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/uio.h>
19 #include <sys/socket.h>
20 #include <sys/stat.h>
21 #include <sys/mman.h>
22 #include <sys/resource.h>
23 
24 #include <ctype.h>
25 #include <fcntl.h>
26 #include <fnmatch.h>
27 #include <limits.h>
28 #include <dirent.h>
29 #include <stdlib.h>
30 #include <stdio.h>
31 #include <string.h>
32 #include <time.h>
33 #include <unistd.h>
34 #include <zlib.h>
35 #include <errno.h>
36 #include <libgen.h>
37 #include <stdint.h>
38 #include <uuid.h>
39 
40 #include "got_compat.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 
49 #include "got_lib_delta.h"
50 #include "got_lib_inflate.h"
51 #include "got_lib_object.h"
52 #include "got_lib_object_parse.h"
53 #include "got_lib_object_create.h"
54 #include "got_lib_pack.h"
55 #include "got_lib_privsep.h"
56 #include "got_lib_sha1.h"
57 #include "got_lib_object_cache.h"
58 #include "got_lib_repository.h"
59 #include "got_lib_gotconfig.h"
60 
61 #ifndef nitems
62 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
63 #endif
64 
65 const char *
66 got_repo_get_path(struct got_repository *repo)
67 {
68 	return repo->path;
69 }
70 
71 const char *
72 got_repo_get_path_git_dir(struct got_repository *repo)
73 {
74 	return repo->path_git_dir;
75 }
76 
77 int
78 got_repo_get_fd(struct got_repository *repo)
79 {
80 	return repo->gitdir_fd;
81 }
82 
83 const char *
84 got_repo_get_gitconfig_author_name(struct got_repository *repo)
85 {
86 	return repo->gitconfig_author_name;
87 }
88 
89 const char *
90 got_repo_get_gitconfig_author_email(struct got_repository *repo)
91 {
92 	return repo->gitconfig_author_email;
93 }
94 
95 const char *
96 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
97 {
98 	return repo->global_gitconfig_author_name;
99 }
100 
101 const char *
102 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
103 {
104 	return repo->global_gitconfig_author_email;
105 }
106 
107 const char *
108 got_repo_get_gitconfig_owner(struct got_repository *repo)
109 {
110 	return repo->gitconfig_owner;
111 }
112 
113 void
114 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
115     struct got_repository *repo)
116 {
117 	*extensions = repo->extensions;
118 	*nextensions = repo->nextensions;
119 }
120 
121 int
122 got_repo_is_bare(struct got_repository *repo)
123 {
124 	return (strcmp(repo->path, repo->path_git_dir) == 0);
125 }
126 
127 static char *
128 get_path_git_child(struct got_repository *repo, const char *basename)
129 {
130 	char *path_child;
131 
132 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
133 	    basename) == -1)
134 		return NULL;
135 
136 	return path_child;
137 }
138 
139 char *
140 got_repo_get_path_objects(struct got_repository *repo)
141 {
142 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
143 }
144 
145 char *
146 got_repo_get_path_objects_pack(struct got_repository *repo)
147 {
148 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
149 }
150 
151 char *
152 got_repo_get_path_refs(struct got_repository *repo)
153 {
154 	return get_path_git_child(repo, GOT_REFS_DIR);
155 }
156 
157 char *
158 got_repo_get_path_packed_refs(struct got_repository *repo)
159 {
160 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
161 }
162 
163 static char *
164 get_path_head(struct got_repository *repo)
165 {
166 	return get_path_git_child(repo, GOT_HEAD_FILE);
167 }
168 
169 char *
170 got_repo_get_path_gitconfig(struct got_repository *repo)
171 {
172 	return get_path_git_child(repo, GOT_GITCONFIG);
173 }
174 
175 char *
176 got_repo_get_path_gotconfig(struct got_repository *repo)
177 {
178 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
179 }
180 
181 const struct got_gotconfig *
182 got_repo_get_gotconfig(struct got_repository *repo)
183 {
184 	return repo->gotconfig;
185 }
186 
187 void
188 got_repo_get_gitconfig_remotes(int *nremotes,
189     const struct got_remote_repo **remotes, struct got_repository *repo)
190 {
191 	*nremotes = repo->ngitconfig_remotes;
192 	*remotes = repo->gitconfig_remotes;
193 }
194 
195 static int
196 is_git_repo(struct got_repository *repo)
197 {
198 	const char *path_git = got_repo_get_path_git_dir(repo);
199 	char *path_objects = got_repo_get_path_objects(repo);
200 	char *path_refs = got_repo_get_path_refs(repo);
201 	char *path_head = get_path_head(repo);
202 	int ret = 0;
203 	struct stat sb;
204 	struct got_reference *head_ref;
205 
206 	if (lstat(path_git, &sb) == -1)
207 		goto done;
208 	if (!S_ISDIR(sb.st_mode))
209 		goto done;
210 
211 	if (lstat(path_objects, &sb) == -1)
212 		goto done;
213 	if (!S_ISDIR(sb.st_mode))
214 		goto done;
215 
216 	if (lstat(path_refs, &sb) == -1)
217 		goto done;
218 	if (!S_ISDIR(sb.st_mode))
219 		goto done;
220 
221 	if (lstat(path_head, &sb) == -1)
222 		goto done;
223 	if (!S_ISREG(sb.st_mode))
224 		goto done;
225 
226 	/* Check if the HEAD reference can be opened. */
227 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
228 		goto done;
229 	got_ref_close(head_ref);
230 
231 	ret = 1;
232 done:
233 	free(path_objects);
234 	free(path_refs);
235 	free(path_head);
236 	return ret;
237 
238 }
239 
240 const struct got_error *
241 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
242     struct got_object *obj)
243 {
244 #ifndef GOT_NO_OBJ_CACHE
245 	const struct got_error *err = NULL;
246 	err = got_object_cache_add(&repo->objcache, id, obj);
247 	if (err) {
248 		if (err->code == GOT_ERR_OBJ_EXISTS ||
249 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
250 			err = NULL;
251 		return err;
252 	}
253 	obj->refcnt++;
254 #endif
255 	return NULL;
256 }
257 
258 struct got_object *
259 got_repo_get_cached_object(struct got_repository *repo,
260     struct got_object_id *id)
261 {
262 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
263 }
264 
265 const struct got_error *
