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 <endian.h>
27 #include <fcntl.h>
28 #include <fnmatch.h>
29 #include <limits.h>
30 #include <dirent.h>
31 #include <stdlib.h>
32 #include <stdio.h>
33 #include <sha1.h>
34 #include <string.h>
35 #include <time.h>
36 #include <unistd.h>
37 #include <zlib.h>
38 #include <errno.h>
39 #include <libgen.h>
40 #include <stdint.h>
41 #include <imsg.h>
42 #include <uuid.h>
43 
44 #include "got_error.h"
45 #include "got_reference.h"
46 #include "got_repository.h"
47 #include "got_path.h"
48 #include "got_cancel.h"
49 #include "got_object.h"
50 
51 #include "got_lib_delta.h"
52 #include "got_lib_inflate.h"
53 #include "got_lib_object.h"
54 #include "got_lib_object_parse.h"
55 #include "got_lib_object_create.h"
56 #include "got_lib_pack.h"
57 #include "got_lib_privsep.h"
58 #include "got_lib_sha1.h"
59 #include "got_lib_object_cache.h"
60 #include "got_lib_repository.h"
61 #include "got_lib_gotconfig.h"
62 
63 #ifndef nitems
64 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
65 #endif
66 
67 const char *
68 got_repo_get_path(struct got_repository *repo)
69 {
70 	return repo->path;
71 }
72 
73 const char *
74 got_repo_get_path_git_dir(struct got_repository *repo)
75 {
76 	return repo->path_git_dir;
77 }
78 
79 int
80 got_repo_get_fd(struct got_repository *repo)
81 {
82 	return repo->gitdir_fd;
83 }
84 
85 const char *
86 got_repo_get_gitconfig_author_name(struct got_repository *repo)
87 {
88 	return repo->gitconfig_author_name;
89 }
90 
91 const char *
92 got_repo_get_gitconfig_author_email(struct got_repository *repo)
93 {
94 	return repo->gitconfig_author_email;
95 }
96 
97 const char *
98 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
99 {
100 	return repo->global_gitconfig_author_name;
101 }
102 
103 const char *
104 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
105 {
106 	return repo->global_gitconfig_author_email;
107 }
108 
109 const char *
110 got_repo_get_gitconfig_owner(struct got_repository *repo)
111 {
112 	return repo->gitconfig_owner;
113 }
114 
115 void
116 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
117     struct got_repository *repo)
118 {
119 	*extensions = repo->extensions;
120 	*nextensions = repo->nextensions;
121 }
122 
123 int
124 got_repo_is_bare(struct got_repository *repo)
125 {
126 	return (strcmp(repo->path, repo->path_git_dir) == 0);
127 }
128 
129 static char *
130 get_path_git_child(struct got_repository *repo, const char *basename)
131 {
132 	char *path_child;
133 
134 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
135 	    basename) == -1)
136 		return NULL;
137 
138 	return path_child;
139 }
140 
141 char *
142 got_repo_get_path_objects(struct got_repository *repo)
143 {
144 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
145 }
146 
147 char *
148 got_repo_get_path_objects_pack(struct got_repository *repo)
149 {
150 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
151 }
152 
153 char *
154 got_repo_get_path_refs(struct got_repository *repo)
155 {
156 	return get_path_git_child(repo, GOT_REFS_DIR);
157 }
158 
159 char *
160 got_repo_get_path_packed_refs(struct got_repository *repo)
161 {
162 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
163 }
164 
165 static char *
166 get_path_head(struct got_repository *repo)
167 {
168 	return get_path_git_child(repo, GOT_HEAD_FILE);
169 }
170 
171 char *
172 got_repo_get_path_gitconfig(struct got_repository *repo)
173 {
174 	return get_path_git_child(repo, GOT_GITCONFIG);
175 }
176 
177 char *
178 got_repo_get_path_gotconfig(struct got_repository *repo)
179 {
180 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
181 }
182 
183 const struct got_gotconfig *
184 got_repo_get_gotconfig(struct got_repository *repo)
185 {
186 	return repo->gotconfig;
187 }
188 
189 void
190 got_repo_get_gitconfig_remotes(int *nremotes,
191     const struct got_remote_repo **remotes, struct got_repository *repo)
192 {
193 	*nremotes = repo->ngitconfig_remotes;
194 	*remotes = repo->gitconfig_remotes;
195 }
196 
197 static int
198 is_git_repo(struct got_repository *repo)
199 {
200 	const char *path_git = got_repo_get_path_git_dir(repo);
201 	char *path_objects = got_repo_get_path_objects(repo);
202 	char *path_refs = got_repo_get_path_refs(repo);
203 	char *path_head = get_path_head(repo);
204 	int ret = 0;
205 	struct stat sb;
206 	struct got_reference *head_ref;
207 
208 	if (lstat(path_git, &sb) == -1)
209 		goto done;
210 	if (!S_ISDIR(sb.st_mode))
211 		goto done;
212 
213 	if (lstat(path_objects, &sb) == -1)
214 		goto done;
215 	if (!S_ISDIR(sb.st_mode))
216 		goto done;
217 
218 	if (lstat(path_refs, &sb) == -1)
219 		goto done;
220 	if (!S_ISDIR(sb.st_mode))
221 		goto done;
222 
223 	if (lstat(path_head, &sb) == -1)
224 		goto done;
225 	if (!S_ISREG(sb.st_mode))
226 		goto done;
227 
228 	/* Check if the HEAD reference can be opened. */
229 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
230 		goto done;
231 	got_ref_close(head_ref);
232 
233 	ret = 1;
234 done:
235 	free(path_objects);
236 	free(path_refs);
237 	free(path_head);
238 	return ret;
239 
240 }
241 
242 const struct got_error *
243 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
244     struct got_object *obj)
245 {
246 #ifndef GOT_NO_OBJ_CACHE
247 	const struct got_error *err = NULL;
248 	err = got_object_cache_add(&repo->objcache, id, obj);
249 	if (err) {
250 		if (err->code == GOT_ERR_OBJ_EXISTS ||
251 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
252 			err = NULL;
253 		return err;
254 	}
255 	obj->refcnt++;
256 #endif
257 	return NULL;
258 }
259 
260 struct got_object *
261 got_repo_get_cached_object(struct got_repository *repo,
262     struct got_object_id *id)
263 {
264 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
265 }
266 
267 const struct got_error *
268 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
269     struct got_tree_object *tree)
270 {
271 #ifndef GOT_NO_OBJ_CACHE
272 	const struct got_error *err = NULL;
273 	err = got_object_cache_add(&repo->treecache, id, tree);
274 	if (err) {
275 		if (err->code == GOT_ERR_OBJ_EXISTS ||
276 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
277 			err = NULL;
278 		return err;
279 	}
280 	tree->refcnt++;
281 #endif
282 	return NULL;
283 }
284 
285 struct got_tree_object *
286 got_repo_get_cached_tree(struct got_repository *repo,
287     struct got_object_id *id)
288 {
289 	return (struct got_tree_object *)got_object_cache_get(
290 	    &repo->treecache, id);
291 }
292 
293 const struct got_error *
294 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
295     struct got_commit_object *commit)
296 {
297 #ifndef GOT_NO_OBJ_CACHE
298 	const struct got_error *err = NULL;
299 	err = got_object_cache_add(&repo->commitcache, id, commit);
300 	if (err) {
301 		if (err->code == GOT_ERR_OBJ_EXISTS ||
302 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
303 			err = NULL;
304 		return err;
305 	}
306 	commit->refcnt++;
307 #endif
308 	return NULL;
309 }
310 
311 struct got_commit_object *
312 got_repo_get_cached_commit(struct got_repository *repo,
313     struct got_object_id *id)
314 {
315 	return (struct got_commit_object *)got_object_cache_get(
316 	    &repo->commitcache, id);
