1 /*
2  * Copyright (c) 2018, 2019 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/queue.h>
19 #include <sys/uio.h>
20 #include <sys/stat.h>
21 #include <sys/mman.h>
22 #include <sys/syslimits.h>
23 
24 #include <ctype.h>
25 #include <fcntl.h>
26 #include <fnmatch.h>
27 #include <limits.h>
28 #include <dirent.h>
29 #include <stdlib.h>
30 #include <stdio.h>
31 #include <sha1.h>
32 #include <string.h>
33 #include <zlib.h>
34 #include <errno.h>
35 #include <libgen.h>
36 #include <stdint.h>
37 #include <imsg.h>
38 #include <uuid.h>
39 
40 #include "got_error.h"
41 #include "got_reference.h"
42 #include "got_repository.h"
43 #include "got_path.h"
44 #include "got_cancel.h"
45 #include "got_worktree.h"
46 #include "got_object.h"
47 
48 #include "got_lib_delta.h"
49 #include "got_lib_inflate.h"
50 #include "got_lib_object.h"
51 #include "got_lib_object_parse.h"
52 #include "got_lib_object_create.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_privsep.h"
55 #include "got_lib_worktree.h"
56 #include "got_lib_sha1.h"
57 #include "got_lib_object_cache.h"
58 #include "got_lib_repository.h"
59 #include "got_lib_gitconfig.h"
60 
61 #ifndef nitems
62 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
63 #endif
64 
65 #define GOT_GIT_DIR	".git"
66 
67 /* Mandatory files and directories inside the git directory. */
68 #define GOT_OBJECTS_DIR		"objects"
69 #define GOT_REFS_DIR		"refs"
70 #define GOT_HEAD_FILE		"HEAD"
71 #define GOT_GITCONFIG		"config"
72 
73 /* Other files and directories inside the git directory. */
74 #define GOT_FETCH_HEAD_FILE	"FETCH_HEAD"
75 #define GOT_ORIG_HEAD_FILE	"ORIG_HEAD"
76 #define GOT_OBJECTS_PACK_DIR	"objects/pack"
77 #define GOT_PACKED_REFS_FILE	"packed-refs"
78 
79 const char *
80 got_repo_get_path(struct got_repository *repo)
81 {
82 	return repo->path;
83 }
84 
85 const char *
86 got_repo_get_path_git_dir(struct got_repository *repo)
87 {
88 	return repo->path_git_dir;
89 }
90 
91 int
92 got_repo_is_bare(struct got_repository *repo)
93 {
94 	return (strcmp(repo->path, repo->path_git_dir) == 0);
95 }
96 
97 static char *
98 get_path_git_child(struct got_repository *repo, const char *basename)
99 {
100 	char *path_child;
101 	
102 	if (asprintf(&path_child, "%s/%s", repo->path_git_dir,
103 	    basename) == -1)
104 		return NULL;
105 
106 	return path_child;
107 }
108 
109 char *
110 got_repo_get_path_objects(struct got_repository *repo)
111 {
112 	return get_path_git_child(repo, GOT_OBJECTS_DIR);
113 }
114 
115 char *
116 got_repo_get_path_objects_pack(struct got_repository *repo)
117 {
118 	return get_path_git_child(repo, GOT_OBJECTS_PACK_DIR);
119 }
120 
121 char *
122 got_repo_get_path_refs(struct got_repository *repo)
123 {
124 	return get_path_git_child(repo, GOT_REFS_DIR);
125 }
126 
127 char *
128 got_repo_get_path_packed_refs(struct got_repository *repo)
129 {
130 	return get_path_git_child(repo, GOT_PACKED_REFS_FILE);
131 }
132 
133 static char *
134 get_path_head(struct got_repository *repo)
135 {
136 	return get_path_git_child(repo, GOT_HEAD_FILE);
137 }
138 
139 static const struct got_error *
140 get_path_gitconfig(char **p, struct got_repository *repo)
141 {
142 	*p = get_path_git_child(repo, GOT_GITCONFIG);
143 	if (*p == NULL)
144 		return got_error_from_errno("asprintf");
145 	return NULL;
146 }
147 
148 static int
149 is_git_repo(struct got_repository *repo)
150 {
151 	const char *path_git = got_repo_get_path_git_dir(repo);
152 	char *path_objects = got_repo_get_path_objects(repo);
153 	char *path_refs = got_repo_get_path_refs(repo);
154 	char *path_head = get_path_head(repo);
155 	int ret = 0;
156 	struct stat sb;
157 	struct got_reference *head_ref;
158 
159 	if (lstat(path_git, &sb) == -1)
160 		goto done;
161 	if (!S_ISDIR(sb.st_mode))
162 		goto done;
163 
164 	if (lstat(path_objects, &sb) == -1)
165 		goto done;
166 	if (!S_ISDIR(sb.st_mode))
167 		goto done;
168 
169 	if (lstat(path_refs, &sb) == -1)
170 		goto done;
171 	if (!S_ISDIR(sb.st_mode))
172 		goto done;
173 
174 	if (lstat(path_head, &sb) == -1)
175 		goto done;
176 	if (!S_ISREG(sb.st_mode))
177 		goto done;
178 
179 	/* Check if the HEAD reference can be opened. */
180 	if (got_ref_open(&head_ref, repo, GOT_REF_HEAD, 0) != NULL)
181 		goto done;
182 	got_ref_close(head_ref);
183 
184 	ret = 1;
185 done:
186 	free(path_objects);
187 	free(path_refs);
188 	free(path_head);
189 	return ret;
190 
191 }
192 
193 const struct got_error *
194 got_repo_cache_object(struct got_repository *repo, struct got_object_id *id,
195     struct got_object *obj)
196 {
197 #ifndef GOT_NO_OBJ_CACHE
198 	const struct got_error *err = NULL;
199 	err = got_object_cache_add(&repo->objcache, id, obj);
200 	if (err) {
201 		if (err->code == GOT_ERR_OBJ_EXISTS ||
202 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
203 			err = NULL;
204 		return err;
205 	}
206 	obj->refcnt++;
207 #endif
