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 static inline int
78 is_boolean_val(const char *val)
79 {
80 	return (strcasecmp(val, "true") == 0 ||
81 	    strcasecmp(val, "false") == 0 ||
82 	    strcasecmp(val, "on") == 0 ||
83     	    strcasecmp(val, "off") == 0 ||
84 	    strcasecmp(val, "yes") == 0 ||
85     	    strcasecmp(val, "no") == 0 ||
86 	    strcasecmp(val, "1") == 0 ||
87 	    strcasecmp(val, "0") == 0);
88 }
89 
90 static inline int
91 get_boolean_val(const char *val)
92 {
93 	return (strcasecmp(val, "true") == 0 ||
94 	    strcasecmp(val, "on") == 0 ||
95 	    strcasecmp(val, "yes") == 0 ||
96 	    strcasecmp(val, "1") == 0);
97 }
98 
99 const char *
100 got_repo_get_path(struct got_repository *repo)
101 {
102 	return repo->path;
103 }
104 
105 const char *
106 got_repo_get_path_git_dir(struct got_repository *repo)
107 {
108 	return repo->path_git_dir;
109 }
110 
111 int
112 got_repo_get_fd(struct got_repository *repo)
113 {
114 	return repo->gitdir_fd;
115 }
116 
117 const char *
118 got_repo_get_gitconfig_author_name(struct got_repository *repo)
119 {
120 	return repo->gitconfig_author_name;
121 }
122 
123 const char *
124 got_repo_get_gitconfig_author_email(struct got_repository *repo)
125 {
126 	return repo->gitconfig_author_email;
127 }
128 
129 const char *
130 got_repo_get_global_gitconfig_author_name(struct got_repository *repo)
131 {
132 	return repo->global_gitconfig_author_name;
133 }
134 
135 const char *
136 got_repo_get_global_gitconfig_author_email(struct got_repository *repo)
137 {
138 	return repo->global_gitconfig_author_email;
139 }
140 
141 const char *
142 got_repo_get_gitconfig_owner(struct got_repository *repo)
143 {
144 	return repo->gitconfig_owner;
145 }
146 
147 int
148 got_repo_has_extension(struct got_repository *repo, const char *ext)
149 {
150 	int i;
151 
152 	for (i = 0; i < repo->nextensions; ++i) {
153 		if (!strcasecmp(ext, repo->extnames[i]))
154 			return get_boolean_val(repo->extvals[i]);
155 	}
156 
157 	return 0;
158 }
159 
160 int
161 got_repo_is_bare(struct got_repository *repo)
162 {
163 	return (strcmp(repo->path, repo->path_git_dir) == 0);
164 }
165 
166 static char *
167 get_path_git_child(struct got_repository *repo, const char *basename)
168 {
169 	char *path_child;
170 
171 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
172 	    basename) == -1)
173 		return NULL;
174 
175 	return path_child;
176 }
177 
178 char *
179 got_repo_get_path_objects(struct got_repository *repo)
180 {
181 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
182 }
183 
184 char *
185 got_repo_get_path_objects_pack(struct got_repository *repo)
186 {
187 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
188 }
189 
190 char *
191 got_repo_get_path_refs(struct got_repository *repo)
192 {
193 	return get_path_git_child(repo, GOT_REFS_DIR);
194 }
195 
196 char *
197 got_repo_get_path_packed_refs(struct got_repository *repo)
198 {
199 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
200 }
201 
202 static char *
203 get_path_head(struct got_repository *repo)
204 {
205 	return get_path_git_child(repo, GOT_HEAD_FILE);
206 }
207 
208 char *
209 got_repo_get_path_gitconfig(struct got_repository *repo)
210 {
211 	return get_path_git_child(repo, GOT_GITCONFIG);
212 }
213 
214 char *
215 got_repo_get_path_gotconfig(struct got_repository *repo)
216 {
217 	return get_path_git_child(repo, GOT_GOTCONFIG_FILENAME);
218 }
219 
220 const struct got_gotconfig *
221 got_repo_get_gotconfig(struct got_repository *repo)
222 {
223 	return repo->gotconfig;
224 }
225 
226 void
227 got_repo_get_gitconfig_remotes(int *nremotes,
228     const struct got_remote_repo **remotes, struct got_repository *repo)
229 {
230 	*nremotes = repo->ngitconfig_remotes;
231 	*remotes = repo->gitconfig_remotes;
232 }
233 
234 static int
235 is_git_repo(struct got_repository *repo)
236 {
237 	const char *path_git = got_repo_get_path_git_dir(repo);
238 	char *path_objects = got_repo_get_path_objects(repo);
239 	char *path_refs = got_repo_get_path_refs(repo);
240 	char *path_head = get_path_head(repo);
241 	int ret = 0;
242 	struct stat sb;
243 	struct got_reference *head_ref;
244 
245 	if (lstat(path_git, &sb) == -1)
246 		goto done;
247 	if (!S_ISDIR(sb.st_mode))
248 		goto done;
249 
250 	if (lstat(path_objects, &sb) == -1)
251 		goto done;
252 	if (!S_ISDIR(sb.st_mode))
253 		goto done;
254 
255 	if (lstat(path_refs, &sb) == -1)
256 		goto done;
257 	if (!S_ISDIR(sb.st_mode))
258 		goto done;
259 
260 	if (lstat(path_head, &sb) == -1)
261 		goto done;
262 	if (!S_ISREG(sb.st_mode))
263 		goto done;
264 
265 	/* Check if the HEAD reference can be opened. */
266 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
267 		goto done;
268 	got_ref_close(head_ref);
269 
270 	ret = 1;
271 done:
272 	free(path_objects);
273 	free(path_refs);
274 	free(path_head);
275 	return ret;
276 
277 }
278 
279 static const struct got_error *
280 close_tempfiles(int *fds, size_t nfds)
281 {
282 	const struct got_error *err = NULL;
283 	int i;
284 
285 	for (i = 0; i < nfds; i++) {
286 		if (fds[i] == -1)
287 			continue;
288 		if (close(fds[i]) == -1) {
289 			err = got_error_from_errno("close");
290 			break;
291 		}
292 	}
293 	free(fds);
294 	return err;
295 }
296 
297 static const struct got_error *
298 open_tempfiles(int **fds, size_t array_size, size_t nfds)
299 {
300 	const struct got_error *err = NULL;
301 	int i;
302 
303 	*fds = calloc(array_size, sizeof(**fds));
304 	if (*fds == NULL)
305 		return got_error_from_errno("calloc");
306 
307 	for (i = 0; i < array_size; i++)
308 		(*fds)[i] = -1;
309 
310 	for (i = 0; i < nfds; i++) {
311 		(*fds)[i] = got_opentempfd();
312 		if ((*fds)[i] == -1) {
313 			err = got_error_from_errno("got_opentempfd");
314 			close_tempfiles(*fds, nfds);
315 			*fds = NULL;
316 			return err;
317 		}
318 	}
319 
320 	return NULL;
321 }
322 
323 static const struct got_error *
324 get_pack_cache_size(int *pack_cache_size)
325 {
326 	struct rlimit rl;
327 
328 	if (getrlimit(RLIMIT_NOFILE, &rl) == -1)
329 		return got_error_from_errno("getrlimit");
330 
331 	*pack_cache_size = GOT_PACK_CACHE_SIZE;
332 	if (*pack_cache_size > rl.rlim_cur / 8)
333 		*pack_cache_size = rl.rlim_cur / 8;
334 
335 	return NULL;
336 }
337 
338 const struct got_error *
339 got_repo_pack_fds_open(int **pack_fds)
340 {
341 	const struct got_error *err;
342 	int nfds;
343 
344 	err = get_pack_cache_size(&nfds);
345 	if (err)
346 		return err;
347 
348 	/*
349 	 * We need one basefd and one accumfd per cached pack.
350 	 * Our constants should be set up in a way such that
351 	 * this error never triggers.
