Gotweb

Blob

Date:: Fri Feb 3 15:22:14 2023 UTC
Message:: use got_object_id_cmp instad of memcmp where possible in object_parse.c ok stsp@
Actions:: History | Blame | Raw File
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/uio.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 #include <sys/mman.h>
24 
25 #include <errno.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <stdint.h>
30 #include <zlib.h>
31 #include <ctype.h>
32 #include <limits.h>
33 #include <time.h>
34 #include <unistd.h>
35 
36 #include "got_compat.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_sha1.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 = SHA1_DIGEST_STRING_LENGTH;
91 
92 	*outbuf = malloc(len);
93 	if (*outbuf == NULL)
94 		return got_error_from_errno("malloc");
95 
96 	if (got_sha1_digest_to_str(id->sha1, *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 void
106 got_object_close(struct got_object *obj)
107 {
108 	if (obj->refcnt > 0) {
109 		obj->refcnt--;
110 		if (obj->refcnt > 0)
111 			return;
112 	}
113 
114 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
115 		struct got_delta *delta;
116 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
117 			delta = STAILQ_FIRST(&obj->deltas.entries);
118 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
119 			free(delta);
120 		}
121 	}
122 	free(obj);
123 }
124 
125 const struct got_error *
126 got_object_raw_close(struct got_raw_object *obj)
127 {
128 	const struct got_error *err = NULL;
129 
130 	if (obj->refcnt > 0) {
131 		obj->refcnt--;
132 		if (obj->refcnt > 0)
133 			return NULL;
134 	}
135 
136 	if (obj->close_cb)
137 		obj->close_cb(obj);
138 
139 	if (obj->f == NULL) {
140 		if (obj->fd != -1) {
141 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
142 				err = got_error_from_errno("munmap");
143 			if (close(obj->fd) == -1 && err == NULL)
144 				err = got_error_from_errno("close");
145 		} else
146 			free(obj->data);
147 	} else {
148 		if (fclose(obj->f) == EOF && err == NULL)
149 			err = got_error_from_errno("fclose");
150 	}
151 	free(obj);
152 	return err;
153 }
154 
155 void
156 got_object_qid_free(struct got_object_qid *qid)
157 {
158 	free(qid);
159 }
160 
161 void
162 got_object_id_queue_free(struct got_object_id_queue *ids)
163 {
164 	struct got_object_qid *qid;
165 
166 	while (!STAILQ_EMPTY(ids)) {
167 		qid = STAILQ_FIRST(ids);
168 		STAILQ_REMOVE_HEAD(ids, entry);
169 		got_object_qid_free(qid);
170 	}
171 }
172 
173 const struct got_error *
174 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
175 {
176 	const char *obj_labels[] = {
177 		GOT_OBJ_LABEL_COMMIT,
178 		GOT_OBJ_LABEL_TREE,
179 		GOT_OBJ_LABEL_BLOB,
180 		GOT_OBJ_LABEL_TAG,
181 	};
182 	const int obj_types[] = {
183 		GOT_OBJ_TYPE_COMMIT,
184 		GOT_OBJ_TYPE_TREE,
185 		GOT_OBJ_TYPE_BLOB,
186 		GOT_OBJ_TYPE_TAG,
187 	};
188 	int type = 0;
189 	size_t size = 0;
190 	size_t i;
191 	char *end;
192 
193 	*obj = NULL;
194 
195 	end = memchr(buf, '\0', len);
196 	if (end == NULL)
197 		return got_error(GOT_ERR_BAD_OBJ_HDR);
198 
199 	for (i = 0; i < nitems(obj_labels); i++) {
