1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/socket.h>
21 #include <sys/stat.h>
22 #include <sys/mman.h>
23 #include <sys/resource.h>
24 
25 #include <ctype.h>
26 #include <fcntl.h>
27 #include <fnmatch.h>
28 #include <limits.h>
29 #include <dirent.h>
30 #include <stdlib.h>
31 #include <stdio.h>
32 #include <string.h>
33 #include <time.h>
34 #include <unistd.h>
35 #include <zlib.h>
36 #include <errno.h>
37 #include <libgen.h>
38 #include <stdint.h>
39 
40 #include "bloom.h"
41 
42 #include "got_error.h"
43 #include "got_reference.h"
44 #include "got_repository.h"
45 #include "got_path.h"
46 #include "got_cancel.h"
47 #include "got_object.h"
48 #include "got_opentemp.h"
49 
50 #include "got_lib_delta.h"
51 #include "got_lib_inflate.h"
52 #include "got_lib_object.h"
53 #include "got_lib_object_parse.h"
54 #include "got_lib_object_create.h"
55 #include "got_lib_pack.h"
56 #include "got_lib_privsep.h"
57 #include "got_lib_sha1.h"
58 #include "got_lib_object_cache.h"
59 #include "got_lib_repository.h"
60 #include "got_lib_gotconfig.h"
61 
62 #ifndef nitems
63 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
64 #endif
65 
66 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
67 
68 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
69     got_packidx_bloom_filter_cmp);
70 
71 const char *
72 got_repo_get_path(struct got_repository *repo)
73 {
74 	return repo->path;
75 }
76 
77 const char *
78 got_repo_get_path_git_dir(struct got_repository *repo)
79 {
80 	return repo->path_git_dir;
81 }
82 
83 int
84 got_repo_get_fd(struct got_repository *repo)
85 {
86 	return repo->gitdir_fd;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_name(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_name;
93 }
94 
95 const char *
96 got_repo_get_gitconfig_author_email(struct got_repository *repo)
97 {
98 	return repo->gitconfig_author_email;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_name;
105 }
106 
107 const char *
108 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
109 {
110 	return repo->global_gitconfig_author_email;
111 }
112 
113 const char *
114 got_repo_get_gitconfig_owner(struct got_repository *repo)
115 {
116 	return repo->gitconfig_owner;
117 }
118 
119 void
120 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
121     struct got_repository *repo)
122 {
123 	*extensions = repo->extensions;
124 	*nextensions = repo->nextensions;
125 }
126 
127 int
128 got_repo_is_bare(struct got_repository *repo)
129 {
130 	return (strcmp(repo->path, repo->path_git_dir) == 0);
131 }
132 
133 static char *
134 get_path_git_child(struct got_repository *repo, const char *basename)
135 {
136 	char *path_child;
137 
138 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
139 	    basename) == -1)
140 		return NULL;
141 
142 	return path_child;
143 }
144 
145 char *
146 got_repo_get_path_objects(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
149 }
150 
151 char *
152 got_repo_get_path_objects_pack(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
155 }
156 
157 char *
158 got_repo_get_path_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_REFS_DIR);
161 }
162 
163 char *
164 got_repo_get_path_packed_refs(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
167 }
168 
169 static char *
170 get_path_head(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_HEAD_FILE);
173 }
174 
175 char *
176 got_repo_get_path_gitconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GITCONFIG);
179 }
180 
181 char *
182 got_repo_get_path_gotconfig(struct got_repository *repo)
183 {
184 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
185 }
186 
187 const struct got_gotconfig *
188 got_repo_get_gotconfig(struct got_repository *repo)
189 {
190 	return repo->gotconfig;
191 }
192 
193 void
194 got_repo_get_gitconfig_remotes(int *nremotes,
195     const struct got_remote_repo **remotes, struct got_repository *repo)
196 {
197 	*nremotes = repo->ngitconfig_remotes;
198 	*remotes = repo->gitconfig_remotes;
199 }
200 
201 static int
202 is_git_repo(struct got_repository *repo)
203 {
204 	const char *path_git = got_repo_get_path_git_dir(repo);
205 	char *path_objects = got_repo_get_path_objects(repo);
206 	char *path_refs = got_repo_get_path_refs(repo);
207 	char *path_head = get_path_head(repo);
208 	int ret = 0;
209 	struct stat sb;
210 	struct got_reference *head_ref;
211 
212 	if (lstat(path_git, &sb) == -1)
213 		goto done;
214 	if (!S_ISDIR(sb.st_mode))
215 		goto done;
216 
217 	if (lstat(path_objects, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_refs, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_head, &sb) == -1)
228 		goto done;
229 	if (!S_ISREG(sb.st_mode))
230 		goto done;
231 
232 	/* Check if the HEAD reference can be opened. */
233 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
234 		goto done;
235 	got_ref_close(head_ref);
236 
237 	ret = 1;
238 done:
239 	free(path_objects);
240 	free(path_refs);
241 	free(path_head);
242 	return ret;
243 
244 }
245 
246 const struct got_error *
247 got_repo_pack_fds_open(int **pack_fds)
248 {
249 	const struct got_error *err = NULL;
250 	int i;
251 
252 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
253 	if (*pack_fds == NULL)
254 		return got_error_from_errno("calloc");
255 
256 	/*
257 	 * got_repo_pack_fds_close will try to close all of the
258 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
259 	 * a value from got_opentempfd(), which would result in a close(0) if
260 	 * we do not initialize to -1 here.
