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 <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/socket.h>
23 #include <sys/wait.h>
24 #include <sys/mman.h>
25 
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdlib.h>
29 #include <string.h>
30 #include <stdint.h>
31 #include <sha1.h>
32 #include <zlib.h>
33 #include <ctype.h>
34 #include <limits.h>
35 #include <imsg.h>
36 #include <time.h>
37 #include <unistd.h>
38 
39 #include "got_error.h"
40 #include "got_object.h"
41 #include "got_repository.h"
42 #include "got_opentemp.h"
43 #include "got_path.h"
44 
45 #include "got_lib_sha1.h"
46 #include "got_lib_delta.h"
47 #include "got_lib_inflate.h"
48 #include "got_lib_object.h"
49 #include "got_lib_object_parse.h"
50 #include "got_lib_object_cache.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_repository.h"
53 
54 #ifndef nitems
55 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
56 #endif
57 
58 struct got_object_id *
59 got_object_id_dup(struct got_object_id *id1)
60 {
61 	struct got_object_id *id2;
62 
63 	id2 = malloc(sizeof(*id2));
64 	if (id2 == NULL)
65 		return NULL;
66 	memcpy(id2, id1, sizeof(*id2));
67 	return id2;
68 }
69 
70 int
71 got_object_id_cmp(const struct got_object_id *id1,
72     const struct got_object_id *id2)
73 {
74 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
75 }
76 
77 const struct got_error *
78 got_object_qid_alloc_partial(struct got_object_qid **qid)
79 {
80 	*qid = malloc(sizeof(**qid));
81 	if (*qid == NULL)
82 		return got_error_from_errno("malloc");
83 
84 	(*qid)->data = NULL;
85 	return NULL;
86 }
87 
88 const struct got_error *
89 got_object_id_str(char **outbuf, struct got_object_id *id)
90 {
91 	static const size_t len = SHA1_DIGEST_STRING_LENGTH;
92 
93 	*outbuf = malloc(len);
94 	if (*outbuf == NULL)
95 		return got_error_from_errno("malloc");
96 
97 	if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
98 		free(*outbuf);
99 		*outbuf = NULL;
100 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
101 	}
102 
103 	return NULL;
104 }
105 
106 void
107 got_object_close(struct got_object *obj)
108 {
109 	if (obj->refcnt > 0) {
110 		obj->refcnt--;
111 		if (obj->refcnt > 0)
112 			return;
113 	}
114 
115 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
116 		struct got_delta *delta;
117 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
118 			delta = STAILQ_FIRST(&obj->deltas.entries);
119 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
120 			free(delta);
121 		}
122 	}
123 	free(obj);
124 }
125 
126 const struct got_error *
127 got_object_raw_close(struct got_raw_object *obj)
128 {
129 	const struct got_error *err = NULL;
130 
131 	if (obj->refcnt > 0) {
132 		obj->refcnt--;
133 		if (obj->refcnt > 0)
134 			return NULL;
135 	}
136 
137 	if (obj->close_cb)
138 		obj->close_cb(obj);
139 
140 	if (obj->f == NULL) {
141 		if (obj->fd != -1) {
142 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
143 				err = got_error_from_errno("munmap");
144 			if (close(obj->fd) == -1 && err == NULL)
145 				err = got_error_from_errno("close");
146 		} else
147 			free(obj->data);
148 	} else {
149 		if (fclose(obj->f) == EOF && err == NULL)
150 			err = got_error_from_errno("fclose");
151 	}
152 	free(obj);
153 	return err;
154 }
155 
156 void
157 got_object_qid_free(struct got_object_qid *qid)
158 {
159 	free(qid);
160 }
161 
162 void
163 got_object_id_queue_free(struct got_object_id_queue *ids)
164 {
165 	struct got_object_qid *qid;
166 
167 	while (!STAILQ_EMPTY(ids)) {
168 		qid = STAILQ_FIRST(ids);
169 		STAILQ_REMOVE_HEAD(ids, entry);
170 		got_object_qid_free(qid);
171 	}
172 }
173 
174 const struct got_error *
