Game of Trees Repos

Blob

Date:: Tue Oct 18 22:12:22 2022 UTC
Message:: allow got_object_parse_tree to reuse entries buffer allocations for speed ok millert@
Actions:: History | Blame | Raw File
1 /*
2  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
3  *
4  * Permission to use, copy, modify, and distribute this software for any
5  * purpose with or without fee is hereby granted, provided that the above
6  * copyright notice and this permission notice appear in all copies.
7  *
8  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
15  */
16 
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
20 #include <sys/uio.h>
21 #include <sys/socket.h>
22 #include <sys/wait.h>
23 #include <sys/mman.h>
24 
25 #include <errno.h>
26 #include <stdio.h>
27 #include <stdlib.h>
28 #include <string.h>
29 #include <stdint.h>
30 #include <zlib.h>
31 #include <ctype.h>
32 #include <limits.h>
33 #include <time.h>
34 #include <unistd.h>
35 
36 #include "got_compat.h"
37 
38 #include "got_error.h"
39 #include "got_object.h"
40 #include "got_repository.h"
41 #include "got_opentemp.h"
42 #include "got_path.h"
43 
44 #include "got_lib_sha1.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_cache.h"
50 #include "got_lib_pack.h"
51 #include "got_lib_repository.h"
52 
53 #ifndef nitems
54 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
55 #endif
56 
57 struct got_object_id *
58 got_object_id_dup(struct got_object_id *id1)
59 {
60 	struct got_object_id *id2;
61 
62 	id2 = malloc(sizeof(*id2));
63 	if (id2 == NULL)
64 		return NULL;
65 	memcpy(id2, id1, sizeof(*id2));
66 	return id2;
67 }
68 
69 int
70 got_object_id_cmp(const struct got_object_id *id1,
71     const struct got_object_id *id2)
72 {
73 	return memcmp(id1->sha1, id2->sha1, SHA1_DIGEST_LENGTH);
74 }
75 
76 const struct got_error *
77 got_object_qid_alloc_partial(struct got_object_qid **qid)
78 {
79 	*qid = malloc(sizeof(**qid));
80 	if (*qid == NULL)
81 		return got_error_from_errno("malloc");
82 
83 	(*qid)->data = NULL;
84 	return NULL;
85 }
86 
87 const struct got_error *
88 got_object_id_str(char **outbuf, struct got_object_id *id)
89 {
90 	static const size_t len = SHA1_DIGEST_STRING_LENGTH;
91 
92 	*outbuf = malloc(len);
93 	if (*outbuf == NULL)
94 		return got_error_from_errno("malloc");
95 
96 	if (got_sha1_digest_to_str(id->sha1, *outbuf, len) == NULL) {
97 		free(*outbuf);
98 		*outbuf = NULL;
99 		return got_error(GOT_ERR_BAD_OBJ_ID_STR);
100 	}
101 
102 	return NULL;
103 }
104 
105 void
106 got_object_close(struct got_object *obj)
107 {
108 	if (obj->refcnt > 0) {
109 		obj->refcnt--;
110 		if (obj->refcnt > 0)
111 			return;
112 	}
113 
114 	if (obj->flags & GOT_OBJ_FLAG_DELTIFIED) {
115 		struct got_delta *delta;
116 		while (!STAILQ_EMPTY(&obj->deltas.entries)) {
117 			delta = STAILQ_FIRST(&obj->deltas.entries);
118 			STAILQ_REMOVE_HEAD(&obj->deltas.entries, entry);
119 			free(delta);
120 		}
121 	}
122 	free(obj);
123 }
124 
125 const struct got_error *
126 got_object_raw_close(struct got_raw_object *obj)
127 {
128 	const struct got_error *err = NULL;
129 
130 	if (obj->refcnt > 0) {
131 		obj->refcnt--;
132 		if (obj->refcnt > 0)
133 			return NULL;
134 	}
135 
136 	if (obj->f == NULL) {
137 		if (obj->fd != -1) {
138 			if (munmap(obj->data, obj->hdrlen + obj->size) == -1)