261 	 */
262 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
263 		(*pack_fds)[i] = -1;
264 
265 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
266 		(*pack_fds)[i] = got_opentempfd();
267 		if ((*pack_fds)[i] == -1) {
268 			err = got_error_from_errno("got_opentempfd");
269 			got_repo_pack_fds_close(*pack_fds);
270 			*pack_fds = NULL;
271 			return err;
272 		}
273 	}
274 
275 	return NULL;
276 }
277 
278 const struct got_error *
279 got_repo_pack_fds_close(int *pack_fds)
280 {
281 	const struct got_error *err = NULL;
282 	int i;
283 
284 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
285 		if (pack_fds[i] == -1)
286 			continue;
287 		if (close(pack_fds[i]) == -1) {
288 			err = got_error_from_errno("close");
289 			break;
290 		}
291 	}
292 	free(pack_fds);
293 	return err;
294 }
295 
296 const struct got_error *
297 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
298     struct got_object *obj)
299 {
300 #ifndef GOT_NO_OBJ_CACHE
301 	const struct got_error *err = NULL;
302 	err = got_object_cache_add(&repo->objcache, id, obj);
303 	if (err) {
304 		if (err->code == GOT_ERR_OBJ_EXISTS ||
305 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
306 			err = NULL;
307 		return err;
308 	}
309 	obj->refcnt++;
310 #endif
311 	return NULL;
312 }
313 
314 struct got_object *
315 got_repo_get_cached_object(struct got_repository *repo,
316     struct got_object_id *id)
317 {
318 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
319 }
320 
321 const struct got_error *
322 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
323     struct got_tree_object *tree)
324 {
325 #ifndef GOT_NO_OBJ_CACHE
326 	const struct got_error *err = NULL;
327 	err = got_object_cache_add(&repo->treecache, id, tree);
328 	if (err) {
329 		if (err->code == GOT_ERR_OBJ_EXISTS ||
330 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
331 			err = NULL;
332 		return err;
333 	}
334 	tree->refcnt++;
335 #endif
336 	return NULL;
337 }
338 
339 struct got_tree_object *
340 got_repo_get_cached_tree(struct got_repository *repo,
341     struct got_object_id *id)
342 {
343 	return (struct got_tree_object *)got_object_cache_get(
344 	    &repo->treecache, id);
345 }
346 
347 const struct got_error *
348 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
349     struct got_commit_object *commit)
350 {
351 #ifndef GOT_NO_OBJ_CACHE
352 	const struct got_error *err = NULL;
353 	err = got_object_cache_add(&repo->commitcache, id, commit);
354 	if (err) {
355 		if (err->code == GOT_ERR_OBJ_EXISTS ||
356 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
357 			err = NULL;
358 		return err;
359 	}
360 	commit->refcnt++;
361 #endif
362 	return NULL;
363 }
364 
365 struct got_commit_object *
366 got_repo_get_cached_commit(struct got_repository *repo,
367     struct got_object_id *id)
368 {
369 	return (struct got_commit_object *)got_object_cache_get(
370 	    &repo->commitcache, id);
371 }
372 
373 const struct got_error *
374 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
375     struct got_tag_object *tag)
376 {
377 #ifndef GOT_NO_OBJ_CACHE
378 	const struct got_error *err = NULL;
379 	err = got_object_cache_add(&repo->tagcache, id, tag);
380 	if (err) {
381 		if (err->code == GOT_ERR_OBJ_EXISTS ||
382 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
383 			err = NULL;
384 		return err;
385 	}
386 	tag->refcnt++;
387 #endif
388 	return NULL;
389 }
390 
391 struct got_tag_object *
392 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
393 {
394 	return (struct got_tag_object *)got_object_cache_get(
395 	    &repo->tagcache, id);
396 }
397 
398 const struct got_error *
399 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
400     struct got_raw_object *raw)
401 {
402 #ifndef GOT_NO_OBJ_CACHE
403 	const struct got_error *err = NULL;
404 	err = got_object_cache_add(&repo->rawcache, id, raw);
405 	if (err) {
406 		if (err->code == GOT_ERR_OBJ_EXISTS ||
407 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
408 			err = NULL;
409 		return err;
410 	}
411 	raw->refcnt++;
412 #endif
413 	return NULL;
414 }
415 
416 
417 struct got_raw_object *
418 got_repo_get_cached_raw_object(struct got_repository *repo,
419     struct got_object_id *id)
420 {
421 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
422 }
423 
424 
425 static const struct got_error *
426 open_repo(struct got_repository *repo, const char *path)
427 {
428 	const struct got_error *err = NULL;
429 
430 	repo->gitdir_fd = -1;
431 
432 	/* bare git repository? */
433 	repo->path_git_dir = strdup(path);
434 	if (repo->path_git_dir == NULL)
435 		return got_error_from_errno("strdup");
436 	if (is_git_repo(repo)) {
437 		repo->path = strdup(repo->path_git_dir);
438 		if (repo->path == NULL) {
439 			err = got_error_from_errno("strdup");
440 			goto done;
441 		}
442 		repo->gitdir_fd = open(repo->path_git_dir,
443 		    O_DIRECTORY | O_CLOEXEC);
444 		if (repo->gitdir_fd == -1) {
445 			err = got_error_from_errno2("open",
446 			    repo->path_git_dir);
447 			goto done;
448 		}
449 		return NULL;
450 	}
451 
452 	/* git repository with working tree? */
453 	free(repo->path_git_dir);
454 	repo->path_git_dir = NULL;
455 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
456 		err = got_error_from_errno("asprintf");
457 		goto done;
458 	}
459 	if (is_git_repo(repo)) {
460 		repo->path = strdup(path);
461 		if (repo->path == NULL) {
462 			err = got_error_from_errno("strdup");
463 			goto done;
464 		}
465 		repo->gitdir_fd = open(repo->path_git_dir,
466 		    O_DIRECTORY | O_CLOEXEC);
467 		if (repo->gitdir_fd == -1) {
468 			err = got_error_from_errno2("open",
469 			    repo->path_git_dir);
470 			goto done;
471 		}
472 		return NULL;
473 	}
474 
475 	err = got_error(GOT_ERR_NOT_GIT_REPO);
476 done:
477 	if (err) {
478 		free(repo->path);
479 		repo->path = NULL;
480 		free(repo->path_git_dir);
481 		repo->path_git_dir = NULL;
482 		if (repo->gitdir_fd != -1)
483 			close(repo->gitdir_fd);
484 		repo->gitdir_fd = -1;
485 
486 	}
487 	return err;
488 }
489 
490 static const struct got_error *
491 parse_gitconfig_file(int *gitconfig_repository_format_version,
492     char **gitconfig_author_name, char **gitconfig_author_email,
493     struct got_remote_repo **remotes, int *nremotes,
494     char **gitconfig_owner, char ***extensions, int *nextensions,
495     const char *gitconfig_path)
496 {
497 	const struct got_error *err = NULL, *child_err = NULL;
498 	int fd = -1;
499 	int imsg_fds[2] = { -1, -1 };
500 	pid_t pid;
501 	struct imsgbuf *ibuf;
502 
503 	*gitconfig_repository_format_version = 0;
504 	if (extensions)
505 		*extensions = NULL;
506 	if (nextensions)
507 		*nextensions = 0;
508 	*gitconfig_author_name = NULL;
509 	*gitconfig_author_email = NULL;
510 	if (remotes)
511 		*remotes = NULL;
512 	if (nremotes)
513 		*nremotes = 0;
514 	if (gitconfig_owner)
515 		*gitconfig_owner = NULL;
516 
517 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
518 	if (fd == -1) {
519 		if (errno == ENOENT)
520 			return NULL;
521 		return got_error_from_errno2("open", gitconfig_path);
522 	}
523 
524 	ibuf = calloc(1, sizeof(*ibuf));
525 	if (ibuf == NULL) {
526 		err = got_error_from_errno("calloc");
527 		goto done;
528 	}
529 
530 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
531 		err = got_error_from_errno("socketpair");
532 		goto done;
533 	}
534 
535 	pid = fork();
536 	if (pid == -1) {
537 		err = got_error_from_errno("fork");
538 		goto done;
539 	} else if (pid == 0) {
540 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
541 		    gitconfig_path);
542 		/* not reached */
543 	}
544 
545 	if (close(imsg_fds[1]) == -1) {
546 		err = got_error_from_errno("close");
547 		goto done;
548 	}
549 	imsg_fds[1] = -1;
550 	imsg_init(ibuf, imsg_fds[0]);
551 
552 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
553 	if (err)
554 		goto done;
555 	fd = -1;
556 
557 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
558 	if (err)
559 		goto done;
560 
561 	err = got_privsep_recv_gitconfig_int(
562 	    gitconfig_repository_format_version, ibuf);
563 	if (err)
564 		goto done;
565 
566 	if (extensions && nextensions) {
567 		err = got_privsep_send_gitconfig_repository_extensions_req(
