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/uio.h>
20 #include <sys/mman.h>
21 
22 #include <fcntl.h>
23 #include <errno.h>
24 #include <stdio.h>
25 #include <stdint.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <limits.h>
29 #include <unistd.h>
30 #include <zlib.h>
31 
32 #include "got_error.h"
33 #include "got_object.h"
34 #include "got_opentemp.h"
35 #include "got_path.h"
36 
37 #include "got_lib_sha1.h"
38 #include "got_lib_delta.h"
39 #include "got_lib_delta_cache.h"
40 #include "got_lib_inflate.h"
41 #include "got_lib_object.h"
42 #include "got_lib_object_parse.h"
43 #include "got_lib_privsep.h"
44 #include "got_lib_pack.h"
45 
46 #ifndef nitems
47 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
48 #endif
49 
50 #ifndef MIN
51 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
52 #endif
53 
54 static const struct got_error *
55 verify_fanout_table(uint32_t *fanout_table)
56 {
57 	int i;
58 
59 	for (i = 0; i < 0xff - 1; i++) {
60 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
61 			return got_error(GOT_ERR_BAD_PACKIDX);
62 	}
63 
64 	return NULL;
65 }
66 
67 const struct got_error *
68 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
69 {
70 	const struct got_error *err = NULL;
71 	struct got_packidx_v2_hdr *h;
72 	SHA1_CTX ctx;
73 	uint8_t sha1[SHA1_DIGEST_LENGTH];
74 	size_t nobj, len_fanout, len_ids, offset, remain;
75 	ssize_t n;
76 	int i;
77 
78 	SHA1Init(&ctx);
79 
80 	h = &p->hdr;
81 	offset = 0;
82 	remain = p->len;
83 
84 	if (remain < sizeof(*h->magic)) {
85 		err = got_error(GOT_ERR_BAD_PACKIDX);
86 		goto done;
87 	}
88 	if (p->map)
89 		h->magic = (uint32_t *)(p->map + offset);
90 	else {
91 		h->magic = malloc(sizeof(*h->magic));
92 		if (h->magic == NULL) {
93 			err = got_error_from_errno("malloc");
94 			goto done;
95 		}
96 		n = read(p->fd, h->magic, sizeof(*h->magic));
97 		if (n < 0) {
98 			err = got_error_from_errno("read");
99 			goto done;
100 		} else if (n != sizeof(*h->magic)) {
101 			err = got_error(GOT_ERR_BAD_PACKIDX);
102 			goto done;
103 		}
104 	}
105 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
106 		err = got_error(GOT_ERR_BAD_PACKIDX);
107 		goto done;
108 	}
109 	offset += sizeof(*h->magic);
110 	remain -= sizeof(*h->magic);
111 
112 	if (verify)
113 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
114 
115 	if (remain < sizeof(*h->version)) {
116 		err = got_error(GOT_ERR_BAD_PACKIDX);
117 		goto done;
118 	}
119 	if (p->map)
120 		h->version = (uint32_t *)(p->map + offset);
121 	else {
122 		h->version = malloc(sizeof(*h->version));
123 		if (h->version == NULL) {
124 			err = got_error_from_errno("malloc");
125 			goto done;
126 		}
127 		n = read(p->fd, h->version, sizeof(*h->version));
128 		if (n < 0) {
129 			err = got_error_from_errno("read");
130 			goto done;
131 		} else if (n != sizeof(*h->version)) {
132 			err = got_error(GOT_ERR_BAD_PACKIDX);
133 			goto done;
134 		}
135 	}
136 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
137 		err = got_error(GOT_ERR_BAD_PACKIDX);
138 		goto done;
139 	}
140 	offset += sizeof(*h->version);
141 	remain -= sizeof(*h->version);
142 
143 	if (verify)
144 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
145 
146 	len_fanout =
147 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
148 	if (remain < len_fanout) {
149 		err = got_error(GOT_ERR_BAD_PACKIDX);
150 		goto done;
151 	}
152 	if (p->map)
153 		h->fanout_table = (uint32_t *)(p->map + offset);
154 	else {
155 		h->fanout_table = malloc(len_fanout);
156 		if (h->fanout_table == NULL) {
157 			err = got_error_from_errno("malloc");
158 			goto done;
159 		}
160 		n = read(p->fd, h->fanout_table, len_fanout);
161 		if (n < 0) {
162 			err = got_error_from_errno("read");
163 			goto done;
164 		} else if (n != len_fanout) {
165 			err = got_error(GOT_ERR_BAD_PACKIDX);
166 			goto done;
167 		}
168 	}
169 	err = verify_fanout_table(h->fanout_table);
170 	if (err)
171 		goto done;
172 	if (verify)
173 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
174 	offset += len_fanout;
175 	remain -= len_fanout;
176 
177 	nobj = be32toh(h->fanout_table[0xff]);
178 	len_ids = nobj * sizeof(*h->sorted_ids);
179 	if (len_ids <= nobj || len_ids > remain) {
180 		err = got_error(GOT_ERR_BAD_PACKIDX);
181 		goto done;
182 	}
183 	if (p->map)
184 		h->sorted_ids =
185 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
186 	else {
187 		h->sorted_ids = malloc(len_ids);
188 		if (h->sorted_ids == NULL) {
189 			err = got_error(GOT_ERR_BAD_PACKIDX);
190 			goto done;
191 		}
192 		n = read(p->fd, h->sorted_ids, len_ids);
193 		if (n < 0)
194 			err = got_error_from_errno("read");
195 		else if (n != len_ids) {
196 			err = got_error(GOT_ERR_BAD_PACKIDX);
197 			goto done;
198 		}
199 	}
200 	if (verify)
201 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
202 	offset += len_ids;
203 	remain -= len_ids;
204 
205 	if (remain < nobj * sizeof(*h->crc32)) {
206 		err = got_error(GOT_ERR_BAD_PACKIDX);
207 		goto done;
208 	}
209 	if (p->map)
210 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
211 	else {
212 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
213 		if (h->crc32 == NULL) {
214 			err = got_error_from_errno("malloc");
215 			goto done;
216 		}
217 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
218 		if (n < 0)
219 			err = got_error_from_errno("read");
220 		else if (n != nobj * sizeof(*h->crc32)) {
221 			err = got_error(GOT_ERR_BAD_PACKIDX);
222 			goto done;
223 		}
224 	}
225 	if (verify)
226 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
227 	remain -= nobj * sizeof(*h->crc32);
228 	offset += nobj * sizeof(*h->crc32);
229 
230 	if (remain < nobj * sizeof(*h->offsets)) {
231 		err = got_error(GOT_ERR_BAD_PACKIDX);
232 		goto done;
233 	}
234 	if (p->map)
235 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
236 	else {
237 		h->offsets = malloc(nobj * sizeof(*h->offsets));
238 		if (h->offsets == NULL) {
239 			err = got_error_from_errno("malloc");
240 			goto done;
241 		}
242 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
243 		if (n < 0)
244 			err = got_error_from_errno("read");
245 		else if (n != nobj * sizeof(*h->offsets)) {
246 			err = got_error(GOT_ERR_BAD_PACKIDX);
