op public repos

Blob

Date:: Sat Sep 3 13:05:15 2022 UTC
Message:: plug mem leak in got_repo_pack_fds_open pack_fds_tmp is not always free'd. However, while here, simplify it to not use a temporary array. ok stsp@
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/tree.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/stat.h>
23 #include <sys/mman.h>
24 #include <sys/resource.h>
25 
26 #include <ctype.h>
27 #include <endian.h>
28 #include <fcntl.h>
29 #include <fnmatch.h>
30 #include <limits.h>
31 #include <dirent.h>
32 #include <stdlib.h>
33 #include <stdio.h>
34 #include <sha1.h>
35 #include <string.h>
36 #include <time.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <errno.h>
40 #include <libgen.h>
41 #include <stdint.h>
42 #include <imsg.h>
43 #include <uuid.h>
44 
45 #include "bloom.h"
46 
47 #include "got_error.h"
48 #include "got_reference.h"
49 #include "got_repository.h"
50 #include "got_path.h"
51 #include "got_cancel.h"
52 #include "got_object.h"
53 #include "got_opentemp.h"
54 
55 #include "got_lib_delta.h"
56 #include "got_lib_inflate.h"
57 #include "got_lib_object.h"
58 #include "got_lib_object_parse.h"
59 #include "got_lib_object_create.h"
60 #include "got_lib_pack.h"
61 #include "got_lib_privsep.h"
62 #include "got_lib_sha1.h"
63 #include "got_lib_object_cache.h"
64 #include "got_lib_repository.h"
65 #include "got_lib_gotconfig.h"
66 
67 #ifndef nitems
68 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
69 #endif
70 
71 #define GOT_PACK_NUM_TEMPFILES		GOT_PACK_CACHE_SIZE * 2
72 
73 RB_PROTOTYPE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
74     got_packidx_bloom_filter_cmp);
75 
76 const char *
77 got_repo_get_path(struct got_repository *repo)
78 {
79 	return repo->path;
80 }
81 
82 const char *
83 got_repo_get_path_git_dir(struct got_repository *repo)
84 {
85 	return repo->path_git_dir;
86 }
87 
88 int
89 got_repo_get_fd(struct got_repository *repo)
90 {
91 	return repo->gitdir_fd;
92 }
93 
94 const char *
95 got_repo_get_gitconfig_author_name(struct got_repository *repo)
96 {
97 	return repo->gitconfig_author_name;
98 }
99 
100 const char *
101 got_repo_get_gitconfig_author_email(struct got_repository *repo)
102 {
103 	return repo->gitconfig_author_email;
104 }
105 
106 const char *
107 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
108 {
109 	return repo->global_gitconfig_author_name;
110 }
111 
112 const char *
113 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
114 {
115 	return repo->global_gitconfig_author_email;
116 }
117 
118 const char *
119 got_repo_get_gitconfig_owner(struct got_repository *repo)
120 {
121 	return repo->gitconfig_owner;
122 }
123 
124 void
125 got_repo_get_gitconfig_extensions(char ***extensions, int *nextensions,
126     struct got_repository *repo)
127 {
128 	*extensions = repo->extensions;
129 	*nextensions = repo->nextensions;
130 }
131 
132 int
133 got_repo_is_bare(struct got_repository *repo)
134 {
135 	return (strcmp(repo->path, repo->path_git_dir) == 0);
136 }
137 
138 static char *
139 get_path_git_child(struct got_repository *repo, const char *basename)
140 {
141 	char *path_child;
142 
143 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
144 	    basename) == -1)
145 		return NULL;
146 
147 	return path_child;
148 }
149 
150 char *
151 got_repo_get_path_objects(struct got_repository *repo)
152 {
153 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
154 }
155 
156 char *
157 got_repo_get_path_objects_pack(struct got_repository *repo)
158 {
159 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
160 }
161 
162 char *
163 got_repo_get_path_refs(struct got_repository *repo)
164 {
165 	return get_path_git_child(repo, GOT_REFS_DIR);
166 }
167 
168 char *
169 got_repo_get_path_packed_refs(struct got_repository *repo)
170 {
171 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
172 }
173 
174 static char *
175 get_path_head(struct got_repository *repo)
176 {
177 	return get_path_git_child(repo, GOT_HEAD_FILE);
178 }
179 
180 char *
181 got_repo_get_path_gitconfig(struct got_repository *repo)
182 {
183 	return get_path_git_child(repo, GOT_GITCONFIG);
184 }
185 
186 char *
187 got_repo_get_path_gotconfig(struct got_repository *repo)
188 {
189 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
190 }
191 
192 const struct got_gotconfig *
193 got_repo_get_gotconfig(struct got_repository *repo)
194 {
195 	return repo->gotconfig;
196 }
197 
198 void
199 got_repo_get_gitconfig_remotes(int *nremotes,
200     const struct got_remote_repo **remotes, struct got_repository *repo)
201 {
202 	*nremotes = repo->ngitconfig_remotes;
203 	*remotes = repo->gitconfig_remotes;
204 }
205 
206 static int
207 is_git_repo(struct got_repository *repo)
208 {
209 	const char *path_git = got_repo_get_path_git_dir(repo);
210 	char *path_objects = got_repo_get_path_objects(repo);
211 	char *path_refs = got_repo_get_path_refs(repo);
212 	char *path_head = get_path_head(repo);
213 	int ret = 0;
214 	struct stat sb;
215 	struct got_reference *head_ref;
216 
217 	if (lstat(path_git, &sb) == -1)
218 		goto done;
219 	if (!S_ISDIR(sb.st_mode))
220 		goto done;
221 
222 	if (lstat(path_objects, &sb) == -1)
223 		goto done;
224 	if (!S_ISDIR(sb.st_mode))
225 		goto done;
226 
227 	if (lstat(path_refs, &sb) == -1)
228 		goto done;
229 	if (!S_ISDIR(sb.st_mode))
230 		goto done;
231 
232 	if (lstat(path_head, &sb) == -1)
233 		goto done;
234 	if (!S_ISREG(sb.st_mode))
235 		goto done;
236 
237 	/* Check if the HEAD reference can be opened. */
238 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
239 		goto done;
240 	got_ref_close(head_ref);
241 
242 	ret = 1;
243 done:
244 	free(path_objects);
245 	free(path_refs);
246 	free(path_head);
247 	return ret;
248 
249 }
250 
251 const struct got_error *
252 got_repo_pack_fds_open(int **pack_fds)
253 {
254 	const struct got_error *err = NULL;
255 	int i;
256 
257 	*pack_fds = calloc(GOT_PACK_NUM_TEMPFILES, sizeof(**pack_fds));
258 	if (*pack_fds == NULL)
259 		return got_error_from_errno("calloc");
260 
261 	/*
262 	 * got_repo_pack_fds_close will try to close all of the
263 	 * GOT_PACK_NUM_TEMPFILES fds, even the ones that didn't manage to get
264 	 * a value from got_opentempfd(), which would result in a close(0) if
265 	 * we do not initialize to -1 here.