568 		    ibuf);
569 		if (err)
570 			goto done;
571 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
572 		if (err)
573 			goto done;
574 		if (*nextensions > 0) {
575 			int i;
576 			*extensions = calloc(*nextensions, sizeof(char *));
577 			if (*extensions == NULL) {
578 				err = got_error_from_errno("calloc");
579 				goto done;
580 			}
581 			for (i = 0; i < *nextensions; i++) {
582 				char *ext;
583 				err = got_privsep_recv_gitconfig_str(&ext,
584 				    ibuf);
585 				if (err)
586 					goto done;
587 				(*extensions)[i] = ext;
588 			}
589 		}
590 	}
591 
592 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
593 	if (err)
594 		goto done;
595 
596 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
597 	if (err)
598 		goto done;
599 
600 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
601 	if (err)
602 		goto done;
603 
604 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
605 	if (err)
606 		goto done;
607 
608 	if (remotes && nremotes) {
609 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
610 		if (err)
611 			goto done;
612 
613 		err = got_privsep_recv_gitconfig_remotes(remotes,
614 		    nremotes, ibuf);
615 		if (err)
616 			goto done;
617 	}
618 
619 	if (gitconfig_owner) {
620 		err = got_privsep_send_gitconfig_owner_req(ibuf);
621 		if (err)
622 			goto done;
623 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
624 		if (err)
625 			goto done;
626 	}
627 
628 	err = got_privsep_send_stop(imsg_fds[0]);
629 	child_err = got_privsep_wait_for_child(pid);
630 	if (child_err && err == NULL)
631 		err = child_err;
632 done:
633 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
634 		err = got_error_from_errno("close");
635 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
636 		err = got_error_from_errno("close");
637 	if (fd != -1 && close(fd) == -1 && err == NULL)
638 		err = got_error_from_errno2("close", gitconfig_path);
639 	free(ibuf);
640 	return err;
641 }
642 
643 static const struct got_error *
644 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
645 {
646 	const struct got_error *err = NULL;
647 	char *repo_gitconfig_path = NULL;
648 
649 	if (global_gitconfig_path) {
650 		/* Read settings from ~/.gitconfig. */
651 		int dummy_repo_version;
652 		err = parse_gitconfig_file(&dummy_repo_version,
653 		    &repo->global_gitconfig_author_name,
654 		    &repo->global_gitconfig_author_email,
655 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
656 		if (err)
657 			return err;
658 	}
659 
660 	/* Read repository's .git/config file. */
661 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
662 	if (repo_gitconfig_path == NULL)
663 		return got_error_from_errno("got_repo_get_path_gitconfig");
664 
665 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
666 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
667 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
668 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
669 	    repo_gitconfig_path);
670 	if (err)
671 		goto done;
672 
673 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
674 		int i;
675 
676 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
677 			got_repo_free_remote_repo_data(
678 			    &repo->gitconfig_remotes[i]);
679 		}
680 		free(repo->gitconfig_remotes);
681 		repo->gitconfig_remotes = NULL;
682 		repo->ngitconfig_remotes = 0;
683 
684 		free(repo->gitconfig_author_name);
685 		repo->gitconfig_author_name = NULL;
686 		free(repo->gitconfig_author_email);
687 		repo->gitconfig_author_email = NULL;
688 
689 		free(repo->global_gitconfig_author_name);
690 		repo->global_gitconfig_author_name = NULL;
691 		free(repo->global_gitconfig_author_email);
692 		repo->global_gitconfig_author_email = NULL;
693 	}
694 
695 done:
696 	free(repo_gitconfig_path);
697 	return err;
698 }
699 
700 static const struct got_error *
701 read_gotconfig(struct got_repository *repo)
702 {
703 	const struct got_error *err = NULL;
704 	char *gotconfig_path;
705 
706 	gotconfig_path = got_repo_get_path_gotconfig(repo);
707 	if (gotconfig_path == NULL)
708 		return got_error_from_errno("got_repo_get_path_gotconfig");
709 
710 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
711 	free(gotconfig_path);
712 	return err;
713 }
714 
715 /* Supported repository format extensions. */
716 static const char *const repo_extensions[] = {
717 	"noop",			/* Got supports repository format version 1. */
718 	"preciousObjects",	/* Supported by gotadmin cleanup. */
719 	"worktreeConfig",	/* Got does not care about Git work trees. */
720 };
721 
722 const struct got_error *
723 got_repo_open(struct got_repository **repop, const char *path,
724     const char *global_gitconfig_path, int *pack_fds)
725 {
726 	struct got_repository *repo = NULL;
727 	const struct got_error *err = NULL;
728 	char *repo_path = NULL;
729 	size_t i, j = 0;
730 	struct rlimit rl;
731 
732 	*repop = NULL;
733 
734 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
735 		return got_error_from_errno("getrlimit");
736 
737 	repo = calloc(1, sizeof(*repo));
738 	if (repo == NULL)
739 		return got_error_from_errno("calloc");
740 
741 	RB_INIT(&repo->packidx_bloom_filters);
742 	TAILQ_INIT(&repo->packidx_paths);
743 
744 	for (i = 0; i < nitems(repo->privsep_children); i++) {
745 		memset(&repo->privsep_children[i], 0,
746 		    sizeof(repo->privsep_children[0]));
747 		repo->privsep_children[i].imsg_fd = -1;
748 	}
749 
750 	err = got_object_cache_init(&repo->objcache,
751 	    GOT_OBJECT_CACHE_TYPE_OBJ);
752 	if (err)
753 		goto done;
754 	err = got_object_cache_init(&repo->treecache,
755 	    GOT_OBJECT_CACHE_TYPE_TREE);
756 	if (err)
757 		goto done;
758 	err = got_object_cache_init(&repo->commitcache,
759 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
760 	if (err)
761 		goto done;
762 	err = got_object_cache_init(&repo->tagcache,
763 	    GOT_OBJECT_CACHE_TYPE_TAG);
764 	if (err)
765 		goto done;
766 	err = got_object_cache_init(&repo->rawcache,
767 	    GOT_OBJECT_CACHE_TYPE_RAW);
768 	if (err)
769 		goto done;
770 
771 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
772 	if (repo->pack_cache_size > rl.rlim_cur / 8)
773 		repo->pack_cache_size = rl.rlim_cur / 8;
774 	for (i = 0; i < nitems(repo->packs); i++) {
775 		if (i < repo->pack_cache_size) {
776 			repo->packs[i].basefd = pack_fds[j++];
777 			repo->packs[i].accumfd = pack_fds[j++];
778 		} else {
779 			repo->packs[i].basefd = -1;
780 			repo->packs[i].accumfd = -1;
781 		}
782 	}
783 	repo->pinned_pack = -1;
784 	repo->pinned_packidx = -1;
785 	repo->pinned_pid = 0;
786 
787 	repo_path = realpath(path, NULL);
788 	if (repo_path == NULL) {
789 		err = got_error_from_errno2("realpath", path);
790 		goto done;
791 	}
792 
793 	for (;;) {
794 		char *parent_path;
795 
796 		err = open_repo(repo, repo_path);
797 		if (err == NULL)
798 			break;
799 		if (err->code != GOT_ERR_NOT_GIT_REPO)
800 			goto done;
801 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
802 			err = got_error(GOT_ERR_NOT_GIT_REPO);
803 			goto done;
804 		}
805 		err = got_path_dirname(&parent_path, repo_path);
806 		if (err)
807 			goto done;
808 		free(repo_path);
809 		repo_path = parent_path;
810 	}
811 
812 	err = read_gotconfig(repo);
813 	if (err)
814 		goto done;
815 
816 	err = read_gitconfig(repo, global_gitconfig_path);
817 	if (err)
818 		goto done;
819 	if (repo->gitconfig_repository_format_version != 0) {
820 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
821 		goto done;
822 	}
823 	for (i = 0; i < repo->nextensions; i++) {
824 		char *ext = repo->extensions[i];
825 		int j, supported = 0;
826 		for (j = 0; j < nitems(repo_extensions); j++) {
827 			if (strcmp(ext, repo_extensions[j]) == 0) {
828 				supported = 1;
829 				break;
830 			}
831 		}
832 		if (!supported) {
833 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
834 			goto done;
835 		}
836 	}
837 
838 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
839 done:
840 	if (err)
841 		got_repo_close(repo);
842 	else
843 		*repop = repo;
844 	free(repo_path);
845 	return err;
846 }
847 
848 const struct got_error *
849 got_repo_close(struct got_repository *repo)
850 {