247 			goto done;
248 		}
249 	}
250 	if (verify)
251 		SHA1Update(&ctx, (uint8_t *)h->offsets,
252 		    nobj * sizeof(*h->offsets));
253 	remain -= nobj * sizeof(*h->offsets);
254 	offset += nobj * sizeof(*h->offsets);
255 
256 	/* Large file offsets are contained only in files > 2GB. */
257 	if (verify || packfile_size > 0x7fffffff) {
258 		for (i = 0; i < nobj; i++) {
259 			uint32_t o = h->offsets[i];
260 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
261 				p->nlargeobj++;
262 		}
263 	}
264 	if (p->nlargeobj == 0)
265 		goto checksum;
266 	else if (packfile_size <= 0x7fffffff) {
267 		err = got_error(GOT_ERR_BAD_PACKIDX);
268 		goto done;
269 	}
270 
271 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
272 		err = got_error(GOT_ERR_BAD_PACKIDX);
273 		goto done;
274 	}
275 	if (p->map)
276 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
277 	else {
278 		h->large_offsets = malloc(p->nlargeobj *
279 		    sizeof(*h->large_offsets));
280 		if (h->large_offsets == NULL) {
281 			err = got_error_from_errno("malloc");
282 			goto done;
283 		}
284 		n = read(p->fd, h->large_offsets,
285 		    p->nlargeobj * sizeof(*h->large_offsets));
286 		if (n < 0)
287 			err = got_error_from_errno("read");
288 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
289 			err = got_error(GOT_ERR_BAD_PACKIDX);
290 			goto done;
291 		}
292 	}
293 	if (verify)
294 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
295 		    p->nlargeobj * sizeof(*h->large_offsets));
296 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
297 	offset += p->nlargeobj * sizeof(*h->large_offsets);
298 
299 checksum:
300 	if (remain < sizeof(*h->trailer)) {
301 		err = got_error(GOT_ERR_BAD_PACKIDX);
302 		goto done;
303 	}
304 	if (p->map)
305 		h->trailer =
306 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
307 	else {
308 		h->trailer = malloc(sizeof(*h->trailer));
309 		if (h->trailer == NULL) {
310 			err = got_error_from_errno("malloc");
311 			goto done;
312 		}
313 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
314 		if (n < 0)
315 			err = got_error_from_errno("read");
316 		else if (n != sizeof(*h->trailer)) {
317 			err = got_error(GOT_ERR_BAD_PACKIDX);
318 			goto done;
319 		}
320 	}
321 	if (verify) {
322 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
323 		SHA1Final(sha1, &ctx);
324 		if (memcmp(h->trailer->packidx_sha1, sha1,
325 		    SHA1_DIGEST_LENGTH) != 0)
326 			err = got_error(GOT_ERR_PACKIDX_CSUM);
327 	}
328 done:
329 	return err;
330 }
331 
332 const struct got_error *
333 got_packidx_open(struct got_packidx **packidx,
334     int dir_fd, const char *relpath, int verify)
335 {
336 	const struct got_error *err = NULL;
337 	struct got_packidx *p = NULL;
338 	char *pack_relpath;
339 	struct stat idx_sb, pack_sb;
340 
341 	*packidx = NULL;
342 
343 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
344 	if (err)
345 		return err;
346 
347 	/*
348 	 * Ensure that a corresponding pack file exists.
349 	 * Some Git repositories have this problem. Git seems to ignore
350 	 * the existence of lonely pack index files but we do not.
351 	 */
352 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
353 		if (errno == ENOENT) {
354 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
355 			    "%s", relpath);
356 		} else
357 			err = got_error_from_errno2("fstatat", pack_relpath);
358 		goto done;
359 	}
360 
361 	p = calloc(1, sizeof(*p));
362 	if (p == NULL) {
363 		err = got_error_from_errno("calloc");
364 		goto done;
365 	}
366 
367 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
368 	if (p->fd == -1) {
369 		err = got_error_from_errno2("openat", relpath);
370 		goto done;
371 	}
372 
373 	if (fstat(p->fd, &idx_sb) != 0) {
374 		err = got_error_from_errno2("fstat", relpath);
375 		goto done;
376 	}
377 	p->len = idx_sb.st_size;
378 	if (p->len < sizeof(p->hdr)) {
379 		err = got_error(GOT_ERR_BAD_PACKIDX);
380 		goto done;
381 	}
382 
383 	p->path_packidx = strdup(relpath);
384 	if (p->path_packidx == NULL) {
385 		err = got_error_from_errno("strdup");
386 		goto done;
387 	}
388 
389 #ifndef GOT_PACK_NO_MMAP
390 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
391 	if (p->map == MAP_FAILED) {
392 		if (errno != ENOMEM) {
393 			err = got_error_from_errno("mmap");
394 			goto done;
395 		}
396 		p->map = NULL; /* fall back to read(2) */
397 	}
398 #endif
399 
400 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
401 done:
402 	if (err) {
403 		if (p)
404 			got_packidx_close(p);
405 	} else
406 		*packidx = p;
407 	free(pack_relpath);
408 	return err;
409 }
410 
411 const struct got_error *
412 got_packidx_close(struct got_packidx *packidx)
413 {
414 	const struct got_error *err = NULL;
415 
416 	free(packidx->path_packidx);
417 	if (packidx->map) {
418 		if (munmap(packidx->map, packidx->len) == -1)
419 			err = got_error_from_errno("munmap");
420 	} else {
421 		free(packidx->hdr.magic);
422 		free(packidx->hdr.version);
423 		free(packidx->hdr.fanout_table);
424 		free(packidx->hdr.sorted_ids);
425 		free(packidx->hdr.crc32);
426 		free(packidx->hdr.offsets);
427 		free(packidx->hdr.large_offsets);
428 		free(packidx->hdr.trailer);
429 	}
430 	if (close(packidx->fd) == -1 && err == NULL)
431 		err = got_error_from_errno("close");
432 	free(packidx->sorted_offsets);
433 	free(packidx->sorted_large_offsets);
434 	free(packidx);
435 
436 	return err;
437 }
438 
439 const struct got_error *
440 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
441 {
442 	size_t size;
443 
444 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
445 	size = strlen(path_packidx) + 2;
446 	if (size < GOT_PACKFILE_NAMELEN + 1)
447 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
448 
449 	*path_packfile = malloc(size);
450 	if (*path_packfile == NULL)
451 		return got_error_from_errno("malloc");
452 
453 	/* Copy up to and excluding ".idx". */
454 	if (strlcpy(*path_packfile, path_packidx,
455 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
456 		return got_error(GOT_ERR_NO_SPACE);
457 
458 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
459 		return got_error(GOT_ERR_NO_SPACE);
460 
461 	return NULL;
462 }
463 
464 off_t
465 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
