1 /*
2  * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org>
3  * Copyright (c) 2020, 2022 Stefan Sperling <stsp@openbsd.org>
4  *
5  * Permission to use, copy, modify, and distribute this software for any
6  * purpose with or without fee is hereby granted, provided that the above
7  * copyright notice and this permission notice appear in all copies.
8  *
9  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
10  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
11  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
12  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
13  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
14  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
15  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
16  */
17 
18 #include "got_compat.h"
19 
20 #include <sys/queue.h>
21 #include <sys/stat.h>
22 #include <sys/time.h>
23 #include <sys/types.h>
24 #include <sys/uio.h>
25 #include <sys/mman.h>
26 
27 #include <stdint.h>
28 #include <errno.h>
29 #include <imsg.h>
30 #include <limits.h>
31 #include <signal.h>
32 #include <stdio.h>
33 #include <stdlib.h>
34 #include <string.h>
35 #include <ctype.h>
36 #include <fcntl.h>
37 #include <unistd.h>
38 #include <zlib.h>
39 #include <err.h>
40 #include <assert.h>
41 #include <dirent.h>
42 
43 #include "got_error.h"
44 #include "got_object.h"
45 
46 #include "got_lib_hash.h"
47 #include "got_lib_delta.h"
48 #include "got_lib_inflate.h"
49 #include "got_lib_object.h"
50 #include "got_lib_object_parse.h"
51 #include "got_lib_object_idset.h"
52 #include "got_lib_privsep.h"
53 #include "got_lib_pack.h"
54 #include "got_lib_ratelimit.h"
55 #include "got_lib_pack_index.h"
56 #include "got_lib_delta_cache.h"
57 
58 struct got_indexed_object {
59 	struct got_object_id id;
60 
61 	/*
62 	 * Has this object been fully resolved?
63 	 * If so, we know its ID, otherwise we don't and 'id' is invalid.
64 	 */
65 	int valid;
66 
67 	/* Offset of type+size field for this object in pack file. */
68 	off_t off;
69 
70 	/* Type+size values parsed from pack file. */
71 	uint8_t type;
72 	uint64_t size;
73 
74 	/* Length of on-disk type+size data. */
75 	size_t tslen;
76 
77 	/* Length of object data following type+size. */
78 	size_t len;
79 
80 	uint32_t crc;
81 
82 	union {
83 		struct {
84 			/* For ref deltas. */
85 			struct got_object_id ref_id;
86 		} ref;
87 		struct {
88 			/* For offset deltas. */
89 			off_t base_offset;
90 			size_t base_offsetlen;
91 		} ofs;
92 	} delta;
93 };
94 
95 static void
96 putbe32(char *b, uint32_t n)
97 {
98 	b[0] = n >> 24;
99 	b[1] = n >> 16;
100 	b[2] = n >> 8;
101 	b[3] = n >> 0;
102 }
103 
104 static const struct got_error *
105 get_obj_type_label(const char **label, int obj_type)
106 {
107 	const struct got_error *err = NULL;
108 
109 	switch (obj_type) {
110 	case GOT_OBJ_TYPE_BLOB:
111 		*label = GOT_OBJ_LABEL_BLOB;
112 		break;
113 	case GOT_OBJ_TYPE_TREE:
114 		*label = GOT_OBJ_LABEL_TREE;
115 		break;
116 	case GOT_OBJ_TYPE_COMMIT:
117 		*label = GOT_OBJ_LABEL_COMMIT;
118 		break;
119 	case GOT_OBJ_TYPE_TAG:
120 		*label = GOT_OBJ_LABEL_TAG;
121 		break;
122 	default:
123 		*label = NULL;
124 		err = got_error(GOT_ERR_OBJ_TYPE);
125 		break;
126 	}
127 
128 	return err;
129 }
130 
131 static const struct got_error *
