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;
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 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
800 	if (close(pack->privsep_child->imsg_fd) == -1 && err == NULL)
801 		err = got_error_from_errno("close");
802 	free(pack->privsep_child->ibuf);
803 	free(pack->privsep_child);
804 	pack->privsep_child = NULL;
805 	return err;
806 }
807 
808 const struct got_error *
809 got_pack_close(struct got_pack *pack)
810 {
811 	const struct got_error *err = NULL;
812 
813 	err = pack_stop_privsep_child(pack);
814 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
815 		err = got_error_from_errno("munmap");
816 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
817 		err = got_error_from_errno("close");
818 	pack->fd = -1;
819 	free(pack->path_packfile);
820 	pack->path_packfile = NULL;
821 	pack->filesize = 0;
822 	if (pack->delta_cache) {
823 		got_delta_cache_free(pack->delta_cache);
824 		pack->delta_cache = NULL;
825 	}
826 
827 	/*
828 	 * Leave accumfd and basefd alone. They are managed by the
829 	 * repository layer and can be reused.
830 	 */
831 
832 	return err;
833 }
834 
835 const struct got_error *
836 got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
837     struct got_pack *pack, off_t offset)
838 {
839 	uint8_t t = 0;
840 	uint64_t s = 0;
841 	uint8_t sizeN;
842 	size_t mapoff = 0;
843 	int i = 0;
844 
845 	*len = 0;
846 
847 	if (offset >= pack->filesize)
848 		return got_error(GOT_ERR_PACK_OFFSET);
849 
850 	if (pack->map) {
851 		mapoff = (size_t)offset;
852 	} else {
853 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
854 			return got_error_from_errno("lseek");
855 	}
856 
857 	do {
858 		/* We do not support size values which don't fit in 64 bit. */
859 		if (i > 9)
860 			return got_error(GOT_ERR_NO_SPACE);
861 
862 		if (pack->map) {
863 			if (mapoff + sizeof(sizeN) >= pack->filesize)
864 				return got_error(GOT_ERR_BAD_PACKFILE);
865 			sizeN = *(pack->map + mapoff);
866 			mapoff += sizeof(sizeN);
867 		} else {
868 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
869 			if (n < 0)
870 				return got_error_from_errno("read");
871 			if (n != sizeof(sizeN))
872 				return got_error(GOT_ERR_BAD_PACKFILE);
873 		}
874 		*len += sizeof(sizeN);
875 
876 		if (i == 0) {
877 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
878 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
879 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
880 		} else {
881 			size_t shift = 4 + 7 * (i - 1);
882 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
883 		}
884 		i++;
885 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
886 
887 	*type = t;
888 	*size = s;
889 	return NULL;
890 }
891 
892 static const struct got_error *
893 open_plain_object(struct got_object **obj, struct got_object_id *id,
894     uint8_t type, off_t offset, size_t size, int idx)
895 {
896 	*obj = calloc(1, sizeof(**obj));
897 	if (*obj == NULL)
898 		return got_error_from_errno("calloc");
899 
900 	(*obj)->type = type;
901 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
902 	(*obj)->pack_idx = idx;
903 	(*obj)->hdrlen = 0;
904 	(*obj)->size = size;
905 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
906 	(*obj)->pack_offset = offset;
907 
908 	return NULL;
909 }
910 
911 static const struct got_error *
912 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
913     off_t delta_offset)
914 {
915 	int64_t o = 0;
916 	uint8_t offN;
917 	int i = 0;
918 
919 	*offset = 0;
920 	*len = 0;
921 
922 	do {
923 		/* We do not support offset values which don't fit in 64 bit. */
924 		if (i > 8)
925 			return got_error(GOT_ERR_NO_SPACE);
926 
927 		if (pack->map) {
928 			size_t mapoff;
929 			mapoff = (size_t)delta_offset + *len;
930 			if (mapoff + sizeof(offN) >= pack->filesize)
931 				return got_error(GOT_ERR_PACK_OFFSET);
932 			offN = *(pack->map + mapoff);
933 		} else {
934 			ssize_t n;
935 			n = read(pack->fd, &offN, sizeof(offN));
936 			if (n < 0)
937 				return got_error_from_errno("read");
938 			if (n != sizeof(offN))