266 	 */
267 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++)
268 		(*pack_fds)[i] = -1;
269 
270 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
271 		(*pack_fds)[i] = got_opentempfd();
272 		if ((*pack_fds)[i] == -1) {
273 			err = got_error_from_errno("got_opentempfd");
274 			got_repo_pack_fds_close(*pack_fds);
275 			*pack_fds = NULL;
276 			return err;
277 		}
278 	}
279 
280 	return NULL;
281 }
282 
283 const struct got_error *
284 got_repo_pack_fds_close(int *pack_fds)
285 {
286 	const struct got_error *err = NULL;
287 	int i;
288 
289 	for (i = 0; i < GOT_PACK_NUM_TEMPFILES; i++) {
290 		if (pack_fds[i] == -1)
291 			continue;
292 		if (close(pack_fds[i]) == -1) {
293 			err = got_error_from_errno("close");
294 			break;
295 		}
296 	}
297 	free(pack_fds);
298 	return err;
299 }
300 
301 const struct got_error *
302 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
303     struct got_object *obj)
304 {
305 #ifndef GOT_NO_OBJ_CACHE
306 	const struct got_error *err = NULL;
307 	err = got_object_cache_add(&repo->objcache, id, obj);
308 	if (err) {
309 		if (err->code == GOT_ERR_OBJ_EXISTS ||
310 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
311 			err = NULL;
312 		return err;
313 	}
314 	obj->refcnt++;
315 #endif
316 	return NULL;
317 }
318 
319 struct got_object *
320 got_repo_get_cached_object(struct got_repository *repo,
321     struct got_object_id *id)
322 {
323 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
324 }
325 
326 const struct got_error *
327 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
328     struct got_tree_object *tree)
329 {
330 #ifndef GOT_NO_OBJ_CACHE
331 	const struct got_error *err = NULL;
332 	err = got_object_cache_add(&repo->treecache, id, tree);
333 	if (err) {
334 		if (err->code == GOT_ERR_OBJ_EXISTS ||
335 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
336 			err = NULL;
337 		return err;
338 	}
339 	tree->refcnt++;
340 #endif
341 	return NULL;
342 }
343 
344 struct got_tree_object *
345 got_repo_get_cached_tree(struct got_repository *repo,
346     struct got_object_id *id)
347 {
348 	return (struct got_tree_object *)got_object_cache_get(
349 	    &repo->treecache, id);
350 }
351 
352 const struct got_error *
353 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
354     struct got_commit_object *commit)
355 {
356 #ifndef GOT_NO_OBJ_CACHE
357 	const struct got_error *err = NULL;
358 	err = got_object_cache_add(&repo->commitcache, id, commit);
359 	if (err) {
360 		if (err->code == GOT_ERR_OBJ_EXISTS ||
361 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
362 			err = NULL;
363 		return err;
364 	}
365 	commit->refcnt++;
366 #endif
367 	return NULL;
368 }
369 
370 struct got_commit_object *
371 got_repo_get_cached_commit(struct got_repository *repo,
372     struct got_object_id *id)
373 {
374 	return (struct got_commit_object *)got_object_cache_get(
375 	    &repo->commitcache, id);
376 }
377 
378 const struct got_error *
379 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
380     struct got_tag_object *tag)
381 {
382 #ifndef GOT_NO_OBJ_CACHE
383 	const struct got_error *err = NULL;
384 	err = got_object_cache_add(&repo->tagcache, id, tag);
385 	if (err) {
386 		if (err->code == GOT_ERR_OBJ_EXISTS ||
387 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
388 			err = NULL;
389 		return err;
390 	}
391 	tag->refcnt++;
392 #endif
393 	return NULL;
394 }
395 
396 struct got_tag_object *
397 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
398 {
399 	return (struct got_tag_object *)got_object_cache_get(
400 	    &repo->tagcache, id);
401 }
402 
403 const struct got_error *
404 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
405     struct got_raw_object *raw)
406 {
407 #ifndef GOT_NO_OBJ_CACHE
408 	const struct got_error *err = NULL;
409 	err = got_object_cache_add(&repo->rawcache, id, raw);
410 	if (err) {
411 		if (err->code == GOT_ERR_OBJ_EXISTS ||
412 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
413 			err = NULL;
414 		return err;
415 	}
416 	raw->refcnt++;
417 #endif
418 	return NULL;
419 }
420 
421 
422 struct got_raw_object *
423 got_repo_get_cached_raw_object(struct got_repository *repo,
424     struct got_object_id *id)
425 {
426 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
427 }
428 
429 
430 static const struct got_error *
431 open_repo(struct got_repository *repo, const char *path)
432 {
433 	const struct got_error *err = NULL;
434 
435 	repo->gitdir_fd = -1;
436 
437 	/* bare git repository? */
438 	repo->path_git_dir = strdup(path);
439 	if (repo->path_git_dir == NULL)
440 		return got_error_from_errno("strdup");
441 	if (is_git_repo(repo)) {
442 		repo->path = strdup(repo->path_git_dir);
443 		if (repo->path == NULL) {
444 			err = got_error_from_errno("strdup");
445 			goto done;
446 		}
447 		repo->gitdir_fd = open(repo->path_git_dir,
448 		    O_DIRECTORY | O_CLOEXEC);
449 		if (repo->gitdir_fd == -1) {
450 			err = got_error_from_errno2("open",
451 			    repo->path_git_dir);
452 			goto done;
453 		}
454 		return NULL;
455 	}
456 
457 	/* git repository with working tree? */
458 	free(repo->path_git_dir);
459 	repo->path_git_dir = NULL;
460 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
461 		err = got_error_from_errno("asprintf");
462 		goto done;
463 	}
464 	if (is_git_repo(repo)) {
465 		repo->path = strdup(path);
466 		if (repo->path == NULL) {
467 			err = got_error_from_errno("strdup");
468 			goto done;
469 		}
470 		repo->gitdir_fd = open(repo->path_git_dir,
471 		    O_DIRECTORY | O_CLOEXEC);
472 		if (repo->gitdir_fd == -1) {
473 			err = got_error_from_errno2("open",
474 			    repo->path_git_dir);
475 			goto done;
476 		}
477 		return NULL;
478 	}
479 
480 	err = got_error(GOT_ERR_NOT_GIT_REPO);
481 done:
482 	if (err) {
483 		free(repo->path);
484 		repo->path = NULL;
485 		free(repo->path_git_dir);
486 		repo->path_git_dir = NULL;
487 		if (repo->gitdir_fd != -1)
488 			close(repo->gitdir_fd);
489 		repo->gitdir_fd = -1;
490 
491 	}
492 	return err;
493 }
494 
495 static const struct got_error *
496 parse_gitconfig_file(int *gitconfig_repository_format_version,
497     char **gitconfig_author_name, char **gitconfig_author_email,
498     struct got_remote_repo **remotes, int *nremotes,
499     char **gitconfig_owner, char ***extensions, int *nextensions,
500     const char *gitconfig_path)
501 {
502 	const struct got_error *err = NULL, *child_err = NULL;
503 	int fd = -1;
504 	int imsg_fds[2] = { -1, -1 };
505 	pid_t pid;
506 	struct imsgbuf *ibuf;
507 
508 	*gitconfig_repository_format_version = 0;
509 	if (extensions)
510 		*extensions = NULL;
511 	if (nextensions)
512 		*nextensions = 0;
513 	*gitconfig_author_name = NULL;
514 	*gitconfig_author_email = NULL;
515 	if (remotes)
516 		*remotes = NULL;
517 	if (nremotes)
518 		*nremotes = 0;
519 	if (gitconfig_owner)
520 		*gitconfig_owner = NULL;
521 
522 	fd = open(gitconfig_path, O_RDONLY | O_CLOEXEC);
523 	if (fd == -1) {
524 		if (errno == ENOENT)
525 			return NULL;
526 		return got_error_from_errno2("open", gitconfig_path);
527 	}
528 
529 	ibuf = calloc(1, sizeof(*ibuf));
530 	if (ibuf == NULL) {
531 		err = got_error_from_errno("calloc");
532 		goto done;
533 	}
534 
535 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
536 		err = got_error_from_errno("socketpair");
537 		goto done;
538 	}
539 
540 	pid = fork();
541 	if (pid == -1) {
542 		err = got_error_from_errno("fork");
543 		goto done;
544 	} else if (pid == 0) {
545 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_GITCONFIG,
546 		    gitconfig_path);
547 		/* not reached */
548 	}
549 
550 	if (close(imsg_fds[1]) == -1) {
551 		err = got_error_from_errno("close");
552 		goto done;
553 	}
554 	imsg_fds[1] = -1;
555 	imsg_init(ibuf, imsg_fds[0]);
556 
557 	err = got_privsep_send_gitconfig_parse_req(ibuf, fd);
558 	if (err)
559 		goto done;
560 	fd = -1;
561 
562 	err = got_privsep_send_gitconfig_repository_format_version_req(ibuf);
563 	if (err)
564 		goto done;
565 
566 	err = got_privsep_recv_gitconfig_int(
567 	    gitconfig_repository_format_version, ibuf);
568 	if (err)
569 		goto done;
570 
571 	if (extensions && nextensions) {
572 		err = got_privsep_send_gitconfig_repository_extensions_req(
573 		    ibuf);
574 		if (err)
575 			goto done;
576 		err = got_privsep_recv_gitconfig_int(nextensions, ibuf);
577 		if (err)
578 			goto done;
579 		if (*nextensions > 0) {
580 			int i;
581 			*extensions = calloc(*nextensions, sizeof(char *));
582 			if (*extensions == NULL) {
583 				err = got_error_from_errno("calloc");
584 				goto done;
585 			}
586 			for (i = 0; i < *nextensions; i++) {
587 				char *ext;
588 				err = got_privsep_recv_gitconfig_str(&ext,
589 				    ibuf);
590 				if (err)
591 					goto done;
592 				(*extensions)[i] = ext;
593 			}
594 		}
595 	}
596 
597 	err = got_privsep_send_gitconfig_author_name_req(ibuf);
598 	if (err)
599 		goto done;
600 
601 	err = got_privsep_recv_gitconfig_str(gitconfig_author_name, ibuf);
602 	if (err)
603 		goto done;
604 
605 	err = got_privsep_send_gitconfig_author_email_req(ibuf);
606 	if (err)
607 		goto done;
608 
609 	err = got_privsep_recv_gitconfig_str(gitconfig_author_email, ibuf);
610 	if (err)
611 		goto done;
612 
613 	if (remotes && nremotes) {
614 		err = got_privsep_send_gitconfig_remotes_req(ibuf);
615 		if (err)
616 			goto done;
617 
618 		err = got_privsep_recv_gitconfig_remotes(remotes,
619 		    nremotes, ibuf);
620 		if (err)
621 			goto done;
622 	}
623 
624 	if (gitconfig_owner) {
625 		err = got_privsep_send_gitconfig_owner_req(ibuf);
626 		if (err)
627 			goto done;
628 		err = got_privsep_recv_gitconfig_str(gitconfig_owner, ibuf);
629 		if (err)
630 			goto done;
631 	}
632 
633 	err = got_privsep_send_stop(imsg_fds[0]);
634 	child_err = got_privsep_wait_for_child(pid);
635 	if (child_err && err == NULL)
636 		err = child_err;
637 done:
638 	if (imsg_fds[0] != -1 && close(imsg_fds[0]) == -1 && err == NULL)
639 		err = got_error_from_errno("close");
640 	if (imsg_fds[1] != -1 && close(imsg_fds[1]) == -1 && err == NULL)
641 		err = got_error_from_errno("close");
642 	if (fd != -1 && close(fd) == -1 && err == NULL)
643 		err = got_error_from_errno2("close", gitconfig_path);