317 }
318 
319 const struct got_error *
320 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
321     struct got_tag_object *tag)
322 {
323 #ifndef GOT_NO_OBJ_CACHE
324 	const struct got_error *err = NULL;
325 	err = got_object_cache_add(&repo->tagcache, id, tag);
326 	if (err) {
327 		if (err->code == GOT_ERR_OBJ_EXISTS ||
328 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
329 			err = NULL;
330 		return err;
331 	}
332 	tag->refcnt++;
333 #endif
334 	return NULL;
335 }
336 
337 struct got_tag_object *
338 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
339 {
340 	return (struct got_tag_object *)got_object_cache_get(
341 	    &repo->tagcache, id);
342 }
343 
344 static const struct got_error *
345 open_repo(struct got_repository *repo, const char *path)
346 {
347 	const struct got_error *err = NULL;
348 
349 	repo->gitdir_fd = -1;
350 
351 	/* bare git repository? */
352 	repo->path_git_dir = strdup(path);
353 	if (repo->path_git_dir == NULL)
354 		return got_error_from_errno("strdup");
355 	if (is_git_repo(repo)) {
356 		repo->path = strdup(repo->path_git_dir);
357 		if (repo->path == NULL) {
358 			err = got_error_from_errno("strdup");
359 			goto done;
360 		}
361 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
362 		if (repo->gitdir_fd == -1) {
363 			err = got_error_from_errno2("open",
364 			    repo->path_git_dir);
365 			goto done;
366 		}
367 		return NULL;
368 	}
369 
370 	/* git repository with working tree? */
371 	free(repo->path_git_dir);
372 	repo->path_git_dir = NULL;
373 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
374 		err = got_error_from_errno("asprintf");
375 		goto done;
376 	}
377 	if (is_git_repo(repo)) {
378 		repo->path = strdup(path);
379 		if (repo->path == NULL) {
380 			err = got_error_from_errno("strdup");
381 			goto done;
382 		}
383 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
384 		if (repo->gitdir_fd == -1) {
385 			err = got_error_from_errno2("open",
386 			    repo->path_git_dir);
387 			goto done;
388 		}
389 		return NULL;
390 	}
391 
392 	err = got_error(GOT_ERR_NOT_GIT_REPO);
393 done:
394 	if (err) {
395 		free(repo->path);
396 		repo->path = NULL;
397 		free(repo->path_git_dir);
398 		repo->path_git_dir = NULL;
399 		if (repo->gitdir_fd != -1)
400 			close(repo->gitdir_fd);
401 		repo->gitdir_fd = -1;
402 
403 	}
404 	return err;
405 }
406 
407 static const struct got_error *
408 parse_gitconfig_file(int *gitconfig_repository_format_version,
409     char **gitconfig_author_name, char **gitconfig_author_email,
410     struct got_remote_repo **remotes, int *nremotes,
411     char **gitconfig_owner, char ***extensions, int *nextensions,
412     const char *gitconfig_path)
413 {
414 	const struct got_error *err = NULL, *child_err = NULL;
415 	int fd = -1;
416 	int imsg_fds[2] = { -1, -1 };
417 	pid_t pid;
418 	struct imsgbuf *ibuf;
419 
420 	*gitconfig_repository_format_version = 0;
421 	if (extensions)
422 		*extensions = NULL;
423 	if (nextensions)
424 		*nextensions = 0;
425 	*gitconfig_author_name = NULL;
426 	*gitconfig_author_email = NULL;
427 	if (remotes)
428 		*remotes = NULL;
429 	if (nremotes)
430 		*nremotes = 0;
431 	if (gitconfig_owner)
432 		*gitconfig_owner = NULL;
433 
434 	fd = open(gitconfig_path, O_RDONLY);
435 	if (fd == -1) {
436 		if (errno == ENOENT)
437 			return NULL;
438 		return got_error_from_errno2("open", gitconfig_path);
439 	}
440 
441 	ibuf = calloc(1, sizeof(*ibuf));
442 	if (ibuf == NULL) {
443 		err = got_error_from_errno("calloc");
444 		goto done;
445 	}
446 
447 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
448 		err = got_error_from_errno("socketpair");
449 		goto done;
450 	}
451 
452 	pid = fork();
453 	if (pid == -1) {
454 		err = got_error_from_errno("fork");
455 		goto done;
456 	} else if (pid == 0) {
457 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
458 		    gitconfig_path);
459 		/* not reached */
460 	}
461 
462 	if (close(imsg_fds[1]) == -1) {
463 		err = got_error_from_errno("close");
464 		goto done;
465 	}
466 	imsg_fds[1] = -1;
467 	imsg_init(ibuf, imsg_fds[0]);
468 
469 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
470 	if (err)
471 		goto done;
472 	fd = -1;
473 
474 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
475 	if (err)
476 		goto done;
477 
478 	err = got_privsep_recv_gitconfig_int(
479 	    gitconfig_repository_format_version, ibuf);
480 	if (err)
481 		goto done;
482 
483 	if (extensions && nextensions) {
484 		err = got_privsep_send_gitconfig_repository_extensions_req(
485 		    ibuf);
486 		if (err)
487 			goto done;
488 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
489 		if (err)
490 			goto done;
491 		if (*nextensions > 0) {
492 			int i;
493 			*extensions = calloc(*nextensions, sizeof(char *));
494 			if (*extensions == NULL) {
495 				err = got_error_from_errno("calloc");
496 				goto done;
497 			}
498 			for (i = 0; i < *nextensions; i++) {
499 				char *ext;
500 				err = got_privsep_recv_gitconfig_str(&ext,
501 				    ibuf);
502 				if (err)
503 					goto done;
504 				(*extensions)[i] = ext;
505 			}
506 		}
507 	}
508 
509 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
510 	if (err)
511 		goto done;
512 
513 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
514 	if (err)
515 		goto done;
516 
517 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
518 	if (err)
519 		goto done;
520 
521 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
522 	if (err)
523 		goto done;
524 
525 	if (remotes && nremotes) {
526 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
527 		if (err)
528 			goto done;
529 
530 		err = got_privsep_recv_gitconfig_remotes(remotes,
531 		    nremotes, ibuf);
532 		if (err)
533 			goto done;
534 	}
535 
536 	if (gitconfig_owner) {
537 		err = got_privsep_send_gitconfig_owner_req(ibuf);
538 		if (err)
539 			goto done;
540 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
541 		if (err)
542 			goto done;
543 	}
544 
545 	imsg_clear(ibuf);
546 	err = got_privsep_send_stop(imsg_fds[0]);
547 	child_err = got_privsep_wait_for_child(pid);
548 	if (child_err && err == NULL)
549 		err = child_err;
550 done:
551 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
552 		err = got_error_from_errno("close");
553 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
554 		err = got_error_from_errno("close");
555 	if (fd != -1 && close(fd) == -1 && err == NULL)
556 		err = got_error_from_errno2("close", gitconfig_path);
557 	free(ibuf);
558 	return err;
559 }
560 
561 static const struct got_error *
562 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
563 {
564 	const struct got_error *err = NULL;
565 	char *repo_gitconfig_path = NULL;
566 
567 	if (global_gitconfig_path) {
568 		/* Read settings from ~/.gitconfig. */
569 		int dummy_repo_version;
570 		err = parse_gitconfig_file(&dummy_repo_version,
571 		    &repo->global_gitconfig_author_name,
572 		    &repo->global_gitconfig_author_email,
573 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
574 		if (err)
575 			return err;
576 	}
577 
578 	/* Read repository's .git/config file. */
579 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
580 	if (repo_gitconfig_path == NULL)
