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(), which would result in a close(0) if
265 	 * we do not initialize to -1 here.
266 	 */
267 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
268 		pack_fds_tmp[i] = -1;
269 
270 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
271 		pack_fds_tmp[i] = got_opentempfd();
272 		if (pack_fds_tmp[i] == -1) {
273 			err = got_error_from_errno("got_opentempfd");
274 			got_repo_pack_fds_close(pack_fds_tmp);
275 			return err;
276 		}
277 	}
278 	memcpy(*pack_fds, pack_fds_tmp, GOT_PACK_NUM_TEMPFILES * sizeof(int));
279 	return err;
280 }
281 
282 const struct got_error *
283 got_repo_pack_fds_close(int *pack_fds)
284 {
285 	const struct got_error *err = NULL;
286 	int i;
287 
288 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
289 		if (pack_fds[i] == -1)
290 			continue;
291 		if (close(pack_fds[i]) == -1) {
292 			err = got_error_from_errno("close");
293 			break;
294 		}
295 	}
296 	free(pack_fds);
297 	return err;
298 }
299 
300 const struct got_error *
301 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
302     struct got_object *obj)
303 {
304 #ifndef GOT_NO_OBJ_CACHE
305 	const struct got_error *err = NULL;
306 	err = got_object_cache_add(&repo->objcache, id, obj);
307 	if (err) {
308 		if (err->code == GOT_ERR_OBJ_EXISTS ||
309 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
310 			err = NULL;
311 		return err;
312 	}
313 	obj->refcnt++;
314 #endif
315 	return NULL;
316 }
317 
318 struct got_object *
319 got_repo_get_cached_object(struct got_repository *repo,
320     struct got_object_id *id)
321 {
322 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
323 }
324 
325 const struct got_error *
326 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
327     struct got_tree_object *tree)
328 {
329 #ifndef GOT_NO_OBJ_CACHE
330 	const struct got_error *err = NULL;
331 	err = got_object_cache_add(&repo->treecache, id, tree);
332 	if (err) {
333 		if (err->code == GOT_ERR_OBJ_EXISTS ||
334 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
335 			err = NULL;
336 		return err;
337 	}
338 	tree->refcnt++;
339 #endif
340 	return NULL;
341 }
342 
343 struct got_tree_object *
344 got_repo_get_cached_tree(struct got_repository *repo,
345     struct got_object_id *id)
346 {
347 	return (struct got_tree_object *)got_object_cache_get(
348 	    &repo->treecache, id);
349 }
350 
351 const struct got_error *
352 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
353     struct got_commit_object *commit)
354 {
355 #ifndef GOT_NO_OBJ_CACHE
356 	const struct got_error *err = NULL;
357 	err = got_object_cache_add(&repo->commitcache, id, commit);
358 	if (err) {
359 		if (err->code == GOT_ERR_OBJ_EXISTS ||
360 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
361 			err = NULL;
362 		return err;
363 	}
364 	commit->refcnt++;
365 #endif
366 	return NULL;
367 }
368 
369 struct got_commit_object *
370 got_repo_get_cached_commit(struct got_repository *repo,
371     struct got_object_id *id)
372 {
373 	return (struct got_commit_object *)got_object_cache_get(
374 	    &repo->commitcache, id);
375 }
376 
377 const struct got_error *
378 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
379     struct got_tag_object *tag)
380 {
381 #ifndef GOT_NO_OBJ_CACHE
382 	const struct got_error *err = NULL;
383 	err = got_object_cache_add(&repo->tagcache, id, tag);
384 	if (err) {
385 		if (err->code == GOT_ERR_OBJ_EXISTS ||
386 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
387 			err = NULL;
388 		return err;
389 	}
390 	tag->refcnt++;
391 #endif
392 	return NULL;
393 }
394 
395 struct got_tag_object *
396 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
397 {
398 	return (struct got_tag_object *)got_object_cache_get(
399 	    &repo->tagcache, id);
400 }
401 
402 const struct got_error *
403 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
404     struct got_raw_object *raw)
405 {
406 #ifndef GOT_NO_OBJ_CACHE
407 	const struct got_error *err = NULL;
408 	err = got_object_cache_add(&repo->rawcache, id, raw);
409 	if (err) {
410 		if (err->code == GOT_ERR_OBJ_EXISTS ||
411 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
412 			err = NULL;
413 		return err;
414 	}
415 	raw->refcnt++;
416 #endif
417 	return NULL;
418 }
419 
420 
421 struct got_raw_object *
422 got_repo_get_cached_raw_object(struct got_repository *repo,
423     struct got_object_id *id)
424 {
425 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
426 }
427 
428 
429 static const struct got_error *
430 open_repo(struct got_repository *repo, const char *path)
431 {
432 	const struct got_error *err = NULL;
433 
434 	repo->gitdir_fd = -1;
435 
436 	/* bare git repository? */
437 	repo->path_git_dir = strdup(path);
438 	if (repo->path_git_dir == NULL)
439 		return got_error_from_errno("strdup");
440 	if (is_git_repo(repo)) {
441 		repo->path = strdup(repo->path_git_dir);
442 		if (repo->path == NULL) {
443 			err = got_error_from_errno("strdup");
444 			goto done;
445 		}
446 		repo->gitdir_fd = open(repo->path_git_dir,
447 		    O_DIRECTORY | O_CLOEXEC);
448 		if (repo->gitdir_fd == -1) {
449 			err = got_error_from_errno2("open",
450 			    repo->path_git_dir);
451 			goto done;
452 		}
453 		return NULL;
454 	}
455 
456 	/* git repository with working tree? */
457 	free(repo->path_git_dir);
458 	repo->path_git_dir = NULL;
459 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
460 		err = got_error_from_errno("asprintf");
461 		goto done;
462 	}
463 	if (is_git_repo(repo)) {
464 		repo->path = strdup(path);
465 		if (repo->path == NULL) {
466 			err = got_error_from_errno("strdup");
467 			goto done;
468 		}
469 		repo->gitdir_fd = open(repo->path_git_dir,
470 		    O_DIRECTORY | O_CLOEXEC);
471 		if (repo->gitdir_fd == -1) {
472 			err = got_error_from_errno2("open",
473 			    repo->path_git_dir);
474 			goto done;
475 		}
476 		return NULL;
477 	}
478 
479 	err = got_error(GOT_ERR_NOT_GIT_REPO);
480 done:
481 	if (err) {
482 		free(repo->path);
483 		repo->path = NULL;
484 		free(repo->path_git_dir);
485 		repo->path_git_dir = NULL;
486 		if (repo->gitdir_fd != -1)
487 			close(repo->gitdir_fd);
488 		repo->gitdir_fd = -1;
489 
490 	}
491 	return err;
492 }
493 
494 static const struct got_error *
495 parse_gitconfig_file(int *gitconfig_repository_format_version,
496     char **gitconfig_author_name, char **gitconfig_author_email,
497     struct got_remote_repo **remotes, int *nremotes,
498     char **gitconfig_owner, char ***extensions, int *nextensions,
499     const char *gitconfig_path)
500 {
501 	const struct got_error *err = NULL, *child_err = NULL;
502 	int fd = -1;
503 	int imsg_fds[2] = { -1, -1 };
504 	pid_t pid;
505 	struct imsgbuf *ibuf;
506 
507 	*gitconfig_repository_format_version = 0;
508 	if (extensions)
509 		*extensions = NULL;
510 	if (nextensions)
511 		*nextensions = 0;
512 	*gitconfig_author_name = NULL;
513 	*gitconfig_author_email = NULL;
514 	if (remotes)
515 		*remotes = NULL;
516 	if (nremotes)
517 		*nremotes = 0;
518 	if (gitconfig_owner)
519 		*gitconfig_owner = NULL;
520 
521 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
522 	if (fd == -1) {
523 		if (errno == ENOENT)
524 			return NULL;
525 		return got_error_from_errno2("open", gitconfig_path);