266 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
267     struct got_tree_object *tree)
268 {
269 #ifndef GOT_NO_OBJ_CACHE
270 	const struct got_error *err = NULL;
271 	err = got_object_cache_add(&repo->treecache, id, tree);
272 	if (err) {
273 		if (err->code == GOT_ERR_OBJ_EXISTS ||
274 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
275 			err = NULL;
276 		return err;
277 	}
278 	tree->refcnt++;
279 #endif
280 	return NULL;
281 }
282 
283 struct got_tree_object *
284 got_repo_get_cached_tree(struct got_repository *repo,
285     struct got_object_id *id)
286 {
287 	return (struct got_tree_object *)got_object_cache_get(
288 	    &repo->treecache, id);
289 }
290 
291 const struct got_error *
292 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
293     struct got_commit_object *commit)
294 {
295 #ifndef GOT_NO_OBJ_CACHE
296 	const struct got_error *err = NULL;
297 	err = got_object_cache_add(&repo->commitcache, id, commit);
298 	if (err) {
299 		if (err->code == GOT_ERR_OBJ_EXISTS ||
300 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
301 			err = NULL;
302 		return err;
303 	}
304 	commit->refcnt++;
305 #endif
306 	return NULL;
307 }
308 
309 struct got_commit_object *
310 got_repo_get_cached_commit(struct got_repository *repo,
311     struct got_object_id *id)
312 {
313 	return (struct got_commit_object *)got_object_cache_get(
314 	    &repo->commitcache, id);
315 }
316 
317 const struct got_error *
318 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
319     struct got_tag_object *tag)
320 {
321 #ifndef GOT_NO_OBJ_CACHE
322 	const struct got_error *err = NULL;
323 	err = got_object_cache_add(&repo->tagcache, id, tag);
324 	if (err) {
325 		if (err->code == GOT_ERR_OBJ_EXISTS ||
326 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
327 			err = NULL;
328 		return err;
329 	}
330 	tag->refcnt++;
331 #endif
332 	return NULL;
333 }
334 
335 struct got_tag_object *
336 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
337 {
338 	return (struct got_tag_object *)got_object_cache_get(
339 	    &repo->tagcache, id);
340 }
341 
342 static const struct got_error *
343 open_repo(struct got_repository *repo, const char *path)
344 {
345 	const struct got_error *err = NULL;
346 
347 	repo->gitdir_fd = -1;
348 
349 	/* bare git repository? */
350 	repo->path_git_dir = strdup(path);
351 	if (repo->path_git_dir == NULL)
352 		return got_error_from_errno("strdup");
353 	if (is_git_repo(repo)) {
354 		repo->path = strdup(repo->path_git_dir);
355 		if (repo->path == NULL) {
356 			err = got_error_from_errno("strdup");
357 			goto done;
358 		}
359 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
360 		if (repo->gitdir_fd == -1) {
361 			err = got_error_from_errno2("open",
362 			    repo->path_git_dir);
363 			goto done;
364 		}
365 		return NULL;
366 	}
367 
368 	/* git repository with working tree? */
369 	free(repo->path_git_dir);
370 	repo->path_git_dir = NULL;
371 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
372 		err = got_error_from_errno("asprintf");
373 		goto done;
374 	}
375 	if (is_git_repo(repo)) {
376 		repo->path = strdup(path);
377 		if (repo->path == NULL) {
378 			err = got_error_from_errno("strdup");
379 			goto done;
380 		}
381 		repo->gitdir_fd = open(repo->path_git_dir, O_DIRECTORY);
382 		if (repo->gitdir_fd == -1) {
383 			err = got_error_from_errno2("open",
384 			    repo->path_git_dir);
385 			goto done;
386 		}
387 		return NULL;
388 	}
389 
390 	err = got_error(GOT_ERR_NOT_GIT_REPO);
391 done:
392 	if (err) {
393 		free(repo->path);
394 		repo->path = NULL;
395 		free(repo->path_git_dir);
396 		repo->path_git_dir = NULL;
397 		if (repo->gitdir_fd != -1)
398 			close(repo->gitdir_fd);
399 		repo->gitdir_fd = -1;
400 
401 	}
402 	return err;
403 }
404 
405 static const struct got_error *
406 parse_gitconfig_file(int *gitconfig_repository_format_version,
407     char **gitconfig_author_name, char **gitconfig_author_email,
408     struct got_remote_repo **remotes, int *nremotes,
409     char **gitconfig_owner, char ***extensions, int *nextensions,
410     const char *gitconfig_path)
411 {
412 	const struct got_error *err = NULL, *child_err = NULL;
413 	int fd = -1;
414 	int imsg_fds[2] = { -1, -1 };
415 	pid_t pid;
416 	struct imsgbuf *ibuf;
417 
418 	*gitconfig_repository_format_version = 0;
419 	if (extensions)
420 		*extensions = NULL;
421 	if (nextensions)
422 		*nextensions = 0;
423 	*gitconfig_author_name = NULL;
424 	*gitconfig_author_email = NULL;
425 	if (remotes)
426 		*remotes = NULL;
427 	if (nremotes)
428 		*nremotes = 0;
429 	if (gitconfig_owner)
430 		*gitconfig_owner = NULL;
431 
432 	fd = open(gitconfig_path, O_RDONLY);
433 	if (fd == -1) {
434 		if (errno == ENOENT)
435 			return NULL;
436 		return got_error_from_errno2("open", gitconfig_path);
437 	}
438 
439 	ibuf = calloc(1, sizeof(*ibuf));
440 	if (ibuf == NULL) {
441 		err = got_error_from_errno("calloc");
442 		goto done;
443 	}
444 
445 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
446 		err = got_error_from_errno("socketpair");
447 		goto done;
448 	}
449 
450 	pid = fork();
451 	if (pid == -1) {
452 		err = got_error_from_errno("fork");
453 		goto done;
454 	} else if (pid == 0) {
455 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
456 		    gitconfig_path);
457 		/* not reached */
458 	}
459 
460 	if (close(imsg_fds[1]) == -1) {
461 		err = got_error_from_errno("close");
462 		goto done;
463 	}
464 	imsg_fds[1] = -1;
465 	imsg_init(ibuf, imsg_fds[0]);
466 
467 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
468 	if (err)
469 		goto done;
470 	fd = -1;
471 
472 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
473 	if (err)
474 		goto done;
475 
476 	err = got_privsep_recv_gitconfig_int(
477 	    gitconfig_repository_format_version, ibuf);