466 {
467 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
468 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
469 		uint64_t loffset;
470 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
471 		if (idx < 0 || idx >= packidx->nlargeobj ||
472 		    packidx->hdr.large_offsets == NULL)
473 			return -1;
474 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
475 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
476 	}
477 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
478 }
479 
480 int
481 got_packidx_get_object_idx(struct got_packidx *packidx,
482     struct got_object_id *id)
483 {
484 	u_int8_t id0 = id->sha1[0];
485 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
486 	int left = 0, right = totobj - 1;
487 
488 	if (id0 > 0)
489 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
490 
491 	while (left <= right) {
492 		struct got_packidx_object_id *oid;
493 		int i, cmp;
494 
495 		i = ((left + right) / 2);
496 		oid = &packidx->hdr.sorted_ids[i];
497 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
498 		if (cmp == 0)
499 			return i;
500 		else if (cmp > 0)
501 			left = i + 1;
502 		else if (cmp < 0)
503 			right = i - 1;
504 	}
505 
506 	return -1;
507 }
508 
509 static int
510 offset_cmp(const void *pa, const void *pb)
511 {
512 	const struct got_pack_offset_index *a, *b;
513 
514 	a = (const struct got_pack_offset_index *)pa;
515 	b = (const struct got_pack_offset_index *)pb;
516 
517 	if (a->offset < b->offset)
518 		return -1;
519 	else if (a->offset > b->offset)
520 		return 1;
521 
522 	return 0;
523 }
524 
525 static int
526 large_offset_cmp(const void *pa, const void *pb)
527 {
528 	const struct got_pack_large_offset_index *a, *b;
529 
530 	a = (const struct got_pack_large_offset_index *)pa;
531 	b = (const struct got_pack_large_offset_index *)pb;
532 
533 	if (a->offset < b->offset)
534 		return -1;
535 	else if (a->offset > b->offset)
536 		return 1;
537 
538 	return 0;
539 }
540 
541 static const struct got_error *
542 build_offset_index(struct got_packidx *p)
543 {
544 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
545 	unsigned int i, j, k;
546 
547 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
548 	    sizeof(p->sorted_offsets[0]));
549 	if (p->sorted_offsets == NULL)
550 		return got_error_from_errno("calloc");
551 
552 	if (p->nlargeobj > 0) {
553 		p->sorted_large_offsets = calloc(p->nlargeobj,
554 		    sizeof(p->sorted_large_offsets[0]));
555 		if (p->sorted_large_offsets == NULL)
556 			return got_error_from_errno("calloc");
557 	}
558 
559 	j = 0;
560 	k = 0;
561 	for (i = 0; i < nobj; i++) {
562 		uint32_t offset = be32toh(p->hdr.offsets[i]);
563 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
564 			uint64_t loffset;
565 			uint32_t idx;
566 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
567 			if (idx >= p->nlargeobj ||
568 			    p->nlargeobj == 0 ||
569 			    p->hdr.large_offsets == NULL)
570 				return got_error(GOT_ERR_BAD_PACKIDX);
571 			loffset = be64toh(p->hdr.large_offsets[idx]);
572 			p->sorted_large_offsets[j].offset = loffset;
573 			p->sorted_large_offsets[j].idx = i;
574 			j++;
575 		} else {
576 			p->sorted_offsets[k].offset = offset;
577 			p->sorted_offsets[k].idx = i;
578 			k++;
579 		}
580 	}
581 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
582 		return got_error(GOT_ERR_BAD_PACKIDX);
583 
584 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
585 	    sizeof(p->sorted_offsets[0]), offset_cmp);
586 
587 	if (p->sorted_large_offsets)
588 		qsort(p->sorted_large_offsets, p->nlargeobj,
589 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
590 
591 	return NULL;
592 }
593 
594 const struct got_error *
595 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
596 {
597 	const struct got_error *err;
598 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
599 	int i, left, right;
600 
601 	*idx = -1;
602 
603 	if (packidx->sorted_offsets == NULL) {
604 		err = build_offset_index(packidx);
605 		if (err)
606 			return err;
607 	}
608 
609 	if (offset >= 0x7fffffff) {
610 		uint64_t lo;
611 		left = 0, right = packidx->nlargeobj - 1;
612 		while (left <= right) {
613 			i = ((left + right) / 2);
614 			lo = packidx->sorted_large_offsets[i].offset;
615 			if (lo == offset) {
616 				*idx = packidx->sorted_large_offsets[i].idx;
617 				break;
618 			} else if (offset > lo)
619 				left = i + 1;
620 			else if (offset < lo)
621 				right = i - 1;
622 		}
623 	} else {
624 		uint32_t o;
625 		left = 0, right = totobj - packidx->nlargeobj - 1;
626 		while (left <= right) {
627 			i = ((left + right) / 2);
628 			o = packidx->sorted_offsets[i].offset;
629 			if (o == offset) {
630 				*idx = packidx->sorted_offsets[i].idx;
631 				break;
632 			} else if (offset > o)
633 				left = i + 1;
634 			else if (offset < o)
635 				right = i - 1;
636 		}
637 	}
638 
639 	return NULL;
640 }
641 
642 const struct got_error *
643 got_packidx_get_object_id(struct got_object_id *id,
644     struct got_packidx *packidx, int idx)
645 {
646 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
647 	struct got_packidx_object_id *oid;
648 
649 	if (idx < 0 || idx >= totobj)
650 		return got_error(GOT_ERR_NO_OBJ);
651 
652 	oid = &packidx->hdr.sorted_ids[idx];
653 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
654 	return NULL;
655 }
656 
657 const struct got_error *
658 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
659     struct got_packidx *packidx, const char *id_str_prefix)
660 {
661 	const struct got_error *err = NULL;
662 	u_int8_t id0;
663 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
664 	char hex[3];
665 	size_t prefix_len = strlen(id_str_prefix);
666 	struct got_packidx_object_id *oid;
667 	uint32_t i = 0;
668 
669 	STAILQ_INIT(matched_ids);
670 
671 	if (prefix_len < 2)
672 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
673 
674 	hex[0] = id_str_prefix[0];
675 	hex[1] = id_str_prefix[1];
676 	hex[2] = '\0';
677 	if (!got_parse_xdigit(&id0, hex))
678 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
679 
680 	if (id0 > 0)
681 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
682 	oid = &packidx->hdr.sorted_ids[i];
683 	while (i < totobj && oid->sha1[0] == id0) {