581 		return got_error_from_errno("got_repo_get_path_gitconfig");
582 
583 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
584 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
585 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
586 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
587 	    repo_gitconfig_path);
588 	if (err)
589 		goto done;
590 done:
591 	free(repo_gitconfig_path);
592 	return err;
593 }
594 
595 static const struct got_error *
596 read_gotconfig(struct got_repository *repo)
597 {
598 	const struct got_error *err = NULL;
599 	char *gotconfig_path;
600 
601 	gotconfig_path = got_repo_get_path_gotconfig(repo);
602 	if (gotconfig_path == NULL)
603 		return got_error_from_errno("got_repo_get_path_gotconfig");
604 
605 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
606 	free(gotconfig_path);
607 	return err;
608 }
609 
610 /* Supported repository format extensions. */
611 static const char *repo_extensions[] = {
612 	"noop",			/* Got supports repository format version 1. */
613 	"preciousObjects",	/* Supported by gotadmin cleanup. */
614 	"worktreeConfig",	/* Got does not care about Git work trees. */
615 };
616 
617 const struct got_error *
618 got_repo_open(struct got_repository **repop, const char *path,
619     const char *global_gitconfig_path)
620 {
621 	struct got_repository *repo = NULL;
622 	const struct got_error *err = NULL;
623 	char *repo_path = NULL;
624 	size_t i;
625 	struct rlimit rl;
626 
627 	*repop = NULL;
628 
629 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
630 		return got_error_from_errno("getrlimit");
631 
632 	repo = calloc(1, sizeof(*repo));
633 	if (repo == NULL) {
634 		err = got_error_from_errno("calloc");
635 		goto done;
636 	}
637 
638 	for (i = 0; i < nitems(repo->privsep_children); i++) {
639 		memset(&repo->privsep_children[i], 0,
640 		    sizeof(repo->privsep_children[0]));
641 		repo->privsep_children[i].imsg_fd = -1;
642 	}
643 
644 	err = got_object_cache_init(&repo->objcache,
645 	    GOT_OBJECT_CACHE_TYPE_OBJ);
646 	if (err)
647 		goto done;
648 	err = got_object_cache_init(&repo->treecache,
649 	    GOT_OBJECT_CACHE_TYPE_TREE);
650 	if (err)
651 		goto done;
652 	err = got_object_cache_init(&repo->commitcache,
653 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
654 	if (err)
655 		goto done;
656 	err = got_object_cache_init(&repo->tagcache,
657 	    GOT_OBJECT_CACHE_TYPE_TAG);
658 	if (err)
659 		goto done;
660 
661 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
662 	if (repo->pack_cache_size > rl.rlim_cur / 8)
663 		repo->pack_cache_size = rl.rlim_cur / 8;
664 
665 	repo_path = realpath(path, NULL);
666 	if (repo_path == NULL) {
667 		err = got_error_from_errno2("realpath", path);
668 		goto done;
669 	}
670 
671 	for (;;) {
672 		char *parent_path;
673 
674 		err = open_repo(repo, repo_path);
675 		if (err == NULL)
676 			break;
677 		if (err->code != GOT_ERR_NOT_GIT_REPO)
678 			goto done;
679 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
680 			err = got_error(GOT_ERR_NOT_GIT_REPO);
681 			goto done;
682 		}
683 		err = got_path_dirname(&parent_path, repo_path);
684 		if (err)
685 			goto done;
686 		free(repo_path);
687 		repo_path = parent_path;
688 	}
689 
690 	err = read_gotconfig(repo);
691 	if (err)
692 		goto done;
693 
694 	err = read_gitconfig(repo, global_gitconfig_path);
695 	if (err)
696 		goto done;
697 	if (repo->gitconfig_repository_format_version != 0)
698 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
699 	for (i = 0; i < repo->nextensions; i++) {
700 		char *ext = repo->extensions[i];
701 		int j, supported = 0;
702 		for (j = 0; j < nitems(repo_extensions); j++) {
703 			if (strcmp(ext, repo_extensions[j]) == 0) {
704 				supported = 1;
705 				break;
706 			}
707 		}
708 		if (!supported) {
709 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
710 			goto done;
711 		}
712 	}
713 done:
714 	if (err)
715 		got_repo_close(repo);
716 	else
717 		*repop = repo;
718 	free(repo_path);
719 	return err;
720 }
721 
722 const struct got_error *
723 got_repo_close(struct got_repository *repo)
724 {
725 	const struct got_error *err = NULL, *child_err;
726 	size_t i;
727 
728 	for (i = 0; i < repo->pack_cache_size; i++) {
729 		if (repo->packidx_cache[i] == NULL)
730 			break;
731 		got_packidx_close(repo->packidx_cache[i]);
732 	}
733 
734 	for (i = 0; i < repo->pack_cache_size; i++) {
735 		if (repo->packs[i].path_packfile == NULL)
736 			break;
737 		got_pack_close(&repo->packs[i]);
738 	}
739 
740 	free(repo->path);
741 	free(repo->path_git_dir);
742 
743 	got_object_cache_close(&repo->objcache);
744 	got_object_cache_close(&repo->treecache);
745 	got_object_cache_close(&repo->commitcache);
746 	got_object_cache_close(&repo->tagcache);
747 
748 	for (i = 0; i < nitems(repo->privsep_children); i++) {
749 		if (repo->privsep_children[i].imsg_fd == -1)
750 			continue;
751 		imsg_clear(repo->privsep_children[i].ibuf);
752 		free(repo->privsep_children[i].ibuf);
753 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
754 		child_err = got_privsep_wait_for_child(
755 		    repo->privsep_children[i].pid);
756 		if (child_err && err == NULL)
757 			err = child_err;
758 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
759 		    err == NULL)
760 			err = got_error_from_errno("close");
761 	}
762 
763 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
764 	    err == NULL)
765 		err = got_error_from_errno("close");
766 
767 	if (repo->gotconfig)
768 		got_gotconfig_free(repo->gotconfig);
769 	free(repo->gitconfig_author_name);
770 	free(repo->gitconfig_author_email);
771 	for (i = 0; i < repo->ngitconfig_remotes; i++)
772 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
773 	free(repo->gitconfig_remotes);
774 	for (i = 0; i < repo->nextensions; i++)
775 		free(repo->extensions[i]);
776 	free(repo->extensions);
777 	free(repo);
778 
779 	return err;
780 }
781 
782 void
783 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
784 {
785 	int i;
786 
787 	free(repo->name);
788 	repo->name = NULL;
789 	free(repo->url);
790 	repo->url = NULL;
791 	for (i = 0; i < repo->nbranches; i++)
792 		free(repo->branches[i]);
793 	free(repo->branches);
794 	repo->branches = NULL;
795 	repo->nbranches = 0;
796 }
797 
798 const struct got_error *
799 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
800     const char *input_path)
801 {
802 	const struct got_error *err = NULL;
803 	const char *repo_abspath = NULL;
804 	size_t repolen, len;
805 	char *canonpath, *path = NULL;
806 
807 	*in_repo_path = NULL;
808 
809 	canonpath = strdup(input_path);
810 	if (canonpath == NULL) {
811 		err = got_error_from_errno("strdup");
812 		goto done;
813 	}
814 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
815 	if (err)
816 		goto done;
817 
818 	repo_abspath = got_repo_get_path(repo);
819 
820 	if (canonpath[0] == '\0') {
821 		path = strdup(canonpath);
822 		if (path == NULL) {
823 			err = got_error_from_errno("strdup");
824 			goto done;
825 		}
826 	} else {
827 		path = realpath(canonpath, NULL);
828 		if (path == NULL) {
829 			if (errno != ENOENT) {
830 				err = got_error_from_errno2("realpath",
831 				    canonpath);
832 				goto done;
833 			}
834 			/*
835 			 * Path is not on disk.
