op public repos

Blob

Date:: Thu Jan 19 02:59:28 2023 UTC
Message:: fix typo: overlapping comparison always evaluates false 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 = SHA1_DIGEST_STRING_LENGTH;
92 
93 	*outbuf = malloc(len);
94 	if (*outbuf == NULL)
95 		return got_error_from_errno("malloc");
96 
97 	if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
98 		free(*outbuf);
99 		*outbuf = NULL;
100 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
101 	}
102 
103 	return NULL;
104 }
105 
106 void
107 got_object_close(struct got_object *obj)
108 {
109 	if (obj->refcnt > 0) {
110 		obj->refcnt--;
111 		if (obj->refcnt > 0)
112 			return;
113 	}
114 
115 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
116 		struct got_delta *delta;
117 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
118 			delta = STAILQ_FIRST(&obj->deltas.entries);
119 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
120 			free(delta);
121 		}
122 	}
123 	free(obj);
124 }
125 
126 const struct got_error *
127 got_object_raw_close(struct got_raw_object *obj)
128 {
129 	const struct got_error *err = NULL;
130 
131 	if (obj->refcnt > 0) {
132 		obj->refcnt--;
133 		if (obj->refcnt > 0)
134 			return NULL;
135 	}
136 
137 	if (obj->close_cb)
138 		obj->close_cb(obj);
139 
140 	if (obj->f == NULL) {
141 		if (obj->fd != -1) {
142 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
143 				err = got_error_from_errno("munmap");
144 			if (close(obj->fd) == -1 && err == NULL)
145 				err = got_error_from_errno("close");
146 		} else
147 			free(obj->data);
148 	} else {
149 		if (fclose(obj->f) == EOF && err == NULL)
150 			err = got_error_from_errno("fclose");
151 	}
152 	free(obj);
153 	return err;
154 }
155 
156 void
157 got_object_qid_free(struct got_object_qid *qid)
158 {
159 	free(qid);
160 }
161 
162 void
163 got_object_id_queue_free(struct got_object_id_queue *ids)
164 {
165 	struct got_object_qid *qid;
166 
167 	while (!STAILQ_EMPTY(ids)) {
168 		qid = STAILQ_FIRST(ids);
169 		STAILQ_REMOVE_HEAD(ids, entry);
170 		got_object_qid_free(qid);
171 	}
172 }
173 
174 const struct got_error *
175 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
176 {
177 	const char *obj_labels[] = {
178 		GOT_OBJ_LABEL_COMMIT,
179 		GOT_OBJ_LABEL_TREE,
180 		GOT_OBJ_LABEL_BLOB,
181 		GOT_OBJ_LABEL_TAG,
182 	};
183 	const int obj_types[] = {
184 		GOT_OBJ_TYPE_COMMIT,
185 		GOT_OBJ_TYPE_TREE,
186 		GOT_OBJ_TYPE_BLOB,
187 		GOT_OBJ_TYPE_TAG,
188 	};
189 	int type = 0;
190 	size_t size = 0;
191 	size_t i;
192 	char *end;
193 
194 	*obj = NULL;
195 
196 	end = memchr(buf, '\0', len);
197 	if (end == NULL)
198 		return got_error(GOT_ERR_BAD_OBJ_HDR);
199 
200 	for (i = 0; i < nitems(obj_labels); i++) {
201 		const char *label = obj_labels[i];
202 		size_t label_len = strlen(label);
203 		const char *errstr;
204 
205 		if (len <= label_len || buf + label_len >= end ||
206 		    strncmp(buf, label, label_len) != 0)
207 			continue;
208 
209 		type = obj_types[i];
210 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
211 		if (errstr != NULL)
212 			return got_error(GOT_ERR_BAD_OBJ_HDR);
213 		break;
214 	}
215 
216 	if (type == 0)
217 		return got_error(GOT_ERR_BAD_OBJ_HDR);
218 
219 	*obj = calloc(1, sizeof(**obj));
220 	if (*obj == NULL)
221 		return got_error_from_errno("calloc");
222 	(*obj)->type = type;
223 	(*obj)->hdrlen = end - buf + 1;
224 	(*obj)->size = size;
225 	return NULL;
226 }
227 
228 const struct got_error *
229 got_object_read_header(struct got_object **obj, int fd)
230 {
231 	const struct got_error *err;
232 	struct got_inflate_buf zb;
233 	uint8_t *buf;
234 	const size_t zbsize = 64;
235 	size_t outlen, totlen;
236 	int nbuf = 1;
237 
238 	*obj = NULL;
239 
240 	buf = malloc(zbsize);
241 	if (buf == NULL)
242 		return got_error_from_errno("malloc");
243 	buf[0] = '\0';
244 
245 	err = got_inflate_init(&zb, buf, zbsize, NULL);
246 	if (err)
247 		return err;
248 
249 	totlen = 0;
250 	do {
251 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
