op public repos

Blob

Date:: Sun Feb 12 14:10:01 2023 UTC
Message:: fix fallout; mostly struct got_object_id zeroing
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 <sha2.h>
33 #include <zlib.h>
34 #include <ctype.h>
35 #include <limits.h>
36 #include <imsg.h>
37 #include <time.h>
38 #include <unistd.h>
39 
40 #include "got_error.h"
41 #include "got_object.h"
42 #include "got_repository.h"
43 #include "got_opentemp.h"
44 #include "got_path.h"
45 
46 #include "got_lib_hash.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_inflate.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_parse.h"
51 #include "got_lib_object_cache.h"
52 #include "got_lib_pack.h"
53 #include "got_lib_repository.h"
54 
55 #ifndef nitems
56 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
57 #endif
58 
59 struct got_object_id *
60 got_object_id_dup(struct got_object_id *id1)
61 {
62 	struct got_object_id *id2;
63 
64 	id2 = malloc(sizeof(*id2));
65 	if (id2 == NULL)
66 		return NULL;
67 	memcpy(id2, id1, sizeof(*id2));
68 	return id2;
69 }
70 
71 int
72 got_object_id_cmp(const struct got_object_id *id1,
73     const struct got_object_id *id2)
74 {
75 	if (id1->algo != id2->algo)
76 		abort(); // return -1;
77 	if (id1->algo == GOT_HASH_SHA1)
78 		return memcmp(id1->hash, id2->hash, SHA1_DIGEST_LENGTH);
79 	if (id1->algo == GOT_HASH_SHA256)
80 		return memcmp(id1->hash, id2->hash, SHA256_DIGEST_LENGTH);
81 	abort();
82 	return -1;
83 }
84 
85 const struct got_error *
86 got_object_qid_alloc_partial(struct got_object_qid **qid)
87 {
88 	*qid = calloc(1, sizeof(**qid));
89 	if (*qid == NULL)
90 		return got_error_from_errno("malloc");
91 
92 	(*qid)->data = NULL;
93 	return NULL;
94 }
95 
96 const struct got_error *
97 got_object_id_str(char **outbuf, struct got_object_id *id)
98 {
99 	static const size_t len = GOT_OBJECT_ID_HEX_MAXLEN;
100 
101 	*outbuf = malloc(len);
102 	if (*outbuf == NULL)
103 		return got_error_from_errno("malloc");
104 
105 	if (got_object_id_hex(id, *outbuf, len) == NULL) {
106 		free(*outbuf);
107 		*outbuf = NULL;
108 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
109 	}
110 
111 	return NULL;
112 }
113 
114 char *
115 got_object_id_hex(struct got_object_id *id, char *buf, size_t len)
116 {
117 	return got_hash_digest_to_str(id->hash, buf, len, id->algo);
118 }
119 
120 void
121 got_object_close(struct got_object *obj)
122 {
123 	if (obj->refcnt > 0) {
124 		obj->refcnt--;
125 		if (obj->refcnt > 0)
126 			return;
127 	}
128 
129 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
130 		struct got_delta *delta;
131 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
132 			delta = STAILQ_FIRST(&obj->deltas.entries);
133 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
134 			free(delta);
135 		}
136 	}
137 	free(obj);
138 }
139 
140 const struct got_error *
141 got_object_raw_close(struct got_raw_object *obj)
142 {
143 	const struct got_error *err = NULL;
144 
145 	if (obj->refcnt > 0) {
146 		obj->refcnt--;
147 		if (obj->refcnt > 0)
148 			return NULL;
149 	}
150 
151 	if (obj->close_cb)
152 		obj->close_cb(obj);
153 
154 	if (obj->f == NULL) {
155 		if (obj->fd != -1) {
156 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
157 				err = got_error_from_errno("munmap");
158 			if (close(obj->fd) == -1 && err == NULL)
159 				err = got_error_from_errno("close");
160 		} else
161 			free(obj->data);
162 	} else {
163 		if (fclose(obj->f) == EOF && err == NULL)
164 			err = got_error_from_errno("fclose");
165 	}
166 	free(obj);
167 	return err;
168 }
169 
170 void
171 got_object_qid_free(struct got_object_qid *qid)
172 {
173 	free(qid);
174 }
175 
176 void
177 got_object_id_queue_free(struct got_object_id_queue *ids)
178 {
179 	struct got_object_qid *qid;
180 
181 	while (!STAILQ_EMPTY(ids)) {