478 	if (err)
479 		goto done;
480 
481 	if (extensions && nextensions) {
482 		err = got_privsep_send_gitconfig_repository_extensions_req(
483 		    ibuf);
484 		if (err)
485 			goto done;
486 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
487 		if (err)
488 			goto done;
489 		if (*nextensions > 0) {
490 			int i;
491 			*extensions = calloc(*nextensions, sizeof(char *));
492 			if (*extensions == NULL) {
493 				err = got_error_from_errno("calloc");
494 				goto done;
495 			}
496 			for (i = 0; i < *nextensions; i++) {
497 				char *ext;
498 				err = got_privsep_recv_gitconfig_str(&ext,
499 				    ibuf);
500 				if (err)
501 					goto done;
502 				(*extensions)[i] = ext;
503 			}
504 		}
505 	}
506 
507 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
508 	if (err)
509 		goto done;
510 
511 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
512 	if (err)
513 		goto done;
514 
515 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
516 	if (err)
517 		goto done;
518 
519 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
520 	if (err)
521 		goto done;
522 
523 	if (remotes && nremotes) {
524 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
525 		if (err)
526 			goto done;
527 
528 		err = got_privsep_recv_gitconfig_remotes(remotes,
529 		    nremotes, ibuf);
530 		if (err)
531 			goto done;
532 	}
533 
534 	if (gitconfig_owner) {
535 		err = got_privsep_send_gitconfig_owner_req(ibuf);
536 		if (err)
537 			goto done;
538 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
539 		if (err)
540 			goto done;
541 	}
542 
543 	imsg_clear(ibuf);
544 	err = got_privsep_send_stop(imsg_fds[0]);
545 	child_err = got_privsep_wait_for_child(pid);
546 	if (child_err && err == NULL)
547 		err = child_err;
548 done:
549 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
550 		err = got_error_from_errno("close");
551 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
552 		err = got_error_from_errno("close");
553 	if (fd != -1 && close(fd) == -1 && err == NULL)
554 		err = got_error_from_errno2("close", gitconfig_path);
555 	free(ibuf);
556 	return err;
557 }
558 
559 static const struct got_error *
560 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
561 {
562 	const struct got_error *err = NULL;
563 	char *repo_gitconfig_path = NULL;
564 
565 	if (global_gitconfig_path) {
566 		/* Read settings from ~/.gitconfig. */
567 		int dummy_repo_version;
568 		err = parse_gitconfig_file(&dummy_repo_version,
569 		    &repo->global_gitconfig_author_name,
570 		    &repo->global_gitconfig_author_email,
571 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
572 		if (err)
573 			return err;
574 	}
575 
576 	/* Read repository's .git/config file. */
577 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
578 	if (repo_gitconfig_path == NULL)
579 		return got_error_from_errno("got_repo_get_path_gitconfig");
580 
581 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
582 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
583 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
584 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
585 	    repo_gitconfig_path);
586 	if (err)
587 		goto done;
588 done:
589 	free(repo_gitconfig_path);
590 	return err;
591 }
592 
593 static const struct got_error *
594 read_gotconfig(struct got_repository *repo)
595 {
596 	const struct got_error *err = NULL;
597 	char *gotconfig_path;
598 
599 	gotconfig_path = got_repo_get_path_gotconfig(repo);
600 	if (gotconfig_path == NULL)
601 		return got_error_from_errno("got_repo_get_path_gotconfig");
602 
603 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
604 	free(gotconfig_path);
605 	return err;
606 }
607 
608 /* Supported repository format extensions. */
609 static const char *repo_extensions[] = {
610 	"noop",			/* Got supports repository format version 1. */
611 	"preciousObjects",	/* Supported by gotadmin cleanup. */
612 	"worktreeConfig",	/* Got does not care about Git work trees. */
613 };
614 
615 const struct got_error *
616 got_repo_open(struct got_repository **repop, const char *path,
617     const char *global_gitconfig_path)
618 {
619 	struct got_repository *repo = NULL;
620 	const struct got_error *err = NULL;
621 	char *repo_path = NULL;
622 	size_t i;
623 	struct rlimit rl;
624 
625 	*repop = NULL;
626 
627 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
628 		return got_error_from_errno("getrlimit");
629 
630 	repo = calloc(1, sizeof(*repo));
631 	if (repo == NULL) {
632 		err = got_error_from_errno("calloc");
633 		goto done;
634 	}
635 
636 	for (i = 0; i < nitems(repo->privsep_children); i++) {
637 		memset(&repo->privsep_children[i], 0,
638 		    sizeof(repo->privsep_children[0]));
639 		repo->privsep_children[i].imsg_fd = -1;
640 	}
641 
642 	err = got_object_cache_init(&repo->objcache,
643 	    GOT_OBJECT_CACHE_TYPE_OBJ);
644 	if (err)
645 		goto done;
646 	err = got_object_cache_init(&repo->treecache,
647 	    GOT_OBJECT_CACHE_TYPE_TREE);
648 	if (err)
649 		goto done;
650 	err = got_object_cache_init(&repo->commitcache,
651 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
652 	if (err)
653 		goto done;
654 	err = got_object_cache_init(&repo->tagcache,
655 	    GOT_OBJECT_CACHE_TYPE_TAG);
656 	if (err)
657 		goto done;
658 
659 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
660 	if (repo->pack_cache_size > rl.rlim_cur / 8)
661 		repo->pack_cache_size = rl.rlim_cur / 8;
662 
663 	repo_path = realpath(path, NULL);
664 	if (repo_path == NULL) {
665 		err = got_error_from_errno2("realpath", path);
666 		goto done;
667 	}
668 
669 	for (;;) {
670 		char *parent_path;
671 
672 		err = open_repo(repo, repo_path);
673 		if (err == NULL)
674 			break;
675 		if (err->code != GOT_ERR_NOT_GIT_REPO)
676 			goto done;
677 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