851 	const struct got_error *err = NULL, *child_err;
852 	struct got_packidx_bloom_filter *bf;
853 	struct got_pathlist_entry *pe;
854 	size_t i;
855 
856 	for (i = 0; i < repo->pack_cache_size; i++) {
857 		if (repo->packidx_cache[i] == NULL)
858 			break;
859 		got_packidx_close(repo->packidx_cache[i]);
860 	}
861 
862 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
863 	    &repo->packidx_bloom_filters))) {
864 		RB_REMOVE(got_packidx_bloom_filter_tree,
865 		    &repo->packidx_bloom_filters, bf);
866 		bloom_free(bf->bloom);
867 		free(bf->bloom);
868 		free(bf);
869 	}
870 
871 	for (i = 0; i < repo->pack_cache_size; i++)
872 		if (repo->packs[i].path_packfile)
873 			if (repo->packs[i].path_packfile)
874 				got_pack_close(&repo->packs[i]);
875 
876 	free(repo->path);
877 	free(repo->path_git_dir);
878 
879 	got_object_cache_close(&repo->objcache);
880 	got_object_cache_close(&repo->treecache);
881 	got_object_cache_close(&repo->commitcache);
882 	got_object_cache_close(&repo->tagcache);
883 	got_object_cache_close(&repo->rawcache);
884 
885 	for (i = 0; i < nitems(repo->privsep_children); i++) {
886 		if (repo->privsep_children[i].imsg_fd == -1)
887 			continue;
888 		imsg_clear(repo->privsep_children[i].ibuf);
889 		free(repo->privsep_children[i].ibuf);
890 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
891 		child_err = got_privsep_wait_for_child(
892 		    repo->privsep_children[i].pid);
893 		if (child_err && err == NULL)
894 			err = child_err;
895 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
896 		    err == NULL)
897 			err = got_error_from_errno("close");
898 	}
899 
900 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
901 	    err == NULL)
902 		err = got_error_from_errno("close");
903 
904 	if (repo->gotconfig)
905 		got_gotconfig_free(repo->gotconfig);
906 	free(repo->gitconfig_author_name);
907 	free(repo->gitconfig_author_email);
908 	for (i = 0; i < repo->ngitconfig_remotes; i++)
909 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
910 	free(repo->gitconfig_remotes);
911 	for (i = 0; i < repo->nextensions; i++)
912 		free(repo->extensions[i]);
913 	free(repo->extensions);
914 
915 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
916 		free((void *)pe->path);
917 	got_pathlist_free(&repo->packidx_paths);
918 	free(repo);
919 
920 	return err;
921 }
922 
923 void
924 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
925 {
926 	int i;
927 
928 	free(repo->name);
929 	repo->name = NULL;
930 	free(repo->fetch_url);
931 	repo->fetch_url = NULL;
932 	free(repo->send_url);
933 	repo->send_url = NULL;
934 	for (i = 0; i < repo->nfetch_branches; i++)
935 		free(repo->fetch_branches[i]);
936 	free(repo->fetch_branches);
937 	repo->fetch_branches = NULL;
938 	repo->nfetch_branches = 0;
939 	for (i = 0; i < repo->nsend_branches; i++)
940 		free(repo->send_branches[i]);
941 	free(repo->send_branches);
942 	repo->send_branches = NULL;
943 	repo->nsend_branches = 0;
944 }
945 
946 const struct got_error *
947 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
948     const char *input_path)
949 {
950 	const struct got_error *err = NULL;
951 	const char *repo_abspath = NULL;
952 	size_t repolen, len;
953 	char *canonpath, *path = NULL;
954 
955 	*in_repo_path = NULL;
956 
957 	canonpath = strdup(input_path);
958 	if (canonpath == NULL) {
959 		err = got_error_from_errno("strdup");
960 		goto done;
961 	}
962 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
963 	if (err)
964 		goto done;
965 
966 	repo_abspath = got_repo_get_path(repo);
967 
968 	if (canonpath[0] == '\0') {
969 		path = strdup(canonpath);
970 		if (path == NULL) {
971 			err = got_error_from_errno("strdup");
972 			goto done;
973 		}
974 	} else {
975 		path = realpath(canonpath, NULL);
976 		if (path == NULL) {
977 			if (errno != ENOENT) {
978 				err = got_error_from_errno2("realpath",
979 				    canonpath);
980 				goto done;
981 			}
982 			/*
983 			 * Path is not on disk.
984 			 * Assume it is already relative to repository root.
985 			 */
986 			path = strdup(canonpath);
987 			if (path == NULL) {
988 				err = got_error_from_errno("strdup");
989 				goto done;
990 			}
991 		}
992 
993 		repolen = strlen(repo_abspath);
994 		len = strlen(path);
995 
996 
997 		if (strcmp(path, repo_abspath) == 0) {
998 			free(path);
999 			path = strdup("");
1000 			if (path == NULL) {
1001 				err = got_error_from_errno("strdup");
1002 				goto done;
1003 			}
1004 		} else if (len > repolen &&
1005 		    got_path_is_child(path, repo_abspath, repolen)) {
1006 			/* Matched an on-disk path inside repository. */
1007 			if (got_repo_is_bare(repo)) {
1008 				/*
1009 				 * Matched an on-disk path inside repository
1010 				 * database. Treat input as repository-relative.
1011 				 */
1012 				free(path);
1013 				path = canonpath;
1014 				canonpath = NULL;
1015 			} else {
1016 				char *child;
1017 				/* Strip common prefix with repository path. */
1018 				err = got_path_skip_common_ancestor(&child,
1019 				    repo_abspath, path);
1020 				if (err)
1021 					goto done;
1022 				free(path);
1023 				path = child;
1024 			}
1025 		} else {
1026 			/*
1027 			 * Matched unrelated on-disk path.
1028 			 * Treat input as repository-relative.
1029 			 */
1030 			free(path);
1031 			path = canonpath;
1032 			canonpath = NULL;
1033 		}
1034 	}
1035 
1036 	/* Make in-repository path absolute */
1037 	if (path[0] != '/') {
1038 		char *abspath;
1039 		if (asprintf(&abspath, "/%s", path) == -1) {
1040 			err = got_error_from_errno("asprintf");
1041 			goto done;
1042 		}
1043 		free(path);
1044 		path = abspath;
1045 	}
1046 
1047 done:
1048 	free(canonpath);
1049 	if (err)
1050 		free(path);
1051 	else
1052 		*in_repo_path = path;
1053 	return err;
1054 }
1055 
1056 static const struct got_error *
1057 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1058     const char *path_packidx)
1059 {
1060 	const struct got_error *err = NULL;
1061 	size_t i;
1062 
1063 	for (i = 0; i < repo->pack_cache_size; i++) {
1064 		if (repo->packidx_cache[i] == NULL)
1065 			break;
1066 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1067 		    path_packidx) == 0) {
1068 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1069 		}
1070 	}
1071 	if (i == repo->pack_cache_size) {
1072 		do {
1073 			i--;
1074 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1075 		    i == repo->pinned_packidx);
1076 		err = got_packidx_close(repo->packidx_cache[i]);
1077 		if (err)
1078 			return err;
1079 	}
1080 
1081 	repo->packidx_cache[i] = packidx;
1082 
1083 	return NULL;
1084 }
1085 
1086 int
1087 got_repo_is_packidx_filename(const char *name, size_t len)
1088 {
1089 	if (len != GOT_PACKIDX_NAMELEN)
1090 		return 0;
1091 
1092 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1093 		return 0;
1094 
1095 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1096 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1097 		return 0;
1098 
1099 	return 1;
1100 }
1101 
1102 static struct got_packidx_bloom_filter *
1103 get_packidx_bloom_filter(struct got_repository *repo,
1104     const char *path, size_t path_len)
1105 {
1106 	struct got_packidx_bloom_filter key;
1107 
1108 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1109 		return NULL; /* XXX */
1110 	key.path_len = path_len;
1111 
1112 	return RB_FIND(got_packidx_bloom_filter_tree,
1113 	    &repo->packidx_bloom_filters, &key);
1114 }
1115 
1116 int
1117 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1118     const char *path_packidx, struct got_object_id *id)
1119 {
1120 	struct got_packidx_bloom_filter *bf;
1121 
1122 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1123 	if (bf)
1124 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1125 
1126 	/* No bloom filter means this pack index must be searched. */
1127 	return 1;
1128 }
1129 
1130 static const struct got_error *
1131 add_packidx_bloom_filter(struct got_repository *repo,
1132     struct got_packidx *packidx, const char *path_packidx)
1133 {
1134 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1135 	struct got_packidx_bloom_filter *bf;
1136 	size_t len;
1137 
1138 	/*
1139 	 * Don't use bloom filters for very large pack index files.