139 				err = got_error_from_errno("munmap");
140 			if (close(obj->fd) == -1 && err == NULL)
141 				err = got_error_from_errno("close");
142 		} else
143 			free(obj->data);
144 	} else {
145 		if (fclose(obj->f) == EOF && err == NULL)
146 			err = got_error_from_errno("fclose");
147 	}
148 	free(obj);
149 	return err;
150 }
151 
152 void
153 got_object_qid_free(struct got_object_qid *qid)
154 {
155 	free(qid);
156 }
157 
158 void
159 got_object_id_queue_free(struct got_object_id_queue *ids)
160 {
161 	struct got_object_qid *qid;
162 
163 	while (!STAILQ_EMPTY(ids)) {
164 		qid = STAILQ_FIRST(ids);
165 		STAILQ_REMOVE_HEAD(ids, entry);
166 		got_object_qid_free(qid);
167 	}
168 }
169 
170 const struct got_error *
171 got_object_parse_header(struct got_object **obj, char *buf, size_t len)
172 {
173 	const char *obj_labels[] = {
174 		GOT_OBJ_LABEL_COMMIT,
175 		GOT_OBJ_LABEL_TREE,
176 		GOT_OBJ_LABEL_BLOB,
177 		GOT_OBJ_LABEL_TAG,
178 	};
179 	const int obj_types[] = {
180 		GOT_OBJ_TYPE_COMMIT,
181 		GOT_OBJ_TYPE_TREE,
182 		GOT_OBJ_TYPE_BLOB,
183 		GOT_OBJ_TYPE_TAG,
184 	};
185 	int type = 0;
186 	size_t size = 0;
187 	size_t i;
188 	char *end;
189 
190 	*obj = NULL;
191 
192 	end = memchr(buf, '\0', len);
193 	if (end == NULL)
194 		return got_error(GOT_ERR_BAD_OBJ_HDR);
195 
196 	for (i = 0; i < nitems(obj_labels); i++) {
197 		const char *label = obj_labels[i];
198 		size_t label_len = strlen(label);
199 		const char *errstr;
200 
201 		if (len <= label_len || buf + label_len >= end ||
202 		    strncmp(buf, label, label_len) != 0)
203 			continue;
204 
205 		type = obj_types[i];
206 		size = strtonum(buf + label_len, 0, LONG_MAX, &errstr);
207 		if (errstr != NULL)
208 			return got_error(GOT_ERR_BAD_OBJ_HDR);
209 		break;
210 	}
211 
212 	if (type == 0)
213 		return got_error(GOT_ERR_BAD_OBJ_HDR);
214 
215 	*obj = calloc(1, sizeof(**obj));
216 	if (*obj == NULL)
217 		return got_error_from_errno("calloc");
218 	(*obj)->type = type;
219 	(*obj)->hdrlen = end - buf + 1;
220 	(*obj)->size = size;
221 	return NULL;
222 }
223 
224 const struct got_error *
225 got_object_read_header(struct got_object **obj, int fd)
226 {
227 	const struct got_error *err;
228 	struct got_inflate_buf zb;
229 	uint8_t *buf;
230 	const size_t zbsize = 64;
231 	size_t outlen, totlen;
232 	int nbuf = 1;
233 
234 	*obj = NULL;
235 
236 	buf = malloc(zbsize);
237 	if (buf == NULL)
238 		return got_error_from_errno("malloc");
239 	buf[0] = '\0';
240 
241 	err = got_inflate_init(&zb, buf, zbsize, NULL);
242 	if (err)
243 		return err;
244 
245 	totlen = 0;
246 	do {
247 		err = got_inflate_read_fd(&zb, fd, &outlen, NULL);
248 		if (err)
249 			goto done;
250 		if (outlen == 0)
251 			break;
252 		totlen += outlen;
253 		if (memchr(zb.outbuf, '\0', outlen) == NULL) {
254 			uint8_t *newbuf;
255 			nbuf++;
256 			newbuf = recallocarray(buf, nbuf - 1, nbuf, zbsize);
257 			if (newbuf == NULL) {
258 				err = got_error_from_errno("recallocarray");
259 				goto done;
260 			}
261 			buf = newbuf;
262 			zb.outbuf = newbuf + totlen;
263 			zb.outlen = (nbuf * zbsize) - totlen;
264 		}
265 	} while (memchr(zb.outbuf, '\0', outlen) == NULL);
266 
267 	err = got_object_parse_header(obj, buf, totlen);
268 done:
269 	free(buf);