684 		char id_str[SHA1_DIGEST_STRING_LENGTH];
685 		struct got_object_qid *qid;
686 		int cmp;
687 
688 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
689 		        return got_error(GOT_ERR_NO_SPACE);
690 
691 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
692 		if (cmp < 0) {
693 			oid = &packidx->hdr.sorted_ids[++i];
694 			continue;
695 		} else if (cmp > 0)
696 			break;
697 
698 		err = got_object_qid_alloc_partial(&qid);
699 		if (err)
700 			break;
701 		memcpy(qid->id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
702 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
703 
704 		oid = &packidx->hdr.sorted_ids[++i];
705 	}
706 
707 	if (err)
708 		got_object_id_queue_free(matched_ids);
709 	return err;
710 }
711 
712 static const struct got_error *
713 pack_stop_privsep_child(struct got_pack *pack)
714 {
715 	const struct got_error *err = NULL;
716 
717 	if (pack->privsep_child == NULL)
718 		return NULL;
719 
720 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
721 	if (err)
722 		return err;
723 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
724 	if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
725 		err = got_error_from_errno("close");
726 	free(pack->privsep_child->ibuf);
727 	free(pack->privsep_child);
728 	pack->privsep_child = NULL;
729 	return err;
730 }
731 
732 const struct got_error *
733 got_pack_close(struct got_pack *pack)
734 {
735 	const struct got_error *err = NULL;
736 
737 	err = pack_stop_privsep_child(pack);
738 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
739 		err = got_error_from_errno("munmap");
740 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
741 		err = got_error_from_errno("close");
742 	pack->fd = -1;
743 	free(pack->path_packfile);
744 	pack->path_packfile = NULL;
745 	pack->filesize = 0;
746 	if (pack->delta_cache) {
747 		got_delta_cache_free(pack->delta_cache);
748 		pack->delta_cache = NULL;
749 	}
750 
751 	return err;
752 }
753 
754 const struct got_error *
755 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
756     struct got_pack *pack, off_t offset)
757 {
758 	uint8_t t = 0;
759 	uint64_t s = 0;
760 	uint8_t sizeN;
761 	size_t mapoff = 0;
762 	int i = 0;
763 
764 	*len = 0;
765 
766 	if (offset >= pack->filesize)
767 		return got_error(GOT_ERR_PACK_OFFSET);
768 
769 	if (pack->map) {
770 		mapoff = (size_t)offset;
771 	} else {
772 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
773 			return got_error_from_errno("lseek");
774 	}
775 
776 	do {
777 		/* We do not support size values which don't fit in 64 bit. */
778 		if (i > 9)
779 			return got_error(GOT_ERR_NO_SPACE);
780 
781 		if (pack->map) {
782 			if (mapoff + sizeof(sizeN) >= pack->filesize)
783 				return got_error(GOT_ERR_BAD_PACKFILE);
784 			sizeN = *(pack->map + mapoff);
785 			mapoff += sizeof(sizeN);
786 		} else {
787 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
788 			if (n < 0)
789 				return got_error_from_errno("read");
790 			if (n != sizeof(sizeN))
791 				return got_error(GOT_ERR_BAD_PACKFILE);
792 		}
793 		*len += sizeof(sizeN);
794 
795 		if (i == 0) {
796 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
797 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
798 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
799 		} else {
800 			size_t shift = 4 + 7 * (i - 1);
801 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
802 		}
803 		i++;
804 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
805 
806 	*type = t;
807 	*size = s;
808 	return NULL;
809 }
810 
811 static const struct got_error *
812 open_plain_object(struct got_object **obj, struct got_object_id *id,
813     uint8_t type, off_t offset, size_t size, int idx)
814 {
815 	*obj = calloc(1, sizeof(**obj));
816 	if (*obj == NULL)
817 		return got_error_from_errno("calloc");
818 
819 	(*obj)->type = type;
820 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
821 	(*obj)->pack_idx = idx;
822 	(*obj)->hdrlen = 0;
823 	(*obj)->size = size;
824 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
825 	(*obj)->pack_offset = offset;
826 
827 	return NULL;
828 }
829 
830 static const struct got_error *
831 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
832     off_t delta_offset)
833 {
834 	int64_t o = 0;
835 	uint8_t offN;
836 	int i = 0;
837 
838 	*offset = 0;
839 	*len = 0;
840 
841 	do {
842 		/* We do not support offset values which don't fit in 64 bit. */
843 		if (i > 8)
844 			return got_error(GOT_ERR_NO_SPACE);
845 
846 		if (pack->map) {
847 			size_t mapoff;
848 			mapoff = (size_t)delta_offset + *len;
849 			if (mapoff + sizeof(offN) >= pack->filesize)
850 				return got_error(GOT_ERR_PACK_OFFSET);
851 			offN = *(pack->map + mapoff);
852 		} else {
853 			ssize_t n;
854 			n = read(pack->fd, &offN, sizeof(offN));
855 			if (n < 0)
856 				return got_error_from_errno("read");
857 			if (n != sizeof(offN))
858 				return got_error(GOT_ERR_BAD_PACKFILE);
859 		}
860 		*len += sizeof(offN);
861 
862 		if (i == 0)
863 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
864 		else {
865 			o++;
866 			o <<= 7;
867 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
868 		}
869 		i++;
870 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
871 
872 	*offset = o;
873 	return NULL;
874 }
875 
876 const struct got_error *
877 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
878     struct got_pack *pack, off_t offset, int tslen)
879 {
880 	const struct got_error *err;
881 	int64_t negoffset;
882 	size_t negofflen;
883 
884 	*len = 0;
885 
886 	err = parse_negative_offset(&negoffset, &negofflen, pack,
887 	    offset + tslen);
888 	if (err)
889 		return err;
890 
891 	/* Compute the base object's offset (must be in the same pack file). */
892 	*base_offset = (offset - negoffset);
893 	if (*base_offset <= 0)
894 		return got_error(GOT_ERR_BAD_PACKFILE);
895 
896 	*len = negofflen;
897 	return NULL;
898 }
899 
900 static const struct got_error *
901 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
902     size_t delta_data_offset, struct got_pack *pack)
903 {
904 	const struct got_error *err = NULL;
905 
906 	if (pack->map) {
907 		if (delta_data_offset >= pack->filesize)
908 			return got_error(GOT_ERR_PACK_OFFSET);
