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/stat.h>
19 #include <sys/queue.h>
20 #include <sys/tree.h>
21 #include <sys/uio.h>
22 #include <sys/mman.h>
23 
24 #include <errno.h>
25 #include <fcntl.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <sha1.h>
30 #include <sha2.h>
31 #include <unistd.h>
32 #include <zlib.h>
33 #include <libgen.h>
34 #include <limits.h>
35 #include <imsg.h>
36 
37 #include "got_error.h"
38 #include "got_object.h"
39 #include "got_repository.h"
40 #include "got_opentemp.h"
41 #include "got_path.h"
42 
43 #include "got_lib_hash.h"
44 #include "got_lib_delta.h"
45 #include "got_lib_inflate.h"
46 #include "got_lib_object.h"
47 #include "got_lib_object_idcache.h"
48 #include "got_lib_object_cache.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_repository.h"
52 
53 #ifndef MIN
54 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
55 #endif
56 
57 #ifndef nitems
58 #define nitems(_a)	(sizeof((_a)) / sizeof((_a)[0]))
59 #endif
60 
61 struct got_object_id *
62 got_object_get_id(struct got_object *obj)
63 {
64 	return &obj->id;
65 }
66 
67 const struct got_error *
68 got_object_get_id_str(char **outbuf, struct got_object *obj)
69 {
70 	return got_object_id_str(outbuf, &obj->id);
71 }
72 
73 const struct got_error *
74 got_object_get_type(int *type, struct got_repository *repo,
75     struct got_object_id *id)
76 {
77 	const struct got_error *err = NULL;
78 	struct got_object *obj;
79 
80 	err = got_object_open(&obj, repo, id);
81 	if (err)
82 		return err;
83 
84 	switch (obj->type) {
85 	case GOT_OBJ_TYPE_COMMIT:
86 	case GOT_OBJ_TYPE_TREE:
87 	case GOT_OBJ_TYPE_BLOB:
88 	case GOT_OBJ_TYPE_TAG:
89 		*type = obj->type;
90 		break;
91 	default:
92 		err = got_error(GOT_ERR_OBJ_TYPE);
93 		break;
94 	}
95 
96 	got_object_close(obj);
97 	return err;
98 }
99 
100 const struct got_error *
101 got_object_get_path(char **path, struct got_object_id *id,
102     struct got_repository *repo)
103 {
104 	const struct got_error *err = NULL;
105 	char *hex = NULL;
106 	char *path_objects;
107 
108 	*path = NULL;
109 
110 	path_objects = got_repo_get_path_objects(repo);
111 	if (path_objects == NULL)
112 		return got_error_from_errno("got_repo_get_path_objects");
113 
114 	err = got_object_id_str(&hex, id);
115 	if (err)
116 		goto done;
117 
118 	if (asprintf(path, "%s/%.2x/%s", path_objects,
119 	    id->sha1[0], hex + 2) == -1)
120 		err = got_error_from_errno("asprintf");
121 
122 done:
123 	free(hex);
124 	free(path_objects);
125 	return err;
126 }
127 
128 const struct got_error *
129 got_object_open_loose_fd(int *fd, struct got_object_id *id,
130     struct got_repository *repo)
131 {
132 	const struct got_error *err = NULL;
133 	char *path;
134 
135 	err = got_object_get_path(&path, id, repo);
136 	if (err)
137 		return err;
138 	*fd = open(path, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
139 	if (*fd == -1) {
140 		err = got_error_from_errno2("open", path);
141 		goto done;
142 	}
143 done:
144 	free(path);
145 	return err;
146 }
147 
148 const struct got_error *
149 got_object_open_by_id_str(struct got_object **obj, struct got_repository *repo,
150     const char *id_str)
151 {
152 	struct got_object_id id;
153 
154 	if (!got_parse_object_id(&id, id_str, GOT_HASH_SHA1))
155 		return got_error_path(id_str, GOT_ERR_BAD_OBJ_ID_STR);
156 
157 	return got_object_open(obj, repo, &id);
158 }
159 
160 const struct got_error *
161 got_object_resolve_id_str(struct got_object_id **id,
162     struct got_repository *repo, const char *id_str)