352 	 */
353 	if (nfds * 2 > GOT_PACK_NUM_TEMPFILES)
354 		return got_error(GOT_ERR_NO_SPACE);
355 
356 	return open_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES, nfds * 2);
357 }
358 
359 const struct got_error *
360 got_repo_pack_fds_close(int *pack_fds)
361 {
362 	return close_tempfiles(pack_fds, GOT_PACK_NUM_TEMPFILES);
363 }
364 
365 const struct got_error *
366 got_repo_temp_fds_open(int **temp_fds)
367 {
368 	return open_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES,
369 	    GOT_REPO_NUM_TEMPFILES);
370 }
371 
372 void
373 got_repo_temp_fds_set(struct got_repository *repo, int *temp_fds)
374 {
375 	int i;
376 
377 	for (i = 0; i < GOT_REPO_NUM_TEMPFILES; i++)
378 		repo->tempfiles[i] = temp_fds[i];
379 }
380 
381 const struct got_error *
382 got_repo_temp_fds_get(int *fd, int *idx, struct got_repository *repo)
383 {
384 	int i;
385 
386 	*fd = -1;
387 	*idx = -1;
388 
389 	for (i = 0; i < nitems(repo->tempfiles); i++) {
390 		if (repo->tempfile_use_mask & (1 << i))
391 			continue;
392 		if (repo->tempfiles[i] != -1) {
393 			if (ftruncate(repo->tempfiles[i], 0L) == -1)
394 				return got_error_from_errno("ftruncate");
395 			*fd = repo->tempfiles[i];
396 			*idx = i;
397 			repo->tempfile_use_mask |= (1 << i);
398 			return NULL;
399 		}
400 	}
401 
402 	return got_error(GOT_ERR_REPO_TEMPFILE);
403 }
404 
405 void
406 got_repo_temp_fds_put(int idx, struct got_repository *repo)
407 {
408 	repo->tempfile_use_mask &= ~(1 << idx);
409 }
410 
411 const struct got_error *
412 got_repo_temp_fds_close(int *temp_fds)
413 {
414 	return close_tempfiles(temp_fds, GOT_REPO_NUM_TEMPFILES);
415 }
416 
417 const struct got_error *
418 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
419     struct got_object *obj)
420 {
421 #ifndef GOT_NO_OBJ_CACHE
422 	const struct got_error *err = NULL;
423 	err = got_object_cache_add(&repo->objcache, id, obj);
424 	if (err) {
425 		if (err->code == GOT_ERR_OBJ_EXISTS ||
426 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
427 			err = NULL;
428 		return err;
429 	}
430 	obj->refcnt++;
431 #endif
432 	return NULL;
433 }
434 
435 struct got_object *
436 got_repo_get_cached_object(struct got_repository *repo,
437     struct got_object_id *id)
438 {
439 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
440 }
441 
442 const struct got_error *
443 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
444     struct got_tree_object *tree)
445 {
446 #ifndef GOT_NO_OBJ_CACHE
447 	const struct got_error *err = NULL;
448 	err = got_object_cache_add(&repo->treecache, id, tree);
449 	if (err) {
450 		if (err->code == GOT_ERR_OBJ_EXISTS ||
451 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
452 			err = NULL;
453 		return err;
454 	}
455 	tree->refcnt++;
456 #endif
457 	return NULL;
458 }
459 
460 struct got_tree_object *
461 got_repo_get_cached_tree(struct got_repository *repo,
462     struct got_object_id *id)
463 {
464 	return (struct got_tree_object *)got_object_cache_get(
465 	    &repo->treecache, id);
466 }
467 
468 const struct got_error *
469 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
470     struct got_commit_object *commit)
471 {
472 #ifndef GOT_NO_OBJ_CACHE
473 	const struct got_error *err = NULL;
474 	err = got_object_cache_add(&repo->commitcache, id, commit);
475 	if (err) {
476 		if (err->code == GOT_ERR_OBJ_EXISTS ||
477 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
478 			err = NULL;
479 		return err;
480 	}
481 	commit->refcnt++;
482 #endif
483 	return NULL;
484 }
485 
486 struct got_commit_object *
487 got_repo_get_cached_commit(struct got_repository *repo,
488     struct got_object_id *id)
489 {
490 	return (struct got_commit_object *)got_object_cache_get(
491 	    &repo->commitcache, id);
492 }
493 
494 const struct got_error *
495 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
496     struct got_tag_object *tag)
497 {
498 #ifndef GOT_NO_OBJ_CACHE
499 	const struct got_error *err = NULL;
500 	err = got_object_cache_add(&repo->tagcache, id, tag);
501 	if (err) {
502 		if (err->code == GOT_ERR_OBJ_EXISTS ||
503 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
504 			err = NULL;
505 		return err;
506 	}
507 	tag->refcnt++;
508 #endif
509 	return NULL;
510 }
511 
512 struct got_tag_object *
513 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
514 {
515 	return (struct got_tag_object *)got_object_cache_get(
516 	    &repo->tagcache, id);
517 }
518 
519 const struct got_error *
520 got_repo_cache_raw_object(struct got_repository *repo, struct got_object_id *id,
521     struct got_raw_object *raw)
522 {
523 #ifndef GOT_NO_OBJ_CACHE
524 	const struct got_error *err = NULL;
525 	err = got_object_cache_add(&repo->rawcache, id, raw);
526 	if (err) {
527 		if (err->code == GOT_ERR_OBJ_EXISTS ||
528 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
529 			err = NULL;
530 		return err;
531 	}
532 	raw->refcnt++;
533 #endif
534 	return NULL;
535 }
536 
537 
538 struct got_raw_object *
539 got_repo_get_cached_raw_object(struct got_repository *repo,
540     struct got_object_id *id)
541 {
542 	return (struct got_raw_object *)got_object_cache_get(&repo->rawcache, id);
543 }
544 
545 
546 static const struct got_error *
547 open_repo(struct got_repository *repo, const char *path)
548 {
549 	const struct got_error *err = NULL;
550 
551 	repo->gitdir_fd = -1;
552 
553 	/* bare git repository? */
554 	repo->path_git_dir = strdup(path);
555 	if (repo->path_git_dir == NULL)
556 		return got_error_from_errno("strdup");
557 	if (is_git_repo(repo)) {
558 		repo->path = strdup(repo->path_git_dir);
559 		if (repo->path == NULL) {
560 			err = got_error_from_errno("strdup");
561 			goto done;
562 		}
563 		repo->gitdir_fd = open(repo->path_git_dir,
564 		    O_DIRECTORY | O_CLOEXEC);
565 		if (repo->gitdir_fd == -1) {
566 			err = got_error_from_errno2("open",
567 			    repo->path_git_dir);
568 			goto done;
569 		}
570 		return NULL;
571 	}
572 
573 	/* git repository with working tree? */
574 	free(repo->path_git_dir);
575 	repo->path_git_dir = NULL;
576 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
577 		err = got_error_from_errno("asprintf");
578 		goto done;
579 	}
580 	if (is_git_repo(repo)) {
581 		repo->path = strdup(path);
582 		if (repo->path == NULL) {
583 			err = got_error_from_errno("strdup");
584 			goto done;
585 		}
586 		repo->gitdir_fd = open(repo->path_git_dir,
587 		    O_DIRECTORY | O_CLOEXEC);
588 		if (repo->gitdir_fd == -1) {
589 			err = got_error_from_errno2("open",
590 			    repo->path_git_dir);
591 			goto done;
592 		}
593 		return NULL;
594 	}
595 
596 	err = got_error(GOT_ERR_NOT_GIT_REPO);
597 done:
598 	if (err) {
599 		free(repo->path);
600 		repo->path = NULL;
601 		free(repo->path_git_dir);
602 		repo->path_git_dir = NULL;
603 		if (repo->gitdir_fd != -1)
604 			close(repo->gitdir_fd);
605 		repo->gitdir_fd = -1;
606 
607 	}
608 	return err;
609 }
610 
611 static const struct got_error *
612 read_gitconfig(struct got_repository *repo, const char *global_gitconfig_path)
613 {
614 	const struct got_error *err = NULL;
615 	char *repo_gitconfig_path = NULL;
616 
617 	if (global_gitconfig_path) {
618 		/* Read settings from ~/.gitconfig. */
619 		int dummy_repo_version;
620 		err = got_repo_read_gitconfig(&dummy_repo_version,
621 		    &repo->global_gitconfig_author_name,
622 		    &repo->global_gitconfig_author_email,
623 		    NULL, NULL, NULL, NULL, NULL, NULL,
624 		    global_gitconfig_path);
625 		if (err)
626 			return err;
627 	}
628 
629 	/* Read repository's .git/config file. */
630 	repo_gitconfig_path = got_repo_get_path_gitconfig(repo);
631 	if (repo_gitconfig_path == NULL)
632 		return got_error_from_errno("got_repo_get_path_gitconfig");
633 
634 	err = got_repo_read_gitconfig(
635 	    &repo->gitconfig_repository_format_version,
636 	    &repo->gitconfig_author_name, &repo->gitconfig_author_email,
637 	    &repo->gitconfig_remotes, &repo->ngitconfig_remotes,
638 	    &repo->gitconfig_owner, &repo->extnames, &repo->extvals,
639 	    &repo->nextensions, repo_gitconfig_path);
640 	if (err)
641 		goto done;
642 
643 	if (getenv("GOT_IGNORE_GITCONFIG") != NULL) {
644 		int i;
645 
646 		for (i = 0; i < repo->ngitconfig_remotes; i++) {
647 			got_repo_free_remote_repo_data(
648 			    &repo->gitconfig_remotes[i]);
649 		}
650 		free(repo->gitconfig_remotes);
651 		repo->gitconfig_remotes = NULL;
652 		repo->ngitconfig_remotes = 0;
653 
654 		free(repo->gitconfig_author_name);
655 		repo->gitconfig_author_name = NULL;
656 		free(repo->gitconfig_author_email);
657 		repo->gitconfig_author_email = NULL;
658 
659 		free(repo->global_gitconfig_author_name);
660 		repo->global_gitconfig_author_name = NULL;
661 		free(repo->global_gitconfig_author_email);
662 		repo->global_gitconfig_author_email = NULL;
663 	}
664 
665 done:
666 	free(repo_gitconfig_path);
667 	return err;
668 }
669 
670 static const struct got_error *
671 read_gotconfig(struct got_repository *repo)
672 {
673 	const struct got_error *err = NULL;
674 	char *gotconfig_path;
675 
676 	gotconfig_path = got_repo_get_path_gotconfig(repo);
677 	if (gotconfig_path == NULL)
678 		return got_error_from_errno("got_repo_get_path_gotconfig");
679 
680 	err = got_gotconfig_read(&repo->gotconfig, gotconfig_path);
