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/mman.h>
22 #include <sys/resource.h>
23 #include <sys/socket.h>
24 
25 #include <fcntl.h>
26 #include <errno.h>
27 #include <stdio.h>
28 #include <stdint.h>
29 #include <stdlib.h>
30 #include <string.h>
31 #include <limits.h>
32 #include <unistd.h>
33 #include <zlib.h>
34 
35 #include "got_error.h"
36 #include "got_object.h"
37 #include "got_path.h"
38 
39 #include "got_lib_sha1.h"
40 #include "got_lib_delta.h"
41 #include "got_lib_delta_cache.h"
42 #include "got_lib_inflate.h"
43 #include "got_lib_object.h"
44 #include "got_lib_object_parse.h"
45 #include "got_lib_privsep.h"
46 #include "got_lib_pack.h"
47 
48 #ifndef nitems
49 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
50 #endif
51 
52 #ifndef MIN
53 #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
54 #endif
55 
56 static const struct got_error *
57 verify_fanout_table(uint32_t *fanout_table)
58 {
59 	int i;
60 
61 	for (i = 0; i < 0xff - 1; i++) {
62 		if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
63 			return got_error(GOT_ERR_BAD_PACKIDX);
64 	}
65 
66 	return NULL;
67 }
68 
69 const struct got_error *
70 got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
71 {
72 	const struct got_error *err = NULL;
73 	struct got_packidx_v2_hdr *h;
74 	SHA1_CTX ctx;
75 	uint8_t sha1[SHA1_DIGEST_LENGTH];
76 	size_t nobj, len_fanout, len_ids, offset, remain;
77 	ssize_t n;
78 	int i;
79 
80 	SHA1Init(&ctx);
81 
82 	h = &p->hdr;
83 	offset = 0;
84 	remain = p->len;
85 
86 	if (remain < sizeof(*h->magic)) {
87 		err = got_error(GOT_ERR_BAD_PACKIDX);
88 		goto done;
89 	}
90 	if (p->map)
91 		h->magic = (uint32_t *)(p->map + offset);
92 	else {
93 		h->magic = malloc(sizeof(*h->magic));
94 		if (h->magic == NULL) {
95 			err = got_error_from_errno("malloc");
96 			goto done;
97 		}
98 		n = read(p->fd, h->magic, sizeof(*h->magic));
99 		if (n < 0) {
100 			err = got_error_from_errno("read");
101 			goto done;
102 		} else if (n != sizeof(*h->magic)) {
103 			err = got_error(GOT_ERR_BAD_PACKIDX);
104 			goto done;
105 		}
106 	}
107 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
108 		err = got_error(GOT_ERR_BAD_PACKIDX);
109 		goto done;
110 	}
111 	offset += sizeof(*h->magic);
112 	remain -= sizeof(*h->magic);
113 
114 	if (verify)
115 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
116 
117 	if (remain < sizeof(*h->version)) {
118 		err = got_error(GOT_ERR_BAD_PACKIDX);
119 		goto done;
120 	}
121 	if (p->map)
122 		h->version = (uint32_t *)(p->map + offset);
123 	else {
124 		h->version = malloc(sizeof(*h->version));
125 		if (h->version == NULL) {
126 			err = got_error_from_errno("malloc");
127 			goto done;
128 		}
129 		n = read(p->fd, h->version, sizeof(*h->version));
130 		if (n < 0) {
131 			err = got_error_from_errno("read");
132 			goto done;
133 		} else if (n != sizeof(*h->version)) {
134 			err = got_error(GOT_ERR_BAD_PACKIDX);
135 			goto done;
136 		}
137 	}
138 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
139 		err = got_error(GOT_ERR_BAD_PACKIDX);
140 		goto done;
141 	}
142 	offset += sizeof(*h->version);
143 	remain -= sizeof(*h->version);
144 
145 	if (verify)
146 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
147 
148 	len_fanout =
149 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
150 	if (remain < len_fanout) {
151 		err = got_error(GOT_ERR_BAD_PACKIDX);
152 		goto done;
153 	}
154 	if (p->map)
155 		h->fanout_table = (uint32_t *)(p->map + offset);
156 	else {
157 		h->fanout_table = malloc(len_fanout);
158 		if (h->fanout_table == NULL) {
159 			err = got_error_from_errno("malloc");
160 			goto done;
161 		}
162 		n = read(p->fd, h->fanout_table, len_fanout);
163 		if (n < 0) {
164 			err = got_error_from_errno("read");
165 			goto done;
166 		} else if (n != len_fanout) {
167 			err = got_error(GOT_ERR_BAD_PACKIDX);
168 			goto done;
169 		}
170 	}
171 	err = verify_fanout_table(h->fanout_table);
172 	if (err)
173 		goto done;
174 	if (verify)
175 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
176 	offset += len_fanout;
177 	remain -= len_fanout;
178 
179 	nobj = be32toh(h->fanout_table[0xff]);
180 	len_ids = nobj * sizeof(*h->sorted_ids);
181 	if (len_ids <= nobj || len_ids > remain) {
182 		err = got_error(GOT_ERR_BAD_PACKIDX);
183 		goto done;
184 	}
185 	if (p->map)
186 		h->sorted_ids =
187 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
188 	else {
189 		h->sorted_ids = malloc(len_ids);
190 		if (h->sorted_ids == NULL) {
191 			err = got_error(GOT_ERR_BAD_PACKIDX);
192 			goto done;
193 		}
194 		n = read(p->fd, h->sorted_ids, len_ids);
195 		if (n < 0)
196 			err = got_error_from_errno("read");
197 		else if (n != len_ids) {
198 			err = got_error(GOT_ERR_BAD_PACKIDX);
199 			goto done;
200 		}
201 	}
202 	if (verify)
203 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
204 	offset += len_ids;
205 	remain -= len_ids;
206 
207 	if (remain < nobj * sizeof(*h->crc32)) {
208 		err = got_error(GOT_ERR_BAD_PACKIDX);
209 		goto done;
210 	}
211 	if (p->map)
212 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
213 	else {
214 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
215 		if (h->crc32 == NULL) {
216 			err = got_error_from_errno("malloc");
217 			goto done;
218 		}
219 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
220 		if (n < 0)
221 			err = got_error_from_errno("read");
222 		else if (n != nobj * sizeof(*h->crc32)) {
223 			err = got_error(GOT_ERR_BAD_PACKIDX);
224 			goto done;
225 		}
226 	}
227 	if (verify)
228 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
229 	remain -= nobj * sizeof(*h->crc32);
230 	offset += nobj * sizeof(*h->crc32);
231 
232 	if (remain < nobj * sizeof(*h->offsets)) {
233 		err = got_error(GOT_ERR_BAD_PACKIDX);
234 		goto done;
235 	}
236 	if (p->map)
237 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
238 	else {
239 		h->offsets = malloc(nobj * sizeof(*h->offsets));
240 		if (h->offsets == NULL) {
241 			err = got_error_from_errno("malloc");
242 			goto done;
243 		}
244 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
245 		if (n < 0)
246 			err = got_error_from_errno("read");
247 		else if (n != nobj * sizeof(*h->offsets)) {
248 			err = got_error(GOT_ERR_BAD_PACKIDX);
249 			goto done;
250 		}
251 	}
252 	if (verify)
253 		SHA1Update(&ctx, (uint8_t *)h->offsets,
254 		    nobj * sizeof(*h->offsets));
255 	remain -= nobj * sizeof(*h->offsets);
256 	offset += nobj * sizeof(*h->offsets);
257 
258 	/* Large file offsets are contained only in files > 2GB. */
259 	if (verify || packfile_size > 0x7fffffff) {
260 		for (i = 0; i < nobj; i++) {
261 			uint32_t o = h->offsets[i];
262 			if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
263 				p->nlargeobj++;
264 		}
265 	}
266 	if (p->nlargeobj == 0)
267 		goto checksum;
268 	else if (packfile_size <= 0x7fffffff) {