200 		const char *label = obj_labels[i];
201 		size_t label_len = strlen(label);
202 		const char *errstr;
203 
204 		if (len <= label_len || buf + label_len >= end ||
205 		    strncmp(buf, label, label_len) != 0)
206 			continue;
207 
208 		type = obj_types[i];
209 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
210 		if (errstr != NULL)
211 			return got_error(GOT_ERR_BAD_OBJ_HDR);
212 		break;
213 	}
214 
215 	if (type == 0)
216 		return got_error(GOT_ERR_BAD_OBJ_HDR);
217 
218 	*obj = calloc(1, sizeof(**obj));
219 	if (*obj == NULL)
220 		return got_error_from_errno("calloc");
221 	(*obj)->type = type;
222 	(*obj)->hdrlen = end - buf + 1;
223 	(*obj)->size = size;
224 	return NULL;
225 }
226 
227 const struct got_error *
228 got_object_read_header(struct got_object **obj, int fd)
229 {
230 	const struct got_error *err;
231 	struct got_inflate_buf zb;
232 	uint8_t *buf;
233 	const size_t zbsize = 64;
234 	size_t outlen, totlen;
235 	int nbuf = 1;
236 
237 	*obj = NULL;
238 
239 	buf = malloc(zbsize);
240 	if (buf == NULL)
241 		return got_error_from_errno("malloc");
242 	buf[0] = '\0';
243 
244 	err = got_inflate_init(&zb, buf, zbsize, NULL);
245 	if (err)
246 		return err;
247 
248 	totlen = 0;
249 	do {
250 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
251 		if (err)
252 			goto done;
253 		if (outlen == 0)
254 			break;
255 		totlen += outlen;
256 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
257 			uint8_t *newbuf;
258 			nbuf++;
259 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
260 			if (newbuf == NULL) {
261 				err = got_error_from_errno("recallocarray");
262 				goto done;
263 			}
264 			buf = newbuf;
265 			zb.outbuf = newbuf + totlen;
266 			zb.outlen = (nbuf * zbsize) - totlen;
267 		}
268 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
269 
270 	err = got_object_parse_header(obj, buf, totlen);
271 done:
272 	free(buf);
273 	got_inflate_end(&zb);
274 	return err;
275 }
276 
277 const struct got_error *
278 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
279     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
280     int infd)
281 {
282 	const struct got_error *err = NULL;
283 	struct got_object *obj;
284 	struct got_inflate_checksum csum;
285 	uint8_t sha1[SHA1_DIGEST_LENGTH];
286 	SHA1_CTX sha1_ctx;
287 	size_t len, consumed;
288 	FILE *f = NULL;
289 
290 	*outbuf = NULL;
291 	*size = 0;
292 	*hdrlen = 0;
293 
294 	SHA1Init(&sha1_ctx);
295 	memset(&csum, 0, sizeof(csum));
296 	csum.output_sha1 = &sha1_ctx;
297 
298 	if (lseek(infd, SEEK_SET, 0) == -1)
299 		return got_error_from_errno("lseek");
300 
301 	err = got_object_read_header(&obj, infd);
302 	if (err)
303 		return err;
304 
305 	if (lseek(infd, SEEK_SET, 0) == -1)
306 		return got_error_from_errno("lseek");
307 
308 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
309 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
310 		    obj->size + obj->hdrlen, infd);
311 	} else {
312 		int fd;
313 		/*
314 		 * XXX This uses an extra file descriptor for no good reason.
315 		 * We should have got_inflate_fd_to_fd().