909 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
910 		    NULL, NULL, pack->map, delta_data_offset,
911 		    pack->filesize - delta_data_offset);
912 	} else {
913 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
914 			return got_error_from_errno("lseek");
915 		err = got_inflate_to_mem_fd(delta_buf, delta_len, NULL,
916 		    NULL, 0, pack->fd);
917 	}
918 	return err;
919 }
920 
921 static const struct got_error *
922 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
923     int delta_type, size_t delta_size, size_t delta_data_offset)
924 {
925 	struct got_delta *delta;
926 
927 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
928 	    delta_data_offset);
929 	if (delta == NULL)
930 		return got_error_from_errno("got_delta_open");
931 	/* delta is freed in got_object_close() */
932 	deltas->nentries++;
933 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
934 	return NULL;
935 }
936 
937 static const struct got_error *
938 resolve_offset_delta(struct got_delta_chain *deltas,
939     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
940     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
941 
942 {
943 	const struct got_error *err;
944 	off_t base_offset;
945 	uint8_t base_type;
946 	uint64_t base_size;
947 	size_t base_tslen;
948 	off_t delta_data_offset;
949 	size_t consumed;
950 
951 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
952 	    delta_offset, tslen);
953 	if (err)
954 		return err;
955 
956 	delta_data_offset = delta_offset + tslen + consumed;
957 	if (delta_data_offset >= pack->filesize)
958 		return got_error(GOT_ERR_PACK_OFFSET);
959 
960 	if (pack->map == NULL) {
961 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
962 		if (delta_data_offset == -1)
963 			return got_error_from_errno("lseek");
964 	}
965 
966 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
967 	    delta_data_offset);
968 	if (err)
969 		return err;
970 
971 	/* An offset delta must be in the same packfile. */
972 	if (base_offset >= pack->filesize)
973 		return got_error(GOT_ERR_PACK_OFFSET);
974 
975 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
976 	    &base_tslen, pack, base_offset);
977 	if (err)
978 		return err;
979 
980 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
981 	    base_tslen, base_type, base_size, recursion - 1);
982 }
983 
984 const struct got_error *
985 got_pack_parse_ref_delta(struct got_object_id *id,
986     struct got_pack *pack, off_t delta_offset, int tslen)
987 {
988 	if (pack->map) {
989 		size_t mapoff = delta_offset + tslen;
990 		if (mapoff + sizeof(*id) >= pack->filesize)
991 			return got_error(GOT_ERR_PACK_OFFSET);
992 		memcpy(id, pack->map + mapoff, sizeof(*id));
993 	} else {
994 		ssize_t n;
995 		n = read(pack->fd, id, sizeof(*id));
996 		if (n < 0)
997 			return got_error_from_errno("read");
998 		if (n != sizeof(*id))
999 			return got_error(GOT_ERR_BAD_PACKFILE);
1000 	}
1001 
1002 	return NULL;
1003 }
1004 
1005 static const struct got_error *
1006 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1007     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1008     size_t delta_size, unsigned int recursion)
1009 {
1010 	const struct got_error *err;
1011 	struct got_object_id id;
1012 	int idx;
1013 	off_t base_offset;
1014 	uint8_t base_type;
1015 	uint64_t base_size;
1016 	size_t base_tslen;
1017 	off_t delta_data_offset;
1018 
1019 	if (delta_offset + tslen >= pack->filesize)
1020 		return got_error(GOT_ERR_PACK_OFFSET);
1021 
1022 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1023 	if (err)
1024 		return err;
1025 	if (pack->map) {
1026 		delta_data_offset = delta_offset + tslen + sizeof(id);
1027 	} else {
1028 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1029 		if (delta_data_offset == -1)
1030 			return got_error_from_errno("lseek");
1031 	}
1032 
1033 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1034 	    delta_data_offset);
1035 	if (err)
1036 		return err;
1037 
1038 	/* Delta base must be in the same pack file. */
1039 	idx = got_packidx_get_object_idx(packidx, &id);
1040 	if (idx == -1)
1041 		return got_error(GOT_ERR_NO_OBJ);
1042 
1043 	base_offset = got_packidx_get_object_offset(packidx, idx);
1044 	if (base_offset == (uint64_t)-1)
1045 		return got_error(GOT_ERR_BAD_PACKIDX);
1046 
1047 	if (base_offset >= pack->filesize)
1048 		return got_error(GOT_ERR_PACK_OFFSET);
1049 
1050 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1051 	    &base_tslen, pack, base_offset);
1052 	if (err)
1053 		return err;
1054 
1055 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1056 	    base_tslen, base_type, base_size, recursion - 1);
1057 }
1058 
1059 const struct got_error *
1060 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1061     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1062     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1063 {
1064 	const struct got_error *err = NULL;
1065 
1066 	if (--recursion == 0)
1067 		return got_error(GOT_ERR_RECURSION);
1068 
1069 	switch (delta_type) {
1070 	case GOT_OBJ_TYPE_COMMIT:
1071 	case GOT_OBJ_TYPE_TREE:
1072 	case GOT_OBJ_TYPE_BLOB:
1073 	case GOT_OBJ_TYPE_TAG:
1074 		/* Plain types are the final delta base. Recursion ends. */
1075 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1076 		    delta_size, 0);
1077 		break;
1078 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1079 		err = resolve_offset_delta(deltas, packidx, pack,
1080 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1081 		break;
1082 	case GOT_OBJ_TYPE_REF_DELTA:
1083 		err = resolve_ref_delta(deltas, packidx, pack,
1084 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1085 		break;
1086 	default:
1087 		return got_error(GOT_ERR_OBJ_TYPE);
1088 	}
1089 
1090 	return err;
1091 }
1092 
1093 static const struct got_error *
1094 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1095     struct got_pack *pack, struct got_object_id *id, off_t offset,
1096     size_t tslen, int delta_type, size_t delta_size, int idx)
1097 {
1098 	const struct got_error *err = NULL;
1099 	int resolved_type;
1100 
1101 	*obj = calloc(1, sizeof(**obj));
1102 	if (*obj == NULL)
1103 		return got_error_from_errno("calloc");
1104 
1105 	(*obj)->flags = 0;
1106 	(*obj)->hdrlen = 0;