269 		err = got_error(GOT_ERR_BAD_PACKIDX);
270 		goto done;
271 	}
272 
273 	if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
274 		err = got_error(GOT_ERR_BAD_PACKIDX);
275 		goto done;
276 	}
277 	if (p->map)
278 		h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
279 	else {
280 		h->large_offsets = malloc(p->nlargeobj *
281 		    sizeof(*h->large_offsets));
282 		if (h->large_offsets == NULL) {
283 			err = got_error_from_errno("malloc");
284 			goto done;
285 		}
286 		n = read(p->fd, h->large_offsets,
287 		    p->nlargeobj * sizeof(*h->large_offsets));
288 		if (n < 0)
289 			err = got_error_from_errno("read");
290 		else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
291 			err = got_error(GOT_ERR_BAD_PACKIDX);
292 			goto done;
293 		}
294 	}
295 	if (verify)
296 		SHA1Update(&ctx, (uint8_t*)h->large_offsets,
297 		    p->nlargeobj * sizeof(*h->large_offsets));
298 	remain -= p->nlargeobj * sizeof(*h->large_offsets);
299 	offset += p->nlargeobj * sizeof(*h->large_offsets);
300 
301 checksum:
302 	if (remain < sizeof(*h->trailer)) {
303 		err = got_error(GOT_ERR_BAD_PACKIDX);
304 		goto done;
305 	}
306 	if (p->map)
307 		h->trailer =
308 		    (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
309 	else {
310 		h->trailer = malloc(sizeof(*h->trailer));
311 		if (h->trailer == NULL) {
312 			err = got_error_from_errno("malloc");
313 			goto done;
314 		}
315 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
316 		if (n < 0)
317 			err = got_error_from_errno("read");
318 		else if (n != sizeof(*h->trailer)) {
319 			err = got_error(GOT_ERR_BAD_PACKIDX);
320 			goto done;
321 		}
322 	}
323 	if (verify) {
324 		SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
325 		SHA1Final(sha1, &ctx);
326 		if (memcmp(h->trailer->packidx_sha1, sha1,
327 		    SHA1_DIGEST_LENGTH) != 0)
328 			err = got_error(GOT_ERR_PACKIDX_CSUM);
329 	}
330 done:
331 	return err;
332 }
333 
334 const struct got_error *
335 got_packidx_open(struct got_packidx **packidx,
336     int dir_fd, const char *relpath, int verify)
337 {
338 	const struct got_error *err = NULL;
339 	struct got_packidx *p = NULL;
340 	char *pack_relpath;
341 	struct stat idx_sb, pack_sb;
342 
343 	*packidx = NULL;
344 
345 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
346 	if (err)
347 		return err;
348 
349 	/*
350 	 * Ensure that a corresponding pack file exists.
351 	 * Some Git repositories have this problem. Git seems to ignore
352 	 * the existence of lonely pack index files but we do not.
353 	 */
354 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
355 		if (errno == ENOENT) {
356 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
357 			    "%s", relpath);
358 		} else
359 			err = got_error_from_errno2("fstatat", pack_relpath);
360 		goto done;
361 	}
362 
363 	p = calloc(1, sizeof(*p));
364 	if (p == NULL) {
365 		err = got_error_from_errno("calloc");
366 		goto done;
367 	}
368 
369 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
370 	if (p->fd == -1) {
371 		err = got_error_from_errno2("openat", relpath);
372 		goto done;
373 	}
374 
375 	if (fstat(p->fd, &idx_sb) != 0) {
376 		err = got_error_from_errno2("fstat", relpath);
377 		goto done;
378 	}
379 	p->len = idx_sb.st_size;
380 	if (p->len < sizeof(p->hdr)) {
381 		err = got_error(GOT_ERR_BAD_PACKIDX);
382 		goto done;
383 	}
384 
385 	p->path_packidx = strdup(relpath);
386 	if (p->path_packidx == NULL) {
387 		err = got_error_from_errno("strdup");
388 		goto done;
389 	}
390 
391 #ifndef GOT_PACK_NO_MMAP
392 	p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
393 	if (p->map == MAP_FAILED) {
394 		if (errno != ENOMEM) {
395 			err = got_error_from_errno("mmap");
396 			goto done;
397 		}
398 		p->map = NULL; /* fall back to read(2) */
399 	}
400 #endif
401 
402 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
403 done:
404 	if (err) {
405 		if (p)
406 			got_packidx_close(p);
407 	} else
408 		*packidx = p;
409 	free(pack_relpath);
410 	return err;
411 }
412 
413 const struct got_error *
414 got_packidx_close(struct got_packidx *packidx)
415 {
416 	const struct got_error *err = NULL;
417 
418 	free(packidx->path_packidx);
419 	if (packidx->map) {
420 		if (munmap(packidx->map, packidx->len) == -1)
421 			err = got_error_from_errno("munmap");
422 	} else {
423 		free(packidx->hdr.magic);
424 		free(packidx->hdr.version);
425 		free(packidx->hdr.fanout_table);
426 		free(packidx->hdr.sorted_ids);
427 		free(packidx->hdr.crc32);
428 		free(packidx->hdr.offsets);
429 		free(packidx->hdr.large_offsets);
430 		free(packidx->hdr.trailer);
431 	}
432 	if (close(packidx->fd) == -1 && err == NULL)
433 		err = got_error_from_errno("close");
434 	free(packidx->sorted_offsets);
435 	free(packidx->sorted_large_offsets);
436 	free(packidx);
437 
438 	return err;
439 }
440 
441 const struct got_error *
442 got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
443 {
444 	size_t size;
445 
446 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
447 	size = strlen(path_packidx) + 2;
448 	if (size < GOT_PACKFILE_NAMELEN + 1)
449 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
450 
451 	*path_packfile = malloc(size);
452 	if (*path_packfile == NULL)
453 		return got_error_from_errno("malloc");
454 
455 	/* Copy up to and excluding ".idx". */
456 	if (strlcpy(*path_packfile, path_packidx,
457 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
458 		return got_error(GOT_ERR_NO_SPACE);
459 
460 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
461 		return got_error(GOT_ERR_NO_SPACE);
462 
463 	return NULL;
464 }
465 
466 off_t
467 got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
468 {
469 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
470 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
471 		uint64_t loffset;
472 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
473 		if (idx < 0 || idx >= packidx->nlargeobj ||
474 		    packidx->hdr.large_offsets == NULL)
475 			return -1;
476 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
477 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
478 	}
479 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
480 }
481 
482 int
483 got_packidx_get_object_idx(struct got_packidx *packidx,
484     struct got_object_id *id)
485 {
486 	u_int8_t id0 = id->sha1[0];
487 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
488 	int left = 0, right = totobj - 1;
489 
490 	if (id0 > 0)
491 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
492 
493 	while (left <= right) {
494 		struct got_packidx_object_id *oid;
495 		int i, cmp;
496 
497 		i = ((left + right) / 2);
498 		oid = &packidx->hdr.sorted_ids[i];
499 		cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
500 		if (cmp == 0)
501 			return i;
502 		else if (cmp > 0)
503 			left = i + 1;
504 		else if (cmp < 0)