208 	return NULL;
209 }
210 
211 struct got_object *
212 got_repo_get_cached_object(struct got_repository *repo,
213     struct got_object_id *id)
214 {
215 	return (struct got_object *)got_object_cache_get(&repo->objcache, id);
216 }
217 
218 const struct got_error *
219 got_repo_cache_tree(struct got_repository *repo, struct got_object_id *id,
220     struct got_tree_object *tree)
221 {
222 #ifndef GOT_NO_OBJ_CACHE
223 	const struct got_error *err = NULL;
224 	err = got_object_cache_add(&repo->treecache, id, tree);
225 	if (err) {
226 		if (err->code == GOT_ERR_OBJ_EXISTS ||
227 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
228 			err = NULL;
229 		return err;
230 	}
231 	tree->refcnt++;
232 #endif
233 	return NULL;
234 }
235 
236 struct got_tree_object *
237 got_repo_get_cached_tree(struct got_repository *repo,
238     struct got_object_id *id)
239 {
240 	return (struct got_tree_object *)got_object_cache_get(
241 	    &repo->treecache, id);
242 }
243 
244 const struct got_error *
245 got_repo_cache_commit(struct got_repository *repo, struct got_object_id *id,
246     struct got_commit_object *commit)
247 {
248 #ifndef GOT_NO_OBJ_CACHE
249 	const struct got_error *err = NULL;
250 	err = got_object_cache_add(&repo->commitcache, id, commit);
251 	if (err) {
252 		if (err->code == GOT_ERR_OBJ_EXISTS ||
253 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
254 			err = NULL;
255 		return err;
256 	}
257 	commit->refcnt++;
258 #endif
259 	return NULL;
260 }
261 
262 struct got_commit_object *
263 got_repo_get_cached_commit(struct got_repository *repo,
264     struct got_object_id *id)
265 {
266 	return (struct got_commit_object *)got_object_cache_get(
267 	    &repo->commitcache, id);
268 }
269 
270 const struct got_error *
271 got_repo_cache_tag(struct got_repository *repo, struct got_object_id *id,
272     struct got_tag_object *tag)
273 {
274 #ifndef GOT_NO_OBJ_CACHE
275 	const struct got_error *err = NULL;
276 	err = got_object_cache_add(&repo->tagcache, id, tag);
277 	if (err) {
278 		if (err->code == GOT_ERR_OBJ_EXISTS ||
279 		    err->code == GOT_ERR_OBJ_TOO_LARGE)
280 			err = NULL;
281 		return err;
282 	}
283 	tag->refcnt++;
284 #endif
285 	return NULL;
286 }
287 
288 struct got_tag_object *
289 got_repo_get_cached_tag(struct got_repository *repo, struct got_object_id *id)
290 {
291 	return (struct got_tag_object *)got_object_cache_get(
292 	    &repo->tagcache, id);
293 }
294 
295 const struct got_error *
296 open_repo(struct got_repository *repo, const char *path)
297 {
298 	const struct got_error *err = NULL;
299 
300 	/* bare git repository? */
301 	repo->path_git_dir = strdup(path);
302 	if (repo->path_git_dir == NULL)
303 		return got_error_from_errno("strdup");
304 	if (is_git_repo(repo)) {
305 		repo->path = strdup(repo->path_git_dir);
306 		if (repo->path == NULL) {
307 			err = got_error_from_errno("strdup");
308 			goto done;
309 		}
310 		return NULL;
311 	}
312 
313 	/* git repository with working tree? */
314 	free(repo->path_git_dir);
315 	if (asprintf(&repo->path_git_dir, "%s/%s", path, GOT_GIT_DIR) == -1) {
316 		err = got_error_from_errno("asprintf");
317 		goto done;
318 	}
319 	if (is_git_repo(repo)) {
320 		repo->path = strdup(path);
321 		if (repo->path == NULL) {
322 			err = got_error_from_errno("strdup");
323 			goto done;
324 		}
325 		return NULL;
326 	}
327 
328 	err = got_error(GOT_ERR_NOT_GIT_REPO);
329 done:
330 	if (err) {
331 		free(repo->path);
332 		repo->path = NULL;
333 		free(repo->path_git_dir);
334 		repo->path_git_dir = NULL;
335 	}
336 	return err;
337 }
338 
339 const struct got_error *
340 got_repo_open(struct got_repository **repop, const char *path)
341 {
342 	struct got_repository *repo = NULL;
343 	const struct got_error *err = NULL;
344 	char *abspath, *gitconfig_path = NULL;
345 	int i, tried_root = 0;
346 
347 	*repop = NULL;
348 
349 	if (got_path_is_absolute(path))
350 		abspath = strdup(path);
351 	else
352 		abspath = got_path_get_absolute(path);
353 	if (abspath == NULL)
354 		return got_error(GOT_ERR_BAD_PATH);
355 
356 	repo = calloc(1, sizeof(*repo));
357 	if (repo == NULL) {
358 		err = got_error_from_errno("calloc");
359 		goto done;
360 	}
361 
362 	for (i = 0; i < nitems(repo->privsep_children); i++) {
363 		memset(&repo->privsep_children[i], 0,
364 		    sizeof(repo->privsep_children[0]));
365 		repo->privsep_children[i].imsg_fd = -1;
366 	}
367 
368 	err = got_object_cache_init(&repo->objcache,
369 	    GOT_OBJECT_CACHE_TYPE_OBJ);
370 	if (err)
371 		goto done;
372 	err = got_object_cache_init(&repo->treecache,
373 	    GOT_OBJECT_CACHE_TYPE_TREE);
374 	if (err)
375 		goto done;
376 	err = got_object_cache_init(&repo->commitcache,
377 	    GOT_OBJECT_CACHE_TYPE_COMMIT);
378 	if (err)
379 		goto done;
380 	err = got_object_cache_init(&repo->tagcache,
381 	    GOT_OBJECT_CACHE_TYPE_TAG);
382 	if (err)
383 		goto done;