939 				return got_error(GOT_ERR_BAD_PACKFILE);
940 		}
941 		*len += sizeof(offN);
942 
943 		if (i == 0)
944 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
945 		else {
946 			o++;
947 			o <<= 7;
948 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
949 		}
950 		i++;
951 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
952 
953 	*offset = o;
954 	return NULL;
955 }
956 
957 const struct got_error *
958 got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
959     struct got_pack *pack, off_t offset, int tslen)
960 {
961 	const struct got_error *err;
962 	int64_t negoffset;
963 	size_t negofflen;
964 
965 	*len = 0;
966 
967 	err = parse_negative_offset(&negoffset, &negofflen, pack,
968 	    offset + tslen);
969 	if (err)
970 		return err;
971 
972 	/* Compute the base object's offset (must be in the same pack file). */
973 	*base_offset = (offset - negoffset);
974 	if (*base_offset <= 0)
975 		return got_error(GOT_ERR_BAD_PACKFILE);
976 
977 	*len = negofflen;
978 	return NULL;
979 }
980 
981 static const struct got_error *
982 read_delta_data(uint8_t **delta_buf, size_t *delta_len,
983     size_t *delta_compressed_len, size_t delta_data_offset,
984     struct got_pack *pack)
985 {
986 	const struct got_error *err = NULL;
987 	size_t consumed = 0;
988 
989 	if (pack->map) {
990 		if (delta_data_offset >= pack->filesize)
991 			return got_error(GOT_ERR_PACK_OFFSET);
992 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
993 		    &consumed, NULL, pack->map, delta_data_offset,
994 		    pack->filesize - delta_data_offset);
995 		if (err)
996 			return err;
997 	} else {
998 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
999 			return got_error_from_errno("lseek");
1000 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1001 		    &consumed, NULL, 0, pack->fd);
1002 		if (err)
1003 			return err;
1004 	}
1005 
1006 	if (delta_compressed_len)
1007 		*delta_compressed_len = consumed;
1008 
1009 	return NULL;
1010 }
1011 
1012 static const struct got_error *
1013 add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1014     int delta_type, size_t delta_size, size_t delta_data_offset)
1015 {
1016 	struct got_delta *delta;
1017 
1018 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1019 	    delta_data_offset);
1020 	if (delta == NULL)
1021 		return got_error_from_errno("got_delta_open");
1022 	/* delta is freed in got_object_close() */
1023 	deltas->nentries++;
1024 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1025 	return NULL;
1026 }
1027 
1028 static const struct got_error *
1029 resolve_offset_delta(struct got_delta_chain *deltas,
1030     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1031     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1032 
1033 {
1034 	const struct got_error *err;
1035 	off_t base_offset;
1036 	uint8_t base_type;
1037 	uint64_t base_size;
1038 	size_t base_tslen;
1039 	off_t delta_data_offset;
1040 	size_t consumed;
1041 
1042 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1043 	    delta_offset, tslen);
1044 	if (err)
1045 		return err;
1046 
1047 	delta_data_offset = delta_offset + tslen + consumed;
1048 	if (delta_data_offset >= pack->filesize)
1049 		return got_error(GOT_ERR_PACK_OFFSET);
1050 
1051 	if (pack->map == NULL) {
1052 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1053 		if (delta_data_offset == -1)
1054 			return got_error_from_errno("lseek");
1055 	}
1056 
1057 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1058 	    delta_data_offset);
1059 	if (err)
1060 		return err;
1061 
1062 	/* An offset delta must be in the same packfile. */
1063 	if (base_offset >= pack->filesize)
1064 		return got_error(GOT_ERR_PACK_OFFSET);
1065 
1066 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1067 	    &base_tslen, pack, base_offset);
1068 	if (err)
1069 		return err;
1070 
1071 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1072 	    base_tslen, base_type, base_size, recursion - 1);
1073 }
1074 
1075 const struct got_error *
1076 got_pack_parse_ref_delta(struct got_object_id *id,
1077     struct got_pack *pack, off_t delta_offset, int tslen)