270 	got_inflate_end(&zb);
271 	return err;
272 }
273 
274 struct got_commit_object *
275 got_object_commit_alloc_partial(void)
276 {
277 	struct got_commit_object *commit;
278 
279 	commit = calloc(1, sizeof(*commit));
280 	if (commit == NULL)
281 		return NULL;
282 	commit->tree_id = malloc(sizeof(*commit->tree_id));
283 	if (commit->tree_id == NULL) {
284 		free(commit);
285 		return NULL;
286 	}
287 
288 	STAILQ_INIT(&commit->parent_ids);
289 
290 	return commit;
291 }
292 
293 const struct got_error *
294 got_object_commit_add_parent(struct got_commit_object *commit,
295     const char *id_str)
296 {
297 	const struct got_error *err = NULL;
298 	struct got_object_qid *qid;
299 
300 	err = got_object_qid_alloc_partial(&qid);
301 	if (err)
302 		return err;
303 
304 	if (!got_parse_sha1_digest(qid->id.sha1, id_str)) {
305 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
306 		got_object_qid_free(qid);
307 		return err;
308 	}
309 
310 	STAILQ_INSERT_TAIL(&commit->parent_ids, qid, entry);
311 	commit->nparents++;
312 
313 	return NULL;
314 }
315 
316 static const struct got_error *
317 parse_gmtoff(time_t *gmtoff, const char *tzstr)
318 {
319 	int sign = 1;
320 	const char *p = tzstr;
321 	time_t h, m;
322 
323 	*gmtoff = 0;
324 
325 	if (*p == '-')
326 		sign = -1;
327 	else if (*p != '+')
328 		return got_error(GOT_ERR_BAD_OBJ_DATA);
329 	p++;
330 	if (!isdigit(*p) && !isdigit(*(p + 1)))
331 		return got_error(GOT_ERR_BAD_OBJ_DATA);
332 	h = (((*p - '0') * 10) + (*(p + 1) - '0'));
333 
334 	p += 2;
335 	if (!isdigit(*p) && !isdigit(*(p + 1)))
336 		return got_error(GOT_ERR_BAD_OBJ_DATA);
337 	m = ((*p - '0') * 10) + (*(p + 1) - '0');
338 
339 	*gmtoff = (h * 60 * 60 + m * 60) * sign;
340 	return NULL;
341 }
342 
343 static const struct got_error *
344 parse_commit_time(time_t *time, time_t *gmtoff, char *committer)
345 {
346 	const struct got_error *err = NULL;
347 	const char *errstr;
348 	char *space, *tzstr;
349 
350 	/* Parse and strip off trailing timezone indicator string. */
351 	space = strrchr(committer, ' ');
352 	if (space == NULL)
353 		return got_error(GOT_ERR_BAD_OBJ_DATA);
354 	tzstr = strdup(space + 1);
355 	if (tzstr == NULL)
356 		return got_error_from_errno("strdup");
357 	err = parse_gmtoff(gmtoff, tzstr);
358 	free(tzstr);
359 	if (err) {
360 		if (err->code != GOT_ERR_BAD_OBJ_DATA)
361 			return err;
362 		/* Old versions of Git omitted the timestamp. */
363 		*time = 0;
364 		*gmtoff = 0;
365 		return NULL;
366 	}
367 	*space = '\0';
368 
369 	/* Timestamp is separated from committer name + email by space. */
370 	space = strrchr(committer, ' ');
371 	if (space == NULL)
372 		return got_error(GOT_ERR_BAD_OBJ_DATA);
373 
374 	/* Timestamp parsed here is expressed as UNIX timestamp (UTC). */
375 	*time = strtonum(space + 1, 0, INT64_MAX, &errstr);
376 	if (errstr)
377 		return got_error(GOT_ERR_BAD_OBJ_DATA);
378 
379 	/* Strip off parsed time information, leaving just author and email. */
380 	*space = '\0';
381 
382 	return NULL;
383 }
384 
385 void
386 got_object_commit_close(struct got_commit_object *commit)
387 {
388 	if (commit->refcnt > 0) {
389 		commit->refcnt--;
390 		if (commit->refcnt > 0)
391 			return;
392 	}
393 
394 	got_object_id_queue_free(&commit->parent_ids);
