeatedly
1212 			 * searching a wrong pack index can be expensive.
1213 			 */
1214 			if (i > 0) {
1215 				memmove(&repo->packidx_cache[1],
1216 				    &repo->packidx_cache[0],
1217 				    i * sizeof(repo->packidx_cache[0]));
1218 				repo->packidx_cache[0] = *packidx;
1219 				if (repo->pinned_packidx >= 0 &&
1220 				    repo->pinned_packidx < i)
1221 					repo->pinned_packidx++;
1222 				else if (repo->pinned_packidx == i)
1223 					repo->pinned_packidx = 0;
1224 			}
1225 			return NULL;
1226 		}
1227 	}
1228 	/* No luck. Search the filesystem. */
1229 
1230 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1231 		const char *path_packidx = pe->path;
1232 		int is_cached = 0;
1233 
1234 		if (!got_repo_check_packidx_bloom_filter(repo,
1235 		    pe->path, id))
1236 			continue; /* object will not be found in this index */
1237 
1238 		for (i = 0; i < repo->pack_cache_size; i++) {
1239 			if (repo->packidx_cache[i] == NULL)
1240 				break;
1241 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1242 			    path_packidx) == 0) {
1243 				is_cached = 1;
1244 				break;
1245 			}
1246 		}
1247 		if (is_cached)
1248 			continue; /* already searched */
1249 
1250 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1251 		    path_packidx, 0);
1252 		if (err)
1253 			goto done;
1254 
1255 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1256 		if (err)
1257 			goto done;
1258 
1259 		err = cache_packidx(repo, *packidx, path_packidx);
1260 		if (err)
1261 			goto done;
1262 
1263 		*idx = got_packidx_get_object_idx(*packidx, id);
1264 		if (*idx != -1) {
1265 			err = NULL; /* found the object */
1266 			goto done;
1267 		}
1268 	}
1269 
1270 	err = got_error_no_obj(id);
1271 done:
1272 	return err;
1273 }
1274 
1275 const struct got_error *
1276 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1277     struct got_repository *repo)
1278 {
1279 	const struct got_error *err = NULL;
1280 	DIR *packdir = NULL;
1281 	struct dirent *dent;
1282 	char *path_packidx = NULL;
1283 	int packdir_fd;
1284 
1285 	packdir_fd = openat(got_repo_get_fd(repo),
1286 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1287 	if (packdir_fd == -1) {
1288 		return got_error_from_errno_fmt("openat: %s/%s",
1289 		    got_repo_get_path_git_dir(repo),
1290 		    GOT_OBJECTS_PACK_DIR);
1291 	}
1292 
1293 	packdir = fdopendir(packdir_fd);
1294 	if (packdir == NULL) {
1295 		err = got_error_from_errno("fdopendir");
1296 		goto done;
1297 	}
1298 
1299 	while ((dent = readdir(packdir)) != NULL) {
1300 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1301 			continue;
1302 
1303 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1304 		    dent->d_name) == -1) {
1305 			err = got_error_from_errno("asprintf");
1306 			path_packidx = NULL;
1307 			break;
1308 		}
1309 
1310 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1311 		if (err)
1312 			break;
1313 	}
1314 done:
1315 	if (err)
1316 		free(path_packidx);
1317 	if (packdir && closedir(packdir) != 0 && err == NULL)
1318 		err = got_error_from_errno("closedir");
1319 	return err;
1320 }
1321 
1322 const struct got_error *
1323 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1324     struct got_repository *repo)
1325 {
1326 	const struct got_error *err;
1327 	size_t i;
1328 
1329 	*packidx = NULL;
1330 
1331 	/* Search pack index cache. */
1332 	for (i = 0; i < repo->pack_cache_size; i++) {
1333 		if (repo->packidx_cache[i] == NULL)
1334 			break;
1335 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1336 		    path_packidx) == 0) {
1337 			*packidx = repo->packidx_cache[i];
1338 			return NULL;
1339 		}
1340 	}
1341 	/* No luck. Search the filesystem. */
1342 
1343 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1344 	    path_packidx, 0);
1345 	if (err)
1346 		return err;
1347 
1348 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1349 	if (err)
1350 		goto done;
1351 
1352 	err = cache_packidx(repo, *packidx, path_packidx);
1353 done:
1354 	if (err) {
1355 		got_packidx_close(*packidx);
1356 		*packidx = NULL;
1357 	}
1358 	return err;
1359 }
1360 
1361 static const struct got_error *