681 	free(gotconfig_path);
682 	return err;
683 }
684 
685 /* Supported repository format extensions. */
686 static const char *const repo_extensions[] = {
687 	"noop",			/* Got supports repository format version 1. */
688 	"preciousObjects",	/* Supported by gotadmin cleanup. */
689 	"worktreeConfig",	/* Got does not care about Git work trees. */
690 };
691 
692 const struct got_error *
693 got_repo_open(struct got_repository **repop, const char *path,
694     const char *global_gitconfig_path, int *pack_fds)
695 {
696 	struct got_repository *repo = NULL;
697 	const struct got_error *err = NULL;
698 	char *repo_path = NULL;
699 	size_t i, j = 0;
700 
701 	*repop = NULL;
702 
703 	repo = calloc(1, sizeof(*repo));
704 	if (repo == NULL)
705 		return got_error_from_errno("calloc");
706 
707 	RB_INIT(&repo->packidx_bloom_filters);
708 	TAILQ_INIT(&repo->packidx_paths);
709 
710 	for (i = 0; i < nitems(repo->privsep_children); i++) {
711 		memset(&repo->privsep_children[i], 0,
712 		    sizeof(repo->privsep_children[0]));
713 		repo->privsep_children[i].imsg_fd = -1;
714 	}
715 
716 	err = got_object_cache_init(&repo->objcache,
717 	    GOT_OBJECT_CACHE_TYPE_OBJ);
718 	if (err)
719 		goto done;
720 	err = got_object_cache_init(&repo->treecache,
721 	    GOT_OBJECT_CACHE_TYPE_TREE);
722 	if (err)
723 		goto done;
724 	err = got_object_cache_init(&repo->commitcache,
725 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
726 	if (err)
727 		goto done;
728 	err = got_object_cache_init(&repo->tagcache,
729 	    GOT_OBJECT_CACHE_TYPE_TAG);
730 	if (err)
731 		goto done;
732 	err = got_object_cache_init(&repo->rawcache,
733 	    GOT_OBJECT_CACHE_TYPE_RAW);
734 	if (err)
735 		goto done;
736 
737 	err = get_pack_cache_size(&repo->pack_cache_size);
738 	if (err)
739 		goto done;
740 	for (i = 0; i < nitems(repo->packs); i++) {
741 		if (pack_fds != NULL && i < repo->pack_cache_size) {
742 			repo->packs[i].basefd = pack_fds[j++];
743 			repo->packs[i].accumfd = pack_fds[j++];
744 		} else {
745 			repo->packs[i].basefd = -1;
746 			repo->packs[i].accumfd = -1;
747 		}
748 	}
749 	for (i = 0; i < nitems(repo->tempfiles); i++)
750 		repo->tempfiles[i] = -1;
751 	repo->pinned_pack = -1;
752 	repo->pinned_packidx = -1;
753 	repo->pinned_pid = 0;
754 
755 	repo_path = realpath(path, NULL);
756 	if (repo_path == NULL) {
757 		err = got_error_from_errno2("realpath", path);
758 		goto done;
759 	}
760 
761 	for (;;) {
762 		char *parent_path;
763 
764 		err = open_repo(repo, repo_path);
765 		if (err == NULL)
766 			break;
767 		if (err->code != GOT_ERR_NOT_GIT_REPO)
768 			goto done;
769 		if (repo_path[0] == '/' && repo_path[1] == '\0') {
770 			err = got_error(GOT_ERR_NOT_GIT_REPO);
771 			goto done;
772 		}
773 		err = got_path_dirname(&parent_path, repo_path);
774 		if (err)
775 			goto done;
776 		free(repo_path);
777 		repo_path = parent_path;
778 	}
779 
780 	err = read_gotconfig(repo);
781 	if (err)
782 		goto done;
783 
784 	err = read_gitconfig(repo, global_gitconfig_path);
785 	if (err)
786 		goto done;
787 	if (repo->gitconfig_repository_format_version != 0) {
788 		err = got_error_path(path, GOT_ERR_GIT_REPO_FORMAT);
789 		goto done;
790 	}
791 	for (i = 0; i < repo->nextensions; i++) {
792 		char *ext = repo->extnames[i];
793 		char *val = repo->extvals[i];
794 		int j, supported = 0;
795 
796 		if (!is_boolean_val(val)) {
797 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
798 			goto done;
799 		}
800 
801 		if (!get_boolean_val(val))
802 			continue;
803 
804 		for (j = 0; j < nitems(repo_extensions); j++) {
805 			if (strcmp(ext, repo_extensions[j]) == 0) {
806 				supported = 1;
807 				break;
808 			}
809 		}
810 		if (!supported) {
811 			err = got_error_path(ext, GOT_ERR_GIT_REPO_EXT);
812 			goto done;
813 		}
814 	}
815 
816 	err = got_repo_list_packidx(&repo->packidx_paths, repo);
817 done:
818 	if (err)
819 		got_repo_close(repo);
820 	else
821 		*repop = repo;
822 	free(repo_path);
823 	return err;
824 }
825 
826 const struct got_error *
827 got_repo_close(struct got_repository *repo)
828 {
829 	const struct got_error *err = NULL, *child_err;
830 	struct got_packidx_bloom_filter *bf;
831 	size_t i;
832 
833 	for (i = 0; i < repo->pack_cache_size; i++) {
834 		if (repo->packidx_cache[i] == NULL)
835 			break;
836 		got_packidx_close(repo->packidx_cache[i]);
837 	}
838 
839 	while ((bf = RB_MIN(got_packidx_bloom_filter_tree,
840 	    &repo->packidx_bloom_filters))) {
841 		RB_REMOVE(got_packidx_bloom_filter_tree,
842 		    &repo->packidx_bloom_filters, bf);
843 		bloom_free(bf->bloom);
844 		free(bf->bloom);
845 		free(bf);
846 	}
847 
848 	for (i = 0; i < repo->pack_cache_size; i++)
849 		if (repo->packs[i].path_packfile)
850 			if (repo->packs[i].path_packfile)
851 				got_pack_close(&repo->packs[i]);
852 
853 	free(repo->path);
854 	free(repo->path_git_dir);
855 
856 	got_object_cache_close(&repo->objcache);
857 	got_object_cache_close(&repo->treecache);
858 	got_object_cache_close(&repo->commitcache);
859 	got_object_cache_close(&repo->tagcache);
860 	got_object_cache_close(&repo->rawcache);
861 
862 	for (i = 0; i < nitems(repo->privsep_children); i++) {
863 		if (repo->privsep_children[i].imsg_fd == -1)
864 			continue;
865 		imsg_clear(repo->privsep_children[i].ibuf);
866 		free(repo->privsep_children[i].ibuf);
867 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
868 		child_err = got_privsep_wait_for_child(
869 		    repo->privsep_children[i].pid);
870 		if (child_err && err == NULL)
871 			err = child_err;
872 		if (close(repo->privsep_children[i].imsg_fd) == -1 &&
873 		    err == NULL)
874 			err = got_error_from_errno("close");
875 	}
876 
877 	if (repo->gitdir_fd != -1 && close(repo->gitdir_fd) == -1 &&
878 	    err == NULL)
879 		err = got_error_from_errno("close");
880 
881 	if (repo->gotconfig)
882 		got_gotconfig_free(repo->gotconfig);
883 	free(repo->gitconfig_author_name);
884 	free(repo->gitconfig_author_email);
885 	for (i = 0; i < repo->ngitconfig_remotes; i++)
886 		got_repo_free_remote_repo_data(&repo->gitconfig_remotes[i]);
887 	free(repo->gitconfig_remotes);
888 	for (i = 0; i < repo->nextensions; i++) {
889 		free(repo->extnames[i]);
890 		free(repo->extvals[i]);
891 	}
892 	free(repo->extnames);
893 	free(repo->extvals);
894 
895 	got_pathlist_free(&repo->packidx_paths, GOT_PATHLIST_FREE_PATH);
896 	free(repo);
897 
898 	return err;
899 }
900 
901 void
902 got_repo_free_remote_repo_data(struct got_remote_repo *repo)
903 {
904 	int i;
905 
906 	free(repo->name);
907 	repo->name = NULL;
908 	free(repo->fetch_url);
909 	repo->fetch_url = NULL;
910 	free(repo->send_url);
911 	repo->send_url = NULL;
912 	for (i = 0; i < repo->nfetch_branches; i++)
913 		free(repo->fetch_branches[i]);
914 	free(repo->fetch_branches);
915 	repo->fetch_branches = NULL;
916 	repo->nfetch_branches = 0;
917 	for (i = 0; i < repo->nsend_branches; i++)
918 		free(repo->send_branches[i]);
919 	free(repo->send_branches);
920 	repo->send_branches = NULL;
921 	repo->nsend_branches = 0;
922 }
923 
924 const struct got_error *
925 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
926     const char *input_path)
927 {
928 	const struct got_error *err = NULL;
929 	const char *repo_abspath = NULL;
930 	size_t repolen, len;
931 	char *canonpath, *path = NULL;
932 
933 	*in_repo_path = NULL;
934 
935 	canonpath = strdup(input_path);
936 	if (canonpath == NULL) {
937 		err = got_error_from_errno("strdup");
938 		goto done;
939 	}
940 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
941 	if (err)
942 		goto done;
943 
944 	repo_abspath = got_repo_get_path(repo);
945 
946 	if (canonpath[0] == '\0') {
947 		path = strdup(canonpath);
948 		if (path == NULL) {
949 			err = got_error_from_errno("strdup");
950 			goto done;
951 		}
952 	} else {
953 		path = realpath(canonpath, NULL);
954 		if (path == NULL) {
955 			if (errno != ENOENT) {
956 				err = got_error_from_errno2("realpath",
957 				    canonpath);
958 				goto done;
959 			}
960 			/*
961 			 * Path is not on disk.
962 			 * Assume it is already relative to repository root.
963 			 */
964 			path = strdup(canonpath);
965 			if (path == NULL) {
966 				err = got_error_from_errno("strdup");
967 				goto done;
968 			}
969 		}
970 
971 		repolen = strlen(repo_abspath);
972 		len = strlen(path);
973 
974 
975 		if (strcmp(path, repo_abspath) == 0) {
976 			free(path);
977 			path = strdup("");
978 			if (path == NULL) {
979 				err = got_error_from_errno("strdup");
980 				goto done;
981 			}
982 		} else if (len > repolen &&
983 		    got_path_is_child(path, repo_abspath, repolen)) {
984 			/* Matched an on-disk path inside repository. */
985 			if (got_repo_is_bare(repo)) {
986 				/*
987 				 * Matched an on-disk path inside repository
988 				 * database. Treat input as repository-relative.
989 				 */
990 				free(path);
991 				path = canonpath;
992 				canonpath = NULL;
993 			} else {
994 				char *child;
995 				/* Strip common prefix with repository path. */
996 				err = got_path_skip_common_ancestor(&child,
997 				    repo_abspath, path);
998 				if (err)
999 					goto done;
1000 				free(path);
1001 				path = child;
1002 			}
1003 		} else {
1004 			/*
1005 			 * Matched unrelated on-disk path.