175 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
176 {
177 	const char *obj_labels[] = {
178 		GOT_OBJ_LABEL_COMMIT,
179 		GOT_OBJ_LABEL_TREE,
180 		GOT_OBJ_LABEL_BLOB,
181 		GOT_OBJ_LABEL_TAG,
182 	};
183 	const int obj_types[] = {
184 		GOT_OBJ_TYPE_COMMIT,
185 		GOT_OBJ_TYPE_TREE,
186 		GOT_OBJ_TYPE_BLOB,
187 		GOT_OBJ_TYPE_TAG,
188 	};
189 	int type = 0;
190 	size_t size = 0;
191 	size_t i;
192 	char *end;
193 
194 	*obj = NULL;
195 
196 	end = memchr(buf, '\0', len);
197 	if (end == NULL)
198 		return got_error(GOT_ERR_BAD_OBJ_HDR);
199 
200 	for (i = 0; i < nitems(obj_labels); i++) {
201 		const char *label = obj_labels[i];
202 		size_t label_len = strlen(label);
203 		const char *errstr;
204 
205 		if (len <= label_len || buf + label_len >= end ||
206 		    strncmp(buf, label, label_len) != 0)
207 			continue;
208 
209 		type = obj_types[i];
210 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
211 		if (errstr != NULL)
212 			return got_error(GOT_ERR_BAD_OBJ_HDR);
213 		break;
214 	}
215 
216 	if (type == 0)
217 		return got_error(GOT_ERR_BAD_OBJ_HDR);
218 
219 	*obj = calloc(1, sizeof(**obj));
220 	if (*obj == NULL)
221 		return got_error_from_errno("calloc");
222 	(*obj)->type = type;
223 	(*obj)->hdrlen = end - buf + 1;
224 	(*obj)->size = size;
225 	return NULL;
226 }
227 
228 const struct got_error *
229 got_object_read_header(struct got_object **obj, int fd)
230 {
231 	const struct got_error *err;
232 	struct got_inflate_buf zb;
233 	uint8_t *buf;
234 	const size_t zbsize = 64;
235 	size_t outlen, totlen;
236 	int nbuf = 1;
237 
238 	*obj = NULL;
239 
240 	buf = malloc(zbsize);
241 	if (buf == NULL)
242 		return got_error_from_errno("malloc");
243 	buf[0] = '\0';
244 
245 	err = got_inflate_init(&zb, buf, zbsize, NULL);
246 	if (err)
247 		return err;
248 
249 	totlen = 0;
250 	do {
251 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
252 		if (err)
253 			goto done;
254 		if (outlen == 0)
255 			break;
256 		totlen += outlen;
257 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
258 			uint8_t *newbuf;
259 			nbuf++;
260 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
261 			if (newbuf == NULL) {
262 				err = got_error_from_errno("recallocarray");
263 				goto done;
264 			}
265 			buf = newbuf;
266 			zb.outbuf = newbuf + totlen;
267 			zb.outlen = (nbuf * zbsize) - totlen;
268 		}
269 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
270 
271 	err = got_object_parse_header(obj, buf, totlen);
272 done:
273 	free(buf);
274 	got_inflate_end(&zb);
275 	return err;
276 }
277 
278 const struct got_error *
279 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
280     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
281     int infd)
282 {
283 	const struct got_error *err = NULL;
284 	struct got_object *obj;
285 	struct got_inflate_checksum csum;
286 	uint8_t sha1[SHA1_DIGEST_LENGTH];
287 	SHA1_CTX sha1_ctx;
288 	size_t len, consumed;
289 	FILE *f = NULL;
290 
291 	*outbuf = NULL;
292 	*size = 0;
293 	*hdrlen = 0;
294 
295 	SHA1Init(&sha1_ctx);
296 	memset(&csum, 0, sizeof(csum));
297 	csum.output_sha1 = &sha1_ctx;
298 
299 	if (lseek(infd, SEEK_SET, 0) == -1)
300 		return got_error_from_errno("lseek");
301 
302 	err = got_object_read_header(&obj, infd);
303 	if (err)
304 		return err;
305 
306 	if (lseek(infd, SEEK_SET, 0) == -1)
307 		return got_error_from_errno("lseek");
308 
309 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
310 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
311 		    obj->size + obj->hdrlen, infd);
312 	} else {
313 		int fd;
314 		/*
315 		 * XXX This uses an extra file descriptor for no good reason.