1362 read_packfile_hdr(int fd, struct got_packidx *packidx)
1363 {
1364 	const struct got_error *err = NULL;
1365 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1366 	struct got_packfile_hdr hdr;
1367 	ssize_t n;
1368 
1369 	n = read(fd, &hdr, sizeof(hdr));
1370 	if (n < 0)
1371 		return got_error_from_errno("read");
1372 	if (n != sizeof(hdr))
1373 		return got_error(GOT_ERR_BAD_PACKFILE);
1374 
1375 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1376 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1377 	    be32toh(hdr.nobjects) != totobj)
1378 		err = got_error(GOT_ERR_BAD_PACKFILE);
1379 
1380 	return err;
1381 }
1382 
1383 static const struct got_error *
1384 open_packfile(int *fd, struct got_repository *repo,
1385     const char *relpath, struct got_packidx *packidx)
1386 {
1387 	const struct got_error *err = NULL;
1388 
1389 	*fd = openat(got_repo_get_fd(repo), relpath,
1390 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1391 	if (*fd == -1)
1392 		return got_error_from_errno_fmt("openat: %s/%s",
1393 		    got_repo_get_path_git_dir(repo), relpath);
1394 
1395 	if (packidx) {
1396 		err = read_packfile_hdr(*fd, packidx);
1397 		if (err) {
1398 			close(*fd);
1399 			*fd = -1;
1400 		}
1401 	}
1402 
1403 	return err;
1404 }
1405 
1406 const struct got_error *
1407 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1408     const char *path_packfile, struct got_packidx *packidx)
1409 {
1410 	const struct got_error *err = NULL;
1411 	struct got_pack *pack = NULL;
1412 	struct stat sb;
1413 	size_t i;
1414 
1415 	if (packp)
1416 		*packp = NULL;
1417 
1418 	for (i = 0; i < repo->pack_cache_size; i++) {
1419 		pack = &repo->packs[i];
1420 		if (pack->path_packfile == NULL)
1421 			break;
1422 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1423 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1424 	}
1425 
1426 	if (i == repo->pack_cache_size) {
1427 		struct got_pack tmp;
1428 		do {
1429 			i--;
1430 		} while (i > 0 && repo->pinned_pack >= 0 &&
1431 		    i == repo->pinned_pack);
1432 		err = got_pack_close(&repo->packs[i]);
1433 		if (err)
1434 			return err;
1435 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1436 			return got_error_from_errno("ftruncate");
1437 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1438 			return got_error_from_errno("ftruncate");
1439 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1440 		memcpy(&repo->packs[i], &repo->packs[0],
1441 		    sizeof(repo->packs[i]));
1442 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1443 		if (repo->pinned_pack == 0)
1444 			repo->pinned_pack = i;
1445 		else if (repo->pinned_pack == i)
1446 			repo->pinned_pack = 0;
1447 		i = 0;
1448 	}
1449 
1450 	pack = &repo->packs[i];
1451 
1452 	pack->path_packfile = strdup(path_packfile);
1453 	if (pack->path_packfile == NULL) {
1454 		err = got_error_from_errno("strdup");
1455 		goto done;
1456 	}
1457 
1458 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1459 	if (err)
1460 		goto done;
1461 
1462 	if (fstat(pack->fd, &sb) != 0) {
1463 		err = got_error_from_errno("fstat");
1464 		goto done;
1465 	}
1466 	pack->filesize = sb.st_size;
1467 
1468 	pack->privsep_child = NULL;
1469 
1470 #ifndef GOT_PACK_NO_MMAP
1471 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1472 	    pack->fd, 0);
1473 	if (pack->map == MAP_FAILED) {
1474 		if (errno != ENOMEM) {
1475 			err = got_error_from_errno("mmap");
1476 			goto done;
1477 		}
1478 		pack->map = NULL; /* fall back to read(2) */
1479 	}
1480 #endif
1481 done:
1482 	if (err) {
1483 		if (pack) {
1484 			free(pack->path_packfile);
1485 			memset(pack, 0, sizeof(*pack));
1486 		}
1487 	} else if (packp)
1488 		*packp = pack;
1489 	return err;
1490 }
1491 
1492 struct got_pack *
1493 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1494 {
1495 	struct got_pack *pack = NULL;
1496 	size_t i;
1497 
1498 	for (i = 0; i < repo->pack_cache_size; i++) {
1499 		pack = &repo->packs[i];
1500 		if (pack->path_packfile == NULL)
1501 			break;
1502 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1503 			return pack;