678 			err = got_error(GOT_ERR_NOT_GIT_REPO);
679 			goto done;
680 		}
681 		err = got_path_dirname(&parent_path, repo_path);
682 		if (err)
683 			goto done;
684 		free(repo_path);
685 		repo_path = parent_path;
686 	}
687 
688 	err = read_gotconfig(repo);
689 	if (err)
690 		goto done;
691 
692 	err = read_gitconfig(repo, global_gitconfig_path);
693 	if (err)
694 		goto done;
695 	if (repo->gitconfig_repository_format_version != 0)
696 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
697 	for (i = 0; i < repo->nextensions; i++) {
698 		char *ext = repo->extensions[i];
699 		int j, supported = 0;
700 		for (j = 0; j < nitems(repo_extensions); j++) {
701 			if (strcmp(ext, repo_extensions[j]) == 0) {
702 				supported = 1;
703 				break;
704 			}
705 		}
706 		if (!supported) {
707 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
708 			goto done;
709 		}
710 	}
711 done:
712 	if (err)
713 		got_repo_close(repo);
714 	else
715 		*repop = repo;
716 	free(repo_path);
717 	return err;
718 }
719 
720 const struct got_error *
721 got_repo_close(struct got_repository *repo)
722 {
723 	const struct got_error *err = NULL, *child_err;
724 	size_t i;
725 
726 	for (i = 0; i < repo->pack_cache_size; i++) {
727 		if (repo->packidx_cache[i] == NULL)
728 			break;
729 		got_packidx_close(repo->packidx_cache[i]);
730 	}
731 
732 	for (i = 0; i < repo->pack_cache_size; i++) {
733 		if (repo->packs[i].path_packfile == NULL)
734 			break;
735 		got_pack_close(&repo->packs[i]);
736 	}
737 
738 	free(repo->path);
739 	free(repo->path_git_dir);
740 
741 	got_object_cache_close(&repo->objcache);
742 	got_object_cache_close(&repo->treecache);
743 	got_object_cache_close(&repo->commitcache);
744 	got_object_cache_close(&repo->tagcache);
745 
746 	for (i = 0; i < nitems(repo->privsep_children); i++) {
747 		if (repo->privsep_children[i].imsg_fd == -1)
748 			continue;
749 		imsg_clear(repo->privsep_children[i].ibuf);
750 		free(repo->privsep_children[i].ibuf);
751 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
752 		child_err = got_privsep_wait_for_child(
753 		    repo->privsep_children[i].pid);
754 		if (child_err && err == NULL)
755 			err = child_err;
756 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
757 		    err == NULL)
758 			err = got_error_from_errno("close");
759 	}
760 
761 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
762 	    err == NULL)
763 		err = got_error_from_errno("close");
764 
765 	if (repo->gotconfig)
766 		got_gotconfig_free(repo->gotconfig);
767 	free(repo->gitconfig_author_name);
768 	free(repo->gitconfig_author_email);
769 	for (i = 0; i < repo->ngitconfig_remotes; i++)
770 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
771 	free(repo->gitconfig_remotes);
772 	for (i = 0; i < repo->nextensions; i++)
773 		free(repo->extensions[i]);
774 	free(repo->extensions);
775 	free(repo);
776 
777 	return err;
778 }
779 
780 void
781 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
782 {
783 	int i;
784 
785 	free(repo->name);
786 	repo->name = NULL;
787 	free(repo->fetch_url);
788 	repo->fetch_url = NULL;
789 	free(repo->send_url);
790 	repo->send_url = NULL;
791 	for (i = 0; i < repo->nfetch_branches; i++)
792 		free(repo->fetch_branches[i]);
793 	free(repo->fetch_branches);
794 	repo->fetch_branches = NULL;
795 	repo->nfetch_branches = 0;
796 	for (i = 0; i < repo->nsend_branches; i++)
797 		free(repo->send_branches[i]);
798 	free(repo->send_branches);
799 	repo->send_branches = NULL;
800 	repo->nsend_branches = 0;
801 }
802 
803 const struct got_error *
804 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
805     const char *input_path)
806 {
807 	const struct got_error *err = NULL;
808 	const char *repo_abspath = NULL;
809 	size_t repolen, len;
810 	char *canonpath, *path = NULL;
811 
812 	*in_repo_path = NULL;
813 
814 	canonpath = strdup(input_path);
815 	if (canonpath == NULL) {
816 		err = got_error_from_errno("strdup");
817 		goto done;
818 	}
819 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
820 	if (err)
821 		goto done;
822 
823 	repo_abspath = got_repo_get_path(repo);
824 
825 	if (canonpath[0] == '\0') {
826 		path = strdup(canonpath);
827 		if (path == NULL) {
828 			err = got_error_from_errno("strdup");
829 			goto done;
830 		}
831 	} else {
832 		path = realpath(canonpath, NULL);
833 		if (path == NULL) {
834 			if (errno != ENOENT) {
835 				err = got_error_from_errno2("realpath",
836 				    canonpath);
837 				goto done;
838 			}
839 			/*
840 			 * Path is not on disk.
841 			 * Assume it is already relative to repository root.
842 			 */
843 			path = strdup(canonpath);
844 			if (path == NULL) {
845 				err = got_error_from_errno("strdup");
846 				goto done;
847 			}
848 		}
849 
850 		repolen = strlen(repo_abspath);
851 		len = strlen(path);
852 
853 
854 		if (strcmp(path, repo_abspath) == 0) {
855 			free(path);
856 			path = strdup("");
857 			if (path == NULL) {
858 				err = got_error_from_errno("strdup");
859 				goto done;
860 			}
861 		} else if (len > repolen &&
862 		    got_path_is_child(path, repo_abspath, repolen)) {
863 			/* Matched an on-disk path inside repository. */
864 			if (got_repo_is_bare(repo)) {
865 				/*
866 				 * Matched an on-disk path inside repository
867 				 * database. Treat input as repository-relative.
868 				 */
869 				free(path);
870 				path = canonpath;
871 				canonpath = NULL;
872 			} else {
873 				char *child;
874 				/* Strip common prefix with repository path. */
875 				err = got_path_skip_common_ancestor(&child,
876 				    repo_abspath, path);
877 				if (err)
878 					goto done;
879 				free(path);
880 				path = child;
881 			}
882 		} else {
883 			/*
884 			 * Matched unrelated on-disk path.
885 			 * Treat input as repository-relative.