182 		qid = STAILQ_FIRST(ids);
183 		STAILQ_REMOVE_HEAD(ids, entry);
184 		got_object_qid_free(qid);
185 	}
186 }
187 
188 const struct got_error *
189 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
190 {
191 	const char *obj_labels[] = {
192 		GOT_OBJ_LABEL_COMMIT,
193 		GOT_OBJ_LABEL_TREE,
194 		GOT_OBJ_LABEL_BLOB,
195 		GOT_OBJ_LABEL_TAG,
196 	};
197 	const int obj_types[] = {
198 		GOT_OBJ_TYPE_COMMIT,
199 		GOT_OBJ_TYPE_TREE,
200 		GOT_OBJ_TYPE_BLOB,
201 		GOT_OBJ_TYPE_TAG,
202 	};
203 	int type = 0;
204 	size_t size = 0;
205 	size_t i;
206 	char *end;
207 
208 	*obj = NULL;
209 
210 	end = memchr(buf, '\0', len);
211 	if (end == NULL)
212 		return got_error(GOT_ERR_BAD_OBJ_HDR);
213 
214 	for (i = 0; i < nitems(obj_labels); i++) {
215 		const char *label = obj_labels[i];
216 		size_t label_len = strlen(label);
217 		const char *errstr;
218 
219 		if (len <= label_len || buf + label_len >= end ||
220 		    strncmp(buf, label, label_len) != 0)
221 			continue;
222 
223 		type = obj_types[i];
224 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
225 		if (errstr != NULL)
226 			return got_error(GOT_ERR_BAD_OBJ_HDR);
227 		break;
228 	}
229 
230 	if (type == 0)
231 		return got_error(GOT_ERR_BAD_OBJ_HDR);
232 
233 	*obj = calloc(1, sizeof(**obj));
234 	if (*obj == NULL)
235 		return got_error_from_errno("calloc");
236 	(*obj)->type = type;
237 	(*obj)->hdrlen = end - buf + 1;
238 	(*obj)->size = size;
239 	return NULL;
240 }
241 
242 const struct got_error *
243 got_object_read_header(struct got_object **obj, int fd)
244 {
245 	const struct got_error *err;
246 	struct got_inflate_buf zb;
247 	uint8_t *buf;
248 	const size_t zbsize = 64;
249 	size_t outlen, totlen;
250 	int nbuf = 1;
251 
252 	*obj = NULL;
253 
254 	buf = malloc(zbsize);
255 	if (buf == NULL)
256 		return got_error_from_errno("malloc");
257 	buf[0] = '\0';
258 
259 	err = got_inflate_init(&zb, buf, zbsize, NULL);
260 	if (err)
261 		return err;
262 
263 	totlen = 0;
264 	do {
265 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
266 		if (err)
267 			goto done;
268 		if (outlen == 0)
269 			break;
270 		totlen += outlen;
271 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
272 			uint8_t *newbuf;
273 			nbuf++;
274 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
275 			if (newbuf == NULL) {
276 				err = got_error_from_errno("recallocarray");
277 				goto done;
278 			}
279 			buf = newbuf;
280 			zb.outbuf = newbuf + totlen;
281 			zb.outlen = (nbuf * zbsize) - totlen;
282 		}
283 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
284 
285 	err = got_object_parse_header(obj, buf, totlen);
286 done:
287 	free(buf);
288 	got_inflate_end(&zb);
289 	return err;
290 }
291 
292 const struct got_error *
293 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
294     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
295     int infd)
296 {
297 	const struct got_error *err = NULL;
298 	struct got_object *obj;
299 	struct got_inflate_checksum csum;
300 	uint8_t hash[GOT_OBJECT_ID_MAXLEN];
301 	struct got_hash ctx;
302 	size_t len, consumed;
303 	FILE *f = NULL;
304 
305 	*outbuf = NULL;
306 	*size = 0;
307 	*hdrlen = 0;
308 
309 	got_hash_init(&ctx, expected_id->algo);
310 	memset(&csum, 0, sizeof(csum));
311 	csum.output_ctx = &ctx;
312 
313 	if (lseek(infd, SEEK_SET, 0) == -1)
314 		return got_error_from_errno("lseek");
315 
316 	err = got_object_read_header(&obj, infd);
317 	if (err)
318 		return err;
319 
320 	if (lseek(infd, SEEK_SET, 0) == -1)
321 		return got_error_from_errno("lseek");
322 
323 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
324 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
325 		    obj->size + obj->hdrlen, infd);
326 	} else {
327 		int fd;
328 		/*
329 		 * XXX This uses an extra file descriptor for no good reason.
