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