886 			 */
887 			free(path);
888 			path = canonpath;
889 			canonpath = NULL;
890 		}
891 	}
892 
893 	/* Make in-repository path absolute */
894 	if (path[0] != '/') {
895 		char *abspath;
896 		if (asprintf(&abspath, "/%s", path) == -1) {
897 			err = got_error_from_errno("asprintf");
898 			goto done;
899 		}
900 		free(path);
901 		path = abspath;
902 	}
903 
904 done:
905 	free(canonpath);
906 	if (err)
907 		free(path);
908 	else
909 		*in_repo_path = path;
910 	return err;
911 }
912 
913 static const struct got_error *
914 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
915     const char *path_packidx)
916 {
917 	const struct got_error *err = NULL;
918 	size_t i;
919 
920 	for (i = 0; i < repo->pack_cache_size; i++) {
921 		if (repo->packidx_cache[i] == NULL)
922 			break;
923 		if (strcmp(repo->packidx_cache[i]->path_packidx,
924 		    path_packidx) == 0) {
925 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
926 		}
927 	}
928 	if (i == repo->pack_cache_size) {
929 		i = repo->pack_cache_size - 1;
930 		err = got_packidx_close(repo->packidx_cache[i]);
931 		if (err)
932 			return err;
933 	}
934 
935 	repo->packidx_cache[i] = packidx;
936 
937 	return NULL;
938 }
939 
940 int
941 got_repo_is_packidx_filename(const char *name, size_t len)
942 {
943 	if (len != GOT_PACKIDX_NAMELEN)
944 		return 0;
945 
946 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
947 		return 0;
948 
949 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
950 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
951 		return 0;
952 
953 	return 1;
954 }
955 
956 const struct got_error *
957 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
958     struct got_repository *repo, struct got_object_id *id)
959 {
960 	const struct got_error *err;
961 	DIR *packdir = NULL;
962 	struct dirent *dent;
963 	char *path_packidx;
964 	size_t i;
965 	int packdir_fd;
966 
967 	/* Search pack index cache. */
968 	for (i = 0; i < repo->pack_cache_size; i++) {
969 		if (repo->packidx_cache[i] == NULL)
970 			break;
971 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
972 		if (*idx != -1) {
973 			*packidx = repo->packidx_cache[i];
974 			/*
975 			 * Move this cache entry to the front. Repeatedly
976 			 * searching a wrong pack index can be expensive.
977 			 */
978 			if (i > 0) {
979 				memmove(&repo->packidx_cache[1],
980 				    &repo->packidx_cache[0],
981 				    i * sizeof(repo->packidx_cache[0]));
982 				repo->packidx_cache[0] = *packidx;
983 			}
984 			return NULL;
985 		}
986 	}
987 	/* No luck. Search the filesystem. */
988 
989 	packdir_fd = openat(got_repo_get_fd(repo),
990 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
991 	if (packdir_fd == -1) {
992 		if (errno == ENOENT)
993 			err = got_error_no_obj(id);
994 		else
995 			err = got_error_from_errno_fmt("openat: %s/%s",
996 			    got_repo_get_path_git_dir(repo),
997 			    GOT_OBJECTS_PACK_DIR);
998 		goto done;
999 	}
1000 
1001 	packdir = fdopendir(packdir_fd);
1002 	if (packdir == NULL) {
1003 		err = got_error_from_errno("fdopendir");
1004 		goto done;
1005 	}
1006 
1007 	while ((dent = readdir(packdir)) != NULL) {
1008 		int is_cached = 0;
1009 
1010 		if (!got_repo_is_packidx_filename(dent->d_name,
1011 		    strlen(dent->d_name)))
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 (!got_repo_is_packidx_filename(dent->d_name,
1280 		    strlen(dent->d_name)))
1281 			continue;
1282 
1283 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1284 		    dent->d_name) == -1) {
1285 			err = got_error_from_errno("strdup");
1286 			break;
1287 		}
1288 
1289 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1290 		    path_packidx, 0);
1291 		free(path_packidx);
1292 		if (err)
1293 			break;
1294 
1295 		err = got_packidx_match_id_str_prefix(&matched_ids,
1296 		    packidx, id_str_prefix);
1297 		if (err) {
1298 			got_packidx_close(packidx);
1299 			break;
1300 		}
1301 		err = got_packidx_close(packidx);
1302 		if (err)
1303 			break;
1304 
1305 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1306 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1307 				int matched_type;
1308 				err = got_object_get_type(&matched_type, repo,
1309 				    qid->id);
1310 				if (err)
1311 					goto done;
1312 				if (matched_type != obj_type)
1313 					continue;
1314 			}
1315 			if (*unique_id == NULL) {
1316 				*unique_id = got_object_id_dup(qid->id);
1317 				if (*unique_id == NULL) {
1318 					err = got_error_from_errno("malloc");
1319 					goto done;
1320 				}
1321 			} else {
1322 				if (got_object_id_cmp(*unique_id, qid->id) == 0)
1323 					continue; /* packed multiple times */
1324 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1325 				goto done;
1326 			}
1327 		}
1328 	}
1329 done:
1330 	got_object_id_queue_free(&matched_ids);
1331 	if (packdir && closedir(packdir) != 0 && err == NULL)
1332 		err = got_error_from_errno("closedir");
1333 	if (err) {
1334 		free(*unique_id);
1335 		*unique_id = NULL;
1336 	}
1337 	return err;
1338 }
1339 
1340 static const struct got_error *
1341 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1342     const char *object_dir, const char *id_str_prefix, int obj_type,
1343     struct got_repository *repo)
1344 {
1345 	const struct got_error *err = NULL;
1346 	char *path;
1347 	DIR *dir = NULL;
1348 	struct dirent *dent;
1349 	struct got_object_id id;
1350 
1351 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1352 		err = got_error_from_errno("asprintf");
1353 		goto done;
1354 	}
1355 
1356 	dir = opendir(path);
1357 	if (dir == NULL) {
1358 		if (errno == ENOENT) {
1359 			err = NULL;
1360 			goto done;
1361 		}
1362 		err = got_error_from_errno2("opendir", path);
1363 		goto done;
1364 	}
1365 	while ((dent = readdir(dir)) != NULL) {
1366 		char *id_str;
1367 		int cmp;
1368 
1369 		if (strcmp(dent->d_name, ".") == 0 ||
1370 		    strcmp(dent->d_name, "..") == 0)
1371 			continue;
1372 
1373 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1374 			err = got_error_from_errno("asprintf");
1375 			goto done;
1376 		}
1377 
1378 		if (!got_parse_sha1_digest(id.sha1, id_str))
1379 			continue;
1380 
1381 		/*
1382 		 * Directory entries do not necessarily appear in
1383 		 * sorted order, so we must iterate over all of them.
