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