1140 	 * Large pack files will contain a relatively large fraction
1141 	 * of our objects so we will likely need to visit them anyway.
1142 	 * The more objects a pack file contains the higher the probability
1143 	 * of a false-positive match from the bloom filter. And reading
1144 	 * all object IDs from a large pack index file can be expensive.
1145 	 */
1146 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1147 		return NULL;
1148 
1149 	/* Do we already have a filter for this pack index? */
1150 	if (get_packidx_bloom_filter(repo, path_packidx,
1151 	    strlen(path_packidx)) != NULL)
1152 		return NULL;
1153 
1154 	bf = calloc(1, sizeof(*bf));
1155 	if (bf == NULL)
1156 		return got_error_from_errno("calloc");
1157 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1158 	if (bf->bloom == NULL) {
1159 		free(bf);
1160 		return got_error_from_errno("calloc");
1161 	}
1162 
1163 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1164 	if (len >= sizeof(bf->path)) {
1165 		free(bf->bloom);
1166 		free(bf);
1167 		return got_error(GOT_ERR_NO_SPACE);
1168 	}
1169 	bf->path_len = len;
1170 
1171 	/* Minimum size supported by our bloom filter is 1000 entries. */
1172 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1173 	for (i = 0; i < nobjects; i++) {
1174 		struct got_packidx_object_id *id;
1175 		id = &packidx->hdr.sorted_ids[i];
1176 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1177 	}
1178 
1179 	RB_INSERT(got_packidx_bloom_filter_tree,
1180 	    &repo->packidx_bloom_filters, bf);
1181 	return NULL;
1182 }
1183 
1184 const struct got_error *
1185 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1186     struct got_repository *repo, struct got_object_id *id)
1187 {
1188 	const struct got_error *err;
1189 	struct got_pathlist_entry *pe;
1190 	size_t i;
1191 
1192 	/* Search pack index cache. */
1193 	for (i = 0; i < repo->pack_cache_size; i++) {
1194 		if (repo->packidx_cache[i] == NULL)
1195 			break;
1196 		if (!got_repo_check_packidx_bloom_filter(repo,
1197 		    repo->packidx_cache[i]->path_packidx, id))
1198 			continue; /* object will not be found in this index */
1199 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1200 		if (*idx != -1) {
1201 			*packidx = repo->packidx_cache[i];
1202 			/*
1203 			 * Move this cache entry to the front. Repeatedly
1204 			 * searching a wrong pack index can be expensive.
1205 			 */
1206 			if (i > 0) {
1207 				memmove(&repo->packidx_cache[1],
1208 				    &repo->packidx_cache[0],
1209 				    i * sizeof(repo->packidx_cache[0]));
1210 				repo->packidx_cache[0] = *packidx;
1211 				if (repo->pinned_packidx >= 0 &&
1212 				    repo->pinned_packidx < i)
1213 					repo->pinned_packidx++;
1214 				else if (repo->pinned_packidx == i)
1215 					repo->pinned_packidx = 0;
1216 			}
1217 			return NULL;
1218 		}
1219 	}
1220 	/* No luck. Search the filesystem. */
1221 
1222 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1223 		const char *path_packidx = pe->path;
1224 		int is_cached = 0;
1225 
1226 		if (!got_repo_check_packidx_bloom_filter(repo,
1227 		    pe->path, id))
1228 			continue; /* object will not be found in this index */
1229 
1230 		for (i = 0; i < repo->pack_cache_size; i++) {
1231 			if (repo->packidx_cache[i] == NULL)
1232 				break;
1233 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1234 			    path_packidx) == 0) {
1235 				is_cached = 1;
1236 				break;
1237 			}
1238 		}
1239 		if (is_cached)
1240 			continue; /* already searched */
1241 
1242 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1243 		    path_packidx, 0);
1244 		if (err)
1245 			goto done;
1246 
1247 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1248 		if (err)
1249 			goto done;
1250 
1251 		err = cache_packidx(repo, *packidx, path_packidx);
1252 		if (err)
1253 			goto done;
1254 
1255 		*idx = got_packidx_get_object_idx(*packidx, id);
1256 		if (*idx != -1) {
1257 			err = NULL; /* found the object */
1258 			goto done;
1259 		}
1260 	}
1261 
1262 	err = got_error_no_obj(id);
1263 done:
1264 	return err;
1265 }
1266 
1267 const struct got_error *
1268 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1269     struct got_repository *repo)
1270 {
1271 	const struct got_error *err = NULL;
1272 	DIR *packdir = NULL;
1273 	struct dirent *dent;
1274 	char *path_packidx = NULL;
1275 	int packdir_fd;
1276 	struct stat sb;
1277 
1278 	packdir_fd = openat(got_repo_get_fd(repo),
1279 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1280 	if (packdir_fd == -1) {
1281 		return got_error_from_errno_fmt("openat: %s/%s",
1282 		    got_repo_get_path_git_dir(repo),
1283 		    GOT_OBJECTS_PACK_DIR);
1284 	}
1285 
1286 	packdir = fdopendir(packdir_fd);
1287 	if (packdir == NULL) {
1288 		err = got_error_from_errno("fdopendir");
1289 		goto done;
1290 	}
1291 
1292 	if (fstat(packdir_fd, &sb) == -1) {
1293 		err = got_error_from_errno("fstat");
1294 		goto done;
1295 	}
1296 	repo->pack_path_mtime = sb.st_mtime;
1297 
1298 	while ((dent = readdir(packdir)) != NULL) {
1299 		if (!got_repo_is_packidx_filename(dent->d_name,
1300 		    strlen(dent->d_name)))
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 void
1611 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1612 {
1613 	struct got_pathlist_entry *pe;
1614 
1615 	while (!TAILQ_EMPTY(packidx_paths)) {
1616 		pe = TAILQ_FIRST(packidx_paths);
1617 		TAILQ_REMOVE(packidx_paths, pe, entry);
1618 		free((char *)pe->path);
1619 		free(pe);
1620 	}
1621 }
1622 
1623 static const struct got_error *
1624 match_packed_object(struct got_object_id **unique_id,
1625     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1626 {
1627 	const struct got_error *err = NULL;
1628 	char *objects_pack_dir = NULL;
1629 	struct got_object_id_queue matched_ids;
1630 	struct got_pathlist_entry *pe;
1631 	struct stat sb;
1632 
1633 	STAILQ_INIT(&matched_ids);
1634 
1635 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1636 	if (objects_pack_dir == NULL)
1637 		return got_error_from_errno("got_repo_get_path_objects_pack");
1638 
1639 	if (stat(objects_pack_dir, &sb) == -1) {
1640 		if (errno != ENOENT) {
1641 			err = got_error_from_errno2("stat", objects_pack_dir);
1642 			goto done;
1643 		}
1644 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1645 		purge_packidx_paths(&repo->packidx_paths);
1646 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1647 		if (err)
1648 			goto done;
1649 	}
1650 
1651 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1652 		const char *path_packidx = pe->path;
1653 		struct got_packidx *packidx;
1654 		struct got_object_qid *qid;
1655 
1656 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1657 		    path_packidx, 0);
1658 		if (err)
1659 			break;
1660 
1661 		err = got_packidx_match_id_str_prefix(&matched_ids,
1662 		    packidx, id_str_prefix);
1663 		if (err) {
1664 			got_packidx_close(packidx);
1665 			break;
1666 		}
1667 		err = got_packidx_close(packidx);
1668 		if (err)