1078 {
1079 	if (pack->map) {
1080 		size_t mapoff = delta_offset + tslen;
1081 		if (mapoff + sizeof(*id) >= pack->filesize)
1082 			return got_error(GOT_ERR_PACK_OFFSET);
1083 		memcpy(id, pack->map + mapoff, sizeof(*id));
1084 	} else {
1085 		ssize_t n;
1086 		n = read(pack->fd, id, sizeof(*id));
1087 		if (n < 0)
1088 			return got_error_from_errno("read");
1089 		if (n != sizeof(*id))
1090 			return got_error(GOT_ERR_BAD_PACKFILE);
1091 	}
1092 
1093 	return NULL;
1094 }
1095 
1096 static const struct got_error *
1097 resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1098     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1099     size_t delta_size, unsigned int recursion)
1100 {
1101 	const struct got_error *err;
1102 	struct got_object_id id;
1103 	int idx;
1104 	off_t base_offset;
1105 	uint8_t base_type;
1106 	uint64_t base_size;
1107 	size_t base_tslen;
1108 	off_t delta_data_offset;
1109 
1110 	if (delta_offset + tslen >= pack->filesize)
1111 		return got_error(GOT_ERR_PACK_OFFSET);
1112 
1113 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1114 	if (err)
1115 		return err;
1116 	if (pack->map) {
1117 		delta_data_offset = delta_offset + tslen + sizeof(id);
1118 	} else {
1119 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1120 		if (delta_data_offset == -1)
1121 			return got_error_from_errno("lseek");
1122 	}
1123 
1124 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1125 	    delta_data_offset);
1126 	if (err)
1127 		return err;
1128 
1129 	/* Delta base must be in the same pack file. */
1130 	idx = got_packidx_get_object_idx(packidx, &id);
1131 	if (idx == -1)
1132 		return got_error(GOT_ERR_NO_OBJ);
1133 
1134 	base_offset = got_packidx_get_object_offset(packidx, idx);
1135 	if (base_offset == (uint64_t)-1)
1136 		return got_error(GOT_ERR_BAD_PACKIDX);
1137 
1138 	if (base_offset >= pack->filesize)
1139 		return got_error(GOT_ERR_PACK_OFFSET);
1140 
1141 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1142 	    &base_tslen, pack, base_offset);
1143 	if (err)
1144 		return err;
1145 
1146 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1147 	    base_tslen, base_type, base_size, recursion - 1);
1148 }
1149 
1150 const struct got_error *
1151 got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1152     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1153     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1154 {
1155 	const struct got_error *err = NULL;
1156 
1157 	if (--recursion == 0)
1158 		return got_error(GOT_ERR_RECURSION);
1159 
1160 	switch (delta_type) {
1161 	case GOT_OBJ_TYPE_COMMIT:
1162 	case GOT_OBJ_TYPE_TREE:
1163 	case GOT_OBJ_TYPE_BLOB:
1164 	case GOT_OBJ_TYPE_TAG:
1165 		/* Plain types are the final delta base. Recursion ends. */
1166 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1167 		    delta_size, 0);
1168 		break;
1169 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1170 		err = resolve_offset_delta(deltas, packidx, pack,
1171 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1172 		break;
1173 	case GOT_OBJ_TYPE_REF_DELTA:
1174 		err = resolve_ref_delta(deltas, packidx, pack,
1175 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1176 		break;
1177 	default:
1178 		return got_error(GOT_ERR_OBJ_TYPE);
1179 	}
1180 
1181 	return err;
1182 }
1183 
1184 static const struct got_error *
1185 open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1186     struct got_pack *pack, struct got_object_id *id, off_t offset,
1187     size_t tslen, int delta_type, size_t delta_size, int idx)
1188 {
1189 	const struct got_error *err = NULL;
1190 	int resolved_type;
1191 
1192 	*obj = calloc(1, sizeof(**obj));
1193 	if (*obj == NULL)
1194 		return got_error_from_errno("calloc");
1195 
1196 	(*obj)->flags = 0;
1197 	(*obj)->hdrlen = 0;
1198 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1199 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1200 	(*obj)->pack_offset = offset + tslen;
1201 
1202 	STAILQ_INIT(&(*obj)->deltas.entries);
1203 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1204 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1205 	(*obj)->pack_idx = idx;