316 		 */
317 		fd = dup(infd);
318 		if (fd == -1)
319 			return got_error_from_errno("dup");
320 		f = fdopen(fd, "r");
321 		if (f == NULL) {
322 			err = got_error_from_errno("fdopen");
323 			abort();
324 			close(fd);
325 			goto done;
326 		}
327 		err = got_inflate_to_fd(&len, f, &csum, outfd);
328 	}
329 	if (err)
330 		goto done;
331 
332 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
333 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
334 		goto done;
335 	}
336 
337 	SHA1Final(sha1, &sha1_ctx);
338 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
339 		char buf[SHA1_DIGEST_STRING_LENGTH];
340 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
341 		    "checksum failure for object %s",
342 		    got_sha1_digest_to_str(expected_id->sha1, buf,
343 		    sizeof(buf)));
344 		goto done;
345 	}
346 
347 	*size = obj->size;
348 	*hdrlen = obj->hdrlen;
349 done:
350 	got_object_close(obj);
351 	if (f && fclose(f) == EOF && err == NULL)
352 		err = got_error_from_errno("fclose");
353 	return err;
354 }
355 
356 struct got_commit_object *
357 got_object_commit_alloc_partial(void)
358 {
359 	struct got_commit_object *commit;
360 
361 	commit = calloc(1, sizeof(*commit));
362 	if (commit == NULL)
363 		return NULL;
364 	commit->tree_id = malloc(sizeof(*commit->tree_id));
365 	if (commit->tree_id == NULL) {
366 		free(commit);
367 		return NULL;
368 	}
369 
370 	STAILQ_INIT(&commit->parent_ids);
371 
372 	return commit;
373 }
374 
375 const struct got_error *
376 got_object_commit_add_parent(struct got_commit_object *commit,
377     const char *id_str)
378 {
379 	const struct got_error *err = NULL;
380 	struct got_object_qid *qid;
381 
382 	err = got_object_qid_alloc_partial(&qid);
383 	if (err)
384 		return err;
385 
386 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
387 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
388 		got_object_qid_free(qid);
389 		return err;
390 	}
391 
392 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
393 	commit->nparents++;
394 
395 	return NULL;
396 }
397 
398 static const struct got_error *
399 parse_gmtoff(time_t *gmtoff, const char *tzstr)
400 {
401 	int sign = 1;
402 	const char *p = tzstr;
403 	time_t h, m;
404 
405 	*gmtoff = 0;
406 
407 	if (*p == '-')
408 		sign = -1;
409 	else if (*p != '+')
410 		return got_error(GOT_ERR_BAD_OBJ_DATA);
411 	p++;
412 	if (!isdigit((unsigned char)*p) &&
413 	    !isdigit((unsigned char)*(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((unsigned char)*p) &&
419 	    !isdigit((unsigned char)*(p + 1)))
420 		return got_error(GOT_ERR_BAD_OBJ_DATA);
421 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
422 
423 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
424 	return NULL;
425 }
426 
427 static const struct got_error *
428 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
429 {
430 	const struct got_error *err = NULL;
431 	const char *errstr;
432 	char *space, *tzstr;
433 
434 	/* Parse and strip off trailing timezone indicator string. */
435 	space = strrchr(committer, ' ');
436 	if (space == NULL)
437 		return got_error(GOT_ERR_BAD_OBJ_DATA);
438 	tzstr = strdup(space + 1);
439 	if (tzstr == NULL)
440 		return got_error_from_errno("strdup");
441 	err = parse_gmtoff(gmtoff, tzstr);
442 	free(tzstr);
443 	if (err) {
444 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
445 			return err;
446 		/* Old versions of Git omitted the timestamp. */
447 		*time = 0;
448 		*gmtoff = 0;
449 		return NULL;
450 	}
451 	*space = '\0';
452 
453 	/* Timestamp is separated from committer name + email by space. */
454 	space = strrchr(committer, ' ');
455 	if (space == NULL)
456 		return got_error(GOT_ERR_BAD_OBJ_DATA);
457 
458 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
459 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
460 	if (errstr)
461 		return got_error(GOT_ERR_BAD_OBJ_DATA);
462 
463 	/* Strip off parsed time information, leaving just author and email. */
464 	*space = '\0';
465 
466 	return NULL;
467 }
468 
469 void
470 got_object_commit_close(struct got_commit_object *commit)
471 {
472 	if (commit->refcnt > 0) {
473 		commit->refcnt--;
474 		if (commit->refcnt > 0)
475 			return;