384 
385 	path = realpath(abspath, NULL);
386 	if (path == NULL) {
387 		err = got_error_from_errno2("realpath", abspath);
388 		goto done;
389 	}
390 
391 	do {
392 		err = open_repo(repo, path);
393 		if (err == NULL)
394 			break;
395 		if (err->code != GOT_ERR_NOT_GIT_REPO)
396 			break;
397 		if (path[0] == '/' && path[1] == '\0') {
398 			if (tried_root) {
399 				err = got_error(GOT_ERR_NOT_GIT_REPO);
400 				break;
401 			}
402 			tried_root = 1;
403 		}
404 		path = dirname(path);
405 		if (path == NULL)
406 			err = got_error_from_errno2("dirname", path);
407 	} while (path);
408 
409 	err = get_path_gitconfig(&gitconfig_path, repo);
410 	if (err)
411 		goto done;
412 
413 #ifdef notyet
414 	err = got_gitconfig_open(&repo->gitconfig, gitconfig_path);
415 	if (err)
416 		goto done;
417 #else
418 	repo->gitconfig = NULL;
419 #endif
420 done:
421 	if (err)
422 		got_repo_close(repo);
423 	else
424 		*repop = repo;
425 	free(abspath);
426 	free(gitconfig_path);
427 	return err;
428 }
429 
430 const struct got_error *
431 got_repo_close(struct got_repository *repo)
432 {
433 	const struct got_error *err = NULL, *child_err;
434 	int i;
435 
436 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
437 		if (repo->packidx_cache[i] == NULL)
438 			break;
439 		got_packidx_close(repo->packidx_cache[i]);
440 	}
441 
442 	for (i = 0; i < nitems(repo->packs); i++) {
443 		if (repo->packs[i].path_packfile == NULL)
444 			break;
445 		got_pack_close(&repo->packs[i]);
446 	}
447 
448 	free(repo->path);
449 	free(repo->path_git_dir);
450 
451 	got_object_cache_close(&repo->objcache);
452 	got_object_cache_close(&repo->treecache);
453 	got_object_cache_close(&repo->commitcache);
454 	got_object_cache_close(&repo->tagcache);
455 
456 	for (i = 0; i < nitems(repo->privsep_children); i++) {
457 		if (repo->privsep_children[i].imsg_fd == -1)
458 			continue;
459 		imsg_clear(repo->privsep_children[i].ibuf);
460 		free(repo->privsep_children[i].ibuf);
461 		err = got_privsep_send_stop(repo->privsep_children[i].imsg_fd);
462 		child_err = got_privsep_wait_for_child(
463 		    repo->privsep_children[i].pid);
464 		if (child_err && err == NULL)
465 			err = child_err;
466 		if (close(repo->privsep_children[i].imsg_fd) != 0 &&
467 		    err == NULL)
468 			err = got_error_from_errno("close");
469 	}
470 	if (repo->gitconfig)
471 		got_gitconfig_close(repo->gitconfig);
472 	free(repo);
473 
474 	return err;
475 }
476 
477 const struct got_error *
478 got_repo_map_path(char **in_repo_path, struct got_repository *repo,
479     const char *input_path, int check_disk)
480 {
481 	const struct got_error *err = NULL;
482 	const char *repo_abspath = NULL;
483 	size_t repolen, cwdlen, len;
484 	char *cwd, *canonpath, *path = NULL;
485 
486 	*in_repo_path = NULL;
487 
488 	cwd = getcwd(NULL, 0);
489 	if (cwd == NULL)
490 		return got_error_from_errno("getcwd");
491 
492 	canonpath = strdup(input_path);
493 	if (canonpath == NULL) {
494 		err = got_error_from_errno("strdup");
495 		goto done;
496 	}
497 	err = got_canonpath(input_path, canonpath, strlen(canonpath) + 1);
498 	if (err)
499 		goto done;
500 
501 	repo_abspath = got_repo_get_path(repo);
502 
503 	if (!check_disk || canonpath[0] == '\0') {
504 		path = strdup(canonpath);
505 		if (path == NULL) {
506 			err = got_error_from_errno("strdup");
507 			goto done;
508 		}
509 	} else {
510 		int is_repo_child = 0, is_cwd_child = 0;
511 
512 		path = realpath(canonpath, NULL);
513 		if (path == NULL) {
514 			if (errno != ENOENT) {
515 				err = got_error_from_errno2("realpath",
516 				    canonpath);
517 				goto done;
518 			}
519 			/*
520 			 * Path is not on disk.
521 			 * Assume it is already relative to repository root.
522 			 */
523 			path = strdup(canonpath);
524 			if (path == NULL) {
525 				err = got_error_from_errno("strdup");
526 				goto done;
527 			}
528 		}
529 
530 		repolen = strlen(repo_abspath);
531 		cwdlen = strlen(cwd);
532 		len = strlen(path);
533 
534 		if (len > repolen && strncmp(path, repo_abspath, repolen) == 0)
535 			is_repo_child = 1;
536 		if (len > cwdlen && strncmp(path, cwd, cwdlen) == 0)
537 			is_cwd_child = 1;
538 
539 		if (strcmp(path, repo_abspath) == 0) {
540 			free(path);
541 			path = strdup("");
542 			if (path == NULL) {
543 				err = got_error_from_errno("strdup");
544 				goto done;
545 			}
546 		} else if (is_repo_child && is_cwd_child) {
547 			char *child;
548 			/* Strip common prefix with repository path. */
549 			err = got_path_skip_common_ancestor(&child,
550 			    repo_abspath, path);
551 			if (err)
552 				goto done;
553 			free(path);
554 			path = child;
555 		} else if (is_repo_child) {
556 			/* Matched an on-disk path inside repository. */
557 			if (got_repo_is_bare(repo)) {
558 				/*
559 				 * Matched an on-disk path inside repository
560 				 * database. Treat as repository-relative.