132 read_checksum(uint32_t *crc, struct got_hash *ctx, int fd, size_t len)
133 {
134 	uint8_t buf[8192];
135 	size_t n;
136 	ssize_t r;
137 
138 	for (n = len; n > 0; n -= r){
139 		r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
140 		if (r == -1)
141 			return got_error_from_errno("read");
142 		if (r == 0)
143 			break;
144 		if (crc)
145 			*crc = crc32(*crc, buf, r);
146 		if (ctx)
147 			got_hash_update(ctx, buf, r);
148 	}
149 
150 	return NULL;
151 }
152 
153 static const struct got_error *
154 read_file_digest(struct got_hash *ctx, FILE *f, size_t len)
155 {
156 	uint8_t buf[8192];
157 	size_t n, r;
158 
159 	for (n = len; n > 0; n -= r) {
160 		r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
161 		if (r == 0) {
162 			if (feof(f))
163 				return NULL;
164 			return got_ferror(f, GOT_ERR_IO);
165 		}
166 		got_hash_update(ctx, buf, r);
167 	}
168 
169 	return NULL;
170 }
171 
172 static const struct got_error *
173 read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
174     FILE *tmpfile, struct got_hash *pack_sha1_ctx)
175 {
176 	const struct got_error *err = NULL;
177 	struct got_hash ctx;
178 	uint8_t *data = NULL;
179 	size_t datalen = 0;
180 	ssize_t n;
181 	char *header;
182 	size_t headerlen;
183 	const char *obj_label;
184 	size_t mapoff = obj->off;
185 	struct got_inflate_checksum csum;
186 
187 	memset(&csum, 0, sizeof(csum));
188 	csum.input_ctx = pack_sha1_ctx;
189 	csum.input_crc = &obj->crc;
190 
191 	err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
192 	    &obj->tslen, pack, obj->off);
193 	if (err)
194 		return err;
195 
196 	if (pack->map) {
197 		obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
198 		got_hash_update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
199 		mapoff += obj->tslen;
200 	} else {
201 		/* XXX Seek back and get the CRC of on-disk type+size bytes. */
202 		if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
203 			return got_error_from_errno("lseek");
204 		err = read_checksum(&obj->crc, pack_sha1_ctx,
205 		    pack->fd, obj->tslen);
206 		if (err)
207 			return err;
208 	}
209 
210 	switch (obj->type) {
211 	case GOT_OBJ_TYPE_BLOB:
212 	case GOT_OBJ_TYPE_COMMIT:
213 	case GOT_OBJ_TYPE_TREE:
214 	case GOT_OBJ_TYPE_TAG:
215 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
216 			if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
217 				err = got_error_from_errno("fseek");
218 				break;
219 			}
220 			if (pack->map) {
221 				err = got_inflate_to_file_mmap(&datalen,
222 				    &obj->len, &csum, pack->map, mapoff,
223 				    pack->filesize - mapoff, tmpfile);
224 			} else {
225 				err = got_inflate_to_file_fd(&datalen,
226 				    &obj->len, &csum, pack->fd, tmpfile);
227 			}
228 		} else {
229 			if (pack->map) {
230 				err = got_inflate_to_mem_mmap(&data, &datalen,
231 				    &obj->len, &csum, pack->map, mapoff,
232 				    pack->filesize - mapoff);
233 			} else {
234 				err = got_inflate_to_mem_fd(&data, &datalen,
235 				    &obj->len, &csum, obj->size, pack->fd);
236 			}
237 		}
238 		if (err)
239 			break;
240 		got_hash_init(&ctx, GOT_HASH_SHA1);
241 		err = get_obj_type_label(&obj_label, obj->type);
242 		if (err) {
243 			free(data);
244 			break;
245 		}
246 		if (asprintf(&header, "%s %lld", obj_label,
247 		    (long long)obj->size) == -1) {
248 			err = got_error_from_errno("asprintf");
