Gotweb

Blob

Date:: Fri Apr 28 11:41:25 2023 UTC
Message:: when finding changed paths iterate tree entries in on-disk order for speed ok op, jamsek
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 "got_compat.h"
18 
19 #include <sys/types.h>
20 #include <sys/stat.h>
21 #include <sys/queue.h>
22 #include <sys/uio.h>
23 #include <sys/socket.h>
24 #include <sys/wait.h>
25 #include <sys/mman.h>
26 
27 #include <errno.h>
28 #include <stdio.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <stdint.h>
32 #include <zlib.h>
33 #include <ctype.h>
34 #include <limits.h>
35 #include <time.h>
36 #include <unistd.h>
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_hash.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_qid.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 const struct got_error *
59 got_object_type_label(const char **label, int obj_type)
60 {
61 	const struct got_error *err = NULL;
62 
63 	switch (obj_type) {
64 	case GOT_OBJ_TYPE_BLOB:
65 		*label = GOT_OBJ_LABEL_BLOB;
66 		break;
67 	case GOT_OBJ_TYPE_TREE:
68 		*label = GOT_OBJ_LABEL_TREE;
69 		break;
70 	case GOT_OBJ_TYPE_COMMIT:
71 		*label = GOT_OBJ_LABEL_COMMIT;
72 		break;
73 	case GOT_OBJ_TYPE_TAG:
74 		*label = GOT_OBJ_LABEL_TAG;
75 		break;
76 	default:
77 		*label = NULL;
78 		err = got_error(GOT_ERR_OBJ_TYPE);
79 		break;
80 	}
81 
82 	return err;
83 }
84 
85 void
86 got_object_close(struct got_object *obj)
87 {
88 	if (obj->refcnt > 0) {
89 		obj->refcnt--;
90 		if (obj->refcnt > 0)
91 			return;
92 	}
93 
94 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
95 		struct got_delta *delta;
96 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
97 			delta = STAILQ_FIRST(&obj->deltas.entries);
98 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
99 			free(delta);
100 		}
101 	}
102 	free(obj);
103 }
104 
105 const struct got_error *
106 got_object_raw_close(struct got_raw_object *obj)
107 {
108 	const struct got_error *err = NULL;
109 
110 	if (obj->refcnt > 0) {
111 		obj->refcnt--;
112 		if (obj->refcnt > 0)
113 			return NULL;
114 	}
115 
116 	if (obj->close_cb)
117 		obj->close_cb(obj);
118 
119 	if (obj->f == NULL) {
120 		if (obj->fd != -1) {
121 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
122 				err = got_error_from_errno("munmap");
123 			if (close(obj->fd) == -1 && err == NULL)
124 				err = got_error_from_errno("close");
125 		} else
126 			free(obj->data);
127 	} else {
128 		if (fclose(obj->f) == EOF && err == NULL)
129 			err = got_error_from_errno("fclose");
130 	}
131 	free(obj);
132 	return err;
133 }
134 
135 const struct got_error *
136 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
137 {
138 	const char *obj_labels[] = {
139 		GOT_OBJ_LABEL_COMMIT,
140 		GOT_OBJ_LABEL_TREE,
141 		GOT_OBJ_LABEL_BLOB,
142 		GOT_OBJ_LABEL_TAG,
143 	};
144 	const int obj_types[] = {
145 		GOT_OBJ_TYPE_COMMIT,
146 		GOT_OBJ_TYPE_TREE,
147 		GOT_OBJ_TYPE_BLOB,
148 		GOT_OBJ_TYPE_TAG,
149 	};
150 	int type = 0;
151 	size_t size = 0;
152 	size_t i;
153 	char *end;
154 
155 	*obj = NULL;
156 
157 	end = memchr(buf, '\0', len);
158 	if (end == NULL)
159 		return got_error(GOT_ERR_BAD_OBJ_HDR);
160 