561 				 */
562 			} else {
563 				char *child;
564 				/* Strip common prefix with repository path. */
565 				err = got_path_skip_common_ancestor(&child,
566 				    repo_abspath, path);
567 				if (err)
568 					goto done;
569 				free(path);
570 				path = child;
571 			}
572 		} else if (is_cwd_child) {
573 			char *child;
574 			/* Strip common prefix with cwd. */
575 			err = got_path_skip_common_ancestor(&child, cwd,
576 			    path);
577 			if (err)
578 				goto done;
579 			free(path);
580 			path = child;
581 		} else {
582 			/*
583 			 * Matched unrelated on-disk path.
584 			 * Treat it as repository-relative.
585 			 */
586 		}
587 	}
588 
589 	/* Make in-repository path absolute */
590 	if (path[0] != '/') {
591 		char *abspath;
592 		if (asprintf(&abspath, "/%s", path) == -1) {
593 			err = got_error_from_errno("asprintf");
594 			goto done;
595 		}
596 		free(path);
597 		path = abspath;
598 	}
599 
600 done:
601 	free(cwd);
602 	free(canonpath);
603 	if (err)
604 		free(path);
605 	else
606 		*in_repo_path = path;
607 	return err;
608 }
609 
610 const struct got_error *
611 got_repo_cache_packidx(struct got_repository *repo, struct got_packidx *packidx)
612 {
613 	const struct got_error *err = NULL;
614 	int i;
615 
616 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
617 		if (repo->packidx_cache[i] == NULL)
618 			break;
619 	}
620 	if (i == nitems(repo->packidx_cache)) {
621 		err = got_packidx_close(repo->packidx_cache[i - 1]);
622 		if (err)
623 			return err;
624 	}
625 
626 	/*
627 	 * Insert the new pack index at the front so it will
628 	 * be searched first in the future.
629 	 */
630 	memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
631 	    sizeof(repo->packidx_cache) -
632 	    sizeof(repo->packidx_cache[0]));
633 	repo->packidx_cache[0] = packidx;
634 
635 	return NULL;
636 }
637 
638 static int
639 is_packidx_filename(const char *name, size_t len)
640 {
641 	if (len != GOT_PACKIDX_NAMELEN)
642 		return 0;
643 
644 	if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
645 		return 0;
646 
647 	if (strcmp(name + strlen(GOT_PACK_PREFIX) +
648 	    SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
649 		return 0;
650 
651 	return 1;
652 }
653 
654 const struct got_error *
655 got_repo_search_packidx(struct got_packidx **packidx, int *idx,
656     struct got_repository *repo, struct got_object_id *id)
657 {
658 	const struct got_error *err;
659 	char *path_packdir;
660 	DIR *packdir;
661 	struct dirent *dent;
662 	char *path_packidx;
663 	int i;
664 
665 	/* Search pack index cache. */
666 	for (i = 0; i < nitems(repo->packidx_cache); i++) {
667 		if (repo->packidx_cache[i] == NULL)
668 			break;
669 		*idx = got_packidx_get_object_idx(repo->packidx_cache[i], id);
670 		if (*idx != -1) {
671 			*packidx = repo->packidx_cache[i];
672 			return NULL;
673 		}
674 	}
675 	/* No luck. Search the filesystem. */
676 
677 	path_packdir = got_repo_get_path_objects_pack(repo);
678 	if (path_packdir == NULL)
679 		return got_error_from_errno("got_repo_get_path_objects_pack");
680 
681 	packdir = opendir(path_packdir);
682 	if (packdir == NULL) {
683 		if (errno == ENOENT)
684 			err = got_error_no_obj(id);
685 		else
686 			err = got_error_from_errno2("opendir", path_packdir);
687 		goto done;
688 	}
689 
690 	while ((dent = readdir(packdir)) != NULL) {
691 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
692 			continue;
693 
694 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
695 		    dent->d_name) == -1) {
696 			err = got_error_from_errno("asprintf");
697 			goto done;
698 		}
699 
700 		err = got_packidx_open(packidx, path_packidx, 0);
701 		free(path_packidx);
702 		if (err)
703 			goto done;
704 
705 		*idx = got_packidx_get_object_idx(*packidx, id);
706 		if (*idx != -1) {
707 			err = NULL; /* found the object */
708 			err = got_repo_cache_packidx(repo, *packidx);
709 			goto done;
710 		}
711 
712 		err = got_packidx_close(*packidx);
713 		*packidx = NULL;
714 		if (err)
715 			goto done;
716 	}
717 
718 	err = got_error_no_obj(id);
719 done:
720 	free(path_packdir);
721 	if (packdir && closedir(packdir) != 0 && err == NULL)
722 		err = got_error_from_errno("closedir");
723 	return err;
724 }
725 
726 static const struct got_error *
727 read_packfile_hdr(int fd, struct got_packidx *packidx)