476 	}
477 
478 	got_object_id_queue_free(&commit->parent_ids);
479 	free(commit->tree_id);
480 	free(commit->author);
481 	free(commit->committer);
482 	free(commit->logmsg);
483 	free(commit);
484 }
485 
486 struct got_object_id *
487 got_object_commit_get_tree_id(struct got_commit_object *commit)
488 {
489 	return commit->tree_id;
490 }
491 
492 int
493 got_object_commit_get_nparents(struct got_commit_object *commit)
494 {
495 	return commit->nparents;
496 }
497 
498 const struct got_object_id_queue *
499 got_object_commit_get_parent_ids(struct got_commit_object *commit)
500 {
501 	return &commit->parent_ids;
502 }
503 
504 const char *
505 got_object_commit_get_author(struct got_commit_object *commit)
506 {
507 	return commit->author;
508 }
509 
510 time_t
511 got_object_commit_get_author_time(struct got_commit_object *commit)
512 {
513 	return commit->author_time;
514 }
515 
516 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
517 {
518 	return commit->author_gmtoff;
519 }
520 
521 const char *
522 got_object_commit_get_committer(struct got_commit_object *commit)
523 {
524 	return commit->committer;
525 }
526 
527 time_t
528 got_object_commit_get_committer_time(struct got_commit_object *commit)
529 {
530 	return commit->committer_time;
531 }
532 
533 time_t
534 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
535 {
536 	return commit->committer_gmtoff;
537 }
538 
539 const struct got_error *
540 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
541 {
542 	const struct got_error *err = NULL;
543 	const char *src;
544 	char *dst;
545 	size_t len;
546 
547 	len = strlen(commit->logmsg);
548 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
549 	if (*logmsg == NULL)
550 		return got_error_from_errno("malloc");
551 
552 	/*
553 	 * Strip out unusual headers. Headers are separated from the commit
554 	 * message body by a single empty line.
555 	 */
556 	src = commit->logmsg;
557 	dst = *logmsg;
558 	while (*src != '\0' && *src != '\n') {
559 		int copy_header = 1, eol = 0;
560 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
561 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
562 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
563 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
564 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
565 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
566 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
567 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
568 			copy_header = 0;
569 
570 		while (*src != '\0' && !eol) {
571 			if (copy_header) {
572 				*dst = *src;
573 				dst++;
574 			}
575 			if (*src == '\n')
576 				eol = 1;
577 			src++;
578 		}
579 	}
580 	*dst = '\0';
581 
582 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
583 		err = got_error(GOT_ERR_NO_SPACE);
584 		goto done;
585 	}
586 
587 	/* Trim redundant trailing whitespace. */
588 	len = strlen(*logmsg);
589 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
590 	    isspace((unsigned char)(*logmsg)[len - 1])) {
591 		(*logmsg)[len - 1] = '\0';
592 		len--;
593 	}
594 
595 	/* Append a trailing newline if missing. */
596 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
597 		(*logmsg)[len] = '\n';
598 		(*logmsg)[len + 1] = '\0';
599 	}
600 done:
601 	if (err) {
602 		free(*logmsg);
603 		*logmsg = NULL;
604 	}
605 	return err;
606 }
607 
608 const char *
609 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
610 {
611 	return commit->logmsg;
612 }
613 
614 const struct got_error *
615 got_object_parse_commit(struct got_commit_object **commit, char *buf,
616     size_t len)
617 {
618 	const struct got_error *err = NULL;
619 	char *s = buf;
620 	size_t label_len;
621 	ssize_t remain = (ssize_t)len;
622 
623 	if (remain == 0)
624 		return got_error(GOT_ERR_BAD_OBJ_DATA);
625 
626 	*commit = got_object_commit_alloc_partial();
627 	if (*commit == NULL)
628 		return got_error_from_errno("got_object_commit_alloc_partial");
629 
630 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
631 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
632 		remain -= label_len;
633 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