526 	}
527 
528 	ibuf = calloc(1, sizeof(*ibuf));
529 	if (ibuf == NULL) {
530 		err = got_error_from_errno("calloc");
531 		goto done;
532 	}
533 
534 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
535 		err = got_error_from_errno("socketpair");
536 		goto done;
537 	}
538 
539 	pid = fork();
540 	if (pid == -1) {
541 		err = got_error_from_errno("fork");
542 		goto done;
543 	} else if (pid == 0) {
544 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
545 		    gitconfig_path);
546 		/* not reached */
547 	}
548 
549 	if (close(imsg_fds[1]) == -1) {
550 		err = got_error_from_errno("close");
551 		goto done;
552 	}
553 	imsg_fds[1] = -1;
554 	imsg_init(ibuf, imsg_fds[0]);
555 
556 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
557 	if (err)
558 		goto done;
559 	fd = -1;
560 
561 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
562 	if (err)
563 		goto done;
564 
565 	err = got_privsep_recv_gitconfig_int(
566 	    gitconfig_repository_format_version, ibuf);
567 	if (err)
568 		goto done;
569 
570 	if (extensions && nextensions) {
571 		err = got_privsep_send_gitconfig_repository_extensions_req(
572 		    ibuf);
573 		if (err)
574 			goto done;
575 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
576 		if (err)
577 			goto done;
578 		if (*nextensions > 0) {
579 			int i;
580 			*extensions = calloc(*nextensions, sizeof(char *));
581 			if (*extensions == NULL) {
582 				err = got_error_from_errno("calloc");
583 				goto done;
584 			}
585 			for (i = 0; i < *nextensions; i++) {
586 				char *ext;
587 				err = got_privsep_recv_gitconfig_str(&ext,
588 				    ibuf);
589 				if (err)
590 					goto done;
591 				(*extensions)[i] = ext;
592 			}
593 		}
594 	}
595 
596 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
597 	if (err)
598 		goto done;
599 
600 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
601 	if (err)
602 		goto done;
603 
604 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
605 	if (err)
606 		goto done;
607 
608 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
609 	if (err)
610 		goto done;
611 
612 	if (remotes && nremotes) {
613 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
614 		if (err)
615 			goto done;
616 
617 		err = got_privsep_recv_gitconfig_remotes(remotes,
618 		    nremotes, ibuf);
619 		if (err)
620 			goto done;
621 	}
622 
623 	if (gitconfig_owner) {
624 		err = got_privsep_send_gitconfig_owner_req(ibuf);
625 		if (err)
626 			goto done;
627 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
628 		if (err)
629 			goto done;
630 	}
631 
632 	err = got_privsep_send_stop(imsg_fds[0]);
633 	child_err = got_privsep_wait_for_child(pid);
634 	if (child_err && err == NULL)
635 		err = child_err;
636 done:
637 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
638 		err = got_error_from_errno("close");
639 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
640 		err = got_error_from_errno("close");
641 	if (fd != -1 && close(fd) == -1 && err == NULL)
642 		err = got_error_from_errno2("close", gitconfig_path);
643 	free(ibuf);
644 	return err;
645 }
646 
647 static const struct got_error *
648 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
649 {
650 	const struct got_error *err = NULL;
651 	char *repo_gitconfig_path = NULL;
652 
653 	if (global_gitconfig_path) {
654 		/* Read settings from ~/.gitconfig. */
655 		int dummy_repo_version;
656 		err = parse_gitconfig_file(&dummy_repo_version,
657 		    &repo->global_gitconfig_author_name,
658 		    &repo->global_gitconfig_author_email,
659 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
660 		if (err)
661 			return err;
662 	}
663 
664 	/* Read repository's .git/config file. */
665 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
666 	if (repo_gitconfig_path == NULL)
667 		return got_error_from_errno("got_repo_get_path_gitconfig");
668 
669 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
670 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
671 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
672 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
673 	    repo_gitconfig_path);
674 	if (err)
675 		goto done;
676 
677 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
678 		int i;
679 
680 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
681 			got_repo_free_remote_repo_data(
682 			    &repo->gitconfig_remotes[i]);
683 		}
684 		free(repo->gitconfig_remotes);
685 		repo->gitconfig_remotes = NULL;
686 		repo->ngitconfig_remotes = 0;
687 
688 		free(repo->gitconfig_author_name);
689 		repo->gitconfig_author_name = NULL;
690 		free(repo->gitconfig_author_email);
691 		repo->gitconfig_author_email = NULL;
692 
693 		free(repo->global_gitconfig_author_name);
694 		repo->global_gitconfig_author_name = NULL;
695 		free(repo->global_gitconfig_author_email);
696 		repo->global_gitconfig_author_email = NULL;
697 	}
698 
699 done:
700 	free(repo_gitconfig_path);
701 	return err;
702 }
703 
704 static const struct got_error *
705 read_gotconfig(struct got_repository *repo)
706 {
707 	const struct got_error *err = NULL;
708 	char *gotconfig_path;
709 
710 	gotconfig_path = got_repo_get_path_gotconfig(repo);
711 	if (gotconfig_path == NULL)
712 		return got_error_from_errno("got_repo_get_path_gotconfig");
713 
714 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
715 	free(gotconfig_path);
716 	return err;
717 }
718 
719 /* Supported repository format extensions. */
720 static const char *const repo_extensions[] = {
721 	"noop",			/* Got supports repository format version 1. */
722 	"preciousObjects",	/* Supported by gotadmin cleanup. */
723 	"worktreeConfig",	/* Got does not care about Git work trees. */
724 };
725 
726 const struct got_error *
727 got_repo_open(struct got_repository **repop, const char *path,
728     const char *global_gitconfig_path, int *pack_fds)
729 {
730 	struct got_repository *repo = NULL;
731 	const struct got_error *err = NULL;
732 	char *repo_path = NULL;
733 	size_t i, j = 0;
734 	struct rlimit rl;
735 
736 	*repop = NULL;
737 
738 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
739 		return got_error_from_errno("getrlimit");
740 
741 	repo = calloc(1, sizeof(*repo));
742 	if (repo == NULL)
743 		return got_error_from_errno("calloc");
744 
745 	RB_INIT(&repo->packidx_bloom_filters);
746 	TAILQ_INIT(&repo->packidx_paths);
747 
748 	for (i = 0; i < nitems(repo->privsep_children); i++) {
749 		memset(&repo->privsep_children[i], 0,
750 		    sizeof(repo->privsep_children[0]));
751 		repo->privsep_children[i].imsg_fd = -1;
752 	}
753 
754 	err = got_object_cache_init(&repo->objcache,
755 	    GOT_OBJECT_CACHE_TYPE_OBJ);
756 	if (err)
757 		goto done;
758 	err = got_object_cache_init(&repo->treecache,
759 	    GOT_OBJECT_CACHE_TYPE_TREE);
760 	if (err)
761 		goto done;
762 	err = got_object_cache_init(&repo->commitcache,
763 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
764 	if (err)
765 		goto done;
766 	err = got_object_cache_init(&repo->tagcache,
767 	    GOT_OBJECT_CACHE_TYPE_TAG);
768 	if (err)
769 		goto done;
770 	err = got_object_cache_init(&repo->rawcache,
771 	    GOT_OBJECT_CACHE_TYPE_RAW);
772 	if (err)
773 		goto done;
774 
775 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