316 		 * We should have got_inflate_fd_to_fd().
317 		 */
318 		fd = dup(infd);
319 		if (fd == -1)
320 			return got_error_from_errno("dup");
321 		f = fdopen(fd, "r");
322 		if (f == NULL) {
323 			err = got_error_from_errno("fdopen");
324 			abort();
325 			close(fd);
326 			goto done;
327 		}
328 		err = got_inflate_to_fd(&len, f, &csum, outfd);
329 	}
330 	if (err)
331 		goto done;
332 
333 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
334 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
335 		goto done;
336 	}
337 
338 	SHA1Final(sha1, &sha1_ctx);
339 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
340 		char buf[SHA1_DIGEST_STRING_LENGTH];
341 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
342 		    "checksum failure for object %s",
343 		    got_sha1_digest_to_str(expected_id->sha1, buf,
344 		    sizeof(buf)));
345 		goto done;
346 	}
347 
348 	*size = obj->size;
349 	*hdrlen = obj->hdrlen;
350 done:
351 	got_object_close(obj);
352 	if (f && fclose(f) == EOF && err == NULL)
353 		err = got_error_from_errno("fclose");
354 	return err;
355 }
356 
357 struct got_commit_object *
358 got_object_commit_alloc_partial(void)
359 {
360 	struct got_commit_object *commit;
361 
362 	commit = calloc(1, sizeof(*commit));
363 	if (commit == NULL)
364 		return NULL;
365 	commit->tree_id = malloc(sizeof(*commit->tree_id));
366 	if (commit->tree_id == NULL) {
367 		free(commit);
368 		return NULL;
369 	}
370 
371 	STAILQ_INIT(&commit->parent_ids);
372 
373 	return commit;
374 }
375 
376 const struct got_error *
377 got_object_commit_add_parent(struct got_commit_object *commit,
378     const char *id_str)
379 {
380 	const struct got_error *err = NULL;
381 	struct got_object_qid *qid;
382 
383 	err = got_object_qid_alloc_partial(&qid);
384 	if (err)
385 		return err;
386 
387 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
388 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
389 		got_object_qid_free(qid);
390 		return err;
391 	}
392 
393 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
394 	commit->nparents++;
395 
396 	return NULL;
397 }
398 
399 static const struct got_error *
400 parse_gmtoff(time_t *gmtoff, const char *tzstr)
401 {
402 	int sign = 1;
403 	const char *p = tzstr;
404 	time_t h, m;
405 
406 	*gmtoff = 0;
407 
408 	if (*p == '-')
409 		sign = -1;
410 	else if (*p != '+')
411 		return got_error(GOT_ERR_BAD_OBJ_DATA);
412 	p++;
413 	if (!isdigit(*p) && !isdigit(*(p + 1)))
414 		return got_error(GOT_ERR_BAD_OBJ_DATA);
415 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
416 
417 	p += 2;
418 	if (!isdigit(*p) && !isdigit(*(p + 1)))
419 		return got_error(GOT_ERR_BAD_OBJ_DATA);
420 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
421 
422 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
423 	return NULL;
424 }
425 
426 static const struct got_error *
427 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
428 {
429 	const struct got_error *err = NULL;
430 	const char *errstr;
431 	char *space, *tzstr;
432 
433 	/* Parse and strip off trailing timezone indicator string. */
434 	space = strrchr(committer, ' ');
435 	if (space == NULL)
436 		return got_error(GOT_ERR_BAD_OBJ_DATA);
437 	tzstr = strdup(space + 1);
438 	if (tzstr == NULL)
439 		return got_error_from_errno("strdup");
440 	err = parse_gmtoff(gmtoff, tzstr);
441 	free(tzstr);
442 	if (err) {
443 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
444 			return err;
445 		/* Old versions of Git omitted the timestamp. */
446 		*time = 0;
447 		*gmtoff = 0;
448 		return NULL;
449 	}
450 	*space = '\0';
451 
452 	/* Timestamp is separated from committer name + email by space. */
453 	space = strrchr(committer, ' ');
454 	if (space == NULL)
455 		return got_error(GOT_ERR_BAD_OBJ_DATA);
456 
457 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
458 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
459 	if (errstr)
460 		return got_error(GOT_ERR_BAD_OBJ_DATA);
461 
462 	/* Strip off parsed time information, leaving just author and email. */
463 	*space = '\0';
464 
465 	return NULL;
466 }
467 
468 void
469 got_object_commit_close(struct got_commit_object *commit)
470 {
471 	if (commit->refcnt > 0) {
472 		commit->refcnt--;
473 		if (commit->refcnt > 0)
474 			return;
475 	}
476 