1504 	}
1505 
1506 	return NULL;
1507 }
1508 
1509 const struct got_error *
1510 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1511     struct got_pack *pack)
1512 {
1513 	size_t i;
1514 	int pinned_pack = -1, pinned_packidx = -1;
1515 
1516 	for (i = 0; i < repo->pack_cache_size; i++) {
1517 		if (repo->packidx_cache[i] &&
1518 		    strcmp(repo->packidx_cache[i]->path_packidx,
1519 		    packidx->path_packidx) == 0)
1520 			pinned_packidx = i;
1521 		if (repo->packs[i].path_packfile &&
1522 		    strcmp(repo->packs[i].path_packfile,
1523 		    pack->path_packfile) == 0)
1524 			pinned_pack = i;
1525 	}
1526 
1527 	if (pinned_packidx == -1 || pinned_pack == -1)
1528 		return got_error(GOT_ERR_PIN_PACK);
1529 
1530 	repo->pinned_pack = pinned_pack;
1531 	repo->pinned_packidx = pinned_packidx;
1532 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1533 	return NULL;
1534 }
1535 
1536 struct got_pack *
1537 got_repo_get_pinned_pack(struct got_repository *repo)
1538 {
1539 	if (repo->pinned_pack >= 0 &&
1540 	    repo->pinned_pack < repo->pack_cache_size)
1541 		return &repo->packs[repo->pinned_pack];
1542 
1543 	return NULL;
1544 }
1545 
1546 void
1547 got_repo_unpin_pack(struct got_repository *repo)
1548 {
1549 	repo->pinned_packidx = -1;
1550 	repo->pinned_pack = -1;
1551 	repo->pinned_pid = 0;
1552 }
1553 
1554 const struct got_error *
1555 got_repo_init(const char *repo_path)
1556 {
1557 	const struct got_error *err = NULL;
1558 	const char *dirnames[] = {
1559 		GOT_OBJECTS_DIR,
1560 		GOT_OBJECTS_PACK_DIR,
1561 		GOT_REFS_DIR,
1562 	};
1563 	const char *description_str = "Unnamed repository; "
1564 	    "edit this file 'description' to name the repository.";
1565 	const char *headref_str = "ref: refs/heads/main";
1566 	const char *gitconfig_str = "[core]\n"
1567 	    "\trepositoryformatversion = 0\n"
1568 	    "\tfilemode = true\n"
1569 	    "\tbare = true\n";
1570 	char *path;
1571 	size_t i;
1572 
1573 	if (!got_path_dir_is_empty(repo_path))
1574 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1575 
1576 	for (i = 0; i < nitems(dirnames); i++) {
1577 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1578 			return got_error_from_errno("asprintf");
1579 		}
1580 		err = got_path_mkdir(path);
1581 		free(path);
1582 		if (err)
1583 			return err;
1584 	}
1585 
1586 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1587 		return got_error_from_errno("asprintf");
1588 	err = got_path_create_file(path, description_str);
1589 	free(path);
1590 	if (err)
1591 		return err;
1592 
1593 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1594 		return got_error_from_errno("asprintf");
1595 	err = got_path_create_file(path, headref_str);
1596 	free(path);
1597 	if (err)
1598 		return err;
1599 
1600 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1601 		return got_error_from_errno("asprintf");
1602 	err = got_path_create_file(path, gitconfig_str);
1603 	free(path);
1604 	if (err)
1605 		return err;
1606 
1607 	return NULL;
1608 }
1609 
1610 static const struct got_error *
1611 match_packed_object(struct got_object_id **unique_id,
1612     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1613 {
1614 	const struct got_error *err = NULL;
1615 	struct got_object_id_queue matched_ids;
1616 	struct got_pathlist_entry *pe;
1617 
1618 	STAILQ_INIT(&matched_ids);
1619 
1620 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1621 		const char *path_packidx = pe->path;
1622 		struct got_packidx *packidx;
1623 		struct got_object_qid *qid;
1624 
1625 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1626 		    path_packidx, 0);
1627 		if (err)
1628 			break;
1629 
1630 		err = got_packidx_match_id_str_prefix(&matched_ids,
1631 		    packidx, id_str_prefix);
1632 		if (err) {
1633 			got_packidx_close(packidx);
1634 			break;
1635 		}
1636 		err = got_packidx_close(packidx);
1637 		if (err)
1638 			break;
1639 
1640 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1641 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1642 				int matched_type;
1643 				err = got_object_get_type(&matched_type, repo,