330 		 * We should have got_inflate_fd_to_fd().
331 		 */
332 		fd = dup(infd);
333 		if (fd == -1)
334 			return got_error_from_errno("dup");
335 		f = fdopen(fd, "r");
336 		if (f == NULL) {
337 			err = got_error_from_errno("fdopen");
338 			abort();
339 			close(fd);
340 			goto done;
341 		}
342 		err = got_inflate_to_fd(&len, f, &csum, outfd);
343 	}
344 	if (err)
345 		goto done;
346 
347 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
348 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
349 		goto done;
350 	}
351 
352 	got_hash_final(&ctx, hash);
353 	if (got_hash_cmp(&ctx, expected_id->hash, hash) != 0) {
354 		err = got_error_checksum(expected_id);
355 		goto done;
356 	}
357 
358 	*size = obj->size;
359 	*hdrlen = obj->hdrlen;
360 done:
361 	got_object_close(obj);
362 	if (f && fclose(f) == EOF && err == NULL)
363 		err = got_error_from_errno("fclose");
364 	return err;
365 }
366 
367 struct got_commit_object *
368 got_object_commit_alloc_partial(void)
369 {
370 	struct got_commit_object *commit;
371 
372 	commit = calloc(1, sizeof(*commit));
373 	if (commit == NULL)
374 		return NULL;
375 	commit->tree_id = malloc(sizeof(*commit->tree_id));
376 	if (commit->tree_id == NULL) {
377 		free(commit);
378 		return NULL;
379 	}
380 
381 	STAILQ_INIT(&commit->parent_ids);
382 
383 	return commit;
384 }
385 
386 const struct got_error *
387 got_object_commit_add_parent(struct got_commit_object *commit,
388     const char *id_str, enum got_hash_algorithm algo)
389 {
390 	const struct got_error *err = NULL;
391 	struct got_object_qid *qid;
392 
393 	err = got_object_qid_alloc_partial(&qid);
394 	if (err)
395 		return err;
396 
397 	qid->id.algo = algo;
398 	if (!got_parse_hash_digest(qid->id.hash, id_str, algo)) {
399 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
400 		got_object_qid_free(qid);
401 		return err;
402 	}
403 
404 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
405 	commit->nparents++;
406 
407 	return NULL;
408 }
409 
410 static const struct got_error *
411 parse_gmtoff(time_t *gmtoff, const char *tzstr)
412 {
413 	int sign = 1;
414 	const char *p = tzstr;
415 	time_t h, m;
416 
417 	*gmtoff = 0;
418 
419 	if (*p == '-')
420 		sign = -1;
421 	else if (*p != '+')
422 		return got_error(GOT_ERR_BAD_OBJ_DATA);
423 	p++;
424 	if (!isdigit((unsigned char)*p) &&
425 	    !isdigit((unsigned char)*(p + 1)))
426 		return got_error(GOT_ERR_BAD_OBJ_DATA);
427 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
428 
429 	p += 2;
430 	if (!isdigit((unsigned char)*p) &&
431 	    !isdigit((unsigned char)*(p + 1)))
432 		return got_error(GOT_ERR_BAD_OBJ_DATA);
433 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
434 
435 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
436 	return NULL;
437 }
438 
439 static const struct got_error *
440 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
441 {
442 	const struct got_error *err = NULL;
443 	const char *errstr;
444 	char *space, *tzstr;
445 
446 	/* Parse and strip off trailing timezone indicator string. */
447 	space = strrchr(committer, ' ');
448 	if (space == NULL)
449 		return got_error(GOT_ERR_BAD_OBJ_DATA);
450 	tzstr = strdup(space + 1);
451 	if (tzstr == NULL)
452 		return got_error_from_errno("strdup");
453 	err = parse_gmtoff(gmtoff, tzstr);
454 	free(tzstr);
455 	if (err) {
456 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
457 			return err;
458 		/* Old versions of Git omitted the timestamp. */
459 		*time = 0;
460 		*gmtoff = 0;
461 		return NULL;
462 	}
463 	*space = '\0';
464 
465 	/* Timestamp is separated from committer name + email by space. */
466 	space = strrchr(committer, ' ');
467 	if (space == NULL)
468 		return got_error(GOT_ERR_BAD_OBJ_DATA);
469 
470 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
471 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
472 	if (errstr)
473 		return got_error(GOT_ERR_BAD_OBJ_DATA);
474 
475 	/* Strip off parsed time information, leaving just author and email. */
476 	*space = '\0';
477 
478 	return NULL;
479 }
480 
481 void
482 got_object_commit_close(struct got_commit_object *commit)
483 {
484 	if (commit->refcnt > 0) {
485 		commit->refcnt--;
486 		if (commit->refcnt > 0)
