1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include "got_compat.h"
18 
19 #include <sys/types.h>
20 #include <sys/stat.h>
21 #include <sys/queue.h>
22 #include <sys/uio.h>
23 #include <sys/socket.h>
24 #include <sys/wait.h>
25 #include <sys/mman.h>
26 
27 #include <errno.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <stdint.h>
32 #include <zlib.h>
33 #include <ctype.h>
34 #include <limits.h>
35 #include <time.h>
36 #include <unistd.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_hash.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_repository.h"
52 
53 #ifndef nitems
54 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
55 #endif
56 
57 struct got_object_id *
58 got_object_id_dup(struct got_object_id *id1)
59 {
60 	struct got_object_id *id2;
61 
62 	id2 = malloc(sizeof(*id2));
63 	if (id2 == NULL)
64 		return NULL;
65 	memcpy(id2, id1, sizeof(*id2));
66 	return id2;
67 }
68 
69 int
70 got_object_id_cmp(const struct got_object_id *id1,
71     const struct got_object_id *id2)
72 {
73 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
74 }
75 
76 const struct got_error *
77 got_object_qid_alloc_partial(struct got_object_qid **qid)
78 {
79 	*qid = malloc(sizeof(**qid));
80 	if (*qid == NULL)
81 		return got_error_from_errno("malloc");
82 
83 	(*qid)->data = NULL;
84 	return NULL;
85 }
86 
87 const struct got_error *
88 got_object_id_str(char **outbuf, struct got_object_id *id)
89 {
90 	static const size_t len = GOT_OBJECT_ID_HEX_MAXLEN;
91 
92 	*outbuf = malloc(len);
93 	if (*outbuf == NULL)
94 		return got_error_from_errno("malloc");
95 
96 	if (got_object_id_hex(id, *outbuf, len) == NULL) {
97 		free(*outbuf);
98 		*outbuf = NULL;
99 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
100 	}
101 
102 	return NULL;
103 }
104 
105 char *
106 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
107 {
108 	return got_sha1_digest_to_str(id->sha1, buf, len);
109 }
110 
111 void
112 got_object_close(struct got_object *obj)
113 {
114 	if (obj->refcnt > 0) {
115 		obj->refcnt--;
116 		if (obj->refcnt > 0)
117 			return;
118 	}
119 
120 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
121 		struct got_delta *delta;
122 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
123 			delta = STAILQ_FIRST(&obj->deltas.entries);
124 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
125 			free(delta);
126 		}
127 	}
128 	free(obj);
129 }
130 
131 const struct got_error *
132 got_object_raw_close(struct got_raw_object *obj)
133 {
134 	const struct got_error *err = NULL;
135 
136 	if (obj->refcnt > 0) {
137 		obj->refcnt--;
138 		if (obj->refcnt > 0)
139 			return NULL;
140 	}
141 
142 	if (obj->close_cb)
143 		obj->close_cb(obj);
144 
145 	if (obj->f == NULL) {
146 		if (obj->fd != -1) {
147 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
148 				err = got_error_from_errno("munmap");
149 			if (close(obj->fd) == -1 && err == NULL)
150 				err = got_error_from_errno("close");
151 		} else
152 			free(obj->data);
153 	} else {
154 		if (fclose(obj->f) == EOF && err == NULL)
155 			err = got_error_from_errno("fclose");
156 	}
157 	free(obj);
158 	return err;
159 }
160 
161 void
162 got_object_qid_free(struct got_object_qid *qid)
163 {
164 	free(qid);
165 }
166 
167 void
168 got_object_id_queue_free(struct got_object_id_queue *ids)
169 {
170 	struct got_object_qid *qid;
171 
172 	while (!STAILQ_EMPTY(ids)) {
173 		qid = STAILQ_FIRST(ids);
174 		STAILQ_REMOVE_HEAD(ids, entry);
175 		got_object_qid_free(qid);
176 	}
177 }
178 
179 const struct got_error *
180 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
181 {