1384 		 */
1385 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1386 		if (cmp != 0) {
1387 			free(id_str);
1388 			continue;
1389 		}
1390 
1391 		if (*unique_id == NULL) {
1392 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1393 				int matched_type;
1394 				err = got_object_get_type(&matched_type, repo,
1395 				    &id);
1396 				if (err)
1397 					goto done;
1398 				if (matched_type != obj_type)
1399 					continue;
1400 			}
1401 			*unique_id = got_object_id_dup(&id);
1402 			if (*unique_id == NULL) {
1403 				err = got_error_from_errno("got_object_id_dup");
1404 				free(id_str);
1405 				goto done;
1406 			}
1407 		} else {
1408 			if (got_object_id_cmp(*unique_id, &id) == 0)
1409 				continue; /* both packed and loose */
1410 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1411 			free(id_str);
1412 			goto done;
1413 		}
1414 	}
1415 done:
1416 	if (dir && closedir(dir) != 0 && err == NULL)
1417 		err = got_error_from_errno("closedir");
1418 	if (err) {
1419 		free(*unique_id);
1420 		*unique_id = NULL;
1421 	}
1422 	free(path);
1423 	return err;
1424 }
1425 
1426 const struct got_error *
1427 got_repo_match_object_id_prefix(struct got_object_id **id,
1428     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1429 {
1430 	const struct got_error *err = NULL;
1431 	char *path_objects = got_repo_get_path_objects(repo);
1432 	char *object_dir = NULL;
1433 	size_t len;
1434 	int i;
1435 
1436 	*id = NULL;
1437 
1438 	for (i = 0; i < strlen(id_str_prefix); i++) {
1439 		if (isxdigit((unsigned char)id_str_prefix[i]))
1440 			continue;
1441 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1442 	}
1443 
1444 	len = strlen(id_str_prefix);
1445 	if (len >= 2) {
1446 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1447 		if (err)
1448 			goto done;
1449 		object_dir = strndup(id_str_prefix, 2);
1450 		if (object_dir == NULL) {
1451 			err = got_error_from_errno("strdup");
1452 			goto done;
1453 		}
1454 		err = match_loose_object(id, path_objects, object_dir,
1455 		    id_str_prefix, obj_type, repo);
1456 	} else if (len == 1) {
1457 		int i;
1458 		for (i = 0; i < 0xf; i++) {
1459 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1460 			    == -1) {
1461 				err = got_error_from_errno("asprintf");
1462 				goto done;
1463 			}
1464 			err = match_packed_object(id, repo, object_dir,
1465 			    obj_type);
1466 			if (err)
1467 				goto done;
1468 			err = match_loose_object(id, path_objects, object_dir,
1469 			    id_str_prefix, obj_type, repo);
1470 			if (err)
1471 				goto done;
1472 		}
1473 	} else {
1474 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1475 		goto done;
1476 	}
1477 done:
1478 	free(object_dir);
1479 	if (err) {
1480 		free(*id);
1481 		*id = NULL;
1482 	} else if (*id == NULL) {
1483 		switch (obj_type) {
1484 		case GOT_OBJ_TYPE_BLOB:
1485 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1486 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1487 			break;
1488 		case GOT_OBJ_TYPE_TREE:
1489 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1490 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1491 			break;
1492 		case GOT_OBJ_TYPE_COMMIT:
1493 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1494 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1495 			break;
1496 		case GOT_OBJ_TYPE_TAG:
1497 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1498 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1499 			break;
1500 		default:
1501 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1502 			break;
1503 		}
1504 	}
1505 
1506 	return err;
1507 }
1508 
1509 const struct got_error *
1510 got_repo_match_object_id(struct got_object_id **id, char **label,
1511     const char *id_str, int obj_type, struct got_reflist_head *refs,
1512     struct got_repository *repo)
1513 {
1514 	const struct got_error *err;
1515 	struct got_tag_object *tag;
1516 	struct got_reference *ref = NULL;
1517 
1518 	*id = NULL;
1519 	if (label)
1520 		*label = NULL;
1521 
1522 	if (refs) {
1523 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1524 		    refs, repo);
1525 		if (err == NULL) {
1526 			*id = got_object_id_dup(
1527 			    got_object_tag_get_object_id(tag));
1528 			if (*id == NULL)
1529 				err = got_error_from_errno("got_object_id_dup");
1530 			else if (label && asprintf(label, "refs/tags/%s",
1531 			    got_object_tag_get_name(tag)) == -1) {
1532 				err = got_error_from_errno("asprintf");
1533 				free(*id);
1534 				*id = NULL;
1535 			}
1536 			got_object_tag_close(tag);
1537 			return err;
1538 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1539 		    err->code != GOT_ERR_NO_OBJ)
1540 			return err;
1541 	}
1542 
1543 	err = got_repo_match_object_id_prefix(id, id_str, obj_type, repo);
1544 	if (err) {
1545 		if (err->code != GOT_ERR_BAD_OBJ_ID_STR)
1546 			return err;
1547 		err = got_ref_open(&ref, repo, id_str, 0);
1548 		if (err != NULL)
1549 			goto done;
1550 		if (label) {
1551 			*label = strdup(got_ref_get_name(ref));
1552 			if (*label == NULL) {
1553 				err = got_error_from_errno("strdup");
1554 				goto done;
1555 			}
1556 		}
1557 		err = got_ref_resolve(id, repo, ref);
1558 	} else if (label) {
1559 		err = got_object_id_str(label, *id);
1560 		if (*label == NULL) {
1561 			err = got_error_from_errno("strdup");
1562 			goto done;
1563 		}
1564 	}
1565 done:
1566 	if (ref)
1567 		got_ref_close(ref);
1568 	return err;
1569 }
1570 
1571 const struct got_error *
1572 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1573     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1574 {
1575 	const struct got_error *err = NULL;
1576 	struct got_reflist_entry *re;
1577 	struct got_object_id *tag_id;
1578 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1579 
1580 	*tag = NULL;
1581 
1582 	TAILQ_FOREACH(re, refs, entry) {
1583 		const char *refname;
1584 		refname = got_ref_get_name(re->ref);
1585 		if (got_ref_is_symbolic(re->ref))
1586 			continue;
1587 		if (strncmp(refname, "refs/tags/", 10) != 0)
1588 			continue;