252 		if (err)
253 			goto done;
254 		if (outlen == 0)
255 			break;
256 		totlen += outlen;
257 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
258 			uint8_t *newbuf;
259 			nbuf++;
260 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
261 			if (newbuf == NULL) {
262 				err = got_error_from_errno("recallocarray");
263 				goto done;
264 			}
265 			buf = newbuf;
266 			zb.outbuf = newbuf + totlen;
267 			zb.outlen = (nbuf * zbsize) - totlen;
268 		}
269 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
270 
271 	err = got_object_parse_header(obj, buf, totlen);
272 done:
273 	free(buf);
274 	got_inflate_end(&zb);
275 	return err;
276 }
277 
278 const struct got_error *
279 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
280     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
281     int infd)
282 {
283 	const struct got_error *err = NULL;
284 	struct got_object *obj;
285 	struct got_inflate_checksum csum;
286 	uint8_t sha1[SHA1_DIGEST_LENGTH];
287 	SHA1_CTX sha1_ctx;
288 	size_t len, consumed;
289 	FILE *f = NULL;
290 
291 	*outbuf = NULL;
292 	*size = 0;
293 	*hdrlen = 0;
294 
295 	SHA1Init(&sha1_ctx);
296 	memset(&csum, 0, sizeof(csum));
297 	csum.output_sha1 = &sha1_ctx;
298 
299 	if (lseek(infd, SEEK_SET, 0) == -1)
300 		return got_error_from_errno("lseek");
301 
302 	err = got_object_read_header(&obj, infd);
303 	if (err)
304 		return err;
305 
306 	if (lseek(infd, SEEK_SET, 0) == -1)
307 		return got_error_from_errno("lseek");
308 
309 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
310 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
311 		    obj->size + obj->hdrlen, infd);
312 	} else {
313 		int fd;
314 		/*
315 		 * XXX This uses an extra file descriptor for no good reason.
316 		 * We should have got_inflate_fd_to_fd().
317 		 */
318 		fd = dup(infd);
319 		if (fd == -1)
320 			return got_error_from_errno("dup");
321 		f = fdopen(fd, "r");
322 		if (f == NULL) {
323 			err = got_error_from_errno("fdopen");
324 			abort();
325 			close(fd);
326 			goto done;
327 		}
328 		err = got_inflate_to_fd(&len, f, &csum, outfd);
329 	}
330 	if (err)
331 		goto done;
332 
333 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
334 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
335 		goto done;
336 	}
337 
338 	SHA1Final(sha1, &sha1_ctx);
339 	if (memcmp(expected_id->sha1, sha1, SHA1_DIGEST_LENGTH) != 0) {
340 		char buf[SHA1_DIGEST_STRING_LENGTH];
341 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
342 		    "checksum failure for object %s",
343 		    got_sha1_digest_to_str(expected_id->sha1, buf,
344 		    sizeof(buf)));
345 		goto done;
346 	}
347 
348 	*size = obj->size;
349 	*hdrlen = obj->hdrlen;
350 done:
351 	got_object_close(obj);
352 	if (f && fclose(f) == EOF && err == NULL)
353 		err = got_error_from_errno("fclose");
354 	return err;
355 }
356 
357 struct got_commit_object *
358 got_object_commit_alloc_partial(void)
359 {
360 	struct got_commit_object *commit;
361 
362 	commit = calloc(1, sizeof(*commit));
363 	if (commit == NULL)
364 		return NULL;
365 	commit->tree_id = malloc(sizeof(*commit->tree_id));
366 	if (commit->tree_id == NULL) {
367 		free(commit);
368 		return NULL;
369 	}
370 
371 	STAILQ_INIT(&commit->parent_ids);
372 
373 	return commit;
374 }
375 
376 const struct got_error *
377 got_object_commit_add_parent(struct got_commit_object *commit,
378     const char *id_str)
379 {
380 	const struct got_error *err = NULL;
381 	struct got_object_qid *qid;
382 
383 	err = got_object_qid_alloc_partial(&qid);
384 	if (err)
385 		return err;
386 
387 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
388 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
389 		got_object_qid_free(qid);
390 		return err;
391 	}
392 
393 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
394 	commit->nparents++;
395 
396 	return NULL;
397 }
398 
399 static const struct got_error *
400 parse_gmtoff(time_t *gmtoff, const char *tzstr)
401 {
402 	int sign = 1;