182 	const char *obj_labels[] = {
183 		GOT_OBJ_LABEL_COMMIT,
184 		GOT_OBJ_LABEL_TREE,
185 		GOT_OBJ_LABEL_BLOB,
186 		GOT_OBJ_LABEL_TAG,
187 	};
188 	const int obj_types[] = {
189 		GOT_OBJ_TYPE_COMMIT,
190 		GOT_OBJ_TYPE_TREE,
191 		GOT_OBJ_TYPE_BLOB,
192 		GOT_OBJ_TYPE_TAG,
193 	};
194 	int type = 0;
195 	size_t size = 0;
196 	size_t i;
197 	char *end;
198 
199 	*obj = NULL;
200 
201 	end = memchr(buf, '\0', len);
202 	if (end == NULL)
203 		return got_error(GOT_ERR_BAD_OBJ_HDR);
204 
205 	for (i = 0; i < nitems(obj_labels); i++) {
206 		const char *label = obj_labels[i];
207 		size_t label_len = strlen(label);
208 		const char *errstr;
209 
210 		if (len <= label_len || buf + label_len >= end ||
211 		    strncmp(buf, label, label_len) != 0)
212 			continue;
213 
214 		type = obj_types[i];
215 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
216 		if (errstr != NULL)
217 			return got_error(GOT_ERR_BAD_OBJ_HDR);
218 		break;
219 	}
220 
221 	if (type == 0)
222 		return got_error(GOT_ERR_BAD_OBJ_HDR);
223 
224 	*obj = calloc(1, sizeof(**obj));
225 	if (*obj == NULL)
226 		return got_error_from_errno("calloc");
227 	(*obj)->type = type;
228 	(*obj)->hdrlen = end - buf + 1;
229 	(*obj)->size = size;
230 	return NULL;
231 }
232 
233 const struct got_error *
234 got_object_read_header(struct got_object **obj, int fd)
235 {
236 	const struct got_error *err;
237 	struct got_inflate_buf zb;
238 	uint8_t *buf;
239 	const size_t zbsize = 64;
240 	size_t outlen, totlen;
241 	int nbuf = 1;
242 
243 	*obj = NULL;
244 
245 	buf = malloc(zbsize);
246 	if (buf == NULL)
247 		return got_error_from_errno("malloc");
248 	buf[0] = '\0';
249 
250 	err = got_inflate_init(&zb, buf, zbsize, NULL);
251 	if (err)
252 		return err;
253 
254 	totlen = 0;
255 	do {
256 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
257 		if (err)
258 			goto done;
259 		if (outlen == 0)
260 			break;
261 		totlen += outlen;
262 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
263 			uint8_t *newbuf;
264 			nbuf++;
265 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
266 			if (newbuf == NULL) {
267 				err = got_error_from_errno("recallocarray");
268 				goto done;
269 			}
270 			buf = newbuf;
271 			zb.outbuf = newbuf + totlen;
272 			zb.outlen = (nbuf * zbsize) - totlen;
273 		}
274 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
275 
276 	err = got_object_parse_header(obj, buf, totlen);
277 done:
278 	free(buf);
279 	got_inflate_end(&zb);
280 	return err;
281 }
282 
283 const struct got_error *
284 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
285     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
286     int infd)
287 {
288 	const struct got_error *err = NULL;
289 	struct got_object *obj;
290 	struct got_inflate_checksum csum;
291 	struct got_object_id id;
292 	struct got_hash ctx;
293 	size_t len, consumed;
294 	FILE *f = NULL;
295 
296 	*outbuf = NULL;
297 	*size = 0;
298 	*hdrlen = 0;
299 
300 	got_hash_init(&ctx, GOT_HASH_SHA1);
301 	memset(&csum, 0, sizeof(csum));
302 	csum.output_ctx = &ctx;
303 
304 	if (lseek(infd, SEEK_SET, 0) == -1)
305 		return got_error_from_errno("lseek");
306 
307 	err = got_object_read_header(&obj, infd);
308 	if (err)
309 		return err;
310 
311 	if (lseek(infd, SEEK_SET, 0) == -1)
312 		return got_error_from_errno("lseek");
313 
314 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
315 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
316 		    obj->size + obj->hdrlen, infd);
317 	} else {
318 		int fd;
319 		/*
320 		 * XXX This uses an extra file descriptor for no good reason.
321 		 * We should have got_inflate_fd_to_fd().