477 	got_object_id_queue_free(&commit->parent_ids);
478 	free(commit->tree_id);
479 	free(commit->author);
480 	free(commit->committer);
481 	free(commit->logmsg);
482 	free(commit);
483 }
484 
485 struct got_object_id *
486 got_object_commit_get_tree_id(struct got_commit_object *commit)
487 {
488 	return commit->tree_id;
489 }
490 
491 int
492 got_object_commit_get_nparents(struct got_commit_object *commit)
493 {
494 	return commit->nparents;
495 }
496 
497 const struct got_object_id_queue *
498 got_object_commit_get_parent_ids(struct got_commit_object *commit)
499 {
500 	return &commit->parent_ids;
501 }
502 
503 const char *
504 got_object_commit_get_author(struct got_commit_object *commit)
505 {
506 	return commit->author;
507 }
508 
509 time_t
510 got_object_commit_get_author_time(struct got_commit_object *commit)
511 {
512 	return commit->author_time;
513 }
514 
515 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
516 {
517 	return commit->author_gmtoff;
518 }
519 
520 const char *
521 got_object_commit_get_committer(struct got_commit_object *commit)
522 {
523 	return commit->committer;
524 }
525 
526 time_t
527 got_object_commit_get_committer_time(struct got_commit_object *commit)
528 {
529 	return commit->committer_time;
530 }
531 
532 time_t
533 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
534 {
535 	return commit->committer_gmtoff;
536 }
537 
538 const struct got_error *
539 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
540 {
541 	const struct got_error *err = NULL;
542 	const char *src;
543 	char *dst;
544 	size_t len;
545 
546 	len = strlen(commit->logmsg);
547 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
548 	if (*logmsg == NULL)
549 		return got_error_from_errno("malloc");
550 
551 	/*
552 	 * Strip out unusual headers. Headers are separated from the commit
553 	 * message body by a single empty line.
554 	 */
555 	src = commit->logmsg;
556 	dst = *logmsg;
557 	while (*src != '\0' && *src != '\n') {
558 		int copy_header = 1, eol = 0;
559 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
560 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
561 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
562 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
563 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
564 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
565 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
566 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
567 			copy_header = 0;
568 
569 		while (*src != '\0' && !eol) {
570 			if (copy_header) {
571 				*dst = *src;
572 				dst++;
573 			}
574 			if (*src == '\n')
575 				eol = 1;
576 			src++;
577 		}
578 	}
579 	*dst = '\0';
580 
581 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
582 		err = got_error(GOT_ERR_NO_SPACE);
583 		goto done;
584 	}
585 
586 	/* Trim redundant trailing whitespace. */
587 	len = strlen(*logmsg);
588 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
589 	    isspace((unsigned char)(*logmsg)[len - 1])) {
590 		(*logmsg)[len - 1] = '\0';
591 		len--;
592 	}
593 
594 	/* Append a trailing newline if missing. */
595 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
596 		(*logmsg)[len] = '\n';
597 		(*logmsg)[len + 1] = '\0';
598 	}
599 done:
600 	if (err) {
601 		free(*logmsg);
602 		*logmsg = NULL;
603 	}
604 	return err;
605 }
606 
607 const char *
608 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
609 {
610 	return commit->logmsg;
611 }
612 
613 const struct got_error *
614 got_object_parse_commit(struct got_commit_object **commit, char *buf,
615     size_t len)
616 {
617 	const struct got_error *err = NULL;
618 	char *s = buf;
619 	size_t label_len;
620 	ssize_t remain = (ssize_t)len;
621 
622 	if (remain == 0)
623 		return got_error(GOT_ERR_BAD_OBJ_DATA);
624 
625 	*commit = got_object_commit_alloc_partial();
626 	if (*commit == NULL)
627 		return got_error_from_errno("got_object_commit_alloc_partial");
628 
629 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
630 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
631 		remain -= label_len;
632 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
633 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
634 			goto done;
635 		}
636 		s += label_len;
637 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