403 	const char *p = tzstr;
404 	time_t h, m;
405 
406 	*gmtoff = 0;
407 
408 	if (*p == '-')
409 		sign = -1;
410 	else if (*p != '+')
411 		return got_error(GOT_ERR_BAD_OBJ_DATA);
412 	p++;
413 	if (!isdigit((unsigned char)*p) &&
414 	    !isdigit((unsigned char)*(p + 1)))
415 		return got_error(GOT_ERR_BAD_OBJ_DATA);
416 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
417 
418 	p += 2;
419 	if (!isdigit((unsigned char)*p) &&
420 	    !isdigit((unsigned char)*(p + 1)))
421 		return got_error(GOT_ERR_BAD_OBJ_DATA);
422 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
423 
424 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
425 	return NULL;
426 }
427 
428 static const struct got_error *
429 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
430 {
431 	const struct got_error *err = NULL;
432 	const char *errstr;
433 	char *space, *tzstr;
434 
435 	/* Parse and strip off trailing timezone indicator string. */
436 	space = strrchr(committer, ' ');
437 	if (space == NULL)
438 		return got_error(GOT_ERR_BAD_OBJ_DATA);
439 	tzstr = strdup(space + 1);
440 	if (tzstr == NULL)
441 		return got_error_from_errno("strdup");
442 	err = parse_gmtoff(gmtoff, tzstr);
443 	free(tzstr);
444 	if (err) {
445 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
446 			return err;
447 		/* Old versions of Git omitted the timestamp. */
448 		*time = 0;
449 		*gmtoff = 0;
450 		return NULL;
451 	}
452 	*space = '\0';
453 
454 	/* Timestamp is separated from committer name + email by space. */
455 	space = strrchr(committer, ' ');
456 	if (space == NULL)
457 		return got_error(GOT_ERR_BAD_OBJ_DATA);
458 
459 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
460 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
461 	if (errstr)
462 		return got_error(GOT_ERR_BAD_OBJ_DATA);
463 
464 	/* Strip off parsed time information, leaving just author and email. */
465 	*space = '\0';
466 
467 	return NULL;
468 }
469 
470 void
471 got_object_commit_close(struct got_commit_object *commit)
472 {
473 	if (commit->refcnt > 0) {
474 		commit->refcnt--;
475 		if (commit->refcnt > 0)
476 			return;
477 	}
478 
479 	got_object_id_queue_free(&commit->parent_ids);
480 	free(commit->tree_id);
481 	free(commit->author);
482 	free(commit->committer);
483 	free(commit->logmsg);
484 	free(commit);
485 }
486 
487 struct got_object_id *
488 got_object_commit_get_tree_id(struct got_commit_object *commit)
489 {
490 	return commit->tree_id;
491 }
492 
493 int
494 got_object_commit_get_nparents(struct got_commit_object *commit)
495 {
496 	return commit->nparents;
497 }
498 
499 const struct got_object_id_queue *
500 got_object_commit_get_parent_ids(struct got_commit_object *commit)
501 {
502 	return &commit->parent_ids;
503 }
504 
505 const char *
506 got_object_commit_get_author(struct got_commit_object *commit)
507 {
508 	return commit->author;
509 }
510 
511 time_t
512 got_object_commit_get_author_time(struct got_commit_object *commit)
513 {
514 	return commit->author_time;
515 }
516 
517 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
518 {
519 	return commit->author_gmtoff;
520 }
521 
522 const char *
523 got_object_commit_get_committer(struct got_commit_object *commit)
524 {
525 	return commit->committer;
526 }
527 
528 time_t
529 got_object_commit_get_committer_time(struct got_commit_object *commit)
530 {
531 	return commit->committer_time;
532 }
533 
534 time_t
535 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
536 {
537 	return commit->committer_gmtoff;
538 }
539 
540 const struct got_error *
541 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
542 {
543 	const struct got_error *err = NULL;
544 	const char *src;
545 	char *dst;
546 	size_t len;
547 
548 	len = strlen(commit->logmsg);
549 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
550 	if (*logmsg == NULL)
551 		return got_error_from_errno("malloc");
552 
553 	/*
554 	 * Strip out unusual headers. Headers are separated from the commit
555 	 * message body by a single empty line.