776 	if (repo->pack_cache_size > rl.rlim_cur / 8)
777 		repo->pack_cache_size = rl.rlim_cur / 8;
778 	for (i = 0; i < nitems(repo->packs); i++) {
779 		if (i < repo->pack_cache_size) {
780 			repo->packs[i].basefd = pack_fds[j++];
781 			repo->packs[i].accumfd = pack_fds[j++];
782 		} else {
783 			repo->packs[i].basefd = -1;
784 			repo->packs[i].accumfd = -1;
785 		}
786 	}
787 	repo->pinned_pack = -1;
788 	repo->pinned_packidx = -1;
789 	repo->pinned_pid = 0;
790 
791 	repo_path = realpath(path, NULL);
792 	if (repo_path == NULL) {
793 		err = got_error_from_errno2("realpath", path);
794 		goto done;
795 	}
796 
797 	for (;;) {
798 		char *parent_path;
799 
800 		err = open_repo(repo, repo_path);
801 		if (err == NULL)
802 			break;
803 		if (err->code != GOT_ERR_NOT_GIT_REPO)
804 			goto done;
805 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
806 			err = got_error(GOT_ERR_NOT_GIT_REPO);
807 			goto done;
808 		}
809 		err = got_path_dirname(&parent_path, repo_path);
810 		if (err)
811 			goto done;
812 		free(repo_path);
813 		repo_path = parent_path;
814 	}
815 
816 	err = read_gotconfig(repo);
817 	if (err)
818 		goto done;
819 
820 	err = read_gitconfig(repo, global_gitconfig_path);
821 	if (err)
822 		goto done;
823 	if (repo->gitconfig_repository_format_version != 0) {
824 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
825 		goto done;
826 	}
827 	for (i = 0; i < repo->nextensions; i++) {
828 		char *ext = repo->extensions[i];
829 		int j, supported = 0;
830 		for (j = 0; j < nitems(repo_extensions); j++) {
831 			if (strcmp(ext, repo_extensions[j]) == 0) {
832 				supported = 1;
833 				break;
834 			}
835 		}
836 		if (!supported) {
837 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
838 			goto done;
839 		}
840 	}
841 
842 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
843 done:
844 	if (err)
845 		got_repo_close(repo);
846 	else
847 		*repop = repo;
848 	free(repo_path);
849 	return err;
850 }
851 
852 const struct got_error *
853 got_repo_close(struct got_repository *repo)
854 {
855 	const struct got_error *err = NULL, *child_err;
856 	struct got_packidx_bloom_filter *bf;
857 	struct got_pathlist_entry *pe;
858 	size_t i;
859 
860 	for (i = 0; i < repo->pack_cache_size; i++) {
861 		if (repo->packidx_cache[i] == NULL)
862 			break;
863 		got_packidx_close(repo->packidx_cache[i]);
864 	}
865 
866 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
867 	    &repo->packidx_bloom_filters))) {
868 		RB_REMOVE(got_packidx_bloom_filter_tree,
869 		    &repo->packidx_bloom_filters, bf);
870 		free(bf->bloom);
871 		free(bf);
872 	}
873 
874 	for (i = 0; i < repo->pack_cache_size; i++)
875 		if (repo->packs[i].path_packfile)
876 			if (repo->packs[i].path_packfile)
877 				got_pack_close(&repo->packs[i]);
878 
879 	free(repo->path);
880 	free(repo->path_git_dir);
881 
882 	got_object_cache_close(&repo->objcache);
883 	got_object_cache_close(&repo->treecache);
884 	got_object_cache_close(&repo->commitcache);
885 	got_object_cache_close(&repo->tagcache);
886 	got_object_cache_close(&repo->rawcache);
887 
888 	for (i = 0; i < nitems(repo->privsep_children); i++) {
889 		if (repo->privsep_children[i].imsg_fd == -1)
890 			continue;
891 		imsg_clear(repo->privsep_children[i].ibuf);
892 		free(repo->privsep_children[i].ibuf);
893 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
894 		child_err = got_privsep_wait_for_child(
895 		    repo->privsep_children[i].pid);
896 		if (child_err && err == NULL)
897 			err = child_err;
898 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
899 		    err == NULL)
900 			err = got_error_from_errno("close");
901 	}
902 
903 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
904 	    err == NULL)
905 		err = got_error_from_errno("close");
906 
907 	if (repo->gotconfig)
908 		got_gotconfig_free(repo->gotconfig);
909 	free(repo->gitconfig_author_name);
910 	free(repo->gitconfig_author_email);
911 	for (i = 0; i < repo->ngitconfig_remotes; i++)
912 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
913 	free(repo->gitconfig_remotes);
914 	for (i = 0; i < repo->nextensions; i++)
915 		free(repo->extensions[i]);
916 	free(repo->extensions);
917 
918 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
919 		free((void *)pe->path);
920 	got_pathlist_free(&repo->packidx_paths);
921 	free(repo);
922 
923 	return err;
924 }
925 
926 void
927 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
928 {
929 	int i;
930 
931 	free(repo->name);
932 	repo->name = NULL;
933 	free(repo->fetch_url);
934 	repo->fetch_url = NULL;
935 	free(repo->send_url);
936 	repo->send_url = NULL;
937 	for (i = 0; i < repo->nfetch_branches; i++)
938 		free(repo->fetch_branches[i]);
939 	free(repo->fetch_branches);
940 	repo->fetch_branches = NULL;
941 	repo->nfetch_branches = 0;
942 	for (i = 0; i < repo->nsend_branches; i++)
943 		free(repo->send_branches[i]);
944 	free(repo->send_branches);
945 	repo->send_branches = NULL;
946 	repo->nsend_branches = 0;
947 }
948 
949 const struct got_error *
950 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
951     const char *input_path)
952 {
953 	const struct got_error *err = NULL;
954 	const char *repo_abspath = NULL;
955 	size_t repolen, len;
956 	char *canonpath, *path = NULL;
957 
958 	*in_repo_path = NULL;
959 
960 	canonpath = strdup(input_path);
961 	if (canonpath == NULL) {
962 		err = got_error_from_errno("strdup");
963 		goto done;
964 	}
965 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
966 	if (err)
967 		goto done;
968 
969 	repo_abspath = got_repo_get_path(repo);
970 
971 	if (canonpath[0] == '\0') {
972 		path = strdup(canonpath);
973 		if (path == NULL) {
974 			err = got_error_from_errno("strdup");
975 			goto done;
976 		}
977 	} else {
978 		path = realpath(canonpath, NULL);
979 		if (path == NULL) {
980 			if (errno != ENOENT) {
981 				err = got_error_from_errno2("realpath",
982 				    canonpath);
983 				goto done;
984 			}
985 			/*
986 			 * Path is not on disk.
987 			 * Assume it is already relative to repository root.
988 			 */
989 			path = strdup(canonpath);
990 			if (path == NULL) {
991 				err = got_error_from_errno("strdup");
992 				goto done;
993 			}
994 		}
995 
996 		repolen = strlen(repo_abspath);
997 		len = strlen(path);
998 
999 
1000 		if (strcmp(path, repo_abspath) == 0) {
1001 			free(path);
1002 			path = strdup("");
1003 			if (path == NULL) {
1004 				err = got_error_from_errno("strdup");
1005 				goto done;
1006 			}
1007 		} else if (len > repolen &&
1008 		    got_path_is_child(path, repo_abspath, repolen)) {
1009 			/* Matched an on-disk path inside repository. */
1010 			if (got_repo_is_bare(repo)) {
1011 				/*
1012 				 * Matched an on-disk path inside repository
1013 				 * database. Treat input as repository-relative.
1014 				 */
1015 				free(path);
1016 				path = canonpath;
1017 				canonpath = NULL;
1018 			} else {
1019 				char *child;
1020 				/* Strip common prefix with repository path. */
1021 				err = got_path_skip_common_ancestor(&child,
1022 				    repo_abspath, path);
1023 				if (err)
1024 					goto done;
1025 				free(path);
1026 				path = child;
1027 			}
1028 		} else {
1029 			/*
1030 			 * Matched unrelated on-disk path.
1031 			 * Treat input as repository-relative.