322 		 */
323 		fd = dup(infd);
324 		if (fd == -1)
325 			return got_error_from_errno("dup");
326 		f = fdopen(fd, "r");
327 		if (f == NULL) {
328 			err = got_error_from_errno("fdopen");
329 			abort();
330 			close(fd);
331 			goto done;
332 		}
333 		err = got_inflate_to_fd(&len, f, &csum, outfd);
334 	}
335 	if (err)
336 		goto done;
337 
338 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
339 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
340 		goto done;
341 	}
342 
343 	got_hash_final_object_id(&ctx, &id);
344 	if (got_object_id_cmp(expected_id, &id) != 0) {
345 		err = got_error_checksum(expected_id);
346 		goto done;
347 	}
348 
349 	*size = obj->size;
350 	*hdrlen = obj->hdrlen;
351 done:
352 	got_object_close(obj);
353 	if (f && fclose(f) == EOF && err == NULL)
354 		err = got_error_from_errno("fclose");
355 	return err;
356 }
357 
358 struct got_commit_object *
359 got_object_commit_alloc_partial(void)
360 {
361 	struct got_commit_object *commit;
362 
363 	commit = calloc(1, sizeof(*commit));
364 	if (commit == NULL)
365 		return NULL;
366 	commit->tree_id = malloc(sizeof(*commit->tree_id));
367 	if (commit->tree_id == NULL) {
368 		free(commit);
369 		return NULL;
370 	}
371 
372 	STAILQ_INIT(&commit->parent_ids);
373 
374 	return commit;
375 }
376 
377 const struct got_error *
378 got_object_commit_add_parent(struct got_commit_object *commit,
379     const char *id_str)
380 {
381 	const struct got_error *err = NULL;
382 	struct got_object_qid *qid;
383 
384 	err = got_object_qid_alloc_partial(&qid);
385 	if (err)
386 		return err;
387 
388 	if (!got_parse_object_id(&qid->id, id_str, GOT_HASH_SHA1)) {
389 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
390 		got_object_qid_free(qid);
391 		return err;
392 	}
393 
394 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
395 	commit->nparents++;
396 
397 	return NULL;
398 }
399 
400 static const struct got_error *
401 parse_gmtoff(time_t *gmtoff, const char *tzstr)
402 {
403 	int sign = 1;
404 	const char *p = tzstr;
405 	time_t h, m;
406 
407 	*gmtoff = 0;
408 
409 	if (*p == '-')
410 		sign = -1;
411 	else if (*p != '+')
412 		return got_error(GOT_ERR_BAD_OBJ_DATA);
413 	p++;
414 	if (!isdigit((unsigned char)*p) &&
415 	    !isdigit((unsigned char)*(p + 1)))
416 		return got_error(GOT_ERR_BAD_OBJ_DATA);
417 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
418 
419 	p += 2;
420 	if (!isdigit((unsigned char)*p) &&
421 	    !isdigit((unsigned char)*(p + 1)))
422 		return got_error(GOT_ERR_BAD_OBJ_DATA);
423 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
424 
425 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
426 	return NULL;
427 }
428 
429 static const struct got_error *
430 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
431 {
432 	const struct got_error *err = NULL;
433 	const char *errstr;
434 	char *space, *tzstr;
435 
436 	/* Parse and strip off trailing timezone indicator string. */
437 	space = strrchr(committer, ' ');
438 	if (space == NULL)
439 		return got_error(GOT_ERR_BAD_OBJ_DATA);
440 	tzstr = strdup(space + 1);
441 	if (tzstr == NULL)
442 		return got_error_from_errno("strdup");
443 	err = parse_gmtoff(gmtoff, tzstr);
444 	free(tzstr);
445 	if (err) {
446 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
447 			return err;
448 		/* Old versions of Git omitted the timestamp. */
449 		*time = 0;
450 		*gmtoff = 0;
451 		return NULL;
452 	}
453 	*space = '\0';
454 
455 	/* Timestamp is separated from committer name + email by space. */
456 	space = strrchr(committer, ' ');
457 	if (space == NULL)
458 		return got_error(GOT_ERR_BAD_OBJ_DATA);
459 
460 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
461 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
462 	if (errstr)
463 		return got_error(GOT_ERR_BAD_OBJ_DATA);
464 
465 	/* Strip off parsed time information, leaving just author and email. */
466 	*space = '\0';
467 
468 	return NULL;
469 }
470 
471 void
472 got_object_commit_close(struct got_commit_object *commit)
473 {
474 	if (commit->refcnt > 0) {
475 		commit->refcnt--;
476 		if (commit->refcnt > 0)
477 			return;