556 	 */
557 	src = commit->logmsg;
558 	dst = *logmsg;
559 	while (*src != '\0' && *src != '\n') {
560 		int copy_header = 1, eol = 0;
561 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
562 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
563 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
564 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
565 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
566 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
567 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
568 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
569 			copy_header = 0;
570 
571 		while (*src != '\0' && !eol) {
572 			if (copy_header) {
573 				*dst = *src;
574 				dst++;
575 			}
576 			if (*src == '\n')
577 				eol = 1;
578 			src++;
579 		}
580 	}
581 	*dst = '\0';
582 
583 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
584 		err = got_error(GOT_ERR_NO_SPACE);
585 		goto done;
586 	}
587 
588 	/* Trim redundant trailing whitespace. */
589 	len = strlen(*logmsg);
590 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
591 	    isspace((unsigned char)(*logmsg)[len - 1])) {
592 		(*logmsg)[len - 1] = '\0';
593 		len--;
594 	}
595 
596 	/* Append a trailing newline if missing. */
597 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
598 		(*logmsg)[len] = '\n';
599 		(*logmsg)[len + 1] = '\0';
600 	}
601 done:
602 	if (err) {
603 		free(*logmsg);
604 		*logmsg = NULL;
605 	}
606 	return err;
607 }
608 
609 const char *
610 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
611 {
612 	return commit->logmsg;
613 }
614 
615 const struct got_error *
616 got_object_parse_commit(struct got_commit_object **commit, char *buf,
617     size_t len)
618 {
619 	const struct got_error *err = NULL;
620 	char *s = buf;
621 	size_t label_len;
622 	ssize_t remain = (ssize_t)len;
623 
624 	if (remain == 0)
625 		return got_error(GOT_ERR_BAD_OBJ_DATA);
626 
627 	*commit = got_object_commit_alloc_partial();
628 	if (*commit == NULL)
629 		return got_error_from_errno("got_object_commit_alloc_partial");
630 
631 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
632 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
633 		remain -= label_len;
634 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
635 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
636 			goto done;
637 		}
638 		s += label_len;
639 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
640 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
641 			goto done;
642 		}
643 		remain -= SHA1_DIGEST_STRING_LENGTH;
644 		s += SHA1_DIGEST_STRING_LENGTH;
645 	} else {
646 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
647 		goto done;
648 	}
649 
650 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
651 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
652 		remain -= label_len;
653 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
654 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
655 			goto done;
656 		}
657 		s += label_len;
658 		err = got_object_commit_add_parent(*commit, s);
659 		if (err)
660 			goto done;
661 
662 		remain -= SHA1_DIGEST_STRING_LENGTH;
663 		s += SHA1_DIGEST_STRING_LENGTH;
664 	}
665 
666 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
667 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
668 		char *p;
669 		size_t slen;
670 
671 		remain -= label_len;
672 		if (remain <= 0) {
673 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
674 			goto done;
675 		}
676 		s += label_len;
677 		p = memchr(s, '\n', remain);
678 		if (p == NULL) {
679 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
680 			goto done;
681 		}
682 		*p = '\0';
683 		slen = strlen(s);
684 		err = parse_commit_time(&(*commit)->author_time,
685 		    &(*commit)->author_gmtoff, s);
686 		if (err)