395 	free(commit->tree_id);
396 	free(commit->author);
397 	free(commit->committer);
398 	free(commit->logmsg);
399 	free(commit);
400 }
401 
402 struct got_object_id *
403 got_object_commit_get_tree_id(struct got_commit_object *commit)
404 {
405 	return commit->tree_id;
406 }
407 
408 int
409 got_object_commit_get_nparents(struct got_commit_object *commit)
410 {
411 	return commit->nparents;
412 }
413 
414 const struct got_object_id_queue *
415 got_object_commit_get_parent_ids(struct got_commit_object *commit)
416 {
417 	return &commit->parent_ids;
418 }
419 
420 const char *
421 got_object_commit_get_author(struct got_commit_object *commit)
422 {
423 	return commit->author;
424 }
425 
426 time_t
427 got_object_commit_get_author_time(struct got_commit_object *commit)
428 {
429 	return commit->author_time;
430 }
431 
432 time_t got_object_commit_get_author_gmtoff(struct got_commit_object *commit)
433 {
434 	return commit->author_gmtoff;
435 }
436 
437 const char *
438 got_object_commit_get_committer(struct got_commit_object *commit)
439 {
440 	return commit->committer;
441 }
442 
443 time_t
444 got_object_commit_get_committer_time(struct got_commit_object *commit)
445 {
446 	return commit->committer_time;
447 }
448 
449 time_t
450 got_object_commit_get_committer_gmtoff(struct got_commit_object *commit)
451 {
452 	return commit->committer_gmtoff;
453 }
454 
455 const struct got_error *
456 got_object_commit_get_logmsg(char **logmsg, struct got_commit_object *commit)
457 {
458 	const struct got_error *err = NULL;
459 	const char *src;
460 	char *dst;
461 	size_t len;
462 
463 	len = strlen(commit->logmsg);
464 	*logmsg = malloc(len + 2); /* leave room for a trailing \n and \0 */
465 	if (*logmsg == NULL)
466 		return got_error_from_errno("malloc");
467 
468 	/*
469 	 * Strip out unusual headers. Headers are separated from the commit
470 	 * message body by a single empty line.
471 	 */
472 	src = commit->logmsg;
473 	dst = *logmsg;
474 	while (*src != '\0' && *src != '\n') {
475 		int copy_header = 1, eol = 0;
476 		if (strncmp(src, GOT_COMMIT_LABEL_TREE,
477 		    strlen(GOT_COMMIT_LABEL_TREE)) != 0 &&
478 		    strncmp(src, GOT_COMMIT_LABEL_AUTHOR,
479 		    strlen(GOT_COMMIT_LABEL_AUTHOR)) != 0 &&
480 		    strncmp(src, GOT_COMMIT_LABEL_PARENT,
481 		    strlen(GOT_COMMIT_LABEL_PARENT)) != 0 &&
482 		    strncmp(src, GOT_COMMIT_LABEL_COMMITTER,
483 		    strlen(GOT_COMMIT_LABEL_COMMITTER)) != 0)
484 			copy_header = 0;
485 
486 		while (*src != '\0' && !eol) {
487 			if (copy_header) {
488 				*dst = *src;
489 				dst++;
490 			}
491 			if (*src == '\n')
492 				eol = 1;
493 			src++;
494 		}
495 	}
496 	*dst = '\0';
497 
498 	if (strlcat(*logmsg, src, len + 1) >= len + 1) {
499 		err = got_error(GOT_ERR_NO_SPACE);
500 		goto done;
501 	}
502 
503 	/* Trim redundant trailing whitespace. */
504 	len = strlen(*logmsg);
505 	while (len > 1 && isspace((unsigned char)(*logmsg)[len - 2]) &&
506 	    isspace((unsigned char)(*logmsg)[len - 1])) {
507 		(*logmsg)[len - 1] = '\0';
508 		len--;
509 	}
510 
511 	/* Append a trailing newline if missing. */
512 	if (len > 0 && (*logmsg)[len - 1] != '\n') {
513 		(*logmsg)[len] = '\n';
514 		(*logmsg)[len + 1] = '\0';
515 	}
516 done:
517 	if (err) {
518 		free(*logmsg);
519 		*logmsg = NULL;