487 			return;
488 	}
489 
490 	got_object_id_queue_free(&commit->parent_ids);
491 	free(commit->tree_id);
492 	free(commit->author);
493 	free(commit->committer);
494 	free(commit->logmsg);
495 	free(commit);
496 }
497 
498 struct got_object_id *
499 got_object_commit_get_tree_id(struct got_commit_object *commit)
500 {
501 	return commit->tree_id;
502 }
503 
504 int
505 got_object_commit_get_nparents(struct got_commit_object *commit)
506 {
507 	return commit->nparents;
508 }
509 
510 const struct got_object_id_queue *
511 got_object_commit_get_parent_ids(struct got_commit_object *commit)
512 {
513 	return &commit->parent_ids;
514 }
515 
516 const char *
517 got_object_commit_get_author(struct got_commit_object *commit)
518 {
519 	return commit->author;
520 }
521 
522 time_t
523 got_object_commit_get_author_time(struct got_commit_object *commit)
524 {
525 	return commit->author_time;
526 }
527 
528 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
529 {
530 	return commit->author_gmtoff;
531 }
532 
533 const char *
534 got_object_commit_get_committer(struct got_commit_object *commit)
535 {
536 	return commit->committer;
537 }
538 
539 time_t
540 got_object_commit_get_committer_time(struct got_commit_object *commit)
541 {
542 	return commit->committer_time;
543 }
544 
545 time_t
546 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
547 {
548 	return commit->committer_gmtoff;
549 }
550 
551 const struct got_error *
552 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
553 {
554 	const struct got_error *err = NULL;
555 	const char *src;
556 	char *dst;
557 	size_t len;
558 
559 	len = strlen(commit->logmsg);
560 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
561 	if (*logmsg == NULL)
562 		return got_error_from_errno("malloc");
563 
564 	/*
565 	 * Strip out unusual headers. Headers are separated from the commit
566 	 * message body by a single empty line.
567 	 */
568 	src = commit->logmsg;
569 	dst = *logmsg;
570 	while (*src != '\0' && *src != '\n') {
571 		int copy_header = 1, eol = 0;
572 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
573 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
574 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
575 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
576 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
577 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
578 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
579 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
580 			copy_header = 0;
581 
582 		while (*src != '\0' && !eol) {
583 			if (copy_header) {
584 				*dst = *src;
585 				dst++;
586 			}
587 			if (*src == '\n')
588 				eol = 1;
589 			src++;
590 		}
591 	}
592 	*dst = '\0';
593 
594 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
595 		err = got_error(GOT_ERR_NO_SPACE);
596 		goto done;
597 	}
598 
599 	/* Trim redundant trailing whitespace. */
600 	len = strlen(*logmsg);
601 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
602 	    isspace((unsigned char)(*logmsg)[len - 1])) {
603 		(*logmsg)[len - 1] = '\0';
604 		len--;
605 	}
606 
607 	/* Append a trailing newline if missing. */
608 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
609 		(*logmsg)[len] = '\n';
610 		(*logmsg)[len + 1] = '\0';
611 	}
612 done:
613 	if (err) {
614 		free(*logmsg);
615 		*logmsg = NULL;
616 	}
617 	return err;
618 }
619 
620 const char *
621 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
622 {
623 	return commit->logmsg;
624 }
625 
626 const struct got_error *
627 got_object_parse_commit(struct got_commit_object **commit, char *buf,
628     size_t len, enum got_hash_algorithm algo)
629 {
630 	const struct got_error *err = NULL;
631 	char *s = buf;
632 	size_t label_len, idlen;
633 	ssize_t remain = (ssize_t)len;
634 
635 	idlen = got_hash_digest_string_length(algo);
636 
637 	if (remain == 0)
638 		return got_error(GOT_ERR_BAD_OBJ_DATA);
639 
640 	*commit = got_object_commit_alloc_partial();
641 	if (*commit == NULL)
642 		return got_error_from_errno("got_object_commit_alloc_partial");
643 
644 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
645 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
646 		remain -= label_len;