163 {
164 	const struct got_error *err = NULL;
165 	struct got_object *obj;
166 
167 	err = got_object_open_by_id_str(&obj, repo, id_str);
168 	if (err)
169 		return err;
170 
171 	*id = got_object_id_dup(got_object_get_id(obj));
172 	got_object_close(obj);
173 	if (*id == NULL)
174 		return got_error_from_errno("got_object_id_dup");
175 
176 	return NULL;
177 }
178 
179 int
180 got_object_tree_get_nentries(struct got_tree_object *tree)
181 {
182 	return tree->nentries;
183 }
184 
185 struct got_tree_entry *
186 got_object_tree_get_first_entry(struct got_tree_object *tree)
187 {
188 	return got_object_tree_get_entry(tree, 0);
189 }
190 
191 struct got_tree_entry *
192 got_object_tree_get_last_entry(struct got_tree_object *tree)
193 {
194 	return got_object_tree_get_entry(tree, tree->nentries - 1);
195 }
196 
197 struct got_tree_entry *
198 got_object_tree_get_entry(struct got_tree_object *tree, int i)
199 {
200 	if (i < 0 || i >= tree->nentries)
201 		return NULL;
202 	return &tree->entries[i];
203 }
204 
205 mode_t
206 got_tree_entry_get_mode(struct got_tree_entry *te)
207 {
208 	return te->mode;
209 }
210 
211 const char *
212 got_tree_entry_get_name(struct got_tree_entry *te)
213 {
214 	return &te->name[0];
215 }
216 
217 struct got_object_id *
218 got_tree_entry_get_id(struct got_tree_entry *te)
219 {
220 	return &te->id;
221 }
222 
223 const struct got_error *
224 got_object_blob_read_to_str(char **s, struct got_blob_object *blob)
225 {
226 	const struct got_error *err = NULL;
227 	size_t len, totlen, hdrlen, offset;
228 
229 	*s = NULL;
230 
231 	hdrlen = got_object_blob_get_hdrlen(blob);
232 	totlen = 0;
233 	offset = 0;
234 	do {
235 		char *p;
236 
237 		err = got_object_blob_read_block(&len, blob);
238 		if (err)
239 			return err;
240 
241 		if (len == 0)
242 			break;
243 
244 		totlen += len - hdrlen;
245 		p = realloc(*s, totlen + 1);
246 		if (p == NULL) {
247 			err = got_error_from_errno("realloc");
248 			free(*s);
249 			*s = NULL;
250 			return err;
251 		}
252 		*s = p;
253 		/* Skip blob object header first time around. */
254 		memcpy(*s + offset,
255 		    got_object_blob_get_read_buf(blob) + hdrlen, len - hdrlen);
256 		hdrlen = 0;
257 		offset = totlen;
258 	} while (len > 0);
259 
260 	(*s)[totlen] = '\0';
261 	return NULL;
262 }
263 
264 const struct got_error *
265 got_tree_entry_get_symlink_target(char **link_target, struct got_tree_entry *te,
266     struct got_repository *repo)
267 {
268 	const struct got_error *err = NULL;
269 	struct got_blob_object *blob = NULL;
270 	int fd = -1;
271 
272 	*link_target = NULL;
273 
274 	if (!got_object_tree_entry_is_symlink(te))
275 		return got_error(GOT_ERR_TREE_ENTRY_TYPE);
276 
277 	fd = got_opentempfd();
278 	if (fd == -1) {
279 		err = got_error_from_errno("got_opentempfd");
280 		goto done;
281 	}
282 
283 	err = got_object_open_as_blob(&blob, repo,
284 	    got_tree_entry_get_id(te), PATH_MAX, fd);
285 	if (err)
286 		goto done;
287 
288 	err = got_object_blob_read_to_str(link_target, blob);
289 done:
290 	if (fd != -1 && close(fd) == -1 && err == NULL)
291 		err = got_error_from_errno("close");
292 	if (blob)
293 		got_object_blob_close(blob);
294 	if (err) {
295 		free(*link_target);
296 		*link_target = NULL;
297 	}
298 	return err;
299 }
300 
301 int
302 got_tree_entry_get_index(struct got_tree_entry *te)
303 {
304 	return te->idx;
305 }
306 
307 struct got_tree_entry *
308 got_tree_entry_get_next(struct got_tree_object *tree,