1589 		if (!name_is_absolute)
1590 			refname += strlen("refs/tags/");
1591 		if (strcmp(refname, name) != 0)
1592 			continue;
1593 		err = got_ref_resolve(&tag_id, repo, re->ref);
1594 		if (err)
1595 			break;
1596 		err = got_object_open_as_tag(tag, repo, tag_id);
1597 		free(tag_id);
1598 		if (err)
1599 			break;
1600 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1601 		    got_object_tag_get_object_type(*tag) == obj_type)
1602 			break;
1603 		got_object_tag_close(*tag);
1604 		*tag = NULL;
1605 	}
1606 
1607 	if (err == NULL && *tag == NULL)
1608 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1609 		    GOT_OBJ_LABEL_TAG, name);
1610 	return err;
1611 }
1612 
1613 static const struct got_error *
1614 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1615     const char *name, mode_t mode, struct got_object_id *blob_id)
1616 {
1617 	const struct got_error *err = NULL;
1618 
1619 	 *new_te = NULL;
1620 
1621 	*new_te = calloc(1, sizeof(**new_te));
1622 	if (*new_te == NULL)
1623 		return got_error_from_errno("calloc");
1624 
1625 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
1626 	    sizeof((*new_te)->name)) {
1627 		err = got_error(GOT_ERR_NO_SPACE);
1628 		goto done;
1629 	}
1630 
1631 	if (S_ISLNK(mode)) {
1632 		(*new_te)->mode = S_IFLNK;
1633 	} else {
1634 		(*new_te)->mode = S_IFREG;
1635 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
1636 	}
1637 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
1638 done:
1639 	if (err && *new_te) {
1640 		free(*new_te);
1641 		*new_te = NULL;
1642 	}
1643 	return err;
1644 }
1645 
1646 static const struct got_error *
1647 import_file(struct got_tree_entry **new_te, struct dirent *de,
1648     const char *path, struct got_repository *repo)
1649 {
1650 	const struct got_error *err;
1651 	struct got_object_id *blob_id = NULL;
1652 	char *filepath;
1653 	struct stat sb;
1654 
1655 	if (asprintf(&filepath, "%s%s%s", path,
1656 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1657 		return got_error_from_errno("asprintf");
1658 
1659 	if (lstat(filepath, &sb) != 0) {
1660 		err = got_error_from_errno2("lstat", path);
1661 		goto done;
1662 	}
1663 
1664 	err = got_object_blob_create(&blob_id, filepath, repo);
1665 	if (err)
1666 		goto done;
1667 
1668 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1669 	    blob_id);
1670 done:
1671 	free(filepath);
1672 	if (err)
1673 		free(blob_id);
1674 	return err;
1675 }
1676 
1677 static const struct got_error *
1678 insert_tree_entry(struct got_tree_entry *new_te,
1679     struct got_pathlist_head *paths)
1680 {
1681 	const struct got_error *err = NULL;
1682 	struct got_pathlist_entry *new_pe;
1683 
1684 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1685 	if (err)
1686 		return err;
1687 	if (new_pe == NULL)
1688 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1689 	return NULL;
1690 }
1691 
1692 static const struct got_error *write_tree(struct got_object_id **,
1693     const char *, struct got_pathlist_head *, struct got_repository *,
1694     got_repo_import_cb progress_cb, void *progress_arg);
1695 
1696 static const struct got_error *
1697 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1698     const char *path, struct got_pathlist_head *ignores,
1699     struct got_repository *repo,
1700     got_repo_import_cb progress_cb, void *progress_arg)
1701 {
1702 	const struct got_error *err;
1703 	struct got_object_id *id = NULL;
1704 	char *subdirpath;
1705 
1706 	if (asprintf(&subdirpath, "%s%s%s", path,
1707 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1708 		return got_error_from_errno("asprintf");
1709 
1710 	(*new_te) = calloc(1, sizeof(**new_te));
1711 	if (*new_te == NULL)
1712 		return got_error_from_errno("calloc");
1713 	(*new_te)->mode = S_IFDIR;
1714 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
1715 	    sizeof((*new_te)->name)) {
1716 		err = got_error(GOT_ERR_NO_SPACE);
1717 		goto done;
1718 	}
1719 	err = write_tree(&id, subdirpath, ignores,  repo,
1720 	    progress_cb, progress_arg);
1721 	if (err)
1722 		goto done;
1723 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
1724 	
1725 done:
1726 	free(id);
1727 	free(subdirpath);
1728 	if (err) {
1729 		free(*new_te);
1730 		*new_te = NULL;
1731 	}
1732 	return err;
1733 }
1734 
1735 static const struct got_error *
1736 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1737     struct got_pathlist_head *ignores, struct got_repository *repo,
1738     got_repo_import_cb progress_cb, void *progress_arg)
1739 {
1740 	const struct got_error *err = NULL;
1741 	DIR *dir;
1742 	struct dirent *de;
1743 	int nentries;
1744 	struct got_tree_entry *new_te = NULL;
1745 	struct got_pathlist_head paths;
1746 	struct got_pathlist_entry *pe;
1747 
1748 	*new_tree_id = NULL;
1749 
1750 	TAILQ_INIT(&paths);
1751 
1752 	dir = opendir(path_dir);
1753 	if (dir == NULL) {
1754 		err = got_error_from_errno2("opendir", path_dir);
1755 		goto done;
1756 	}
1757 
1758 	nentries = 0;
1759 	while ((de = readdir(dir)) != NULL) {
1760 		int ignore = 0;
1761 		int type;
1762 
1763 		if (strcmp(de->d_name, ".") == 0 ||
1764 		    strcmp(de->d_name, "..") == 0)
1765 			continue;
1766 
1767 		TAILQ_FOREACH(pe, ignores, entry) {
1768 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1769 				ignore = 1;
1770 				break;
1771 			}
1772 		}
1773 		if (ignore)
1774 			continue;
1775 
1776 		err = got_path_dirent_type(&type, path_dir, de);
1777 		if (err)
1778 			goto done;
1779 
1780 		if (type == DT_DIR) {
1781 			err = import_subdir(&new_te, de, path_dir,
1782 			    ignores, repo, progress_cb, progress_arg);
1783 			if (err) {
1784 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
1785 					goto done;
1786 				err = NULL;
1787 				continue;
1788 			}
1789 		} else if (type == DT_REG || type == DT_LNK) {
1790 			err = import_file(&new_te, de, path_dir, repo);
1791 			if (err)
1792 				goto done;
1793 		} else
1794 			continue;
1795 
1796 		err = insert_tree_entry(new_te, &paths);
1797 		if (err)
1798 			goto done;
1799 		nentries++;
1800 	}
1801 