505 			right = i - 1;
506 	}
507 
508 	return -1;
509 }
510 
511 static int
512 offset_cmp(const void *pa, const void *pb)
513 {
514 	const struct got_pack_offset_index *a, *b;
515 
516 	a = (const struct got_pack_offset_index *)pa;
517 	b = (const struct got_pack_offset_index *)pb;
518 
519 	if (a->offset < b->offset)
520 		return -1;
521 	else if (a->offset > b->offset)
522 		return 1;
523 
524 	return 0;
525 }
526 
527 static int
528 large_offset_cmp(const void *pa, const void *pb)
529 {
530 	const struct got_pack_large_offset_index *a, *b;
531 
532 	a = (const struct got_pack_large_offset_index *)pa;
533 	b = (const struct got_pack_large_offset_index *)pb;
534 
535 	if (a->offset < b->offset)
536 		return -1;
537 	else if (a->offset > b->offset)
538 		return 1;
539 
540 	return 0;
541 }
542 
543 static const struct got_error *
544 build_offset_index(struct got_packidx *p)
545 {
546 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
547 	unsigned int i, j, k;
548 
549 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
550 	    sizeof(p->sorted_offsets[0]));
551 	if (p->sorted_offsets == NULL)
552 		return got_error_from_errno("calloc");
553 
554 	if (p->nlargeobj > 0) {
555 		p->sorted_large_offsets = calloc(p->nlargeobj,
556 		    sizeof(p->sorted_large_offsets[0]));
557 		if (p->sorted_large_offsets == NULL)
558 			return got_error_from_errno("calloc");
559 	}
560 
561 	j = 0;
562 	k = 0;
563 	for (i = 0; i < nobj; i++) {
564 		uint32_t offset = be32toh(p->hdr.offsets[i]);
565 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
566 			uint64_t loffset;
567 			uint32_t idx;
568 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
569 			if (idx >= p->nlargeobj ||
570 			    p->nlargeobj == 0 ||
571 			    p->hdr.large_offsets == NULL)
572 				return got_error(GOT_ERR_BAD_PACKIDX);
573 			loffset = be64toh(p->hdr.large_offsets[idx]);
574 			p->sorted_large_offsets[j].offset = loffset;
575 			p->sorted_large_offsets[j].idx = i;
576 			j++;
577 		} else {
578 			p->sorted_offsets[k].offset = offset;
579 			p->sorted_offsets[k].idx = i;
580 			k++;
581 		}
582 	}
583 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
584 		return got_error(GOT_ERR_BAD_PACKIDX);
585 
586 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
587 	    sizeof(p->sorted_offsets[0]), offset_cmp);
588 
589 	if (p->sorted_large_offsets)
590 		qsort(p->sorted_large_offsets, p->nlargeobj,
591 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
592 
593 	return NULL;
594 }
595 
596 const struct got_error *
597 got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
598 {
599 	const struct got_error *err;
600 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
601 	int i, left, right;
602 
603 	*idx = -1;
604 
605 	if (packidx->sorted_offsets == NULL) {
606 		err = build_offset_index(packidx);
607 		if (err)
608 			return err;
609 	}
610 
611 	if (offset >= 0x7fffffff) {
612 		uint64_t lo;
613 		left = 0, right = packidx->nlargeobj - 1;
614 		while (left <= right) {
615 			i = ((left + right) / 2);
616 			lo = packidx->sorted_large_offsets[i].offset;
617 			if (lo == offset) {
618 				*idx = packidx->sorted_large_offsets[i].idx;
619 				break;
620 			} else if (offset > lo)
621 				left = i + 1;
622 			else if (offset < lo)
623 				right = i - 1;
624 		}
625 	} else {
626 		uint32_t o;
627 		left = 0, right = totobj - packidx->nlargeobj - 1;
628 		while (left <= right) {
629 			i = ((left + right) / 2);
630 			o = packidx->sorted_offsets[i].offset;
631 			if (o == offset) {
632 				*idx = packidx->sorted_offsets[i].idx;
633 				break;
634 			} else if (offset > o)
635 				left = i + 1;
636 			else if (offset < o)
637 				right = i - 1;
638 		}
639 	}
640 
641 	return NULL;
642 }
643 
644 const struct got_error *
645 got_packidx_get_object_id(struct got_object_id *id,
646     struct got_packidx *packidx, int idx)
647 {
648 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
649 	struct got_packidx_object_id *oid;
650 
651 	if (idx < 0 || idx >= totobj)
652 		return got_error(GOT_ERR_NO_OBJ);
653 
654 	oid = &packidx->hdr.sorted_ids[idx];
655 	memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
656 	return NULL;
657 }
658 
659 const struct got_error *
660 got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
661     struct got_packidx *packidx, const char *id_str_prefix)
662 {
663 	const struct got_error *err = NULL;
664 	u_int8_t id0;
665 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
666 	char hex[3];
667 	size_t prefix_len = strlen(id_str_prefix);
668 	struct got_packidx_object_id *oid;
669 	uint32_t i = 0;
670 
671 	STAILQ_INIT(matched_ids);
672 
673 	if (prefix_len < 2)
674 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
675 
676 	hex[0] = id_str_prefix[0];
677 	hex[1] = id_str_prefix[1];
678 	hex[2] = '\0';
679 	if (!got_parse_xdigit(&id0, hex))
680 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
681 
682 	if (id0 > 0)
683 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
684 	oid = &packidx->hdr.sorted_ids[i];
685 	while (i < totobj && oid->sha1[0] == id0) {
686 		char id_str[SHA1_DIGEST_STRING_LENGTH];
687 		struct got_object_qid *qid;
688 		int cmp;
689 
690 		if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
691 		        return got_error(GOT_ERR_NO_SPACE);
692 
693 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
694 		if (cmp < 0) {
695 			oid = &packidx->hdr.sorted_ids[++i];
696 			continue;
697 		} else if (cmp > 0)
698 			break;
699 
700 		err = got_object_qid_alloc_partial(&qid);
701 		if (err)
702 			break;
703 		memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
704 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
705 
706 		oid = &packidx->hdr.sorted_ids[++i];
707 	}
708 
709 	if (err)
710 		got_object_id_queue_free(matched_ids);
711 	return err;
712 }
713 
714 static void
715 set_max_datasize(void)
716 {
717 	struct rlimit rl;
718 
719 	if (getrlimit(RLIMIT_DATA, &rl) != 0)
720 		return;
721 
722 	rl.rlim_cur = rl.rlim_max;
723 	setrlimit(RLIMIT_DATA, &rl);
724 }
725 
726 const struct got_error *
727 got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
728 {
729 	const struct got_error *err = NULL;
730 	int imsg_fds[2];
731 	pid_t pid;
732 	struct imsgbuf *ibuf;
733 
734 	ibuf = calloc(1, sizeof(*ibuf));
735 	if (ibuf == NULL)
736 		return got_error_from_errno("calloc");
737 
738 	pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
739 	if (pack->privsep_child == NULL) {
740 		err = got_error_from_errno("calloc");
741 		free(ibuf);
742 		return err;
743 	}
744 	pack->child_has_tempfiles = 0;
745 	pack->child_has_delta_outfd = 0;
746 
747 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
748 		err = got_error_from_errno("socketpair");
749 		goto done;
750 	}
751 
752 	pid = fork();
753 	if (pid == -1) {
754 		err = got_error_from_errno("fork");