634 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
635 			goto done;
636 		}
637 		s += label_len;
638 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
639 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
640 			goto done;
641 		}
642 		remain -= SHA1_DIGEST_STRING_LENGTH;
643 		s += SHA1_DIGEST_STRING_LENGTH;
644 	} else {
645 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
646 		goto done;
647 	}
648 
649 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
650 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
651 		remain -= label_len;
652 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
653 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
654 			goto done;
655 		}
656 		s += label_len;
657 		err = got_object_commit_add_parent(*commit, s);
658 		if (err)
659 			goto done;
660 
661 		remain -= SHA1_DIGEST_STRING_LENGTH;
662 		s += SHA1_DIGEST_STRING_LENGTH;
663 	}
664 
665 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
666 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
667 		char *p;
668 		size_t slen;
669 
670 		remain -= label_len;
671 		if (remain <= 0) {
672 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
673 			goto done;
674 		}
675 		s += label_len;
676 		p = memchr(s, '\n', remain);
677 		if (p == NULL) {
678 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
679 			goto done;
680 		}
681 		*p = '\0';
682 		slen = strlen(s);
683 		err = parse_commit_time(&(*commit)->author_time,
684 		    &(*commit)->author_gmtoff, s);
685 		if (err)
686 			goto done;
687 		(*commit)->author = strdup(s);
688 		if ((*commit)->author == NULL) {
689 			err = got_error_from_errno("strdup");
690 			goto done;
691 		}
692 		s += slen + 1;
693 		remain -= slen + 1;
694 	}
695 
696 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
697 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
698 		char *p;
699 		size_t slen;
700 
701 		remain -= label_len;
702 		if (remain <= 0) {
703 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
704 			goto done;
705 		}
706 		s += label_len;
707 		p = memchr(s, '\n', remain);
708 		if (p == NULL) {
709 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
710 			goto done;
711 		}
712 		*p = '\0';
713 		slen = strlen(s);
714 		err = parse_commit_time(&(*commit)->committer_time,
715 		    &(*commit)->committer_gmtoff, s);
716 		if (err)
717 			goto done;
718 		(*commit)->committer = strdup(s);
719 		if ((*commit)->committer == NULL) {
720 			err = got_error_from_errno("strdup");
721 			goto done;
722 		}
723 		s += slen + 1;
724 		remain -= slen + 1;
725 	}
726 
727 	(*commit)->logmsg = strndup(s, remain);
728 	if ((*commit)->logmsg == NULL) {
729 		err = got_error_from_errno("strndup");
730 		goto done;
731 	}
732 done:
733 	if (err) {
734 		got_object_commit_close(*commit);
735 		*commit = NULL;
736 	}
737 	return err;
738 }
739 
740 const struct got_error *
741 got_object_read_commit(struct got_commit_object **commit, int fd,
742     struct got_object_id *expected_id, size_t expected_size)
743 {
744 	struct got_object *obj = NULL;
745 	const struct got_error *err = NULL;
746 	size_t len;
747 	uint8_t *p;
748 	struct got_inflate_checksum csum;
749 	SHA1_CTX sha1_ctx;
750 	struct got_object_id id;
751 
752 	SHA1Init(&sha1_ctx);
753 	memset(&csum, 0, sizeof(csum));
754 	csum.output_sha1 = &sha1_ctx;
755 
756 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
757 	if (err)
758 		return err;
759 
760 	SHA1Final(id.sha1, &sha1_ctx);
761 	if (got_object_id_cmp(expected_id, &id) != 0) {
762 		char buf[SHA1_DIGEST_STRING_LENGTH];
763 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
764 		    "checksum failure for object %s",
765 		    got_sha1_digest_to_str(expected_id->sha1, buf,
766 		    sizeof(buf)));
767 		goto done;
768 	}
769 
770 	err = got_object_parse_header(&obj, p, len);
771 	if (err)
772 		goto done;
773 
774 	if (len < obj->hdrlen + obj->size) {
775 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
776 		goto done;
777 	}
778 
779 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
780 		err = got_error(GOT_ERR_OBJ_TYPE);
781 		goto done;
782 	}
783 
784 	/* Skip object header. */
785 	len -= obj->hdrlen;
786 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