249 			free(data);
250 			break;
251 		}
252 		headerlen = strlen(header) + 1;
253 		got_hash_update(&ctx, header, headerlen);
254 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
255 			err = read_file_digest(&ctx, tmpfile, datalen);
256 			if (err) {
257 				free(header);
258 				free(data);
259 				break;
260 			}
261 		} else
262 			got_hash_update(&ctx, data, datalen);
263 		got_hash_final_object_id(&ctx, &obj->id);
264 		free(header);
265 		free(data);
266 		break;
267 	case GOT_OBJ_TYPE_REF_DELTA:
268 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
269 		if (pack->map) {
270 			if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
271 				err = got_error(GOT_ERR_BAD_PACKFILE);
272 				break;
273 			}
274 			if (mapoff + SHA1_DIGEST_LENGTH > SIZE_MAX) {
275 				err = got_error_fmt(GOT_ERR_RANGE,
276 				    "mapoff %lld would overflow size_t",
277 				    (long long)mapoff + SHA1_DIGEST_LENGTH);
278 				break;
279 			}
280 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
281 			    SHA1_DIGEST_LENGTH);
282 			obj->crc = crc32(obj->crc, pack->map + mapoff,
283 			    SHA1_DIGEST_LENGTH);
284 			got_hash_update(pack_sha1_ctx, pack->map + mapoff,
285 			    SHA1_DIGEST_LENGTH);
286 			mapoff += SHA1_DIGEST_LENGTH;
287 			err = got_inflate_to_mem_mmap(NULL, &datalen,
288 			    &obj->len, &csum, pack->map, mapoff,
289 			    pack->filesize - mapoff);
290 			if (err)
291 				break;
292 		} else {
293 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
294 			    SHA1_DIGEST_LENGTH);
295 			if (n == -1) {
296 				err = got_error_from_errno("read");
297 				break;
298 			}
299 			if (n < sizeof(obj->id)) {
300 				err = got_error(GOT_ERR_BAD_PACKFILE);
301 				break;
302 			}
303 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
304 			    SHA1_DIGEST_LENGTH);
305 			got_hash_update(pack_sha1_ctx,
306 			    obj->delta.ref.ref_id.sha1, SHA1_DIGEST_LENGTH);
307 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
308 			    &csum, obj->size, pack->fd);
309 			if (err)
310 				break;
311 		}
312 		obj->len += SHA1_DIGEST_LENGTH;
313 		break;
314 	case GOT_OBJ_TYPE_OFFSET_DELTA:
315 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
316 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
317 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
318 		    obj->tslen);
319 		if (err)
320 			break;
321 
322 		if (pack->map) {
323 			if (mapoff + obj->delta.ofs.base_offsetlen >=
324 			    pack->filesize) {
325 				err = got_error(GOT_ERR_BAD_PACKFILE);
326 				break;
327 			}
328 
329 			if (mapoff + obj->delta.ofs.base_offsetlen >
330 			    SIZE_MAX) {
331 				err = got_error_fmt(GOT_ERR_RANGE,
332 				    "mapoff %lld would overflow size_t",
333 				    (long long)mapoff
334 				    + obj->delta.ofs.base_offsetlen);
335 			}
336 
337 			obj->crc = crc32(obj->crc, pack->map + mapoff,
338 			    obj->delta.ofs.base_offsetlen);
339 			got_hash_update(pack_sha1_ctx, pack->map + mapoff,
340 			    obj->delta.ofs.base_offsetlen);
341 			mapoff += obj->delta.ofs.base_offsetlen;
342 			err = got_inflate_to_mem_mmap(NULL, &datalen,
343 			    &obj->len, &csum, pack->map, mapoff,
344 			    pack->filesize - mapoff);
345 			if (err)
346 				break;
347 		} else {
348 			/*
349 			 * XXX Seek back and get CRC and SHA1 of on-disk
350 			 * offset bytes.