1644 				    &qid->id);
1645 				if (err)
1646 					goto done;
1647 				if (matched_type != obj_type)
1648 					continue;
1649 			}
1650 			if (*unique_id == NULL) {
1651 				*unique_id = got_object_id_dup(&qid->id);
1652 				if (*unique_id == NULL) {
1653 					err = got_error_from_errno("malloc");
1654 					goto done;
1655 				}
1656 			} else {
1657 				if (got_object_id_cmp(*unique_id,
1658 				    &qid->id) == 0)
1659 					continue; /* packed multiple times */
1660 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1661 				goto done;
1662 			}
1663 		}
1664 	}
1665 done:
1666 	got_object_id_queue_free(&matched_ids);
1667 	if (err) {
1668 		free(*unique_id);
1669 		*unique_id = NULL;
1670 	}
1671 	return err;
1672 }
1673 
1674 static const struct got_error *
1675 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1676     const char *object_dir, const char *id_str_prefix, int obj_type,
1677     struct got_repository *repo)
1678 {
1679 	const struct got_error *err = NULL;
1680 	char *path;
1681 	DIR *dir = NULL;
1682 	struct dirent *dent;
1683 	struct got_object_id id;
1684 
1685 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1686 		err = got_error_from_errno("asprintf");
1687 		goto done;
1688 	}
1689 
1690 	dir = opendir(path);
1691 	if (dir == NULL) {
1692 		if (errno == ENOENT) {
1693 			err = NULL;
1694 			goto done;
1695 		}
1696 		err = got_error_from_errno2("opendir", path);
1697 		goto done;
1698 	}
1699 	while ((dent = readdir(dir)) != NULL) {
1700 		char *id_str;
1701 		int cmp;
1702 
1703 		if (strcmp(dent->d_name, ".") == 0 ||
1704 		    strcmp(dent->d_name, "..") == 0)
1705 			continue;
1706 
1707 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1708 			err = got_error_from_errno("asprintf");
1709 			goto done;
1710 		}
1711 
1712 		if (!got_parse_sha1_digest(id.sha1, id_str))
1713 			continue;
1714 
1715 		/*
1716 		 * Directory entries do not necessarily appear in
1717 		 * sorted order, so we must iterate over all of them.
1718 		 */
1719 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1720 		if (cmp != 0) {
1721 			free(id_str);
1722 			continue;
1723 		}
1724 
1725 		if (*unique_id == NULL) {
1726 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1727 				int matched_type;
1728 				err = got_object_get_type(&matched_type, repo,
1729 				    &id);
1730 				if (err)
1731 					goto done;
1732 				if (matched_type != obj_type)
1733 					continue;
1734 			}
1735 			*unique_id = got_object_id_dup(&id);
1736 			if (*unique_id == NULL) {
1737 				err = got_error_from_errno("got_object_id_dup");
1738 				free(id_str);
1739 				goto done;
1740 			}
1741 		} else {
1742 			if (got_object_id_cmp(*unique_id, &id) == 0)
1743 				continue; /* both packed and loose */
1744 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1745 			free(id_str);
1746 			goto done;
1747 		}
1748 	}
1749 done:
1750 	if (dir && closedir(dir) != 0 && err == NULL)
1751 		err = got_error_from_errno("closedir");
1752 	if (err) {
1753 		free(*unique_id);
1754 		*unique_id = NULL;
1755 	}
1756 	free(path);
1757 	return err;
1758 }
1759 
1760 const struct got_error *
1761 got_repo_match_object_id_prefix(struct got_object_id **id,
1762     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1763 {
1764 	const struct got_error *err = NULL;
1765 	char *path_objects = got_repo_get_path_objects(repo);
1766 	char *object_dir = NULL;
1767 	size_t len;
1768 	int i;
1769 
1770 	*id = NULL;
1771 
1772 	len = strlen(id_str_prefix);
1773 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1774 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1775 
1776 	for (i = 0; i < len; i++) {
1777 		if (isxdigit((unsigned char)id_str_prefix[i]))
1778 			continue;
1779 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1780 	}
1781 
1782 	if (len >= 2) {
1783 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1784 		if (err)
1785 			goto done;
1786 		object_dir = strndup(id_str_prefix, 2);
1787 		if (object_dir == NULL) {
1788 			err = got_error_from_errno("strdup");
1789 			goto done;
1790 		}
1791 		err = match_loose_object(id, path_objects, object_dir,