478 	}
479 
480 	got_object_id_queue_free(&commit->parent_ids);
481 	free(commit->tree_id);
482 	free(commit->author);
483 	free(commit->committer);
484 	free(commit->logmsg);
485 	free(commit);
486 }
487 
488 struct got_object_id *
489 got_object_commit_get_tree_id(struct got_commit_object *commit)
490 {
491 	return commit->tree_id;
492 }
493 
494 int
495 got_object_commit_get_nparents(struct got_commit_object *commit)
496 {
497 	return commit->nparents;
498 }
499 
500 const struct got_object_id_queue *
501 got_object_commit_get_parent_ids(struct got_commit_object *commit)
502 {
503 	return &commit->parent_ids;
504 }
505 
506 const char *
507 got_object_commit_get_author(struct got_commit_object *commit)
508 {
509 	return commit->author;
510 }
511 
512 time_t
513 got_object_commit_get_author_time(struct got_commit_object *commit)
514 {
515 	return commit->author_time;
516 }
517 
518 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
519 {
520 	return commit->author_gmtoff;
521 }
522 
523 const char *
524 got_object_commit_get_committer(struct got_commit_object *commit)
525 {
526 	return commit->committer;
527 }
528 
529 time_t
530 got_object_commit_get_committer_time(struct got_commit_object *commit)
531 {
532 	return commit->committer_time;
533 }
534 
535 time_t
536 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
537 {
538 	return commit->committer_gmtoff;
539 }
540 
541 const struct got_error *
542 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
543 {
544 	const struct got_error *err = NULL;
545 	const char *src;
546 	char *dst;
547 	size_t len;
548 
549 	len = strlen(commit->logmsg);
550 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
551 	if (*logmsg == NULL)
552 		return got_error_from_errno("malloc");
553 
554 	/*
555 	 * Strip out unusual headers. Headers are separated from the commit
556 	 * message body by a single empty line.
557 	 */
558 	src = commit->logmsg;
559 	dst = *logmsg;
560 	while (*src != '\0' && *src != '\n') {
561 		int copy_header = 1, eol = 0;
562 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
563 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
564 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
565 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
566 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
567 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
568 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
569 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
570 			copy_header = 0;
571 
572 		while (*src != '\0' && !eol) {
573 			if (copy_header) {
574 				*dst = *src;
575 				dst++;
576 			}
577 			if (*src == '\n')
578 				eol = 1;
579 			src++;
580 		}
581 	}
582 	*dst = '\0';
583 
584 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
585 		err = got_error(GOT_ERR_NO_SPACE);
586 		goto done;
587 	}
588 
589 	/* Trim redundant trailing whitespace. */
590 	len = strlen(*logmsg);
591 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
592 	    isspace((unsigned char)(*logmsg)[len - 1])) {
593 		(*logmsg)[len - 1] = '\0';
594 		len--;
595 	}
596 
597 	/* Append a trailing newline if missing. */
598 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
599 		(*logmsg)[len] = '\n';
600 		(*logmsg)[len + 1] = '\0';
601 	}
602 done:
603 	if (err) {
604 		free(*logmsg);
605 		*logmsg = NULL;
606 	}
607 	return err;
608 }
609 
610 const char *
611 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
612 {
613 	return commit->logmsg;
614 }
615 
616 const struct got_error *
617 got_object_parse_commit(struct got_commit_object **commit, char *buf,
618     size_t len)
619 {
620 	const struct got_error *err = NULL;
621 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
622 	char *s = buf;
623 	size_t label_len;
624 	ssize_t remain = (ssize_t)len;
625 
626 	if (remain == 0)
627 		return got_error(GOT_ERR_BAD_OBJ_DATA);
628 
629 	*commit = got_object_commit_alloc_partial();
630 	if (*commit == NULL)
631 		return got_error_from_errno("got_object_commit_alloc_partial");
632 
633 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
634 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