687 			goto done;
688 		(*commit)->author = strdup(s);
689 		if ((*commit)->author == NULL) {
690 			err = got_error_from_errno("strdup");
691 			goto done;
692 		}
693 		s += slen + 1;
694 		remain -= slen + 1;
695 	}
696 
697 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
698 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
699 		char *p;
700 		size_t slen;
701 
702 		remain -= label_len;
703 		if (remain <= 0) {
704 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
705 			goto done;
706 		}
707 		s += label_len;
708 		p = memchr(s, '\n', remain);
709 		if (p == NULL) {
710 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
711 			goto done;
712 		}
713 		*p = '\0';
714 		slen = strlen(s);
715 		err = parse_commit_time(&(*commit)->committer_time,
716 		    &(*commit)->committer_gmtoff, s);
717 		if (err)
718 			goto done;
719 		(*commit)->committer = strdup(s);
720 		if ((*commit)->committer == NULL) {
721 			err = got_error_from_errno("strdup");
722 			goto done;
723 		}
724 		s += slen + 1;
725 		remain -= slen + 1;
726 	}
727 
728 	(*commit)->logmsg = strndup(s, remain);
729 	if ((*commit)->logmsg == NULL) {
730 		err = got_error_from_errno("strndup");
731 		goto done;
732 	}
733 done:
734 	if (err) {
735 		got_object_commit_close(*commit);
736 		*commit = NULL;
737 	}
738 	return err;
739 }
740 
741 const struct got_error *
742 got_object_read_commit(struct got_commit_object **commit, int fd,
743     struct got_object_id *expected_id, size_t expected_size)
744 {
745 	struct got_object *obj = NULL;
746 	const struct got_error *err = NULL;
747 	size_t len;
748 	uint8_t *p;
749 	struct got_inflate_checksum csum;
750 	SHA1_CTX sha1_ctx;
751 	struct got_object_id id;
752 
753 	SHA1Init(&sha1_ctx);
754 	memset(&csum, 0, sizeof(csum));
755 	csum.output_sha1 = &sha1_ctx;
756 
757 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
758 	if (err)
759 		return err;
760 
761 	SHA1Final(id.sha1, &sha1_ctx);
762 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
763 		char buf[SHA1_DIGEST_STRING_LENGTH];
764 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
765 		    "checksum failure for object %s",
766 		    got_sha1_digest_to_str(expected_id->sha1, buf,
767 		    sizeof(buf)));
768 		goto done;
769 	}
770 
771 	err = got_object_parse_header(&obj, p, len);
772 	if (err)
773 		goto done;
774 
775 	if (len < obj->hdrlen + obj->size) {
776 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
777 		goto done;
778 	}
779 
780 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
781 		err = got_error(GOT_ERR_OBJ_TYPE);
782 		goto done;
783 	}
784 
785 	/* Skip object header. */
786 	len -= obj->hdrlen;
787 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
788 done:
789 	free(p);
790 	if (obj)
791 		got_object_close(obj);
792 	return err;
793 }
794 
795 void
796 got_object_tree_close(struct got_tree_object *tree)
797 {
798 	if (tree->refcnt > 0) {
799 		tree->refcnt--;
800 		if (tree->refcnt > 0)
801 			return;
802 	}
803 
804 	free(tree->entries);
805 	free(tree);
806 }
807 
808 static const struct got_error *
809 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
810     size_t maxlen)
811 {
812 	char *p, *space;
813 
814 	*elen = 0;
815 
816 	*elen = strnlen(buf, maxlen) + 1;
817 	if (*elen > maxlen)
818 		return got_error(GOT_ERR_BAD_OBJ_DATA);
819 
820 	space = memchr(buf, ' ', *elen);
821 	if (space == NULL || space <= buf)
822 		return got_error(GOT_ERR_BAD_OBJ_DATA);
823 
824 	pte->mode = 0;
825 	p = buf;
826 	while (p < space) {
827 		if (*p < '0' || *p > '7')
828 			return got_error(GOT_ERR_BAD_OBJ_DATA);
829 		pte->mode <<= 3;
830 		pte->mode |= *p - '0';
831 		p++;
832 	}
833 
834 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
835 		return got_error(GOT_ERR_BAD_OBJ_DATA);
836 
837 	pte->name = space + 1;