755 		goto done;
756 	} else if (pid == 0) {
757 		set_max_datasize();
758 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
759 		    pack->path_packfile);
760 		/* not reached */
761 	}
762 
763 	if (close(imsg_fds[1]) == -1)
764 		return got_error_from_errno("close");
765 	pack->privsep_child->imsg_fd = imsg_fds[0];
766 	pack->privsep_child->pid = pid;
767 	imsg_init(ibuf, imsg_fds[0]);
768 	pack->privsep_child->ibuf = ibuf;
769 
770 	err = got_privsep_init_pack_child(ibuf, pack, packidx);
771 	if (err) {
772 		const struct got_error *child_err;
773 		err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
774 		child_err = got_privsep_wait_for_child(
775 		    pack->privsep_child->pid);
776 		if (child_err && err == NULL)
777 			err = child_err;
778 	}
779 done:
780 	if (err) {
781 		free(ibuf);
782 		free(pack->privsep_child);
783 		pack->privsep_child = NULL;
784 	}
785 	return err;
786 }
787 
788 static const struct got_error *
789 pack_stop_privsep_child(struct got_pack *pack)
790 {
791 	const struct got_error *err = NULL, *close_err = NULL;
792 
793 	if (pack->privsep_child == NULL)
794 		return NULL;
795 
796 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
797 	if (err)
798 		return err;
799 	if (close(pack->privsep_child->imsg_fd) == -1)
800 		close_err = got_error_from_errno("close");
801 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
802 	if (close_err && err == NULL)
803 		err = close_err;
804 	imsg_clear(pack->privsep_child->ibuf);
805 	free(pack->privsep_child->ibuf);
806 	free(pack->privsep_child);
807 	pack->privsep_child = NULL;
808 	return err;
809 }
810 
811 const struct got_error *
812 got_pack_close(struct got_pack *pack)
813 {
814 	const struct got_error *err = NULL;
815 
816 	err = pack_stop_privsep_child(pack);
817 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
818 		err = got_error_from_errno("munmap");
819 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
820 		err = got_error_from_errno("close");
821 	pack->fd = -1;
822 	free(pack->path_packfile);
823 	pack->path_packfile = NULL;
824 	pack->filesize = 0;
825 	if (pack->delta_cache) {
826 		got_delta_cache_free(pack->delta_cache);
827 		pack->delta_cache = NULL;
828 	}
829 
830 	/*
831 	 * Leave accumfd and basefd alone. They are managed by the
832 	 * repository layer and can be reused.
833 	 */
834 
835 	return err;
836 }
837 
838 const struct got_error *
839 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
840     struct got_pack *pack, off_t offset)
841 {
842 	uint8_t t = 0;
843 	uint64_t s = 0;
844 	uint8_t sizeN;
845 	size_t mapoff = 0;
846 	int i = 0;
847 
848 	*len = 0;
849 
850 	if (offset >= pack->filesize)
851 		return got_error(GOT_ERR_PACK_OFFSET);
852 
853 	if (pack->map) {
854 		mapoff = (size_t)offset;
855 	} else {
856 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
857 			return got_error_from_errno("lseek");
858 	}
859 
860 	do {
861 		/* We do not support size values which don't fit in 64 bit. */
862 		if (i > 9)
863 			return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
864 			    "packfile offset %lld", (long long)offset);
865 
866 		if (pack->map) {
867 			if (mapoff + sizeof(sizeN) >= pack->filesize)
868 				return got_error(GOT_ERR_BAD_PACKFILE);
869 			sizeN = *(pack->map + mapoff);
870 			mapoff += sizeof(sizeN);
871 		} else {
872 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
873 			if (n < 0)
874 				return got_error_from_errno("read");
875 			if (n != sizeof(sizeN))
876 				return got_error(GOT_ERR_BAD_PACKFILE);
877 		}
878 		*len += sizeof(sizeN);
879 
880 		if (i == 0) {
881 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
882 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
883 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
884 		} else {
885 			size_t shift = 4 + 7 * (i - 1);
886 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
887 		}
888 		i++;
889 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
890 
891 	*type = t;
892 	*size = s;
893 	return NULL;
894 }
895 
896 static const struct got_error *
897 open_plain_object(struct got_object **obj, struct got_object_id *id,
898     uint8_t type, off_t offset, size_t size, int idx)
899 {
900 	*obj = calloc(1, sizeof(**obj));
901 	if (*obj == NULL)
902 		return got_error_from_errno("calloc");
903 
904 	(*obj)->type = type;
905 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
906 	(*obj)->pack_idx = idx;
907 	(*obj)->hdrlen = 0;
908 	(*obj)->size = size;
909 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
910 	(*obj)->pack_offset = offset;
911 
912 	return NULL;
913 }
914 
915 static const struct got_error *
916 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
917     off_t delta_offset)
918 {
919 	int64_t o = 0;
920 	uint8_t offN;
921 	int i = 0;
922 
923 	*offset = 0;
924 	*len = 0;
925 
926 	do {
927 		/* We do not support offset values which don't fit in 64 bit. */
928 		if (i > 8)
929 			return got_error(GOT_ERR_NO_SPACE);
930 
931 		if (pack->map) {
932 			size_t mapoff;
933 			mapoff = (size_t)delta_offset + *len;
934 			if (mapoff + sizeof(offN) >= pack->filesize)
935 				return got_error(GOT_ERR_PACK_OFFSET);
936 			offN = *(pack->map + mapoff);
937 		} else {
938 			ssize_t n;
939 			n = read(pack->fd, &offN, sizeof(offN));
940 			if (n < 0)
941 				return got_error_from_errno("read");
942 			if (n != sizeof(offN))
943 				return got_error(GOT_ERR_BAD_PACKFILE);
944 		}
945 		*len += sizeof(offN);
946 
947 		if (i == 0)
948 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
949 		else {
950 			o++;
951 			o <<= 7;
952 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
953 		}
954 		i++;
955 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
956 
957 	*offset = o;
958 	return NULL;
959 }
960 
961 const struct got_error *
962 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
963     struct got_pack *pack, off_t offset, size_t tslen)
964 {
965 	const struct got_error *err;
966 	int64_t negoffset;
967 	size_t negofflen;
968 
969 	*len = 0;
970 
971 	err = parse_negative_offset(&negoffset, &negofflen, pack,
972 	    offset + tslen);
973 	if (err)
974 		return err;
975 
976 	/* Compute the base object's offset (must be in the same pack file). */
977 	*base_offset = (offset - negoffset);
978 	if (*base_offset <= 0)
979 		return got_error(GOT_ERR_BAD_PACKFILE);
980 
981 	*len = negofflen;
982 	return NULL;
983 }
984 
985 static const struct got_error *
986 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
987     size_t *delta_compressed_len, size_t delta_data_offset,
988     struct got_pack *pack)
989 {
990 	const struct got_error *err = NULL;
991 	size_t consumed = 0;
992 
993 	if (pack->map) {
994 		if (delta_data_offset >= pack->filesize)
995 			return got_error(GOT_ERR_PACK_OFFSET);
996 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