1006 			 * Treat input as repository-relative.
1007 			 */
1008 			free(path);
1009 			path = canonpath;
1010 			canonpath = NULL;
1011 		}
1012 	}
1013 
1014 	/* Make in-repository path absolute */
1015 	if (path[0] != '/') {
1016 		char *abspath;
1017 		if (asprintf(&abspath, "/%s", path) == -1) {
1018 			err = got_error_from_errno("asprintf");
1019 			goto done;
1020 		}
1021 		free(path);
1022 		path = abspath;
1023 	}
1024 
1025 done:
1026 	free(canonpath);
1027 	if (err)
1028 		free(path);
1029 	else
1030 		*in_repo_path = path;
1031 	return err;
1032 }
1033 
1034 static const struct got_error *
1035 cache_packidx(struct got_repository *repo, struct got_packidx *packidx,
1036     const char *path_packidx)
1037 {
1038 	const struct got_error *err = NULL;
1039 	size_t i;
1040 
1041 	for (i = 0; i < repo->pack_cache_size; i++) {
1042 		if (repo->packidx_cache[i] == NULL)
1043 			break;
1044 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1045 		    path_packidx) == 0) {
1046 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1047 		}
1048 	}
1049 	if (i == repo->pack_cache_size) {
1050 		do {
1051 			i--;
1052 		} while (i > 0 && repo->pinned_packidx >= 0 &&
1053 		    i == repo->pinned_packidx);
1054 		err = got_packidx_close(repo->packidx_cache[i]);
1055 		if (err)
1056 			return err;
1057 	}
1058 
1059 	repo->packidx_cache[i] = packidx;
1060 
1061 	return NULL;
1062 }
1063 
1064 int
1065 got_repo_is_packidx_filename(const char *name, size_t len)
1066 {
1067 	if (len != GOT_PACKIDX_NAMELEN)
1068 		return 0;
1069 
1070 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
1071 		return 0;
1072 
1073 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
1074 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
1075 		return 0;
1076 
1077 	return 1;
1078 }
1079 
1080 static struct got_packidx_bloom_filter *
1081 get_packidx_bloom_filter(struct got_repository *repo,
1082     const char *path, size_t path_len)
1083 {
1084 	struct got_packidx_bloom_filter key;
1085 
1086 	if (strlcpy(key.path, path, sizeof(key.path)) >= sizeof(key.path))
1087 		return NULL; /* XXX */
1088 	key.path_len = path_len;
1089 
1090 	return RB_FIND(got_packidx_bloom_filter_tree,
1091 	    &repo->packidx_bloom_filters, &key);
1092 }
1093 
1094 int
1095 got_repo_check_packidx_bloom_filter(struct got_repository *repo,
1096     const char *path_packidx, struct got_object_id *id)
1097 {
1098 	struct got_packidx_bloom_filter *bf;
1099 
1100 	bf = get_packidx_bloom_filter(repo, path_packidx, strlen(path_packidx));
1101 	if (bf)
1102 		return bloom_check(bf->bloom, id->sha1, sizeof(id->sha1));
1103 
1104 	/* No bloom filter means this pack index must be searched. */
1105 	return 1;
1106 }
1107 
1108 static const struct got_error *
1109 add_packidx_bloom_filter(struct got_repository *repo,
1110     struct got_packidx *packidx, const char *path_packidx)
1111 {
1112 	int i, nobjects = be32toh(packidx->hdr.fanout_table[0xff]);
1113 	struct got_packidx_bloom_filter *bf;
1114 	size_t len;
1115 
1116 	/*
1117 	 * Don't use bloom filters for very large pack index files.
1118 	 * Large pack files will contain a relatively large fraction
1119 	 * of our objects so we will likely need to visit them anyway.
1120 	 * The more objects a pack file contains the higher the probability
1121 	 * of a false-positive match from the bloom filter. And reading
1122 	 * all object IDs from a large pack index file can be expensive.
1123 	 */
1124 	if (nobjects > 100000) /* cut-off at about 2MB, at 20 bytes per ID */
1125 		return NULL;
1126 
1127 	/* Do we already have a filter for this pack index? */
1128 	if (get_packidx_bloom_filter(repo, path_packidx,
1129 	    strlen(path_packidx)) != NULL)
1130 		return NULL;
1131 
1132 	bf = calloc(1, sizeof(*bf));
1133 	if (bf == NULL)
1134 		return got_error_from_errno("calloc");
1135 	bf->bloom = calloc(1, sizeof(*bf->bloom));
1136 	if (bf->bloom == NULL) {
1137 		free(bf);
1138 		return got_error_from_errno("calloc");
1139 	}
1140 
1141 	len = strlcpy(bf->path, path_packidx, sizeof(bf->path));
1142 	if (len >= sizeof(bf->path)) {
1143 		free(bf->bloom);
1144 		free(bf);
1145 		return got_error(GOT_ERR_NO_SPACE);
1146 	}
1147 	bf->path_len = len;
1148 
1149 	/* Minimum size supported by our bloom filter is 1000 entries. */
1150 	bloom_init(bf->bloom, nobjects < 1000 ? 1000 : nobjects, 0.1);
1151 	for (i = 0; i < nobjects; i++) {
1152 		struct got_packidx_object_id *id;
1153 		id = &packidx->hdr.sorted_ids[i];
1154 		bloom_add(bf->bloom, id->sha1, sizeof(id->sha1));
1155 	}
1156 
1157 	RB_INSERT(got_packidx_bloom_filter_tree,
1158 	    &repo->packidx_bloom_filters, bf);
1159 	return NULL;
1160 }
1161 
1162 static void
1163 purge_packidx_paths(struct got_pathlist_head *packidx_paths)
1164 {
1165 	struct got_pathlist_entry *pe;
1166 
1167 	while (!TAILQ_EMPTY(packidx_paths)) {
1168 		pe = TAILQ_FIRST(packidx_paths);
1169 		TAILQ_REMOVE(packidx_paths, pe, entry);
1170 		free((char *)pe->path);
1171 		free(pe);
1172 	}
1173 }
1174 
1175 static const struct got_error *
1176 refresh_packidx_paths(struct got_repository *repo)
1177 {
1178 	const struct got_error *err = NULL;
1179 	char *objects_pack_dir = NULL;
1180 	struct stat sb;
1181 
1182 	objects_pack_dir = got_repo_get_path_objects_pack(repo);
1183 	if (objects_pack_dir == NULL)
1184 		return got_error_from_errno("got_repo_get_path_objects_pack");
1185 
1186 	if (stat(objects_pack_dir, &sb) == -1) {
1187 		if (errno != ENOENT) {
1188 			err = got_error_from_errno2("stat", objects_pack_dir);
1189 			goto done;
1190 		}
1191 	} else if (TAILQ_EMPTY(&repo->packidx_paths) ||
1192 	    sb.st_mtim.tv_sec != repo->pack_path_mtime.tv_sec ||
1193 	    sb.st_mtim.tv_nsec != repo->pack_path_mtime.tv_nsec) {
1194 		purge_packidx_paths(&repo->packidx_paths);
1195 		err = got_repo_list_packidx(&repo->packidx_paths, repo);
1196 		if (err)
1197 			goto done;
1198 	}
1199 done:
1200 	free(objects_pack_dir);
1201 	return err;
1202 }
1203 
1204 const struct got_error *
1205 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
1206     struct got_repository *repo, struct got_object_id *id)
1207 {
1208 	const struct got_error *err;
1209 	struct got_pathlist_entry *pe;
1210 	size_t i;
1211 
1212 	/* Search pack index cache. */
1213 	for (i = 0; i < repo->pack_cache_size; i++) {
1214 		if (repo->packidx_cache[i] == NULL)
1215 			break;
1216 		if (!got_repo_check_packidx_bloom_filter(repo,
1217 		    repo->packidx_cache[i]->path_packidx, id))
1218 			continue; /* object will not be found in this index */
1219 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
1220 		if (*idx != -1) {
1221 			*packidx = repo->packidx_cache[i];
1222 			/*
1223 			 * Move this cache entry to the front. Repeatedly
1224 			 * searching a wrong pack index can be expensive.