728 {
729 	const struct got_error *err = NULL;
730 	uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
731 	struct got_packfile_hdr hdr;
732 	ssize_t n;
733 
734 	n = read(fd, &hdr, sizeof(hdr));
735 	if (n < 0)
736 		return got_error_from_errno("read");
737 	if (n != sizeof(hdr))
738 		return got_error(GOT_ERR_BAD_PACKFILE);
739 
740 	if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
741 	    betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
742 	    betoh32(hdr.nobjects) != totobj)
743 		err = got_error(GOT_ERR_BAD_PACKFILE);
744 
745 	return err;
746 }
747 
748 static const struct got_error *
749 open_packfile(int *fd, const char *path_packfile, struct got_packidx *packidx)
750 {
751 	const struct got_error *err = NULL;
752 
753 	*fd = open(path_packfile, O_RDONLY | O_NOFOLLOW);
754 	if (*fd == -1)
755 		return got_error_from_errno2("open", path_packfile);
756 
757 	if (packidx) {
758 		err = read_packfile_hdr(*fd, packidx);
759 		if (err) {
760 			close(*fd);
761 			*fd = -1;
762 		}
763 	}
764 
765 	return err;
766 }
767 
768 const struct got_error *
769 got_repo_cache_pack(struct got_pack **packp, struct got_repository *repo,
770     const char *path_packfile, struct got_packidx *packidx)
771 {
772 	const struct got_error *err = NULL;
773 	struct got_pack *pack = NULL;
774 	struct stat sb;
775 	int i;
776 
777 	if (packp)
778 		*packp = NULL;
779 
780 	for (i = 0; i < nitems(repo->packs); i++) {
781 		pack = &repo->packs[i];
782 		if (pack->path_packfile == NULL)
783 			break;
784 		if (strcmp(pack->path_packfile, path_packfile) == 0)
785 			return NULL;
786 	}
787 
788 	if (i == nitems(repo->packs) - 1) {
789 		err = got_pack_close(&repo->packs[i - 1]);
790 		if (err)
791 			return err;
792 		memmove(&repo->packs[1], &repo->packs[0],
793 		    sizeof(repo->packs) - sizeof(repo->packs[0]));
794 		i = 0;
795 	}
796 
797 	pack = &repo->packs[i];
798 
799 	pack->path_packfile = strdup(path_packfile);
800 	if (pack->path_packfile == NULL) {
801 		err = got_error_from_errno("strdup");
802 		goto done;
803 	}
804 
805 	err = open_packfile(&pack->fd, path_packfile, packidx);
806 	if (err)
807 		goto done;
808 
809 	if (fstat(pack->fd, &sb) != 0) {
810 		err = got_error_from_errno("fstat");
811 		goto done;
812 	}
813 	pack->filesize = sb.st_size;
814 
815 	pack->privsep_child = NULL;
816 
817 #ifndef GOT_PACK_NO_MMAP
818 	pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
819 	    pack->fd, 0);
820 	if (pack->map == MAP_FAILED) {
821 		if (errno != ENOMEM) {
822 			err = got_error_from_errno("mmap");
823 			goto done;
824 		}
825 		pack->map = NULL; /* fall back to read(2) */
826 	}
827 #endif
828 done:
829 	if (err) {
830 		if (pack) {
831 			free(pack->path_packfile);
832 			memset(pack, 0, sizeof(*pack));
833 		}
834 	} else if (packp)
835 		*packp = pack;
836 	return err;
837 }
838 
839 struct got_pack *
840 got_repo_get_cached_pack(struct got_repository *repo, const char *path_packfile)
841 {
842 	struct got_pack *pack = NULL;
843 	int i;
844 
845 	for (i = 0; i < nitems(repo->packs); i++) {
846 		pack = &repo->packs[i];
847 		if (pack->path_packfile == NULL)
848 			break;
849 		if (strcmp(pack->path_packfile, path_packfile) == 0)
850 			return pack;
851 	}
852 
853 	return NULL;
854 }
855 
856 const struct got_error *
857 got_repo_init(const char *repo_path)
858 {
859 	const struct got_error *err = NULL;
860 	const char *dirnames[] = {
861 		GOT_OBJECTS_DIR,
862 		GOT_OBJECTS_PACK_DIR,
863 		GOT_REFS_DIR,
864 	};
865 	const char *description_str = "Unnamed repository; "
866 	    "edit this file 'description' to name the repository.";
867 	const char *headref_str = "ref: refs/heads/master";
868 	const char *gitconfig_str = "[core]\n"
869 	    "\trepositoryformatversion = 0\n"
870 	    "\tfilemode = true\n"
871 	    "\tbare = true\n";
872 	char *path;
873 	int i;
874 
875 	if (!got_path_dir_is_empty(repo_path))
876 		return got_error(GOT_ERR_DIR_NOT_EMPTY);
877 
878 	for (i = 0; i < nitems(dirnames); i++) {
879 		if (asprintf(&path, "%s/%s", repo_path, dirnames[i]) == -1) {
880 			return got_error_from_errno("asprintf");
881 		}
882 		err = got_path_mkdir(path);
883 		free(path);
884 		if (err)
885 			return err;
886 	}
887 
888 	if (asprintf(&path, "%s/%s", repo_path, "description") == -1)