644 	free(ibuf);
645 	return err;
646 }
647 
648 static const struct got_error *
649 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
650 {
651 	const struct got_error *err = NULL;
652 	char *repo_gitconfig_path = NULL;
653 
654 	if (global_gitconfig_path) {
655 		/* Read settings from ~/.gitconfig. */
656 		int dummy_repo_version;
657 		err = parse_gitconfig_file(&dummy_repo_version,
658 		    &repo->global_gitconfig_author_name,
659 		    &repo->global_gitconfig_author_email,
660 		    NULL, NULL, NULL, NULL, NULL, global_gitconfig_path);
661 		if (err)
662 			return err;
663 	}
664 
665 	/* Read repository's .git/config file. */
666 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
667 	if (repo_gitconfig_path == NULL)
668 		return got_error_from_errno("got_repo_get_path_gitconfig");
669 
670 	err = parse_gitconfig_file(&repo->gitconfig_repository_format_version,
671 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
672 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
673 	    &repo->gitconfig_owner, &repo->extensions, &repo->nextensions,
674 	    repo_gitconfig_path);
675 	if (err)
676 		goto done;
677 
678 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
679 		int i;
680 
681 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
682 			got_repo_free_remote_repo_data(
683 			    &repo->gitconfig_remotes[i]);
684 		}
685 		free(repo->gitconfig_remotes);
686 		repo->gitconfig_remotes = NULL;
687 		repo->ngitconfig_remotes = 0;
688 
689 		free(repo->gitconfig_author_name);
690 		repo->gitconfig_author_name = NULL;
691 		free(repo->gitconfig_author_email);
692 		repo->gitconfig_author_email = NULL;
693 
694 		free(repo->global_gitconfig_author_name);
695 		repo->global_gitconfig_author_name = NULL;
696 		free(repo->global_gitconfig_author_email);
697 		repo->global_gitconfig_author_email = NULL;
698 	}
699 
700 done:
701 	free(repo_gitconfig_path);
702 	return err;
703 }
704 
705 static const struct got_error *
706 read_gotconfig(struct got_repository *repo)
707 {
708 	const struct got_error *err = NULL;
709 	char *gotconfig_path;
710 
711 	gotconfig_path = got_repo_get_path_gotconfig(repo);
712 	if (gotconfig_path == NULL)
713 		return got_error_from_errno("got_repo_get_path_gotconfig");
714 
715 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
716 	free(gotconfig_path);
717 	return err;
718 }
719 
720 /* Supported repository format extensions. */
721 static const char *const repo_extensions[] = {
722 	"noop",			/* Got supports repository format version 1. */
723 	"preciousObjects",	/* Supported by gotadmin cleanup. */
724 	"worktreeConfig",	/* Got does not care about Git work trees. */
725 };
726 
727 const struct got_error *
728 got_repo_open(struct got_repository **repop, const char *path,
729     const char *global_gitconfig_path, int *pack_fds)
730 {
731 	struct got_repository *repo = NULL;
732 	const struct got_error *err = NULL;
733 	char *repo_path = NULL;
734 	size_t i, j = 0;
735 	struct rlimit rl;
736 
737 	*repop = NULL;
738 
739 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
740 		return got_error_from_errno("getrlimit");
741 
742 	repo = calloc(1, sizeof(*repo));
743 	if (repo == NULL)
744 		return got_error_from_errno("calloc");
745 
746 	RB_INIT(&repo->packidx_bloom_filters);
747 	TAILQ_INIT(&repo->packidx_paths);
748 
749 	for (i = 0; i < nitems(repo->privsep_children); i++) {
750 		memset(&repo->privsep_children[i], 0,
751 		    sizeof(repo->privsep_children[0]));
752 		repo->privsep_children[i].imsg_fd = -1;
753 	}
754 
755 	err = got_object_cache_init(&repo->objcache,
756 	    GOT_OBJECT_CACHE_TYPE_OBJ);
757 	if (err)
758 		goto done;
759 	err = got_object_cache_init(&repo->treecache,
760 	    GOT_OBJECT_CACHE_TYPE_TREE);
761 	if (err)
762 		goto done;
763 	err = got_object_cache_init(&repo->commitcache,
764 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
765 	if (err)
766 		goto done;
767 	err = got_object_cache_init(&repo->tagcache,
768 	    GOT_OBJECT_CACHE_TYPE_TAG);
769 	if (err)
770 		goto done;
771 	err = got_object_cache_init(&repo->rawcache,
772 	    GOT_OBJECT_CACHE_TYPE_RAW);
773 	if (err)
774 		goto done;
775 
776 	repo->pack_cache_size = GOT_PACK_CACHE_SIZE;
777 	if (repo->pack_cache_size > rl.rlim_cur / 8)
778 		repo->pack_cache_size = rl.rlim_cur / 8;
779 	for (i = 0; i < nitems(repo->packs); i++) {
780 		if (i < repo->pack_cache_size) {
781 			repo->packs[i].basefd = pack_fds[j++];
782 			repo->packs[i].accumfd = pack_fds[j++];
783 		} else {
784 			repo->packs[i].basefd = -1;
785 			repo->packs[i].accumfd = -1;
786 		}
787 	}
788 	repo->pinned_pack = -1;
789 	repo->pinned_packidx = -1;
790 	repo->pinned_pid = 0;
791 
792 	repo_path = realpath(path, NULL);
793 	if (repo_path == NULL) {
794 		err = got_error_from_errno2("realpath", path);
795 		goto done;
796 	}
797 
798 	for (;;) {
799 		char *parent_path;
800 
801 		err = open_repo(repo, repo_path);
802 		if (err == NULL)
803 			break;
804 		if (err->code != GOT_ERR_NOT_GIT_REPO)
805 			goto done;
806 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
807 			err = got_error(GOT_ERR_NOT_GIT_REPO);
808 			goto done;
809 		}
810 		err = got_path_dirname(&parent_path, repo_path);
811 		if (err)
812 			goto done;
813 		free(repo_path);
814 		repo_path = parent_path;
815 	}
816 
817 	err = read_gotconfig(repo);
818 	if (err)
819 		goto done;
820 
821 	err = read_gitconfig(repo, global_gitconfig_path);
822 	if (err)
823 		goto done;
824 	if (repo->gitconfig_repository_format_version != 0) {
825 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
826 		goto done;
827 	}
828 	for (i = 0; i < repo->nextensions; i++) {
829 		char *ext = repo->extensions[i];
830 		int j, supported = 0;
831 		for (j = 0; j < nitems(repo_extensions); j++) {
832 			if (strcmp(ext, repo_extensions[j]) == 0) {
833 				supported = 1;
834 				break;
835 			}
836 		}
837 		if (!supported) {
838 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
839 			goto done;
840 		}
841 	}
842 
843 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
844 done:
845 	if (err)
846 		got_repo_close(repo);
847 	else
848 		*repop = repo;
849 	free(repo_path);
850 	return err;
851 }
852 
853 const struct got_error *
854 got_repo_close(struct got_repository *repo)
855 {
856 	const struct got_error *err = NULL, *child_err;
857 	struct got_packidx_bloom_filter *bf;
858 	struct got_pathlist_entry *pe;
859 	size_t i;
860 
861 	for (i = 0; i < repo->pack_cache_size; i++) {
862 		if (repo->packidx_cache[i] == NULL)
863 			break;
864 		got_packidx_close(repo->packidx_cache[i]);
865 	}
866 
867 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
868 	    &repo->packidx_bloom_filters))) {
869 		RB_REMOVE(got_packidx_bloom_filter_tree,
870 		    &repo->packidx_bloom_filters, bf);
871 		bloom_free(bf->bloom);
872 		free(bf->bloom);
873 		free(bf);
874 	}
875 
876 	for (i = 0; i < repo->pack_cache_size; i++)
877 		if (repo->packs[i].path_packfile)
878 			if (repo->packs[i].path_packfile)
879 				got_pack_close(&repo->packs[i]);
880 
881 	free(repo->path);
882 	free(repo->path_git_dir);
883 
884 	got_object_cache_close(&repo->objcache);
885 	got_object_cache_close(&repo->treecache);
886 	got_object_cache_close(&repo->commitcache);
887 	got_object_cache_close(&repo->tagcache);
888 	got_object_cache_close(&repo->rawcache);
889 
890 	for (i = 0; i < nitems(repo->privsep_children); i++) {
891 		if (repo->privsep_children[i].imsg_fd == -1)
892 			continue;
893 		imsg_clear(repo->privsep_children[i].ibuf);
894 		free(repo->privsep_children[i].ibuf);
895 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
896 		child_err = got_privsep_wait_for_child(
897 		    repo->privsep_children[i].pid);
898 		if (child_err && err == NULL)
899 			err = child_err;
900 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
901 		    err == NULL)
902 			err = got_error_from_errno("close");
903 	}
904 
905 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
906 	    err == NULL)
907 		err = got_error_from_errno("close");
908 
909 	if (repo->gotconfig)
910 		got_gotconfig_free(repo->gotconfig);
911 	free(repo->gitconfig_author_name);
912 	free(repo->gitconfig_author_email);
913 	for (i = 0; i < repo->ngitconfig_remotes; i++)
914 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
915 	free(repo->gitconfig_remotes);
916 	for (i = 0; i < repo->nextensions; i++)
917 		free(repo->extensions[i]);
918 	free(repo->extensions);
919 
920 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry)
921 		free((void *)pe->path);
922 	got_pathlist_free(&repo->packidx_paths);
923 	free(repo);
924 
925 	return err;
926 }
927 
928 void
929 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
930 {
931 	int i;
932 
933 	free(repo->name);
934 	repo->name = NULL;
935 	free(repo->fetch_url);
936 	repo->fetch_url = NULL;
937 	free(repo->send_url);
938 	repo->send_url = NULL;
939 	for (i = 0; i < repo->nfetch_branches; i++)
940 		free(repo->fetch_branches[i]);
941 	free(repo->fetch_branches);
942 	repo->fetch_branches = NULL;
943 	repo->nfetch_branches = 0;
944 	for (i = 0; i < repo->nsend_branches; i++)
945 		free(repo->send_branches[i]);
946 	free(repo->send_branches);
947 	repo->send_branches = NULL;
948 	repo->nsend_branches = 0;
949 }
950 
951 const struct got_error *
952 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
953     const char *input_path)
954 {
955 	const struct got_error *err = NULL;
956 	const char *repo_abspath = NULL;
957 	size_t repolen, len;
958 	char *canonpath, *path = NULL;
959 
960 	*in_repo_path = NULL;
961 
962 	canonpath = strdup(input_path);
963 	if (canonpath == NULL) {
964 		err = got_error_from_errno("strdup");
965 		goto done;
966 	}
967 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
968 	if (err)
969 		goto done;
970 
971 	repo_abspath = got_repo_get_path(repo);
972 
973 	if (canonpath[0] == '\0') {
974 		path = strdup(canonpath);
975 		if (path == NULL) {
976 			err = got_error_from_errno("strdup");
977 			goto done;
978 		}
979 	} else {
980 		path = realpath(canonpath, NULL);
981 		if (path == NULL) {
982 			if (errno != ENOENT) {
983 				err = got_error_from_errno2("realpath",
984 				    canonpath);
985 				goto done;
986 			}
987 			/*
988 			 * Path is not on disk.