1225 			 */
1226 			if (i > 0) {
1227 				memmove(&repo->packidx_cache[1],
1228 				    &repo->packidx_cache[0],
1229 				    i * sizeof(repo->packidx_cache[0]));
1230 				repo->packidx_cache[0] = *packidx;
1231 				if (repo->pinned_packidx >= 0 &&
1232 				    repo->pinned_packidx < i)
1233 					repo->pinned_packidx++;
1234 				else if (repo->pinned_packidx == i)
1235 					repo->pinned_packidx = 0;
1236 			}
1237 			return NULL;
1238 		}
1239 	}
1240 	/* No luck. Search the filesystem. */
1241 
1242 	err = refresh_packidx_paths(repo);
1243 	if (err)
1244 		return err;
1245 
1246 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1247 		const char *path_packidx = pe->path;
1248 		int is_cached = 0;
1249 
1250 		if (!got_repo_check_packidx_bloom_filter(repo,
1251 		    pe->path, id))
1252 			continue; /* object will not be found in this index */
1253 
1254 		for (i = 0; i < repo->pack_cache_size; i++) {
1255 			if (repo->packidx_cache[i] == NULL)
1256 				break;
1257 			if (strcmp(repo->packidx_cache[i]->path_packidx,
1258 			    path_packidx) == 0) {
1259 				is_cached = 1;
1260 				break;
1261 			}
1262 		}
1263 		if (is_cached)
1264 			continue; /* already searched */
1265 
1266 		err = got_packidx_open(packidx, got_repo_get_fd(repo),
1267 		    path_packidx, 0);
1268 		if (err)
1269 			goto done;
1270 
1271 		err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1272 		if (err)
1273 			goto done;
1274 
1275 		err = cache_packidx(repo, *packidx, path_packidx);
1276 		if (err)
1277 			goto done;
1278 
1279 		*idx = got_packidx_get_object_idx(*packidx, id);
1280 		if (*idx != -1) {
1281 			err = NULL; /* found the object */
1282 			goto done;
1283 		}
1284 	}
1285 
1286 	err = got_error_no_obj(id);
1287 done:
1288 	return err;
1289 }
1290 
1291 const struct got_error *
1292 got_repo_list_packidx(struct got_pathlist_head *packidx_paths,
1293     struct got_repository *repo)
1294 {
1295 	const struct got_error *err = NULL;
1296 	DIR *packdir = NULL;
1297 	struct dirent *dent;
1298 	char *path_packidx = NULL;
1299 	int packdir_fd;
1300 	struct stat sb;
1301 
1302 	packdir_fd = openat(got_repo_get_fd(repo),
1303 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY | O_CLOEXEC);
1304 	if (packdir_fd == -1) {
1305 		return got_error_from_errno_fmt("openat: %s/%s",
1306 		    got_repo_get_path_git_dir(repo),
1307 		    GOT_OBJECTS_PACK_DIR);
1308 	}
1309 
1310 	packdir = fdopendir(packdir_fd);
1311 	if (packdir == NULL) {
1312 		err = got_error_from_errno("fdopendir");
1313 		goto done;
1314 	}
1315 
1316 	if (fstat(packdir_fd, &sb) == -1) {
1317 		err = got_error_from_errno("fstat");
1318 		goto done;
1319 	}
1320 	repo->pack_path_mtime.tv_sec = sb.st_mtim.tv_sec;
1321 	repo->pack_path_mtime.tv_nsec = sb.st_mtim.tv_nsec;
1322 
1323 	while ((dent = readdir(packdir)) != NULL) {
1324 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
1325 			continue;
1326 
1327 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
1328 		    dent->d_name) == -1) {
1329 			err = got_error_from_errno("asprintf");
1330 			path_packidx = NULL;
1331 			break;
1332 		}
1333 
1334 		err = got_pathlist_append(packidx_paths, path_packidx, NULL);
1335 		if (err)
1336 			break;
1337 	}
1338 done:
1339 	if (err)
1340 		free(path_packidx);
1341 	if (packdir && closedir(packdir) != 0 && err == NULL)
1342 		err = got_error_from_errno("closedir");
1343 	return err;
1344 }
1345 
1346 const struct got_error *
1347 got_repo_get_packidx(struct got_packidx **packidx, const char *path_packidx,
1348     struct got_repository *repo)
1349 {
1350 	const struct got_error *err;
1351 	size_t i;
1352 
1353 	*packidx = NULL;
1354 
1355 	/* Search pack index cache. */
1356 	for (i = 0; i < repo->pack_cache_size; i++) {
1357 		if (repo->packidx_cache[i] == NULL)
1358 			break;
1359 		if (strcmp(repo->packidx_cache[i]->path_packidx,
1360 		    path_packidx) == 0) {
1361 			*packidx = repo->packidx_cache[i];
1362 			return NULL;
1363 		}
1364 	}
1365 	/* No luck. Search the filesystem. */
1366 
1367 	err = got_packidx_open(packidx, got_repo_get_fd(repo),
1368 	    path_packidx, 0);
1369 	if (err)
1370 		return err;
1371 
1372 	err = add_packidx_bloom_filter(repo, *packidx, path_packidx);
1373 	if (err)
1374 		goto done;
1375 
1376 	err = cache_packidx(repo, *packidx, path_packidx);
1377 done:
1378 	if (err) {
1379 		got_packidx_close(*packidx);
1380 		*packidx = NULL;
1381 	}
1382 	return err;
1383 }
1384 
1385 static const struct got_error *
1386 read_packfile_hdr(int fd, struct got_packidx *packidx)
1387 {
1388 	const struct got_error *err = NULL;
1389 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
1390 	struct got_packfile_hdr hdr;
1391 	ssize_t n;
1392 
1393 	n = read(fd, &hdr, sizeof(hdr));
1394 	if (n < 0)
1395 		return got_error_from_errno("read");
1396 	if (n != sizeof(hdr))
1397 		return got_error(GOT_ERR_BAD_PACKFILE);
1398 
1399 	if (be32toh(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
1400 	    be32toh(hdr.version) != GOT_PACKFILE_VERSION ||
1401 	    be32toh(hdr.nobjects) != totobj)
1402 		err = got_error(GOT_ERR_BAD_PACKFILE);
1403 
1404 	return err;
1405 }
1406 
1407 static const struct got_error *
1408 open_packfile(int *fd, struct got_repository *repo,
1409     const char *relpath, struct got_packidx *packidx)
1410 {
1411 	const struct got_error *err = NULL;
1412 
1413 	*fd = openat(got_repo_get_fd(repo), relpath,
1414 	    O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
1415 	if (*fd == -1)
1416 		return got_error_from_errno_fmt("openat: %s/%s",
1417 		    got_repo_get_path_git_dir(repo), relpath);
1418 
1419 	if (packidx) {
1420 		err = read_packfile_hdr(*fd, packidx);
1421 		if (err) {
1422 			close(*fd);
1423 			*fd = -1;
1424 		}
1425 	}
1426 
1427 	return err;
1428 }
1429 
1430 const struct got_error *
1431 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
1432     const char *path_packfile, struct got_packidx *packidx)
1433 {
1434 	const struct got_error *err = NULL;
1435 	struct got_pack *pack = NULL;
1436 	struct stat sb;
1437 	size_t i;
1438 
1439 	if (packp)
1440 		*packp = NULL;
1441 
1442 	for (i = 0; i < repo->pack_cache_size; i++) {
1443 		pack = &repo->packs[i];
1444 		if (pack->path_packfile == NULL)
1445 			break;
1446 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1447 			return got_error(GOT_ERR_CACHE_DUP_ENTRY);
1448 	}
1449 
1450 	if (i == repo->pack_cache_size) {
1451 		struct got_pack tmp;
1452 		do {
1453 			i--;
1454 		} while (i > 0 && repo->pinned_pack >= 0 &&
1455 		    i == repo->pinned_pack);
1456 		err = got_pack_close(&repo->packs[i]);
1457 		if (err)
1458 			return err;
1459 		if (ftruncate(repo->packs[i].basefd, 0L) == -1)
1460 			return got_error_from_errno("ftruncate");
1461 		if (ftruncate(repo->packs[i].accumfd, 0L) == -1)
1462 			return got_error_from_errno("ftruncate");
1463 		memcpy(&tmp, &repo->packs[i], sizeof(tmp));
1464 		memcpy(&repo->packs[i], &repo->packs[0],
1465 		    sizeof(repo->packs[i]));
1466 		memcpy(&repo->packs[0], &tmp, sizeof(repo->packs[0]));
1467 		if (repo->pinned_pack == 0)
1468 			repo->pinned_pack = i;
1469 		else if (repo->pinned_pack == i)
1470 			repo->pinned_pack = 0;
1471 		i = 0;
1472 	}
1473 
1474 	pack = &repo->packs[i];
1475 
1476 	pack->path_packfile = strdup(path_packfile);
1477 	if (pack->path_packfile == NULL) {
1478 		err = got_error_from_errno("strdup");
1479 		goto done;
1480 	}
1481 
1482 	err = open_packfile(&pack->fd, repo, path_packfile, packidx);
1483 	if (err)
1484 		goto done;
1485 
1486 	if (fstat(pack->fd, &sb) != 0) {
1487 		err = got_error_from_errno("fstat");
1488 		goto done;
1489 	}
1490 	pack->filesize = sb.st_size;
1491 
1492 	pack->privsep_child = NULL;
1493 
1494 	err = got_delta_cache_alloc(&pack->delta_cache);
1495 	if (err)
1496 		goto done;
1497 
1498 #ifndef GOT_PACK_NO_MMAP
1499 	if (pack->filesize > 0 && pack->filesize <= SIZE_MAX) {
1500 		pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
1501 		    pack->fd, 0);
1502 		if (pack->map == MAP_FAILED) {
1503 			if (errno != ENOMEM) {
1504 				err = got_error_from_errno("mmap");
1505 				goto done;
1506 			}
1507 			pack->map = NULL; /* fall back to read(2) */
1508 		}
1509 	}
1510 #endif
1511 done:
1512 	if (err) {
1513 		if (pack)
1514 			got_pack_close(pack);
1515 	} else if (packp)
1516 		*packp = pack;
1517 	return err;
1518 }
1519 
1520 struct got_pack *
1521 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
1522 {
1523 	struct got_pack *pack = NULL;
1524 	size_t i;
1525 
1526 	for (i = 0; i < repo->pack_cache_size; i++) {
1527 		pack = &repo->packs[i];
1528 		if (pack->path_packfile == NULL)