1032 			 */
1033 			free(path);
1034 			path = canonpath;
1035 			canonpath = NULL;
1036 		}
1037 	}
1038 
1039 	/* Make in-repository path absolute */
1040 	if (path[0] != '/') {
1041 		char *abspath;
1042 		if (asprintf(&abspath, "/%s", path) == -1) {
1043 			err = got_error_from_errno("asprintf");
1044 			goto done;
1045 		}
1046 		free(path);
1047 		path = abspath;
1048 	}
1049 
1050 done:
1051 	free(canonpath);
1052 	if (err)
1053 		free(path);
1054 	else
1055 		*in_repo_path = path;
1056 	return err;
1057 }
1058 
1059 static const struct got_error *
1060 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1061     const char *path_packidx)
1062 {
1063 	const struct got_error *err = NULL;
1064 	size_t i;
1065 
1066 	for (i = 0; i < repo->pack_cache_size; i++) {
1067 		if (repo->packidx_cache[i] == NULL)
1068 			break;
1069 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1070 		    path_packidx) == 0) {
1071 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1072 		}
1073 	}
1074 	if (i == repo->pack_cache_size) {
1075 		do {
1076 			i--;
1077 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1078 		    i == repo->pinned_packidx);
1079 		err = got_packidx_close(repo->packidx_cache[i]);
1080 		if (err)
1081 			return err;
1082 	}
1083 
1084 	repo->packidx_cache[i] = packidx;
1085 
1086 	return NULL;
1087 }
1088 
1089 int
1090 got_repo_is_packidx_filename(const char *name, size_t len)
1091 {
1092 	if (len != GOT_PACKIDX_NAMELEN)
1093 		return 0;
1094 
1095 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1096 		return 0;
1097 
1098 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1099 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1100 		return 0;
1101 
1102 	return 1;
1103 }
1104 
1105 static struct got_packidx_bloom_filter *
1106 get_packidx_bloom_filter(struct got_repository *repo,
1107     const char *path, size_t path_len)
1108 {
1109 	struct got_packidx_bloom_filter key;
1110 
1111 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1112 		return NULL; /* XXX */
1113 	key.path_len = path_len;
1114 
1115 	return RB_FIND(got_packidx_bloom_filter_tree,
1116 	    &repo->packidx_bloom_filters, &key);
1117 }
1118 
1119 int
1120 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1121     const char *path_packidx, struct got_object_id *id)
1122 {
1123 	struct got_packidx_bloom_filter *bf;
1124 
1125 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1126 	if (bf)
1127 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1128 
1129 	/* No bloom filter means this pack index must be searched. */
1130 	return 1;
1131 }
1132 
1133 static const struct got_error *
1134 add_packidx_bloom_filter(struct got_repository *repo,
1135     struct got_packidx *packidx, const char *path_packidx)
1136 {
1137 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1138 	struct got_packidx_bloom_filter *bf;
1139 	size_t len;
1140 
1141 	/*
1142 	 * Don't use bloom filters for very large pack index files.
1143 	 * Large pack files will contain a relatively large fraction
1144 	 * of our objects so we will likely need to visit them anyway.
1145 	 * The more objects a pack file contains the higher the probability
1146 	 * of a false-positive match from the bloom filter. And reading
1147 	 * all object IDs from a large pack index file can be expensive.
1148 	 */
1149 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1150 		return NULL;
1151 
1152 	/* Do we already have a filter for this pack index? */
1153 	if (get_packidx_bloom_filter(repo, path_packidx,
1154 	    strlen(path_packidx)) != NULL)
1155 		return NULL;
1156 
1157 	bf = calloc(1, sizeof(*bf));
1158 	if (bf == NULL)
1159 		return got_error_from_errno("calloc");
1160 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1161 	if (bf->bloom == NULL) {
1162 		free(bf);
1163 		return got_error_from_errno("calloc");
1164 	}
1165 	
1166 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1167 	if (len >= sizeof(bf->path)) {
1168 		free(bf->bloom);
1169 		free(bf);
1170 		return got_error(GOT_ERR_NO_SPACE);
1171 	}
1172 	bf->path_len = len;
1173 
1174 	/* Minimum size supported by our bloom filter is 1000 entries. */
1175 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1176 	for (i = 0; i < nobjects; i++) {
1177 		struct got_packidx_object_id *id;
1178 		id = &packidx->hdr.sorted_ids[i];
1179 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1180 	}
1181 
1182 	RB_INSERT(got_packidx_bloom_filter_tree,
1183 	    &repo->packidx_bloom_filters, bf);
1184 	return NULL;
1185 }
1186 
1187 const struct got_error *
1188 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1189     struct got_repository *repo, struct got_object_id *id)
1190 {
1191 	const struct got_error *err;
1192 	struct got_pathlist_entry *pe;
1193 	size_t i;
1194 
1195 	/* Search pack index cache. */
1196 	for (i = 0; i < repo->pack_cache_size; i++) {
1197 		if (repo->packidx_cache[i] == NULL)
1198 			break;
1199 		if (!got_repo_check_packidx_bloom_filter(repo,
1200 		    repo->packidx_cache[i]->path_packidx, id))
1201 			continue; /* object will not be found in this index */
1202 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1203 		if (*idx != -1) {
1204 			*packidx = repo->packidx_cache[i];
1205 			/*
1206 			 * Move this cache entry to the front. Repeatedly
1207 			 * searching a wrong pack index can be expensive.
1208 			 */
1209 			if (i > 0) {
1210 				memmove(&repo->packidx_cache[1],
1211 				    &repo->packidx_cache[0],
1212 				    i * sizeof(repo->packidx_cache[0]));
1213 				repo->packidx_cache[0] = *packidx;
1214 				if (repo->pinned_packidx >= 0 &&
1215 				    repo->pinned_packidx < i)
1216 					repo->pinned_packidx++;
1217 				else if (repo->pinned_packidx == i)
1218 					repo->pinned_packidx = 0;
1219 			}
1220 			return NULL;
1221 		}
1222 	}
1223 	/* No luck. Search the filesystem. */
1224 
1225 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1226 		const char *path_packidx = pe->path;
1227 		int is_cached = 0;
1228 
1229 		if (!got_repo_check_packidx_bloom_filter(repo,
1230 		    pe->path, id))
1231 			continue; /* object will not be found in this index */
1232 
1233 		for (i = 0; i < repo->pack_cache_size; i++) {
1234 			if (repo->packidx_cache[i] == NULL)
1235 				break;
1236 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1237 			    path_packidx) == 0) {
1238 				is_cached = 1;
1239 				break;
1240 			}
1241 		}
1242 		if (is_cached)
1243 			continue; /* already searched */
1244 
1245 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1246 		    path_packidx, 0);
1247 		if (err)
1248 			goto done;
1249 
1250 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1251 		if (err)
1252 			goto done;
1253 
1254 		err = cache_packidx(repo, *packidx, path_packidx);
1255 		if (err)
1256 			goto done;
1257 
1258 		*idx = got_packidx_get_object_idx(*packidx, id);
1259 		if (*idx != -1) {
1260 			err = NULL; /* found the object */
1261 			goto done;
1262 		}
1263 	}
1264 
1265 	err = got_error_no_obj(id);
1266 done:
1267 	return err;
1268 }
1269 
1270 const struct got_error *
1271 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1272     struct got_repository *repo)
1273 {
1274 	const struct got_error *err = NULL;
1275 	DIR *packdir = NULL;
1276 	struct dirent *dent;
1277 	char *path_packidx = NULL;
1278 	int packdir_fd;
1279 
1280 	packdir_fd = openat(got_repo_get_fd(repo),
1281 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1282 	if (packdir_fd == -1) {
1283 		return got_error_from_errno_fmt("openat: %s/%s",
1284 		    got_repo_get_path_git_dir(repo),
1285 		    GOT_OBJECTS_PACK_DIR);
1286 	}
1287 
1288 	packdir = fdopendir(packdir_fd);