989 			 * Assume it is already relative to repository root.
990 			 */
991 			path = strdup(canonpath);
992 			if (path == NULL) {
993 				err = got_error_from_errno("strdup");
994 				goto done;
995 			}
996 		}
997 
998 		repolen = strlen(repo_abspath);
999 		len = strlen(path);
1000 
1001 
1002 		if (strcmp(path, repo_abspath) == 0) {
1003 			free(path);
1004 			path = strdup("");
1005 			if (path == NULL) {
1006 				err = got_error_from_errno("strdup");
1007 				goto done;
1008 			}
1009 		} else if (len > repolen &&
1010 		    got_path_is_child(path, repo_abspath, repolen)) {
1011 			/* Matched an on-disk path inside repository. */
1012 			if (got_repo_is_bare(repo)) {
1013 				/*
1014 				 * Matched an on-disk path inside repository
1015 				 * database. Treat input as repository-relative.
1016 				 */
1017 				free(path);
1018 				path = canonpath;
1019 				canonpath = NULL;
1020 			} else {
1021 				char *child;
1022 				/* Strip common prefix with repository path. */
1023 				err = got_path_skip_common_ancestor(&child,
1024 				    repo_abspath, path);
1025 				if (err)
1026 					goto done;
1027 				free(path);
1028 				path = child;
1029 			}
1030 		} else {
1031 			/*
1032 			 * Matched unrelated on-disk path.
1033 			 * Treat input as repository-relative.
1034 			 */
1035 			free(path);
1036 			path = canonpath;
1037 			canonpath = NULL;
1038 		}
1039 	}
1040 
1041 	/* Make in-repository path absolute */
1042 	if (path[0] != '/') {
1043 		char *abspath;
1044 		if (asprintf(&abspath, "/%s", path) == -1) {
1045 			err = got_error_from_errno("asprintf");
1046 			goto done;
1047 		}
1048 		free(path);
1049 		path = abspath;
1050 	}
1051 
1052 done:
1053 	free(canonpath);
1054 	if (err)
1055 		free(path);
1056 	else
1057 		*in_repo_path = path;
1058 	return err;
1059 }
1060 
1061 static const struct got_error *
1062 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1063     const char *path_packidx)
1064 {
1065 	const struct got_error *err = NULL;
1066 	size_t i;
1067 
1068 	for (i = 0; i < repo->pack_cache_size; i++) {
1069 		if (repo->packidx_cache[i] == NULL)
1070 			break;
1071 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1072 		    path_packidx) == 0) {
1073 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1074 		}
1075 	}
1076 	if (i == repo->pack_cache_size) {
1077 		do {
1078 			i--;
1079 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1080 		    i == repo->pinned_packidx);
1081 		err = got_packidx_close(repo->packidx_cache[i]);
1082 		if (err)
1083 			return err;
1084 	}
1085 
1086 	repo->packidx_cache[i] = packidx;
1087 
1088 	return NULL;
1089 }
1090 
1091 int
1092 got_repo_is_packidx_filename(const char *name, size_t len)
1093 {
1094 	if (len != GOT_PACKIDX_NAMELEN)
1095 		return 0;
1096 
1097 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1098 		return 0;
1099 
1100 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1101 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1102 		return 0;
1103 
1104 	return 1;
1105 }
1106 
1107 static struct got_packidx_bloom_filter *
1108 get_packidx_bloom_filter(struct got_repository *repo,
1109     const char *path, size_t path_len)
1110 {
1111 	struct got_packidx_bloom_filter key;
1112 
1113 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1114 		return NULL; /* XXX */
1115 	key.path_len = path_len;
1116 
1117 	return RB_FIND(got_packidx_bloom_filter_tree,
1118 	    &repo->packidx_bloom_filters, &key);
1119 }
1120 
1121 int
1122 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1123     const char *path_packidx, struct got_object_id *id)
1124 {
1125 	struct got_packidx_bloom_filter *bf;
1126 
1127 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1128 	if (bf)
1129 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1130 
1131 	/* No bloom filter means this pack index must be searched. */
1132 	return 1;
1133 }
1134 
1135 static const struct got_error *
1136 add_packidx_bloom_filter(struct got_repository *repo,
1137     struct got_packidx *packidx, const char *path_packidx)
1138 {
1139 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1140 	struct got_packidx_bloom_filter *bf;
1141 	size_t len;
1142 
1143 	/*
1144 	 * Don't use bloom filters for very large pack index files.
1145 	 * Large pack files will contain a relatively large fraction
1146 	 * of our objects so we will likely need to visit them anyway.
1147 	 * The more objects a pack file contains the higher the probability
1148 	 * of a false-positive match from the bloom filter. And reading
1149 	 * all object IDs from a large pack index file can be expensive.
1150 	 */
1151 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1152 		return NULL;
1153 
1154 	/* Do we already have a filter for this pack index? */
1155 	if (get_packidx_bloom_filter(repo, path_packidx,
1156 	    strlen(path_packidx)) != NULL)
1157 		return NULL;
1158 
1159 	bf = calloc(1, sizeof(*bf));
1160 	if (bf == NULL)
1161 		return got_error_from_errno("calloc");
1162 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1163 	if (bf->bloom == NULL) {
1164 		free(bf);
1165 		return got_error_from_errno("calloc");
1166 	}
1167 
1168 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1169 	if (len >= sizeof(bf->path)) {
1170 		free(bf->bloom);
1171 		free(bf);
1172 		return got_error(GOT_ERR_NO_SPACE);
1173 	}
1174 	bf->path_len = len;
1175 
1176 	/* Minimum size supported by our bloom filter is 1000 entries. */
1177 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1178 	for (i = 0; i < nobjects; i++) {
1179 		struct got_packidx_object_id *id;
1180 		id = &packidx->hdr.sorted_ids[i];
1181 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1182 	}
1183 
1184 	RB_INSERT(got_packidx_bloom_filter_tree,
1185 	    &repo->packidx_bloom_filters, bf);
1186 	return NULL;
1187 }
1188 
1189 const struct got_error *
1190 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1191     struct got_repository *repo, struct got_object_id *id)
1192 {
1193 	const struct got_error *err;
1194 	struct got_pathlist_entry *pe;
1195 	size_t i;
1196 
1197 	/* Search pack index cache. */
1198 	for (i = 0; i < repo->pack_cache_size; i++) {
1199 		if (repo->packidx_cache[i] == NULL)
1200 			break;
1201 		if (!got_repo_check_packidx_bloom_filter(repo,
1202 		    repo->packidx_cache[i]->path_packidx, id))
1203 			continue; /* object will not be found in this index */
1204 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1205 		if (*idx != -1) {
1206 			*packidx = repo->packidx_cache[i];
1207 			/*
1208 			 * Move this cache entry to the front. Repeatedly
1209 			 * searching a wrong pack index can be expensive.
