op public repos

Blob

Date:: Sat Feb 4 07:41:11 2023 UTC
Message:: introduce got_error_checksum 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/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 = GOT_OBJECT_ID_HEX_MAXLEN;
92 
93 	*outbuf = malloc(len);
94 	if (*outbuf == NULL)
95 		return got_error_from_errno("malloc");
96 
97 	if (got_object_id_hex(id, *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 char *
107 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
108 {
109 	return got_sha1_digest_to_str(id->sha1, buf, len);
110 }
111 
112 void
113 got_object_close(struct got_object *obj)
114 {
115 	if (obj->refcnt > 0) {
116 		obj->refcnt--;
117 		if (obj->refcnt > 0)
118 			return;
119 	}
120 
121 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
122 		struct got_delta *delta;
123 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
124 			delta = STAILQ_FIRST(&obj->deltas.entries);
125 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
126 			free(delta);
127 		}
128 	}
129 	free(obj);
130 }
131 
132 const struct got_error *
133 got_object_raw_close(struct got_raw_object *obj)
134 {
135 	const struct got_error *err = NULL;
136 
137 	if (obj->refcnt > 0) {
138 		obj->refcnt--;
139 		if (obj->refcnt > 0)
140 			return NULL;
141 	}
142 
143 	if (obj->close_cb)
144 		obj->close_cb(obj);
145 
146 	if (obj->f == NULL) {
147 		if (obj->fd != -1) {
148 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
149 				err = got_error_from_errno("munmap");
150 			if (close(obj->fd) == -1 && err == NULL)
151 				err = got_error_from_errno("close");
152 		} else
153 			free(obj->data);
154 	} else {
155 		if (fclose(obj->f) == EOF && err == NULL)
156 			err = got_error_from_errno("fclose");
157 	}
158 	free(obj);
159 	return err;
160 }
161 
162 void
163 got_object_qid_free(struct got_object_qid *qid)
164 {
165 	free(qid);
166 }
167 
168 void
169 got_object_id_queue_free(struct got_object_id_queue *ids)
170 {
171 	struct got_object_qid *qid;
172 
173 	while (!STAILQ_EMPTY(ids)) {
174 		qid = STAILQ_FIRST(ids);
175 		STAILQ_REMOVE_HEAD(ids, entry);
176 		got_object_qid_free(qid);
177 	}
178 }
179 
180 const struct got_error *
181 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
182 {
183 	const char *obj_labels[] = {
184 		GOT_OBJ_LABEL_COMMIT,
185 		GOT_OBJ_LABEL_TREE,
186 		GOT_OBJ_LABEL_BLOB,
187 		GOT_OBJ_LABEL_TAG,
188 	};
189 	const int obj_types[] = {
190 		GOT_OBJ_TYPE_COMMIT,
191 		GOT_OBJ_TYPE_TREE,
192 		GOT_OBJ_TYPE_BLOB,
193 		GOT_OBJ_TYPE_TAG,
194 	};
195 	int type = 0;
196 	size_t size = 0;
197 	size_t i;
198 	char *end;
199 
200 	*obj = NULL;
201 
202 	end = memchr(buf, '\0', len);
203 	if (end == NULL)
204 		return got_error(GOT_ERR_BAD_OBJ_HDR);
205 
206 	for (i = 0; i < nitems(obj_labels); i++) {
207 		const char *label = obj_labels[i];
208 		size_t label_len = strlen(label);
209 		const char *errstr;
210 
211 		if (len <= label_len || buf + label_len >= end ||
212 		    strncmp(buf, label, label_len) != 0)
213 			continue;
214 
215 		type = obj_types[i];
216 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
217 		if (errstr != NULL)
218 			return got_error(GOT_ERR_BAD_OBJ_HDR);
219 		break;
220 	}
221 
222 	if (type == 0)
223 		return got_error(GOT_ERR_BAD_OBJ_HDR);
224 
225 	*obj = calloc(1, sizeof(**obj));
226 	if (*obj == NULL)
227 		return got_error_from_errno("calloc");
228 	(*obj)->type = type;
229 	(*obj)->hdrlen = end - buf + 1;
230 	(*obj)->size = size;
231 	return NULL;
232 }
233 
234 const struct got_error *
235 got_object_read_header(struct got_object **obj, int fd)
236 {
237 	const struct got_error *err;
238 	struct got_inflate_buf zb;
239 	uint8_t *buf;
240 	const size_t zbsize = 64;
241 	size_t outlen, totlen;
242 	int nbuf = 1;
243 
244 	*obj = NULL;
245 
246 	buf = malloc(zbsize);
247 	if (buf == NULL)
248 		return got_error_from_errno("malloc");
249 	buf[0] = '\0';
250 
251 	err = got_inflate_init(&zb, buf, zbsize, NULL);
252 	if (err)