520 	}
521 	return err;
522 }
523 
524 const char *
525 got_object_commit_get_logmsg_raw(struct got_commit_object *commit)
526 {
527 	return commit->logmsg;
528 }
529 
530 const struct got_error *
531 got_object_parse_commit(struct got_commit_object **commit, char *buf,
532     size_t len)
533 {
534 	const struct got_error *err = NULL;
535 	char *s = buf;
536 	size_t label_len;
537 	ssize_t remain = (ssize_t)len;
538 
539 	if (remain == 0)
540 		return got_error(GOT_ERR_BAD_OBJ_DATA);
541 
542 	*commit = got_object_commit_alloc_partial();
543 	if (*commit == NULL)
544 		return got_error_from_errno("got_object_commit_alloc_partial");
545 
546 	label_len = strlen(GOT_COMMIT_LABEL_TREE);
547 	if (strncmp(s, GOT_COMMIT_LABEL_TREE, label_len) == 0) {
548 		remain -= label_len;
549 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
550 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
551 			goto done;
552 		}
553 		s += label_len;
554 		if (!got_parse_sha1_digest((*commit)->tree_id->sha1, s)) {
555 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
556 			goto done;
557 		}
558 		remain -= SHA1_DIGEST_STRING_LENGTH;
559 		s += SHA1_DIGEST_STRING_LENGTH;
560 	} else {
561 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
562 		goto done;
563 	}
564 
565 	label_len = strlen(GOT_COMMIT_LABEL_PARENT);
566 	while (strncmp(s, GOT_COMMIT_LABEL_PARENT, label_len) == 0) {
567 		remain -= label_len;
568 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
569 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
570 			goto done;
571 		}
572 		s += label_len;
573 		err = got_object_commit_add_parent(*commit, s);
574 		if (err)
575 			goto done;
576 
577 		remain -= SHA1_DIGEST_STRING_LENGTH;
578 		s += SHA1_DIGEST_STRING_LENGTH;
579 	}
580 
581 	label_len = strlen(GOT_COMMIT_LABEL_AUTHOR);
582 	if (strncmp(s, GOT_COMMIT_LABEL_AUTHOR, label_len) == 0) {
583 		char *p;
584 		size_t slen;
585 
586 		remain -= label_len;
587 		if (remain <= 0) {
588 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
589 			goto done;
590 		}
591 		s += label_len;
592 		p = memchr(s, '\n', remain);
593 		if (p == NULL) {
594 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
595 			goto done;
596 		}
597 		*p = '\0';
598 		slen = strlen(s);
599 		err = parse_commit_time(&(*commit)->author_time,
600 		    &(*commit)->author_gmtoff, s);
601 		if (err)
602 			goto done;
603 		(*commit)->author = strdup(s);
604 		if ((*commit)->author == NULL) {
605 			err = got_error_from_errno("strdup");
606 			goto done;
607 		}
608 		s += slen + 1;
609 		remain -= slen + 1;
610 	}
611 
612 	label_len = strlen(GOT_COMMIT_LABEL_COMMITTER);
613 	if (strncmp(s, GOT_COMMIT_LABEL_COMMITTER, label_len) == 0) {
614 		char *p;
615 		size_t slen;
616 
617 		remain -= label_len;
618 		if (remain <= 0) {
619 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
620 			goto done;
621 		}
622 		s += label_len;
623 		p = memchr(s, '\n', remain);
624 		if (p == NULL) {
625 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
626 			goto done;
627 		}
628 		*p = '\0';
629 		slen = strlen(s);
630 		err = parse_commit_time(&(*commit)->committer_time,
631 		    &(*commit)->committer_gmtoff, s);
632 		if (err)
633 			goto done;
634 		(*commit)->committer = strdup(s);
