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