638 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
639 			goto done;
640 		}
641 		remain -= SHA1_DIGEST_STRING_LENGTH;
642 		s += SHA1_DIGEST_STRING_LENGTH;
643 	} else {
644 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
645 		goto done;
646 	}
647 
648 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
649 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
650 		remain -= label_len;
651 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
652 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
653 			goto done;
654 		}
655 		s += label_len;
656 		err = got_object_commit_add_parent(*commit, s);
657 		if (err)
658 			goto done;
659 
660 		remain -= SHA1_DIGEST_STRING_LENGTH;
661 		s += SHA1_DIGEST_STRING_LENGTH;
662 	}
663 
664 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
665 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
666 		char *p;
667 		size_t slen;
668 
669 		remain -= label_len;
670 		if (remain <= 0) {
671 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
672 			goto done;
673 		}
674 		s += label_len;
675 		p = memchr(s, '\n', remain);
676 		if (p == NULL) {
677 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
678 			goto done;
679 		}
680 		*p = '\0';
681 		slen = strlen(s);
682 		err = parse_commit_time(&(*commit)->author_time,
683 		    &(*commit)->author_gmtoff, s);
684 		if (err)
685 			goto done;
686 		(*commit)->author = strdup(s);
687 		if ((*commit)->author == NULL) {
688 			err = got_error_from_errno("strdup");
689 			goto done;
690 		}
691 		s += slen + 1;
692 		remain -= slen + 1;
693 	}
694 
695 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
696 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
697 		char *p;
698 		size_t slen;
699 
700 		remain -= label_len;
701 		if (remain <= 0) {
702 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
703 			goto done;
704 		}
705 		s += label_len;
706 		p = memchr(s, '\n', remain);
707 		if (p == NULL) {
708 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
709 			goto done;
710 		}
711 		*p = '\0';
712 		slen = strlen(s);
713 		err = parse_commit_time(&(*commit)->committer_time,
714 		    &(*commit)->committer_gmtoff, s);
715 		if (err)
716 			goto done;
717 		(*commit)->committer = strdup(s);
718 		if ((*commit)->committer == NULL) {
719 			err = got_error_from_errno("strdup");
720 			goto done;
721 		}
722 		s += slen + 1;
723 		remain -= slen + 1;
724 	}
725 
726 	(*commit)->logmsg = strndup(s, remain);
727 	if ((*commit)->logmsg == NULL) {
728 		err = got_error_from_errno("strndup");
729 		goto done;
730 	}
731 done:
732 	if (err) {
733 		got_object_commit_close(*commit);
734 		*commit = NULL;
735 	}
736 	return err;
737 }
738 
739 const struct got_error *
740 got_object_read_commit(struct got_commit_object **commit, int fd,
741     struct got_object_id *expected_id, size_t expected_size)
742 {
743 	struct got_object *obj = NULL;
744 	const struct got_error *err = NULL;
745 	size_t len;
746 	uint8_t *p;
747 	struct got_inflate_checksum csum;
748 	SHA1_CTX sha1_ctx;
749 	struct got_object_id id;
750 
751 	SHA1Init(&sha1_ctx);
752 	memset(&csum, 0, sizeof(csum));
753 	csum.output_sha1 = &sha1_ctx;
754 
755 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
756 	if (err)
757 		return err;
758 
759 	SHA1Final(id.sha1, &sha1_ctx);
760 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
761 		char buf[SHA1_DIGEST_STRING_LENGTH];
762 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
763 		    "checksum failure for object %s",
764 		    got_sha1_digest_to_str(expected_id->sha1, buf,
765 		    sizeof(buf)));
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 	SHA1_CTX sha1_ctx;
922 	struct got_object_id id;
923 
924 	SHA1Init(&sha1_ctx);
925 	memset(&csum, 0, sizeof(csum));
926 	csum.output_sha1 = &sha1_ctx;
927 
928 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
929 	if (err)
930 		return err;
931 
932 	SHA1Final(id.sha1, &sha1_ctx);
933 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
934 		char buf[SHA1_DIGEST_STRING_LENGTH];
935 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
936 		    "checksum failure for object %s",
937 		    got_sha1_digest_to_str(expected_id->sha1, buf,
938 		    sizeof(buf)));
939 		goto done;
940 	}
941 
942 	err = got_object_parse_header(&obj, *p, len);