836 			 * Assume it is already relative to repository root.
837 			 */
838 			path = strdup(canonpath);
839 			if (path == NULL) {
840 				err = got_error_from_errno("strdup");
841 				goto done;
842 			}
843 		}
844 
845 		repolen = strlen(repo_abspath);
846 		len = strlen(path);
847 
848 
849 		if (strcmp(path, repo_abspath) == 0) {
850 			free(path);
851 			path = strdup("");
852 			if (path == NULL) {
853 				err = got_error_from_errno("strdup");
854 				goto done;
855 			}
856 		} else if (len > repolen &&
857 		    got_path_is_child(path, repo_abspath, repolen)) {
858 			/* Matched an on-disk path inside repository. */
859 			if (got_repo_is_bare(repo)) {
860 				/*
861 				 * Matched an on-disk path inside repository
862 				 * database. Treat input as repository-relative.
863 				 */
864 				 free(path);
865 				 path = canonpath;
866 				 canonpath = NULL;
867 			} else {
868 				char *child;
869 				/* Strip common prefix with repository path. */
870 				err = got_path_skip_common_ancestor(&child,
871 				    repo_abspath, path);
872 				if (err)
873 					goto done;
874 				free(path);
875 				path = child;
876 			}
877 		} else {
878 			/*
879 			 * Matched unrelated on-disk path.
880 			 * Treat input as repository-relative.
881 			 */
882 			 free(path);
883 			 path = canonpath;
884 			 canonpath = NULL;
885 		}
886 	}
887 
888 	/* Make in-repository path absolute */
889 	if (path[0] != '/') {
890 		char *abspath;
891 		if (asprintf(&abspath, "/%s", path) == -1) {
892 			err = got_error_from_errno("asprintf");
893 			goto done;
894 		}
895 		free(path);
896 		path = abspath;
897 	}
898 
899 done:
900 	free(canonpath);
901 	if (err)
902 		free(path);
903 	else
904 		*in_repo_path = path;
905 	return err;
906 }
907 
908 static const struct got_error *
909 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
910     const char *path_packidx)
911 {
912 	const struct got_error *err = NULL;
913 	size_t i;
914 
915 	for (i = 0; i < repo->pack_cache_size; i++) {
916 		if (repo->packidx_cache[i] == NULL)
917 			break;
918 		if (strcmp(repo->packidx_cache[i]->path_packidx,
919 		    path_packidx) == 0) {
920 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
921 		}
922 	}
923 	if (i == repo->pack_cache_size) {
924 		err = got_packidx_close(repo->packidx_cache[i - 1]);
925 		if (err)
926 			return err;
927 	}
928 
929 	/*
930 	 * Insert the new pack index at the front so it will
931 	 * be searched first in the future.
932 	 */
933 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
934 	    sizeof(repo->packidx_cache) -
935 	    sizeof(repo->packidx_cache[0]));
936 	repo->packidx_cache[0] = packidx;
937 
938 	return NULL;
939 }
940 
941 static int
942 is_packidx_filename(const char *name, size_t len)
943 {
944 	if (len != GOT_PACKIDX_NAMELEN)
945 		return 0;
946 
947 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
948 		return 0;
949 
950 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
951 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
952 		return 0;
953 
954 	return 1;
955 }
956 
957 const struct got_error *
958 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
959     struct got_repository *repo, struct got_object_id *id)
960 {
961 	const struct got_error *err;
962 	DIR *packdir = NULL;
963 	struct dirent *dent;
964 	char *path_packidx;
965 	size_t i;
966 	int packdir_fd;
967 
968 	/* Search pack index cache. */
969 	for (i = 0; i < repo->pack_cache_size; i++) {
970 		if (repo->packidx_cache[i] == NULL)
971 			break;
972 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
973 		if (*idx != -1) {
974 			*packidx = repo->packidx_cache[i];
975 			/*
976 			 * Move this cache entry to the front. Repeatedly
977 			 * searching a wrong pack index can be expensive.