1802 	if (TAILQ_EMPTY(&paths)) {
1803 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
1804 		    "cannot create tree without any entries");
1805 		goto done;
1806 	}
1807 
1808 	TAILQ_FOREACH(pe, &paths, entry) {
1809 		struct got_tree_entry *te = pe->data;
1810 		char *path;
1811 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
1812 			continue;
1813 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1814 			err = got_error_from_errno("asprintf");
1815 			goto done;
1816 		}
1817 		err = (*progress_cb)(progress_arg, path);
1818 		free(path);
1819 		if (err)
1820 			goto done;
1821 	}
1822 
1823 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
1824 done:
1825 	if (dir)
1826 		closedir(dir);
1827 	got_pathlist_free(&paths);
1828 	return err;
1829 }
1830 
1831 const struct got_error *
1832 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1833     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1834     struct got_repository *repo, got_repo_import_cb progress_cb,
1835     void *progress_arg)
1836 {
1837 	const struct got_error *err;
1838 	struct got_object_id *new_tree_id;
1839 
1840 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1841 	    progress_cb, progress_arg);
1842 	if (err)
1843 		return err;
1844 
1845 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1846 	    author, time(NULL), author, time(NULL), logmsg, repo);
1847 	free(new_tree_id);
1848 	return err;
1849 }
1850 
1851 const struct got_error *
1852 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
1853     struct got_repository *repo)
1854 {
1855 	const struct got_error *err = NULL;
1856 	char *path_objects = NULL, *path = NULL;
1857 	DIR *dir = NULL;
1858 	struct got_object_id id;
1859 	int i;
1860 
1861 	*nobjects = 0;
1862 	*ondisk_size = 0;
1863 
1864 	path_objects = got_repo_get_path_objects(repo);
1865 	if (path_objects == NULL)
1866 		return got_error_from_errno("got_repo_get_path_objects");
1867 
1868 	for (i = 0; i <= 0xff; i++) {
1869 		struct dirent *dent;
1870 
1871 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
1872 			err = got_error_from_errno("asprintf");
1873 			break;
1874 		}
1875 
1876 		dir = opendir(path);
1877 		if (dir == NULL) {
1878 			if (errno == ENOENT) {
1879 				err = NULL;
1880 				continue;
1881 			}
1882 			err = got_error_from_errno2("opendir", path);
1883 			break;
1884 		}
1885 
1886 		while ((dent = readdir(dir)) != NULL) {
1887 			char *id_str;
1888 			int fd;
1889 			struct stat sb;
1890 
1891 			if (strcmp(dent->d_name, ".") == 0 ||
1892 			    strcmp(dent->d_name, "..") == 0)
1893 				continue;
1894 
1895 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
1896 				err = got_error_from_errno("asprintf");
1897 				goto done;
1898 			}
1899 
1900 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
1901 				free(id_str);
1902 				continue;
1903 			}
1904 			free(id_str);
1905 
1906 			err = got_object_open_loose_fd(&fd, &id, repo);
1907 			if (err)
1908 				goto done;
1909 
1910 			if (fstat(fd, &sb) == -1) {
1911 				err = got_error_from_errno("fstat");
1912 				close(fd);
1913 				goto done;
1914 			}
1915 			(*nobjects)++;
1916 			(*ondisk_size) += sb.st_size;
1917 
1918 			if (close(fd) == -1) {
1919 				err = got_error_from_errno("close");
1920 				goto done;
1921 			}
1922 		}
1923 
1924 		if (closedir(dir) != 0) {
1925 			err = got_error_from_errno("closedir");
1926 			goto done;
1927 		}
1928 		dir = NULL;
1929 
1930 		free(path);
1931 		path = NULL;
1932 	}
1933 done:
1934 	if (dir && closedir(dir) != 0 && err == NULL)
1935 		err = got_error_from_errno("closedir");
1936 
1937 	if (err) {
1938 		*nobjects = 0;
1939 		*ondisk_size = 0;
1940 	}
1941 	free(path_objects);
1942 	free(path);
1943 	return err;
1944 }
1945 
1946 const struct got_error *
1947 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
1948     off_t *total_packsize, struct got_repository *repo)
1949 {
1950 	const struct got_error *err = NULL;
1951 	DIR *packdir = NULL;
1952 	struct dirent *dent;
1953 	struct got_packidx *packidx = NULL;
1954 	char *path_packidx;
1955 	char *path_packfile;
1956 	int packdir_fd;
1957 	struct stat sb;
1958 
1959 	*npackfiles = 0;
1960 	*nobjects = 0;
1961 	*total_packsize = 0;
1962 
1963 	packdir_fd = openat(got_repo_get_fd(repo),
1964 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
1965 	if (packdir_fd == -1) {
1966 		return got_error_from_errno_fmt("openat: %s/%s",
1967 		    got_repo_get_path_git_dir(repo),
1968 		    GOT_OBJECTS_PACK_DIR);
1969 	}
1970 
1971 	packdir = fdopendir(packdir_fd);
1972 	if (packdir == NULL) {
1973 		err = got_error_from_errno("fdopendir");
1974 		goto done;
1975 	}
1976 
1977 	while ((dent = readdir(packdir)) != NULL) {
1978 		if (!got_repo_is_packidx_filename(dent->d_name,
1979 		    strlen(dent->d_name)))
1980 			continue;
1981 
1982 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1983 		    dent->d_name) == -1) {
1984 			err = got_error_from_errno("asprintf");
1985 			goto done;
1986 		}
1987 
1988 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1989 		    path_packidx, 0);
1990 		free(path_packidx);
1991 		if (err)
1992 			goto done;
1993 
1994 		if (fstat(packidx->fd, &sb) == -1)
1995 			goto done;
1996 		*total_packsize += sb.st_size;
1997 
1998 		err = got_packidx_get_packfile_path(&path_packfile,
1999 		    packidx->path_packidx);
2000 		if (err)
2001 			goto done;
2002 
2003 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2004 		    0) == -1) {
2005 			free(path_packfile);
2006 			goto done;
2007 		}
2008 		free(path_packfile);
2009 		*total_packsize += sb.st_size;
2010 
2011 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2012 
2013 		(*npackfiles)++;
2014 
2015 		got_packidx_close(packidx);
2016 		packidx = NULL;
2017 	}
2018 done:
2019 	if (packidx)
2020 		got_packidx_close(packidx);
2021 	if (packdir && closedir(packdir) != 0 && err == NULL)
2022 		err = got_error_from_errno("closedir");
2023 	if (err) {
2024 		*npackfiles = 0;
2025 		*nobjects = 0;
2026 		*total_packsize = 0;
2027 	}
2028 	return err;
2029 }