1210 			 */
1211 			if (i > 0) {
1212 				memmove(&repo->packidx_cache[1],
1213 				    &repo->packidx_cache[0],
1214 				    i * sizeof(repo->packidx_cache[0]));
1215 				repo->packidx_cache[0] = *packidx;
1216 				if (repo->pinned_packidx >= 0 &&
1217 				    repo->pinned_packidx < i)
1218 					repo->pinned_packidx++;
1219 				else if (repo->pinned_packidx == i)
1220 					repo->pinned_packidx = 0;
1221 			}
1222 			return NULL;
1223 		}
1224 	}
1225 	/* No luck. Search the filesystem. */
1226 
1227 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1228 		const char *path_packidx = pe->path;
1229 		int is_cached = 0;
1230 
1231 		if (!got_repo_check_packidx_bloom_filter(repo,
1232 		    pe->path, id))
1233 			continue; /* object will not be found in this index */
1234 
1235 		for (i = 0; i < repo->pack_cache_size; i++) {
1236 			if (repo->packidx_cache[i] == NULL)
1237 				break;
1238 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1239 			    path_packidx) == 0) {
1240 				is_cached = 1;
1241 				break;
1242 			}
1243 		}
1244 		if (is_cached)
1245 			continue; /* already searched */
1246 
1247 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1248 		    path_packidx, 0);
1249 		if (err)
1250 			goto done;
1251 
1252 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1253 		if (err)
1254 			goto done;
1255 
1256 		err = cache_packidx(repo, *packidx, path_packidx);
1257 		if (err)
1258 			goto done;
1259 
1260 		*idx = got_packidx_get_object_idx(*packidx, id);
1261 		if (*idx != -1) {
1262 			err = NULL; /* found the object */
1263 			goto done;
1264 		}
1265 	}
1266 
1267 	err = got_error_no_obj(id);
1268 done:
1269 	return err;
1270 }
1271 
1272 const struct got_error *
1273 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1274     struct got_repository *repo)
1275 {
1276 	const struct got_error *err = NULL;
1277 	DIR *packdir = NULL;
1278 	struct dirent *dent;
1279 	char *path_packidx = NULL;
1280 	int packdir_fd;
1281 	struct stat sb;
1282 
1283 	packdir_fd = openat(got_repo_get_fd(repo),
1284 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1285 	if (packdir_fd == -1) {
1286 		return got_error_from_errno_fmt("openat: %s/%s",
1287 		    got_repo_get_path_git_dir(repo),
1288 		    GOT_OBJECTS_PACK_DIR);
1289 	}
1290 
1291 	packdir = fdopendir(packdir_fd);
1292 	if (packdir == NULL) {
1293 		err = got_error_from_errno("fdopendir");
1294 		goto done;
1295 	}
1296 
1297 	if (fstat(packdir_fd, &sb) == -1) {
1298 		err = got_error_from_errno("fstat");
1299 		goto done;
1300 	}
1301 	repo->pack_path_mtime = sb.st_mtime;
1302 
1303 	while ((dent = readdir(packdir)) != NULL) {
1304 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1305 			continue;
1306 
1307 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1308 		    dent->d_name) == -1) {
1309 			err = got_error_from_errno("asprintf");
1310 			path_packidx = NULL;
1311 			break;
1312 		}
1313 
1314 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1315 		if (err)
1316 			break;
1317 	}
1318 done:
1319 	if (err)
1320 		free(path_packidx);
1321 	if (packdir && closedir(packdir) != 0 && err == NULL)
1322 		err = got_error_from_errno("closedir");
1323 	return err;
1324 }
1325 
1326 const struct got_error *
1327 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1328     struct got_repository *repo)
1329 {
1330 	const struct got_error *err;
1331 	size_t i;
1332 
1333 	*packidx = NULL;
1334 
1335 	/* Search pack index cache. */
1336 	for (i = 0; i < repo->pack_cache_size; i++) {
1337 		if (repo->packidx_cache[i] == NULL)
1338 			break;
1339 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1340 		    path_packidx) == 0) {
1341 			*packidx = repo->packidx_cache[i];
1342 			return NULL;
1343 		}
1344 	}
1345 	/* No luck. Search the filesystem. */
1346 
1347 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1348 	    path_packidx, 0);
1349 	if (err)
1350 		return err;
1351 
1352 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1353 	if (err)
1354 		goto done;
1355 
1356 	err = cache_packidx(repo, *packidx, path_packidx);
1357 done:
1358 	if (err) {
1359 		got_packidx_close(*packidx);
1360 		*packidx = NULL;
1361 	}
1362 	return err;
1363 }
1364 
1365 static const struct got_error *
1366 read_packfile_hdr(int fd, struct got_packidx *packidx)
1367 {
1368 	const struct got_error *err = NULL;
1369 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1370 	struct got_packfile_hdr hdr;
1371 	ssize_t n;
1372 
1373 	n = read(fd, &hdr, sizeof(hdr));
1374 	if (n < 0)
1375 		return got_error_from_errno("read");
1376 	if (n != sizeof(hdr))
1377 		return got_error(GOT_ERR_BAD_PACKFILE);
1378 
1379 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1380 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1381 	    be32toh(hdr.nobjects) != totobj)
1382 		err = got_error(GOT_ERR_BAD_PACKFILE);
1383 
1384 	return err;
1385 }
1386 
1387 static const struct got_error *
1388 open_packfile(int *fd, struct got_repository *repo,
1389     const char *relpath, struct got_packidx *packidx)
1390 {
1391 	const struct got_error *err = NULL;
1392 
1393 	*fd = openat(got_repo_get_fd(repo), relpath,
1394 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1395 	if (*fd == -1)
1396 		return got_error_from_errno_fmt("openat: %s/%s",
1397 		    got_repo_get_path_git_dir(repo), relpath);
1398 
1399 	if (packidx) {
1400 		err = read_packfile_hdr(*fd, packidx);
1401 		if (err) {
1402 			close(*fd);
1403 			*fd = -1;
1404 		}
1405 	}
1406 
1407 	return err;
1408 }
1409 
1410 const struct got_error *
1411 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1412     const char *path_packfile, struct got_packidx *packidx)
1413 {
1414 	const struct got_error *err = NULL;
1415 	struct got_pack *pack = NULL;
1416 	struct stat sb;
1417 	size_t i;
1418 
1419 	if (packp)
1420 		*packp = NULL;
1421 
1422 	for (i = 0; i < repo->pack_cache_size; i++) {
1423 		pack = &repo->packs[i];
1424 		if (pack->path_packfile == NULL)
1425 			break;
1426 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1427 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1428 	}
1429 
1430 	if (i == repo->pack_cache_size) {
1431 		struct got_pack tmp;
1432 		do {
1433 			i--;
1434 		} while (i > 0 && repo->pinned_pack >= 0 &&
1435 		    i == repo->pinned_pack);
1436 		err = got_pack_close(&repo->packs[i]);
1437 		if (err)
1438 			return err;
1439 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1440 			return got_error_from_errno("ftruncate");
1441 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1442 			return got_error_from_errno("ftruncate");
1443 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1444 		memcpy(&repo->packs[i], &repo->packs[0],
1445 		    sizeof(repo->packs[i]));
1446 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1447 		if (repo->pinned_pack == 0)
1448 			repo->pinned_pack = i;
1449 		else if (repo->pinned_pack == i)
1450 			repo->pinned_pack = 0;
1451 		i = 0;
1452 	}
1453 
1454 	pack = &repo->packs[i];
1455 
1456 	pack->path_packfile = strdup(path_packfile);
1457 	if (pack->path_packfile == NULL) {
1458 		err = got_error_from_errno("strdup");
1459 		goto done;
1460 	}
1461 
1462 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1463 	if (err)
1464 		goto done;
1465 
1466 	if (fstat(pack->fd, &sb) != 0) {
1467 		err = got_error_from_errno("fstat");
1468 		goto done;
1469 	}
1470 	pack->filesize = sb.st_size;
1471 
1472 	pack->privsep_child = NULL;
1473 
1474 #ifndef GOT_PACK_NO_MMAP
1475 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1476 	    pack->fd, 0);
1477 	if (pack->map == MAP_FAILED) {
1478 		if (errno != ENOMEM) {