978 			 */
979 			if (i > 0) {
980 				struct got_packidx *p;
981 				p = repo->packidx_cache[0];
982 				repo->packidx_cache[0] = *packidx;
983 				repo->packidx_cache[i] = p;
984 			}
985 			return NULL;
986 		}
987 	}
988 	/* No luck. Search the filesystem. */
989 
990 	packdir_fd = openat(got_repo_get_fd(repo),
991 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
992 	if (packdir_fd == -1) {
993 		if (errno == ENOENT)
994 			err = got_error_no_obj(id);
995 		else
996 			err = got_error_from_errno_fmt("openat: %s/%s",
997 			    got_repo_get_path_git_dir(repo),
998 			    GOT_OBJECTS_PACK_DIR);
999 		goto done;
1000 	}
1001 
1002 	packdir = fdopendir(packdir_fd);
1003 	if (packdir == NULL) {
1004 		err = got_error_from_errno("fdopendir");
1005 		goto done;
1006 	}
1007 
1008 	while ((dent = readdir(packdir)) != NULL) {
1009 		int is_cached = 0;
1010 
1011 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1012 			continue;
1013 
1014 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1015 		    dent->d_name) == -1) {
1016 			err = got_error_from_errno("asprintf");
1017 			goto done;
1018 		}
1019 
1020 		for (i = 0; i < repo->pack_cache_size; i++) {
1021 			if (repo->packidx_cache[i] == NULL)
1022 				break;
1023 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1024 			    path_packidx) == 0) {
1025 				is_cached = 1;
1026 				break;
1027 			}
1028 		}
1029 		if (is_cached) {
1030 			free(path_packidx);
1031 			continue; /* already searched */
1032 		}
1033 
1034 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1035 		    path_packidx, 0);
1036 		if (err) {
1037 			free(path_packidx);
1038 			goto done;
1039 		}
1040 
1041 		err = cache_packidx(repo, *packidx, path_packidx);
1042 		free(path_packidx);
1043 		if (err)
1044 			goto done;
1045 
1046 		*idx = got_packidx_get_object_idx(*packidx, id);
1047 		if (*idx != -1) {
1048 			err = NULL; /* found the object */
1049 			goto done;
1050 		}
1051 	}
1052 
1053 	err = got_error_no_obj(id);
1054 done:
1055 	if (packdir && closedir(packdir) != 0 && err == NULL)
1056 		err = got_error_from_errno("closedir");
1057 	return err;
1058 }
1059 
1060 static const struct got_error *
1061 read_packfile_hdr(int fd, struct got_packidx *packidx)
1062 {
1063 	const struct got_error *err = NULL;
1064 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1065 	struct got_packfile_hdr hdr;
1066 	ssize_t n;
1067 
1068 	n = read(fd, &hdr, sizeof(hdr));
1069 	if (n < 0)
1070 		return got_error_from_errno("read");
1071 	if (n != sizeof(hdr))
1072 		return got_error(GOT_ERR_BAD_PACKFILE);
1073 
1074 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1075 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1076 	    be32toh(hdr.nobjects) != totobj)
1077 		err = got_error(GOT_ERR_BAD_PACKFILE);
1078 
1079 	return err;
1080 }
1081 
1082 static const struct got_error *
1083 open_packfile(int *fd, struct got_repository *repo,
1084     const char *relpath, struct got_packidx *packidx)
1085 {
1086 	const struct got_error *err = NULL;
1087 
1088 	*fd = openat(got_repo_get_fd(repo), relpath, O_RDONLY | O_NOFOLLOW);
1089 	if (*fd == -1)
1090 		return got_error_from_errno_fmt("openat: %s/%s",
1091 		    got_repo_get_path_git_dir(repo), relpath);
1092 
1093 	if (packidx) {
1094 		err = read_packfile_hdr(*fd, packidx);
1095 		if (err) {
1096 			close(*fd);
1097 			*fd = -1;
1098 		}
1099 	}
1100 
1101 	return err;
1102 }
1103 
1104 const struct got_error *
1105 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1106     const char *path_packfile, struct got_packidx *packidx)
1107 {
1108 	const struct got_error *err = NULL;
1109 	struct got_pack *pack = NULL;
1110 	struct stat sb;
1111 	size_t i;
1112 
1113 	if (packp)
1114 		*packp = NULL;
1115 
1116 	for (i = 0; i < repo->pack_cache_size; i++) {
1117 		pack = &repo->packs[i];
1118 		if (pack->path_packfile == NULL)
1119 			break;
1120 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1121 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1122 	}
1123 
1124 	if (i == repo->pack_cache_size) {
1125 		err = got_pack_close(&repo->packs[i - 1]);
1126 		if (err)
1127 			return err;
1128 		memmove(&repo->packs[1], &repo->packs[0],
1129 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
1130 		i = 0;
1131 	}
1132 
1133 	pack = &repo->packs[i];
1134 
1135 	pack->path_packfile = strdup(path_packfile);
1136 	if (pack->path_packfile == NULL) {
1137 		err = got_error_from_errno("strdup");
1138 		goto done;
1139 	}
1140 
1141 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1142 	if (err)
1143 		goto done;
1144 
1145 	if (fstat(pack->fd, &sb) != 0) {
1146 		err = got_error_from_errno("fstat");
1147 		goto done;
1148 	}
1149 	pack->filesize = sb.st_size;
1150 
1151 	pack->privsep_child = NULL;
1152 
1153 #ifndef GOT_PACK_NO_MMAP
1154 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1155 	    pack->fd, 0);
1156 	if (pack->map == MAP_FAILED) {
1157 		if (errno != ENOMEM) {
1158 			err = got_error_from_errno("mmap");
1159 			goto done;
1160 		}
1161 		pack->map = NULL; /* fall back to read(2) */
1162 	}
1163 #endif
1164 done:
1165 	if (err) {
1166 		if (pack) {
1167 			free(pack->path_packfile);
1168 			memset(pack, 0, sizeof(*pack));
1169 		}
1170 	} else if (packp)
1171 		*packp = pack;
1172 	return err;
1173 }
1174 
1175 struct got_pack *
1176 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1177 {
1178 	struct got_pack *pack = NULL;
1179 	size_t i;
1180 
1181 	for (i = 0; i < repo->pack_cache_size; i++) {
1182 		pack = &repo->packs[i];
1183 		if (pack->path_packfile == NULL)
1184 			break;
1185 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1186 			return pack;
1187 	}
1188 
1189 	return NULL;
1190 }
1191 
1192 const struct got_error *
1193 got_repo_init(const char *repo_path)
1194 {
1195 	const struct got_error *err = NULL;
1196 	const char *dirnames[] = {
1197 		GOT_OBJECTS_DIR,
1198 		GOT_OBJECTS_PACK_DIR,
1199 		GOT_REFS_DIR,
1200 	};
1201 	const char *description_str = "Unnamed repository; "
1202 	    "edit this file 'description' to name the repository.";
1203 	const char *headref_str = "ref: refs/heads/main";
1204 	const char *gitconfig_str = "[core]\n"
1205 	    "\trepositoryformatversion = 0\n"
1206 	    "\tfilemode = true\n"
1207 	    "\tbare = true\n";
1208 	char *path;
1209 	size_t i;
1210 
1211 	if (!got_path_dir_is_empty(repo_path))
1212 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1213 
1214 	for (i = 0; i < nitems(dirnames); i++) {
1215 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1216 			return got_error_from_errno("asprintf");
1217 		}
1218 		err = got_path_mkdir(path);
1219 		free(path);
1220 		if (err)
1221 			return err;
1222 	}
1223 
1224 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1225 		return got_error_from_errno("asprintf");
1226 	err = got_path_create_file(path, description_str);
1227 	free(path);
1228 	if (err)
1229 		return err;
1230 
1231 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1232 		return got_error_from_errno("asprintf");
1233 	err = got_path_create_file(path, headref_str);
1234 	free(path);
1235 	if (err)
1236 		return err;
1237 