1206 
1207 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1208 	    offset, tslen, delta_type, delta_size,
1209 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1210 	if (err)
1211 		goto done;
1212 
1213 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1214 	if (err)
1215 		goto done;
1216 	(*obj)->type = resolved_type;
1217 done:
1218 	if (err) {
1219 		got_object_close(*obj);
1220 		*obj = NULL;
1221 	}
1222 	return err;
1223 }
1224 
1225 const struct got_error *
1226 got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1227     struct got_packidx *packidx, int idx, struct got_object_id *id)
1228 {
1229 	const struct got_error *err = NULL;
1230 	off_t offset;
1231 	uint8_t type;
1232 	uint64_t size;
1233 	size_t tslen;
1234 
1235 	*obj = NULL;
1236 
1237 	offset = got_packidx_get_object_offset(packidx, idx);
1238 	if (offset == (uint64_t)-1)
1239 		return got_error(GOT_ERR_BAD_PACKIDX);
1240 
1241 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1242 	    pack, offset);
1243 	if (err)
1244 		return err;
1245 
1246 	switch (type) {
1247 	case GOT_OBJ_TYPE_COMMIT:
1248 	case GOT_OBJ_TYPE_TREE:
1249 	case GOT_OBJ_TYPE_BLOB:
1250 	case GOT_OBJ_TYPE_TAG:
1251 		err = open_plain_object(obj, id, type, offset + tslen,
1252 		    size, idx);
1253 		break;
1254 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1255 	case GOT_OBJ_TYPE_REF_DELTA:
1256 		err = open_delta_object(obj, packidx, pack, id, offset,
1257 		    tslen, type, size, idx);
1258 		break;
1259 	default:
1260 		err = got_error(GOT_ERR_OBJ_TYPE);
1261 		break;
1262 	}
1263 
1264 	return err;
1265 }
1266 
1267 const struct got_error *
1268 got_pack_get_delta_chain_max_size(uint64_t *max_size,
1269     struct got_delta_chain *deltas, struct got_pack *pack)
1270 {
1271 	struct got_delta *delta;
1272 	uint64_t base_size = 0, result_size = 0;
1273 
1274 	*max_size = 0;
1275 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1276 		/* Plain object types are the delta base. */
1277 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1278 		    delta->type != GOT_OBJ_TYPE_TREE &&
1279 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1280 		    delta->type != GOT_OBJ_TYPE_TAG) {
1281 			const struct got_error *err;
1282 			uint8_t *delta_buf;
1283 			size_t delta_len;
1284 			int cached = 1;
1285 
1286 			got_delta_cache_get(&delta_buf, &delta_len,
1287 			    pack->delta_cache, delta->data_offset);
1288 			if (delta_buf == NULL) {
1289 				cached = 0;
1290 				err = read_delta_data(&delta_buf, &delta_len,
1291 				    NULL, delta->data_offset, pack);
1292 				if (err)
1293 					return err;
1294 				err = got_delta_cache_add(pack->delta_cache,
1295 				    delta->data_offset, delta_buf, delta_len);
1296 				if (err == NULL)
1297 					cached = 1;
1298 				else if (err->code != GOT_ERR_NO_SPACE) {
1299 					free(delta_buf);
1300 					return err;
1301 				}
1302 			}
1303 			err = got_delta_get_sizes(&base_size, &result_size,
1304 			    delta_buf, delta_len);
1305 			if (!cached)
1306 				free(delta_buf);
1307 			if (err)
1308 				return err;
1309 		} else
1310 			base_size = delta->size;
1311 		if (base_size > *max_size)
1312 			*max_size = base_size;
1313 		if (result_size > *max_size)
1314 			*max_size = result_size;
1315 	}
1316 
1317 	return NULL;
1318 }
1319 
1320 const struct got_error *
1321 got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1322     struct got_pack *pack)
1323 {
1324 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1325 		return got_error(GOT_ERR_OBJ_TYPE);
1326 
1327 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1328 }
1329 
1330 const struct got_error *
1331 got_pack_dump_delta_chain_to_file(size_t *result_size,
1332     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1333     FILE *base_file, FILE *accum_file)
1334 {
1335 	const struct got_error *err = NULL;
1336 	struct got_delta *delta;
1337 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1338 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1339 	/* We process small enough files entirely in memory for speed. */