1289 	if (packdir == NULL) {
1290 		err = got_error_from_errno("fdopendir");
1291 		goto done;
1292 	}
1293 
1294 	while ((dent = readdir(packdir)) != NULL) {
1295 		if (!got_repo_is_packidx_filename(dent->d_name,
1296 		    strlen(dent->d_name)))
1297 			continue;
1298 
1299 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1300 		    dent->d_name) == -1) {
1301 			err = got_error_from_errno("asprintf");
1302 			path_packidx = NULL;
1303 			break;
1304 		}
1305 
1306 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1307 		if (err)
1308 			break;
1309 	}
1310 done:
1311 	if (err)
1312 		free(path_packidx);
1313 	if (packdir && closedir(packdir) != 0 && err == NULL)
1314 		err = got_error_from_errno("closedir");
1315 	return err;
1316 }
1317 
1318 const struct got_error *
1319 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1320     struct got_repository *repo)
1321 {
1322 	const struct got_error *err;
1323 	size_t i;
1324 
1325 	*packidx = NULL;
1326 
1327 	/* Search pack index cache. */
1328 	for (i = 0; i < repo->pack_cache_size; i++) {
1329 		if (repo->packidx_cache[i] == NULL)
1330 			break;
1331 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1332 		    path_packidx) == 0) {
1333 			*packidx = repo->packidx_cache[i];
1334 			return NULL;
1335 		}
1336 	}
1337 	/* No luck. Search the filesystem. */
1338 
1339 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1340 	    path_packidx, 0);
1341 	if (err)
1342 		return err;
1343 
1344 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1345 	if (err)
1346 		goto done;
1347 
1348 	err = cache_packidx(repo, *packidx, path_packidx);
1349 done:
1350 	if (err) {
1351 		got_packidx_close(*packidx);
1352 		*packidx = NULL;
1353 	}
1354 	return err;
1355 }
1356 
1357 static const struct got_error *
1358 read_packfile_hdr(int fd, struct got_packidx *packidx)
1359 {
1360 	const struct got_error *err = NULL;
1361 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1362 	struct got_packfile_hdr hdr;
1363 	ssize_t n;
1364 
1365 	n = read(fd, &hdr, sizeof(hdr));
1366 	if (n < 0)
1367 		return got_error_from_errno("read");
1368 	if (n != sizeof(hdr))
1369 		return got_error(GOT_ERR_BAD_PACKFILE);
1370 
1371 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1372 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1373 	    be32toh(hdr.nobjects) != totobj)
1374 		err = got_error(GOT_ERR_BAD_PACKFILE);
1375 
1376 	return err;
1377 }
1378 
1379 static const struct got_error *
1380 open_packfile(int *fd, struct got_repository *repo,
1381     const char *relpath, struct got_packidx *packidx)
1382 {
1383 	const struct got_error *err = NULL;
1384 
1385 	*fd = openat(got_repo_get_fd(repo), relpath,
1386 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1387 	if (*fd == -1)
1388 		return got_error_from_errno_fmt("openat: %s/%s",
1389 		    got_repo_get_path_git_dir(repo), relpath);
1390 
1391 	if (packidx) {
1392 		err = read_packfile_hdr(*fd, packidx);
1393 		if (err) {
1394 			close(*fd);
1395 			*fd = -1;
1396 		}
1397 	}
1398 
1399 	return err;
1400 }
1401 
1402 const struct got_error *
1403 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1404     const char *path_packfile, struct got_packidx *packidx)
1405 {
1406 	const struct got_error *err = NULL;
1407 	struct got_pack *pack = NULL;
1408 	struct stat sb;
1409 	size_t i;
1410 
1411 	if (packp)
1412 		*packp = NULL;
1413 
1414 	for (i = 0; i < repo->pack_cache_size; i++) {
1415 		pack = &repo->packs[i];
1416 		if (pack->path_packfile == NULL)
1417 			break;
1418 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1419 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1420 	}
1421 
1422 	if (i == repo->pack_cache_size) {
1423 		struct got_pack tmp;
1424 		do {
1425 			i--;
1426 		} while (i > 0 && repo->pinned_pack >= 0 &&
1427 		    i == repo->pinned_pack);
1428 		err = got_pack_close(&repo->packs[i]);
1429 		if (err)
1430 			return err;
1431 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1432 			return got_error_from_errno("ftruncate");
1433 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1434 			return got_error_from_errno("ftruncate");
1435 		memcpy(&tmp, &repo->packs[i], sizeof(tmp)); 
1436 		memcpy(&repo->packs[i], &repo->packs[0],
1437 		    sizeof(repo->packs[i]));
1438 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1439 		if (repo->pinned_pack == 0)
1440 			repo->pinned_pack = i;
1441 		else if (repo->pinned_pack == i)
1442 			repo->pinned_pack = 0;
1443 		i = 0;
1444 	}
1445 
1446 	pack = &repo->packs[i];
1447 
1448 	pack->path_packfile = strdup(path_packfile);
1449 	if (pack->path_packfile == NULL) {
1450 		err = got_error_from_errno("strdup");
1451 		goto done;
1452 	}
1453 
1454 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1455 	if (err)
1456 		goto done;
1457 
1458 	if (fstat(pack->fd, &sb) != 0) {
1459 		err = got_error_from_errno("fstat");
1460 		goto done;
1461 	}
1462 	pack->filesize = sb.st_size;
1463 
1464 	pack->privsep_child = NULL;
1465 
1466 #ifndef GOT_PACK_NO_MMAP
1467 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1468 	    pack->fd, 0);
1469 	if (pack->map == MAP_FAILED) {
1470 		if (errno != ENOMEM) {
1471 			err = got_error_from_errno("mmap");
1472 			goto done;
1473 		}
1474 		pack->map = NULL; /* fall back to read(2) */
1475 	}
1476 #endif
1477 done:
1478 	if (err) {
1479 		if (pack) {
1480 			free(pack->path_packfile);
1481 			memset(pack, 0, sizeof(*pack));
1482 		}
1483 	} else if (packp)
1484 		*packp = pack;
1485 	return err;
1486 }
1487 
1488 struct got_pack *
1489 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1490 {
1491 	struct got_pack *pack = NULL;
1492 	size_t i;
1493 
1494 	for (i = 0; i < repo->pack_cache_size; i++) {
1495 		pack = &repo->packs[i];
1496 		if (pack->path_packfile == NULL)
1497 			break;
1498 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1499 			return pack;
1500 	}
1501 
1502 	return NULL;
1503 }
1504 
1505 const struct got_error *
1506 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1507     struct got_pack *pack)
1508 {
1509 	size_t i;
1510 	int pinned_pack = -1, pinned_packidx = -1;
1511 
1512 	for (i = 0; i < repo->pack_cache_size; i++) {
1513 		if (repo->packidx_cache[i] &&
1514 		    strcmp(repo->packidx_cache[i]->path_packidx,
1515 		    packidx->path_packidx) == 0)
1516 			pinned_packidx = i;
1517 		if (repo->packs[i].path_packfile &&
1518 		    strcmp(repo->packs[i].path_packfile,
1519 		    pack->path_packfile) == 0)
1520 			pinned_pack = i;
1521 	}
1522 
1523 	if (pinned_packidx == -1 || pinned_pack == -1)
1524 		return got_error(GOT_ERR_PIN_PACK);
1525 
1526 	repo->pinned_pack = pinned_pack;
1527 	repo->pinned_packidx = pinned_packidx;
1528 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1529 	return NULL;
1530 }
1531 
1532 struct got_pack *
1533 got_repo_get_pinned_pack(struct got_repository *repo)
1534 {
1535 	if (repo->pinned_pack >= 0 &&
1536 	    repo->pinned_pack < repo->pack_cache_size)
1537 		return &repo->packs[repo->pinned_pack];
1538 
1539 	return NULL;
1540 }
1541 
1542 void
1543 got_repo_unpin_pack(struct got_repository *repo)
1544 {
1545 	repo->pinned_packidx = -1;
1546 	repo->pinned_pack = -1;
1547 	repo->pinned_pid = 0;
1548 }
1549 
1550 const struct got_error *
1551 got_repo_init(const char *repo_path)
1552 {
1553 	const struct got_error *err = NULL;
1554 	const char *dirnames[] = {
1555 		GOT_OBJECTS_DIR,
1556 		GOT_OBJECTS_PACK_DIR,
1557 		GOT_REFS_DIR,
1558 	};
1559 	const char *description_str = "Unnamed repository; "
1560 	    "edit this file 'description' to name the repository.";
1561 	const char *headref_str = "ref: refs/heads/main";
