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