1669 			break;
1670 
1671 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1672 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1673 				int matched_type;
1674 				err = got_object_get_type(&matched_type, repo,
1675 				    &qid->id);
1676 				if (err)
1677 					goto done;
1678 				if (matched_type != obj_type)
1679 					continue;
1680 			}
1681 			if (*unique_id == NULL) {
1682 				*unique_id = got_object_id_dup(&qid->id);
1683 				if (*unique_id == NULL) {
1684 					err = got_error_from_errno("malloc");
1685 					goto done;
1686 				}
1687 			} else {
1688 				if (got_object_id_cmp(*unique_id,
1689 				    &qid->id) == 0)
1690 					continue; /* packed multiple times */
1691 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1692 				goto done;
1693 			}
1694 		}
1695 	}
1696 done:
1697 	free(objects_pack_dir);
1698 	got_object_id_queue_free(&matched_ids);
1699 	if (err) {
1700 		free(*unique_id);
1701 		*unique_id = NULL;
1702 	}
1703 	return err;
1704 }
1705 
1706 static const struct got_error *
1707 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1708     const char *object_dir, const char *id_str_prefix, int obj_type,
1709     struct got_repository *repo)
1710 {
1711 	const struct got_error *err = NULL;
1712 	char *path, *id_str = NULL;
1713 	DIR *dir = NULL;
1714 	struct dirent *dent;
1715 	struct got_object_id id;
1716 
1717 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1718 		err = got_error_from_errno("asprintf");
1719 		goto done;
1720 	}
1721 
1722 	dir = opendir(path);
1723 	if (dir == NULL) {
1724 		if (errno == ENOENT) {
1725 			err = NULL;
1726 			goto done;
1727 		}
1728 		err = got_error_from_errno2("opendir", path);
1729 		goto done;
1730 	}
1731 	while ((dent = readdir(dir)) != NULL) {
1732 		int cmp;
1733 
1734 		free(id_str);
1735 		id_str = NULL;
1736 
1737 		if (strcmp(dent->d_name, ".") == 0 ||
1738 		    strcmp(dent->d_name, "..") == 0)
1739 			continue;
1740 
1741 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1742 			err = got_error_from_errno("asprintf");
1743 			goto done;
1744 		}
1745 
1746 		if (!got_parse_sha1_digest(id.sha1, id_str))
1747 			continue;
1748 
1749 		/*
1750 		 * Directory entries do not necessarily appear in
1751 		 * sorted order, so we must iterate over all of them.
1752 		 */
1753 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1754 		if (cmp != 0)
1755 			continue;
1756 
1757 		if (*unique_id == NULL) {
1758 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1759 				int matched_type;
1760 				err = got_object_get_type(&matched_type, repo,
1761 				    &id);
1762 				if (err)
1763 					goto done;
1764 				if (matched_type != obj_type)
1765 					continue;
1766 			}
1767 			*unique_id = got_object_id_dup(&id);
1768 			if (*unique_id == NULL) {
1769 				err = got_error_from_errno("got_object_id_dup");
1770 				goto done;
1771 			}
1772 		} else {
1773 			if (got_object_id_cmp(*unique_id, &id) == 0)
1774 				continue; /* both packed and loose */
1775 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1776 			goto done;
1777 		}
1778 	}
1779 done:
1780 	if (dir && closedir(dir) != 0 && err == NULL)
1781 		err = got_error_from_errno("closedir");
1782 	if (err) {
1783 		free(*unique_id);
1784 		*unique_id = NULL;
1785 	}
1786 	free(id_str);
1787 	free(path);
1788 	return err;
1789 }
1790 
1791 const struct got_error *
1792 got_repo_match_object_id_prefix(struct got_object_id **id,
1793     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1794 {
1795 	const struct got_error *err = NULL;
1796 	char *path_objects = NULL, *object_dir = NULL;
1797 	size_t len;
1798 	int i;
1799 
1800 	*id = NULL;
1801 
1802 	path_objects = got_repo_get_path_objects(repo);
1803 
1804 	len = strlen(id_str_prefix);
1805 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1806 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1807 		goto done;
1808 	}
1809 
1810 	for (i = 0; i < len; i++) {
1811 		if (isxdigit((unsigned char)id_str_prefix[i]))
1812 			continue;
1813 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1814 		goto done;
1815 	}
1816 
1817 	if (len >= 2) {
1818 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1819 		if (err)
1820 			goto done;
1821 		object_dir = strndup(id_str_prefix, 2);
1822 		if (object_dir == NULL) {
1823 			err = got_error_from_errno("strdup");
1824 			goto done;
1825 		}
1826 		err = match_loose_object(id, path_objects, object_dir,
1827 		    id_str_prefix, obj_type, repo);
1828 	} else if (len == 1) {
1829 		int i;
1830 		for (i = 0; i < 0xf; i++) {
1831 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1832 			    == -1) {
1833 				err = got_error_from_errno("asprintf");
1834 				goto done;
1835 			}
1836 			err = match_packed_object(id, repo, object_dir,
1837 			    obj_type);
1838 			if (err)
1839 				goto done;
1840 			err = match_loose_object(id, path_objects, object_dir,
1841 			    id_str_prefix, obj_type, repo);
1842 			if (err)
1843 				goto done;
1844 		}
1845 	} else {
1846 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1847 		goto done;
1848 	}
1849 done:
1850 	free(path_objects);
1851 	free(object_dir);
1852 	if (err) {
1853 		free(*id);
1854 		*id = NULL;
1855 	} else if (*id == NULL) {
1856 		switch (obj_type) {
1857 		case GOT_OBJ_TYPE_BLOB:
1858 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1859 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1860 			break;
1861 		case GOT_OBJ_TYPE_TREE:
1862 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1863 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1864 			break;
1865 		case GOT_OBJ_TYPE_COMMIT:
1866 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1867 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1868 			break;
1869 		case GOT_OBJ_TYPE_TAG:
1870 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1871 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1872 			break;
1873 		default:
1874 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1875 			break;
1876 		}
1877 	}
1878 
1879 	return err;
1880 }
1881 
1882 const struct got_error *
1883 got_repo_match_object_id(struct got_object_id **id, char **label,
1884     const char *id_str, int obj_type, struct got_reflist_head *refs,
1885     struct got_repository *repo)
1886 {
1887 	const struct got_error *err;
1888 	struct got_tag_object *tag;
1889 	struct got_reference *ref = NULL;
1890 
1891 	*id = NULL;
1892 	if (label)
1893 		*label = NULL;
1894 
1895 	if (refs) {
1896 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1897 		    refs, repo);
1898 		if (err == NULL) {
1899 			*id = got_object_id_dup(
1900 			    got_object_tag_get_object_id(tag));
1901 			if (*id == NULL)
1902 				err = got_error_from_errno("got_object_id_dup");
1903 			else if (label && asprintf(label, "refs/tags/%s",
1904 			    got_object_tag_get_name(tag)) == -1) {
1905 				err = got_error_from_errno("asprintf");
1906 				free(*id);
1907 				*id = NULL;
1908 			}
1909 			got_object_tag_close(tag);