635 		remain -= label_len;
636 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
637 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
638 			goto done;
639 		}
640 		s += label_len;
641 		if (!got_parse_object_id((*commit)->tree_id, s, algo)) {
642 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
643 			goto done;
644 		}
645 		remain -= SHA1_DIGEST_STRING_LENGTH;
646 		s += SHA1_DIGEST_STRING_LENGTH;
647 	} else {
648 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
649 		goto done;
650 	}
651 
652 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
653 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
654 		remain -= label_len;
655 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
656 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
657 			goto done;
658 		}
659 		s += label_len;
660 		err = got_object_commit_add_parent(*commit, s);
661 		if (err)
662 			goto done;
663 
664 		remain -= SHA1_DIGEST_STRING_LENGTH;
665 		s += SHA1_DIGEST_STRING_LENGTH;
666 	}
667 
668 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
669 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
670 		char *p;
671 		size_t slen;
672 
673 		remain -= label_len;
674 		if (remain <= 0) {
675 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
676 			goto done;
677 		}
678 		s += label_len;
679 		p = memchr(s, '\n', remain);
680 		if (p == NULL) {
681 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
682 			goto done;
683 		}
684 		*p = '\0';
685 		slen = strlen(s);
686 		err = parse_commit_time(&(*commit)->author_time,
687 		    &(*commit)->author_gmtoff, s);
688 		if (err)
689 			goto done;
690 		(*commit)->author = strdup(s);
691 		if ((*commit)->author == NULL) {
692 			err = got_error_from_errno("strdup");
693 			goto done;
694 		}
695 		s += slen + 1;
696 		remain -= slen + 1;
697 	}
698 
699 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
700 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
701 		char *p;
702 		size_t slen;
703 
704 		remain -= label_len;
705 		if (remain <= 0) {
706 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
707 			goto done;
708 		}
709 		s += label_len;
710 		p = memchr(s, '\n', remain);
711 		if (p == NULL) {
712 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
713 			goto done;
714 		}
715 		*p = '\0';
716 		slen = strlen(s);
717 		err = parse_commit_time(&(*commit)->committer_time,
718 		    &(*commit)->committer_gmtoff, s);
719 		if (err)
720 			goto done;
721 		(*commit)->committer = strdup(s);
722 		if ((*commit)->committer == NULL) {
723 			err = got_error_from_errno("strdup");
724 			goto done;
725 		}
726 		s += slen + 1;
727 		remain -= slen + 1;
728 	}
729 
730 	(*commit)->logmsg = strndup(s, remain);
731 	if ((*commit)->logmsg == NULL) {
732 		err = got_error_from_errno("strndup");
733 		goto done;
734 	}
735 done:
736 	if (err) {
737 		got_object_commit_close(*commit);
738 		*commit = NULL;
739 	}
740 	return err;
741 }
742 
743 const struct got_error *
744 got_object_read_commit(struct got_commit_object **commit, int fd,
745     struct got_object_id *expected_id, size_t expected_size)
746 {
747 	struct got_object *obj = NULL;
748 	const struct got_error *err = NULL;
749 	size_t len;
750 	uint8_t *p;
751 	struct got_inflate_checksum csum;
752 	struct got_hash ctx;
753 	struct got_object_id id;
754 
755 	got_hash_init(&ctx, GOT_HASH_SHA1);
756 	memset(&csum, 0, sizeof(csum));
757 	csum.output_ctx = &ctx;
758 
759 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
760 	if (err)
761 		return err;
762 
763 	got_hash_final_object_id(&ctx, &id);
764 	if (got_object_id_cmp(expected_id, &id) != 0) {
765 		err = got_error_checksum(expected_id);
766 		goto done;
767 	}
768 
769 	err = got_object_parse_header(&obj, p, len);
770 	if (err)
771 		goto done;
772 
773 	if (len < obj->hdrlen + obj->size) {
774 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
775 		goto done;
776 	}
777 
778 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
779 		err = got_error(GOT_ERR_OBJ_TYPE);
780 		goto done;
781 	}
782 
783 	/* Skip object header. */