635 		if ((*commit)->committer == NULL) {
636 			err = got_error_from_errno("strdup");
637 			goto done;
638 		}
639 		s += slen + 1;
640 		remain -= slen + 1;
641 	}
642 
643 	(*commit)->logmsg = strndup(s, remain);
644 	if ((*commit)->logmsg == NULL) {
645 		err = got_error_from_errno("strndup");
646 		goto done;
647 	}
648 done:
649 	if (err) {
650 		got_object_commit_close(*commit);
651 		*commit = NULL;
652 	}
653 	return err;
654 }
655 
656 void
657 got_object_tree_close(struct got_tree_object *tree)
658 {
659 	if (tree->refcnt > 0) {
660 		tree->refcnt--;
661 		if (tree->refcnt > 0)
662 			return;
663 	}
664 
665 	free(tree->entries);
666 	free(tree);
667 }
668 
669 static const struct got_error *
670 parse_tree_entry(struct got_parsed_tree_entry *pte, size_t *elen, char *buf,
671     size_t maxlen)
672 {
673 	char *p, *space;
674 
675 	*elen = 0;
676 
677 	*elen = strnlen(buf, maxlen) + 1;
678 	if (*elen > maxlen)
679 		return got_error(GOT_ERR_BAD_OBJ_DATA);
680 
681 	space = memchr(buf, ' ', *elen);
682 	if (space == NULL || space <= buf)
683 		return got_error(GOT_ERR_BAD_OBJ_DATA);
684 
685 	pte->mode = 0;
686 	p = buf;
687 	while (p < space) {
688 		if (*p < '0' && *p > '7')
689 			return got_error(GOT_ERR_BAD_OBJ_DATA);
690 		pte->mode <<= 3;
691 		pte->mode |= *p - '0';
692 		p++;
693 	}
694 
695 	if (*elen > maxlen || maxlen - *elen < SHA1_DIGEST_LENGTH)
696 		return got_error(GOT_ERR_BAD_OBJ_DATA);
697 
698 	pte->name = space + 1;
699 	pte->namelen = strlen(pte->name);
700 	buf += *elen;
701 	pte->id = buf;
702 	*elen += SHA1_DIGEST_LENGTH;
703 	return NULL;
704 }
705 
706 static int
707 pte_cmp(const void *pa, const void *pb)
708 {
709 	const struct got_parsed_tree_entry *a = pa, *b = pb;
710 
711 	return got_path_cmp(a->name, b->name, a->namelen, b->namelen);
712 }
713 
714 const struct got_error *
715 got_object_parse_tree(struct got_parsed_tree_entry **entries, size_t *nentries,
716     size_t *nentries_alloc, uint8_t *buf, size_t len)
717 {
718 	const struct got_error *err = NULL;
719 	size_t remain = len;
720 	const size_t nalloc = 16;
721 	struct got_parsed_tree_entry *pte;
722 	int i;
723 
724 	*nentries = 0;
725 	if (remain == 0)
726 		return NULL; /* tree is empty */
727 
728 	while (remain > 0) {
729 		size_t elen;
730 
731 		if (*nentries >= *nentries_alloc) {
732 			pte = recallocarray(*entries, *nentries_alloc,
733 			    *nentries_alloc + nalloc, sizeof(**entries));
734 			if (pte == NULL) {
735 				err = got_error_from_errno("recallocarray");
736 				goto done;
737 			}
738 			*entries = pte;
739 			*nentries_alloc += nalloc;
740 		}
741 
742 		pte = &(*entries)[*nentries];
743 		err = parse_tree_entry(pte, &elen, buf, remain);
744 		if (err)
745 			goto done;
746 		buf += elen;
747 		remain -= elen;
748 		(*nentries)++;
749 	}
750 
751 	if (remain != 0) {
752 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
753 		goto done;
754 	}
755 
756 	if (*nentries > 1) {
757 		mergesort(*entries, *nentries, sizeof(**entries), pte_cmp);
758 
759 		for (i = 0; i < *nentries - 1; i++) {
760 			struct got_parsed_tree_entry *prev = &(*entries)[i];
761 			pte = &(*entries)[i + 1];
762 			if (got_path_cmp(prev->name, pte->name,
763 			    prev->namelen, pte->namelen) == 0) {
764 				err = got_error(GOT_ERR_TREE_DUP_ENTRY);