1529 			break;
1530 		if (strcmp(pack->path_packfile, path_packfile) == 0)
1531 			return pack;
1532 	}
1533 
1534 	return NULL;
1535 }
1536 
1537 const struct got_error *
1538 got_repo_pin_pack(struct got_repository *repo, struct got_packidx *packidx,
1539     struct got_pack *pack)
1540 {
1541 	size_t i;
1542 	int pinned_pack = -1, pinned_packidx = -1;
1543 
1544 	for (i = 0; i < repo->pack_cache_size; i++) {
1545 		if (repo->packidx_cache[i] &&
1546 		    strcmp(repo->packidx_cache[i]->path_packidx,
1547 		    packidx->path_packidx) == 0)
1548 			pinned_packidx = i;
1549 		if (repo->packs[i].path_packfile &&
1550 		    strcmp(repo->packs[i].path_packfile,
1551 		    pack->path_packfile) == 0)
1552 			pinned_pack = i;
1553 	}
1554 
1555 	if (pinned_packidx == -1 || pinned_pack == -1)
1556 		return got_error(GOT_ERR_PIN_PACK);
1557 
1558 	repo->pinned_pack = pinned_pack;
1559 	repo->pinned_packidx = pinned_packidx;
1560 	if (repo->packs[pinned_pack].privsep_child)
1561 		repo->pinned_pid = repo->packs[pinned_pack].privsep_child->pid;
1562 	return NULL;
1563 }
1564 
1565 struct got_pack *
1566 got_repo_get_pinned_pack(struct got_repository *repo)
1567 {
1568 	if (repo->pinned_pack >= 0 &&
1569 	    repo->pinned_pack < repo->pack_cache_size)
1570 		return &repo->packs[repo->pinned_pack];
1571 
1572 	return NULL;
1573 }
1574 
1575 void
1576 got_repo_unpin_pack(struct got_repository *repo)
1577 {
1578 	repo->pinned_packidx = -1;
1579 	repo->pinned_pack = -1;
1580 	repo->pinned_pid = 0;
1581 }
1582 
1583 const struct got_error *
1584 got_repo_init(const char *repo_path, const char *head_name)
1585 {
1586 	const struct got_error *err = NULL;
1587 	const char *dirnames[] = {
1588 		GOT_OBJECTS_DIR,
1589 		GOT_OBJECTS_PACK_DIR,
1590 		GOT_REFS_DIR,
1591 	};
1592 	const char *description_str = "Unnamed repository; "
1593 	    "edit this file 'description' to name the repository.";
1594 	const char *headref = "ref: refs/heads/";
1595 	const char *gitconfig_str = "[core]\n"
1596 	    "\trepositoryformatversion = 0\n"
1597 	    "\tfilemode = true\n"
1598 	    "\tbare = true\n";
1599 	char *headref_str, *path;
1600 	size_t i;
1601 
1602 	if (!got_path_dir_is_empty(repo_path))
1603 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
1604 
1605 	for (i = 0; i < nitems(dirnames); i++) {
1606 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
1607 			return got_error_from_errno("asprintf");
1608 		}
1609 		err = got_path_mkdir(path);
1610 		free(path);
1611 		if (err)
1612 			return err;
1613 	}
1614 
1615 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
1616 		return got_error_from_errno("asprintf");
1617 	err = got_path_create_file(path, description_str);
1618 	free(path);
1619 	if (err)
1620 		return err;
1621 
1622 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
1623 		return got_error_from_errno("asprintf");
1624 	if (asprintf(&headref_str, "%s%s", headref,
1625 	    head_name ? head_name : "main") == -1) {
1626 		free(path);
1627 		return got_error_from_errno("asprintf");
1628 	}
1629 	err = got_path_create_file(path, headref_str);
1630 	free(headref_str);
1631 	free(path);
1632 	if (err)
1633 		return err;
1634 
1635 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
1636 		return got_error_from_errno("asprintf");
1637 	err = got_path_create_file(path, gitconfig_str);
1638 	free(path);
1639 	if (err)
1640 		return err;
1641 
1642 	return NULL;
1643 }
1644 
1645 static const struct got_error *
1646 match_packed_object(struct got_object_id **unique_id,
1647     struct got_repository *repo, const char *id_str_prefix, int obj_type)
1648 {
1649 	const struct got_error *err = NULL;
1650 	struct got_object_id_queue matched_ids;
1651 	struct got_pathlist_entry *pe;
1652 
1653 	STAILQ_INIT(&matched_ids);
1654 
1655 	err = refresh_packidx_paths(repo);
1656 	if (err)
1657 		return err;
1658 
1659 	TAILQ_FOREACH(pe, &repo->packidx_paths, entry) {
1660 		const char *path_packidx = pe->path;
1661 		struct got_packidx *packidx;
1662 		struct got_object_qid *qid;
1663 
1664 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
1665 		    path_packidx, 0);
1666 		if (err)
1667 			break;
1668 
1669 		err = got_packidx_match_id_str_prefix(&matched_ids,
1670 		    packidx, id_str_prefix);
1671 		if (err) {
1672 			got_packidx_close(packidx);
1673 			break;
1674 		}
1675 		err = got_packidx_close(packidx);
1676 		if (err)
1677 			break;
1678 
1679 		STAILQ_FOREACH(qid, &matched_ids, entry) {
1680 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1681 				int matched_type;
1682 				err = got_object_get_type(&matched_type, repo,
1683 				    &qid->id);
1684 				if (err)
1685 					goto done;
1686 				if (matched_type != obj_type)
1687 					continue;
1688 			}
1689 			if (*unique_id == NULL) {
1690 				*unique_id = got_object_id_dup(&qid->id);
1691 				if (*unique_id == NULL) {
1692 					err = got_error_from_errno("malloc");
1693 					goto done;
1694 				}
1695 			} else {
1696 				if (got_object_id_cmp(*unique_id,
1697 				    &qid->id) == 0)
1698 					continue; /* packed multiple times */
1699 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
1700 				goto done;
1701 			}
1702 		}
1703 	}
1704 done:
1705 	got_object_id_queue_free(&matched_ids);
1706 	if (err) {
1707 		free(*unique_id);
1708 		*unique_id = NULL;
1709 	}
1710 	return err;
1711 }
1712 
1713 static const struct got_error *
1714 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1715     const char *object_dir, const char *id_str_prefix, int obj_type,
1716     struct got_repository *repo)
1717 {
1718 	const struct got_error *err = NULL;
1719 	char *path, *id_str = NULL;
1720 	DIR *dir = NULL;
1721 	struct dirent *dent;
1722 	struct got_object_id id;
1723 
1724 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1725 		err = got_error_from_errno("asprintf");
1726 		goto done;
1727 	}
1728 
1729 	dir = opendir(path);
1730 	if (dir == NULL) {
1731 		if (errno == ENOENT) {
1732 			err = NULL;
1733 			goto done;
1734 		}
1735 		err = got_error_from_errno2("opendir", path);
1736 		goto done;
1737 	}
1738 	while ((dent = readdir(dir)) != NULL) {
1739 		int cmp;
1740 
1741 		free(id_str);
1742 		id_str = NULL;
1743 
1744 		if (strcmp(dent->d_name, ".") == 0 ||
1745 		    strcmp(dent->d_name, "..") == 0)
1746 			continue;
1747 
1748 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1749 			err = got_error_from_errno("asprintf");
1750 			goto done;
1751 		}
1752 
1753 		if (!got_parse_sha1_digest(id.sha1, id_str))
1754 			continue;
1755 
1756 		/*
1757 		 * Directory entries do not necessarily appear in
1758 		 * sorted order, so we must iterate over all of them.
1759 		 */
1760 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1761 		if (cmp != 0)
1762 			continue;
1763 
1764 		if (*unique_id == NULL) {
1765 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1766 				int matched_type;
1767 				err = got_object_get_type(&matched_type, repo,
1768 				    &id);
1769 				if (err)
1770 					goto done;
1771 				if (matched_type != obj_type)
1772 					continue;
1773 			}
1774 			*unique_id = got_object_id_dup(&id);
1775 			if (*unique_id == NULL) {
1776 				err = got_error_from_errno("got_object_id_dup");
1777 				goto done;
1778 			}
1779 		} else {
1780 			if (got_object_id_cmp(*unique_id, &id) == 0)
1781 				continue; /* both packed and loose */
1782 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1783 			goto done;
1784 		}
1785 	}
1786 done:
1787 	if (dir && closedir(dir) != 0 && err == NULL)
1788 		err = got_error_from_errno("closedir");
1789 	if (err) {
1790 		free(*unique_id);
1791 		*unique_id = NULL;
1792 	}
1793 	free(id_str);
1794 	free(path);
1795 	return err;
1796 }
1797 
1798 const struct got_error *
1799 got_repo_match_object_id_prefix(struct got_object_id **id,
1800     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1801 {
1802 	const struct got_error *err = NULL;
1803 	char *path_objects = NULL, *object_dir = NULL;
1804 	size_t len;
1805 	int i;
1806 
1807 	*id = NULL;
1808 
1809 	path_objects = got_repo_get_path_objects(repo);
1810 
1811 	len = strlen(id_str_prefix);
1812 	if (len > SHA1_DIGEST_STRING_LENGTH - 1) {
1813 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1814 		goto done;
1815 	}
1816 
1817 	for (i = 0; i < len; i++) {
1818 		if (isxdigit((unsigned char)id_str_prefix[i]))
1819 			continue;
1820 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1821 		goto done;
1822 	}
1823 
1824 	if (len >= 2) {
1825 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1826 		if (err)
1827 			goto done;
1828 		object_dir = strndup(id_str_prefix, 2);