1340 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1341 	uint64_t max_size = 0;
1342 	int n = 0;
1343 
1344 	*result_size = 0;
1345 
1346 	if (STAILQ_EMPTY(&deltas->entries))
1347 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1348 
1349 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1350 		return got_error_from_errno("fseeko");
1351 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1352 		return got_error_from_errno("fseeko");
1353 
1354 	/* Deltas are ordered in ascending order. */
1355 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1356 		uint64_t base_size, result_size = 0;
1357 		int cached = 1;
1358 		if (n == 0) {
1359 			size_t mapoff;
1360 			off_t delta_data_offset;
1361 
1362 			/* Plain object types are the delta base. */
1363 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1364 			    delta->type != GOT_OBJ_TYPE_TREE &&
1365 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1366 			    delta->type != GOT_OBJ_TYPE_TAG) {
1367 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1368 				goto done;
1369 			}
1370 
1371 			delta_data_offset = delta->offset + delta->tslen;
1372 			if (delta_data_offset >= pack->filesize) {
1373 				err = got_error(GOT_ERR_PACK_OFFSET);
1374 				goto done;
1375 			}
1376 			if (pack->map == NULL) {
1377 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1378 				    == -1) {
1379 					err = got_error_from_errno("lseek");
1380 					goto done;
1381 				}
1382 			}
1383 			if (delta->size > max_size)
1384 				max_size = delta->size;
1385 			if (max_size > max_bufsize) {
1386 				if (pack->map) {
1387 					mapoff = (size_t)delta_data_offset;
1388 					err = got_inflate_to_file_mmap(
1389 					    &base_bufsz, NULL, NULL, pack->map,
1390 					    mapoff, pack->filesize - mapoff,
1391 					    base_file);
1392 				} else
1393 					err = got_inflate_to_file_fd(
1394 					    &base_bufsz, NULL, NULL, pack->fd,
1395 					    base_file);
1396 			} else {
1397 				accum_buf = malloc(max_size);
1398 				if (accum_buf == NULL) {
1399 					err = got_error_from_errno("malloc");
1400 					goto done;
1401 				}
1402 				accum_bufsz = max_size;
1403 				if (pack->map) {
1404 					mapoff = (size_t)delta_data_offset;
1405 					err = got_inflate_to_mem_mmap(&base_buf,
1406 					    &base_bufsz, NULL, NULL,
1407 					    pack->map, mapoff,
1408 					    pack->filesize - mapoff);
1409 				} else
1410 					err = got_inflate_to_mem_fd(&base_buf,
1411 					    &base_bufsz, NULL, NULL, max_size,
1412 					    pack->fd);
1413 			}
1414 			if (err)
1415 				goto done;
1416 			n++;
1417 			if (base_buf == NULL)
1418 				rewind(base_file);
1419 			continue;
1420 		}
1421 
1422 		got_delta_cache_get(&delta_buf, &delta_len,
1423 		    pack->delta_cache, delta->data_offset);
1424 		if (delta_buf == NULL) {
1425 			cached = 0;
1426 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1427 			    delta->data_offset, pack);
1428 			if (err)
1429 				goto done;
1430 			err = got_delta_cache_add(pack->delta_cache,
1431 			    delta->data_offset, delta_buf, delta_len);
1432 			if (err == NULL)
1433 				cached = 1;
1434 			else if (err->code != GOT_ERR_NO_SPACE) {
1435 				free(delta_buf);
1436 				goto done;
1437 			}
1438 		}
1439 
1440 		err = got_delta_get_sizes(&base_size, &result_size,
1441 		    delta_buf, delta_len);
1442 		if (err)
1443 			goto done;
1444 		if (base_size > max_size)
1445 			max_size = base_size;
1446 		if (result_size > max_size)
1447 			max_size = result_size;
1448 
1449 		if (base_buf && max_size > max_bufsize) {
1450 			/* Switch from buffers to temporary files. */
1451 			size_t w = fwrite(base_buf, 1, base_bufsz,
1452 			    base_file);
1453 			if (w != base_bufsz) {
1454 				err = got_ferror(outfile, GOT_ERR_IO);
1455 				goto done;
1456 			}
1457 			free(base_buf);
1458 			base_buf = NULL;
1459 			free(accum_buf);
1460 			accum_buf = NULL;
1461 		}
1462 
1463 		if (base_buf && max_size > base_bufsz) {
1464 			uint8_t *p = realloc(base_buf, max_size);
1465 			if (p == NULL) {
1466 				err = got_error_from_errno("realloc");