161 	for (i = 0; i < nitems(obj_labels); i++) {
162 		const char *label = obj_labels[i];
163 		size_t label_len = strlen(label);
164 		const char *errstr;
165 
166 		if (len <= label_len || buf + label_len >= end ||
167 		    strncmp(buf, label, label_len) != 0)
168 			continue;
169 
170 		type = obj_types[i];
171 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
172 		if (errstr != NULL)
173 			return got_error(GOT_ERR_BAD_OBJ_HDR);
174 		break;
175 	}
176 
177 	if (type == 0)
178 		return got_error(GOT_ERR_BAD_OBJ_HDR);
179 
180 	*obj = calloc(1, sizeof(**obj));
181 	if (*obj == NULL)
182 		return got_error_from_errno("calloc");
183 	(*obj)->type = type;
184 	(*obj)->hdrlen = end - buf + 1;
185 	(*obj)->size = size;
186 	return NULL;
187 }
188 
189 const struct got_error *
190 got_object_read_header(struct got_object **obj, int fd)
191 {
192 	const struct got_error *err;
193 	struct got_inflate_buf zb;
194 	uint8_t *buf;
195 	const size_t zbsize = 64;
196 	size_t outlen, totlen;
197 	int nbuf = 1;
198 
199 	*obj = NULL;
200 
201 	buf = malloc(zbsize);
202 	if (buf == NULL)
203 		return got_error_from_errno("malloc");
204 	buf[0] = '\0';
205 
206 	err = got_inflate_init(&zb, buf, zbsize, NULL);
207 	if (err)
208 		return err;
209 
210 	totlen = 0;
211 	do {
212 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
213 		if (err)
214 			goto done;
215 		if (outlen == 0)
216 			break;
217 		totlen += outlen;
218 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
219 			uint8_t *newbuf;
220 			nbuf++;
221 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
222 			if (newbuf == NULL) {
223 				err = got_error_from_errno("recallocarray");
224 				goto done;
225 			}
226 			buf = newbuf;
227 			zb.outbuf = newbuf + totlen;
228 			zb.outlen = (nbuf * zbsize) - totlen;
229 		}
230 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
231 
232 	err = got_object_parse_header(obj, buf, totlen);
233 done:
234 	free(buf);
235 	got_inflate_end(&zb);
236 	return err;
237 }
238 
239 const struct got_error *
240 got_object_read_raw(uint8_t **outbuf, off_t *size, size_t *hdrlen,
241     size_t max_in_mem_size, int outfd, struct got_object_id *expected_id,
242     int infd)
243 {
244 	const struct got_error *err = NULL;
245 	struct got_object *obj;
246 	struct got_inflate_checksum csum;
247 	struct got_object_id id;
248 	struct got_hash ctx;
249 	size_t len, consumed;
250 	FILE *f = NULL;
251 
252 	*outbuf = NULL;
253 	*size = 0;
254 	*hdrlen = 0;
255 
256 	got_hash_init(&ctx, GOT_HASH_SHA1);
257 	memset(&csum, 0, sizeof(csum));
258 	csum.output_ctx = &ctx;
259 
260 	if (lseek(infd, SEEK_SET, 0) == -1)
261 		return got_error_from_errno("lseek");
262 
263 	err = got_object_read_header(&obj, infd);
264 	if (err)
265 		return err;
266 
267 	if (lseek(infd, SEEK_SET, 0) == -1)
268 		return got_error_from_errno("lseek");
269 
270 	if (obj->size + obj->hdrlen <= max_in_mem_size) {
271 		err = got_inflate_to_mem_fd(outbuf, &len, &consumed, &csum,
272 		    obj->size + obj->hdrlen, infd);
273 	} else {
274 		int fd;
275 		/*
276 		 * XXX This uses an extra file descriptor for no good reason.
277 		 * We should have got_inflate_fd_to_fd().