309     struct got_tree_entry *te)
310 {
311 	return got_object_tree_get_entry(tree, te->idx + 1);
312 }
313 
314 struct got_tree_entry *
315 got_tree_entry_get_prev(struct got_tree_object *tree,
316     struct got_tree_entry *te)
317 {
318 	return got_object_tree_get_entry(tree, te->idx - 1);
319 }
320 
321 const struct got_error *
322 got_object_blob_close(struct got_blob_object *blob)
323 {
324 	const struct got_error *err = NULL;
325 	free(blob->read_buf);
326 	if (blob->f && fclose(blob->f) == EOF)
327 		err = got_error_from_errno("fclose");
328 	free(blob->data);
329 	free(blob);
330 	return err;
331 }
332 
333 void
334 got_object_blob_rewind(struct got_blob_object *blob)
335 {
336 	if (blob->f)
337 		rewind(blob->f);
338 }
339 
340 char *
341 got_object_blob_id_str(struct got_blob_object *blob, char *buf, size_t size)
342 {
343 	return got_object_id_hex(&blob->id, buf, size);
344 }
345 
346 size_t
347 got_object_blob_get_hdrlen(struct got_blob_object *blob)
348 {
349 	return blob->hdrlen;
350 }
351 
352 const uint8_t *
353 got_object_blob_get_read_buf(struct got_blob_object *blob)
354 {
355 	return blob->read_buf;
356 }
357 
358 const struct got_error *
359 got_object_blob_read_block(size_t *outlenp, struct got_blob_object *blob)
360 {
361 	size_t n;
362 
363 	n = fread(blob->read_buf, 1, blob->blocksize, blob->f);
364 	if (n == 0 && ferror(blob->f))
365 		return got_ferror(blob->f, GOT_ERR_IO);
366 	*outlenp = n;
367 	return NULL;
368 }
369 
370 const struct got_error *
371 got_object_blob_is_binary(int *binary, struct got_blob_object *blob)
372 {
373 	const struct got_error *err;
374 	size_t hdrlen, len;
375 
376 	*binary = 0;
377 	hdrlen = got_object_blob_get_hdrlen(blob);
378 
379 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
380 		return got_error_from_errno("fseeko");
381 
382 	err = got_object_blob_read_block(&len, blob);
383 	if (err)
384 		return err;
385 
386 	*binary = memchr(blob->read_buf, '\0', len) != NULL;
387 
388 	if (fseeko(blob->f, hdrlen, SEEK_SET) == -1)
389 		return got_error_from_errno("fseeko");
390 	return NULL;
391 }
392 
393 const struct got_error *
394 got_object_blob_getline(char **line, ssize_t *linelen, size_t *linesize,
395     struct got_blob_object *blob)
396 {
397 	*linelen = getline(line, linesize, blob->f);
398 	if (*linelen == -1 && !feof(blob->f))
399 		return got_error_from_errno("getline");
400 	return NULL;
401 }
402 
403 const struct got_error *
404 got_object_blob_dump_to_file(off_t *filesize, int *nlines,
405     off_t **line_offsets, FILE *outfile, struct got_blob_object *blob)
406 {
407 	const struct got_error *err = NULL;
408 	size_t n, len, hdrlen;
409 	const uint8_t *buf;
410 	int i;
411 	const int alloc_chunksz = 512;
412 	size_t nalloc = 0;
413 	off_t off = 0, total_len = 0;
414 
415 	if (line_offsets)
416 		*line_offsets = NULL;
417 	if (filesize)
418 		*filesize = 0;
419 	if (nlines)
420 		*nlines = 0;
421 
422 	hdrlen = got_object_blob_get_hdrlen(blob);
423 	do {
424 		err = got_object_blob_read_block(&len, blob);
425 		if (err)
426 			return err;
427 		if (len == 0)
428 			break;
429 		buf = got_object_blob_get_read_buf(blob);
430 		i = hdrlen;
431 		if (nlines) {
432 			if (line_offsets && *line_offsets == NULL) {
433 				/* Have some data but perhaps no '\n'. */
434 				*nlines = 1;
435 				nalloc = alloc_chunksz;
436 				*line_offsets = calloc(nalloc,
437 				    sizeof(**line_offsets));
438 				if (*line_offsets == NULL)
439 					return got_error_from_errno("calloc");
440 