889 		return got_error_from_errno("asprintf");
890 	err = got_path_create_file(path, description_str);
891 	free(path);
892 	if (err)
893 		return err;
894 
895 	if (asprintf(&path, "%s/%s", repo_path, GOT_HEAD_FILE) == -1)
896 		return got_error_from_errno("asprintf");
897 	err = got_path_create_file(path, headref_str);
898 	free(path);
899 	if (err)
900 		return err;
901 
902 	if (asprintf(&path, "%s/%s", repo_path, "config") == -1)
903 		return got_error_from_errno("asprintf");
904 	err = got_path_create_file(path, gitconfig_str);
905 	free(path);
906 	if (err)
907 		return err;
908 
909 	return NULL;
910 }
911 
912 static const struct got_error *
913 match_packed_object(struct got_object_id **unique_id,
914     struct got_repository *repo, const char *id_str_prefix, int obj_type)
915 {
916 	const struct got_error *err = NULL;
917 	char *path_packdir;
918 	DIR *packdir;
919 	struct dirent *dent;
920 	char *path_packidx;
921 	struct got_object_id_queue matched_ids;
922 
923 	SIMPLEQ_INIT(&matched_ids);
924 
925 	path_packdir = got_repo_get_path_objects_pack(repo);
926 	if (path_packdir == NULL)
927 		return got_error_from_errno("got_repo_get_path_objects_pack");
928 
929 	packdir = opendir(path_packdir);
930 	if (packdir == NULL) {
931 		if (errno != ENOENT)
932 			err = got_error_from_errno2("opendir", path_packdir);
933 		goto done;
934 	}
935 
936 	while ((dent = readdir(packdir)) != NULL) {
937 		struct got_packidx *packidx;
938 		struct got_object_qid *qid;
939 
940 
941 		if (!is_packidx_filename(dent->d_name, dent->d_namlen))
942 			continue;
943 
944 		if (asprintf(&path_packidx, "%s/%s", path_packdir,
945 		    dent->d_name) == -1) {
946 			err = got_error_from_errno("asprintf");
947 			break;
948 		}
949 
950 		err = got_packidx_open(&packidx, path_packidx, 0);
951 		free(path_packidx);
952 		if (err)
953 			break;
954 
955 		err = got_packidx_match_id_str_prefix(&matched_ids,
956 		    packidx, id_str_prefix);
957 		if (err) {
958 			got_packidx_close(packidx);
959 			break;
960 		}
961 		err = got_packidx_close(packidx);
962 		if (err)
963 			break;
964 
965 		SIMPLEQ_FOREACH(qid, &matched_ids, entry) {
966 			if (obj_type != GOT_OBJ_TYPE_ANY) {
967 				int matched_type;
968 				err = got_object_get_type(&matched_type, repo,
969 				    qid->id);
970 				if (err)
971 					goto done;
972 				if (matched_type != obj_type)
973 					continue;
974 			}
975 			if (*unique_id == NULL) {
976 				*unique_id = got_object_id_dup(qid->id);
977 				if (*unique_id == NULL) {
978 					err = got_error_from_errno("malloc");
979 					goto done;
980 				}
981 			} else {
982 				err = got_error(GOT_ERR_AMBIGUOUS_ID);
983 				goto done;
984 			}
985 		}
986 	}
987 done:
988 	got_object_id_queue_free(&matched_ids);
989 	free(path_packdir);
990 	if (packdir && closedir(packdir) != 0 && err == NULL)
991 		err = got_error_from_errno("closedir");
992 	if (err) {
993 		free(*unique_id);
994 		*unique_id = NULL;
995 	}
996 	return err;
997 }
998 
999 static const struct got_error *
1000 match_loose_object(struct got_object_id **unique_id, const char *path_objects,
1001     const char *object_dir, const char *id_str_prefix, int obj_type,
1002     struct got_repository *repo)
1003 {
1004 	const struct got_error *err = NULL;
1005 	char *path;
1006 	DIR *dir = NULL;
1007 	struct dirent *dent;
1008 	struct got_object_id id;
1009 
1010 	if (asprintf(&path, "%s/%s", path_objects, object_dir) == -1) {
1011 		err = got_error_from_errno("asprintf");
1012 		goto done;
1013 	}
1014 
1015 	dir = opendir(path);
1016 	if (dir == NULL) {
1017 		if (errno == ENOENT) {
1018 			err = NULL;
1019 			goto done;
1020 		}
1021 		err = got_error_from_errno2("opendir", path);
1022 		goto done;
1023 	}
1024 	while ((dent = readdir(dir)) != NULL) {
1025 		char *id_str;
1026 		int cmp;
1027 
1028 		if (strcmp(dent->d_name, ".") == 0 ||
1029 		    strcmp(dent->d_name, "..") == 0)
1030 			continue;
1031 
1032 		if (asprintf(&id_str, "%s%s", object_dir, dent->d_name) == -1) {
1033 			err = got_error_from_errno("asprintf");
1034 			goto done;
1035 		}
1036 
1037 		if (!got_parse_sha1_digest(id.sha1, id_str))
1038 			continue;
1039 
1040 		/*
1041 		 * Directory entries do not necessarily appear in
1042 		 * sorted order, so we must iterate over all of them.