278 		 */
279 		fd = dup(infd);
280 		if (fd == -1)
281 			return got_error_from_errno("dup");
282 		f = fdopen(fd, "r");
283 		if (f == NULL) {
284 			err = got_error_from_errno("fdopen");
285 			abort();
286 			close(fd);
287 			goto done;
288 		}
289 		err = got_inflate_to_fd(&len, f, &csum, outfd);
290 	}
291 	if (err)
292 		goto done;
293 
294 	if (len < obj->hdrlen || len != obj->hdrlen + obj->size) {
295 		err = got_error(GOT_ERR_BAD_OBJ_HDR);
296 		goto done;
297 	}
298 
299 	got_hash_final_object_id(&ctx, &id);
300 	if (got_object_id_cmp(expected_id, &id) != 0) {
301 		err = got_error_checksum(expected_id);
302 		goto done;
303 	}
304 
305 	*size = obj->size;
306 	*hdrlen = obj->hdrlen;
307 done:
308 	got_object_close(obj);
309 	if (f && fclose(f) == EOF && err == NULL)
310 		err = got_error_from_errno("fclose");
311 	return err;
312 }
313 
314 struct got_commit_object *
315 got_object_commit_alloc_partial(void)
316 {
317 	struct got_commit_object *commit;
318 
319 	commit = calloc(1, sizeof(*commit));
320 	if (commit == NULL)
321 		return NULL;
322 	commit->tree_id = malloc(sizeof(*commit->tree_id));
323 	if (commit->tree_id == NULL) {
324 		free(commit);
325 		return NULL;
326 	}
327 
328 	STAILQ_INIT(&commit->parent_ids);
329 
330 	return commit;
331 }
332 
333 const struct got_error *
334 got_object_commit_add_parent(struct got_commit_object *commit,
335     const char *id_str)
336 {
337 	const struct got_error *err = NULL;
338 	struct got_object_qid *qid;
339 
340 	err = got_object_qid_alloc_partial(&qid);
341 	if (err)
342 		return err;
343 
344 	if (!got_parse_object_id(&qid->id, id_str, GOT_HASH_SHA1)) {
345 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
346 		got_object_qid_free(qid);
347 		return err;
348 	}
349 
350 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
351 	commit->nparents++;
352 
353 	return NULL;
354 }
355 
356 static const struct got_error *
357 parse_gmtoff(time_t *gmtoff, const char *tzstr)
358 {
359 	int sign = 1;
360 	const char *p = tzstr;
361 	time_t h, m;
362 
363 	*gmtoff = 0;
364 
365 	if (*p == '-')
366 		sign = -1;
367 	else if (*p != '+')
368 		return got_error(GOT_ERR_BAD_OBJ_DATA);
369 	p++;
370 	if (!isdigit((unsigned char)*p) &&
371 	    !isdigit((unsigned char)*(p + 1)))
372 		return got_error(GOT_ERR_BAD_OBJ_DATA);
373 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
374 
375 	p += 2;
376 	if (!isdigit((unsigned char)*p) &&
377 	    !isdigit((unsigned char)*(p + 1)))
378 		return got_error(GOT_ERR_BAD_OBJ_DATA);
379 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
380 
381 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
382 	return NULL;
383 }
384 
385 static const struct got_error *
386 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
387 {
388 	const struct got_error *err = NULL;
389 	const char *errstr;
390 	char *space, *tzstr;
391 
392 	/* Parse and strip off trailing timezone indicator string. */
393 	space = strrchr(committer, ' ');
394 	if (space == NULL)
395 		return got_error(GOT_ERR_BAD_OBJ_DATA);
396 	tzstr = strdup(space + 1);
397 	if (tzstr == NULL)
398 		return got_error_from_errno("strdup");
399 	err = parse_gmtoff(gmtoff, tzstr);
400 	free(tzstr);
401 	if (err) {
402 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
403 			return err;
404 		/* Old versions of Git omitted the timestamp. */
405 		*time = 0;
406 		*gmtoff = 0;
407 		return NULL;
408 	}
409 	*space = '\0';
410 
411 	/* Timestamp is separated from committer name + email by space. */
412 	space = strrchr(committer, ' ');
413 	if (space == NULL)
414 		return got_error(GOT_ERR_BAD_OBJ_DATA);
415 
416 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
417 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
418 	if (errstr)
419 		return got_error(GOT_ERR_BAD_OBJ_DATA);
420 
421 	/* Strip off parsed time information, leaving just author and email. */
422 	*space = '\0';
423 
424 	return NULL;
425 }