1910 			return err;
1911 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1912 		    err->code != GOT_ERR_NO_OBJ)
1913 			return err;
1914 	}
1915 
1916 	err = got_ref_open(&ref, repo, id_str, 0);
1917 	if (err == NULL) {
1918 		err = got_ref_resolve(id, repo, ref);
1919 		if (err)
1920 			goto done;
1921 		if (label) {
1922 			*label = strdup(got_ref_get_name(ref));
1923 			if (*label == NULL) {
1924 				err = got_error_from_errno("strdup");
1925 				goto done;
1926 			}
1927 		}
1928 	} else {
1929 		if (err->code != GOT_ERR_NOT_REF &&
1930 		    err->code != GOT_ERR_BAD_REF_NAME)
1931 			goto done;
1932 		err = got_repo_match_object_id_prefix(id, id_str,
1933 		    obj_type, repo);
1934 		if (err) {
1935 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1936 				err = got_error_not_ref(id_str);
1937 			goto done;
1938 		}
1939 		if (label) {
1940 			err = got_object_id_str(label, *id);
1941 			if (*label == NULL) {
1942 				err = got_error_from_errno("strdup");
1943 				goto done;
1944 			}
1945 		}
1946 	}
1947 done:
1948 	if (ref)
1949 		got_ref_close(ref);
1950 	return err;
1951 }
1952 
1953 const struct got_error *
1954 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1955     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1956 {
1957 	const struct got_error *err = NULL;
1958 	struct got_reflist_entry *re;
1959 	struct got_object_id *tag_id;
1960 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1961 
1962 	*tag = NULL;
1963 
1964 	TAILQ_FOREACH(re, refs, entry) {
1965 		const char *refname;
1966 		refname = got_ref_get_name(re->ref);
1967 		if (got_ref_is_symbolic(re->ref))
1968 			continue;
1969 		if (strncmp(refname, "refs/tags/", 10) != 0)
1970 			continue;
1971 		if (!name_is_absolute)
1972 			refname += strlen("refs/tags/");
1973 		if (strcmp(refname, name) != 0)
1974 			continue;
1975 		err = got_ref_resolve(&tag_id, repo, re->ref);
1976 		if (err)
1977 			break;
1978 		err = got_object_open_as_tag(tag, repo, tag_id);
1979 		free(tag_id);
1980 		if (err)
1981 			break;
1982 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1983 		    got_object_tag_get_object_type(*tag) == obj_type)
1984 			break;
1985 		got_object_tag_close(*tag);
1986 		*tag = NULL;
1987 	}
1988 
1989 	if (err == NULL && *tag == NULL)
1990 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1991 		    GOT_OBJ_LABEL_TAG, name);
1992 	return err;
1993 }
1994 
1995 static const struct got_error *
1996 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1997     const char *name, mode_t mode, struct got_object_id *blob_id)
1998 {
1999 	const struct got_error *err = NULL;
2000 
2001 	 *new_te = NULL;
2002 
2003 	*new_te = calloc(1, sizeof(**new_te));
2004 	if (*new_te == NULL)
2005 		return got_error_from_errno("calloc");
2006 
2007 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2008 	    sizeof((*new_te)->name)) {
2009 		err = got_error(GOT_ERR_NO_SPACE);
2010 		goto done;
2011 	}
2012 
2013 	if (S_ISLNK(mode)) {
2014 		(*new_te)->mode = S_IFLNK;
2015 	} else {
2016 		(*new_te)->mode = S_IFREG;
2017 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2018 	}
2019 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2020 done:
2021 	if (err && *new_te) {
2022 		free(*new_te);
2023 		*new_te = NULL;
2024 	}
2025 	return err;
2026 }
2027 
2028 static const struct got_error *
2029 import_file(struct got_tree_entry **new_te, struct dirent *de,
2030     const char *path, struct got_repository *repo)
2031 {
2032 	const struct got_error *err;
2033 	struct got_object_id *blob_id = NULL;
2034 	char *filepath;
2035 	struct stat sb;
2036 
2037 	if (asprintf(&filepath, "%s%s%s", path,
2038 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2039 		return got_error_from_errno("asprintf");
2040 
2041 	if (lstat(filepath, &sb) != 0) {
2042 		err = got_error_from_errno2("lstat", path);
2043 		goto done;
2044 	}
2045 
2046 	err = got_object_blob_create(&blob_id, filepath, repo);
2047 	if (err)
2048 		goto done;
2049 
2050 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2051 	    blob_id);
2052 done:
2053 	free(filepath);
2054 	if (err)
2055 		free(blob_id);
2056 	return err;
2057 }
2058 
2059 static const struct got_error *
2060 insert_tree_entry(struct got_tree_entry *new_te,
2061     struct got_pathlist_head *paths)
2062 {
2063 	const struct got_error *err = NULL;
2064 	struct got_pathlist_entry *new_pe;
2065 
2066 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2067 	if (err)
2068 		return err;
2069 	if (new_pe == NULL)
2070 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2071 	return NULL;
2072 }
2073 
2074 static const struct got_error *write_tree(struct got_object_id **,
2075     const char *, struct got_pathlist_head *, struct got_repository *,
2076     got_repo_import_cb progress_cb, void *progress_arg);
2077 
2078 static const struct got_error *
2079 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2080     const char *path, struct got_pathlist_head *ignores,
2081     struct got_repository *repo,
2082     got_repo_import_cb progress_cb, void *progress_arg)
2083 {
2084 	const struct got_error *err;
2085 	struct got_object_id *id = NULL;
2086 	char *subdirpath;
2087 
2088 	if (asprintf(&subdirpath, "%s%s%s", path,
2089 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2090 		return got_error_from_errno("asprintf");
2091 
2092 	(*new_te) = calloc(1, sizeof(**new_te));
2093 	if (*new_te == NULL)
2094 		return got_error_from_errno("calloc");
2095 	(*new_te)->mode = S_IFDIR;
2096 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2097 	    sizeof((*new_te)->name)) {
2098 		err = got_error(GOT_ERR_NO_SPACE);
2099 		goto done;
2100 	}
2101 	err = write_tree(&id, subdirpath, ignores,  repo,
2102 	    progress_cb, progress_arg);
2103 	if (err)
2104 		goto done;
2105 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2106 
2107 done:
2108 	free(id);
2109 	free(subdirpath);
2110 	if (err) {
2111 		free(*new_te);
2112 		*new_te = NULL;
2113 	}
2114 	return err;
2115 }
2116 
2117 static const struct got_error *
2118 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2119     struct got_pathlist_head *ignores, struct got_repository *repo,
2120     got_repo_import_cb progress_cb, void *progress_arg)
2121 {
2122 	const struct got_error *err = NULL;
2123 	DIR *dir;
2124 	struct dirent *de;
2125 	int nentries;
2126 	struct got_tree_entry *new_te = NULL;
2127 	struct got_pathlist_head paths;
2128 	struct got_pathlist_entry *pe;
2129 
2130 	*new_tree_id = NULL;
2131 
2132 	TAILQ_INIT(&paths);
2133 
2134 	dir = opendir(path_dir);
2135 	if (dir == NULL) {
2136 		err = got_error_from_errno2("opendir", path_dir);
2137 		goto done;
2138 	}
2139 
2140 	nentries = 0;
2141 	while ((de = readdir(dir)) != NULL) {
2142 		int ignore = 0;
2143 		int type;
2144 
2145 		if (strcmp(de->d_name, ".") == 0 ||
2146 		    strcmp(de->d_name, "..") == 0)
2147 			continue;
2148 
2149 		TAILQ_FOREACH(pe, ignores, entry) {