765 				break;
766 			}
767 		}
768 	}
769 done:
770 	if (err)
771 		*nentries = 0;
772 	return err;
773 }
774 
775 void
776 got_object_tag_close(struct got_tag_object *tag)
777 {
778 	if (tag->refcnt > 0) {
779 		tag->refcnt--;
780 		if (tag->refcnt > 0)
781 			return;
782 	}
783 
784 	free(tag->tag);
785 	free(tag->tagger);
786 	free(tag->tagmsg);
787 	free(tag);
788 }
789 
790 const struct got_error *
791 got_object_parse_tag(struct got_tag_object **tag, uint8_t *buf, size_t len)
792 {
793 	const struct got_error *err = NULL;
794 	size_t remain = len;
795 	char *s = buf;
796 	size_t label_len;
797 
798 	if (remain == 0)
799 		return got_error(GOT_ERR_BAD_OBJ_DATA);
800 
801 	*tag = calloc(1, sizeof(**tag));
802 	if (*tag == NULL)
803 		return got_error_from_errno("calloc");
804 
805 	label_len = strlen(GOT_TAG_LABEL_OBJECT);
806 	if (strncmp(s, GOT_TAG_LABEL_OBJECT, label_len) == 0) {
807 		remain -= label_len;
808 		if (remain < SHA1_DIGEST_STRING_LENGTH) {
809 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
810 			goto done;
811 		}
812 		s += label_len;
813 		if (!got_parse_sha1_digest((*tag)->id.sha1, s)) {
814 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
815 			goto done;
816 		}
817 		remain -= SHA1_DIGEST_STRING_LENGTH;
818 		s += SHA1_DIGEST_STRING_LENGTH;
819 	} else {
820 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
821 		goto done;
822 	}
823 
824 	if (remain <= 0) {
825 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
826 		goto done;
827 	}
828 
829 	label_len = strlen(GOT_TAG_LABEL_TYPE);
830 	if (strncmp(s, GOT_TAG_LABEL_TYPE, label_len) == 0) {
831 		remain -= label_len;
832 		if (remain <= 0) {
833 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
834 			goto done;
835 		}
836 		s += label_len;
837 		if (strncmp(s, GOT_OBJ_LABEL_COMMIT,
838 		    strlen(GOT_OBJ_LABEL_COMMIT)) == 0) {
839 			(*tag)->obj_type = GOT_OBJ_TYPE_COMMIT;
840 			label_len = strlen(GOT_OBJ_LABEL_COMMIT);
841 			s += label_len;
842 			remain -= label_len;
843 		} else if (strncmp(s, GOT_OBJ_LABEL_TREE,
844 		    strlen(GOT_OBJ_LABEL_TREE)) == 0) {
845 			(*tag)->obj_type = GOT_OBJ_TYPE_TREE;
846 			label_len = strlen(GOT_OBJ_LABEL_TREE);
847 			s += label_len;
848 			remain -= label_len;
849 		} else if (strncmp(s, GOT_OBJ_LABEL_BLOB,
850 		    strlen(GOT_OBJ_LABEL_BLOB)) == 0) {
851 			(*tag)->obj_type = GOT_OBJ_TYPE_BLOB;
852 			label_len = strlen(GOT_OBJ_LABEL_BLOB);
853 			s += label_len;
854 			remain -= label_len;
855 		} else if (strncmp(s, GOT_OBJ_LABEL_TAG,
856 		    strlen(GOT_OBJ_LABEL_TAG)) == 0) {
857 			(*tag)->obj_type = GOT_OBJ_TYPE_TAG;
858 			label_len = strlen(GOT_OBJ_LABEL_TAG);
859 			s += label_len;
860 			remain -= label_len;
861 		} else {
862 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
863 			goto done;
864 		}
865 
866 		if (remain <= 0 || *s != '\n') {
867 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
868 			goto done;
869 		}
870 		s++;
871 		remain--;
872 		if (remain <= 0) {
873 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
874 			goto done;
875 		}
876 	} else {
877 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
878 		goto done;
879 	}
880 
881 	label_len = strlen(GOT_TAG_LABEL_TAG);