997 		    &consumed, NULL, pack->map, delta_data_offset,
998 		    pack->filesize - delta_data_offset);
999 		if (err)
1000 			return err;
1001 	} else {
1002 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1003 			return got_error_from_errno("lseek");
1004 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1005 		    &consumed, NULL, 0, pack->fd);
1006 		if (err)
1007 			return err;
1008 	}
1009 
1010 	if (delta_compressed_len)
1011 		*delta_compressed_len = consumed;
1012 
1013 	return NULL;
1014 }
1015 
1016 static const struct got_error *
1017 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1018     int delta_type, size_t delta_size, size_t delta_data_offset)
1019 {
1020 	struct got_delta *delta;
1021 
1022 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1023 	    delta_data_offset);
1024 	if (delta == NULL)
1025 		return got_error_from_errno("got_delta_open");
1026 	/* delta is freed in got_object_close() */
1027 	deltas->nentries++;
1028 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1029 	return NULL;
1030 }
1031 
1032 static const struct got_error *
1033 resolve_offset_delta(struct got_delta_chain *deltas,
1034     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1035     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1036 
1037 {
1038 	const struct got_error *err;
1039 	off_t base_offset;
1040 	uint8_t base_type;
1041 	uint64_t base_size;
1042 	size_t base_tslen;
1043 	off_t delta_data_offset;
1044 	size_t consumed;
1045 
1046 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1047 	    delta_offset, tslen);
1048 	if (err)
1049 		return err;
1050 
1051 	delta_data_offset = delta_offset + tslen + consumed;
1052 	if (delta_data_offset >= pack->filesize)
1053 		return got_error(GOT_ERR_PACK_OFFSET);
1054 
1055 	if (pack->map == NULL) {
1056 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1057 		if (delta_data_offset == -1)
1058 			return got_error_from_errno("lseek");
1059 	}
1060 
1061 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1062 	    delta_data_offset);
1063 	if (err)
1064 		return err;
1065 
1066 	/* An offset delta must be in the same packfile. */
1067 	if (base_offset >= pack->filesize)
1068 		return got_error(GOT_ERR_PACK_OFFSET);
1069 
1070 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1071 	    &base_tslen, pack, base_offset);
1072 	if (err)
1073 		return err;
1074 
1075 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1076 	    base_tslen, base_type, base_size, recursion - 1);
1077 }
1078 
1079 const struct got_error *
1080 got_pack_parse_ref_delta(struct got_object_id *id,
1081     struct got_pack *pack, off_t delta_offset, int tslen)
1082 {
1083 	if (pack->map) {
1084 		size_t mapoff = delta_offset + tslen;
1085 		if (mapoff + sizeof(*id) >= pack->filesize)
1086 			return got_error(GOT_ERR_PACK_OFFSET);
1087 		memcpy(id, pack->map + mapoff, sizeof(*id));
1088 	} else {
1089 		ssize_t n;
1090 		n = read(pack->fd, id, sizeof(*id));
1091 		if (n < 0)
1092 			return got_error_from_errno("read");
1093 		if (n != sizeof(*id))
1094 			return got_error(GOT_ERR_BAD_PACKFILE);
1095 	}
1096 
1097 	return NULL;
1098 }
1099 
1100 static const struct got_error *
1101 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1102     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1103     size_t delta_size, unsigned int recursion)
1104 {
1105 	const struct got_error *err;
1106 	struct got_object_id id;
1107 	int idx;
1108 	off_t base_offset;
1109 	uint8_t base_type;
1110 	uint64_t base_size;
1111 	size_t base_tslen;
1112 	off_t delta_data_offset;
1113 
1114 	if (delta_offset + tslen >= pack->filesize)
1115 		return got_error(GOT_ERR_PACK_OFFSET);
1116 
1117 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1118 	if (err)
1119 		return err;
1120 	if (pack->map) {
1121 		delta_data_offset = delta_offset + tslen + sizeof(id);
1122 	} else {
1123 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1124 		if (delta_data_offset == -1)
1125 			return got_error_from_errno("lseek");
1126 	}
1127 
1128 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1129 	    delta_data_offset);
1130 	if (err)
1131 		return err;
1132 
1133 	/* Delta base must be in the same pack file. */
1134 	idx = got_packidx_get_object_idx(packidx, &id);
1135 	if (idx == -1)
1136 		return got_error(GOT_ERR_NO_OBJ);
1137 
1138 	base_offset = got_packidx_get_object_offset(packidx, idx);
1139 	if (base_offset == -1)
1140 		return got_error(GOT_ERR_BAD_PACKIDX);
1141 
1142 	if (base_offset >= pack->filesize)
1143 		return got_error(GOT_ERR_PACK_OFFSET);
1144 
1145 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1146 	    &base_tslen, pack, base_offset);
1147 	if (err)
1148 		return err;
1149 
1150 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1151 	    base_tslen, base_type, base_size, recursion - 1);
1152 }
1153 
1154 const struct got_error *
1155 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1156     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1157     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1158 {
1159 	const struct got_error *err = NULL;
1160 
1161 	if (--recursion == 0)
1162 		return got_error(GOT_ERR_RECURSION);
1163 
1164 	switch (delta_type) {
1165 	case GOT_OBJ_TYPE_COMMIT:
1166 	case GOT_OBJ_TYPE_TREE:
1167 	case GOT_OBJ_TYPE_BLOB:
1168 	case GOT_OBJ_TYPE_TAG:
1169 		/* Plain types are the final delta base. Recursion ends. */
1170 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1171 		    delta_size, 0);
1172 		break;
1173 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1174 		err = resolve_offset_delta(deltas, packidx, pack,
1175 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1176 		break;
1177 	case GOT_OBJ_TYPE_REF_DELTA:
1178 		err = resolve_ref_delta(deltas, packidx, pack,
1179 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1180 		break;
1181 	default:
1182 		return got_error(GOT_ERR_OBJ_TYPE);
1183 	}
1184 
1185 	return err;
1186 }
1187 
1188 static const struct got_error *
1189 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1190     struct got_pack *pack, struct got_object_id *id, off_t offset,
1191     size_t tslen, int delta_type, size_t delta_size, int idx)
1192 {
1193 	const struct got_error *err = NULL;
1194 	int resolved_type;
1195 
1196 	*obj = calloc(1, sizeof(**obj));
1197 	if (*obj == NULL)
1198 		return got_error_from_errno("calloc");
1199 
1200 	(*obj)->flags = 0;
1201 	(*obj)->hdrlen = 0;
1202 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1203 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1204 	(*obj)->pack_offset = offset + tslen;
1205 
1206 	STAILQ_INIT(&(*obj)->deltas.entries);
1207 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1208 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1209 	(*obj)->pack_idx = idx;
1210 