1467 				goto done;
1468 			}
1469 			base_buf = p;
1470 			base_bufsz = max_size;
1471 		}
1472 
1473 		if (accum_buf && max_size > accum_bufsz) {
1474 			uint8_t *p = realloc(accum_buf, max_size);
1475 			if (p == NULL) {
1476 				err = got_error_from_errno("realloc");
1477 				goto done;
1478 			}
1479 			accum_buf = p;
1480 			accum_bufsz = max_size;
1481 		}
1482 
1483 		if (base_buf) {
1484 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1485 			    delta_buf, delta_len, accum_buf,
1486 			    &accum_size, max_size);
1487 			n++;
1488 		} else {
1489 			err = got_delta_apply(base_file, delta_buf,
1490 			    delta_len,
1491 			    /* Final delta application writes to output file. */
1492 			    ++n < deltas->nentries ? accum_file : outfile,
1493 			    &accum_size);
1494 		}
1495 		if (!cached)
1496 			free(delta_buf);
1497 		if (err)
1498 			goto done;
1499 
1500 		if (n < deltas->nentries) {
1501 			/* Accumulated delta becomes the new base. */
1502 			if (base_buf) {
1503 				uint8_t *tmp = accum_buf;
1504 				size_t tmp_size = accum_bufsz;
1505 				accum_buf = base_buf;
1506 				accum_bufsz = base_bufsz;
1507 				base_buf = tmp;
1508 				base_bufsz = tmp_size;
1509 			} else {
1510 				FILE *tmp = accum_file;
1511 				accum_file = base_file;
1512 				base_file = tmp;
1513 				rewind(base_file);
1514 				rewind(accum_file);
1515 			}
1516 		}
1517 	}
1518 
1519 done:
1520 	free(base_buf);
1521 	if (accum_buf) {
1522 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1523 		free(accum_buf);
1524 		if (len != accum_size)
1525 			err = got_ferror(outfile, GOT_ERR_IO);
1526 	}
1527 	rewind(outfile);
1528 	if (err == NULL)
1529 		*result_size = accum_size;
1530 	return err;
1531 }
1532 
1533 const struct got_error *
1534 got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1535     struct got_delta_chain *deltas, struct got_pack *pack)
1536 {
1537 	const struct got_error *err = NULL;
1538 	struct got_delta *delta;
1539 	uint8_t *base_buf = NULL, *accum_buf = NULL, *delta_buf;
1540 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1541 	uint64_t max_size = 0;
1542 	int n = 0;
1543 
1544 	*outbuf = NULL;
1545 	*outlen = 0;
1546 
1547 	if (STAILQ_EMPTY(&deltas->entries))
1548 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1549 
1550 	/* Deltas are ordered in ascending order. */
1551 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1552 		uint64_t base_size, result_size = 0;
1553 		int cached = 1;
1554 		if (n == 0) {
1555 			size_t delta_data_offset;
1556 
1557 			/* Plain object types are the delta base. */
1558 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1559 			    delta->type != GOT_OBJ_TYPE_TREE &&
1560 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1561 			    delta->type != GOT_OBJ_TYPE_TAG) {
1562 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1563 				goto done;
1564 			}
1565 
1566 			delta_data_offset = delta->offset + delta->tslen;
1567 			if (delta_data_offset >= pack->filesize) {
1568 				err = got_error(GOT_ERR_PACK_OFFSET);
1569 				goto done;
1570 			}
1571 
1572 			if (delta->size > max_size)
1573 				max_size = delta->size;
1574 
1575 			if (pack->map) {
1576 				size_t mapoff = (size_t)delta_data_offset;
1577 				err = got_inflate_to_mem_mmap(&base_buf,
1578 				    &base_bufsz, NULL, NULL, pack->map,
1579 				    mapoff, pack->filesize - mapoff);
1580 			} else {
1581 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1582 				    == -1) {
1583 					err = got_error_from_errno("lseek");
1584 					goto done;
1585 				}
1586 				err = got_inflate_to_mem_fd(&base_buf,
1587 				    &base_bufsz, NULL, NULL, max_size,
1588 				    pack->fd);
1589 			}
1590 			if (err)
1591 				goto done;
1592 			n++;
1593 			continue;
1594 		}
1595 
1596 		got_delta_cache_get(&delta_buf, &delta_len,
1597 		    pack->delta_cache, delta->data_offset);
1598 		if (delta_buf == NULL) {
1599 			cached = 0;
1600 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1601 			    delta->data_offset, pack);
1602 			if (err)
1603 				goto done;
1604 			err = got_delta_cache_add(pack->delta_cache,