838 	pte->namelen = strlen(pte->name);
839 	buf += *elen;
840 	pte->id = buf;
841 	*elen += SHA1_DIGEST_LENGTH;
842 	return NULL;
843 }
844 
845 static int
846 pte_cmp(const void *pa, const void *pb)
847 {
848 	const struct got_parsed_tree_entry *a = pa, *b = pb;
849 
850 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
851 }
852 
853 const struct got_error *
854 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
855     size_t *nentries_alloc, uint8_t *buf, size_t len)
856 {
857 	const struct got_error *err = NULL;
858 	size_t remain = len;
859 	const size_t nalloc = 16;
860 	struct got_parsed_tree_entry *pte;
861 	int i;
862 
863 	*nentries = 0;
864 	if (remain == 0)
865 		return NULL; /* tree is empty */
866 
867 	while (remain > 0) {
868 		size_t elen;
869 
870 		if (*nentries >= *nentries_alloc) {
871 			pte = recallocarray(*entries, *nentries_alloc,
872 			    *nentries_alloc + nalloc, sizeof(**entries));
873 			if (pte == NULL) {
874 				err = got_error_from_errno("recallocarray");
875 				goto done;
876 			}
877 			*entries = pte;
878 			*nentries_alloc += nalloc;
879 		}
880 
881 		pte = &(*entries)[*nentries];
882 		err = parse_tree_entry(pte, &elen, buf, remain);
883 		if (err)
884 			goto done;
885 		buf += elen;
886 		remain -= elen;
887 		(*nentries)++;
888 	}
889 
890 	if (remain != 0) {
891 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
892 		goto done;
893 	}
894 
895 	if (*nentries > 1) {
896 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
897 
898 		for (i = 0; i < *nentries - 1; i++) {
899 			struct got_parsed_tree_entry *prev = &(*entries)[i];
900 			pte = &(*entries)[i + 1];
901 			if (got_path_cmp(prev->name, pte->name,
902 			    prev->namelen, pte->namelen) == 0) {
903 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
904 				break;
905 			}
906 		}
907 	}
908 done:
909 	if (err)
910 		*nentries = 0;
911 	return err;
912 }
913 
914 const struct got_error *
915 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
916     size_t *nentries_alloc, uint8_t **p, int fd,
917     struct got_object_id *expected_id)
918 {
919 	const struct got_error *err = NULL;
920 	struct got_object *obj = NULL;
921 	size_t len;
922 	struct got_inflate_checksum csum;
923 	SHA1_CTX sha1_ctx;
924 	struct got_object_id id;
925 
926 	SHA1Init(&sha1_ctx);
927 	memset(&csum, 0, sizeof(csum));
928 	csum.output_sha1 = &sha1_ctx;
929 
930 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
931 	if (err)
932 		return err;
933 
934 	SHA1Final(id.sha1, &sha1_ctx);
935 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
936 		char buf[SHA1_DIGEST_STRING_LENGTH];
937 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
938 		    "checksum failure for object %s",
939 		    got_sha1_digest_to_str(expected_id->sha1, buf,
940 		    sizeof(buf)));
941 		goto done;
942 	}
943 
944 	err = got_object_parse_header(&obj, *p, len);
945 	if (err)
946 		goto done;
947 
948 	if (len < obj->hdrlen + obj->size) {
949 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
950 		goto done;
951 	}
952 
953 	/* Skip object header. */
954 	len -= obj->hdrlen;
955 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
956 	    *p + obj->hdrlen, len);
957 done:
958 	if (obj)
959 		got_object_close(obj);
960 	return err;
961 }
962 
963 void
964 got_object_tag_close(struct got_tag_object *tag)
965 {
966 	if (tag->refcnt > 0) {
967 		tag->refcnt--;
968 		if (tag->refcnt > 0)
969 			return;
970 	}
971 
972 	free(tag->tag);
973 	free(tag->tagger);
974 	free(tag->tagmsg);
975 	free(tag);
976 }
977 
978 const struct got_error *
979 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
980 {
981 	const struct got_error *err = NULL;
982 	size_t remain = len;
983 	char *s = buf;
984 	size_t label_len;
985 
986 	if (remain == 0)
987 		return got_error(GOT_ERR_BAD_OBJ_DATA);
988 