1238 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1239 		return got_error_from_errno("asprintf");
1240 	err = got_path_create_file(path, gitconfig_str);
1241 	free(path);
1242 	if (err)
1243 		return err;
1244 
1245 	return NULL;
1246 }
1247 
1248 static const struct got_error *
1249 match_packed_object(struct got_object_id **unique_id,
1250     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1251 {
1252 	const struct got_error *err = NULL;
1253 	DIR *packdir = NULL;
1254 	struct dirent *dent;
1255 	char *path_packidx;
1256 	struct got_object_id_queue matched_ids;
1257 	int packdir_fd;
1258 
1259 	STAILQ_INIT(&matched_ids);
1260 
1261 	packdir_fd = openat(got_repo_get_fd(repo),
1262 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1263 	if (packdir_fd == -1) {
1264 		if (errno != ENOENT)
1265 			err = got_error_from_errno2("openat", GOT_OBJECTS_PACK_DIR);
1266 		goto done;
1267 	}
1268 
1269 	packdir = fdopendir(packdir_fd);
1270 	if (packdir == NULL) {
1271 		err = got_error_from_errno("fdopendir");
1272 		goto done;
1273 	}
1274 
1275 	while ((dent = readdir(packdir)) != NULL) {
1276 		struct got_packidx *packidx;
1277 		struct got_object_qid *qid;
1278 
1279 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1280 			continue;
1281 
1282 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1283 		    dent->d_name) == -1) {
1284 			err = got_error_from_errno("strdup");
1285 			break;
1286 		}
1287 
1288 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1289 		    path_packidx, 0);
1290 		free(path_packidx);
1291 		if (err)
1292 			break;
1293 
1294 		err = got_packidx_match_id_str_prefix(&matched_ids,
1295 		    packidx, id_str_prefix);
1296 		if (err) {
1297 			got_packidx_close(packidx);
1298 			break;
1299 		}
1300 		err = got_packidx_close(packidx);
1301 		if (err)
1302 			break;
1303 
1304 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1305 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1306 				int matched_type;
1307 				err = got_object_get_type(&matched_type, repo,
1308 				    qid->id);
1309 				if (err)
1310 					goto done;
1311 				if (matched_type != obj_type)
1312 					continue;
1313 			}
1314 			if (*unique_id == NULL) {
1315 				*unique_id = got_object_id_dup(qid->id);
1316 				if (*unique_id == NULL) {
1317 					err = got_error_from_errno("malloc");
1318 					goto done;
1319 				}
1320 			} else {
1321 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1322 					continue; /* packed multiple times */
1323 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1324 				goto done;
1325 			}
1326 		}
1327 	}
1328 done:
1329 	got_object_id_queue_free(&matched_ids);
1330 	if (packdir && closedir(packdir) != 0 && err == NULL)
1331 		err = got_error_from_errno("closedir");
1332 	if (err) {
1333 		free(*unique_id);
1334 		*unique_id = NULL;
1335 	}
1336 	return err;
1337 }
1338 
1339 static const struct got_error *
1340 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1341     const char *object_dir, const char *id_str_prefix, int obj_type,
1342     struct got_repository *repo)
1343 {
1344 	const struct got_error *err = NULL;
1345 	char *path;
1346 	DIR *dir = NULL;
1347 	struct dirent *dent;
1348 	struct got_object_id id;
1349 
1350 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1351 		err = got_error_from_errno("asprintf");
1352 		goto done;
1353 	}
1354 
1355 	dir = opendir(path);
1356 	if (dir == NULL) {
1357 		if (errno == ENOENT) {
1358 			err = NULL;
1359 			goto done;
1360 		}
1361 		err = got_error_from_errno2("opendir", path);
1362 		goto done;
1363 	}
1364 	while ((dent = readdir(dir)) != NULL) {
1365 		char *id_str;
1366 		int cmp;
1367 
1368 		if (strcmp(dent->d_name, ".") == 0 ||
1369 		    strcmp(dent->d_name, "..") == 0)
1370 			continue;
1371 
1372 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1373 			err = got_error_from_errno("asprintf");
1374 			goto done;
1375 		}
1376 
1377 		if (!got_parse_sha1_digest(id.sha1, id_str))
1378 			continue;
1379 
1380 		/*
1381 		 * Directory entries do not necessarily appear in
1382 		 * sorted order, so we must iterate over all of them.
1383 		 */
1384 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1385 		if (cmp != 0) {
1386 			free(id_str);
1387 			continue;
1388 		}
1389 
1390 		if (*unique_id == NULL) {
1391 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1392 				int matched_type;
1393 				err = got_object_get_type(&matched_type, repo,
1394 				    &id);
1395 				if (err)
1396 					goto done;
1397 				if (matched_type != obj_type)
1398 					continue;
1399 			}
1400 			*unique_id = got_object_id_dup(&id);
1401 			if (*unique_id == NULL) {
1402 				err = got_error_from_errno("got_object_id_dup");
1403 				free(id_str);
1404 				goto done;
1405 			}
1406 		} else {
1407 			if (got_object_id_cmp(*unique_id, &id) == 0)
1408 				continue; /* both packed and loose */
1409 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1410 			free(id_str);
1411 			goto done;
1412 		}
1413 	}
1414 done:
1415 	if (dir && closedir(dir) != 0 && err == NULL)
1416 		err = got_error_from_errno("closedir");
1417 	if (err) {
1418 		free(*unique_id);
1419 		*unique_id = NULL;
1420 	}
1421 	free(path);
1422 	return err;
1423 }
1424 
1425 const struct got_error *
1426 got_repo_match_object_id_prefix(struct got_object_id **id,
1427     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1428 {
1429 	const struct got_error *err = NULL;
1430 	char *path_objects = got_repo_get_path_objects(repo);
1431 	char *object_dir = NULL;
1432 	size_t len;
1433 	int i;
1434 
1435 	*id = NULL;
1436 
1437 	for (i = 0; i < strlen(id_str_prefix); i++) {
1438 		if (isxdigit((unsigned char)id_str_prefix[i]))
1439 			continue;
1440 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1441 	}
1442 
1443 	len = strlen(id_str_prefix);
1444 	if (len >= 2) {
1445 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1446 		if (err)
1447 			goto done;
1448 		object_dir = strndup(id_str_prefix, 2);
1449 		if (object_dir == NULL) {
1450 			err = got_error_from_errno("strdup");
1451 			goto done;
1452 		}
1453 		err = match_loose_object(id, path_objects, object_dir,
1454 		    id_str_prefix, obj_type, repo);
1455 	} else if (len == 1) {
1456 		int i;
1457 		for (i = 0; i < 0xf; i++) {
1458 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1459 			    == -1) {
1460 				err = got_error_from_errno("asprintf");
1461 				goto done;
1462 			}
1463 			err = match_packed_object(id, repo, object_dir,
1464 			    obj_type);
1465 			if (err)
1466 				goto done;
1467 			err = match_loose_object(id, path_objects, object_dir,
1468 			    id_str_prefix, obj_type, repo);
1469 			if (err)
1470 				goto done;
1471 		}
1472 	} else {
1473 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1474 		goto done;
1475 	}
1476 done:
1477 	free(object_dir);
1478 	if (err) {
1479 		free(*id);
1480 		*id = NULL;
1481 	} else if (*id == NULL)
1482 		err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1483 
1484 	return err;
1485 }
1486 
1487 const struct got_error *
1488 got_repo_match_object_id(struct got_object_id **id, char **label,
1489     const char *id_str, int obj_type, struct got_reflist_head *refs,