1792 		    id_str_prefix, obj_type, repo);
1793 	} else if (len == 1) {
1794 		int i;
1795 		for (i = 0; i < 0xf; i++) {
1796 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1797 			    == -1) {
1798 				err = got_error_from_errno("asprintf");
1799 				goto done;
1800 			}
1801 			err = match_packed_object(id, repo, object_dir,
1802 			    obj_type);
1803 			if (err)
1804 				goto done;
1805 			err = match_loose_object(id, path_objects, object_dir,
1806 			    id_str_prefix, obj_type, repo);
1807 			if (err)
1808 				goto done;
1809 		}
1810 	} else {
1811 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1812 		goto done;
1813 	}
1814 done:
1815 	free(object_dir);
1816 	if (err) {
1817 		free(*id);
1818 		*id = NULL;
1819 	} else if (*id == NULL) {
1820 		switch (obj_type) {
1821 		case GOT_OBJ_TYPE_BLOB:
1822 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1823 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1824 			break;
1825 		case GOT_OBJ_TYPE_TREE:
1826 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1827 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1828 			break;
1829 		case GOT_OBJ_TYPE_COMMIT:
1830 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1831 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1832 			break;
1833 		case GOT_OBJ_TYPE_TAG:
1834 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1835 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1836 			break;
1837 		default:
1838 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1839 			break;
1840 		}
1841 	}
1842 
1843 	return err;
1844 }
1845 
1846 const struct got_error *
1847 got_repo_match_object_id(struct got_object_id **id, char **label,
1848     const char *id_str, int obj_type, struct got_reflist_head *refs,
1849     struct got_repository *repo)
1850 {
1851 	const struct got_error *err;
1852 	struct got_tag_object *tag;
1853 	struct got_reference *ref = NULL;
1854 
1855 	*id = NULL;
1856 	if (label)
1857 		*label = NULL;
1858 
1859 	if (refs) {
1860 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1861 		    refs, repo);
1862 		if (err == NULL) {
1863 			*id = got_object_id_dup(
1864 			    got_object_tag_get_object_id(tag));
1865 			if (*id == NULL)
1866 				err = got_error_from_errno("got_object_id_dup");
1867 			else if (label && asprintf(label, "refs/tags/%s",
1868 			    got_object_tag_get_name(tag)) == -1) {
1869 				err = got_error_from_errno("asprintf");
1870 				free(*id);
1871 				*id = NULL;
1872 			}
1873 			got_object_tag_close(tag);
1874 			return err;
1875 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1876 		    err->code != GOT_ERR_NO_OBJ)
1877 			return err;
1878 	}
1879 
1880 	err = got_ref_open(&ref, repo, id_str, 0);
1881 	if (err == NULL) {
1882 		err = got_ref_resolve(id, repo, ref);
1883 		if (err)
1884 			goto done;
1885 		if (label) {
1886 			*label = strdup(got_ref_get_name(ref));
1887 			if (*label == NULL) {
1888 				err = got_error_from_errno("strdup");
1889 				goto done;
1890 			}
1891 		}
1892 	} else {
1893 		if (err->code != GOT_ERR_NOT_REF &&
1894 		    err->code != GOT_ERR_BAD_REF_NAME)
1895 			goto done;
1896 		err = got_repo_match_object_id_prefix(id, id_str,
1897 		    obj_type, repo);
1898 		if (err) {
1899 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1900 				err = got_error_not_ref(id_str);
1901 			goto done;
1902 		}
1903 		if (label) {
1904 			err = got_object_id_str(label, *id);
1905 			if (*label == NULL) {
1906 				err = got_error_from_errno("strdup");
1907 				goto done;
1908 			}
1909 		}
1910 	}
1911 done:
1912 	if (ref)
1913 		got_ref_close(ref);
1914 	return err;
1915 }
1916 
1917 const struct got_error *
1918 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1919     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1920 {
1921 	const struct got_error *err = NULL;
1922 	struct got_reflist_entry *re;
1923 	struct got_object_id *tag_id;
1924 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1925 
1926 	*tag = NULL;
1927 
1928 	TAILQ_FOREACH(re, refs, entry) {
1929 		const char *refname;
1930 		refname = got_ref_get_name(re->ref);