989 	*tag = calloc(1, sizeof(**tag));
990 	if (*tag == NULL)
991 		return got_error_from_errno("calloc");
992 
993 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
994 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
995 		remain -= label_len;
996 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
997 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
998 			goto done;
999 		}
1000 		s += label_len;
1001 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
1002 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1003 			goto done;
1004 		}
1005 		remain -= SHA1_DIGEST_STRING_LENGTH;
1006 		s += SHA1_DIGEST_STRING_LENGTH;
1007 	} else {
1008 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1009 		goto done;
1010 	}
1011 
1012 	if (remain <= 0) {
1013 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1014 		goto done;
1015 	}
1016 
1017 	label_len = strlen(GOT_TAG_LABEL_TYPE);
1018 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
1019 		remain -= label_len;
1020 		if (remain <= 0) {
1021 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1022 			goto done;
1023 		}
1024 		s += label_len;
1025 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
1026 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
1027 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
1028 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
1029 			s += label_len;
1030 			remain -= label_len;
1031 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
1032 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
1033 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
1034 			label_len = strlen(GOT_OBJ_LABEL_TREE);
1035 			s += label_len;
1036 			remain -= label_len;
1037 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
1038 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
1039 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
1040 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
1041 			s += label_len;
1042 			remain -= label_len;
1043 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
1044 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
1045 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
1046 			label_len = strlen(GOT_OBJ_LABEL_TAG);
1047 			s += label_len;
1048 			remain -= label_len;
1049 		} else {
1050 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1051 			goto done;
1052 		}
1053 
1054 		if (remain <= 0 || *s != '\n') {
1055 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1056 			goto done;
1057 		}
1058 		s++;
1059 		remain--;
1060 		if (remain <= 0) {
1061 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1062 			goto done;
1063 		}
1064 	} else {
1065 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1066 		goto done;
1067 	}
1068 
1069 	label_len = strlen(GOT_TAG_LABEL_TAG);
1070 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1071 		char *p;
1072 		size_t slen;
1073 		remain -= label_len;
1074 		if (remain <= 0) {
1075 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1076 			goto done;
1077 		}
1078 		s += label_len;
1079 		p = memchr(s, '\n', remain);
1080 		if (p == NULL) {
1081 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1082 			goto done;
1083 		}
1084 		*p = '\0';
1085 		slen = strlen(s);
1086 		(*tag)->tag = strndup(s, slen);
1087 		if ((*tag)->tag == NULL) {
1088 			err = got_error_from_errno("strndup");
1089 			goto done;
1090 		}
1091 		s += slen + 1;
1092 		remain -= slen + 1;
1093 		if (remain <= 0) {
1094 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1095 			goto done;
1096 		}
1097 	} else {
1098 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1099 		goto done;
1100 	}
1101 
1102 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1103 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1104 		char *p;
1105 		size_t slen;
1106 
1107 		remain -= label_len;