1829 		if (object_dir == NULL) {
1830 			err = got_error_from_errno("strdup");
1831 			goto done;
1832 		}
1833 		err = match_loose_object(id, path_objects, object_dir,
1834 		    id_str_prefix, obj_type, repo);
1835 	} else if (len == 1) {
1836 		int i;
1837 		for (i = 0; i < 0xf; i++) {
1838 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1839 			    == -1) {
1840 				err = got_error_from_errno("asprintf");
1841 				goto done;
1842 			}
1843 			err = match_packed_object(id, repo, object_dir,
1844 			    obj_type);
1845 			if (err)
1846 				goto done;
1847 			err = match_loose_object(id, path_objects, object_dir,
1848 			    id_str_prefix, obj_type, repo);
1849 			if (err)
1850 				goto done;
1851 		}
1852 	} else {
1853 		err = got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
1854 		goto done;
1855 	}
1856 done:
1857 	free(path_objects);
1858 	free(object_dir);
1859 	if (err) {
1860 		free(*id);
1861 		*id = NULL;
1862 	} else if (*id == NULL) {
1863 		switch (obj_type) {
1864 		case GOT_OBJ_TYPE_BLOB:
1865 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1866 			    GOT_OBJ_LABEL_BLOB, id_str_prefix);
1867 			break;
1868 		case GOT_OBJ_TYPE_TREE:
1869 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1870 			    GOT_OBJ_LABEL_TREE, id_str_prefix);
1871 			break;
1872 		case GOT_OBJ_TYPE_COMMIT:
1873 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1874 			    GOT_OBJ_LABEL_COMMIT, id_str_prefix);
1875 			break;
1876 		case GOT_OBJ_TYPE_TAG:
1877 			err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1878 			    GOT_OBJ_LABEL_TAG, id_str_prefix);
1879 			break;
1880 		default:
1881 			err = got_error_path(id_str_prefix, GOT_ERR_NO_OBJ);
1882 			break;
1883 		}
1884 	}
1885 
1886 	return err;
1887 }
1888 
1889 const struct got_error *
1890 got_repo_match_object_id(struct got_object_id **id, char **label,
1891     const char *id_str, int obj_type, struct got_reflist_head *refs,
1892     struct got_repository *repo)
1893 {
1894 	const struct got_error *err;
1895 	struct got_tag_object *tag;
1896 	struct got_reference *ref = NULL;
1897 
1898 	*id = NULL;
1899 	if (label)
1900 		*label = NULL;
1901 
1902 	if (refs) {
1903 		err = got_repo_object_match_tag(&tag, id_str, obj_type,
1904 		    refs, repo);
1905 		if (err == NULL) {
1906 			*id = got_object_id_dup(
1907 			    got_object_tag_get_object_id(tag));
1908 			if (*id == NULL)
1909 				err = got_error_from_errno("got_object_id_dup");
1910 			else if (label && asprintf(label, "refs/tags/%s",
1911 			    got_object_tag_get_name(tag)) == -1) {
1912 				err = got_error_from_errno("asprintf");
1913 				free(*id);
1914 				*id = NULL;
1915 			}
1916 			got_object_tag_close(tag);
1917 			return err;
1918 		} else if (err->code != GOT_ERR_OBJ_TYPE &&
1919 		    err->code != GOT_ERR_NO_OBJ)
1920 			return err;
1921 	}
1922 
1923 	err = got_ref_open(&ref, repo, id_str, 0);
1924 	if (err == NULL) {
1925 		err = got_ref_resolve(id, repo, ref);
1926 		if (err)
1927 			goto done;
1928 		if (label) {
1929 			*label = strdup(got_ref_get_name(ref));
1930 			if (*label == NULL) {
1931 				err = got_error_from_errno("strdup");
1932 				goto done;
1933 			}
1934 		}
1935 	} else {
1936 		if (err->code != GOT_ERR_NOT_REF &&
1937 		    err->code != GOT_ERR_BAD_REF_NAME)
1938 			goto done;
1939 		err = got_repo_match_object_id_prefix(id, id_str,
1940 		    obj_type, repo);
1941 		if (err) {
1942 			if (err->code == GOT_ERR_BAD_OBJ_ID_STR)
1943 				err = got_error_not_ref(id_str);
1944 			goto done;
1945 		}
1946 		if (label) {
1947 			err = got_object_id_str(label, *id);
1948 			if (*label == NULL) {
1949 				err = got_error_from_errno("strdup");
1950 				goto done;
1951 			}
1952 		}
1953 	}
1954 done:
1955 	if (ref)
1956 		got_ref_close(ref);
1957 	return err;
1958 }
1959 
1960 const struct got_error *
1961 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1962     int obj_type, struct got_reflist_head *refs, struct got_repository *repo)
1963 {
1964 	const struct got_error *err = NULL;
1965 	struct got_reflist_entry *re;
1966 	struct got_object_id *tag_id;
1967 	int name_is_absolute = (strncmp(name, "refs/", 5) == 0);
1968 
1969 	*tag = NULL;
1970 
1971 	TAILQ_FOREACH(re, refs, entry) {
1972 		const char *refname;
1973 		refname = got_ref_get_name(re->ref);
1974 		if (got_ref_is_symbolic(re->ref))
1975 			continue;
1976 		if (strncmp(refname, "refs/tags/", 10) != 0)
1977 			continue;
1978 		if (!name_is_absolute)
1979 			refname += strlen("refs/tags/");
1980 		if (strcmp(refname, name) != 0)
1981 			continue;
1982 		err = got_ref_resolve(&tag_id, repo, re->ref);
1983 		if (err)
1984 			break;
1985 		err = got_object_open_as_tag(tag, repo, tag_id);
1986 		free(tag_id);
1987 		if (err)
1988 			break;
1989 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1990 		    got_object_tag_get_object_type(*tag) == obj_type)
1991 			break;
1992 		got_object_tag_close(*tag);
1993 		*tag = NULL;
1994 	}
1995 
1996 	if (err == NULL && *tag == NULL)
1997 		err = got_error_fmt(GOT_ERR_NO_OBJ, "%s %s",
1998 		    GOT_OBJ_LABEL_TAG, name);
1999 	return err;
2000 }
2001 
2002 static const struct got_error *
2003 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
2004     const char *name, mode_t mode, struct got_object_id *blob_id)
2005 {
2006 	const struct got_error *err = NULL;
2007 
2008 	 *new_te = NULL;
2009 
2010 	*new_te = calloc(1, sizeof(**new_te));
2011 	if (*new_te == NULL)
2012 		return got_error_from_errno("calloc");
2013 
2014 	if (strlcpy((*new_te)->name, name, sizeof((*new_te)->name)) >=
2015 	    sizeof((*new_te)->name)) {
2016 		err = got_error(GOT_ERR_NO_SPACE);
2017 		goto done;
2018 	}
2019 
2020 	if (S_ISLNK(mode)) {
2021 		(*new_te)->mode = S_IFLNK;
2022 	} else {
2023 		(*new_te)->mode = S_IFREG;
2024 		(*new_te)->mode |= (mode & (S_IRWXU | S_IRWXG | S_IRWXO));
2025 	}
2026 	memcpy(&(*new_te)->id, blob_id, sizeof((*new_te)->id));
2027 done:
2028 	if (err && *new_te) {
2029 		free(*new_te);
2030 		*new_te = NULL;
2031 	}
2032 	return err;
2033 }
2034 
2035 static const struct got_error *
2036 import_file(struct got_tree_entry **new_te, struct dirent *de,
2037     const char *path, struct got_repository *repo)
2038 {
2039 	const struct got_error *err;
2040 	struct got_object_id *blob_id = NULL;
2041 	char *filepath;
2042 	struct stat sb;
2043 
2044 	if (asprintf(&filepath, "%s%s%s", path,
2045 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2046 		return got_error_from_errno("asprintf");
2047 
2048 	if (lstat(filepath, &sb) != 0) {
2049 		err = got_error_from_errno2("lstat", path);
2050 		goto done;
2051 	}
2052 
2053 	err = got_object_blob_create(&blob_id, filepath, repo);
2054 	if (err)
2055 		goto done;
2056 
2057 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
2058 	    blob_id);
2059 done:
2060 	free(filepath);
2061 	if (err)
2062 		free(blob_id);
2063 	return err;
2064 }
2065 
2066 static const struct got_error *
2067 insert_tree_entry(struct got_tree_entry *new_te,
2068     struct got_pathlist_head *paths)
2069 {
2070 	const struct got_error *err = NULL;
2071 	struct got_pathlist_entry *new_pe;
2072 
2073 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
2074 	if (err)
2075 		return err;
2076 	if (new_pe == NULL)
2077 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
2078 	return NULL;
2079 }
2080 
2081 static const struct got_error *write_tree(struct got_object_id **,
2082     const char *, struct got_pathlist_head *, struct got_repository *,
2083     got_repo_import_cb progress_cb, void *progress_arg);
2084 
2085 static const struct got_error *
2086 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
2087     const char *path, struct got_pathlist_head *ignores,
2088     struct got_repository *repo,
2089     got_repo_import_cb progress_cb, void *progress_arg)
2090 {
2091 	const struct got_error *err;
2092 	struct got_object_id *id = NULL;
2093 	char *subdirpath;
2094 
2095 	if (asprintf(&subdirpath, "%s%s%s", path,
2096 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
2097 		return got_error_from_errno("asprintf");
2098 
2099 	(*new_te) = calloc(1, sizeof(**new_te));
2100 	if (*new_te == NULL)
2101 		return got_error_from_errno("calloc");
2102 	(*new_te)->mode = S_IFDIR;
2103 	if (strlcpy((*new_te)->name, de->d_name, sizeof((*new_te)->name)) >=
2104 	    sizeof((*new_te)->name)) {
2105 		err = got_error(GOT_ERR_NO_SPACE);
2106 		goto done;
2107 	}
2108 	err = write_tree(&id, subdirpath, ignores,  repo,
2109 	    progress_cb, progress_arg);
2110 	if (err)
2111 		goto done;
2112 	memcpy(&(*new_te)->id, id, sizeof((*new_te)->id));
2113 