1931 		if (got_ref_is_symbolic(re->ref))
1932 			continue;
1933 		if (strncmp(refname, "refs/tags/", 10) != 0)
1934 			continue;
1935 		if (!name_is_absolute)
1936 			refname += strlen("refs/tags/");
1937 		if (strcmp(refname, name) != 0)
1938 			continue;
1939 		err = got_ref_resolve(&tag_id, repo, re->ref);
1940 		if (err)
1941 			break;
1942 		err = got_object_open_as_tag(tag, repo, tag_id);
1943 		free(tag_id);
1944 		if (err)
1945 			break;
1946 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1947 		    got_object_tag_get_object_type(*tag) == obj_type)
1948 			break;
1949 		got_object_tag_close(*tag);
1950 		*tag = NULL;
1951 	}
1952 
1953 	if (err == NULL && *tag == NULL)
1954 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1955 		    GOT_OBJ_LABEL_TAG, name);
1956 	return err;
1957 }
1958 
1959 static const struct got_error *
1960 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1961     const char *name, mode_t mode, struct got_object_id *blob_id)
1962 {
1963 	const struct got_error *err = NULL;
1964 
1965 	 *new_te = NULL;
1966 
1967 	*new_te = calloc(1, sizeof(**new_te));
1968 	if (*new_te == NULL)
1969 		return got_error_from_errno("calloc");
1970 
1971 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1972 	    sizeof((*new_te)->name)) {
1973 		err = got_error(GOT_ERR_NO_SPACE);
1974 		goto done;
1975 	}
1976 
1977 	if (S_ISLNK(mode)) {
1978 		(*new_te)->mode = S_IFLNK;
1979 	} else {
1980 		(*new_te)->mode = S_IFREG;
1981 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1982 	}
1983 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1984 done:
1985 	if (err && *new_te) {
1986 		free(*new_te);
1987 		*new_te = NULL;
1988 	}
1989 	return err;
1990 }
1991 
1992 static const struct got_error *
1993 import_file(struct got_tree_entry **new_te, struct dirent *de,
1994     const char *path, struct got_repository *repo)
1995 {
1996 	const struct got_error *err;
1997 	struct got_object_id *blob_id = NULL;
1998 	char *filepath;
1999 	struct stat sb;
2000 
2001 	if (asprintf(&filepath, "%s%s%s", path,
2002 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2003 		return got_error_from_errno("asprintf");
2004 
2005 	if (lstat(filepath, &sb) != 0) {
2006 		err = got_error_from_errno2("lstat", path);
2007 		goto done;
2008 	}
2009 
2010 	err = got_object_blob_create(&blob_id, filepath, repo);
2011 	if (err)
2012 		goto done;
2013 
2014 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2015 	    blob_id);
2016 done:
2017 	free(filepath);
2018 	if (err)
2019 		free(blob_id);
2020 	return err;
2021 }
2022 
2023 static const struct got_error *
2024 insert_tree_entry(struct got_tree_entry *new_te,
2025     struct got_pathlist_head *paths)
2026 {
2027 	const struct got_error *err = NULL;
2028 	struct got_pathlist_entry *new_pe;
2029 
2030 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2031 	if (err)
2032 		return err;
2033 	if (new_pe == NULL)
2034 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2035 	return NULL;
2036 }
2037 
2038 static const struct got_error *write_tree(struct got_object_id **,
2039     const char *, struct got_pathlist_head *, struct got_repository *,
2040     got_repo_import_cb progress_cb, void *progress_arg);
2041 
2042 static const struct got_error *
2043 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2044     const char *path, struct got_pathlist_head *ignores,
2045     struct got_repository *repo,
2046     got_repo_import_cb progress_cb, void *progress_arg)
2047 {
2048 	const struct got_error *err;
2049 	struct got_object_id *id = NULL;
2050 	char *subdirpath;
2051 
2052 	if (asprintf(&subdirpath, "%s%s%s", path,
2053 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2054 		return got_error_from_errno("asprintf");
2055 
2056 	(*new_te) = calloc(1, sizeof(**new_te));
2057 	if (*new_te == NULL)
2058 		return got_error_from_errno("calloc");
2059 	(*new_te)->mode = S_IFDIR;
2060 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2061 	    sizeof((*new_te)->name)) {
2062 		err = got_error(GOT_ERR_NO_SPACE);
2063 		goto done;
2064 	}