1107 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1108 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1109 	(*obj)->pack_offset = offset + tslen;
1110 
1111 	STAILQ_INIT(&(*obj)->deltas.entries);
1112 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1113 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1114 	(*obj)->pack_idx = idx;
1115 
1116 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1117 	    offset, tslen, delta_type, delta_size,
1118 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1119 	if (err)
1120 		goto done;
1121 
1122 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1123 	if (err)
1124 		goto done;
1125 	(*obj)->type = resolved_type;
1126 done:
1127 	if (err) {
1128 		got_object_close(*obj);
1129 		*obj = NULL;
1130 	}
1131 	return err;
1132 }
1133 
1134 const struct got_error *
1135 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1136     struct got_packidx *packidx, int idx, struct got_object_id *id)
1137 {
1138 	const struct got_error *err = NULL;
1139 	off_t offset;
1140 	uint8_t type;
1141 	uint64_t size;
1142 	size_t tslen;
1143 
1144 	*obj = NULL;
1145 
1146 	offset = got_packidx_get_object_offset(packidx, idx);
1147 	if (offset == (uint64_t)-1)
1148 		return got_error(GOT_ERR_BAD_PACKIDX);
1149 
1150 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1151 	    pack, offset);
1152 	if (err)
1153 		return err;
1154 
1155 	switch (type) {
1156 	case GOT_OBJ_TYPE_COMMIT:
1157 	case GOT_OBJ_TYPE_TREE:
1158 	case GOT_OBJ_TYPE_BLOB:
1159 	case GOT_OBJ_TYPE_TAG:
1160 		err = open_plain_object(obj, id, type, offset + tslen,
1161 		    size, idx);
1162 		break;
1163 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1164 	case GOT_OBJ_TYPE_REF_DELTA:
1165 		err = open_delta_object(obj, packidx, pack, id, offset,
1166 		    tslen, type, size, idx);
1167 		break;
1168 	default:
1169 		err = got_error(GOT_ERR_OBJ_TYPE);
1170 		break;
1171 	}
1172 
1173 	return err;
1174 }
1175 
1176 const struct got_error *
1177 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1178     struct got_delta_chain *deltas, struct got_pack *pack)
1179 {
1180 	struct got_delta *delta;
1181 	uint64_t base_size = 0, result_size = 0;
1182 
1183 	*max_size = 0;
1184 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1185 		/* Plain object types are the delta base. */
1186 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1187 		    delta->type != GOT_OBJ_TYPE_TREE &&
1188 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1189 		    delta->type != GOT_OBJ_TYPE_TAG) {
1190 			const struct got_error *err;
1191 			uint8_t *delta_buf;
1192 			size_t delta_len;
1193 			int cached = 1;
1194 
1195 			got_delta_cache_get(&delta_buf, &delta_len,
1196 			    pack->delta_cache, delta->data_offset);
1197 			if (delta_buf == NULL) {
1198 				cached = 0;
1199 				err = read_delta_data(&delta_buf, &delta_len,
1200 				    delta->data_offset, pack);
1201 				if (err)
1202 					return err;
1203 				err = got_delta_cache_add(pack->delta_cache,
1204 				    delta->data_offset, delta_buf, delta_len);
1205 				if (err == NULL)
1206 					cached = 1;
1207 				else if (err->code != GOT_ERR_NO_SPACE) {
1208 					free(delta_buf);
1209 					return err;
1210 				}
1211 			}
1212 			err = got_delta_get_sizes(&base_size, &result_size,
1213 			    delta_buf, delta_len);
1214 			if (!cached)
1215 				free(delta_buf);
1216 			if (err)
1217 				return err;
1218 		} else
1219 			base_size = delta->size;
1220 		if (base_size > *max_size)
1221 			*max_size = base_size;
1222 		if (result_size > *max_size)
1223 			*max_size = result_size;
1224 	}
1225 
1226 	return NULL;
1227 }
1228 
1229 const struct got_error *
1230 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1231     struct got_pack *pack)
1232 {
1233 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1234 		return got_error(GOT_ERR_OBJ_TYPE);
1235 
1236 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1237 }
1238 
1239 const struct got_error *
1240 got_pack_dump_delta_chain_to_file(size_t *result_size,
1241     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1242     FILE *base_file, FILE *accum_file)
1243 {
1244 	const struct got_error *err = NULL;
1245 	struct got_delta *delta;
1246 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1247 	size_t base_bufsz = 0, accum_size = 0, delta_len;
1248 	uint64_t max_size;
1249 	int n = 0;
1250 
1251 	*result_size = 0;
1252 
1253 	if (STAILQ_EMPTY(&deltas->entries))
1254 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1255 
1256 	/* We process small enough files entirely in memory for speed. */
1257 	err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1258 	if (err)
1259 		return err;
1260 	if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1261 		accum_buf = malloc(max_size);
1262 		if (accum_buf == NULL)
1263 			return got_error_from_errno("malloc");
1264 		base_file = NULL;
1265 		accum_file = NULL;
1266 	} else {
1267 		if (fseeko(base_file, 0L, SEEK_SET) == -1)
1268 			return got_error_from_errno("fseeko");
1269 		if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1270 			return got_error_from_errno("fseeko");
1271 	}
1272 
1273 	/* Deltas are ordered in ascending order. */
1274 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1275 		int cached = 1;
1276 		if (n == 0) {
1277 			size_t mapoff;
1278 			off_t delta_data_offset;
1279 
1280 			/* Plain object types are the delta base. */
1281 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1282 			    delta->type != GOT_OBJ_TYPE_TREE &&
1283 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1284 			    delta->type != GOT_OBJ_TYPE_TAG) {
1285 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1286 				goto done;
1287 			}
1288 
1289 			delta_data_offset = delta->offset + delta->tslen;
1290 			if (delta_data_offset >= pack->filesize) {
1291 				err = got_error(GOT_ERR_PACK_OFFSET);
1292 				goto done;
1293 			}
1294 			if (pack->map == NULL) {
1295 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1296 				    == -1) {
1297 					err = got_error_from_errno("lseek");
1298 					goto done;
1299 				}
1300 			}
1301 			if (base_file) {
1302 				if (pack->map) {
1303 					mapoff = (size_t)delta_data_offset;
1304 					err = got_inflate_to_file_mmap(