1479 			err = got_error_from_errno("mmap");
1480 			goto done;
1481 		}
1482 		pack->map = NULL; /* fall back to read(2) */
1483 	}
1484 #endif
1485 done:
1486 	if (err) {
1487 		if (pack) {
1488 			free(pack->path_packfile);
1489 			memset(pack, 0, sizeof(*pack));
1490 		}
1491 	} else if (packp)
1492 		*packp = pack;
1493 	return err;
1494 }
1495 
1496 struct got_pack *
1497 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1498 {
1499 	struct got_pack *pack = NULL;
1500 	size_t i;
1501 
1502 	for (i = 0; i < repo->pack_cache_size; i++) {
1503 		pack = &repo->packs[i];
1504 		if (pack->path_packfile == NULL)
1505 			break;
1506 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1507 			return pack;
1508 	}
1509 
1510 	return NULL;
1511 }
1512 
1513 const struct got_error *
1514 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1515     struct got_pack *pack)
1516 {
1517 	size_t i;
1518 	int pinned_pack = -1, pinned_packidx = -1;
1519 
1520 	for (i = 0; i < repo->pack_cache_size; i++) {
1521 		if (repo->packidx_cache[i] &&
1522 		    strcmp(repo->packidx_cache[i]->path_packidx,
1523 		    packidx->path_packidx) == 0)
1524 			pinned_packidx = i;
1525 		if (repo->packs[i].path_packfile &&
1526 		    strcmp(repo->packs[i].path_packfile,
1527 		    pack->path_packfile) == 0)
1528 			pinned_pack = i;
1529 	}
1530 
1531 	if (pinned_packidx == -1 || pinned_pack == -1)
1532 		return got_error(GOT_ERR_PIN_PACK);
1533 
1534 	repo->pinned_pack = pinned_pack;
1535 	repo->pinned_packidx = pinned_packidx;
1536 	repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1537 	return NULL;
1538 }
1539 
1540 struct got_pack *
1541 got_repo_get_pinned_pack(struct got_repository *repo)
1542 {
1543 	if (repo->pinned_pack >= 0 &&
1544 	    repo->pinned_pack < repo->pack_cache_size)
1545 		return &repo->packs[repo->pinned_pack];
1546 
1547 	return NULL;
1548 }
1549 
1550 void
1551 got_repo_unpin_pack(struct got_repository *repo)
1552 {
1553 	repo->pinned_packidx = -1;
1554 	repo->pinned_pack = -1;
1555 	repo->pinned_pid = 0;
1556 }
1557 
1558 const struct got_error *
1559 got_repo_init(const char *repo_path)
1560 {
1561 	const struct got_error *err = NULL;
1562 	const char *dirnames[] = {
1563 		GOT_OBJECTS_DIR,
1564 		GOT_OBJECTS_PACK_DIR,
1565 		GOT_REFS_DIR,
1566 	};
1567 	const char *description_str = "Unnamed repository; "
1568 	    "edit this file 'description' to name the repository.";
1569 	const char *headref_str = "ref: refs/heads/main";
1570 	const char *gitconfig_str = "[core]\n"
1571 	    "\trepositoryformatversion = 0\n"
1572 	    "\tfilemode = true\n"
1573 	    "\tbare = true\n";
1574 	char *path;
1575 	size_t i;
1576 
1577 	if (!got_path_dir_is_empty(repo_path))
1578 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1579 
1580 	for (i = 0; i < nitems(dirnames); i++) {
1581 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1582 			return got_error_from_errno("asprintf");
1583 		}
1584 		err = got_path_mkdir(path);
1585 		free(path);
1586 		if (err)
1587 			return err;
1588 	}
1589 
1590 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1591 		return got_error_from_errno("asprintf");
1592 	err = got_path_create_file(path, description_str);
1593 	free(path);
1594 	if (err)
1595 		return err;
1596 
1597 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1598 		return got_error_from_errno("asprintf");
1599 	err = got_path_create_file(path, headref_str);
1600 	free(path);
1601 	if (err)
1602 		return err;
1603 
1604 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1605 		return got_error_from_errno("asprintf");
1606 	err = got_path_create_file(path, gitconfig_str);
1607 	free(path);
1608 	if (err)
1609 		return err;
1610 
1611 	return NULL;
1612 }
1613 
1614 static void
1615 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1616 {
1617 	struct got_pathlist_entry *pe;
1618 
1619 	while (!TAILQ_EMPTY(packidx_paths)) {
1620 		pe = TAILQ_FIRST(packidx_paths);
1621 		TAILQ_REMOVE(packidx_paths, pe, entry);
1622 		free((char *)pe->path);
1623 		free(pe);
1624 	}
1625 }
1626 
1627 static const struct got_error *
1628 match_packed_object(struct got_object_id **unique_id,
1629     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1630 {
1631 	const struct got_error *err = NULL;
1632 	struct got_object_id_queue matched_ids;
1633 	struct got_pathlist_entry *pe;
1634 	struct stat sb;
1635 
1636 	STAILQ_INIT(&matched_ids);
1637 
1638 	if (stat(got_repo_get_path_objects_pack(repo), &sb) == -1) {
1639 		if (errno != ENOENT)
1640 			return got_error_from_errno2("stat",
1641 			    got_repo_get_path_objects_pack(repo));
1642 	} else if (sb.st_mtime != repo->pack_path_mtime) {
1643 		purge_packidx_paths(&repo->packidx_paths);
1644 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1645 		if (err)
1646 			return err;
1647 	}
1648 
1649 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1650 		const char *path_packidx = pe->path;
1651 		struct got_packidx *packidx;
1652 		struct got_object_qid *qid;
1653 
1654 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1655 		    path_packidx, 0);
1656 		if (err)
1657 			break;
1658 
1659 		err = got_packidx_match_id_str_prefix(&matched_ids,
1660 		    packidx, id_str_prefix);
1661 		if (err) {
1662 			got_packidx_close(packidx);
1663 			break;
1664 		}
1665 		err = got_packidx_close(packidx);
1666 		if (err)
1667 			break;
1668 
1669 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1670 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1671 				int matched_type;
1672 				err = got_object_get_type(&matched_type, repo,
1673 				    &qid->id);
1674 				if (err)
1675 					goto done;
1676 				if (matched_type != obj_type)
1677 					continue;
1678 			}
1679 			if (*unique_id == NULL) {
1680 				*unique_id = got_object_id_dup(&qid->id);
1681 				if (*unique_id == NULL) {
1682 					err = got_error_from_errno("malloc");
1683 					goto done;
1684 				}
1685 			} else {
1686 				if (got_object_id_cmp(*unique_id,
1687 				    &qid->id) == 0)
1688 					continue; /* packed multiple times */
1689 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1690 				goto done;
1691 			}
1692 		}
1693 	}
1694 done:
1695 	got_object_id_queue_free(&matched_ids);
1696 	if (err) {
1697 		free(*unique_id);
1698 		*unique_id = NULL;
1699 	}
1700 	return err;
1701 }
1702 
1703 static const struct got_error *
1704 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1705     const char *object_dir, const char *id_str_prefix, int obj_type,
1706     struct got_repository *repo)
1707 {
1708 	const struct got_error *err = NULL;
1709 	char *path, *id_str = NULL;
1710 	DIR *dir = NULL;
1711 	struct dirent *dent;
1712 	struct got_object_id id;
1713 
1714 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1715 		err = got_error_from_errno("asprintf");
1716 		goto done;
1717 	}
1718 
1719 	dir = opendir(path);
1720 	if (dir == NULL) {
1721 		if (errno == ENOENT) {
1722 			err = NULL;
1723 			goto done;
1724 		}
1725 		err = got_error_from_errno2("opendir", path);
1726 		goto done;
1727 	}
1728 	while ((dent = readdir(dir)) != NULL) {
1729 		int cmp;
1730 
1731 		free(id_str);
1732 		id_str = NULL;
1733 
1734 		if (strcmp(dent->d_name, ".") == 0 ||
1735 		    strcmp(dent->d_name, "..") == 0)
1736 			continue;
1737 
1738 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1739 			err = got_error_from_errno("asprintf");
1740 			goto done;
1741 		}
1742 
1743 		if (!got_parse_sha1_digest(id.sha1, id_str))
1744 			continue;
1745 
1746 		/*
1747 		 * Directory entries do not necessarily appear in
1748 		 * sorted order, so we must iterate over all of them.