787 done:
788 	free(p);
789 	if (obj)
790 		got_object_close(obj);
791 	return err;
792 }
793 
794 void
795 got_object_tree_close(struct got_tree_object *tree)
796 {
797 	if (tree->refcnt > 0) {
798 		tree->refcnt--;
799 		if (tree->refcnt > 0)
800 			return;
801 	}
802 
803 	free(tree->entries);
804 	free(tree);
805 }
806 
807 static const struct got_error *
808 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
809     size_t maxlen)
810 {
811 	char *p, *space;
812 
813 	*elen = 0;
814 
815 	*elen = strnlen(buf, maxlen) + 1;
816 	if (*elen > maxlen)
817 		return got_error(GOT_ERR_BAD_OBJ_DATA);
818 
819 	space = memchr(buf, ' ', *elen);
820 	if (space == NULL || space <= buf)
821 		return got_error(GOT_ERR_BAD_OBJ_DATA);
822 
823 	pte->mode = 0;
824 	p = buf;
825 	while (p < space) {
826 		if (*p < '0' || *p > '7')
827 			return got_error(GOT_ERR_BAD_OBJ_DATA);
828 		pte->mode <<= 3;
829 		pte->mode |= *p - '0';
830 		p++;
831 	}
832 
833 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
834 		return got_error(GOT_ERR_BAD_OBJ_DATA);
835 
836 	pte->name = space + 1;
837 	pte->namelen = strlen(pte->name);
838 	buf += *elen;
839 	pte->id = buf;
840 	*elen += SHA1_DIGEST_LENGTH;
841 	return NULL;
842 }
843 
844 static int
845 pte_cmp(const void *pa, const void *pb)
846 {
847 	const struct got_parsed_tree_entry *a = pa, *b = pb;
848 
849 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
850 }
851 
852 const struct got_error *
853 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
854     size_t *nentries_alloc, uint8_t *buf, size_t len)
855 {
856 	const struct got_error *err = NULL;
857 	size_t remain = len;
858 	const size_t nalloc = 16;
859 	struct got_parsed_tree_entry *pte;
860 	int i;
861 
862 	*nentries = 0;
863 	if (remain == 0)
864 		return NULL; /* tree is empty */
865 
866 	while (remain > 0) {
867 		size_t elen;
868 
869 		if (*nentries >= *nentries_alloc) {
870 			pte = recallocarray(*entries, *nentries_alloc,
871 			    *nentries_alloc + nalloc, sizeof(**entries));
872 			if (pte == NULL) {
873 				err = got_error_from_errno("recallocarray");
874 				goto done;
875 			}
876 			*entries = pte;
877 			*nentries_alloc += nalloc;
878 		}
879 
880 		pte = &(*entries)[*nentries];
881 		err = parse_tree_entry(pte, &elen, buf, remain);
882 		if (err)
883 			goto done;
884 		buf += elen;
885 		remain -= elen;
886 		(*nentries)++;
887 	}
888 
889 	if (remain != 0) {
890 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
891 		goto done;
892 	}
893 
894 	if (*nentries > 1) {
895 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
896 
897 		for (i = 0; i < *nentries - 1; i++) {
898 			struct got_parsed_tree_entry *prev = &(*entries)[i];
899 			pte = &(*entries)[i + 1];
900 			if (got_path_cmp(prev->name, pte->name,
901 			    prev->namelen, pte->namelen) == 0) {
902 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
903 				break;
904 			}
905 		}
906 	}
907 done:
908 	if (err)
909 		*nentries = 0;
910 	return err;
911 }
912 
913 const struct got_error *
914 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
915     size_t *nentries_alloc, uint8_t **p, int fd,
916     struct got_object_id *expected_id)
917 {
918 	const struct got_error *err = NULL;
919 	struct got_object *obj = NULL;
920 	size_t len;
921 	struct got_inflate_checksum csum;
922 	SHA1_CTX sha1_ctx;
923 	struct got_object_id id;
924 
925 	SHA1Init(&sha1_ctx);
926 	memset(&csum, 0, sizeof(csum));
927 	csum.output_sha1 = &sha1_ctx;
928 
929 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
930 	if (err)
931 		return err;
932 
933 	SHA1Final(id.sha1, &sha1_ctx);
934 	if (got_object_id_cmp(expected_id, &id) != 0) {
935 		char buf[SHA1_DIGEST_STRING_LENGTH];
936 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
937 		    "checksum failure for object %s",
938 		    got_sha1_digest_to_str(expected_id->sha1, buf,
939 		    sizeof(buf)));
940 		goto done;
941 	}
942 
943 	err = got_object_parse_header(&obj, *p, len);
944 	if (err)
945 		goto done;
946 
947 	if (len < obj->hdrlen + obj->size) {
948 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
949 		goto done;
950 	}
951 
952 	/* Skip object header. */