2065 	err = write_tree(&id, subdirpath, ignores,  repo,
2066 	    progress_cb, progress_arg);
2067 	if (err)
2068 		goto done;
2069 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2070 
2071 done:
2072 	free(id);
2073 	free(subdirpath);
2074 	if (err) {
2075 		free(*new_te);
2076 		*new_te = NULL;
2077 	}
2078 	return err;
2079 }
2080 
2081 static const struct got_error *
2082 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2083     struct got_pathlist_head *ignores, struct got_repository *repo,
2084     got_repo_import_cb progress_cb, void *progress_arg)
2085 {
2086 	const struct got_error *err = NULL;
2087 	DIR *dir;
2088 	struct dirent *de;
2089 	int nentries;
2090 	struct got_tree_entry *new_te = NULL;
2091 	struct got_pathlist_head paths;
2092 	struct got_pathlist_entry *pe;
2093 
2094 	*new_tree_id = NULL;
2095 
2096 	TAILQ_INIT(&paths);
2097 
2098 	dir = opendir(path_dir);
2099 	if (dir == NULL) {
2100 		err = got_error_from_errno2("opendir", path_dir);
2101 		goto done;
2102 	}
2103 
2104 	nentries = 0;
2105 	while ((de = readdir(dir)) != NULL) {
2106 		int ignore = 0;
2107 		int type;
2108 
2109 		if (strcmp(de->d_name, ".") == 0 ||
2110 		    strcmp(de->d_name, "..") == 0)
2111 			continue;
2112 
2113 		TAILQ_FOREACH(pe, ignores, entry) {
2114 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2115 				ignore = 1;
2116 				break;
2117 			}
2118 		}
2119 		if (ignore)
2120 			continue;
2121 
2122 		err = got_path_dirent_type(&type, path_dir, de);
2123 		if (err)
2124 			goto done;
2125 
2126 		if (type == DT_DIR) {
2127 			err = import_subdir(&new_te, de, path_dir,
2128 			    ignores, repo, progress_cb, progress_arg);
2129 			if (err) {
2130 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2131 					goto done;
2132 				err = NULL;
2133 				continue;
2134 			}
2135 		} else if (type == DT_REG || type == DT_LNK) {
2136 			err = import_file(&new_te, de, path_dir, repo);
2137 			if (err)
2138 				goto done;
2139 		} else
2140 			continue;
2141 
2142 		err = insert_tree_entry(new_te, &paths);
2143 		if (err)
2144 			goto done;
2145 		nentries++;
2146 	}
2147 
2148 	if (TAILQ_EMPTY(&paths)) {
2149 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2150 		    "cannot create tree without any entries");
2151 		goto done;
2152 	}
2153 
2154 	TAILQ_FOREACH(pe, &paths, entry) {
2155 		struct got_tree_entry *te = pe->data;
2156 		char *path;
2157 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2158 			continue;
2159 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2160 			err = got_error_from_errno("asprintf");
2161 			goto done;
2162 		}
2163 		err = (*progress_cb)(progress_arg, path);
2164 		free(path);
2165 		if (err)
2166 			goto done;
2167 	}
2168 
2169 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2170 done:
2171 	if (dir)
2172 		closedir(dir);
2173 	got_pathlist_free(&paths);
2174 	return err;
2175 }
2176 
2177 const struct got_error *
2178 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2179     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2180     struct got_repository *repo, got_repo_import_cb progress_cb,
2181     void *progress_arg)
2182 {
2183 	const struct got_error *err;
2184 	struct got_object_id *new_tree_id;
2185 
2186 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2187 	    progress_cb, progress_arg);
2188 	if (err)
2189 		return err;
2190 
2191 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2192 	    author, time(NULL), author, time(NULL), logmsg, repo);
2193 	free(new_tree_id);
2194 	return err;
2195 }
2196 
2197 const struct got_error *
2198 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2199     struct got_repository *repo)
2200 {
2201 	const struct got_error *err = NULL;
2202 	char *path_objects = NULL, *path = NULL;
2203 	DIR *dir = NULL;
2204 	struct got_object_id id;
2205 	int i;
2206 
2207 	*nobjects = 0;
2208 	*ondisk_size = 0;
2209 
2210 	path_objects = got_repo_get_path_objects(repo);
2211 	if (path_objects == NULL)
2212 		return got_error_from_errno("got_repo_get_path_objects");
2213 
2214 	for (i = 0; i <= 0xff; i++) {
2215 		struct dirent *dent;
2216 