2114 done:
2115 	free(id);
2116 	free(subdirpath);
2117 	if (err) {
2118 		free(*new_te);
2119 		*new_te = NULL;
2120 	}
2121 	return err;
2122 }
2123 
2124 static const struct got_error *
2125 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
2126     struct got_pathlist_head *ignores, struct got_repository *repo,
2127     got_repo_import_cb progress_cb, void *progress_arg)
2128 {
2129 	const struct got_error *err = NULL;
2130 	DIR *dir;
2131 	struct dirent *de;
2132 	int nentries;
2133 	struct got_tree_entry *new_te = NULL;
2134 	struct got_pathlist_head paths;
2135 	struct got_pathlist_entry *pe;
2136 
2137 	*new_tree_id = NULL;
2138 
2139 	TAILQ_INIT(&paths);
2140 
2141 	dir = opendir(path_dir);
2142 	if (dir == NULL) {
2143 		err = got_error_from_errno2("opendir", path_dir);
2144 		goto done;
2145 	}
2146 
2147 	nentries = 0;
2148 	while ((de = readdir(dir)) != NULL) {
2149 		int ignore = 0;
2150 		int type;
2151 
2152 		if (strcmp(de->d_name, ".") == 0 ||
2153 		    strcmp(de->d_name, "..") == 0)
2154 			continue;
2155 
2156 		err = got_path_dirent_type(&type, path_dir, de);
2157 		if (err)
2158 			goto done;
2159 
2160 		TAILQ_FOREACH(pe, ignores, entry) {
2161 			if (type == DT_DIR && pe->path_len > 0 &&
2162 			    pe->path[pe->path_len - 1] == '/') {
2163 				char stripped[PATH_MAX];
2164 
2165 				if (strlcpy(stripped, pe->path,
2166 				    sizeof(stripped)) >= sizeof(stripped)) {
2167 					err = got_error(GOT_ERR_NO_SPACE);
2168 					goto done;
2169 				}
2170 				got_path_strip_trailing_slashes(stripped);
2171 				if (fnmatch(stripped, de->d_name, 0) == 0) {
2172 					ignore = 1;
2173 					break;
2174 				}
2175 			} else if (fnmatch(pe->path, de->d_name, 0) == 0) {
2176 				ignore = 1;
2177 				break;
2178 			}
2179 		}
2180 		if (ignore)
2181 			continue;
2182 
2183 		if (type == DT_DIR) {
2184 			err = import_subdir(&new_te, de, path_dir,
2185 			    ignores, repo, progress_cb, progress_arg);
2186 			if (err) {
2187 				if (err->code != GOT_ERR_NO_TREE_ENTRY)
2188 					goto done;
2189 				err = NULL;
2190 				continue;
2191 			}
2192 		} else if (type == DT_REG || type == DT_LNK) {
2193 			err = import_file(&new_te, de, path_dir, repo);
2194 			if (err)
2195 				goto done;
2196 		} else
2197 			continue;
2198 
2199 		err = insert_tree_entry(new_te, &paths);
2200 		if (err)
2201 			goto done;
2202 		nentries++;
2203 	}
2204 
2205 	if (TAILQ_EMPTY(&paths)) {
2206 		err = got_error_msg(GOT_ERR_NO_TREE_ENTRY,
2207 		    "cannot create tree without any entries");
2208 		goto done;
2209 	}
2210 
2211 	TAILQ_FOREACH(pe, &paths, entry) {
2212 		struct got_tree_entry *te = pe->data;
2213 		char *path;
2214 		if (!S_ISREG(te->mode) && !S_ISLNK(te->mode))
2215 			continue;
2216 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
2217 			err = got_error_from_errno("asprintf");
2218 			goto done;
2219 		}
2220 		err = (*progress_cb)(progress_arg, path);
2221 		free(path);
2222 		if (err)
2223 			goto done;
2224 	}
2225 
2226 	err = got_object_tree_create(new_tree_id, &paths, nentries, repo);
2227 done:
2228 	if (dir)
2229 		closedir(dir);
2230 	got_pathlist_free(&paths, GOT_PATHLIST_FREE_NONE);
2231 	return err;
2232 }
2233 
2234 const struct got_error *
2235 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
2236     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
2237     struct got_repository *repo, got_repo_import_cb progress_cb,
2238     void *progress_arg)
2239 {
2240 	const struct got_error *err;
2241 	struct got_object_id *new_tree_id;
2242 
2243 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
2244 	    progress_cb, progress_arg);
2245 	if (err)
2246 		return err;
2247 
2248 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
2249 	    author, time(NULL), author, time(NULL), logmsg, repo);
2250 	free(new_tree_id);
2251 	return err;
2252 }
2253 
2254 const struct got_error *
2255 got_repo_get_loose_object_info(int *nobjects, off_t *ondisk_size,
2256     struct got_repository *repo)
2257 {
2258 	const struct got_error *err = NULL;
2259 	char *path_objects = NULL, *path = NULL;
2260 	DIR *dir = NULL;
2261 	struct got_object_id id;
2262 	int i;
2263 
2264 	*nobjects = 0;
2265 	*ondisk_size = 0;
2266 
2267 	path_objects = got_repo_get_path_objects(repo);
2268 	if (path_objects == NULL)
2269 		return got_error_from_errno("got_repo_get_path_objects");
2270 
2271 	for (i = 0; i <= 0xff; i++) {
2272 		struct dirent *dent;
2273 
2274 		if (asprintf(&path, "%s/%.2x", path_objects, i) == -1) {
2275 			err = got_error_from_errno("asprintf");
2276 			break;
2277 		}
2278 
2279 		dir = opendir(path);
2280 		if (dir == NULL) {
2281 			if (errno == ENOENT) {
2282 				err = NULL;
2283 				continue;
2284 			}
2285 			err = got_error_from_errno2("opendir", path);
2286 			break;
2287 		}
2288 
2289 		while ((dent = readdir(dir)) != NULL) {
2290 			char *id_str;
2291 			int fd;
2292 			struct stat sb;
2293 
2294 			if (strcmp(dent->d_name, ".") == 0 ||
2295 			    strcmp(dent->d_name, "..") == 0)
2296 				continue;
2297 
2298 			if (asprintf(&id_str, "%.2x%s", i, dent->d_name) == -1) {
2299 				err = got_error_from_errno("asprintf");
2300 				goto done;
2301 			}
2302 
2303 			if (!got_parse_sha1_digest(id.sha1, id_str)) {
2304 				free(id_str);
2305 				continue;
2306 			}
2307 			free(id_str);
2308 
2309 			err = got_object_open_loose_fd(&fd, &id, repo);
2310 			if (err)
2311 				goto done;
2312 
2313 			if (fstat(fd, &sb) == -1) {
2314 				err = got_error_from_errno("fstat");
2315 				close(fd);
2316 				goto done;
2317 			}
2318 			(*nobjects)++;
2319 			(*ondisk_size) += sb.st_size;
2320 
2321 			if (close(fd) == -1) {
2322 				err = got_error_from_errno("close");
2323 				goto done;
2324 			}
2325 		}
2326 
2327 		if (closedir(dir) != 0) {
2328 			err = got_error_from_errno("closedir");
2329 			goto done;
2330 		}
2331 		dir = NULL;
2332 
2333 		free(path);
2334 		path = NULL;
2335 	}
2336 done:
2337 	if (dir && closedir(dir) != 0 && err == NULL)
2338 		err = got_error_from_errno("closedir");
2339 
2340 	if (err) {
2341 		*nobjects = 0;
2342 		*ondisk_size = 0;
2343 	}
2344 	free(path_objects);
2345 	free(path);
2346 	return err;
2347 }
2348 
2349 const struct got_error *
2350 got_repo_get_packfile_info(int *npackfiles, int *nobjects,
2351     off_t *total_packsize, struct got_repository *repo)
2352 {
2353 	const struct got_error *err = NULL;
2354 	DIR *packdir = NULL;
2355 	struct dirent *dent;
2356 	struct got_packidx *packidx = NULL;
2357 	char *path_packidx;
2358 	char *path_packfile;
2359 	int packdir_fd;
2360 	struct stat sb;
2361 
2362 	*npackfiles = 0;
2363 	*nobjects = 0;
2364 	*total_packsize = 0;
2365 
2366 	packdir_fd = openat(got_repo_get_fd(repo),
2367 	    GOT_OBJECTS_PACK_DIR, O_DIRECTORY);
2368 	if (packdir_fd == -1) {
2369 		return got_error_from_errno_fmt("openat: %s/%s",
2370 		    got_repo_get_path_git_dir(repo),
2371 		    GOT_OBJECTS_PACK_DIR);
2372 	}
2373 
2374 	packdir = fdopendir(packdir_fd);
2375 	if (packdir == NULL) {
2376 		err = got_error_from_errno("fdopendir");
2377 		goto done;
2378 	}
2379 
2380 	while ((dent = readdir(packdir)) != NULL) {
2381 		if (!got_repo_is_packidx_filename(dent->d_name, dent->d_namlen))
2382 			continue;
2383 
2384 		if (asprintf(&path_packidx, "%s/%s", GOT_OBJECTS_PACK_DIR,
2385 		    dent->d_name) == -1) {
2386 			err = got_error_from_errno("asprintf");
2387 			goto done;
2388 		}
2389 
2390 		err = got_packidx_open(&packidx, got_repo_get_fd(repo),
2391 		    path_packidx, 0);
2392 		free(path_packidx);
2393 		if (err)
2394 			goto done;
2395 
2396 		if (fstat(packidx->fd, &sb) == -1)
2397 			goto done;
2398 		*total_packsize += sb.st_size;
2399 
2400 		err = got_packidx_get_packfile_path(&path_packfile,
2401 		    packidx->path_packidx);
2402 		if (err)
2403 			goto done;
2404 
2405 		if (fstatat(got_repo_get_fd(repo), path_packfile, &sb,
2406 		    0) == -1) {
2407 			free(path_packfile);
2408 			goto done;
2409 		}
2410 		free(path_packfile);
2411 		*total_packsize += sb.st_size;
2412 
2413 		*nobjects += be32toh(packidx->hdr.fanout_table[0xff]);
2414 
2415 		(*npackfiles)++;
2416 
2417 		got_packidx_close(packidx);
2418 		packidx = NULL;
2419 	}
2420 done:
2421 	if (packidx)
2422 		got_packidx_close(packidx);
2423 	if (packdir && closedir(packdir) != 0 && err == NULL)
2424 		err = got_error_from_errno("closedir");
2425 	if (err) {
2426 		*npackfiles = 0;
2427 		*nobjects = 0;
2428 		*total_packsize = 0;
2429 	}
2430 	return err;
2431 }
2432 
2433 RB_GENERATE(got_packidx_bloom_filter_tree, got_packidx_bloom_filter, entry,
2434     got_packidx_bloom_filter_cmp);