953 	len -= obj->hdrlen;
954 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
955 	    *p + obj->hdrlen, len);
956 done:
957 	if (obj)
958 		got_object_close(obj);
959 	return err;
960 }
961 
962 void
963 got_object_tag_close(struct got_tag_object *tag)
964 {
965 	if (tag->refcnt > 0) {
966 		tag->refcnt--;
967 		if (tag->refcnt > 0)
968 			return;
969 	}
970 
971 	free(tag->tag);
972 	free(tag->tagger);
973 	free(tag->tagmsg);
974 	free(tag);
975 }
976 
977 const struct got_error *
978 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
979 {
980 	const struct got_error *err = NULL;
981 	size_t remain = len;
982 	char *s = buf;
983 	size_t label_len;
984 
985 	if (remain == 0)
986 		return got_error(GOT_ERR_BAD_OBJ_DATA);
987 
988 	*tag = calloc(1, sizeof(**tag));
989 	if (*tag == NULL)
990 		return got_error_from_errno("calloc");
991 
992 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
993 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
994 		remain -= label_len;
995 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
996 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
997 			goto done;
998 		}
999 		s += label_len;
1000 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
1001 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1002 			goto done;
1003 		}
1004 		remain -= SHA1_DIGEST_STRING_LENGTH;
1005 		s += SHA1_DIGEST_STRING_LENGTH;
1006 	} else {
1007 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1008 		goto done;
1009 	}
1010 
1011 	if (remain <= 0) {
1012 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1013 		goto done;
1014 	}
1015 
1016 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1017 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1018 		remain -= label_len;
1019 		if (remain <= 0) {
1020 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1021 			goto done;
1022 		}
1023 		s += label_len;
1024 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1025 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1026 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1027 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1028 			s += label_len;
1029 			remain -= label_len;
1030 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1031 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1032 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1033 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1034 			s += label_len;
1035 			remain -= label_len;
1036 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1037 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1038 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1039 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1040 			s += label_len;
1041 			remain -= label_len;
1042 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1043 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1044 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1045 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1046 			s += label_len;
1047 			remain -= label_len;
1048 		} else {
1049 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1050 			goto done;
1051 		}
1052 
1053 		if (remain <= 0 || *s != '\n') {
1054 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1055 			goto done;
1056 		}
1057 		s++;
1058 		remain--;
1059 		if (remain <= 0) {
1060 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1061 			goto done;
1062 		}
1063 	} else {
1064 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1065 		goto done;
1066 	}
1067 
1068 	label_len = strlen(GOT_TAG_LABEL_TAG);
1069 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1070 		char *p;
1071 		size_t slen;
1072 		remain -= label_len;
1073 		if (remain <= 0) {
1074 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1075 			goto done;
1076 		}
1077 		s += label_len;
1078 		p = memchr(s, '\n', remain);
1079 		if (p == NULL) {
1080 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1081 			goto done;
1082 		}
1083 		*p = '\0';
1084 		slen = strlen(s);
1085 		(*tag)->tag = strndup(s, slen);
1086 		if ((*tag)->tag == NULL) {
1087 			err = got_error_from_errno("strndup");
1088 			goto done;
1089 		}
1090 		s += slen + 1;
1091 		remain -= slen + 1;