1043 		 */
1044 		cmp = strncmp(id_str, id_str_prefix, strlen(id_str_prefix));
1045 		if (cmp != 0) {
1046 			free(id_str);
1047 			continue;
1048 		}
1049 
1050 		if (*unique_id == NULL) {
1051 			if (obj_type != GOT_OBJ_TYPE_ANY) {
1052 				int matched_type;
1053 				err = got_object_get_type(&matched_type, repo,
1054 				    &id);
1055 				if (err)
1056 					goto done;
1057 				if (matched_type != obj_type)
1058 					continue;
1059 			}
1060 			*unique_id = got_object_id_dup(&id);
1061 			if (*unique_id == NULL) {
1062 				err = got_error_from_errno("got_object_id_dup");
1063 				free(id_str);
1064 				goto done;
1065 			}
1066 		} else {
1067 			err = got_error(GOT_ERR_AMBIGUOUS_ID);
1068 			free(id_str);
1069 			goto done;
1070 		}
1071 	}
1072 done:
1073 	if (dir && closedir(dir) != 0 && err == NULL)
1074 		err = got_error_from_errno("closedir");
1075 	if (err) {
1076 		free(*unique_id);
1077 		*unique_id = NULL;
1078 	}
1079 	free(path);
1080 	return err;
1081 }
1082 
1083 const struct got_error *
1084 got_repo_match_object_id_prefix(struct got_object_id **id,
1085     const char *id_str_prefix, int obj_type, struct got_repository *repo)
1086 {
1087 	const struct got_error *err = NULL;
1088 	char *path_objects = got_repo_get_path_objects(repo);
1089 	char *object_dir = NULL;
1090 	size_t len;
1091 	int i;
1092 
1093 	*id = NULL;
1094 
1095 	for (i = 0; i < strlen(id_str_prefix); i++) {
1096 		if (isxdigit((unsigned char)id_str_prefix[i]))
1097 			continue;
1098 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
1099 	}
1100 
1101 	len = strlen(id_str_prefix);
1102 	if (len >= 2) {
1103 		err = match_packed_object(id, repo, id_str_prefix, obj_type);
1104 		if (err)
1105 			goto done;
1106 		object_dir = strndup(id_str_prefix, 2);
1107 		if (object_dir == NULL) {
1108 			err = got_error_from_errno("strdup");
1109 			goto done;
1110 		}
1111 		err = match_loose_object(id, path_objects, object_dir,
1112 		    id_str_prefix, obj_type, repo);
1113 	} else if (len == 1) {
1114 		int i;
1115 		for (i = 0; i < 0xf; i++) {
1116 			if (asprintf(&object_dir, "%s%.1x", id_str_prefix, i)
1117 			    == -1) {
1118 				err = got_error_from_errno("asprintf");
1119 				goto done;
1120 			}
1121 			err = match_packed_object(id, repo, object_dir,
1122 			    obj_type);
1123 			if (err)
1124 				goto done;
1125 			err = match_loose_object(id, path_objects, object_dir,
1126 			    id_str_prefix, obj_type, repo);
1127 			if (err)
1128 				goto done;
1129 		}
1130 	} else {
1131 		err = got_error(GOT_ERR_BAD_OBJ_ID_STR);
1132 		goto done;
1133 	}
1134 done:
1135 	free(object_dir);
1136 	if (err) {
1137 		free(*id);
1138 		*id = NULL;
1139 	} else if (*id == NULL)
1140 		err = got_error(GOT_ERR_NO_OBJ);
1141 
1142 	return err;
1143 }
1144 
1145 const struct got_error *
1146 got_repo_object_match_tag(struct got_tag_object **tag, const char *name,
1147     int obj_type, struct got_repository *repo)
1148 {
1149 	const struct got_error *err;
1150 	struct got_reflist_head refs;
1151 	struct got_reflist_entry *re;
1152 	struct got_object_id *tag_id;
1153 
1154 	SIMPLEQ_INIT(&refs);
1155 	*tag = NULL;
1156 
1157 	err = got_ref_list(&refs, repo, "refs/tags", got_ref_cmp_by_name, NULL);
1158 	if (err)
1159 		return err;
1160 
1161 	SIMPLEQ_FOREACH(re, &refs, entry) {
1162 		const char *refname;
1163 		refname = got_ref_get_name(re->ref);
1164 		if (got_ref_is_symbolic(re->ref))
1165 			continue;
1166 		refname += strlen("refs/tags/");
1167 		if (strcmp(refname, name) != 0)
1168 			continue;
1169 		err = got_ref_resolve(&tag_id, repo, re->ref);
1170 		if (err)
1171 			break;
1172 		err = got_object_open_as_tag(tag, repo, tag_id);
1173 		free(tag_id);
1174 		if (err)
1175 			break;
1176 		if (obj_type == GOT_OBJ_TYPE_ANY ||
1177 		    got_object_tag_get_object_type(*tag) == obj_type)
1178 			break;
1179 		got_object_tag_close(*tag);
1180 		*tag = NULL;
1181 	}
1182 
1183 	got_ref_list_free(&refs);
1184 	if (err == NULL && *tag == NULL)
1185 		err = got_error(GOT_ERR_NO_OBJ);
1186 	return err;
1187 }
1188 
1189 static const struct got_error *
1190 alloc_added_blob_tree_entry(struct got_tree_entry **new_te,
1191     const char *name, mode_t mode, struct got_object_id *blob_id)
1192 {
1193 	const struct got_error *err = NULL;
1194 
1195 	 *new_te = NULL;
1196 
1197 	*new_te = calloc(1, sizeof(**new_te));
1198 	if (*new_te == NULL)
1199 		return got_error_from_errno("calloc");
1200 
1201 	(*new_te)->name = strdup(name);
1202 	if ((*new_te)->name == NULL) {
1203 		err = got_error_from_errno("strdup");
1204 		goto done;
1205 	}
1206 
1207 	(*new_te)->mode = S_IFREG | (mode & ((S_IRWXU | S_IRWXG | S_IRWXO)));
1208 	(*new_te)->id = blob_id;
1209 done:
1210 	if (err && *new_te) {
1211 		got_object_tree_entry_close(*new_te);
1212 		*new_te = NULL;
1213 	}
1214 	return err;
1215 }
1216 
1217 static const struct got_error *
1218 import_file(struct got_tree_entry **new_te, struct dirent *de,
1219     const char *path, struct got_repository *repo)
1220 {
1221 	const struct got_error *err;
1222 	struct got_object_id *blob_id = NULL;
1223 	char *filepath;
1224 	struct stat sb;
1225 
1226 	if (asprintf(&filepath, "%s%s%s", path,