1305 					    &base_bufsz, NULL, NULL, pack->map,
1306 					    mapoff, pack->filesize - mapoff,
1307 					    base_file);
1308 				} else
1309 					err = got_inflate_to_file_fd(
1310 					    &base_bufsz, NULL, NULL, pack->fd,
1311 					    base_file);
1312 			} else {
1313 				if (pack->map) {
1314 					mapoff = (size_t)delta_data_offset;
1315 					err = got_inflate_to_mem_mmap(&base_buf,
1316 					    &base_bufsz, NULL, NULL,
1317 					    pack->map, mapoff,
1318 					    pack->filesize - mapoff);
1319 				} else
1320 					err = got_inflate_to_mem_fd(&base_buf,
1321 					    &base_bufsz, NULL, NULL, max_size,
1322 					    pack->fd);
1323 			}
1324 			if (err)
1325 				goto done;
1326 			n++;
1327 			if (base_file)
1328 				rewind(base_file);
1329 			continue;
1330 		}
1331 
1332 		got_delta_cache_get(&delta_buf, &delta_len,
1333 		    pack->delta_cache, delta->data_offset);
1334 		if (delta_buf == NULL) {
1335 			cached = 0;
1336 			err = read_delta_data(&delta_buf, &delta_len,
1337 			    delta->data_offset, pack);
1338 			if (err)
1339 				goto done;
1340 			err = got_delta_cache_add(pack->delta_cache,
1341 			    delta->data_offset, delta_buf, delta_len);
1342 			if (err == NULL)
1343 				cached = 1;
1344 			else if (err->code != GOT_ERR_NO_SPACE) {
1345 				free(delta_buf);
1346 				goto done;
1347 			}
1348 		}
1349 		if (base_buf) {
1350 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1351 			    delta_buf, delta_len, accum_buf,
1352 			    &accum_size, max_size);
1353 			n++;
1354 		} else {
1355 			err = got_delta_apply(base_file, delta_buf,
1356 			    delta_len,
1357 			    /* Final delta application writes to output file. */
1358 			    ++n < deltas->nentries ? accum_file : outfile,
1359 			    &accum_size);
1360 		}
1361 		if (!cached)
1362 			free(delta_buf);
1363 		if (err)
1364 			goto done;
1365 
1366 		if (n < deltas->nentries) {
1367 			/* Accumulated delta becomes the new base. */
1368 			if (base_buf) {
1369 				uint8_t *tmp = accum_buf;
1370 				/*
1371 				 * Base buffer switches roles with accumulation
1372 				 * buffer. Ensure it can hold the largest
1373 				 * result in the delta chain. The initial
1374 				 * allocation might have been smaller.
1375 				 */
1376 				if (base_bufsz < max_size) {
1377 					uint8_t *p;
1378 					p = reallocarray(base_buf, 1, max_size);
1379 					if (p == NULL) {
1380 						err = got_error_from_errno(
1381 						    "reallocarray");
1382 						goto done;
1383 					}
1384 					base_buf = p;
1385 					base_bufsz = max_size;
1386 				}
1387 				accum_buf = base_buf;
1388 				base_buf = tmp;
1389 			} else {
1390 				FILE *tmp = accum_file;
1391 				accum_file = base_file;
1392 				base_file = tmp;
1393 				rewind(base_file);
1394 				rewind(accum_file);
1395 			}
1396 		}
1397 	}
1398 
1399 done:
1400 	free(base_buf);
1401 	if (accum_buf) {
1402 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1403 		free(accum_buf);
1404 		if (len != accum_size)
1405 			err = got_ferror(outfile, GOT_ERR_IO);
1406 	}
1407 	rewind(outfile);
1408 	if (err == NULL)
1409 		*result_size = accum_size;
1410 	return err;
1411 }
1412 
1413 const struct got_error *
1414 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1415     struct got_delta_chain *deltas, struct got_pack *pack)
1416 {
1417 	const struct got_error *err = NULL;
1418 	struct got_delta *delta;
1419 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1420 	size_t base_bufsz = 0, accum_size = 0, delta_len;
1421 	uint64_t max_size;
1422 	int n = 0;
1423 
1424 	*outbuf = NULL;
1425 	*outlen = 0;
1426 
1427 	if (STAILQ_EMPTY(&deltas->entries))
1428 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1429 
1430 	err = got_pack_get_delta_chain_max_size(&max_size, deltas, pack);
1431 	if (err)
1432 		return err;
1433 	accum_buf = malloc(max_size);
1434 	if (accum_buf == NULL)
1435 		return got_error_from_errno("malloc");
1436 
1437 	/* Deltas are ordered in ascending order. */
1438 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1439 		int cached = 1;
1440 		if (n == 0) {
1441 			size_t delta_data_offset;
1442 
1443 			/* Plain object types are the delta base. */
1444 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1445 			    delta->type != GOT_OBJ_TYPE_TREE &&
1446 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1447 			    delta->type != GOT_OBJ_TYPE_TAG) {
1448 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1449 				goto done;
1450 			}
1451 
1452 			delta_data_offset = delta->offset + delta->tslen;
1453 			if (delta_data_offset >= pack->filesize) {
1454 				err = got_error(GOT_ERR_PACK_OFFSET);
1455 				goto done;
1456 			}
1457 			if (pack->map) {
1458 				size_t mapoff = (size_t)delta_data_offset;
1459 				err = got_inflate_to_mem_mmap(&base_buf,
1460 				    &base_bufsz, NULL, NULL, pack->map,
1461 				    mapoff, pack->filesize - mapoff);
1462 			} else {
1463 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1464 				    == -1) {
1465 					err = got_error_from_errno("lseek");
1466 					goto done;
1467 				}
1468 				err = got_inflate_to_mem_fd(&base_buf,
1469 				    &base_bufsz, NULL, NULL, max_size,
1470 				    pack->fd);
1471 			}
1472 			if (err)
1473 				goto done;
1474 			n++;
1475 			continue;
1476 		}
1477 
1478 		got_delta_cache_get(&delta_buf, &delta_len,
1479 		    pack->delta_cache, delta->data_offset);
1480 		if (delta_buf == NULL) {
1481 			cached = 0;
1482 			err = read_delta_data(&delta_buf, &delta_len,
1483 			    delta->data_offset, pack);
1484 			if (err)
1485 				goto done;
1486 			err = got_delta_cache_add(pack->delta_cache,
1487 			    delta->data_offset, delta_buf, delta_len);
1488 			if (err == NULL)
1489 				cached = 1;
1490 			else if (err->code != GOT_ERR_NO_SPACE) {
1491 				free(delta_buf);
1492 				goto done;
1493 			}
1494 		}
1495 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1496 		    delta_buf, delta_len, accum_buf,
1497 		    &accum_size, max_size);
1498 		if (!cached)
1499 			free(delta_buf);
1500 		n++;
1501 		if (err)
1502 			goto done;
1503 
1504 		if (n < deltas->nentries) {
1505 			/* Accumulated delta becomes the new base. */
1506 			uint8_t *tmp = accum_buf;
1507 			/*
1508 			 * Base buffer switches roles with accumulation buffer.