351 			 */
352 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
353 			    == -1) {
354 				err = got_error_from_errno("lseek");
355 				break;
356 			}
357 			err = read_checksum(&obj->crc, pack_sha1_ctx,
358 			    pack->fd, obj->delta.ofs.base_offsetlen);
359 			if (err)
360 				break;
361 
362 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
363 			    &csum, obj->size, pack->fd);
364 			if (err)
365 				break;
366 		}
367 		obj->len += obj->delta.ofs.base_offsetlen;
368 		break;
369 	default:
370 		err = got_error(GOT_ERR_OBJ_TYPE);
371 		break;
372 	}
373 
374 	return err;
375 }
376 
377 const struct got_error *
378 got_pack_hwrite(int fd, void *buf, int len, struct got_hash *ctx)
379 {
380 	ssize_t w;
381 
382 	got_hash_update(ctx, buf, len);
383 
384 	w = write(fd, buf, len);
385 	if (w == -1)
386 		return got_error_from_errno("write");
387 	if (w != len)
388 		return got_error(GOT_ERR_IO);
389 
390 	return NULL;
391 }
392 
393 static const struct got_error *
394 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
395     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
396     FILE *delta_accum_file)
397 {
398 	const struct got_error *err = NULL;
399 	struct got_delta_chain deltas;
400 	struct got_delta *delta;
401 	uint8_t *buf = NULL;
402 	size_t len = 0;
403 	struct got_hash ctx;
404 	char *header = NULL;
405 	size_t headerlen;
406 	uint64_t max_size;
407 	int base_obj_type;
408 	const char *obj_label;
409 
410 	deltas.nentries = 0;
411 	STAILQ_INIT(&deltas.entries);
412 
413 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
414 	    obj->off, obj->tslen, obj->type, obj->size,
415 	    GOT_DELTA_CHAIN_RECURSION_MAX);
416 	if (err)
417 		goto done;
418 
419 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
420 	if (err)
421 		goto done;
422 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
423 		rewind(tmpfile);
424 		rewind(delta_base_file);
425 		rewind(delta_accum_file);
426 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
427 		    pack, tmpfile, delta_base_file, delta_accum_file);
428 		if (err)
429 			goto done;
430 	} else {
431 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
432 		    &deltas, pack);
433 	}
434 	if (err)
435 		goto done;
436 
437 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
438 	if (err)
439 		goto done;
440 	err = get_obj_type_label(&obj_label, base_obj_type);
441 	if (err)
442 		goto done;
443 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
444 		err = got_error_from_errno("asprintf");
445 		goto done;
446 	}
447 	headerlen = strlen(header) + 1;
448 	got_hash_init(&ctx, GOT_HASH_SHA1);
449 	got_hash_update(&ctx, header, headerlen);
450 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
451 		err = read_file_digest(&ctx, tmpfile, len);
452 		if (err)
453 			goto done;
454 	} else
455 		got_hash_update(&ctx, buf, len);
456 	got_hash_final_object_id(&ctx, &obj->id);
457 done:
458 	free(buf);
459 	free(header);
460 	while (!STAILQ_EMPTY(&deltas.entries)) {
461 		delta = STAILQ_FIRST(&deltas.entries);
462 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
463 		free(delta);
464 	}
465 	return err;
466 }
467 
468 /* Determine the slot in the pack index a given object ID should use. */
469 static int
470 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
471 {
472 	u_int8_t id0 = sha1[0];
473 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
474 	int left = 0, right = nindexed - 1;
475 	int cmp = 0, i = 0;
476 
477 	if (id0 > 0)
478 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
479 
480 	while (left <= right) {
481 		struct got_packidx_object_id *oid;
482 
483 		i = ((left + right) / 2);
484 		oid = &packidx->hdr.sorted_ids[i];
485 
486 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
487 		if (cmp == 0)
488 			return -1; /* object already indexed */
489 		else if (cmp > 0)
490 			left = i + 1;
491 		else if (cmp < 0)
492 			right = i - 1;