441 				/* Skip forward over end of first line. */
442 				while (i < len) {
443 					if (buf[i] == '\n')
444 						break;
445 					i++;
446 				}
447 			}
448 			/* Scan '\n' offsets in remaining chunk of data. */
449 			while (i < len) {
450 				if (buf[i] != '\n') {
451 					i++;
452 					continue;
453 				}
454 				(*nlines)++;
455 				if (line_offsets && nalloc < *nlines) {
456 					size_t n = *nlines + alloc_chunksz;
457 					off_t *o = recallocarray(*line_offsets,
458 					    nalloc, n, sizeof(**line_offsets));
459 					if (o == NULL) {
460 						free(*line_offsets);
461 						*line_offsets = NULL;
462 						return got_error_from_errno(
463 						    "recallocarray");
464 					}
465 					*line_offsets = o;
466 					nalloc = n;
467 				}
468 				if (line_offsets) {
469 					off = total_len + i - hdrlen + 1;
470 					(*line_offsets)[*nlines - 1] = off;
471 				}
472 				i++;
473 			}
474 		}
475 		/* Skip blob object header first time around. */
476 		n = fwrite(buf + hdrlen, 1, len - hdrlen, outfile);
477 		if (n != len - hdrlen)
478 			return got_ferror(outfile, GOT_ERR_IO);
479 		total_len += len - hdrlen;
480 		hdrlen = 0;
481 	} while (len != 0);
482 
483 	if (fflush(outfile) != 0)
484 		return got_error_from_errno("fflush");
485 	rewind(outfile);
486 
487 	if (filesize)
488 		*filesize = total_len;
489 
490 	return NULL;
491 }
492 
493 const char *
494 got_object_tag_get_name(struct got_tag_object *tag)
495 {
496 	return tag->tag;
497 }
498 
499 int
500 got_object_tag_get_object_type(struct got_tag_object *tag)
501 {
502 	return tag->obj_type;
503 }
504 
505 struct got_object_id *
506 got_object_tag_get_object_id(struct got_tag_object *tag)
507 {
508 	return &tag->id;
509 }
510 
511 time_t
512 got_object_tag_get_tagger_time(struct got_tag_object *tag)
513 {
514 	return tag->tagger_time;
515 }
516 
517 time_t
518 got_object_tag_get_tagger_gmtoff(struct got_tag_object *tag)
519 {
520 	return tag->tagger_gmtoff;
521 }
522 
523 const char *
524 got_object_tag_get_tagger(struct got_tag_object *tag)
525 {
526 	return tag->tagger;
527 }
528 
529 const char *
530 got_object_tag_get_message(struct got_tag_object *tag)
531 {
532 	return tag->tagmsg;
533 }
534 
535 static struct got_tree_entry *
536 find_entry_by_name(struct got_tree_object *tree, const char *name, size_t len)
537 {
538 	int i;
539 
540 	/* Note that tree entries are sorted in strncmp() order. */
541 	for (i = 0; i < tree->nentries; i++) {
542 		struct got_tree_entry *te = &tree->entries[i];
543 		int cmp = strncmp(te->name, name, len);
544 		if (cmp < 0)
545 			continue;
546 		if (cmp > 0)
547 			break;
548 		if (te->name[len] == '\0')
549 			return te;
550 	}
551 	return NULL;
552 }
553 
554 struct got_tree_entry *
555 got_object_tree_find_entry(struct got_tree_object *tree, const char *name)
556 {
557 	return find_entry_by_name(tree, name, strlen(name));
558 }
559 
560 const struct got_error *
561 got_object_tree_find_path(struct got_object_id **id, mode_t *mode,
562     struct got_repository *repo, struct got_tree_object *tree,
563     const char *path)
564 {
565 	const struct got_error *err = NULL;
566 	struct got_tree_object *subtree = NULL;
567 	struct got_tree_entry *te = NULL;
568 	const char *seg, *s;
569 	size_t seglen;
570 
571 	*id = NULL;
572 
573 	s = path;
574 	while (s[0] == '/')
575 		s++;
576 	seg = s;
577 	seglen = 0;
578 	subtree = tree;
579 	while (*s) {
580 		struct got_tree_object *next_tree;
581 
582 		if (*s != '/') {
583 			s++;
584 			seglen++;
585 			if (*s)