426 
427 void
428 got_object_commit_close(struct got_commit_object *commit)
429 {
430 	if (commit->refcnt > 0) {
431 		commit->refcnt--;
432 		if (commit->refcnt > 0)
433 			return;
434 	}
435 
436 	got_object_id_queue_free(&commit->parent_ids);
437 	free(commit->tree_id);
438 	free(commit->author);
439 	free(commit->committer);
440 	free(commit->logmsg);
441 	free(commit);
442 }
443 
444 struct got_object_id *
445 got_object_commit_get_tree_id(struct got_commit_object *commit)
446 {
447 	return commit->tree_id;
448 }
449 
450 int
451 got_object_commit_get_nparents(struct got_commit_object *commit)
452 {
453 	return commit->nparents;
454 }
455 
456 const struct got_object_id_queue *
457 got_object_commit_get_parent_ids(struct got_commit_object *commit)
458 {
459 	return &commit->parent_ids;
460 }
461 
462 const char *
463 got_object_commit_get_author(struct got_commit_object *commit)
464 {
465 	return commit->author;
466 }
467 
468 time_t
469 got_object_commit_get_author_time(struct got_commit_object *commit)
470 {
471 	return commit->author_time;
472 }
473 
474 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
475 {
476 	return commit->author_gmtoff;
477 }
478 
479 const char *
480 got_object_commit_get_committer(struct got_commit_object *commit)
481 {
482 	return commit->committer;
483 }
484 
485 time_t
486 got_object_commit_get_committer_time(struct got_commit_object *commit)
487 {
488 	return commit->committer_time;
489 }
490 
491 time_t
492 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
493 {
494 	return commit->committer_gmtoff;
495 }
496 
497 const struct got_error *
498 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
499 {
500 	const struct got_error *err = NULL;
501 	const char *src;
502 	char *dst;
503 	size_t len;
504 
505 	len = strlen(commit->logmsg);
506 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
507 	if (*logmsg == NULL)
508 		return got_error_from_errno("malloc");
509 
510 	/*
511 	 * Strip out unusual headers. Headers are separated from the commit
512 	 * message body by a single empty line.
513 	 */
514 	src = commit->logmsg;
515 	dst = *logmsg;
516 	while (*src != '\0' && *src != '\n') {
517 		int copy_header = 1, eol = 0;
518 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
519 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
520 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
521 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
522 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
523 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
524 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
525 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
526 			copy_header = 0;
527 
528 		while (*src != '\0' && !eol) {
529 			if (copy_header) {
530 				*dst = *src;
531 				dst++;
532 			}
533 			if (*src == '\n')
534 				eol = 1;
535 			src++;
536 		}
537 	}
538 	*dst = '\0';
539 
540 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
541 		err = got_error(GOT_ERR_NO_SPACE);
542 		goto done;
543 	}
544 
545 	/* Trim redundant trailing whitespace. */
546 	len = strlen(*logmsg);
547 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
548 	    isspace((unsigned char)(*logmsg)[len - 1])) {
549 		(*logmsg)[len - 1] = '\0';
550 		len--;
551 	}
552 
553 	/* Append a trailing newline if missing. */
554 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
555 		(*logmsg)[len] = '\n';
556 		(*logmsg)[len + 1] = '\0';
557 	}
558 done:
559 	if (err) {
560 		free(*logmsg);
561 		*logmsg = NULL;
562 	}
563 	return err;
564 }
565 
566 const char *
567 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
568 {
569 	return commit->logmsg;
570 }
571 
572 const struct got_error *
573 got_object_parse_commit(struct got_commit_object **commit, char *buf,
574     size_t len)
575 {
576 	const struct got_error *err = NULL;
577 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
578 	char *s = buf;
579 	size_t label_len;