1749 		 */
1750 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1751 		if (cmp != 0)
1752 			continue;
1753 
1754 		if (*unique_id == NULL) {
1755 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1756 				int matched_type;
1757 				err = got_object_get_type(&matched_type, repo,
1758 				    &id);
1759 				if (err)
1760 					goto done;
1761 				if (matched_type != obj_type)
1762 					continue;
1763 			}
1764 			*unique_id = got_object_id_dup(&id);
1765 			if (*unique_id == NULL) {
1766 				err = got_error_from_errno("got_object_id_dup");
1767 				goto done;
1768 			}
1769 		} else {
1770 			if (got_object_id_cmp(*unique_id, &id) == 0)
1771 				continue; /* both packed and loose */
1772 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1773 			goto done;
1774 		}
1775 	}
1776 done:
1777 	if (dir && closedir(dir) != 0 && err == NULL)
1778 		err = got_error_from_errno("closedir");
1779 	if (err) {
1780 		free(*unique_id);
1781 		*unique_id = NULL;
1782 	}
1783 	free(id_str);
1784 	free(path);
1785 	return err;
1786 }
1787 
1788 const struct got_error *
1789 got_repo_match_object_id_prefix(struct got_object_id **id,
1790     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1791 {
1792 	const struct got_error *err = NULL;
1793 	char *path_objects = got_repo_get_path_objects(repo);
1794 	char *object_dir = NULL;
1795 	size_t len;
1796 	int i;
1797 
1798 	*id = NULL;
1799 
1800 	len = strlen(id_str_prefix);
1801 	if (len > SHA1_DIGEST_STRING_LENGTH - 1)
1802 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1803 
1804 	for (i = 0; i < len; i++) {
1805 		if (isxdigit((unsigned char)id_str_prefix[i]))
1806 			continue;
1807 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1808 	}
1809 
1810 	if (len >= 2) {
1811 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1812 		if (err)
1813 			goto done;
1814 		object_dir = strndup(id_str_prefix, 2);
1815 		if (object_dir == NULL) {
1816 			err = got_error_from_errno("strdup");
1817 			goto done;
1818 		}
1819 		err = match_loose_object(id, path_objects, object_dir,
1820 		    id_str_prefix, obj_type, repo);
1821 	} else if (len == 1) {
1822 		int i;
1823 		for (i = 0; i < 0xf; i++) {
1824 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1825 			    == -1) {
1826 				err = got_error_from_errno("asprintf");
1827 				goto done;
1828 			}
1829 			err = match_packed_object(id, repo, object_dir,
1830 			    obj_type);
1831 			if (err)
1832 				goto done;
1833 			err = match_loose_object(id, path_objects, object_dir,
1834 			    id_str_prefix, obj_type, repo);
1835 			if (err)
1836 				goto done;
1837 		}
1838 	} else {
1839 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1840 		goto done;
1841 	}
1842 done:
1843 	free(object_dir);
1844 	if (err) {
1845 		free(*id);
1846 		*id = NULL;
1847 	} else if (*id == NULL) {
1848 		switch (obj_type) {
1849 		case GOT_OBJ_TYPE_BLOB:
1850 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1851 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1852 			break;
1853 		case GOT_OBJ_TYPE_TREE:
1854 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1855 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1856 			break;
1857 		case GOT_OBJ_TYPE_COMMIT:
1858 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1859 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1860 			break;
1861 		case GOT_OBJ_TYPE_TAG:
1862 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1863 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1864 			break;
1865 		default:
1866 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1867 			break;
1868 		}
1869 	}
1870 
1871 	return err;
1872 }
1873 
1874 const struct got_error *
1875 got_repo_match_object_id(struct got_object_id **id, char **label,
1876     const char *id_str, int obj_type, struct got_reflist_head *refs,
1877     struct got_repository *repo)
1878 {
1879 	const struct got_error *err;
1880 	struct got_tag_object *tag;
1881 	struct got_reference *ref = NULL;
1882 
1883 	*id = NULL;
1884 	if (label)
1885 		*label = NULL;
1886 
1887 	if (refs) {
1888 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1889 		    refs, repo);
1890 		if (err == NULL) {
1891 			*id = got_object_id_dup(
1892 			    got_object_tag_get_object_id(tag));
1893 			if (*id == NULL)
1894 				err = got_error_from_errno("got_object_id_dup");
1895 			else if (label && asprintf(label, "refs/tags/%s",
1896 			    got_object_tag_get_name(tag)) == -1) {
1897 				err = got_error_from_errno("asprintf");
1898 				free(*id);
1899 				*id = NULL;
1900 			}
1901 			got_object_tag_close(tag);
1902 			return err;
1903 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1904 		    err->code != GOT_ERR_NO_OBJ)
1905 			return err;
1906 	}
1907 
1908 	err = got_ref_open(&ref, repo, id_str, 0);
1909 	if (err == NULL) {
1910 		err = got_ref_resolve(id, repo, ref);
1911 		if (err)
1912 			goto done;
1913 		if (label) {
1914 			*label = strdup(got_ref_get_name(ref));
1915 			if (*label == NULL) {
1916 				err = got_error_from_errno("strdup");
1917 				goto done;
1918 			}
1919 		}
1920 	} else {
1921 		if (err->code != GOT_ERR_NOT_REF &&
1922 		    err->code != GOT_ERR_BAD_REF_NAME)
1923 			goto done;
1924 		err = got_repo_match_object_id_prefix(id, id_str,
1925 		    obj_type, repo);
1926 		if (err) {
1927 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1928 				err = got_error_not_ref(id_str);
1929 			goto done;
1930 		}
1931 		if (label) {
1932 			err = got_object_id_str(label, *id);
1933 			if (*label == NULL) {
1934 				err = got_error_from_errno("strdup");
1935 				goto done;
1936 			}
1937 		}
1938 	}
1939 done:
1940 	if (ref)
1941 		got_ref_close(ref);
1942 	return err;
1943 }
1944 
1945 const struct got_error *
1946 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1947     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1948 {
1949 	const struct got_error *err = NULL;
1950 	struct got_reflist_entry *re;
1951 	struct got_object_id *tag_id;
1952 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1953 
1954 	*tag = NULL;
1955 
1956 	TAILQ_FOREACH(re, refs, entry) {
1957 		const char *refname;
1958 		refname = got_ref_get_name(re->ref);
1959 		if (got_ref_is_symbolic(re->ref))
1960 			continue;
1961 		if (strncmp(refname, "refs/tags/", 10) != 0)
1962 			continue;
1963 		if (!name_is_absolute)
1964 			refname += strlen("refs/tags/");
1965 		if (strcmp(refname, name) != 0)
1966 			continue;
1967 		err = got_ref_resolve(&tag_id, repo, re->ref);
1968 		if (err)
1969 			break;
1970 		err = got_object_open_as_tag(tag, repo, tag_id);
1971 		free(tag_id);
1972 		if (err)
1973 			break;
1974 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1975 		    got_object_tag_get_object_type(*tag) == obj_type)
1976 			break;
1977 		got_object_tag_close(*tag);
1978 		*tag = NULL;
1979 	}
1980 
1981 	if (err == NULL && *tag == NULL)
1982 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1983 		    GOT_OBJ_LABEL_TAG, name);
1984 	return err;
1985 }
1986 
1987 static const struct got_error *
1988 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1989     const char *name, mode_t mode, struct got_object_id *blob_id)
1990 {
1991 	const struct got_error *err = NULL;
1992 
1993 	 *new_te = NULL;
1994 
1995 	*new_te = calloc(1, sizeof(**new_te));
1996 	if (*new_te == NULL)
1997 		return got_error_from_errno("calloc");
1998 
1999 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2000 	    sizeof((*new_te)->name)) {
2001 		err = got_error(GOT_ERR_NO_SPACE);
2002 		goto done;
2003 	}
2004 
2005 	if (S_ISLNK(mode)) {
2006 		(*new_te)->mode = S_IFLNK;
2007 	} else {
2008 		(*new_te)->mode = S_IFREG;
2009 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2010 	}
2011 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2012 done:
2013 	if (err && *new_te) {
2014 		free(*new_te);
2015 		*new_te = NULL;
2016 	}
2017 	return err;
2018 }
2019 
2020 static const struct got_error *
2021 import_file(struct got_tree_entry **new_te, struct dirent *de,
2022     const char *path, struct got_repository *repo)
2023 {
2024 	const struct got_error *err;
2025 	struct got_object_id *blob_id = NULL;
2026 	char *filepath;
2027 	struct stat sb;
2028 
2029 	if (asprintf(&filepath, "%s%s%s", path,
2030 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2031 		return got_error_from_errno("asprintf");
2032 
2033 	if (lstat(filepath, &sb) != 0) {
2034 		err = got_error_from_errno2("lstat", path);
2035 		goto done;
2036 	}
2037 
2038 	err = got_object_blob_create(&blob_id, filepath, repo);
2039 	if (err)
2040 		goto done;
2041 
2042 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2043 	    blob_id);
2044 done:
2045 	free(filepath);
2046 	if (err)
2047 		free(blob_id);
2048 	return err;
2049 }
2050 
2051 static const struct got_error *
2052 insert_tree_entry(struct got_tree_entry *new_te,
2053     struct got_pathlist_head *paths)
2054 {
2055 	const struct got_error *err = NULL;
2056 	struct got_pathlist_entry *new_pe;
2057 
2058 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2059 	if (err)
2060 		return err;
2061 	if (new_pe == NULL)
2062 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2063 	return NULL;
2064 }
2065 
2066 static const struct got_error *write_tree(struct got_object_id **,
