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