2150 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2151 				ignore = 1;
2152 				break;
2153 			}
2154 		}
2155 		if (ignore)
2156 			continue;
2157 
2158 		err = got_path_dirent_type(&type, path_dir, de);
2159 		if (err)
2160 			goto done;
2161 
2162 		if (type == DT_DIR) {
2163 			err = import_subdir(&new_te, de, path_dir,
2164 			    ignores, repo, progress_cb, progress_arg);
2165 			if (err) {
2166 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2167 					goto done;
2168 				err = NULL;
2169 				continue;
2170 			}
2171 		} else if (type == DT_REG || type == DT_LNK) {
2172 			err = import_file(&new_te, de, path_dir, repo);
2173 			if (err)
2174 				goto done;
2175 		} else
2176 			continue;
2177 
2178 		err = insert_tree_entry(new_te, &paths);
2179 		if (err)
2180 			goto done;
2181 		nentries++;
2182 	}
2183 
2184 	if (TAILQ_EMPTY(&paths)) {
2185 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2186 		    "cannot create tree without any entries");
2187 		goto done;
2188 	}
2189 
2190 	TAILQ_FOREACH(pe, &paths, entry) {
2191 		struct got_tree_entry *te = pe->data;
2192 		char *path;
2193 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2194 			continue;
2195 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2196 			err = got_error_from_errno("asprintf");
2197 			goto done;
2198 		}
2199 		err = (*progress_cb)(progress_arg, path);
2200 		free(path);
2201 		if (err)
2202 			goto done;
2203 	}
2204 
2205 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2206 done:
2207 	if (dir)
2208 		closedir(dir);
2209 	got_pathlist_free(&paths);
2210 	return err;
2211 }
2212 
2213 const struct got_error *
2214 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2215     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2216     struct got_repository *repo, got_repo_import_cb progress_cb,
2217     void *progress_arg)
2218 {
2219 	const struct got_error *err;
2220 	struct got_object_id *new_tree_id;
2221 
2222 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2223 	    progress_cb, progress_arg);
2224 	if (err)
2225 		return err;
2226 
2227 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2228 	    author, time(NULL), author, time(NULL), logmsg, repo);
2229 	free(new_tree_id);
2230 	return err;
2231 }
2232 
2233 const struct got_error *
2234 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2235     struct got_repository *repo)
2236 {
2237 	const struct got_error *err = NULL;
2238 	char *path_objects = NULL, *path = NULL;
2239 	DIR *dir = NULL;
2240 	struct got_object_id id;
2241 	int i;
2242 
2243 	*nobjects = 0;
2244 	*ondisk_size = 0;
2245 
2246 	path_objects = got_repo_get_path_objects(repo);
2247 	if (path_objects == NULL)
2248 		return got_error_from_errno("got_repo_get_path_objects");
2249 
2250 	for (i = 0; i <= 0xff; i++) {
2251 		struct dirent *dent;
2252 
2253 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2254 			err = got_error_from_errno("asprintf");
2255 			break;
2256 		}
2257 
2258 		dir = opendir(path);
2259 		if (dir == NULL) {
2260 			if (errno == ENOENT) {
2261 				err = NULL;
2262 				continue;
2263 			}
2264 			err = got_error_from_errno2("opendir", path);
2265 			break;
2266 		}
2267 
2268 		while ((dent = readdir(dir)) != NULL) {
2269 			char *id_str;
2270 			int fd;
2271 			struct stat sb;
2272 
2273 			if (strcmp(dent->d_name, ".") == 0 ||
2274 			    strcmp(dent->d_name, "..") == 0)
2275 				continue;
2276 
2277 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2278 				err = got_error_from_errno("asprintf");
2279 				goto done;
2280 			}
2281 
2282 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2283 				free(id_str);
2284 				continue;
2285 			}
2286 			free(id_str);
2287 
2288 			err = got_object_open_loose_fd(&fd, &id, repo);
2289 			if (err)
2290 				goto done;
2291 
2292 			if (fstat(fd, &sb) == -1) {
2293 				err = got_error_from_errno("fstat");
2294 				close(fd);
2295 				goto done;
2296 			}
2297 			(*nobjects)++;
2298 			(*ondisk_size) += sb.st_size;
2299 
2300 			if (close(fd) == -1) {
2301 				err = got_error_from_errno("close");
2302 				goto done;
2303 			}
2304 		}
2305 
2306 		if (closedir(dir) != 0) {
2307 			err = got_error_from_errno("closedir");
2308 			goto done;
2309 		}
2310 		dir = NULL;
2311 
2312 		free(path);
2313 		path = NULL;
2314 	}
2315 done:
2316 	if (dir && closedir(dir) != 0 && err == NULL)
2317 		err = got_error_from_errno("closedir");
2318 
2319 	if (err) {
2320 		*nobjects = 0;
2321 		*ondisk_size = 0;
2322 	}
2323 	free(path_objects);
2324 	free(path);
2325 	return err;
2326 }
2327 
2328 const struct got_error *
2329 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2330     off_t *total_packsize, struct got_repository *repo)
2331 {
2332 	const struct got_error *err = NULL;
2333 	DIR *packdir = NULL;
2334 	struct dirent *dent;
2335 	struct got_packidx *packidx = NULL;
2336 	char *path_packidx;
2337 	char *path_packfile;
2338 	int packdir_fd;
2339 	struct stat sb;
2340 
2341 	*npackfiles = 0;
2342 	*nobjects = 0;
2343 	*total_packsize = 0;
2344 
2345 	packdir_fd = openat(got_repo_get_fd(repo),
2346 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2347 	if (packdir_fd == -1) {
2348 		return got_error_from_errno_fmt("openat: %s/%s",
2349 		    got_repo_get_path_git_dir(repo),
2350 		    GOT_OBJECTS_PACK_DIR);
2351 	}
2352 
2353 	packdir = fdopendir(packdir_fd);
2354 	if (packdir == NULL) {
2355 		err = got_error_from_errno("fdopendir");
2356 		goto done;
2357 	}
2358 
2359 	while ((dent = readdir(packdir)) != NULL) {
2360 		if (!got_repo_is_packidx_filename(dent->d_name,
2361 		    strlen(dent->d_name)))
2362 			continue;
2363 
2364 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2365 		    dent->d_name) == -1) {
2366 			err = got_error_from_errno("asprintf");
2367 			goto done;
2368 		}
2369 
2370 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2371 		    path_packidx, 0);
2372 		free(path_packidx);
2373 		if (err)
2374 			goto done;
2375 
2376 		if (fstat(packidx->fd, &sb) == -1)
2377 			goto done;
2378 		*total_packsize += sb.st_size;
2379 
2380 		err = got_packidx_get_packfile_path(&path_packfile,
2381 		    packidx->path_packidx);
2382 		if (err)
2383 			goto done;
2384 
2385 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2386 		    0) == -1) {
2387 			free(path_packfile);
2388 			goto done;
2389 		}
2390 		free(path_packfile);
2391 		*total_packsize += sb.st_size;
2392 
2393 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2394 
2395 		(*npackfiles)++;
2396 
2397 		got_packidx_close(packidx);
2398 		packidx = NULL;
2399 	}
2400 done:
2401 	if (packidx)
2402 		got_packidx_close(packidx);
2403 	if (packdir && closedir(packdir) != 0 && err == NULL)
2404 		err = got_error_from_errno("closedir");
2405 	if (err) {
2406 		*npackfiles = 0;
2407 		*nobjects = 0;
2408 		*total_packsize = 0;
2409 	}
2410 	return err;
2411 }
2412 
2413 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2414     got_packidx_bloom_filter_cmp);