580 	ssize_t remain = (ssize_t)len;
581 
582 	if (remain == 0)
583 		return got_error(GOT_ERR_BAD_OBJ_DATA);
584 
585 	*commit = got_object_commit_alloc_partial();
586 	if (*commit == NULL)
587 		return got_error_from_errno("got_object_commit_alloc_partial");
588 
589 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
590 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
591 		remain -= label_len;
592 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
593 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
594 			goto done;
595 		}
596 		s += label_len;
597 		if (!got_parse_object_id((*commit)->tree_id, s, algo)) {
598 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
599 			goto done;
600 		}
601 		remain -= SHA1_DIGEST_STRING_LENGTH;
602 		s += SHA1_DIGEST_STRING_LENGTH;
603 	} else {
604 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
605 		goto done;
606 	}
607 
608 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
609 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
610 		remain -= label_len;
611 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
612 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
613 			goto done;
614 		}
615 		s += label_len;
616 		err = got_object_commit_add_parent(*commit, s);
617 		if (err)
618 			goto done;
619 
620 		remain -= SHA1_DIGEST_STRING_LENGTH;
621 		s += SHA1_DIGEST_STRING_LENGTH;
622 	}
623 
624 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
625 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
626 		char *p;
627 		size_t slen;
628 
629 		remain -= label_len;
630 		if (remain <= 0) {
631 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
632 			goto done;
633 		}
634 		s += label_len;
635 		p = memchr(s, '\n', remain);
636 		if (p == NULL) {
637 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
638 			goto done;
639 		}
640 		*p = '\0';
641 		slen = strlen(s);
642 		err = parse_commit_time(&(*commit)->author_time,
643 		    &(*commit)->author_gmtoff, s);
644 		if (err)
645 			goto done;
646 		(*commit)->author = strdup(s);
647 		if ((*commit)->author == NULL) {
648 			err = got_error_from_errno("strdup");
649 			goto done;
650 		}
651 		s += slen + 1;
652 		remain -= slen + 1;
653 	}
654 
655 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
656 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
657 		char *p;
658 		size_t slen;
659 
660 		remain -= label_len;
661 		if (remain <= 0) {
662 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
663 			goto done;
664 		}
665 		s += label_len;
666 		p = memchr(s, '\n', remain);
667 		if (p == NULL) {
668 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
669 			goto done;
670 		}
671 		*p = '\0';
672 		slen = strlen(s);
673 		err = parse_commit_time(&(*commit)->committer_time,
674 		    &(*commit)->committer_gmtoff, s);
675 		if (err)
676 			goto done;
677 		(*commit)->committer = strdup(s);
678 		if ((*commit)->committer == NULL) {
679 			err = got_error_from_errno("strdup");
680 			goto done;
681 		}
682 		s += slen + 1;
683 		remain -= slen + 1;
684 	}
685 
686 	(*commit)->logmsg = strndup(s, remain);
687 	if ((*commit)->logmsg == NULL) {
688 		err = got_error_from_errno("strndup");
689 		goto done;
690 	}
691 done:
692 	if (err) {
693 		got_object_commit_close(*commit);
694 		*commit = NULL;
695 	}
696 	return err;
697 }
698 
699 const struct got_error *
700 got_object_read_commit(struct got_commit_object **commit, int fd,
701     struct got_object_id *expected_id, size_t expected_size)
702 {
703 	struct got_object *obj = NULL;
704 	const struct got_error *err = NULL;
705 	size_t len;
706 	uint8_t *p;
707 	struct got_inflate_checksum csum;
708 	struct got_hash ctx;
709 	struct got_object_id id;
710 
711 	got_hash_init(&ctx, GOT_HASH_SHA1);
712 	memset(&csum, 0, sizeof(csum));
713 	csum.output_ctx = &ctx;
714 
715 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum, expected_size, fd);
716 	if (err)