1092 		if (remain <= 0) {
1093 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1094 			goto done;
1095 		}
1096 	} else {
1097 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1098 		goto done;
1099 	}
1100 
1101 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1102 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1103 		char *p;
1104 		size_t slen;
1105 
1106 		remain -= label_len;
1107 		if (remain <= 0) {
1108 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1109 			goto done;
1110 		}
1111 		s += label_len;
1112 		p = memchr(s, '\n', remain);
1113 		if (p == NULL) {
1114 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1115 			goto done;
1116 		}
1117 		*p = '\0';
1118 		slen = strlen(s);
1119 		err = parse_commit_time(&(*tag)->tagger_time,
1120 		    &(*tag)->tagger_gmtoff, s);
1121 		if (err)
1122 			goto done;
1123 		(*tag)->tagger = strdup(s);
1124 		if ((*tag)->tagger == NULL) {
1125 			err = got_error_from_errno("strdup");
1126 			goto done;
1127 		}
1128 		s += slen + 1;
1129 		remain -= slen + 1;
1130 		if (remain < 0) {
1131 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1132 			goto done;
1133 		}
1134 	} else {
1135 		/* Some old tags in the Linux git repo have no tagger. */
1136 		(*tag)->tagger = strdup("");
1137 		if ((*tag)->tagger == NULL) {
1138 			err = got_error_from_errno("strdup");
1139 			goto done;
1140 		}
1141 	}
1142 
1143 	(*tag)->tagmsg = strndup(s, remain);
1144 	if ((*tag)->tagmsg == NULL) {
1145 		err = got_error_from_errno("strndup");
1146 		goto done;
1147 	}
1148 done:
1149 	if (err) {
1150 		got_object_tag_close(*tag);
1151 		*tag = NULL;
1152 	}
1153 	return err;
1154 }
1155 
1156 const struct got_error *
1157 got_object_read_tag(struct got_tag_object **tag, int fd,
1158     struct got_object_id *expected_id, size_t expected_size)
1159 {
1160 	const struct got_error *err = NULL;
1161 	struct got_object *obj = NULL;
1162 	size_t len;
1163 	uint8_t *p;
1164 	struct got_inflate_checksum csum;
1165 	SHA1_CTX sha1_ctx;
1166 	struct got_object_id id;
1167 
1168 	SHA1Init(&sha1_ctx);
1169 	memset(&csum, 0, sizeof(csum));
1170 	csum.output_sha1 = &sha1_ctx;
1171 
1172 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1173 	    expected_size, fd);
1174 	if (err)
1175 		return err;
1176 
1177 	SHA1Final(id.sha1, &sha1_ctx);
1178 	if (got_object_id_cmp(expected_id, &id) != 0) {
1179 		char buf[SHA1_DIGEST_STRING_LENGTH];
1180 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
1181 		    "checksum failure for object %s",
1182 		    got_sha1_digest_to_str(expected_id->sha1, buf,
1183 		    sizeof(buf)));
1184 		goto done;
1185 	}
1186 
1187 	err = got_object_parse_header(&obj, p, len);
1188 	if (err)
1189 		goto done;
1190 
1191 	if (len < obj->hdrlen + obj->size) {
1192 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1193 		goto done;
1194 	}
1195 
1196 	/* Skip object header. */
1197 	len -= obj->hdrlen;
1198 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1199 done:
1200 	free(p);
1201 	if (obj)
1202 		got_object_close(obj);
1203 	return err;
1204 }
1205 
1206 const struct got_error *
1207 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1208 {
1209 	const struct got_error *err = NULL;
1210 	static const size_t blocksize = 512;
1211 	size_t n, total, remain;
1212 	uint8_t *buf;
1213 
1214 	*outbuf = NULL;
1215 	*outlen = 0;
1216 
1217 	buf = malloc(blocksize);
1218 	if (buf == NULL)
1219 		return got_error_from_errno("malloc");
1220 
1221 	remain = blocksize;
1222 	total = 0;
1223 	for (;;) {
1224 		if (remain == 0) {
1225 			uint8_t *newbuf;
1226 			newbuf = reallocarray(buf, 1, total + blocksize);
1227 			if (newbuf == NULL) {
1228 				err = got_error_from_errno("reallocarray");
1229 				goto done;
1230 			}
1231 			buf = newbuf;
1232 			remain += blocksize;
1233 		}
1234 		n = fread(buf + total, 1, remain, f);
1235 		if (n == 0) {
1236 			if (ferror(f)) {
1237 				err = got_ferror(f, GOT_ERR_IO);
1238 				goto done;
1239 			}
1240 			break; /* EOF */
1241 		}
1242 		remain -= n;
1243 		total += n;
1244 	};
1245 
1246 done:
1247 	if (err == NULL) {
1248 		*outbuf = buf;
1249 		*outlen = total;
1250 	} else
1251 		free(buf);
1252 	return err;
1253 }