647 		if (remain < idlen) {
648 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
649 			goto done;
650 		}
651 		s += label_len;
652 		if (!got_parse_hash_digest((*commit)->tree_id->hash, s, algo)) {
653 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
654 			goto done;
655 		}
656 		(*commit)->tree_id->algo = algo;
657 		remain -= idlen;
658 		s += idlen;
659 	} else {
660 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
661 		goto done;
662 	}
663 
664 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
665 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
666 		remain -= label_len;
667 		if (remain < idlen) {
668 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
669 			goto done;
670 		}
671 		s += label_len;
672 		err = got_object_commit_add_parent(*commit, s, algo);
673 		if (err)
674 			goto done;
675 
676 		remain -= idlen;
677 		s += idlen;
678 	}
679 
680 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
681 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
682 		char *p;
683 		size_t slen;
684 
685 		remain -= label_len;
686 		if (remain <= 0) {
687 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
688 			goto done;
689 		}
690 		s += label_len;
691 		p = memchr(s, '\n', remain);
692 		if (p == NULL) {
693 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
694 			goto done;
695 		}
696 		*p = '\0';
697 		slen = strlen(s);
698 		err = parse_commit_time(&(*commit)->author_time,
699 		    &(*commit)->author_gmtoff, s);
700 		if (err)
701 			goto done;
702 		(*commit)->author = strdup(s);
703 		if ((*commit)->author == NULL) {
704 			err = got_error_from_errno("strdup");
705 			goto done;
706 		}
707 		s += slen + 1;
708 		remain -= slen + 1;
709 	}
710 
711 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
712 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
713 		char *p;
714 		size_t slen;
715 
716 		remain -= label_len;
717 		if (remain <= 0) {
718 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
719 			goto done;
720 		}
721 		s += label_len;
722 		p = memchr(s, '\n', remain);
723 		if (p == NULL) {
724 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
725 			goto done;
726 		}
727 		*p = '\0';
728 		slen = strlen(s);
729 		err = parse_commit_time(&(*commit)->committer_time,
730 		    &(*commit)->committer_gmtoff, s);
731 		if (err)
732 			goto done;
733 		(*commit)->committer = strdup(s);
734 		if ((*commit)->committer == NULL) {
735 			err = got_error_from_errno("strdup");
736 			goto done;
737 		}
738 		s += slen + 1;
739 		remain -= slen + 1;
740 	}
741 
742 	(*commit)->logmsg = strndup(s, remain);
743 	if ((*commit)->logmsg == NULL) {
744 		err = got_error_from_errno("strndup");
745 		goto done;
746 	}
747 done:
748 	if (err) {
749 		got_object_commit_close(*commit);
750 		*commit = NULL;
751 	}
752 	return err;
753 }
754 
755 const struct got_error *
756 got_object_read_commit(struct got_commit_object **commit, int fd,
757     struct got_object_id *expected_id, size_t expected_size)
758 {
759 	struct got_object *obj = NULL;
760 	const struct got_error *err = NULL;
761 	size_t len;
762 	uint8_t *p;
763 	struct got_inflate_checksum csum;
764 	struct got_hash ctx;
765 	struct got_object_id id;
766 
767 	memset(&id, 0, sizeof(id));
768 	id.algo = expected_id->algo;
769 
770 	got_hash_init(&ctx, expected_id->algo);
771 	memset(&csum, 0, sizeof(csum));
772 	csum.output_ctx = &ctx;
773 
774 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
775 	if (err)
776 		return err;
777 
778 	got_hash_final(&ctx, id.hash);
779 	if (got_object_id_cmp(expected_id, &id) != 0) {
780 		err = got_error_checksum(expected_id);
781 		goto done;
782 	}
783 
784 	err = got_object_parse_header(&obj, p, len);
785 	if (err)
786 		goto done;
787 
788 	if (len < obj->hdrlen + obj->size) {
789 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
790 		goto done;
791 	}
792 
793 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
794 		err = got_error(GOT_ERR_OBJ_TYPE);
795 		goto done;
796 	}
797 
798 	/* Skip object header. */
799 	len -= obj->hdrlen;
800 	err = got_object_parse_commit(commit, p + obj->hdrlen, len,
801 	    expected_id->algo);
802 done:
803 	free(p);
804 	if (obj)
805 		got_object_close(obj);