717 		return err;
718 
719 	got_hash_final_object_id(&ctx, &id);
720 	if (got_object_id_cmp(expected_id, &id) != 0) {
721 		err = got_error_checksum(expected_id);
722 		goto done;
723 	}
724 
725 	err = got_object_parse_header(&obj, p, len);
726 	if (err)
727 		goto done;
728 
729 	if (len < obj->hdrlen + obj->size) {
730 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
731 		goto done;
732 	}
733 
734 	if (obj->type != GOT_OBJ_TYPE_COMMIT) {
735 		err = got_error(GOT_ERR_OBJ_TYPE);
736 		goto done;
737 	}
738 
739 	/* Skip object header. */
740 	len -= obj->hdrlen;
741 	err = got_object_parse_commit(commit, p + obj->hdrlen, len);
742 done:
743 	free(p);
744 	if (obj)
745 		got_object_close(obj);
746 	return err;
747 }
748 
749 void
750 got_object_tree_close(struct got_tree_object *tree)
751 {
752 	if (tree->refcnt > 0) {
753 		tree->refcnt--;
754 		if (tree->refcnt > 0)
755 			return;
756 	}
757 
758 	free(tree->entries);
759 	free(tree);
760 }
761 
762 const struct got_error *
763 got_object_parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen,
764     char *buf, size_t maxlen)
765 {
766 	char *p, *space;
767 
768 	*elen = 0;
769 
770 	*elen = strnlen(buf, maxlen) + 1;
771 	if (*elen > maxlen)
772 		return got_error(GOT_ERR_BAD_OBJ_DATA);
773 
774 	space = memchr(buf, ' ', *elen);
775 	if (space == NULL || space <= buf)
776 		return got_error(GOT_ERR_BAD_OBJ_DATA);
777 
778 	pte->mode = 0;
779 	p = buf;
780 	while (p < space) {
781 		if (*p < '0' || *p > '7')
782 			return got_error(GOT_ERR_BAD_OBJ_DATA);
783 		pte->mode <<= 3;
784 		pte->mode |= *p - '0';
785 		p++;
786 	}
787 
788 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
789 		return got_error(GOT_ERR_BAD_OBJ_DATA);
790 
791 	pte->name = space + 1;
792 	pte->namelen = strlen(pte->name);
793 	buf += *elen;
794 	pte->id = buf;
795 	*elen += SHA1_DIGEST_LENGTH;
796 	return NULL;
797 }
798 
799 static int
800 pte_cmp(const void *pa, const void *pb)
801 {
802 	const struct got_parsed_tree_entry *a = pa, *b = pb;
803 
804 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
805 }
806 
807 const struct got_error *
808 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
809     size_t *nentries_alloc, uint8_t *buf, size_t len)
810 {
811 	const struct got_error *err = NULL;
812 	size_t remain = len;
813 	const size_t nalloc = 16;
814 	struct got_parsed_tree_entry *pte;
815 	int i;
816 
817 	*nentries = 0;
818 	if (remain == 0)
819 		return NULL; /* tree is empty */
820 
821 	while (remain > 0) {
822 		size_t elen;
823 
824 		if (*nentries >= *nentries_alloc) {
825 			pte = recallocarray(*entries, *nentries_alloc,
826 			    *nentries_alloc + nalloc, sizeof(**entries));
827 			if (pte == NULL) {
828 				err = got_error_from_errno("recallocarray");
829 				goto done;
830 			}
831 			*entries = pte;
832 			*nentries_alloc += nalloc;
833 		}
834 
835 		pte = &(*entries)[*nentries];
836 		err = got_object_parse_tree_entry(pte, &elen, buf, remain);
837 		if (err)
838 			goto done;
839 		buf += elen;
840 		remain -= elen;
841 		(*nentries)++;
842 	}
843 
844 	if (remain != 0) {
845 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
846 		goto done;
847 	}
848 
849 	if (*nentries > 1) {
850 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
851 
852 		for (i = 0; i < *nentries - 1; i++) {
853 			struct got_parsed_tree_entry *prev = &(*entries)[i];
854 			pte = &(*entries)[i + 1];
855 			if (got_path_cmp(prev->name, pte->name,
856 			    prev->namelen, pte->namelen) == 0) {
857 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
858 				break;
859 			}
860 		}
861 	}
862 done:
863 	if (err)
864 		*nentries = 0;
865 	return err;
866 }
867 
868 const struct got_error *
869 got_object_read_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
870     size_t *nentries_alloc, uint8_t **p, int fd,
871     struct got_object_id *expected_id)
872 {
873 	const struct got_error *err = NULL;