1108 		if (remain <= 0) {
1109 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1110 			goto done;
1111 		}
1112 		s += label_len;
1113 		p = memchr(s, '\n', remain);
1114 		if (p == NULL) {
1115 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1116 			goto done;
1117 		}
1118 		*p = '\0';
1119 		slen = strlen(s);
1120 		err = parse_commit_time(&(*tag)->tagger_time,
1121 		    &(*tag)->tagger_gmtoff, s);
1122 		if (err)
1123 			goto done;
1124 		(*tag)->tagger = strdup(s);
1125 		if ((*tag)->tagger == NULL) {
1126 			err = got_error_from_errno("strdup");
1127 			goto done;
1128 		}
1129 		s += slen + 1;
1130 		remain -= slen + 1;
1131 		if (remain < 0) {
1132 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1133 			goto done;
1134 		}
1135 	} else {
1136 		/* Some old tags in the Linux git repo have no tagger. */
1137 		(*tag)->tagger = strdup("");
1138 		if ((*tag)->tagger == NULL) {
1139 			err = got_error_from_errno("strdup");
1140 			goto done;
1141 		}
1142 	}
1143 
1144 	(*tag)->tagmsg = strndup(s, remain);
1145 	if ((*tag)->tagmsg == NULL) {
1146 		err = got_error_from_errno("strndup");
1147 		goto done;
1148 	}
1149 done:
1150 	if (err) {
1151 		got_object_tag_close(*tag);
1152 		*tag = NULL;
1153 	}
1154 	return err;
1155 }
1156 
1157 const struct got_error *
1158 got_object_read_tag(struct got_tag_object **tag, int fd,
1159     struct got_object_id *expected_id, size_t expected_size)
1160 {
1161 	const struct got_error *err = NULL;
1162 	struct got_object *obj = NULL;
1163 	size_t len;
1164 	uint8_t *p;
1165 	struct got_inflate_checksum csum;
1166 	SHA1_CTX sha1_ctx;
1167 	struct got_object_id id;
1168 
1169 	SHA1Init(&sha1_ctx);
1170 	memset(&csum, 0, sizeof(csum));
1171 	csum.output_sha1 = &sha1_ctx;
1172 
1173 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1174 	    expected_size, fd);
1175 	if (err)
1176 		return err;
1177 
1178 	SHA1Final(id.sha1, &sha1_ctx);
1179 	if (memcmp(expected_id->sha1, id.sha1, SHA1_DIGEST_LENGTH) != 0) {
1180 		char buf[SHA1_DIGEST_STRING_LENGTH];
1181 		err = got_error_fmt(GOT_ERR_OBJ_CSUM,
1182 		    "checksum failure for object %s",
1183 		    got_sha1_digest_to_str(expected_id->sha1, buf,
1184 		    sizeof(buf)));
1185 		goto done;
1186 	}
1187 
1188 	err = got_object_parse_header(&obj, p, len);
1189 	if (err)
1190 		goto done;
1191 
1192 	if (len < obj->hdrlen + obj->size) {
1193 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1194 		goto done;
1195 	}
1196 
1197 	/* Skip object header. */
1198 	len -= obj->hdrlen;
1199 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1200 done:
1201 	free(p);
1202 	if (obj)
1203 		got_object_close(obj);
1204 	return err;
1205 }
1206 
1207 const struct got_error *
1208 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
1209 {
1210 	const struct got_error *err = NULL;
1211 	static const size_t blocksize = 512;
1212 	size_t n, total, remain;
1213 	uint8_t *buf;
1214 
1215 	*outbuf = NULL;
1216 	*outlen = 0;
1217 
1218 	buf = malloc(blocksize);
1219 	if (buf == NULL)
1220 		return got_error_from_errno("malloc");
1221 
1222 	remain = blocksize;
1223 	total = 0;
1224 	for (;;) {
1225 		if (remain == 0) {
1226 			uint8_t *newbuf;
1227 			newbuf = reallocarray(buf, 1, total + blocksize);
1228 			if (newbuf == NULL) {
1229 				err = got_error_from_errno("reallocarray");
1230 				goto done;
1231 			}
1232 			buf = newbuf;
1233 			remain += blocksize;
1234 		}
1235 		n = fread(buf + total, 1, remain, f);
1236 		if (n == 0) {
1237 			if (ferror(f)) {
1238 				err = got_ferror(f, GOT_ERR_IO);
1239 				goto done;
1240 			}
1241 			break; /* EOF */
1242 		}
1243 		remain -= n;
1244 		total += n;
1245 	};
1246 
1247 done:
1248 	if (err == NULL) {
1249 		*outbuf = buf;
1250 		*outlen = total;
1251 	} else
1252 		free(buf);
1253 	return err;
1254 }