2217 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2218 			err = got_error_from_errno("asprintf");
2219 			break;
2220 		}
2221 
2222 		dir = opendir(path);
2223 		if (dir == NULL) {
2224 			if (errno == ENOENT) {
2225 				err = NULL;
2226 				continue;
2227 			}
2228 			err = got_error_from_errno2("opendir", path);
2229 			break;
2230 		}
2231 
2232 		while ((dent = readdir(dir)) != NULL) {
2233 			char *id_str;
2234 			int fd;
2235 			struct stat sb;
2236 
2237 			if (strcmp(dent->d_name, ".") == 0 ||
2238 			    strcmp(dent->d_name, "..") == 0)
2239 				continue;
2240 
2241 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2242 				err = got_error_from_errno("asprintf");
2243 				goto done;
2244 			}
2245 
2246 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2247 				free(id_str);
2248 				continue;
2249 			}
2250 			free(id_str);
2251 
2252 			err = got_object_open_loose_fd(&fd, &id, repo);
2253 			if (err)
2254 				goto done;
2255 
2256 			if (fstat(fd, &sb) == -1) {
2257 				err = got_error_from_errno("fstat");
2258 				close(fd);
2259 				goto done;
2260 			}
2261 			(*nobjects)++;
2262 			(*ondisk_size) += sb.st_size;
2263 
2264 			if (close(fd) == -1) {
2265 				err = got_error_from_errno("close");
2266 				goto done;
2267 			}
2268 		}
2269 
2270 		if (closedir(dir) != 0) {
2271 			err = got_error_from_errno("closedir");
2272 			goto done;
2273 		}
2274 		dir = NULL;
2275 
2276 		free(path);
2277 		path = NULL;
2278 	}
2279 done:
2280 	if (dir && closedir(dir) != 0 && err == NULL)
2281 		err = got_error_from_errno("closedir");
2282 
2283 	if (err) {
2284 		*nobjects = 0;
2285 		*ondisk_size = 0;
2286 	}
2287 	free(path_objects);
2288 	free(path);
2289 	return err;
2290 }
2291 
2292 const struct got_error *
2293 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2294     off_t *total_packsize, struct got_repository *repo)
2295 {
2296 	const struct got_error *err = NULL;
2297 	DIR *packdir = NULL;
2298 	struct dirent *dent;
2299 	struct got_packidx *packidx = NULL;
2300 	char *path_packidx;
2301 	char *path_packfile;
2302 	int packdir_fd;
2303 	struct stat sb;
2304 
2305 	*npackfiles = 0;
2306 	*nobjects = 0;
2307 	*total_packsize = 0;
2308 
2309 	packdir_fd = openat(got_repo_get_fd(repo),
2310 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2311 	if (packdir_fd == -1) {
2312 		return got_error_from_errno_fmt("openat: %s/%s",
2313 		    got_repo_get_path_git_dir(repo),
2314 		    GOT_OBJECTS_PACK_DIR);
2315 	}
2316 
2317 	packdir = fdopendir(packdir_fd);
2318 	if (packdir == NULL) {
2319 		err = got_error_from_errno("fdopendir");
2320 		goto done;
2321 	}
2322 
2323 	while ((dent = readdir(packdir)) != NULL) {
2324 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2325 			continue;
2326 
2327 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2328 		    dent->d_name) == -1) {
2329 			err = got_error_from_errno("asprintf");
2330 			goto done;
2331 		}
2332 
2333 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2334 		    path_packidx, 0);
2335 		free(path_packidx);
2336 		if (err)
2337 			goto done;
2338 
2339 		if (fstat(packidx->fd, &sb) == -1)
2340 			goto done;
2341 		*total_packsize += sb.st_size;
2342 
2343 		err = got_packidx_get_packfile_path(&path_packfile,
2344 		    packidx->path_packidx);
2345 		if (err)
2346 			goto done;
2347 
2348 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2349 		    0) == -1) {
2350 			free(path_packfile);
2351 			goto done;
2352 		}
2353 		free(path_packfile);
2354 		*total_packsize += sb.st_size;
2355 
2356 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2357 
2358 		(*npackfiles)++;
2359 
2360 		got_packidx_close(packidx);
2361 		packidx = NULL;
2362 	}
2363 done:
2364 	if (packidx)
2365 		got_packidx_close(packidx);
2366 	if (packdir && closedir(packdir) != 0 && err == NULL)
2367 		err = got_error_from_errno("closedir");
2368 	if (err) {
2369 		*npackfiles = 0;
2370 		*nobjects = 0;
2371 		*total_packsize = 0;
2372 	}
2373 	return err;
2374 }
2375 
2376 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2377     got_packidx_bloom_filter_cmp);