1605 			    delta->data_offset, delta_buf, delta_len);
1606 			if (err == NULL)
1607 				cached = 1;
1608 			else if (err->code != GOT_ERR_NO_SPACE) {
1609 				free(delta_buf);
1610 				goto done;
1611 			}
1612 		}
1613 
1614 		err = got_delta_get_sizes(&base_size, &result_size,
1615 		    delta_buf, delta_len);
1616 		if (err)
1617 			goto done;
1618 		if (base_size > max_size)
1619 			max_size = base_size;
1620 		if (result_size > max_size)
1621 			max_size = result_size;
1622 
1623 		if (max_size > base_bufsz) {
1624 			uint8_t *p = realloc(base_buf, max_size);
1625 			if (p == NULL) {
1626 				err = got_error_from_errno("realloc");
1627 				goto done;
1628 			}
1629 			base_buf = p;
1630 			base_bufsz = max_size;
1631 		}
1632 
1633 		if (max_size > accum_bufsz) {
1634 			uint8_t *p = realloc(accum_buf, max_size);
1635 			if (p == NULL) {
1636 				err = got_error_from_errno("realloc");
1637 				goto done;
1638 			}
1639 			accum_buf = p;
1640 			accum_bufsz = max_size;
1641 		}
1642 
1643 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1644 		    delta_buf, delta_len, accum_buf,
1645 		    &accum_size, max_size);
1646 		if (!cached)
1647 			free(delta_buf);
1648 		n++;
1649 		if (err)
1650 			goto done;
1651 
1652 		if (n < deltas->nentries) {
1653 			/* Accumulated delta becomes the new base. */
1654 			uint8_t *tmp = accum_buf;
1655 			size_t tmp_size = accum_bufsz;
1656 			accum_buf = base_buf;
1657 			accum_bufsz = base_bufsz;
1658 			base_buf = tmp;
1659 			base_bufsz = tmp_size;
1660 		}
1661 	}
1662 
1663 done:
1664 	free(base_buf);
1665 	if (err) {
1666 		free(accum_buf);
1667 		*outbuf = NULL;
1668 		*outlen = 0;
1669 	} else {
1670 		*outbuf = accum_buf;
1671 		*outlen = accum_size;
1672 	}
1673 	return err;
1674 }
1675 
1676 const struct got_error *
1677 got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1678     FILE *outfile, FILE *base_file, FILE *accum_file)
1679 {
1680 	const struct got_error *err = NULL;
1681 
1682 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1683 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1684 
1685 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1686 		if (obj->pack_offset >= pack->filesize)
1687 			return got_error(GOT_ERR_PACK_OFFSET);
1688 
1689 		if (pack->map) {
1690 			size_t mapoff = (size_t)obj->pack_offset;
1691 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1692 			    pack->map, mapoff, pack->filesize - mapoff,
1693 			    outfile);
1694 		} else {
1695 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1696 				return got_error_from_errno("lseek");
1697 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1698 			    pack->fd, outfile);
1699 		}
1700 	} else
1701 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1702 		    &obj->deltas, pack, outfile, base_file, accum_file);
1703 
1704 	return err;
1705 }
1706 
1707 const struct got_error *
1708 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1709     struct got_object *obj, struct got_pack *pack)
1710 {
1711 	const struct got_error *err = NULL;
1712 
1713 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1714 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1715 
1716 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1717 		if (obj->pack_offset >= pack->filesize)
1718 			return got_error(GOT_ERR_PACK_OFFSET);
1719 		if (pack->map) {
1720 			size_t mapoff = (size_t)obj->pack_offset;
1721 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1722 			    pack->map, mapoff, pack->filesize - mapoff);
1723 		} else {
1724 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1725 				return got_error_from_errno("lseek");
1726 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1727 			    obj->size, pack->fd);
1728 		}
1729 	} else
1730 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1731 		    pack);
1732 
1733 	return err;
1734 }
1735 
1736 static const struct got_error *
1737 read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1738     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1739     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1740 {