943 	if (err)
944 		goto done;
945 
946 	if (len < obj->hdrlen + obj->size) {
947 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
948 		goto done;
949 	}
950 
951 	/* Skip object header. */
952 	len -= obj->hdrlen;
953 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
954 	    *p + obj->hdrlen, len);
955 done:
956 	if (obj)
957 		got_object_close(obj);
958 	return err;
959 }
960 
961 void
962 got_object_tag_close(struct got_tag_object *tag)
963 {
964 	if (tag->refcnt > 0) {
965 		tag->refcnt--;
966 		if (tag->refcnt > 0)
967 			return;
968 	}
969 
970 	free(tag->tag);
971 	free(tag->tagger);
972 	free(tag->tagmsg);
973 	free(tag);
974 }
975 
976 const struct got_error *
977 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
978 {
979 	const struct got_error *err = NULL;
980 	size_t remain = len;
981 	char *s = buf;
982 	size_t label_len;
983 
984 	if (remain == 0)
985 		return got_error(GOT_ERR_BAD_OBJ_DATA);
986 
987 	*tag = calloc(1, sizeof(**tag));
988 	if (*tag == NULL)
989 		return got_error_from_errno("calloc");
990 
991 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
992 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
993 		remain -= label_len;
994 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
995 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
996 			goto done;
997 		}
998 		s += label_len;
999 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
1000 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1001 			goto done;
1002 		}
1003 		remain -= SHA1_DIGEST_STRING_LENGTH;
1004 		s += SHA1_DIGEST_STRING_LENGTH;
1005 	} else {
1006 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1007 		goto done;
1008 	}
1009 
1010 	if (remain <= 0) {
1011 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1012 		goto done;
1013 	}
1014 
1015 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1016 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1017 		remain -= label_len;
1018 		if (remain <= 0) {
1019 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1020 			goto done;
1021 		}
1022 		s += label_len;
1023 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1024 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1025 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1026 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1027 			s += label_len;
1028 			remain -= label_len;
1029 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1030 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1031 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1032 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1033 			s += label_len;
1034 			remain -= label_len;
1035 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1036 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1037 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1038 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1039 			s += label_len;
1040 			remain -= label_len;
1041 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1042 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1043 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1044 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1045 			s += label_len;
1046 			remain -= label_len;
1047 		} else {
1048 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1049 			goto done;
1050 		}
1051 
1052 		if (remain <= 0 || *s != '\n') {
1053 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1054 			goto done;
1055 		}
1056 		s++;
1057 		remain--;
1058 		if (remain <= 0) {
1059 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1060 			goto done;
1061 		}
1062 	} else {
1063 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1064 		goto done;
1065 	}
1066 
1067 	label_len = strlen(GOT_TAG_LABEL_TAG);
1068 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1069 		char *p;
1070 		size_t slen;
1071 		remain -= label_len;
1072 		if (remain <= 0) {
1073 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1074 			goto done;
1075 		}
1076 		s += label_len;
1077 		p = memchr(s, '\n', remain);
1078 		if (p == NULL) {
1079 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1080 			goto done;
1081 		}
1082 		*p = '\0';
1083 		slen = strlen(s);
1084 		(*tag)->tag = strndup(s, slen);
1085 		if ((*tag)->tag == NULL) {
1086 			err = got_error_from_errno("strndup");