784 	len -= obj->hdrlen;
785 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
786 done:
787 	free(p);
788 	if (obj)
789 		got_object_close(obj);
790 	return err;
791 }
792 
793 void
794 got_object_tree_close(struct got_tree_object *tree)
795 {
796 	if (tree->refcnt > 0) {
797 		tree->refcnt--;
798 		if (tree->refcnt > 0)
799 			return;
800 	}
801 
802 	free(tree->entries);
803 	free(tree);
804 }
805 
806 static const struct got_error *
807 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
808     size_t maxlen)
809 {
810 	char *p, *space;
811 
812 	*elen = 0;
813 
814 	*elen = strnlen(buf, maxlen) + 1;
815 	if (*elen > maxlen)
816 		return got_error(GOT_ERR_BAD_OBJ_DATA);
817 
818 	space = memchr(buf, ' ', *elen);
819 	if (space == NULL || space <= buf)
820 		return got_error(GOT_ERR_BAD_OBJ_DATA);
821 
822 	pte->mode = 0;
823 	p = buf;
824 	while (p < space) {
825 		if (*p < '0' || *p > '7')
826 			return got_error(GOT_ERR_BAD_OBJ_DATA);
827 		pte->mode <<= 3;
828 		pte->mode |= *p - '0';
829 		p++;
830 	}
831 
832 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
833 		return got_error(GOT_ERR_BAD_OBJ_DATA);
834 
835 	pte->name = space + 1;
836 	pte->namelen = strlen(pte->name);
837 	buf += *elen;
838 	pte->id = buf;
839 	*elen += SHA1_DIGEST_LENGTH;
840 	return NULL;
841 }
842 
843 static int
844 pte_cmp(const void *pa, const void *pb)
845 {
846 	const struct got_parsed_tree_entry *a = pa, *b = pb;
847 
848 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
849 }
850 
851 const struct got_error *
852 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
853     size_t *nentries_alloc, uint8_t *buf, size_t len)
854 {
855 	const struct got_error *err = NULL;
856 	size_t remain = len;
857 	const size_t nalloc = 16;
858 	struct got_parsed_tree_entry *pte;
859 	int i;
860 
861 	*nentries = 0;
862 	if (remain == 0)
863 		return NULL; /* tree is empty */
864 
865 	while (remain > 0) {
866 		size_t elen;
867 
868 		if (*nentries >= *nentries_alloc) {
869 			pte = recallocarray(*entries, *nentries_alloc,
870 			    *nentries_alloc + nalloc, sizeof(**entries));
871 			if (pte == NULL) {
872 				err = got_error_from_errno("recallocarray");
873 				goto done;
874 			}
875 			*entries = pte;
876 			*nentries_alloc += nalloc;
877 		}
878 
879 		pte = &(*entries)[*nentries];
880 		err = parse_tree_entry(pte, &elen, buf, remain);
881 		if (err)
882 			goto done;
883 		buf += elen;
884 		remain -= elen;
885 		(*nentries)++;
886 	}
887 
888 	if (remain != 0) {
889 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
890 		goto done;
891 	}
892 
893 	if (*nentries > 1) {
894 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
895 
896 		for (i = 0; i < *nentries - 1; i++) {
897 			struct got_parsed_tree_entry *prev = &(*entries)[i];
898 			pte = &(*entries)[i + 1];
899 			if (got_path_cmp(prev->name, pte->name,
900 			    prev->namelen, pte->namelen) == 0) {
901 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
902 				break;
903 			}
904 		}
905 	}
906 done:
907 	if (err)
908 		*nentries = 0;
909 	return err;
910 }
911 
912 const struct got_error *
913 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
914     size_t *nentries_alloc, uint8_t **p, int fd,
915     struct got_object_id *expected_id)
916 {
917 	const struct got_error *err = NULL;
918 	struct got_object *obj = NULL;
919 	size_t len;
920 	struct got_inflate_checksum csum;
921 	struct got_hash ctx;
922 	struct got_object_id id;
923 
924 	got_hash_init(&ctx, GOT_HASH_SHA1);
925 	memset(&csum, 0, sizeof(csum));
926 	csum.output_ctx = &ctx;
927 
928 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
929 	if (err)
930 		return err;
931 
932 	got_hash_final_object_id(&ctx, &id);
933 	if (got_object_id_cmp(expected_id, &id) != 0) {
934 		err = got_error_checksum(expected_id);
935 		goto done;
936 	}
937 
938 	err = got_object_parse_header(&obj, *p, len);
939 	if (err)
940 		goto done;
941 
942 	if (len < obj->hdrlen + obj->size) {
943 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
944 		goto done;
945 	}
946 
947 	/* Skip object header. */
948 	len -= obj->hdrlen;