1211 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1212 	    offset, tslen, delta_type, delta_size,
1213 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1214 	if (err)
1215 		goto done;
1216 
1217 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1218 	if (err)
1219 		goto done;
1220 	(*obj)->type = resolved_type;
1221 done:
1222 	if (err) {
1223 		got_object_close(*obj);
1224 		*obj = NULL;
1225 	}
1226 	return err;
1227 }
1228 
1229 const struct got_error *
1230 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1231     struct got_packidx *packidx, int idx, struct got_object_id *id)
1232 {
1233 	const struct got_error *err = NULL;
1234 	off_t offset;
1235 	uint8_t type;
1236 	uint64_t size;
1237 	size_t tslen;
1238 
1239 	*obj = NULL;
1240 
1241 	offset = got_packidx_get_object_offset(packidx, idx);
1242 	if (offset == -1)
1243 		return got_error(GOT_ERR_BAD_PACKIDX);
1244 
1245 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1246 	    pack, offset);
1247 	if (err)
1248 		return err;
1249 
1250 	switch (type) {
1251 	case GOT_OBJ_TYPE_COMMIT:
1252 	case GOT_OBJ_TYPE_TREE:
1253 	case GOT_OBJ_TYPE_BLOB:
1254 	case GOT_OBJ_TYPE_TAG:
1255 		err = open_plain_object(obj, id, type, offset + tslen,
1256 		    size, idx);
1257 		break;
1258 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1259 	case GOT_OBJ_TYPE_REF_DELTA:
1260 		err = open_delta_object(obj, packidx, pack, id, offset,
1261 		    tslen, type, size, idx);
1262 		break;
1263 	default:
1264 		err = got_error(GOT_ERR_OBJ_TYPE);
1265 		break;
1266 	}
1267 
1268 	return err;
1269 }
1270 
1271 const struct got_error *
1272 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1273     struct got_delta_chain *deltas, struct got_pack *pack)
1274 {
1275 	struct got_delta *delta;
1276 	uint64_t base_size = 0, result_size = 0;
1277 
1278 	*max_size = 0;
1279 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
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 			const struct got_error *err;
1286 			uint8_t *delta_buf = NULL;
1287 			size_t delta_len;
1288 			int cached = 1;
1289 
1290 			if (pack->delta_cache) {
1291 				got_delta_cache_get(&delta_buf, &delta_len,
1292 				    pack->delta_cache, delta->data_offset);
1293 			}
1294 			if (delta_buf == NULL) {
1295 				cached = 0;
1296 				err = read_delta_data(&delta_buf, &delta_len,
1297 				    NULL, delta->data_offset, pack);
1298 				if (err)
1299 					return err;
1300 			}
1301 			if (pack->delta_cache && !cached) {
1302 				err = got_delta_cache_add(pack->delta_cache,
1303 				    delta->data_offset, delta_buf, delta_len);
1304 				if (err == NULL)
1305 					cached = 1;
1306 				else if (err->code != GOT_ERR_NO_SPACE) {
1307 					free(delta_buf);
1308 					return err;
1309 				}
1310 			}
1311 			err = got_delta_get_sizes(&base_size, &result_size,
1312 			    delta_buf, delta_len);
1313 			if (!cached)
1314 				free(delta_buf);
1315 			if (err)
1316 				return err;
1317 		} else
1318 			base_size = delta->size;
1319 		if (base_size > *max_size)
1320 			*max_size = base_size;
1321 		if (result_size > *max_size)
1322 			*max_size = result_size;
1323 	}
1324 
1325 	return NULL;
1326 }
1327 
1328 const struct got_error *
1329 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1330     struct got_pack *pack)
1331 {
1332 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1333 		return got_error(GOT_ERR_OBJ_TYPE);
1334 
1335 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1336 }
1337 
1338 const struct got_error *
1339 got_pack_dump_delta_chain_to_file(size_t *result_size,
1340     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1341     FILE *base_file, FILE *accum_file)
1342 {
1343 	const struct got_error *err = NULL;
1344 	struct got_delta *delta;
1345 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1346 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1347 	/* We process small enough files entirely in memory for speed. */
1348 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1349 	uint64_t max_size = 0;
1350 	int n = 0;
1351 
1352 	*result_size = 0;
1353 
1354 	if (STAILQ_EMPTY(&deltas->entries))
1355 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1356 
1357 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1358 		return got_error_from_errno("fseeko");
1359 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1360 		return got_error_from_errno("fseeko");
1361 
1362 	/* Deltas are ordered in ascending order. */
1363 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1364 		uint8_t *delta_buf = NULL;
1365 		uint64_t base_size, result_size = 0;
1366 		int cached = 1;
1367 		if (n == 0) {
1368 			size_t mapoff;
1369 			off_t delta_data_offset;
1370 
1371 			/* Plain object types are the delta base. */
1372 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1373 			    delta->type != GOT_OBJ_TYPE_TREE &&
1374 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1375 			    delta->type != GOT_OBJ_TYPE_TAG) {
1376 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1377 				goto done;
1378 			}
1379 
1380 			delta_data_offset = delta->offset + delta->tslen;
1381 			if (delta_data_offset >= pack->filesize) {
1382 				err = got_error(GOT_ERR_PACK_OFFSET);
1383 				goto done;
1384 			}
1385 			if (pack->map == NULL) {
1386 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1387 				    == -1) {
1388 					err = got_error_from_errno("lseek");
1389 					goto done;
1390 				}
1391 			}
1392 			if (delta->size > max_size)
1393 				max_size = delta->size;
1394 			if (max_size > max_bufsize) {
1395 				if (pack->map) {
1396 					mapoff = (size_t)delta_data_offset;
1397 					err = got_inflate_to_file_mmap(
1398 					    &base_bufsz, NULL, NULL, pack->map,
1399 					    mapoff, pack->filesize - mapoff,
1400 					    base_file);
1401 				} else
1402 					err = got_inflate_to_file_fd(
1403 					    &base_bufsz, NULL, NULL, pack->fd,
1404 					    base_file);
1405 			} else {
1406 				accum_buf = malloc(max_size);
1407 				if (accum_buf == NULL) {
1408 					err = got_error_from_errno("malloc");
1409 					goto done;
1410 				}
1411 				accum_bufsz = max_size;
1412 				if (pack->map) {
1413 					mapoff = (size_t)delta_data_offset;
1414 					err = got_inflate_to_mem_mmap(&base_buf,
1415 					    &base_bufsz, NULL, NULL,
1416 					    pack->map, mapoff,
1417 					    pack->filesize - mapoff);
1418 				} else
1419 					err = got_inflate_to_mem_fd(&base_buf,
1420 					    &base_bufsz, NULL, NULL, max_size,
1421 					    pack->fd);
1422 			}
1423 			if (err)
1424 				goto done;
1425 			n++;
1426 			if (base_buf == NULL)
1427 				rewind(base_file);
1428 			continue;
1429 		}
1430 
1431 		if (pack->delta_cache) {
1432 			got_delta_cache_get(&delta_buf, &delta_len,
1433 			    pack->delta_cache, delta->data_offset);
1434 		}
1435 		if (delta_buf == NULL) {
1436 			cached = 0;
1437 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1438 			    delta->data_offset, pack);