1087 			goto done;
1088 		}
1089 		s += slen + 1;
1090 		remain -= slen + 1;
1091 		if (remain <= 0) {
1092 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1093 			goto done;
1094 		}
1095 	} else {
1096 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1097 		goto done;
1098 	}
1099 
1100 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1101 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1102 		char *p;
1103 		size_t slen;
1104 
1105 		remain -= label_len;
1106 		if (remain <= 0) {
1107 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1108 			goto done;
1109 		}
1110 		s += label_len;
1111 		p = memchr(s, '\n', remain);
1112 		if (p == NULL) {
1113 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1114 			goto done;
1115 		}
1116 		*p = '\0';
1117 		slen = strlen(s);
1118 		err = parse_commit_time(&(*tag)->tagger_time,
1119 		    &(*tag)->tagger_gmtoff, s);
1120 		if (err)
1121 			goto done;
1122 		(*tag)->tagger = strdup(s);
1123 		if ((*tag)->tagger == NULL) {
1124 			err = got_error_from_errno("strdup");
1125 			goto done;
1126 		}
1127 		s += slen + 1;
1128 		remain -= slen + 1;
1129 		if (remain < 0) {
1130 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1131 			goto done;
1132 		}
1133 	} else {
1134 		/* Some old tags in the Linux git repo have no tagger. */
1135 		(*tag)->tagger = strdup("");
1136 		if ((*tag)->tagger == NULL) {
1137 			err = got_error_from_errno("strdup");
1138 			goto done;
1139 		}
1140 	}
1141 
1142 	(*tag)->tagmsg = strndup(s, remain);
1143 	if ((*tag)->tagmsg == NULL) {
1144 		err = got_error_from_errno("strndup");
1145 		goto done;
1146 	}
1147 done:
1148 	if (err) {
1149 		got_object_tag_close(*tag);
1150 		*tag = NULL;
1151 	}
1152 	return err;
1153 }
1154 
1155 const struct got_error *
1156 got_object_read_tag(struct got_tag_object **tag, int fd, 
1157     struct got_object_id *expected_id, size_t expected_size)
1158 {
1159 	const struct got_error *err = NULL;
1160 	struct got_object *obj = NULL;
1161 	size_t len;
1162 	uint8_t *p;
1163 	struct got_inflate_checksum csum;
1164 	SHA1_CTX sha1_ctx;
1165 	struct got_object_id id;
1166 
1167 	SHA1Init(&sha1_ctx);
1168 	memset(&csum, 0, sizeof(csum));
1169 	csum.output_sha1 = &sha1_ctx;
1170 
1171 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1172 	    expected_size, fd);
1173 	if (err)
1174 		return err;
1175 
1176 	SHA1Final(id.sha1, &sha1_ctx);
1177 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
1178 		char buf[SHA1_DIGEST_STRING_LENGTH];
1179 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
1180 		    "checksum failure for object %s",
1181 		    got_sha1_digest_to_str(expected_id->sha1, buf,
1182 		    sizeof(buf)));
1183 		goto done;
1184 	}
1185 
1186 	err = got_object_parse_header(&obj, p, len);
1187 	if (err)
1188 		goto done;
1189 
1190 	if (len < obj->hdrlen + obj->size) {
1191 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1192 		goto done;
1193 	}
1194 
1195 	/* Skip object header. */
1196 	len -= obj->hdrlen;
1197 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1198 done:
1199 	free(p);
1200 	if (obj)
1201 		got_object_close(obj);
1202 	return err;
1203 }
1204 
1205 const struct got_error *
1206 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1207 {
1208 	const struct got_error *err = NULL;
1209 	static const size_t blocksize = 512;
1210 	size_t n, total, remain;
1211 	uint8_t *buf;
1212 
1213 	*outbuf = NULL;
1214 	*outlen = 0;
1215 
1216 	buf = malloc(blocksize);
1217 	if (buf == NULL)
1218 		return got_error_from_errno("malloc");
1219 
1220 	remain = blocksize;
1221 	total = 0;
1222 	for (;;) {
1223 		if (remain == 0) {
1224 			uint8_t *newbuf;
1225 			newbuf = reallocarray(buf, 1, total + blocksize);
1226 			if (newbuf == NULL) {
1227 				err = got_error_from_errno("reallocarray");
1228 				goto done;
1229 			}
1230 			buf = newbuf;
1231 			remain += blocksize;
1232 		}
1233 		n = fread(buf + total, 1, remain, f);
1234 		if (n == 0) {
1235 			if (ferror(f)) {
1236 				err = got_ferror(f, GOT_ERR_IO);
1237 				goto done;
1238 			}
1239 			break; /* EOF */
1240 		}
1241 		remain -= n;
1242 		total += n;
1243 	};
1244 
1245 done:
1246 	if (err == NULL) {
1247 		*outbuf = buf;
1248 		*outlen = total;
1249 	} else
1250 		free(buf);
1251 	return err;
1252 }