253 		return err;
254 
255 	totlen = 0;
256 	do {
257 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
258 		if (err)
259 			goto done;
260 		if (outlen == 0)
261 			break;
262 		totlen += outlen;
263 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
264 			uint8_t *newbuf;
265 			nbuf++;
266 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
267 			if (newbuf == NULL) {
268 				err = got_error_from_errno("recallocarray");
269 				goto done;
270 			}
271 			buf = newbuf;
272 			zb.outbuf = newbuf + totlen;
273 			zb.outlen = (nbuf * zbsize) - totlen;
274 		}
275 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
276 
277 	err = got_object_parse_header(obj, buf, totlen);
278 done:
279 	free(buf);
280 	got_inflate_end(&zb);
281 	return err;
282 }
283 
284 const struct got_error *
285 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
286     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
287     int infd)
288 {
289 	const struct got_error *err = NULL;
290 	struct got_object *obj;
291 	struct got_inflate_checksum csum;
292 	uint8_t sha1[SHA1_DIGEST_LENGTH];
293 	SHA1_CTX sha1_ctx;
294 	size_t len, consumed;
295 	FILE *f = NULL;
296 
297 	*outbuf = NULL;
298 	*size = 0;
299 	*hdrlen = 0;
300 
301 	SHA1Init(&sha1_ctx);
302 	memset(&csum, 0, sizeof(csum));
303 	csum.output_sha1 = &sha1_ctx;
304 
305 	if (lseek(infd, SEEK_SET, 0) == -1)
306 		return got_error_from_errno("lseek");
307 
308 	err = got_object_read_header(&obj, infd);
309 	if (err)
310 		return err;
311 
312 	if (lseek(infd, SEEK_SET, 0) == -1)
313 		return got_error_from_errno("lseek");
314 
315 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
316 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
317 		    obj->size + obj->hdrlen, infd);
318 	} else {
319 		int fd;
320 		/*
321 		 * XXX This uses an extra file descriptor for no good reason.
322 		 * We should have got_inflate_fd_to_fd().
323 		 */
324 		fd = dup(infd);
325 		if (fd == -1)
326 			return got_error_from_errno("dup");
327 		f = fdopen(fd, "r");
328 		if (f == NULL) {
329 			err = got_error_from_errno("fdopen");
330 			abort();
331 			close(fd);
332 			goto done;
333 		}
334 		err = got_inflate_to_fd(&len, f, &csum, outfd);
335 	}
336 	if (err)
337 		goto done;
338 
339 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
340 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
341 		goto done;
342 	}
343 
344 	SHA1Final(sha1, &sha1_ctx);
345 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
346 		err = got_error_checksum(expected_id);
347 		goto done;
348 	}
349 
350 	*size = obj->size;
351 	*hdrlen = obj->hdrlen;
352 done:
353 	got_object_close(obj);
354 	if (f && fclose(f) == EOF && err == NULL)
355 		err = got_error_from_errno("fclose");
356 	return err;
357 }
358 
359 struct got_commit_object *
360 got_object_commit_alloc_partial(void)
361 {
362 	struct got_commit_object *commit;
363 
364 	commit = calloc(1, sizeof(*commit));
365 	if (commit == NULL)
366 		return NULL;
367 	commit->tree_id = malloc(sizeof(*commit->tree_id));
368 	if (commit->tree_id == NULL) {
369 		free(commit);
370 		return NULL;
371 	}
372 
373 	STAILQ_INIT(&commit->parent_ids);
374 
375 	return commit;
376 }
377 
378 const struct got_error *
379 got_object_commit_add_parent(struct got_commit_object *commit,
380     const char *id_str)
381 {
382 	const struct got_error *err = NULL;
383 	struct got_object_qid *qid;
384 
385 	err = got_object_qid_alloc_partial(&qid);
386 	if (err)
387 		return err;
388 
389 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
390 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
391 		got_object_qid_free(qid);
392 		return err;
393 	}
394 
395 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
396 	commit->nparents++;
397 
398 	return NULL;
399 }
400 
401 static const struct got_error *
402 parse_gmtoff(time_t *gmtoff, const char *tzstr)
403 {
404 	int sign = 1;
405 	const char *p = tzstr;
406 	time_t h, m;
407 