1741 	const struct got_error *err = NULL;
1742 
1743 	/* Validate decompression and obtain the decompressed size. */
1744 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1745 	    delta_data_offset, pack);
1746 	if (err)
1747 		return err;
1748 
1749 	/* Read delta base/result sizes from head of delta stream. */
1750 	err = got_delta_get_sizes(base_size, result_size,
1751 	    *delta_buf, *delta_len);
1752 	if (err)
1753 		goto done;
1754 
1755 	/* Discard decompressed delta and read it again in compressed form. */
1756 	free(*delta_buf);
1757 	*delta_buf = malloc(*delta_len_compressed);
1758 	if (*delta_buf == NULL) {
1759 		err = got_error_from_errno("malloc");
1760 		goto done;
1761 	}
1762 	if (pack->map) {
1763 		if (delta_data_offset >= pack->filesize)
1764 			err = got_error(GOT_ERR_PACK_OFFSET);
1765 		memcpy(*delta_buf, pack->map + delta_data_offset,
1766 		    *delta_len_compressed);
1767 	} else {
1768 		ssize_t n;
1769 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1770 			err = got_error_from_errno("lseek");
1771 			goto done;
1772 		}
1773 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1774 		if (n < 0) {
1775 			err = got_error_from_errno("read");
1776 			goto done;
1777 		} else if (n != *delta_len_compressed) {
1778 			err = got_error(GOT_ERR_IO);
1779 			goto done;
1780 		}
1781 	}
1782 done:
1783 	if (err) {
1784 		free(*delta_buf);
1785 		*delta_buf = NULL;
1786 		*delta_len = 0;
1787 		*delta_len_compressed = 0;
1788 		*base_size = 0;
1789 		*result_size = 0;
1790 	}
1791 	return err;
1792 }
1793 
1794 const struct got_error *
1795 got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1796     size_t *delta_compressed_size, off_t *delta_offset, off_t *base_offset,
1797     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1798     struct got_pack *pack, struct got_packidx *packidx, int idx)
1799 {
1800 	const struct got_error *err = NULL;
1801 	off_t offset;
1802 	uint8_t type;
1803 	uint64_t size;
1804 	size_t tslen, delta_hdrlen;
1805 	off_t delta_data_offset;
1806 
1807 	*delta_buf = NULL;
1808 	*delta_size = 0;
1809 	*delta_compressed_size = 0;
1810 	*delta_offset = 0;
1811 	*base_offset = 0;
1812 	*base_size = 0;
1813 	*result_size = 0;
1814 
1815 	offset = got_packidx_get_object_offset(packidx, idx);
1816 	if (offset == (uint64_t)-1)
1817 		return got_error(GOT_ERR_BAD_PACKIDX);
1818 
1819 	if (offset >= pack->filesize)
1820 		return got_error(GOT_ERR_PACK_OFFSET);
1821 
1822 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1823 	    pack, offset);
1824 	if (err)
1825 		return err;
1826 
1827 	if (tslen + size < tslen || offset + size < size ||
1828 	    tslen + offset < tslen)
1829 		return got_error(GOT_ERR_PACK_OFFSET);
1830 
1831 	switch (type) {
1832 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1833 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1834 		    pack, offset, tslen);
1835 		if (err)
1836 			return err;
1837 		break;
1838 	case GOT_OBJ_TYPE_REF_DELTA:
1839 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1840 		if (err)
1841 			return err;
1842 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1843 		break;
1844 	default:
1845 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1846 		    "non-delta object type %d found at offset %llu",
1847 		    type, offset);
1848 	}
1849 
1850 	if (tslen + delta_hdrlen < delta_hdrlen ||
1851 	    offset + delta_hdrlen < delta_hdrlen)
1852 		return got_error(GOT_ERR_BAD_DELTA);
1853 
1854 	delta_data_offset = offset + tslen + delta_hdrlen;
1855 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1856 	    base_size, result_size, delta_data_offset, pack, packidx);
1857 	if (err)
1858 		return err;
1859 
1860 	if (*delta_size != size) {
1861 		err = got_error(GOT_ERR_BAD_DELTA);
1862 		goto done;
1863 	}
1864 
1865 	*delta_offset = offset;
1866 done:
1867 	if (err) {
1868 		free(*delta_buf);
1869 		*delta_buf = NULL;
1870 		*delta_size = 0;
1871 		*delta_compressed_size = 0;
1872 		*delta_offset = 0;
1873 		*base_offset = 0;
1874 		*base_size = 0;
1875 		*result_size = 0;
1876 	}
1877 	return err;
1878 }