586 				continue;
587 		}
588 
589 		te = find_entry_by_name(subtree, seg, seglen);
590 		if (te == NULL) {
591 			err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
592 			goto done;
593 		}
594 
595 		if (*s == '\0')
596 			break;
597 
598 		seg = s + 1;
599 		seglen = 0;
600 		s++;
601 		if (*s) {
602 			err = got_object_open_as_tree(&next_tree, repo,
603 			    &te->id);
604 			te = NULL;
605 			if (err)
606 				goto done;
607 			if (subtree != tree)
608 				got_object_tree_close(subtree);
609 			subtree = next_tree;
610 		}
611 	}
612 
613 	if (te) {
614 		*id = got_object_id_dup(&te->id);
615 		if (*id == NULL)
616 			return got_error_from_errno("got_object_id_dup");
617 		if (mode)
618 			*mode = te->mode;
619 	} else
620 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
621 done:
622 	if (subtree && subtree != tree)
623 		got_object_tree_close(subtree);
624 	return err;
625 }
626 
627 const struct got_error *
628 got_object_id_by_path(struct got_object_id **id, struct got_repository *repo,
629     struct got_commit_object *commit, const char *path)
630 {
631 	const struct got_error *err = NULL;
632 	struct got_tree_object *tree = NULL;
633 
634 	*id = NULL;
635 
636 	/* Handle opening of root of commit's tree. */
637 	if (got_path_is_root_dir(path)) {
638 		*id = got_object_id_dup(commit->tree_id);
639 		if (*id == NULL)
640 			err = got_error_from_errno("got_object_id_dup");
641 	} else {
642 		err = got_object_open_as_tree(&tree, repo, commit->tree_id);
643 		if (err)
644 			goto done;
645 		err = got_object_tree_find_path(id, NULL, repo, tree, path);
646 	}
647 done:
648 	if (tree)
649 		got_object_tree_close(tree);
650 	return err;
651 }
652 
653 /*
654  * Normalize file mode bits to avoid false positive tree entry differences
655  * in case tree entries have unexpected mode bits set.
656  */
657 static mode_t
658 normalize_mode_for_comparison(mode_t mode)
659 {
660 	/*
661 	 * For directories, the only relevant bit is the IFDIR bit.
662 	 * This allows us to detect paths changing from a directory
663 	 * to a file and vice versa.
664 	 */
665 	if (S_ISDIR(mode))
666 		return mode & S_IFDIR;
667 
668 	/*
669 	 * For symlinks, the only relevant bit is the IFLNK bit.
670 	 * This allows us to detect paths changing from a symlinks
671 	 * to a file or directory and vice versa.
672 	 */
673 	if (S_ISLNK(mode))
674 		return mode & S_IFLNK;
675 
676 	/* For files, the only change we care about is the executable bit. */
677 	return mode & S_IXUSR;
678 }
679 
680 const struct got_error *
681 got_object_tree_path_changed(int *changed,
682     struct got_tree_object *tree01, struct got_tree_object *tree02,
683     const char *path, struct got_repository *repo)
684 {
685 	const struct got_error *err = NULL;
686 	struct got_tree_object *tree1 = NULL, *tree2 = NULL;
687 	struct got_tree_entry *te1 = NULL, *te2 = NULL;
688 	const char *seg, *s;
689 	size_t seglen;
690 
691 	*changed = 0;
692 
693 	/* We not do support comparing the root path. */
694 	if (got_path_is_root_dir(path))
695 		return got_error_path(path, GOT_ERR_BAD_PATH);
696 
697 	tree1 = tree01;
698 	tree2 = tree02;
699 	s = path;
700 	while (*s == '/')
701 		s++;
702 	seg = s;
703 	seglen = 0;
704 	while (*s) {
705 		struct got_tree_object *next_tree1, *next_tree2;
706 		mode_t mode1, mode2;
707 
708 		if (*s != '/') {
709 			s++;
710 			seglen++;
711 			if (*s)
712 				continue;
713 		}
714 
715 		te1 = find_entry_by_name(tree1, seg, seglen);
716 		if (te1 == NULL) {