874 	struct got_object *obj = NULL;
875 	size_t len;
876 	struct got_inflate_checksum csum;
877 	struct got_hash ctx;
878 	struct got_object_id id;
879 
880 	got_hash_init(&ctx, GOT_HASH_SHA1);
881 	memset(&csum, 0, sizeof(csum));
882 	csum.output_ctx = &ctx;
883 
884 	err = got_inflate_to_mem_fd(p, &len, NULL, &csum, 0, fd);
885 	if (err)
886 		return err;
887 
888 	got_hash_final_object_id(&ctx, &id);
889 	if (got_object_id_cmp(expected_id, &id) != 0) {
890 		err = got_error_checksum(expected_id);
891 		goto done;
892 	}
893 
894 	err = got_object_parse_header(&obj, *p, len);
895 	if (err)
896 		goto done;
897 
898 	if (len < obj->hdrlen + obj->size) {
899 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
900 		goto done;
901 	}
902 
903 	/* Skip object header. */
904 	len -= obj->hdrlen;
905 	err = got_object_parse_tree(entries, nentries, nentries_alloc,
906 	    *p + obj->hdrlen, len);
907 done:
908 	if (obj)
909 		got_object_close(obj);
910 	return err;
911 }
912 
913 void
914 got_object_tag_close(struct got_tag_object *tag)
915 {
916 	if (tag->refcnt > 0) {
917 		tag->refcnt--;
918 		if (tag->refcnt > 0)
919 			return;
920 	}
921 
922 	free(tag->tag);
923 	free(tag->tagger);
924 	free(tag->tagmsg);
925 	free(tag);
926 }
927 
928 const struct got_error *
929 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
930 {
931 	const struct got_error *err = NULL;
932 	enum got_hash_algorithm algo = GOT_HASH_SHA1;
933 	size_t remain = len;
934 	char *s = buf;
935 	size_t label_len;
936 
937 	if (remain == 0)
938 		return got_error(GOT_ERR_BAD_OBJ_DATA);
939 
940 	*tag = calloc(1, sizeof(**tag));
941 	if (*tag == NULL)
942 		return got_error_from_errno("calloc");
943 
944 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
945 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
946 		remain -= label_len;
947 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
948 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
949 			goto done;
950 		}
951 		s += label_len;
952 		if (!got_parse_object_id(&(*tag)->id, s, algo)) {
953 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
954 			goto done;
955 		}
956 		remain -= SHA1_DIGEST_STRING_LENGTH;
957 		s += SHA1_DIGEST_STRING_LENGTH;
958 	} else {
959 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
960 		goto done;
961 	}
962 
963 	if (remain <= 0) {
964 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
965 		goto done;
966 	}
967 
968 	label_len = strlen(GOT_TAG_LABEL_TYPE);
969 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
970 		remain -= label_len;
971 		if (remain <= 0) {
972 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
973 			goto done;
974 		}
975 		s += label_len;
976 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
977 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
978 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
979 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
980 			s += label_len;
981 			remain -= label_len;
982 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
983 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
984 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
985 			label_len = strlen(GOT_OBJ_LABEL_TREE);
986 			s += label_len;
987 			remain -= label_len;
988 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
989 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
990 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
991 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
992 			s += label_len;
993 			remain -= label_len;
994 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
995 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
996 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
997 			label_len = strlen(GOT_OBJ_LABEL_TAG);
998 			s += label_len;
999 			remain -= label_len;
1000 		} else {
1001 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1002 			goto done;
1003 		}
1004 
1005 		if (remain <= 0 || *s != '\n') {