1562 	const char *gitconfig_str = "[core]\n"
1563 	    "\trepositoryformatversion = 0\n"
1564 	    "\tfilemode = true\n"
1565 	    "\tbare = true\n";
1566 	char *path;
1567 	size_t i;
1568 
1569 	if (!got_path_dir_is_empty(repo_path))
1570 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1571 
1572 	for (i = 0; i < nitems(dirnames); i++) {
1573 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1574 			return got_error_from_errno("asprintf");
1575 		}
1576 		err = got_path_mkdir(path);
1577 		free(path);
1578 		if (err)
1579 			return err;
1580 	}
1581 
1582 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1583 		return got_error_from_errno("asprintf");
1584 	err = got_path_create_file(path, description_str);
1585 	free(path);
1586 	if (err)
1587 		return err;
1588 
1589 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1590 		return got_error_from_errno("asprintf");
1591 	err = got_path_create_file(path, headref_str);
1592 	free(path);
1593 	if (err)
1594 		return err;
1595 
1596 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1597 		return got_error_from_errno("asprintf");
1598 	err = got_path_create_file(path, gitconfig_str);
1599 	free(path);
1600 	if (err)
1601 		return err;
1602 
1603 	return NULL;
1604 }
1605 
1606 static const struct got_error *
1607 match_packed_object(struct got_object_id **unique_id,
1608     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1609 {
1610 	const struct got_error *err = NULL;
1611 	struct got_object_id_queue matched_ids;
1612 	struct got_pathlist_entry *pe;
1613 
1614 	STAILQ_INIT(&matched_ids);
1615 
1616 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1617 		const char *path_packidx = pe->path;
1618 		struct got_packidx *packidx;
1619 		struct got_object_qid *qid;
1620 
1621 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1622 		    path_packidx, 0);
1623 		if (err)
1624 			break;
1625 
1626 		err = got_packidx_match_id_str_prefix(&matched_ids,
1627 		    packidx, id_str_prefix);
1628 		if (err) {
1629 			got_packidx_close(packidx);
1630 			break;
1631 		}
1632 		err = got_packidx_close(packidx);
1633 		if (err)
1634 			break;
1635 
1636 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1637 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1638 				int matched_type;
1639 				err = got_object_get_type(&matched_type, repo,
1640 				    &qid->id);
1641 				if (err)
1642 					goto done;
1643 				if (matched_type != obj_type)
1644 					continue;
1645 			}
1646 			if (*unique_id == NULL) {
1647 				*unique_id = got_object_id_dup(&qid->id);
1648 				if (*unique_id == NULL) {
1649 					err = got_error_from_errno("malloc");
1650 					goto done;
1651 				}
1652 			} else {
1653 				if (got_object_id_cmp(*unique_id,
1654 				    &qid->id) == 0)
1655 					continue; /* packed multiple times */
1656 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1657 				goto done;
1658 			}
1659 		}
1660 	}
1661 done:
1662 	got_object_id_queue_free(&matched_ids);
1663 	if (err) {
1664 		free(*unique_id);
1665 		*unique_id = NULL;
1666 	}
1667 	return err;
1668 }
1669 
1670 static const struct got_error *
1671 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1672     const char *object_dir, const char *id_str_prefix, int obj_type,
1673     struct got_repository *repo)
1674 {
1675 	const struct got_error *err = NULL;
1676 	char *path;
1677 	DIR *dir = NULL;
1678 	struct dirent *dent;
1679 	struct got_object_id id;
1680 
1681 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1682 		err = got_error_from_errno("asprintf");
1683 		goto done;
1684 	}
1685 
1686 	dir = opendir(path);
1687 	if (dir == NULL) {
1688 		if (errno == ENOENT) {
1689 			err = NULL;
1690 			goto done;
1691 		}
1692 		err = got_error_from_errno2("opendir", path);
1693 		goto done;
1694 	}
1695 	while ((dent = readdir(dir)) != NULL) {
1696 		char *id_str;
1697 		int cmp;
1698 
1699 		if (strcmp(dent->d_name, ".") == 0 ||
1700 		    strcmp(dent->d_name, "..") == 0)
1701 			continue;
1702 
1703 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1704 			err = got_error_from_errno("asprintf");
1705 			goto done;
1706 		}
1707 
1708 		if (!got_parse_sha1_digest(id.sha1, id_str))
1709 			continue;
1710 
1711 		/*
1712 		 * Directory entries do not necessarily appear in
1713 		 * sorted order, so we must iterate over all of them.
1714 		 */
1715 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1716 		if (cmp != 0) {
1717 			free(id_str);
1718 			continue;
1719 		}
1720 
1721 		if (*unique_id == NULL) {
1722 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1723 				int matched_type;
1724 				err = got_object_get_type(&matched_type, repo,
1725 				    &id);
1726 				if (err)
1727 					goto done;
1728 				if (matched_type != obj_type)
1729 					continue;
1730 			}
1731 			*unique_id = got_object_id_dup(&id);
1732 			if (*unique_id == NULL) {
1733 				err = got_error_from_errno("got_object_id_dup");
1734 				free(id_str);
1735 				goto done;
1736 			}
1737 		} else {
1738 			if (got_object_id_cmp(*unique_id, &id) == 0)
1739 				continue; /* both packed and loose */
1740 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1741 			free(id_str);
1742 			goto done;
1743 		}
1744 	}
1745 done:
1746 	if (dir && closedir(dir) != 0 && err == NULL)
1747 		err = got_error_from_errno("closedir");
1748 	if (err) {
1749 		free(*unique_id);
1750 		*unique_id = NULL;
1751 	}
1752 	free(path);
1753 	return err;
1754 }
1755 
1756 const struct got_error *
1757 got_repo_match_object_id_prefix(struct got_object_id **id,
1758     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1759 {
1760 	const struct got_error *err = NULL;
1761 	char *path_objects = got_repo_get_path_objects(repo);
1762 	char *object_dir = NULL;
1763 	size_t len;
1764 	int i;
1765 
1766 	*id = NULL;
1767 
1768 	len = strlen(id_str_prefix);
1769 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1770 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1771 
1772 	for (i = 0; i < len; i++) {
1773 		if (isxdigit((unsigned char)id_str_prefix[i]))
1774 			continue;
1775 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1776 	}
1777 
1778 	if (len >= 2) {
1779 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1780 		if (err)
1781 			goto done;
1782 		object_dir = strndup(id_str_prefix, 2);
1783 		if (object_dir == NULL) {
1784 			err = got_error_from_errno("strdup");
1785 			goto done;
1786 		}
1787 		err = match_loose_object(id, path_objects, object_dir,
1788 		    id_str_prefix, obj_type, repo);
1789 	} else if (len == 1) {
1790 		int i;
1791 		for (i = 0; i < 0xf; i++) {
1792 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1793 			    == -1) {
1794 				err = got_error_from_errno("asprintf");
1795 				goto done;
1796 			}
1797 			err = match_packed_object(id, repo, object_dir,
1798 			    obj_type);
1799 			if (err)
1800 				goto done;
1801 			err = match_loose_object(id, path_objects, object_dir,
1802 			    id_str_prefix, obj_type, repo);
1803 			if (err)
1804 				goto done;
1805 		}
1806 	} else {
1807 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1808 		goto done;
1809 	}
1810 done:
1811 	free(object_dir);
1812 	if (err) {
1813 		free(*id);
1814 		*id = NULL;
1815 	} else if (*id == NULL) {
1816 		switch (obj_type) {
1817 		case GOT_OBJ_TYPE_BLOB:
1818 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1819 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1820 			break;
1821 		case GOT_OBJ_TYPE_TREE:
1822 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1823 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1824 			break;
1825 		case GOT_OBJ_TYPE_COMMIT:
1826 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1827 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1828 			break;
1829 		case GOT_OBJ_TYPE_TAG:
1830 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1831 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1832 			break;
1833 		default:
1834 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1835 			break;
1836 		}
1837 	}
1838 
1839 	return err;
1840 }
1841 
1842 const struct got_error *
1843 got_repo_match_object_id(struct got_object_id **id, char **label,
1844     const char *id_str, int obj_type, struct got_reflist_head *refs,
1845     struct got_repository *repo)
1846 {
1847 	const struct got_error *err;
1848 	struct got_tag_object *tag;
1849 	struct got_reference *ref = NULL;
1850 
1851 	*id = NULL;
1852 	if (label)
1853 		*label = NULL;
1854 
1855 	if (refs) {
1856 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1857 		    refs, repo);
1858 		if (err == NULL) {
1859 			*id = got_object_id_dup(
1860 			    got_object_tag_get_object_id(tag));
1861 			if (*id == NULL)
1862 				err = got_error_from_errno("got_object_id_dup");
1863 			else if (label && asprintf(label, "refs/tags/%s",
1864 			    got_object_tag_get_name(tag)) == -1) {
1865 				err = got_error_from_errno("asprintf");
1866 				free(*id);
1867 				*id = NULL;
1868 			}
1869 			got_object_tag_close(tag);
1870 			return err;
1871 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1872 		    err->code != GOT_ERR_NO_OBJ)
1873 			return err;
1874 	}
1875 
1876 	err = got_ref_open(&ref, repo, id_str, 0);
1877 	if (err == NULL) {
1878 		err = got_ref_resolve(id, repo, ref);
1879 		if (err)
1880 			goto done;
1881 		if (label) {
1882 			*label = strdup(got_ref_get_name(ref));
1883 			if (*label == NULL) {
1884 				err = got_error_from_errno("strdup");
1885 				goto done;
1886 			}
1887 		}
1888 	} else {
1889 		if (err->code != GOT_ERR_NOT_REF &&
1890 		    err->code != GOT_ERR_BAD_REF_NAME)
1891 			goto done;
1892 		err = got_repo_match_object_id_prefix(id, id_str,
1893 		    obj_type, repo);
1894 		if (err) {
1895 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1896 				err = got_error_not_ref(id_str);
1897 			goto done;
1898 		}
1899 		if (label) {
1900 			err = got_object_id_str(label, *id);
1901 			if (*label == NULL) {
1902 				err = got_error_from_errno("strdup");
1903 				goto done;
1904 			}
1905 		}
1906 	}
1907 done:
1908 	if (ref)
1909 		got_ref_close(ref);
1910 	return err;
1911 }
1912 
1913 const struct got_error *
1914 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1915     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1916 {
1917 	const struct got_error *err = NULL;
1918 	struct got_reflist_entry *re;
1919 	struct got_object_id *tag_id;
1920 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1921 
1922 	*tag = NULL;
1923 
1924 	TAILQ_FOREACH(re, refs, entry) {
1925 		const char *refname;
1926 		refname = got_ref_get_name(re->ref);
1927 		if (got_ref_is_symbolic(re->ref))
1928 			continue;
1929 		if (strncmp(refname, "refs/tags/", 10) != 0)
1930 			continue;
1931 		if (!name_is_absolute)
1932 			refname += strlen("refs/tags/");
1933 		if (strcmp(refname, name) != 0)
1934 			continue;
1935 		err = got_ref_resolve(&tag_id, repo, re->ref);
1936 		if (err)
1937 			break;
1938 		err = got_object_open_as_tag(tag, repo, tag_id);
1939 		free(tag_id);
1940 		if (err)
1941 			break;
1942 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1943 		    got_object_tag_get_object_type(*tag) == obj_type)
1944 			break;
1945 		got_object_tag_close(*tag);
1946 		*tag = NULL;
1947 	}
1948 
1949 	if (err == NULL && *tag == NULL)
1950 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1951 		    GOT_OBJ_LABEL_TAG, name);
1952 	return err;
1953 }
1954 
1955 static const struct got_error *
1956 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1957     const char *name, mode_t mode, struct got_object_id *blob_id)
1958 {
1959 	const struct got_error *err = NULL;
1960 
1961 	 *new_te = NULL;
1962 
1963 	*new_te = calloc(1, sizeof(**new_te));
1964 	if (*new_te == NULL)
1965 		return got_error_from_errno("calloc");
1966 
1967 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1968 	    sizeof((*new_te)->name)) {
1969 		err = got_error(GOT_ERR_NO_SPACE);
1970 		goto done;
1971 	}
1972 
1973 	if (S_ISLNK(mode)) {
1974 		(*new_te)->mode = S_IFLNK;
1975 	} else {
1976 		(*new_te)->mode = S_IFREG;
1977 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1978 	}
1979 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1980 done:
1981 	if (err && *new_te) {
1982 		free(*new_te);
1983 		*new_te = NULL;
1984 	}
1985 	return err;
1986 }
1987 
1988 static const struct got_error *
1989 import_file(struct got_tree_entry **new_te, struct dirent *de,
1990     const char *path, struct got_repository *repo)
1991 {
1992 	const struct got_error *err;
1993 	struct got_object_id *blob_id = NULL;
1994 	char *filepath;
1995 	struct stat sb;
1996 
1997 	if (asprintf(&filepath, "%s%s%s", path,
1998 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1999 		return got_error_from_errno("asprintf");
2000 
2001 	if (lstat(filepath, &sb) != 0) {
2002 		err = got_error_from_errno2("lstat", path);
2003 		goto done;
2004 	}
2005 
2006 	err = got_object_blob_create(&blob_id, filepath, repo);
2007 	if (err)
2008 		goto done;
2009 
2010 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2011 	    blob_id);
2012 done:
2013 	free(filepath);
2014 	if (err)
2015 		free(blob_id);
2016 	return err;
2017 }
2018 
2019 static const struct got_error *
2020 insert_tree_entry(struct got_tree_entry *new_te,
2021     struct got_pathlist_head *paths)
2022 {
2023 	const struct got_error *err = NULL;
2024 	struct got_pathlist_entry *new_pe;
2025 
2026 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2027 	if (err)
2028 		return err;
2029 	if (new_pe == NULL)
2030 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2031 	return NULL;
2032 }
2033 
2034 static const struct got_error *write_tree(struct got_object_id **,
2035     const char *, struct got_pathlist_head *, struct got_repository *,
2036     got_repo_import_cb progress_cb, void *progress_arg);
2037 
2038 static const struct got_error *
2039 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2040     const char *path, struct got_pathlist_head *ignores,
2041     struct got_repository *repo,
2042     got_repo_import_cb progress_cb, void *progress_arg)
2043 {
2044 	const struct got_error *err;
2045 	struct got_object_id *id = NULL;
2046 	char *subdirpath;
2047 
2048 	if (asprintf(&subdirpath, "%s%s%s", path,
2049 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2050 		return got_error_from_errno("asprintf");
2051 
2052 	(*new_te) = calloc(1, sizeof(**new_te));
2053 	if (*new_te == NULL)
2054 		return got_error_from_errno("calloc");
2055 	(*new_te)->mode = S_IFDIR;
2056 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2057 	    sizeof((*new_te)->name)) {
2058 		err = got_error(GOT_ERR_NO_SPACE);
2059 		goto done;
2060 	}
2061 	err = write_tree(&id, subdirpath, ignores,  repo,
2062 	    progress_cb, progress_arg);
2063 	if (err)
2064 		goto done;
2065 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2066 	
2067 done:
2068 	free(id);
2069 	free(subdirpath);
2070 	if (err) {
2071 		free(*new_te);
2072 		*new_te = NULL;
2073 	}
2074 	return err;
2075 }
2076 
2077 static const struct got_error *
2078 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2079     struct got_pathlist_head *ignores, struct got_repository *repo,
2080     got_repo_import_cb progress_cb, void *progress_arg)
2081 {
2082 	const struct got_error *err = NULL;