717 			err = got_error(GOT_ERR_NO_OBJ);
718 			goto done;
719 		}
720 
721 		if (tree2)
722 			te2 = find_entry_by_name(tree2, seg, seglen);
723 
724 		if (te2) {
725 			mode1 = normalize_mode_for_comparison(te1->mode);
726 			mode2 = normalize_mode_for_comparison(te2->mode);
727 			if (mode1 != mode2) {
728 				*changed = 1;
729 				goto done;
730 			}
731 
732 			if (got_object_id_cmp(&te1->id, &te2->id) == 0) {
733 				*changed = 0;
734 				goto done;
735 			}
736 		}
737 
738 		if (*s == '\0') { /* final path element */
739 			*changed = 1;
740 			goto done;
741 		}
742 
743 		seg = s + 1;
744 		s++;
745 		seglen = 0;
746 		if (*s) {
747 			err = got_object_open_as_tree(&next_tree1, repo,
748 			    &te1->id);
749 			te1 = NULL;
750 			if (err)
751 				goto done;
752 			if (tree1 != tree01)
753 				got_object_tree_close(tree1);
754 			tree1 = next_tree1;
755 
756 			if (te2) {
757 				err = got_object_open_as_tree(&next_tree2, repo,
758 				    &te2->id);
759 				te2 = NULL;
760 				if (err)
761 					goto done;
762 				if (tree2 != tree02)
763 					got_object_tree_close(tree2);
764 				tree2 = next_tree2;
765 			} else if (tree2) {
766 				if (tree2 != tree02)
767 					got_object_tree_close(tree2);
768 				tree2 = NULL;
769 			}
770 		}
771 	}
772 done:
773 	if (tree1 && tree1 != tree01)
774 		got_object_tree_close(tree1);
775 	if (tree2 && tree2 != tree02)
776 		got_object_tree_close(tree2);
777 	return err;
778 }
779 
780 const struct got_error *
781 got_object_tree_entry_dup(struct got_tree_entry **new_te,
782     struct got_tree_entry *te)
783 {
784 	const struct got_error *err = NULL;
785 
786 	*new_te = calloc(1, sizeof(**new_te));
787 	if (*new_te == NULL)
788 		return got_error_from_errno("calloc");
789 
790 	(*new_te)->mode = te->mode;
791 	memcpy((*new_te)->name, te->name, sizeof((*new_te)->name));
792 	memcpy(&(*new_te)->id, &te->id, sizeof((*new_te)->id));
793 	return err;
794 }
795 
796 int
797 got_object_tree_entry_is_submodule(struct got_tree_entry *te)
798 {
799 	return (te->mode & S_IFMT) == (S_IFDIR | S_IFLNK);
800 }
801 
802 int
803 got_object_tree_entry_is_symlink(struct got_tree_entry *te)
804 {
805 	/* S_IFDIR check avoids confusing symlinks with submodules. */
806 	return ((te->mode & (S_IFDIR | S_IFLNK)) == S_IFLNK);
807 }
808 
809 static const struct got_error *
810 resolve_symlink(char **link_target, const char *path,
811     struct got_commit_object *commit, struct got_repository *repo)
812 {
813 	const struct got_error *err = NULL;
814 	char buf[PATH_MAX];
815 	char *name, *parent_path = NULL;
816 	struct got_object_id *tree_obj_id = NULL;
817 	struct got_tree_object *tree = NULL;
818 	struct got_tree_entry *te = NULL;
819 
820 	*link_target = NULL;
821 
822 	if (strlcpy(buf, path, sizeof(buf)) >= sizeof(buf))
823 		return got_error(GOT_ERR_NO_SPACE);
824 
825 	name = basename(buf);
826 	if (name == NULL)
827 		return got_error_from_errno2("basename", path);
828 
829 	err = got_path_dirname(&parent_path, path);
830 	if (err)
831 		return err;
832 
833 	err = got_object_id_by_path(&tree_obj_id, repo, commit,
834 	    parent_path);
835 	if (err) {
836 		if (err->code == GOT_ERR_NO_TREE_ENTRY) {
837 			/* Display the complete path in error message. */
838 			err = got_error_path(path, err->code);
839 		}
840 		goto done;
841 	}
842 
843 	err = got_object_open_as_tree(&tree, repo, tree_obj_id);
844 	if (err)
845 		goto done;
846 
847 	te = got_object_tree_find_entry(tree, name);
848 	if (te == NULL) {
849 		err = got_error_path(path, GOT_ERR_NO_TREE_ENTRY);