949 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
950 	    *p + obj->hdrlen, len);
951 done:
952 	if (obj)
953 		got_object_close(obj);
954 	return err;
955 }
956 
957 void
958 got_object_tag_close(struct got_tag_object *tag)
959 {
960 	if (tag->refcnt > 0) {
961 		tag->refcnt--;
962 		if (tag->refcnt > 0)
963 			return;
964 	}
965 
966 	free(tag->tag);
967 	free(tag->tagger);
968 	free(tag->tagmsg);
969 	free(tag);
970 }
971 
972 const struct got_error *
973 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
974 {
975 	const struct got_error *err = NULL;
976 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
977 	size_t remain = len;
978 	char *s = buf;
979 	size_t label_len;
980 
981 	if (remain == 0)
982 		return got_error(GOT_ERR_BAD_OBJ_DATA);
983 
984 	*tag = calloc(1, sizeof(**tag));
985 	if (*tag == NULL)
986 		return got_error_from_errno("calloc");
987 
988 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
989 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
990 		remain -= label_len;
991 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
992 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
993 			goto done;
994 		}
995 		s += label_len;
996 		if (!got_parse_object_id(&(*tag)->id, s, algo)) {
997 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
998 			goto done;
999 		}
1000 		remain -= SHA1_DIGEST_STRING_LENGTH;
1001 		s += SHA1_DIGEST_STRING_LENGTH;
1002 	} else {
1003 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1004 		goto done;
1005 	}
1006 
1007 	if (remain <= 0) {
1008 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1009 		goto done;
1010 	}
1011 
1012 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1013 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1014 		remain -= label_len;
1015 		if (remain <= 0) {
1016 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1017 			goto done;
1018 		}
1019 		s += label_len;
1020 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1021 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1022 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1023 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1024 			s += label_len;
1025 			remain -= label_len;
1026 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1027 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1028 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1029 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1030 			s += label_len;
1031 			remain -= label_len;
1032 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1033 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1034 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1035 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1036 			s += label_len;
1037 			remain -= label_len;
1038 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1039 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1040 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1041 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1042 			s += label_len;
1043 			remain -= label_len;
1044 		} else {
1045 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1046 			goto done;
1047 		}
1048 
1049 		if (remain <= 0 || *s != '\n') {
1050 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1051 			goto done;
1052 		}
1053 		s++;
1054 		remain--;
1055 		if (remain <= 0) {
1056 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1057 			goto done;
1058 		}
1059 	} else {
1060 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1061 		goto done;
1062 	}
1063 
1064 	label_len = strlen(GOT_TAG_LABEL_TAG);
1065 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1066 		char *p;
1067 		size_t slen;
1068 		remain -= label_len;
1069 		if (remain <= 0) {
1070 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1071 			goto done;
1072 		}
1073 		s += label_len;
1074 		p = memchr(s, '\n', remain);
1075 		if (p == NULL) {
1076 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1077 			goto done;
1078 		}
1079 		*p = '\0';
1080 		slen = strlen(s);
1081 		(*tag)->tag = strndup(s, slen);
1082 		if ((*tag)->tag == NULL) {
1083 			err = got_error_from_errno("strndup");