408 	*gmtoff = 0;
409 
410 	if (*p == '-')
411 		sign = -1;
412 	else if (*p != '+')
413 		return got_error(GOT_ERR_BAD_OBJ_DATA);
414 	p++;
415 	if (!isdigit((unsigned char)*p) &&
416 	    !isdigit((unsigned char)*(p + 1)))
417 		return got_error(GOT_ERR_BAD_OBJ_DATA);
418 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
419 
420 	p += 2;
421 	if (!isdigit((unsigned char)*p) &&
422 	    !isdigit((unsigned char)*(p + 1)))
423 		return got_error(GOT_ERR_BAD_OBJ_DATA);
424 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
425 
426 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
427 	return NULL;
428 }
429 
430 static const struct got_error *
431 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
432 {
433 	const struct got_error *err = NULL;
434 	const char *errstr;
435 	char *space, *tzstr;
436 
437 	/* Parse and strip off trailing timezone indicator string. */
438 	space = strrchr(committer, ' ');
439 	if (space == NULL)
440 		return got_error(GOT_ERR_BAD_OBJ_DATA);
441 	tzstr = strdup(space + 1);
442 	if (tzstr == NULL)
443 		return got_error_from_errno("strdup");
444 	err = parse_gmtoff(gmtoff, tzstr);
445 	free(tzstr);
446 	if (err) {
447 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
448 			return err;
449 		/* Old versions of Git omitted the timestamp. */
450 		*time = 0;
451 		*gmtoff = 0;
452 		return NULL;
453 	}
454 	*space = '\0';
455 
456 	/* Timestamp is separated from committer name + email by space. */
457 	space = strrchr(committer, ' ');
458 	if (space == NULL)
459 		return got_error(GOT_ERR_BAD_OBJ_DATA);
460 
461 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
462 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
463 	if (errstr)
464 		return got_error(GOT_ERR_BAD_OBJ_DATA);
465 
466 	/* Strip off parsed time information, leaving just author and email. */
467 	*space = '\0';
468 
469 	return NULL;
470 }
471 
472 void
473 got_object_commit_close(struct got_commit_object *commit)
474 {
475 	if (commit->refcnt > 0) {
476 		commit->refcnt--;
477 		if (commit->refcnt > 0)
478 			return;
479 	}
480 
481 	got_object_id_queue_free(&commit->parent_ids);
482 	free(commit->tree_id);
483 	free(commit->author);
484 	free(commit->committer);
485 	free(commit->logmsg);
486 	free(commit);
487 }
488 
489 struct got_object_id *
490 got_object_commit_get_tree_id(struct got_commit_object *commit)
491 {
492 	return commit->tree_id;
493 }
494 
495 int
496 got_object_commit_get_nparents(struct got_commit_object *commit)
497 {
498 	return commit->nparents;
499 }
500 
501 const struct got_object_id_queue *
502 got_object_commit_get_parent_ids(struct got_commit_object *commit)
503 {
504 	return &commit->parent_ids;
505 }
506 
507 const char *
508 got_object_commit_get_author(struct got_commit_object *commit)
509 {
510 	return commit->author;
511 }
512 
513 time_t
514 got_object_commit_get_author_time(struct got_commit_object *commit)
515 {
516 	return commit->author_time;
517 }
518 
519 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
520 {
521 	return commit->author_gmtoff;
522 }
523 
524 const char *
525 got_object_commit_get_committer(struct got_commit_object *commit)
526 {
527 	return commit->committer;
528 }
529 
530 time_t
531 got_object_commit_get_committer_time(struct got_commit_object *commit)
532 {
533 	return commit->committer_time;
534 }
535 
536 time_t
537 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
538 {
539 	return commit->committer_gmtoff;
540 }
541 
542 const struct got_error *
543 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
544 {
545 	const struct got_error *err = NULL;
546 	const char *src;
547 	char *dst;
548 	size_t len;
549 
550 	len = strlen(commit->logmsg);
551 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
552 	if (*logmsg == NULL)
553 		return got_error_from_errno("malloc");
554 
555 	/*
556 	 * Strip out unusual headers. Headers are separated from the commit
557 	 * message body by a single empty line.