1490     struct got_repository *repo)
1491 {
1492 	const struct got_error *err;
1493 	struct got_tag_object *tag;
1494 	struct got_reference *ref = NULL;
1495 
1496 	*id = NULL;
1497 	if (label)
1498 		*label = NULL;
1499 
1500 	if (refs) {
1501 		err = got_repo_object_match_tag(&tag, id_str, GOT_OBJ_TYPE_ANY,
1502 		    refs, repo);
1503 		if (err == NULL) {
1504 			*id = got_object_id_dup(
1505 			    got_object_tag_get_object_id(tag));
1506 			if (*id == NULL)
1507 				err = got_error_from_errno("got_object_id_dup");
1508 			else if (label && asprintf(label, "refs/tags/%s",
1509 			    got_object_tag_get_name(tag)) == -1) {
1510 				err = got_error_from_errno("asprintf");
1511 				free(*id);
1512 				*id = NULL;
1513 			}
1514 			got_object_tag_close(tag);
1515 			return err;
1516 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1517 		    err->code != GOT_ERR_NO_OBJ)
1518 			return err;
1519 	}
1520 
1521 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1522 	if (err) {
1523 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1524 			return err;
1525 		err = got_ref_open(&ref, repo, id_str, 0);
1526 		if (err != NULL)
1527 			goto done;
1528 		if (label) {
1529 			*label = strdup(got_ref_get_name(ref));
1530 			if (*label == NULL) {
1531 				err = got_error_from_errno("strdup");
1532 				goto done;
1533 			}
1534 		}
1535 		err = got_ref_resolve(id, repo, ref);
1536 	} else if (label) {
1537 		err = got_object_id_str(label, *id);
1538 		if (*label == NULL) {
1539 			err = got_error_from_errno("strdup");
1540 			goto done;
1541 		}
1542 	}
1543 done:
1544 	if (ref)
1545 		got_ref_close(ref);
1546 	return err;
1547 }
1548 
1549 const struct got_error *
1550 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1551     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1552 {
1553 	const struct got_error *err = NULL;
1554 	struct got_reflist_entry *re;
1555 	struct got_object_id *tag_id;
1556 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1557 
1558 	*tag = NULL;
1559 
1560 	TAILQ_FOREACH(re, refs, entry) {
1561 		const char *refname;
1562 		refname = got_ref_get_name(re->ref);
1563 		if (got_ref_is_symbolic(re->ref))
1564 			continue;
1565 		if (strncmp(refname, "refs/tags/", 10) != 0)
1566 			continue;
1567 		if (!name_is_absolute)
1568 			refname += strlen("refs/tags/");
1569 		if (strcmp(refname, name) != 0)
1570 			continue;
1571 		err = got_ref_resolve(&tag_id, repo, re->ref);
1572 		if (err)
1573 			break;
1574 		err = got_object_open_as_tag(tag, repo, tag_id);
1575 		free(tag_id);
1576 		if (err)
1577 			break;
1578 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1579 		    got_object_tag_get_object_type(*tag) == obj_type)
1580 			break;
1581 		got_object_tag_close(*tag);
1582 		*tag = NULL;
1583 	}
1584 
1585 	if (err == NULL && *tag == NULL)
1586 		err = got_error_path(name, GOT_ERR_NO_OBJ);
1587 	return err;
1588 }
1589 
1590 static const struct got_error *
1591 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1592     const char *name, mode_t mode, struct got_object_id *blob_id)
1593 {
1594 	const struct got_error *err = NULL;
1595 
1596 	 *new_te = NULL;
1597 
1598 	*new_te = calloc(1, sizeof(**new_te));
1599 	if (*new_te == NULL)
1600 		return got_error_from_errno("calloc");
1601 
1602 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1603 	    sizeof((*new_te)->name)) {
1604 		err = got_error(GOT_ERR_NO_SPACE);
1605 		goto done;
1606 	}
1607 
1608 	if (S_ISLNK(mode)) {
1609 		(*new_te)->mode = S_IFLNK;
1610 	} else {
1611 		(*new_te)->mode = S_IFREG;
1612 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1613 	}
1614 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1615 done:
1616 	if (err && *new_te) {
1617 		free(*new_te);
1618 		*new_te = NULL;
1619 	}
1620 	return err;
1621 }
1622 
1623 static const struct got_error *
1624 import_file(struct got_tree_entry **new_te, struct dirent *de,
1625     const char *path, struct got_repository *repo)
1626 {
1627 	const struct got_error *err;
1628 	struct got_object_id *blob_id = NULL;
1629 	char *filepath;
1630 	struct stat sb;
1631 
1632 	if (asprintf(&filepath, "%s%s%s", path,
1633 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1634 		return got_error_from_errno("asprintf");
1635 
1636 	if (lstat(filepath, &sb) != 0) {
1637 		err = got_error_from_errno2("lstat", path);
1638 		goto done;
1639 	}
1640 
1641 	err = got_object_blob_create(&blob_id, filepath, repo);
1642 	if (err)
1643 		goto done;
1644 
1645 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1646 	    blob_id);
1647 done:
1648 	free(filepath);
1649 	if (err)
1650 		free(blob_id);
1651 	return err;
1652 }
1653 
1654 static const struct got_error *
1655 insert_tree_entry(struct got_tree_entry *new_te,
1656     struct got_pathlist_head *paths)
1657 {
1658 	const struct got_error *err = NULL;
1659 	struct got_pathlist_entry *new_pe;
1660 
1661 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1662 	if (err)
1663 		return err;
1664 	if (new_pe == NULL)
1665 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1666 	return NULL;
1667 }
1668 
1669 static const struct got_error *write_tree(struct got_object_id **,
1670     const char *, struct got_pathlist_head *, struct got_repository *,
1671     got_repo_import_cb progress_cb, void *progress_arg);
1672 
1673 static const struct got_error *
1674 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1675     const char *path, struct got_pathlist_head *ignores,
1676     struct got_repository *repo,
1677     got_repo_import_cb progress_cb, void *progress_arg)
1678 {
1679 	const struct got_error *err;
1680 	struct got_object_id *id = NULL;
1681 	char *subdirpath;
1682 
1683 	if (asprintf(&subdirpath, "%s%s%s", path,
1684 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1685 		return got_error_from_errno("asprintf");
1686 
1687 	(*new_te) = calloc(1, sizeof(**new_te));
1688 	if (*new_te == NULL)
1689 		return got_error_from_errno("calloc");
1690 	(*new_te)->mode = S_IFDIR;
1691 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1692 	    sizeof((*new_te)->name)) {
1693 		err = got_error(GOT_ERR_NO_SPACE);
1694 		goto done;
1695 	}
1696 	err = write_tree(&id, subdirpath, ignores,  repo,
1697 	    progress_cb, progress_arg);
1698 	if (err)
1699 		goto done;
1700 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1701 	
1702 done:
1703 	free(id);
1704 	free(subdirpath);
1705 	if (err) {
1706 		free(*new_te);
1707 		*new_te = NULL;
1708 	}
1709 	return err;
1710 }
1711 
1712 static const struct got_error *
1713 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1714     struct got_pathlist_head *ignores, struct got_repository *repo,
1715     got_repo_import_cb progress_cb, void *progress_arg)
1716 {
1717 	const struct got_error *err = NULL;
1718 	DIR *dir;
1719 	struct dirent *de;
1720 	int nentries;
1721 	struct got_tree_entry *new_te = NULL;
1722 	struct got_pathlist_head paths;
1723 	struct got_pathlist_entry *pe;
1724 
1725 	*new_tree_id = NULL;
1726 
1727 	TAILQ_INIT(&paths);
1728 
1729 	dir = opendir(path_dir);
1730 	if (dir == NULL) {
1731 		err = got_error_from_errno2("opendir", path_dir);