1084 			goto done;
1085 		}
1086 		s += slen + 1;
1087 		remain -= slen + 1;
1088 		if (remain <= 0) {
1089 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1090 			goto done;
1091 		}
1092 	} else {
1093 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1094 		goto done;
1095 	}
1096 
1097 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1098 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1099 		char *p;
1100 		size_t slen;
1101 
1102 		remain -= label_len;
1103 		if (remain <= 0) {
1104 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1105 			goto done;
1106 		}
1107 		s += label_len;
1108 		p = memchr(s, '\n', remain);
1109 		if (p == NULL) {
1110 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1111 			goto done;
1112 		}
1113 		*p = '\0';
1114 		slen = strlen(s);
1115 		err = parse_commit_time(&(*tag)->tagger_time,
1116 		    &(*tag)->tagger_gmtoff, s);
1117 		if (err)
1118 			goto done;
1119 		(*tag)->tagger = strdup(s);
1120 		if ((*tag)->tagger == NULL) {
1121 			err = got_error_from_errno("strdup");
1122 			goto done;
1123 		}
1124 		s += slen + 1;
1125 		remain -= slen + 1;
1126 		if (remain < 0) {
1127 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1128 			goto done;
1129 		}
1130 	} else {
1131 		/* Some old tags in the Linux git repo have no tagger. */
1132 		(*tag)->tagger = strdup("");
1133 		if ((*tag)->tagger == NULL) {
1134 			err = got_error_from_errno("strdup");
1135 			goto done;
1136 		}
1137 	}
1138 
1139 	(*tag)->tagmsg = strndup(s, remain);
1140 	if ((*tag)->tagmsg == NULL) {
1141 		err = got_error_from_errno("strndup");
1142 		goto done;
1143 	}
1144 done:
1145 	if (err) {
1146 		got_object_tag_close(*tag);
1147 		*tag = NULL;
1148 	}
1149 	return err;
1150 }
1151 
1152 const struct got_error *
1153 got_object_read_tag(struct got_tag_object **tag, int fd,
1154     struct got_object_id *expected_id, size_t expected_size)
1155 {
1156 	const struct got_error *err = NULL;
1157 	struct got_object *obj = NULL;
1158 	size_t len;
1159 	uint8_t *p;
1160 	struct got_inflate_checksum csum;
1161 	struct got_hash ctx;
1162 	struct got_object_id id;
1163 
1164 	got_hash_init(&ctx, GOT_HASH_SHA1);
1165 	memset(&csum, 0, sizeof(csum));
1166 	csum.output_ctx = &ctx;
1167 
1168 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1169 	    expected_size, fd);
1170 	if (err)
1171 		return err;
1172 
1173 	got_hash_final_object_id(&ctx, &id);
1174 	if (got_object_id_cmp(expected_id, &id) != 0) {
1175 		err = got_error_checksum(expected_id);
1176 		goto done;
1177 	}
1178 
1179 	err = got_object_parse_header(&obj, p, len);
1180 	if (err)
1181 		goto done;
1182 
1183 	if (len < obj->hdrlen + obj->size) {
1184 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1185 		goto done;
1186 	}
1187 
1188 	/* Skip object header. */
1189 	len -= obj->hdrlen;
1190 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1191 done:
1192 	free(p);
1193 	if (obj)
1194 		got_object_close(obj);
1195 	return err;
1196 }
1197 
1198 const struct got_error *
1199 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1200 {
1201 	const struct got_error *err = NULL;
1202 	static const size_t blocksize = 512;
1203 	size_t n, total, remain;
1204 	uint8_t *buf;
1205 
1206 	*outbuf = NULL;
1207 	*outlen = 0;
1208 
1209 	buf = malloc(blocksize);
1210 	if (buf == NULL)
1211 		return got_error_from_errno("malloc");
1212 
1213 	remain = blocksize;
1214 	total = 0;
1215 	for (;;) {
1216 		if (remain == 0) {
1217 			uint8_t *newbuf;
1218 			newbuf = reallocarray(buf, 1, total + blocksize);
1219 			if (newbuf == NULL) {
1220 				err = got_error_from_errno("reallocarray");
1221 				goto done;
1222 			}
1223 			buf = newbuf;
1224 			remain += blocksize;
1225 		}
1226 		n = fread(buf + total, 1, remain, f);
1227 		if (n == 0) {
1228 			if (ferror(f)) {
1229 				err = got_ferror(f, GOT_ERR_IO);
1230 				goto done;
1231 			}
1232 			break; /* EOF */
1233 		}
1234 		remain -= n;
1235 		total += n;
1236 	};
1237 
1238 done:
1239 	if (err == NULL) {
1240 		*outbuf = buf;
1241 		*outlen = total;
1242 	} else
1243 		free(buf);
1244 	return err;
1245 }