1439 			if (err)
1440 				goto done;
1441 		}
1442 		if (pack->delta_cache && !cached) {
1443 			err = got_delta_cache_add(pack->delta_cache,
1444 			    delta->data_offset, delta_buf, delta_len);
1445 			if (err == NULL)
1446 				cached = 1;
1447 			else if (err->code != GOT_ERR_NO_SPACE) {
1448 				free(delta_buf);
1449 				goto done;
1450 			}
1451 		}
1452 
1453 		err = got_delta_get_sizes(&base_size, &result_size,
1454 		    delta_buf, delta_len);
1455 		if (err)
1456 			goto done;
1457 		if (base_size > max_size)
1458 			max_size = base_size;
1459 		if (result_size > max_size)
1460 			max_size = result_size;
1461 
1462 		if (base_buf && max_size > max_bufsize) {
1463 			/* Switch from buffers to temporary files. */
1464 			size_t w = fwrite(base_buf, 1, base_bufsz,
1465 			    base_file);
1466 			if (w != base_bufsz) {
1467 				err = got_ferror(outfile, GOT_ERR_IO);
1468 				goto done;
1469 			}
1470 			free(base_buf);
1471 			base_buf = NULL;
1472 			free(accum_buf);
1473 			accum_buf = NULL;
1474 		}
1475 
1476 		if (base_buf && max_size > base_bufsz) {
1477 			uint8_t *p = realloc(base_buf, max_size);
1478 			if (p == NULL) {
1479 				err = got_error_from_errno("realloc");
1480 				goto done;
1481 			}
1482 			base_buf = p;
1483 			base_bufsz = max_size;
1484 		}
1485 
1486 		if (accum_buf && max_size > accum_bufsz) {
1487 			uint8_t *p = realloc(accum_buf, max_size);
1488 			if (p == NULL) {
1489 				err = got_error_from_errno("realloc");
1490 				goto done;
1491 			}
1492 			accum_buf = p;
1493 			accum_bufsz = max_size;
1494 		}
1495 
1496 		if (base_buf) {
1497 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1498 			    delta_buf, delta_len, accum_buf,
1499 			    &accum_size, max_size);
1500 			n++;
1501 		} else {
1502 			err = got_delta_apply(base_file, delta_buf,
1503 			    delta_len,
1504 			    /* Final delta application writes to output file. */
1505 			    ++n < deltas->nentries ? accum_file : outfile,
1506 			    &accum_size);
1507 		}
1508 		if (!cached)
1509 			free(delta_buf);
1510 		if (err)
1511 			goto done;
1512 
1513 		if (n < deltas->nentries) {
1514 			/* Accumulated delta becomes the new base. */
1515 			if (base_buf) {
1516 				uint8_t *tmp = accum_buf;
1517 				size_t tmp_size = accum_bufsz;
1518 				accum_buf = base_buf;
1519 				accum_bufsz = base_bufsz;
1520 				base_buf = tmp;
1521 				base_bufsz = tmp_size;
1522 			} else {
1523 				FILE *tmp = accum_file;
1524 				accum_file = base_file;
1525 				base_file = tmp;
1526 				rewind(base_file);
1527 				rewind(accum_file);
1528 			}
1529 		}
1530 	}
1531 
1532 done:
1533 	free(base_buf);
1534 	if (accum_buf) {
1535 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1536 		free(accum_buf);
1537 		if (len != accum_size)
1538 			err = got_ferror(outfile, GOT_ERR_IO);
1539 	}
1540 	rewind(outfile);
1541 	if (err == NULL)
1542 		*result_size = accum_size;
1543 	return err;
1544 }
1545 
1546 const struct got_error *
1547 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1548     struct got_delta_chain *deltas, struct got_pack *pack)
1549 {
1550 	const struct got_error *err = NULL;
1551 	struct got_delta *delta;
1552 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1553 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1554 	uint64_t max_size = 0;
1555 	int n = 0;
1556 
1557 	*outbuf = NULL;
1558 	*outlen = 0;
1559 
1560 	if (STAILQ_EMPTY(&deltas->entries))
1561 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1562 
1563 	/* Deltas are ordered in ascending order. */
1564 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1565 		uint8_t *delta_buf = NULL;
1566 		uint64_t base_size, result_size = 0;
1567 		int cached = 1;
1568 		if (n == 0) {
1569 			size_t delta_data_offset;
1570 
1571 			/* Plain object types are the delta base. */
1572 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1573 			    delta->type != GOT_OBJ_TYPE_TREE &&
1574 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1575 			    delta->type != GOT_OBJ_TYPE_TAG) {
1576 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1577 				goto done;
1578 			}
1579 
1580 			delta_data_offset = delta->offset + delta->tslen;
1581 			if (delta_data_offset >= pack->filesize) {
1582 				err = got_error(GOT_ERR_PACK_OFFSET);
1583 				goto done;
1584 			}
1585 
1586 			if (delta->size > max_size)
1587 				max_size = delta->size;
1588 
1589 			if (pack->map) {
1590 				size_t mapoff = (size_t)delta_data_offset;
1591 				err = got_inflate_to_mem_mmap(&base_buf,
1592 				    &base_bufsz, NULL, NULL, pack->map,
1593 				    mapoff, pack->filesize - mapoff);
1594 			} else {
1595 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1596 				    == -1) {
1597 					err = got_error_from_errno("lseek");
1598 					goto done;
1599 				}
1600 				err = got_inflate_to_mem_fd(&base_buf,
1601 				    &base_bufsz, NULL, NULL, max_size,
1602 				    pack->fd);
1603 			}
1604 			if (err)
1605 				goto done;
1606 			n++;
1607 			continue;
1608 		}
1609 
1610 		if (pack->delta_cache) {
1611 			got_delta_cache_get(&delta_buf, &delta_len,
1612 			    pack->delta_cache, delta->data_offset);
1613 		}
1614 		if (delta_buf == NULL) {
1615 			cached = 0;
1616 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1617 			    delta->data_offset, pack);
1618 			if (err)
1619 				goto done;
1620 		}
1621 		if (pack->delta_cache && !cached) {
1622 			err = got_delta_cache_add(pack->delta_cache,
1623 			    delta->data_offset, delta_buf, delta_len);
1624 			if (err == NULL)
1625 				cached = 1;
1626 			else if (err->code != GOT_ERR_NO_SPACE) {
1627 				free(delta_buf);
1628 				goto done;
1629 			}
1630 		}
1631 
1632 		err = got_delta_get_sizes(&base_size, &result_size,
1633 		    delta_buf, delta_len);
1634 		if (err)
1635 			goto done;
1636 		if (base_size > max_size)
1637 			max_size = base_size;
1638 		if (result_size > max_size)
1639 			max_size = result_size;
1640 
1641 		if (max_size > base_bufsz) {
1642 			uint8_t *p = realloc(base_buf, max_size);
1643 			if (p == NULL) {
1644 				err = got_error_from_errno("realloc");
1645 				goto done;
1646 			}
1647 			base_buf = p;
1648 			base_bufsz = max_size;
1649 		}
1650 
1651 		if (max_size > accum_bufsz) {
1652 			uint8_t *p = realloc(accum_buf, max_size);
1653 			if (p == NULL) {
1654 				err = got_error_from_errno("realloc");
1655 				goto done;
1656 			}
1657 			accum_buf = p;
1658 			accum_bufsz = max_size;
1659 		}
1660 
1661 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1662 		    delta_buf, delta_len, accum_buf,
1663 		    &accum_size, max_size);
1664 		if (!cached)
1665 			free(delta_buf);
1666 		n++;
1667 		if (err)
1668 			goto done;
1669 
1670 		if (n < deltas->nentries) {
1671 			/* Accumulated delta becomes the new base. */
1672 			uint8_t *tmp = accum_buf;