806 	return err;
807 }
808 
809 void
810 got_object_tree_close(struct got_tree_object *tree)
811 {
812 	if (tree->refcnt > 0) {
813 		tree->refcnt--;
814 		if (tree->refcnt > 0)
815 			return;
816 	}
817 
818 	free(tree->entries);
819 	free(tree);
820 }
821 
822 static const struct got_error *
823 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
824     size_t maxlen, enum got_hash_algorithm algo, size_t idlen)
825 {
826 	char *p, *space;
827 
828 	*elen = 0;
829 
830 	*elen = strnlen(buf, maxlen) + 1;
831 	if (*elen > maxlen)
832 		return got_error(GOT_ERR_BAD_OBJ_DATA);
833 
834 	space = memchr(buf, ' ', *elen);
835 	if (space == NULL || space <= buf)
836 		return got_error(GOT_ERR_BAD_OBJ_DATA);
837 
838 	pte->mode = 0;
839 	p = buf;
840 	while (p < space) {
841 		if (*p < '0' || *p > '7')
842 			return got_error(GOT_ERR_BAD_OBJ_DATA);
843 		pte->mode <<= 3;
844 		pte->mode |= *p - '0';
845 		p++;
846 	}
847 
848 	if (*elen > maxlen || maxlen - *elen < idlen)
849 		return got_error(GOT_ERR_BAD_OBJ_DATA);
850 
851 	pte->name = space + 1;
852 	pte->namelen = strlen(pte->name);
853 	buf += *elen;
854 	pte->id = buf;
855 	pte->idlen = idlen;
856 	pte->algo = algo;
857 	*elen += idlen;
858 	return NULL;
859 }
860 
861 static int
862 pte_cmp(const void *pa, const void *pb)
863 {
864 	const struct got_parsed_tree_entry *a = pa, *b = pb;
865 
866 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
867 }
868 
869 const struct got_error *
870 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
871     size_t *nentries_alloc, uint8_t *buf, size_t len,
872     enum got_hash_algorithm algo)
873 {
874 	const struct got_error *err = NULL;
875 	size_t idlen, remain = len;
876 	const size_t nalloc = 16;
877 	struct got_parsed_tree_entry *pte;
878 	int i;
879 
880 	idlen = got_hash_digest_length(algo);
881 
882 	*nentries = 0;
883 	if (remain == 0)
884 		return NULL; /* tree is empty */
885 
886 	while (remain > 0) {
887 		size_t elen;
888 
889 		if (*nentries >= *nentries_alloc) {
890 			pte = recallocarray(*entries, *nentries_alloc,
891 			    *nentries_alloc + nalloc, sizeof(**entries));
892 			if (pte == NULL) {
893 				err = got_error_from_errno("recallocarray");
894 				goto done;
895 			}
896 			*entries = pte;
897 			*nentries_alloc += nalloc;
898 		}
899 
900 		pte = &(*entries)[*nentries];
901 		err = parse_tree_entry(pte, &elen, buf, remain, algo, idlen);
902 		if (err)
903 			goto done;
904 		buf += elen;
905 		remain -= elen;
906 		(*nentries)++;
907 	}
908 
909 	if (remain != 0) {
910 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
911 		goto done;
912 	}
913 
914 	if (*nentries > 1) {
915 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
916 
917 		for (i = 0; i < *nentries - 1; i++) {
918 			struct got_parsed_tree_entry *prev = &(*entries)[i];
919 			pte = &(*entries)[i + 1];
920 			if (got_path_cmp(prev->name, pte->name,
921 			    prev->namelen, pte->namelen) == 0) {
922 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
923 				break;
924 			}
925 		}
926 	}
927 done:
928 	if (err)
929 		*nentries = 0;
930 	return err;
931 }
932 
933 const struct got_error *
934 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
935     size_t *nentries_alloc, uint8_t **p, int fd,
936     struct got_object_id *expected_id)
937 {
938 	const struct got_error *err = NULL;
939 	struct got_object *obj = NULL;
940 	size_t len;
941 	struct got_inflate_checksum csum;
942 	struct got_hash ctx;
943 	struct got_object_id id;
944 
945 	memset(&id, 0, sizeof(id));
946 	id.algo = expected_id->algo;
947 
948 	got_hash_init(&ctx, expected_id->algo);
949 	memset(&csum, 0, sizeof(csum));
950 	csum.output_ctx = &ctx;
951 
952 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
953 	if (err)
954 		return err;
955 
956 	got_hash_final(&ctx, id.hash);
957 	if (got_object_id_cmp(expected_id, &id) != 0) {
958 		err = got_error_checksum(expected_id);
959 		goto done;
960 	}
961 
962 	err = got_object_parse_header(&obj, *p, len);
963 	if (err)
964 		goto done;
965 
966 	if (len < obj->hdrlen + obj->size) {
967 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
968 		goto done;
969 	}
970 
971 	/* Skip object header. */
972 	len -= obj->hdrlen;