2067     const char *, struct got_pathlist_head *, struct got_repository *,
2068     got_repo_import_cb progress_cb, void *progress_arg);
2069 
2070 static const struct got_error *
2071 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2072     const char *path, struct got_pathlist_head *ignores,
2073     struct got_repository *repo,
2074     got_repo_import_cb progress_cb, void *progress_arg)
2075 {
2076 	const struct got_error *err;
2077 	struct got_object_id *id = NULL;
2078 	char *subdirpath;
2079 
2080 	if (asprintf(&subdirpath, "%s%s%s", path,
2081 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2082 		return got_error_from_errno("asprintf");
2083 
2084 	(*new_te) = calloc(1, sizeof(**new_te));
2085 	if (*new_te == NULL)
2086 		return got_error_from_errno("calloc");
2087 	(*new_te)->mode = S_IFDIR;
2088 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2089 	    sizeof((*new_te)->name)) {
2090 		err = got_error(GOT_ERR_NO_SPACE);
2091 		goto done;
2092 	}
2093 	err = write_tree(&id, subdirpath, ignores,  repo,
2094 	    progress_cb, progress_arg);
2095 	if (err)
2096 		goto done;
2097 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2098 
2099 done:
2100 	free(id);
2101 	free(subdirpath);
2102 	if (err) {
2103 		free(*new_te);
2104 		*new_te = NULL;
2105 	}
2106 	return err;
2107 }
2108 
2109 static const struct got_error *
2110 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2111     struct got_pathlist_head *ignores, struct got_repository *repo,
2112     got_repo_import_cb progress_cb, void *progress_arg)
2113 {
2114 	const struct got_error *err = NULL;
2115 	DIR *dir;
2116 	struct dirent *de;
2117 	int nentries;
2118 	struct got_tree_entry *new_te = NULL;
2119 	struct got_pathlist_head paths;
2120 	struct got_pathlist_entry *pe;
2121 
2122 	*new_tree_id = NULL;
2123 
2124 	TAILQ_INIT(&paths);
2125 
2126 	dir = opendir(path_dir);
2127 	if (dir == NULL) {
2128 		err = got_error_from_errno2("opendir", path_dir);
2129 		goto done;
2130 	}
2131 
2132 	nentries = 0;
2133 	while ((de = readdir(dir)) != NULL) {
2134 		int ignore = 0;
2135 		int type;
2136 
2137 		if (strcmp(de->d_name, ".") == 0 ||
2138 		    strcmp(de->d_name, "..") == 0)
2139 			continue;
2140 
2141 		TAILQ_FOREACH(pe, ignores, entry) {
2142 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
2143 				ignore = 1;
2144 				break;
2145 			}
2146 		}
2147 		if (ignore)
2148 			continue;
2149 
2150 		err = got_path_dirent_type(&type, path_dir, de);
2151 		if (err)
2152 			goto done;
2153 
2154 		if (type == DT_DIR) {
2155 			err = import_subdir(&new_te, de, path_dir,
2156 			    ignores, repo, progress_cb, progress_arg);
2157 			if (err) {
2158 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2159 					goto done;
2160 				err = NULL;
2161 				continue;
2162 			}
2163 		} else if (type == DT_REG || type == DT_LNK) {
2164 			err = import_file(&new_te, de, path_dir, repo);
2165 			if (err)
2166 				goto done;
2167 		} else
2168 			continue;
2169 
2170 		err = insert_tree_entry(new_te, &paths);
2171 		if (err)
2172 			goto done;
2173 		nentries++;
2174 	}
2175 
2176 	if (TAILQ_EMPTY(&paths)) {
2177 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2178 		    "cannot create tree without any entries");
2179 		goto done;
2180 	}
2181 
2182 	TAILQ_FOREACH(pe, &paths, entry) {
2183 		struct got_tree_entry *te = pe->data;
2184 		char *path;
2185 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2186 			continue;
2187 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2188 			err = got_error_from_errno("asprintf");
2189 			goto done;
2190 		}
2191 		err = (*progress_cb)(progress_arg, path);
2192 		free(path);
2193 		if (err)
2194 			goto done;
2195 	}
2196 
2197 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2198 done:
2199 	if (dir)
2200 		closedir(dir);
2201 	got_pathlist_free(&paths);
2202 	return err;
2203 }
2204 
2205 const struct got_error *
2206 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2207     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2208     struct got_repository *repo, got_repo_import_cb progress_cb,
2209     void *progress_arg)
2210 {
2211 	const struct got_error *err;
2212 	struct got_object_id *new_tree_id;
2213 
2214 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2215 	    progress_cb, progress_arg);
2216 	if (err)
2217 		return err;
2218 
2219 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2220 	    author, time(NULL), author, time(NULL), logmsg, repo);
2221 	free(new_tree_id);
2222 	return err;
2223 }
2224 
2225 const struct got_error *
2226 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2227     struct got_repository *repo)
2228 {
2229 	const struct got_error *err = NULL;
2230 	char *path_objects = NULL, *path = NULL;
2231 	DIR *dir = NULL;
2232 	struct got_object_id id;
2233 	int i;
2234 
2235 	*nobjects = 0;
2236 	*ondisk_size = 0;
2237 
2238 	path_objects = got_repo_get_path_objects(repo);
2239 	if (path_objects == NULL)
2240 		return got_error_from_errno("got_repo_get_path_objects");
2241 
2242 	for (i = 0; i <= 0xff; i++) {
2243 		struct dirent *dent;
2244 
2245 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2246 			err = got_error_from_errno("asprintf");
2247 			break;
2248 		}
2249 
2250 		dir = opendir(path);
2251 		if (dir == NULL) {
2252 			if (errno == ENOENT) {
2253 				err = NULL;
2254 				continue;
2255 			}
2256 			err = got_error_from_errno2("opendir", path);
2257 			break;
2258 		}
2259 
2260 		while ((dent = readdir(dir)) != NULL) {
2261 			char *id_str;
2262 			int fd;
2263 			struct stat sb;
2264 
2265 			if (strcmp(dent->d_name, ".") == 0 ||
2266 			    strcmp(dent->d_name, "..") == 0)
2267 				continue;
2268 
2269 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2270 				err = got_error_from_errno("asprintf");
2271 				goto done;
2272 			}
2273 
2274 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2275 				free(id_str);
2276 				continue;
2277 			}
2278 			free(id_str);
2279 
2280 			err = got_object_open_loose_fd(&fd, &id, repo);
2281 			if (err)
2282 				goto done;
2283 
2284 			if (fstat(fd, &sb) == -1) {
2285 				err = got_error_from_errno("fstat");
2286 				close(fd);
2287 				goto done;
2288 			}
2289 			(*nobjects)++;
2290 			(*ondisk_size) += sb.st_size;
2291 
2292 			if (close(fd) == -1) {
2293 				err = got_error_from_errno("close");
2294 				goto done;
2295 			}
2296 		}
2297 
2298 		if (closedir(dir) != 0) {
2299 			err = got_error_from_errno("closedir");
2300 			goto done;
2301 		}
2302 		dir = NULL;
2303 
2304 		free(path);
2305 		path = NULL;
2306 	}
2307 done:
2308 	if (dir && closedir(dir) != 0 && err == NULL)
2309 		err = got_error_from_errno("closedir");
2310 
2311 	if (err) {
2312 		*nobjects = 0;
2313 		*ondisk_size = 0;
2314 	}
2315 	free(path_objects);
2316 	free(path);
2317 	return err;
2318 }
2319 
2320 const struct got_error *
2321 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2322     off_t *total_packsize, struct got_repository *repo)
2323 {
2324 	const struct got_error *err = NULL;
2325 	DIR *packdir = NULL;
2326 	struct dirent *dent;
2327 	struct got_packidx *packidx = NULL;
2328 	char *path_packidx;
2329 	char *path_packfile;
2330 	int packdir_fd;
2331 	struct stat sb;
2332 
2333 	*npackfiles = 0;
2334 	*nobjects = 0;
2335 	*total_packsize = 0;
2336 
2337 	packdir_fd = openat(got_repo_get_fd(repo),
2338 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2339 	if (packdir_fd == -1) {
2340 		return got_error_from_errno_fmt("openat: %s/%s",
2341 		    got_repo_get_path_git_dir(repo),
2342 		    GOT_OBJECTS_PACK_DIR);
2343 	}
2344 
2345 	packdir = fdopendir(packdir_fd);
2346 	if (packdir == NULL) {
2347 		err = got_error_from_errno("fdopendir");
2348 		goto done;
2349 	}
2350 
2351 	while ((dent = readdir(packdir)) != NULL) {
2352 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2353 			continue;
2354 
2355 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2356 		    dent->d_name) == -1) {
2357 			err = got_error_from_errno("asprintf");
2358 			goto done;
2359 		}
2360 
2361 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2362 		    path_packidx, 0);
2363 		free(path_packidx);
2364 		if (err)
2365 			goto done;
2366 
2367 		if (fstat(packidx->fd, &sb) == -1)
2368 			goto done;
2369 		*total_packsize += sb.st_size;
2370 
2371 		err = got_packidx_get_packfile_path(&path_packfile,
2372 		    packidx->path_packidx);
2373 		if (err)
2374 			goto done;
2375 
2376 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2377 		    0) == -1) {
2378 			free(path_packfile);
2379 			goto done;
2380 		}
2381 		free(path_packfile);
2382 		*total_packsize += sb.st_size;
2383 
2384 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2385 
2386 		(*npackfiles)++;
2387 
2388 		got_packidx_close(packidx);
2389 		packidx = NULL;
2390 	}
2391 done:
2392 	if (packidx)
2393 		got_packidx_close(packidx);
2394 	if (packdir && closedir(packdir) != 0 && err == NULL)
2395 		err = got_error_from_errno("closedir");
2396 	if (err) {
2397 		*npackfiles = 0;
2398 		*nobjects = 0;
2399 		*total_packsize = 0;
2400 	}
2401 	return err;
2402 }
2403 
2404 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2405     got_packidx_bloom_filter_cmp);