558 	 */
559 	src = commit->logmsg;
560 	dst = *logmsg;
561 	while (*src != '\0' && *src != '\n') {
562 		int copy_header = 1, eol = 0;
563 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
564 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
565 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
566 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
567 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
568 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
569 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
570 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
571 			copy_header = 0;
572 
573 		while (*src != '\0' && !eol) {
574 			if (copy_header) {
575 				*dst = *src;
576 				dst++;
577 			}
578 			if (*src == '\n')
579 				eol = 1;
580 			src++;
581 		}
582 	}
583 	*dst = '\0';
584 
585 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
586 		err = got_error(GOT_ERR_NO_SPACE);
587 		goto done;
588 	}
589 
590 	/* Trim redundant trailing whitespace. */
591 	len = strlen(*logmsg);
592 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
593 	    isspace((unsigned char)(*logmsg)[len - 1])) {
594 		(*logmsg)[len - 1] = '\0';
595 		len--;
596 	}
597 
598 	/* Append a trailing newline if missing. */
599 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
600 		(*logmsg)[len] = '\n';
601 		(*logmsg)[len + 1] = '\0';
602 	}
603 done:
604 	if (err) {
605 		free(*logmsg);
606 		*logmsg = NULL;
607 	}
608 	return err;
609 }
610 
611 const char *
612 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
613 {
614 	return commit->logmsg;
615 }
616 
617 const struct got_error *
618 got_object_parse_commit(struct got_commit_object **commit, char *buf,
619     size_t len)
620 {
621 	const struct got_error *err = NULL;
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_sha1_digest((*commit)->tree_id->sha1, s)) {
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 	SHA1_CTX sha1_ctx;
753 	struct got_object_id id;
754 
755 	SHA1Init(&sha1_ctx);
756 	memset(&csum, 0, sizeof(csum));
757 	csum.output_sha1 = &sha1_ctx;
758 
759 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
760 	if (err)
761 		return err;
762 
763 	SHA1Final(id.sha1, &sha1_ctx);
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 	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 (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 	size_t remain = len;
977 	char *s = buf;
978 	size_t label_len;
979 
980 	if (remain == 0)
981 		return got_error(GOT_ERR_BAD_OBJ_DATA);
982 
983 	*tag = calloc(1, sizeof(**tag));
984 	if (*tag == NULL)
985 		return got_error_from_errno("calloc");
986 
987 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
988 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
989 		remain -= label_len;
990 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
991 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
992 			goto done;
993 		}
994 		s += label_len;
995 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
996 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
997 			goto done;
998 		}
999 		remain -= SHA1_DIGEST_STRING_LENGTH;
1000 		s += SHA1_DIGEST_STRING_LENGTH;
1001 	} else {
1002 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1003 		goto done;
1004 	}
1005 
1006 	if (remain <= 0) {
1007 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1008 		goto done;
1009 	}
1010 
1011 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1012 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1013 		remain -= label_len;
1014 		if (remain <= 0) {
1015 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1016 			goto done;
1017 		}
1018 		s += label_len;
1019 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1020 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1021 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1022 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1023 			s += label_len;
1024 			remain -= label_len;
1025 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1026 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1027 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1028 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1029 			s += label_len;
1030 			remain -= label_len;
1031 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1032 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1033 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1034 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1035 			s += label_len;
1036 			remain -= label_len;
1037 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1038 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1039 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1040 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1041 			s += label_len;
1042 			remain -= label_len;
1043 		} else {
1044 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1045 			goto done;
1046 		}
1047 
1048 		if (remain <= 0 || *s != '\n') {
1049 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1050 			goto done;