882 	if (strncmp(s, GOT_TAG_LABEL_TAG, label_len) == 0) {
883 		char *p;
884 		size_t slen;
885 		remain -= label_len;
886 		if (remain <= 0) {
887 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
888 			goto done;
889 		}
890 		s += label_len;
891 		p = memchr(s, '\n', remain);
892 		if (p == NULL) {
893 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
894 			goto done;
895 		}
896 		*p = '\0';
897 		slen = strlen(s);
898 		(*tag)->tag = strndup(s, slen);
899 		if ((*tag)->tag == NULL) {
900 			err = got_error_from_errno("strndup");
901 			goto done;
902 		}
903 		s += slen + 1;
904 		remain -= slen + 1;
905 		if (remain <= 0) {
906 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
907 			goto done;
908 		}
909 	} else {
910 		err = got_error(GOT_ERR_BAD_OBJ_DATA);
911 		goto done;
912 	}
913 
914 	label_len = strlen(GOT_TAG_LABEL_TAGGER);
915 	if (strncmp(s, GOT_TAG_LABEL_TAGGER, label_len) == 0) {
916 		char *p;
917 		size_t slen;
918 
919 		remain -= label_len;
920 		if (remain <= 0) {
921 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
922 			goto done;
923 		}
924 		s += label_len;
925 		p = memchr(s, '\n', remain);
926 		if (p == NULL) {
927 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
928 			goto done;
929 		}
930 		*p = '\0';
931 		slen = strlen(s);
932 		err = parse_commit_time(&(*tag)->tagger_time,
933 		    &(*tag)->tagger_gmtoff, s);
934 		if (err)
935 			goto done;
936 		(*tag)->tagger = strdup(s);
937 		if ((*tag)->tagger == NULL) {
938 			err = got_error_from_errno("strdup");
939 			goto done;
940 		}
941 		s += slen + 1;
942 		remain -= slen + 1;
943 		if (remain < 0) {
944 			err = got_error(GOT_ERR_BAD_OBJ_DATA);
945 			goto done;
946 		}
947 	} else {
948 		/* Some old tags in the Linux git repo have no tagger. */
949 		(*tag)->tagger = strdup("");
950 		if ((*tag)->tagger == NULL) {
951 			err = got_error_from_errno("strdup");
952 			goto done;
953 		}
954 	}
955 
956 	(*tag)->tagmsg = strndup(s, remain);
957 	if ((*tag)->tagmsg == NULL) {
958 		err = got_error_from_errno("strndup");
959 		goto done;
960 	}
961 done:
962 	if (err) {
963 		got_object_tag_close(*tag);
964 		*tag = NULL;
965 	}
966 	return err;
967 }
968 
969 const struct got_error *
970 got_read_file_to_mem(uint8_t **outbuf, size_t *outlen, FILE *f)
971 {
972 	const struct got_error *err = NULL;
973 	static const size_t blocksize = 512;
974 	size_t n, total, remain;
975 	uint8_t *buf;
976 
977 	*outbuf = NULL;
978 	*outlen = 0;
979 
980 	buf = malloc(blocksize);
981 	if (buf == NULL)
982 		return got_error_from_errno("malloc");
983 
984 	remain = blocksize;
985 	total = 0;
986 	for (;;) {
987 		if (remain == 0) {
988 			uint8_t *newbuf;
989 			newbuf = reallocarray(buf, 1, total + blocksize);
990 			if (newbuf == NULL) {
991 				err = got_error_from_errno("reallocarray");
992 				goto done;
993 			}
994 			buf = newbuf;
995 			remain += blocksize;
996 		}
997 		n = fread(buf + total, 1, remain, f);
998 		if (n == 0) {
999 			if (ferror(f)) {
1000 				err = got_ferror(f, GOT_ERR_IO);
1001 				goto done;
1002 			}
1003 			break; /* EOF */
1004 		}
1005 		remain -= n;
1006 		total += n;
1007 	};
1008 
1009 done:
1010 	if (err == NULL) {
1011 		*outbuf = buf;
1012 		*outlen = total;
1013 	} else
1014 		free(buf);
1015 	return err;
1016 }