2083 	DIR *dir;
2084 	struct dirent *de;
2085 	int nentries;
2086 	struct got_tree_entry *new_te = NULL;
2087 	struct got_pathlist_head paths;
2088 	struct got_pathlist_entry *pe;
2089 
2090 	*new_tree_id = NULL;
2091 
2092 	TAILQ_INIT(&paths);
2093 
2094 	dir = opendir(path_dir);
2095 	if (dir == NULL) {
2096 		err = got_error_from_errno2("opendir", path_dir);
2097 		goto done;
2098 	}
2099 
2100 	nentries = 0;
2101 	while ((de = readdir(dir)) != NULL) {
2102 		int ignore = 0;
2103 		int type;
2104 
2105 		if (strcmp(de->d_name, ".") == 0 ||
2106 		    strcmp(de->d_name, "..") == 0)
2107 			continue;
2108 
2109 		TAILQ_FOREACH(pe, ignores, entry) {
2110 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2111 				ignore = 1;
2112 				break;
2113 			}
2114 		}
2115 		if (ignore)
2116 			continue;
2117 
2118 		err = got_path_dirent_type(&type, path_dir, de);
2119 		if (err)
2120 			goto done;
2121 
2122 		if (type == DT_DIR) {
2123 			err = import_subdir(&new_te, de, path_dir,
2124 			    ignores, repo, progress_cb, progress_arg);
2125 			if (err) {
2126 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2127 					goto done;
2128 				err = NULL;
2129 				continue;
2130 			}
2131 		} else if (type == DT_REG || type == DT_LNK) {
2132 			err = import_file(&new_te, de, path_dir, repo);
2133 			if (err)
2134 				goto done;
2135 		} else
2136 			continue;
2137 
2138 		err = insert_tree_entry(new_te, &paths);
2139 		if (err)
2140 			goto done;
2141 		nentries++;
2142 	}
2143 
2144 	if (TAILQ_EMPTY(&paths)) {
2145 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2146 		    "cannot create tree without any entries");
2147 		goto done;
2148 	}
2149 
2150 	TAILQ_FOREACH(pe, &paths, entry) {
2151 		struct got_tree_entry *te = pe->data;
2152 		char *path;
2153 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2154 			continue;
2155 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2156 			err = got_error_from_errno("asprintf");
2157 			goto done;
2158 		}
2159 		err = (*progress_cb)(progress_arg, path);
2160 		free(path);
2161 		if (err)
2162 			goto done;
2163 	}
2164 
2165 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2166 done:
2167 	if (dir)
2168 		closedir(dir);
2169 	got_pathlist_free(&paths);
2170 	return err;
2171 }
2172 
2173 const struct got_error *
2174 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2175     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2176     struct got_repository *repo, got_repo_import_cb progress_cb,
2177     void *progress_arg)
2178 {
2179 	const struct got_error *err;
2180 	struct got_object_id *new_tree_id;
2181 
2182 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2183 	    progress_cb, progress_arg);
2184 	if (err)
2185 		return err;
2186 
2187 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2188 	    author, time(NULL), author, time(NULL), logmsg, repo);
2189 	free(new_tree_id);
2190 	return err;
2191 }
2192 
2193 const struct got_error *
2194 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2195     struct got_repository *repo)
2196 {
2197 	const struct got_error *err = NULL;
2198 	char *path_objects = NULL, *path = NULL;
2199 	DIR *dir = NULL;
2200 	struct got_object_id id;
2201 	int i;
2202 
2203 	*nobjects = 0;
2204 	*ondisk_size = 0;
2205 
2206 	path_objects = got_repo_get_path_objects(repo);
2207 	if (path_objects == NULL)
2208 		return got_error_from_errno("got_repo_get_path_objects");
2209 
2210 	for (i = 0; i <= 0xff; i++) {
2211 		struct dirent *dent;
2212 
2213 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2214 			err = got_error_from_errno("asprintf");
2215 			break;
2216 		}
2217 
2218 		dir = opendir(path);
2219 		if (dir == NULL) {
2220 			if (errno == ENOENT) {
2221 				err = NULL;
2222 				continue;
2223 			}
2224 			err = got_error_from_errno2("opendir", path);
2225 			break;
2226 		}
2227 
2228 		while ((dent = readdir(dir)) != NULL) {
2229 			char *id_str;
2230 			int fd;
2231 			struct stat sb;
2232 
2233 			if (strcmp(dent->d_name, ".") == 0 ||
2234 			    strcmp(dent->d_name, "..") == 0)
2235 				continue;
2236 
2237 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2238 				err = got_error_from_errno("asprintf");
2239 				goto done;
2240 			}
2241 
2242 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2243 				free(id_str);
2244 				continue;
2245 			}
2246 			free(id_str);
2247 
2248 			err = got_object_open_loose_fd(&fd, &id, repo);
2249 			if (err)
2250 				goto done;
2251 
2252 			if (fstat(fd, &sb) == -1) {
2253 				err = got_error_from_errno("fstat");
2254 				close(fd);
2255 				goto done;
2256 			}
2257 			(*nobjects)++;
2258 			(*ondisk_size) += sb.st_size;
2259 
2260 			if (close(fd) == -1) {
2261 				err = got_error_from_errno("close");
2262 				goto done;
2263 			}
2264 		}
2265 
2266 		if (closedir(dir) != 0) {
2267 			err = got_error_from_errno("closedir");
2268 			goto done;
2269 		}
2270 		dir = NULL;
2271 
2272 		free(path);
2273 		path = NULL;
2274 	}
2275 done:
2276 	if (dir && closedir(dir) != 0 && err == NULL)
2277 		err = got_error_from_errno("closedir");
2278 
2279 	if (err) {
2280 		*nobjects = 0;
2281 		*ondisk_size = 0;
2282 	}
2283 	free(path_objects);
2284 	free(path);
2285 	return err;
2286 }
2287 
2288 const struct got_error *
2289 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2290     off_t *total_packsize, struct got_repository *repo)
2291 {
2292 	const struct got_error *err = NULL;
2293 	DIR *packdir = NULL;
2294 	struct dirent *dent;
2295 	struct got_packidx *packidx = NULL;
2296 	char *path_packidx;
2297 	char *path_packfile;
2298 	int packdir_fd;
2299 	struct stat sb;
2300 
2301 	*npackfiles = 0;
2302 	*nobjects = 0;
2303 	*total_packsize = 0;
2304 
2305 	packdir_fd = openat(got_repo_get_fd(repo),
2306 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2307 	if (packdir_fd == -1) {
2308 		return got_error_from_errno_fmt("openat: %s/%s",
2309 		    got_repo_get_path_git_dir(repo),
2310 		    GOT_OBJECTS_PACK_DIR);
2311 	}
2312 
2313 	packdir = fdopendir(packdir_fd);
2314 	if (packdir == NULL) {
2315 		err = got_error_from_errno("fdopendir");
2316 		goto done;
2317 	}
2318 
2319 	while ((dent = readdir(packdir)) != NULL) {
2320 		if (!got_repo_is_packidx_filename(dent->d_name,
2321 		    strlen(dent->d_name)))
2322 			continue;
2323 
2324 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2325 		    dent->d_name) == -1) {
2326 			err = got_error_from_errno("asprintf");
2327 			goto done;
2328 		}
2329 
2330 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2331 		    path_packidx, 0);
2332 		free(path_packidx);
2333 		if (err)
2334 			goto done;
2335 
2336 		if (fstat(packidx->fd, &sb) == -1)
2337 			goto done;
2338 		*total_packsize += sb.st_size;
2339 
2340 		err = got_packidx_get_packfile_path(&path_packfile,
2341 		    packidx->path_packidx);
2342 		if (err)
2343 			goto done;
2344 
2345 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2346 		    0) == -1) {
2347 			free(path_packfile);
2348 			goto done;
2349 		}
2350 		free(path_packfile);
2351 		*total_packsize += sb.st_size;
2352 
2353 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2354 
2355 		(*npackfiles)++;
2356 
2357 		got_packidx_close(packidx);
2358 		packidx = NULL;
2359 	}
2360 done:
2361 	if (packidx)
2362 		got_packidx_close(packidx);
2363 	if (packdir && closedir(packdir) != 0 && err == NULL)
2364 		err = got_error_from_errno("closedir");
2365 	if (err) {
2366 		*npackfiles = 0;
2367 		*nobjects = 0;
2368 		*total_packsize = 0;
2369 	}
2370 	return err;
2371 }
2372 
2373 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2374     got_packidx_bloom_filter_cmp);