1006 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1007 			goto done;
1008 		}
1009 		s++;
1010 		remain--;
1011 		if (remain <= 0) {
1012 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1013 			goto done;
1014 		}
1015 	} else {
1016 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1017 		goto done;
1018 	}
1019 
1020 	label_len = strlen(GOT_TAG_LABEL_TAG);
1021 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
1022 		char *p;
1023 		size_t slen;
1024 		remain -= label_len;
1025 		if (remain <= 0) {
1026 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1027 			goto done;
1028 		}
1029 		s += label_len;
1030 		p = memchr(s, '\n', remain);
1031 		if (p == NULL) {
1032 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1033 			goto done;
1034 		}
1035 		*p = '\0';
1036 		slen = strlen(s);
1037 		(*tag)->tag = strndup(s, slen);
1038 		if ((*tag)->tag == NULL) {
1039 			err = got_error_from_errno("strndup");
1040 			goto done;
1041 		}
1042 		s += slen + 1;
1043 		remain -= slen + 1;
1044 		if (remain <= 0) {
1045 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1046 			goto done;
1047 		}
1048 	} else {
1049 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1050 		goto done;
1051 	}
1052 
1053 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
1054 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
1055 		char *p;
1056 		size_t slen;
1057 
1058 		remain -= label_len;
1059 		if (remain <= 0) {
1060 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1061 			goto done;
1062 		}
1063 		s += label_len;
1064 		p = memchr(s, '\n', remain);
1065 		if (p == NULL) {
1066 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1067 			goto done;
1068 		}
1069 		*p = '\0';
1070 		slen = strlen(s);
1071 		err = parse_commit_time(&(*tag)->tagger_time,
1072 		    &(*tag)->tagger_gmtoff, s);
1073 		if (err)
1074 			goto done;
1075 		(*tag)->tagger = strdup(s);
1076 		if ((*tag)->tagger == NULL) {
1077 			err = got_error_from_errno("strdup");
1078 			goto done;
1079 		}
1080 		s += slen + 1;
1081 		remain -= slen + 1;
1082 		if (remain < 0) {
1083 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
1084 			goto done;
1085 		}
1086 	} else {
1087 		/* Some old tags in the Linux git repo have no tagger. */
1088 		(*tag)->tagger = strdup("");
1089 		if ((*tag)->tagger == NULL) {
1090 			err = got_error_from_errno("strdup");
1091 			goto done;
1092 		}
1093 	}
1094 
1095 	(*tag)->tagmsg = strndup(s, remain);
1096 	if ((*tag)->tagmsg == NULL) {
1097 		err = got_error_from_errno("strndup");
1098 		goto done;
1099 	}
1100 done:
1101 	if (err) {
1102 		got_object_tag_close(*tag);
1103 		*tag = NULL;
1104 	}
1105 	return err;
1106 }
1107 
1108 const struct got_error *
1109 got_object_read_tag(struct got_tag_object **tag, int fd,
1110     struct got_object_id *expected_id, size_t expected_size)
1111 {
1112 	const struct got_error *err = NULL;
1113 	struct got_object *obj = NULL;
1114 	size_t len;
1115 	uint8_t *p;
1116 	struct got_inflate_checksum csum;
1117 	struct got_hash ctx;
1118 	struct got_object_id id;
1119 
1120 	got_hash_init(&ctx, GOT_HASH_SHA1);
1121 	memset(&csum, 0, sizeof(csum));
1122 	csum.output_ctx = &ctx;
1123 
1124 	err = got_inflate_to_mem_fd(&p, &len, NULL, &csum,
1125 	    expected_size, fd);
1126 	if (err)
1127 		return err;
1128 
1129 	got_hash_final_object_id(&ctx, &id);
1130 	if (got_object_id_cmp(expected_id, &id) != 0) {
1131 		err = got_error_checksum(expected_id);
1132 		goto done;
1133 	}
1134 
1135 	err = got_object_parse_header(&obj, p, len);
1136 	if (err)
1137 		goto done;
1138 
1139 	if (len < obj->hdrlen + obj->size) {
1140 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
1141 		goto done;
1142 	}
1143 
1144 	/* Skip object header. */
1145 	len -= obj->hdrlen;
1146 	err = got_object_parse_tag(tag, p + obj->hdrlen, len);
1147 done:
1148 	free(p);
1149 	if (obj)
1150 		got_object_close(obj);
1151 	return err;
1152 }