1673 			size_t tmp_size = accum_bufsz;
1674 			accum_buf = base_buf;
1675 			accum_bufsz = base_bufsz;
1676 			base_buf = tmp;
1677 			base_bufsz = tmp_size;
1678 		}
1679 	}
1680 
1681 done:
1682 	free(base_buf);
1683 	if (err) {
1684 		free(accum_buf);
1685 		*outbuf = NULL;
1686 		*outlen = 0;
1687 	} else {
1688 		*outbuf = accum_buf;
1689 		*outlen = accum_size;
1690 	}
1691 	return err;
1692 }
1693 
1694 const struct got_error *
1695 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1696     FILE *outfile, FILE *base_file, FILE *accum_file)
1697 {
1698 	const struct got_error *err = NULL;
1699 
1700 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1701 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1702 
1703 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1704 		if (obj->pack_offset >= pack->filesize)
1705 			return got_error(GOT_ERR_PACK_OFFSET);
1706 
1707 		if (pack->map) {
1708 			size_t mapoff = (size_t)obj->pack_offset;
1709 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1710 			    pack->map, mapoff, pack->filesize - mapoff,
1711 			    outfile);
1712 		} else {
1713 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1714 				return got_error_from_errno("lseek");
1715 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1716 			    pack->fd, outfile);
1717 		}
1718 	} else
1719 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1720 		    &obj->deltas, pack, outfile, base_file, accum_file);
1721 
1722 	return err;
1723 }
1724 
1725 const struct got_error *
1726 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1727     struct got_object *obj, struct got_pack *pack)
1728 {
1729 	const struct got_error *err = NULL;
1730 
1731 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1732 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1733 
1734 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1735 		if (obj->pack_offset >= pack->filesize)
1736 			return got_error(GOT_ERR_PACK_OFFSET);
1737 		if (pack->map) {
1738 			size_t mapoff = (size_t)obj->pack_offset;
1739 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1740 			    pack->map, mapoff, pack->filesize - mapoff);
1741 		} else {
1742 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1743 				return got_error_from_errno("lseek");
1744 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1745 			    obj->size, pack->fd);
1746 		}
1747 	} else
1748 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1749 		    pack);
1750 
1751 	return err;
1752 }
1753 
1754 static const struct got_error *
1755 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1756     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1757     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1758 {
1759 	const struct got_error *err = NULL;
1760 
1761 	/* Validate decompression and obtain the decompressed size. */
1762 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1763 	    delta_data_offset, pack);
1764 	if (err)
1765 		return err;
1766 
1767 	/* Read delta base/result sizes from head of delta stream. */
1768 	err = got_delta_get_sizes(base_size, result_size,
1769 	    *delta_buf, *delta_len);
1770 	if (err)
1771 		goto done;
1772 
1773 	/* Discard decompressed delta and read it again in compressed form. */
1774 	free(*delta_buf);
1775 	*delta_buf = malloc(*delta_len_compressed);
1776 	if (*delta_buf == NULL) {
1777 		err = got_error_from_errno("malloc");
1778 		goto done;
1779 	}
1780 	if (pack->map) {
1781 		if (delta_data_offset >= pack->filesize) {
1782 			err = got_error(GOT_ERR_PACK_OFFSET);
1783 			goto done;
1784 		}
1785 		memcpy(*delta_buf, pack->map + delta_data_offset,
1786 		    *delta_len_compressed);
1787 	} else {
1788 		ssize_t n;
1789 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1790 			err = got_error_from_errno("lseek");
1791 			goto done;
1792 		}
1793 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1794 		if (n < 0) {
1795 			err = got_error_from_errno("read");
1796 			goto done;
1797 		} else if (n != *delta_len_compressed) {
1798 			err = got_error(GOT_ERR_IO);
1799 			goto done;
1800 		}
1801 	}
1802 done:
1803 	if (err) {
1804 		free(*delta_buf);
1805 		*delta_buf = NULL;
1806 		*delta_len = 0;
1807 		*delta_len_compressed = 0;
1808 		*base_size = 0;
1809 		*result_size = 0;
1810 	}
1811 	return err;
1812 }
1813 
1814 const struct got_error *
1815 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1816     size_t *delta_compressed_size, off_t *delta_offset, off_t *base_offset,
1817     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1818     struct got_pack *pack, struct got_packidx *packidx, int idx)
1819 {
1820 	const struct got_error *err = NULL;
1821 	off_t offset;
1822 	uint8_t type;
1823 	uint64_t size;
1824 	size_t tslen, delta_hdrlen;
1825 	off_t delta_data_offset;
1826 
1827 	*delta_buf = NULL;
1828 	*delta_size = 0;
1829 	*delta_compressed_size = 0;
1830 	*delta_offset = 0;
1831 	*base_offset = 0;
1832 	*base_size = 0;
1833 	*result_size = 0;
1834 
1835 	offset = got_packidx_get_object_offset(packidx, idx);
1836 	if (offset == -1)
1837 		return got_error(GOT_ERR_BAD_PACKIDX);
1838 
1839 	if (offset >= pack->filesize)
1840 		return got_error(GOT_ERR_PACK_OFFSET);
1841 
1842 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1843 	    pack, offset);
1844 	if (err)
1845 		return err;
1846 
1847 	if (tslen + size < tslen || offset + size < size ||
1848 	    tslen + offset < tslen)
1849 		return got_error(GOT_ERR_PACK_OFFSET);
1850 
1851 	switch (type) {
1852 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1853 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1854 		    pack, offset, tslen);
1855 		if (err)
1856 			return err;
1857 		break;
1858 	case GOT_OBJ_TYPE_REF_DELTA:
1859 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1860 		if (err)
1861 			return err;
1862 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1863 		break;
1864 	default:
1865 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1866 		    "non-delta object type %d found at offset %lld",
1867 		    type, (long long)offset);
1868 	}
1869 
1870 	if (tslen + delta_hdrlen < delta_hdrlen ||
1871 	    offset + delta_hdrlen < delta_hdrlen)
1872 		return got_error(GOT_ERR_BAD_DELTA);
1873 
1874 	delta_data_offset = offset + tslen + delta_hdrlen;
1875 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1876 	    base_size, result_size, delta_data_offset, pack, packidx);
1877 	if (err)
1878 		return err;
1879 
1880 	if (*delta_size != size) {
1881 		err = got_error(GOT_ERR_BAD_DELTA);
1882 		goto done;
1883 	}
1884 
1885 	*delta_offset = offset;
1886 done:
1887 	if (err) {
1888 		free(*delta_buf);
1889 		*delta_buf = NULL;
1890 		*delta_size = 0;
1891 		*delta_compressed_size = 0;
1892 		*delta_offset = 0;
1893 		*base_offset = 0;
1894 		*base_size = 0;
1895 		*result_size = 0;
1896 	}
1897 	return err;
1898 }