973 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
974 	    *p + obj->hdrlen, len, expected_id->algo);
975 done:
976 	if (obj)
977 		got_object_close(obj);
978 	return err;
979 }
980 
981 void
982 got_object_tag_close(struct got_tag_object *tag)
983 {
984 	if (tag->refcnt > 0) {
985 		tag->refcnt--;
986 		if (tag->refcnt > 0)
987 			return;
988 	}
989 
990 	free(tag->tag);
991 	free(tag->tagger);
992 	free(tag->tagmsg);
993 	free(tag);
994 }
995 
996 const struct got_error *
997 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len,
998     enum got_hash_algorithm algo)
999 {
1000 	const struct got_error *err = NULL;
1001 	size_t remain = len;
1002 	char *s = buf;
1003 	size_t label_len, id_len;
1004 
1005 	id_len = got_hash_digest_string_length(algo);
1006 
1007 	if (remain == 0)
1008 		return got_error(GOT_ERR_BAD_OBJ_DATA);
1009 
1010 	*tag = calloc(1, sizeof(**tag));
1011 	if (*tag == NULL)
1012 		return got_error_from_errno("calloc");
1013 
1014 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
1015 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
1016 		remain -= label_len;
1017 		if (remain < id_len) {
1018 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1019 			goto done;
1020 		}
1021 		s += label_len;
1022 		if (!got_parse_hash_digest((*tag)->id.hash, s, algo)) {
1023 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1024 			goto done;
1025 		}
1026 		(*tag)->id.algo = algo;
1027 		remain -= id_len;
1028 		s += id_len;
1029 	} else {
1030 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1031 		goto done;
1032 	}
1033 
1034 	if (remain <= 0) {
1035 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1036 		goto done;
1037 	}
1038 
1039 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1040 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1041 		remain -= label_len;
1042 		if (remain <= 0) {
1043 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1044 			goto done;
1045 		}
1046 		s += label_len;
1047 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1048 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1049 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1050 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1051 			s += label_len;
1052 			remain -= label_len;
1053 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1054 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1055 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1056 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1057 			s += label_len;
1058 			remain -= label_len;
1059 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1060 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1061 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1062 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1063 			s += label_len;
1064 			remain -= label_len;
1065 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1066 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1067 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1068 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1069 			s += label_len;
1070 			remain -= label_len;
1071 		} else {
1072 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1073 			goto done;
1074 		}
1075 
1076 		if (remain <= 0 || *s != '\n') {
1077 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1078 			goto done;
1079 		}
1080 		s++;
1081 		remain--;
1082 		if (remain <= 0) {
1083 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1084 			goto done;
1085 		}
1086 	} else {
1087 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1088 		goto done;
1089 	}
1090 
1091 	label_len = strlen(GOT_TAG_LABEL_TAG);
1092 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1093 		char *p;
1094 		size_t slen;
1095 		remain -= label_len;
1096 		if (remain <= 0) {
1097 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1098 			goto done;
1099 		}
1100 		s += label_len;
1101 		p = memchr(s, '\n', remain);
1102 		if (p == NULL) {
1103 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1104 			goto done;
1105 		}
1106 		*p = '\0';
1107 		slen = strlen(s);
1108 		(*tag)->tag = strndup(s, slen);
1109 		if ((*tag)->tag == NULL) {
1110 			err = got_error_from_errno("strndup");