1227 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1228 		return got_error_from_errno("asprintf");
1229 
1230 	if (lstat(filepath, &sb) != 0) {
1231 		err = got_error_from_errno2("lstat", path);
1232 		goto done;
1233 	}
1234 
1235 	err = got_object_blob_create(&blob_id, filepath, repo);
1236 	if (err)
1237 		goto done;
1238 
1239 	err = alloc_added_blob_tree_entry(new_te, de->d_name, sb.st_mode,
1240 	    blob_id);
1241 done:
1242 	free(filepath);
1243 	if (err)
1244 		free(blob_id);
1245 	return err;
1246 }
1247 
1248 static const struct got_error *
1249 insert_tree_entry(struct got_tree_entry *new_te,
1250     struct got_pathlist_head *paths)
1251 {
1252 	const struct got_error *err = NULL;
1253 	struct got_pathlist_entry *new_pe;
1254 
1255 	err = got_pathlist_insert(&new_pe, paths, new_te->name, new_te);
1256 	if (err)
1257 		return err;
1258 	if (new_pe == NULL)
1259 		return got_error(GOT_ERR_TREE_DUP_ENTRY);
1260 	return NULL;
1261 }
1262 
1263 static const struct got_error *write_tree(struct got_object_id **,
1264     const char *, struct got_pathlist_head *, struct got_repository *,
1265     got_repo_import_cb progress_cb, void *progress_arg);
1266 
1267 static const struct got_error *
1268 import_subdir(struct got_tree_entry **new_te, struct dirent *de,
1269     const char *path, struct got_pathlist_head *ignores,
1270     struct got_repository *repo,
1271     got_repo_import_cb progress_cb, void *progress_arg)
1272 {
1273 	const struct got_error *err;
1274 	char *subdirpath;
1275 
1276 	if (asprintf(&subdirpath, "%s%s%s", path,
1277 	    path[0] == '\0' ? "" : "/", de->d_name) == -1)
1278 		return got_error_from_errno("asprintf");
1279 
1280 	(*new_te) = calloc(1, sizeof(**new_te));
1281 	(*new_te)->mode = S_IFDIR;
1282 	(*new_te)->name = strdup(de->d_name);
1283 	if ((*new_te)->name == NULL) {
1284 		err = got_error_from_errno("strdup");
1285 		goto done;
1286 	}
1287 
1288 	err = write_tree(&(*new_te)->id, subdirpath, ignores,  repo,
1289 	    progress_cb, progress_arg);
1290 done:
1291 	free(subdirpath);
1292 	if (err) {
1293 		got_object_tree_entry_close(*new_te);
1294 		*new_te = NULL;
1295 	}
1296 	return err;
1297 }
1298 
1299 static const struct got_error *
1300 write_tree(struct got_object_id **new_tree_id, const char *path_dir,
1301     struct got_pathlist_head *ignores, struct got_repository *repo,
1302     got_repo_import_cb progress_cb, void *progress_arg)
1303 {
1304 	const struct got_error *err = NULL;
1305 	DIR *dir;
1306 	struct dirent *de;
1307 	struct got_tree_entries new_tree_entries;
1308 	struct got_tree_entry *new_te = NULL;
1309 	struct got_pathlist_head paths;
1310 	struct got_pathlist_entry *pe;
1311 
1312 	*new_tree_id = NULL;
1313 
1314 	TAILQ_INIT(&paths);
1315 	new_tree_entries.nentries = 0;
1316 	SIMPLEQ_INIT(&new_tree_entries.head);
1317 
1318 	dir = opendir(path_dir);
1319 	if (dir == NULL) {
1320 		err = got_error_from_errno2("opendir", path_dir);
1321 		goto done;
1322 	}
1323 
1324 	while ((de = readdir(dir)) != NULL) {
1325 		int ignore = 0;
1326 
1327 		if (strcmp(de->d_name, ".") == 0 ||
1328 		    strcmp(de->d_name, "..") == 0)
1329 			continue;
1330 
1331 		TAILQ_FOREACH(pe, ignores, entry) {
1332 			if (fnmatch(pe->path, de->d_name, 0) == 0) {
1333 				ignore = 1;
1334 				break;
1335 			}
1336 		}
1337 		if (ignore)
1338 			continue;
1339 		if (de->d_type == DT_DIR) {
1340 			err = import_subdir(&new_te, de, path_dir,
1341 			    ignores, repo, progress_cb, progress_arg);
1342 			if (err)
1343 				goto done;
1344 		} else if (de->d_type == DT_REG) {
1345 			err = import_file(&new_te, de, path_dir, repo);
1346 			if (err)
1347 				goto done;
1348 		} else
1349 			continue;
1350 
1351 		err = insert_tree_entry(new_te, &paths);
1352 		if (err)
1353 			goto done;
1354 	}
1355 
1356 	TAILQ_FOREACH(pe, &paths, entry) {
1357 		struct got_tree_entry *te = pe->data;
1358 		char *path;
1359 		new_tree_entries.nentries++;
1360 		SIMPLEQ_INSERT_TAIL(&new_tree_entries.head, te, entry);
1361 		if (!S_ISREG(te->mode))
1362 			continue;
1363 		if (asprintf(&path, "%s/%s", path_dir, pe->path) == -1) {
1364 			err = got_error_from_errno("asprintf");
1365 			goto done;
1366 		}
1367 		err = (*progress_cb)(progress_arg, path);
1368 		free(path);
1369 		if (err)
1370 			goto done;
1371 	}
1372 
1373 	err = got_object_tree_create(new_tree_id, &new_tree_entries, repo);
1374 done:
1375 	if (dir)
1376 		closedir(dir);
1377 	got_object_tree_entries_close(&new_tree_entries);
1378 	got_pathlist_free(&paths);
1379 	return err;
1380 }
1381 
1382 const struct got_error *
1383 got_repo_import(struct got_object_id **new_commit_id, const char *path_dir,
1384     const char *logmsg, const char *author, struct got_pathlist_head *ignores,
1385     struct got_repository *repo, got_repo_import_cb progress_cb,
1386     void *progress_arg)
1387 {
1388 	const struct got_error *err;
1389 	struct got_object_id *new_tree_id;
1390 
1391 	err = write_tree(&new_tree_id, path_dir, ignores, repo,
1392 	    progress_cb, progress_arg);
1393 	if (err)
1394 		return err;
1395 
1396 	err = got_object_commit_create(new_commit_id, new_tree_id, NULL, 0,
1397 	    author, time(NULL), author, time(NULL), logmsg, repo);
1398 	free(new_tree_id);
1399 	return err;
1400 }