850 		goto done;
851 	}
852 
853 	if (got_object_tree_entry_is_symlink(te)) {
854 		err = got_tree_entry_get_symlink_target(link_target, te, repo);
855 		if (err)
856 			goto done;
857 		if (!got_path_is_absolute(*link_target)) {
858 			char *abspath;
859 			if (asprintf(&abspath, "%s/%s", parent_path,
860 			    *link_target) == -1) {
861 				err = got_error_from_errno("asprintf");
862 				goto done;
863 			}
864 			free(*link_target);
865 			*link_target = malloc(PATH_MAX);
866 			if (*link_target == NULL) {
867 				err = got_error_from_errno("malloc");
868 				goto done;
869 			}
870 			err = got_canonpath(abspath, *link_target, PATH_MAX);
871 			free(abspath);
872 			if (err)
873 				goto done;
874 		}
875 	}
876 done:
877 	free(parent_path);
878 	free(tree_obj_id);
879 	if (tree)
880 		got_object_tree_close(tree);
881 	if (err) {
882 		free(*link_target);
883 		*link_target = NULL;
884 	}
885 	return err;
886 }
887 
888 const struct got_error *
889 got_object_resolve_symlinks(char **link_target, const char *path,
890     struct got_commit_object *commit, struct got_repository *repo)
891 {
892 	const struct got_error *err = NULL;
893 	char *next_target = NULL;
894 	int max_recursion = 40; /* matches Git */
895 
896 	*link_target = NULL;
897 
898 	do {
899 		err = resolve_symlink(&next_target,
900 		    *link_target ? *link_target : path, commit, repo);
901 		if (err)
902 			break;
903 		if (next_target) {
904 			free(*link_target);
905 			if (--max_recursion == 0) {
906 				err = got_error_path(path, GOT_ERR_RECURSION);
907 				*link_target = NULL;
908 				break;
909 			}
910 			*link_target = next_target;
911 		}
912 	} while (next_target);
913 
914 	return err;
915 }
916 
917 void
918 got_object_commit_retain(struct got_commit_object *commit)
919 {
920 	commit->refcnt++;
921 }
922 
923 const struct got_error *
924 got_object_raw_alloc(struct got_raw_object **obj, uint8_t *outbuf, int *outfd,
925     size_t max_in_mem_size, size_t hdrlen, off_t size)
926 {
927 	const struct got_error *err = NULL;
928 	off_t tot;
929 
930 	tot = hdrlen + size;
931 
932 	*obj = calloc(1, sizeof(**obj));
933 	if (*obj == NULL) {
934 		err = got_error_from_errno("calloc");
935 		goto done;
936 	}
937 	(*obj)->fd = -1;
938 	(*obj)->tempfile_idx = -1;
939 
940 	if (outbuf) {
941 		(*obj)->data = outbuf;
942 	} else {
943 		struct stat sb;
944 		if (fstat(*outfd, &sb) == -1) {
945 			err = got_error_from_errno("fstat");
946 			goto done;
947 		}
948 
949 		if (sb.st_size != tot) {
950 			err = got_error_msg(GOT_ERR_BAD_OBJ_HDR,
951 			    "raw object has unexpected size");
952 			goto done;
953 		}
954 #ifndef GOT_PACK_NO_MMAP
955 		if (tot > 0 && tot <= max_in_mem_size) {
956 			(*obj)->data = mmap(NULL, tot, PROT_READ,
957 			    MAP_PRIVATE, *outfd, 0);
958 			if ((*obj)->data == MAP_FAILED) {
959 				if (errno != ENOMEM) {
960 					err = got_error_from_errno("mmap");
961 					goto done;
962 				}
963 				(*obj)->data = NULL;
964 			} else {
965 				(*obj)->fd = *outfd;
966 				*outfd = -1;
967 			}
968 		}
969 #endif
970 		if (*outfd != -1) {
971 			(*obj)->f = fdopen(*outfd, "r");
972 			if ((*obj)->f == NULL) {
973 				err = got_error_from_errno("fdopen");
974 				goto done;
975 			}
976 			*outfd = -1;
977 		}
978 	}
979 	(*obj)->hdrlen = hdrlen;
980 	(*obj)->size = size;
981 done:
982 	if (err) {
983 		if (*obj) {
984 			got_object_raw_close(*obj);
985 			*obj = NULL;
986 		}
987 	} else
988 		(*obj)->refcnt++;
989 	return err;
990 }