1111 			goto done;
1112 		}
1113 		s += slen + 1;
1114 		remain -= slen + 1;
1115 		if (remain <= 0) {
1116 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1117 			goto done;
1118 		}
1119 	} else {
1120 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1121 		goto done;
1122 	}
1123 
1124 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1125 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1126 		char *p;
1127 		size_t slen;
1128 
1129 		remain -= label_len;
1130 		if (remain <= 0) {
1131 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1132 			goto done;
1133 		}
1134 		s += label_len;
1135 		p = memchr(s, '\n', remain);
1136 		if (p == NULL) {
1137 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1138 			goto done;
1139 		}
1140 		*p = '\0';
1141 		slen = strlen(s);
1142 		err = parse_commit_time(&(*tag)->tagger_time,
1143 		    &(*tag)->tagger_gmtoff, s);
1144 		if (err)
1145 			goto done;
1146 		(*tag)->tagger = strdup(s);
1147 		if ((*tag)->tagger == NULL) {
1148 			err = got_error_from_errno("strdup");
1149 			goto done;
1150 		}
1151 		s += slen + 1;
1152 		remain -= slen + 1;
1153 		if (remain < 0) {
1154 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1155 			goto done;
1156 		}
1157 	} else {
1158 		/* Some old tags in the Linux git repo have no tagger. */
1159 		(*tag)->tagger = strdup("");
1160 		if ((*tag)->tagger == NULL) {
1161 			err = got_error_from_errno("strdup");
1162 			goto done;
1163 		}
1164 	}
1165 
1166 	(*tag)->tagmsg = strndup(s, remain);
1167 	if ((*tag)->tagmsg == NULL) {
1168 		err = got_error_from_errno("strndup");
1169 		goto done;
1170 	}
1171 done:
1172 	if (err) {
1173 		got_object_tag_close(*tag);
1174 		*tag = NULL;
1175 	}
1176 	return err;
1177 }
1178 
1179 const struct got_error *
1180 got_object_read_tag(struct got_tag_object **tag, int fd,
1181     struct got_object_id *expected_id, size_t expected_size)
1182 {
1183 	const struct got_error *err = NULL;
1184 	struct got_object *obj = NULL;
1185 	size_t len;
1186 	uint8_t *p;
1187 	struct got_inflate_checksum csum;
1188 	struct got_hash ctx;
1189 	struct got_object_id id;
1190 
1191 	memset(&id, 0, sizeof(id));
1192 	id.algo = expected_id->algo;
1193 
1194 	got_hash_init(&ctx, expected_id->algo);
1195 	memset(&csum, 0, sizeof(csum));
1196 	csum.output_ctx = &ctx;
1197 
1198 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1199 	    expected_size, fd);
1200 	if (err)
1201 		return err;
1202 
1203 	got_hash_final(&ctx, id.hash);
1204 	if (got_object_id_cmp(expected_id, &id) != 0) {
1205 		err = got_error_checksum(expected_id);
1206 		goto done;
1207 	}
1208 
1209 	err = got_object_parse_header(&obj, p, len);
1210 	if (err)
1211 		goto done;
1212 
1213 	if (len < obj->hdrlen + obj->size) {
1214 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1215 		goto done;
1216 	}
1217 
1218 	/* Skip object header. */
1219 	len -= obj->hdrlen;
1220 	err = got_object_parse_tag(tag, p + obj->hdrlen, len,
1221 	    expected_id->algo);
1222 done:
1223 	free(p);
1224 	if (obj)
1225 		got_object_close(obj);
1226 	return err;
1227 }
1228 
1229 const struct got_error *
1230 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1231 {
1232 	const struct got_error *err = NULL;
1233 	static const size_t blocksize = 512;
1234 	size_t n, total, remain;
1235 	uint8_t *buf;
1236 
1237 	*outbuf = NULL;
1238 	*outlen = 0;
1239 
1240 	buf = malloc(blocksize);
1241 	if (buf == NULL)
1242 		return got_error_from_errno("malloc");
1243 
1244 	remain = blocksize;
1245 	total = 0;
1246 	for (;;) {
1247 		if (remain == 0) {
1248 			uint8_t *newbuf;
1249 			newbuf = reallocarray(buf, 1, total + blocksize);
1250 			if (newbuf == NULL) {
1251 				err = got_error_from_errno("reallocarray");
1252 				goto done;
1253 			}
1254 			buf = newbuf;
1255 			remain += blocksize;
1256 		}
1257 		n = fread(buf + total, 1, remain, f);
1258 		if (n == 0) {
1259 			if (ferror(f)) {
1260 				err = got_ferror(f, GOT_ERR_IO);
1261 				goto done;
1262 			}
1263 			break; /* EOF */
1264 		}
1265 		remain -= n;
1266 		total += n;
1267 	};
1268 
1269 done:
1270 	if (err == NULL) {
1271 		*outbuf = buf;
1272 		*outlen = total;
1273 	} else
1274 		free(buf);
1275 	return err;
1276 }