1051 		}
1052 		s++;
1053 		remain--;
1054 		if (remain <= 0) {
1055 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1056 			goto done;
1057 		}
1058 	} else {
1059 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1060 		goto done;
1061 	}
1062 
1063 	label_len = strlen(GOT_TAG_LABEL_TAG);
1064 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1065 		char *p;
1066 		size_t slen;
1067 		remain -= label_len;
1068 		if (remain <= 0) {
1069 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1070 			goto done;
1071 		}
1072 		s += label_len;
1073 		p = memchr(s, '\n', remain);
1074 		if (p == NULL) {
1075 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1076 			goto done;
1077 		}
1078 		*p = '\0';
1079 		slen = strlen(s);
1080 		(*tag)->tag = strndup(s, slen);
1081 		if ((*tag)->tag == NULL) {
1082 			err = got_error_from_errno("strndup");
1083 			goto done;
1084 		}
1085 		s += slen + 1;
1086 		remain -= slen + 1;
1087 		if (remain <= 0) {
1088 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1089 			goto done;
1090 		}
1091 	} else {
1092 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1093 		goto done;
1094 	}
1095 
1096 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1097 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1098 		char *p;
1099 		size_t slen;
1100 
1101 		remain -= label_len;
1102 		if (remain <= 0) {
1103 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1104 			goto done;
1105 		}
1106 		s += label_len;
1107 		p = memchr(s, '\n', remain);
1108 		if (p == NULL) {
1109 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1110 			goto done;
1111 		}
1112 		*p = '\0';
1113 		slen = strlen(s);
1114 		err = parse_commit_time(&(*tag)->tagger_time,
1115 		    &(*tag)->tagger_gmtoff, s);
1116 		if (err)
1117 			goto done;
1118 		(*tag)->tagger = strdup(s);
1119 		if ((*tag)->tagger == NULL) {
1120 			err = got_error_from_errno("strdup");
1121 			goto done;
1122 		}
1123 		s += slen + 1;
1124 		remain -= slen + 1;
1125 		if (remain < 0) {
1126 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1127 			goto done;
1128 		}
1129 	} else {
1130 		/* Some old tags in the Linux git repo have no tagger. */
1131 		(*tag)->tagger = strdup("");
1132 		if ((*tag)->tagger == NULL) {
1133 			err = got_error_from_errno("strdup");
1134 			goto done;
1135 		}
1136 	}
1137 
1138 	(*tag)->tagmsg = strndup(s, remain);
1139 	if ((*tag)->tagmsg == NULL) {
1140 		err = got_error_from_errno("strndup");
1141 		goto done;
1142 	}
1143 done:
1144 	if (err) {
1145 		got_object_tag_close(*tag);
1146 		*tag = NULL;
1147 	}
1148 	return err;
1149 }
1150 
1151 const struct got_error *
1152 got_object_read_tag(struct got_tag_object **tag, int fd,
1153     struct got_object_id *expected_id, size_t expected_size)
1154 {
1155 	const struct got_error *err = NULL;
1156 	struct got_object *obj = NULL;
1157 	size_t len;
1158 	uint8_t *p;
1159 	struct got_inflate_checksum csum;
1160 	SHA1_CTX sha1_ctx;
1161 	struct got_object_id id;
1162 
1163 	SHA1Init(&sha1_ctx);
1164 	memset(&csum, 0, sizeof(csum));
1165 	csum.output_sha1 = &sha1_ctx;
1166 
1167 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1168 	    expected_size, fd);
1169 	if (err)
1170 		return err;
1171 
1172 	SHA1Final(id.sha1, &sha1_ctx);
1173 	if (got_object_id_cmp(expected_id, &id) != 0) {
1174 		err = got_error_checksum(expected_id);
1175 		goto done;
1176 	}
1177 
1178 	err = got_object_parse_header(&obj, p, len);
1179 	if (err)
1180 		goto done;
1181 
1182 	if (len < obj->hdrlen + obj->size) {
1183 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1184 		goto done;
1185 	}
1186 
1187 	/* Skip object header. */
1188 	len -= obj->hdrlen;
1189 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1190 done:
1191 	free(p);
1192 	if (obj)
1193 		got_object_close(obj);
1194 	return err;
1195 }
1196 
1197 const struct got_error *
1198 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1199 {
1200 	const struct got_error *err = NULL;
1201 	static const size_t blocksize = 512;
1202 	size_t n, total, remain;
1203 	uint8_t *buf;
1204 
1205 	*outbuf = NULL;
1206 	*outlen = 0;
1207 
1208 	buf = malloc(blocksize);
1209 	if (buf == NULL)
1210 		return got_error_from_errno("malloc");
1211 
1212 	remain = blocksize;
1213 	total = 0;
1214 	for (;;) {
1215 		if (remain == 0) {
1216 			uint8_t *newbuf;
1217 			newbuf = reallocarray(buf, 1, total + blocksize);
1218 			if (newbuf == NULL) {
1219 				err = got_error_from_errno("reallocarray");
1220 				goto done;
1221 			}
1222 			buf = newbuf;
1223 			remain += blocksize;
1224 		}
1225 		n = fread(buf + total, 1, remain, f);
1226 		if (n == 0) {
1227 			if (ferror(f)) {
1228 				err = got_ferror(f, GOT_ERR_IO);
1229 				goto done;
1230 			}
1231 			break; /* EOF */
1232 		}
1233 		remain -= n;
1234 		total += n;
1235 	};
1236 
1237 done:
1238 	if (err == NULL) {
1239 		*outbuf = buf;
1240 		*outlen = total;
1241 	} else
1242 		free(buf);
1243 	return err;
1244 }