1732 		goto done;
1733 	}
1734 
1735 	nentries = 0;
1736 	while ((de = readdir(dir)) != NULL) {
1737 		int ignore = 0;
1738 		int type;
1739 
1740 		if (strcmp(de->d_name, ".") == 0 ||
1741 		    strcmp(de->d_name, "..") == 0)
1742 			continue;
1743 
1744 		TAILQ_FOREACH(pe, ignores, entry) {
1745 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1746 				ignore = 1;
1747 				break;
1748 			}
1749 		}
1750 		if (ignore)
1751 			continue;
1752 
1753 		err = got_path_dirent_type(&type, path_dir, de);
1754 		if (err)
1755 			goto done;
1756 
1757 		if (type == DT_DIR) {
1758 			err = import_subdir(&new_te, de, path_dir,
1759 			    ignores, repo, progress_cb, progress_arg);
1760 			if (err) {
1761 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1762 					goto done;
1763 				err = NULL;
1764 				continue;
1765 			}
1766 		} else if (type == DT_REG || type == DT_LNK) {
1767 			err = import_file(&new_te, de, path_dir, repo);
1768 			if (err)
1769 				goto done;
1770 		} else
1771 			continue;
1772 
1773 		err = insert_tree_entry(new_te, &paths);
1774 		if (err)
1775 			goto done;
1776 		nentries++;
1777 	}
1778 
1779 	if (TAILQ_EMPTY(&paths)) {
1780 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1781 		    "cannot create tree without any entries");
1782 		goto done;
1783 	}
1784 
1785 	TAILQ_FOREACH(pe, &paths, entry) {
1786 		struct got_tree_entry *te = pe->data;
1787 		char *path;
1788 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1789 			continue;
1790 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1791 			err = got_error_from_errno("asprintf");
1792 			goto done;
1793 		}
1794 		err = (*progress_cb)(progress_arg, path);
1795 		free(path);
1796 		if (err)
1797 			goto done;
1798 	}
1799 
1800 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1801 done:
1802 	if (dir)
1803 		closedir(dir);
1804 	got_pathlist_free(&paths);
1805 	return err;
1806 }
1807 
1808 const struct got_error *
1809 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1810     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1811     struct got_repository *repo, got_repo_import_cb progress_cb,
1812     void *progress_arg)
1813 {
1814 	const struct got_error *err;
1815 	struct got_object_id *new_tree_id;
1816 
1817 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1818 	    progress_cb, progress_arg);
1819 	if (err)
1820 		return err;
1821 
1822 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1823 	    author, time(NULL), author, time(NULL), logmsg, repo);
1824 	free(new_tree_id);
1825 	return err;
1826 }
1827 
1828 const struct got_error *
1829 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
1830     struct got_repository *repo)
1831 {
1832 	const struct got_error *err = NULL;
1833 	char *path_objects = NULL, *path = NULL;
1834 	DIR *dir = NULL;
1835 	struct got_object_id id;
1836 	int i;
1837 
1838 	*nobjects = 0;
1839 	*ondisk_size = 0;
1840 
1841 	path_objects = got_repo_get_path_objects(repo);
1842 	if (path_objects == NULL)
1843 		return got_error_from_errno("got_repo_get_path_objects");
1844 
1845 	for (i = 0; i <= 0xff; i++) {
1846 		struct dirent *dent;
1847 
1848 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
1849 			err = got_error_from_errno("asprintf");
1850 			break;
1851 		}
1852 
1853 		dir = opendir(path);
1854 		if (dir == NULL) {
1855 			if (errno == ENOENT) {
1856 				err = NULL;
1857 				continue;
1858 			}
1859 			err = got_error_from_errno2("opendir", path);
1860 			break;
1861 		}
1862 
1863 		while ((dent = readdir(dir)) != NULL) {
1864 			char *id_str;
1865 			int fd;
1866 			struct stat sb;
1867 
1868 			if (strcmp(dent->d_name, ".") == 0 ||
1869 			    strcmp(dent->d_name, "..") == 0)
1870 				continue;
1871 
1872 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
1873 				err = got_error_from_errno("asprintf");
1874 				goto done;
1875 			}
1876 
1877 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
1878 				free(id_str);
1879 				continue;
1880 			}
1881 			free(id_str);
1882 
1883 			err = got_object_open_loose_fd(&fd, &id, repo);
1884 			if (err)
1885 				goto done;
1886 
1887 			if (fstat(fd, &sb) == -1) {
1888 				err = got_error_from_errno("fstat");
1889 				close(fd);
1890 				goto done;
1891 			}
1892 			(*nobjects)++;
1893 			(*ondisk_size) += sb.st_size;
1894 
1895 			if (close(fd) == -1) {
1896 				err = got_error_from_errno("close");
1897 				goto done;
1898 			}
1899 		}
1900 
1901 		if (closedir(dir) != 0) {
1902 			err = got_error_from_errno("closedir");
1903 			goto done;
1904 		}
1905 		dir = NULL;
1906 
1907 		free(path);
1908 		path = NULL;
1909 	}
1910 done:
1911 	if (dir && closedir(dir) != 0 && err == NULL)
1912 		err = got_error_from_errno("closedir");
1913 
1914 	if (err) {
1915 		*nobjects = 0;
1916 		*ondisk_size = 0;
1917 	}
1918 	free(path_objects);
1919 	free(path);
1920 	return err;
1921 }
1922 
1923 const struct got_error *
1924 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
1925     off_t *total_packsize, struct got_repository *repo)
1926 {
1927 	const struct got_error *err = NULL;
1928 	DIR *packdir = NULL;
1929 	struct dirent *dent;
1930 	struct got_packidx *packidx = NULL;
1931 	char *path_packidx;
1932 	char *path_packfile;
1933 	int packdir_fd;
1934 	struct stat sb;
1935 
1936 	*npackfiles = 0;
1937 	*nobjects = 0;
1938 	*total_packsize = 0;
1939 
1940 	packdir_fd = openat(got_repo_get_fd(repo),
1941 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1942 	if (packdir_fd == -1) {
1943 		return got_error_from_errno_fmt("openat: %s/%s",
1944 		    got_repo_get_path_git_dir(repo),
1945 		    GOT_OBJECTS_PACK_DIR);
1946 	}
1947 
1948 	packdir = fdopendir(packdir_fd);
1949 	if (packdir == NULL) {
1950 		err = got_error_from_errno("fdopendir");
1951 		goto done;
1952 	}
1953 
1954 	while ((dent = readdir(packdir)) != NULL) {
1955 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
1956 			continue;
1957 
1958 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1959 		    dent->d_name) == -1) {
1960 			err = got_error_from_errno("asprintf");
1961 			goto done;
1962 		}
1963 
1964 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1965 		    path_packidx, 0);
1966 		free(path_packidx);
1967 		if (err)
1968 			goto done;
1969 
1970 		if (fstat(packidx->fd, &sb) == -1)
1971 			goto done;
1972 		*total_packsize += sb.st_size;
1973 
1974 		err = got_packidx_get_packfile_path(&path_packfile, packidx);
1975 		if (err)
1976 			goto done;
1977 
1978 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
1979 		    0) == -1) {
1980 			free(path_packfile);
1981 			goto done;
1982 		}
1983 		free(path_packfile);
1984 		*total_packsize += sb.st_size;
1985 
1986 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
1987 
1988 		(*npackfiles)++;
1989 
1990 		got_packidx_close(packidx);
1991 		packidx = NULL;
1992 	}
1993 done:
1994 	if (packidx)
1995 		got_packidx_close(packidx);
1996 	if (packdir && closedir(packdir) != 0 && err == NULL)
1997 		err = got_error_from_errno("closedir");
1998 	if (err) {
1999 		*npackfiles = 0;
2000 		*nobjects = 0;
2001 		*total_packsize = 0;
2002 	}
2003 	return err;
2004 }