1509 			 * Ensure it can hold the largest result in the delta
1510 			 * chain. Initial allocation might have been smaller.
1511 			 */
1512 			if (base_bufsz < max_size) {
1513 				uint8_t *p;
1514 				p = reallocarray(base_buf, 1, max_size);
1515 				if (p == NULL) {
1516 					err = got_error_from_errno(
1517 					    "reallocarray");
1518 					goto done;
1519 				}
1520 				base_buf = p;
1521 				base_bufsz = max_size;
1522 			}
1523 			accum_buf = base_buf;
1524 			base_buf = tmp;
1525 		}
1526 	}
1527 
1528 done:
1529 	free(base_buf);
1530 	if (err) {
1531 		free(accum_buf);
1532 		*outbuf = NULL;
1533 		*outlen = 0;
1534 	} else {
1535 		*outbuf = accum_buf;
1536 		*outlen = accum_size;
1537 	}
1538 	return err;
1539 }
1540 
1541 const struct got_error *
1542 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1543     FILE *outfile, FILE *base_file, FILE *accum_file)
1544 {
1545 	const struct got_error *err = NULL;
1546 
1547 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1548 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1549 
1550 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1551 		if (obj->pack_offset >= pack->filesize)
1552 			return got_error(GOT_ERR_PACK_OFFSET);
1553 
1554 		if (pack->map) {
1555 			size_t mapoff = (size_t)obj->pack_offset;
1556 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1557 			    pack->map, mapoff, pack->filesize - mapoff,
1558 			    outfile);
1559 		} else {
1560 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1561 				return got_error_from_errno("lseek");
1562 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1563 			    pack->fd, outfile);
1564 		}
1565 	} else
1566 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1567 		    &obj->deltas, pack, outfile, base_file, accum_file);
1568 
1569 	return err;
1570 }
1571 
1572 const struct got_error *
1573 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1574     struct got_object *obj, struct got_pack *pack)
1575 {
1576 	const struct got_error *err = NULL;
1577 
1578 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1579 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1580 
1581 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1582 		if (obj->pack_offset >= pack->filesize)
1583 			return got_error(GOT_ERR_PACK_OFFSET);
1584 		if (pack->map) {
1585 			size_t mapoff = (size_t)obj->pack_offset;
1586 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1587 			    pack->map, mapoff, pack->filesize - mapoff);
1588 		} else {
1589 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1590 				return got_error_from_errno("lseek");
1591 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1592 			    obj->size, pack->fd);
1593 		}
1594 	} else
1595 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1596 		    pack);
1597 
1598 	return err;
1599 }
1600 
1601 const struct got_error *
1602 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1603     off_t *delta_offset, off_t *base_offset, struct got_object_id *base_id,
1604     uint64_t *base_size, uint64_t *result_size, struct got_pack *pack,
1605     struct got_packidx *packidx, int idx)
1606 {
1607 	const struct got_error *err = NULL;
1608 	off_t offset;
1609 	uint8_t type;
1610 	uint64_t size;
1611 	size_t tslen, delta_hdrlen;
1612 
1613 	*delta_buf = NULL;
1614 	*delta_size = 0;
1615 	*delta_offset = 0;
1616 	*base_offset = 0;
1617 	*base_size = 0;
1618 	*result_size = 0;
1619 
1620 	offset = got_packidx_get_object_offset(packidx, idx);
1621 	if (offset == (uint64_t)-1)
1622 		return got_error(GOT_ERR_BAD_PACKIDX);
1623 
1624 	if (offset >= pack->filesize)
1625 		return got_error(GOT_ERR_PACK_OFFSET);
1626 
1627 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1628 	    pack, offset);
1629 	if (err)
1630 		return err;
1631 
1632 	if (tslen + size < tslen || offset + size < size ||
1633 	    tslen + offset < tslen)
1634 		return got_error(GOT_ERR_PACK_OFFSET);
1635 
1636 	switch (type) {
1637 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1638 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1639 		    pack, offset, tslen);
1640 		if (err)
1641 			return err;
1642 		break;
1643 	case GOT_OBJ_TYPE_REF_DELTA:
1644 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1645 		if (err)
1646 			return err;
1647 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1648 		break;
1649 	default:
1650 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1651 		    "non-delta object type %d found at offset %llu",
1652 		    type, offset);
1653 	}
1654 
1655 	if (tslen + delta_hdrlen < delta_hdrlen ||
1656 	    offset + delta_hdrlen < delta_hdrlen)
1657 		return got_error(GOT_ERR_BAD_DELTA);
1658 
1659 	err = read_delta_data(delta_buf, delta_size,
1660 	    offset + tslen + delta_hdrlen, pack);
1661 	if (err)
1662 		return err;
1663 
1664 	if (*delta_size != size) {
1665 		err = got_error(GOT_ERR_BAD_DELTA);
1666 		goto done;
1667 	}
1668 
1669 	err = got_delta_get_sizes(base_size, result_size, *delta_buf, size);
1670 	if (err)
1671 		goto done;
1672 
1673 	*delta_offset = offset;
1674 done:
1675 	if (err) {
1676 		free(*delta_buf);
1677 		*delta_buf = NULL;
1678 	}
1679 	return err;
1680 }