493 	}
494 
495 	return left;
496 }
497 
498 #if 0
499 static void
500 print_packidx(struct got_packidx *packidx)
501 {
502 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
503 	int i;
504 
505 	fprintf(stderr, "object IDs:\n");
506 	for (i = 0; i < nindexed; i++) {
507 		char hex[SHA1_DIGEST_STRING_LENGTH];
508 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
509 		    hex, sizeof(hex));
510 		fprintf(stderr, "%s\n", hex);
511 	}
512 	fprintf(stderr, "\n");
513 
514 	fprintf(stderr, "object offsets:\n");
515 	for (i = 0; i < nindexed; i++) {
516 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
517 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
518 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
519 			fprintf(stderr, "%u -> %llu\n", offset,
520 			    be64toh(packidx->hdr.large_offsets[j]));
521 		} else
522 			fprintf(stderr, "%u\n", offset);
523 	}
524 	fprintf(stderr, "\n");
525 
526 	fprintf(stderr, "fanout table:");
527 	for (i = 0; i <= 0xff; i++)
528 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
529 	fprintf(stderr, "\n");
530 }
531 #endif
532 
533 static void
534 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
535     struct got_indexed_object *obj)
536 {
537 	int i;
538 
539 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
540 	    SHA1_DIGEST_LENGTH);
541 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
542 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
543 		packidx->hdr.offsets[idx] = htobe32(obj->off);
544 	else {
545 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
546 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
547 		packidx->hdr.large_offsets[packidx->nlargeobj] =
548 		    htobe64(obj->off);
549 		packidx->nlargeobj++;
550 	}
551 
552 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
553 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
554 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
555 	}
556 }
557 
558 static int
559 indexed_obj_cmp(const void *pa, const void *pb)
560 {
561 	struct got_indexed_object *a, *b;
562 
563 	a = (struct got_indexed_object *)pa;
564 	b = (struct got_indexed_object *)pb;
565 	return got_object_id_cmp(&a->id, &b->id);
566 }
567 
568 static void
569 make_packidx(struct got_packidx *packidx, uint32_t nobj,
570     struct got_indexed_object *objects)
571 {
572 	struct got_indexed_object *obj;
573 	int i;
574 	uint32_t idx = 0;
575 
576 	qsort(objects, nobj, sizeof(struct got_indexed_object),
577 	    indexed_obj_cmp);
578 
579 	memset(packidx->hdr.fanout_table, 0,
580 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
581 	packidx->nlargeobj = 0;
582 
583 	for (i = 0; i < nobj; i++) {
584 		obj = &objects[i];
585 		if (obj->valid)
586 			add_indexed_object(packidx, idx++, obj);
587 	}
588 }
589 
590 static void
591 update_packidx(struct got_packidx *packidx, uint32_t nobj,
592     struct got_indexed_object *obj)
593 {
594 	int idx;
595 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
596 
597 	idx = find_object_idx(packidx, obj->id.sha1);
598 	if (idx == -1)
599 		return; /* object already indexed */
600 
601 	memmove(&packidx->hdr.sorted_ids[idx + 1],
602 	    &packidx->hdr.sorted_ids[idx],
603 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
604 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
605 	    sizeof(uint32_t) * (nindexed - idx));
606 
607 	add_indexed_object(packidx, idx, obj);
608 }
609 
610 static const struct got_error *
611 report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
612     uint32_t nobj_resolved, struct got_ratelimit *rl,
613     got_pack_index_progress_cb progress_cb, void *progress_arg)
614 {
615 	const struct got_error *err;
616 	int elapsed = 0;
617 
618 	if (rl) {
619 		err = got_ratelimit_check(&elapsed, rl);
620 		if (err || !elapsed)
621 			return err;
622 	}
623 
624 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
625 	    nobj_resolved);
626 }
627 
628 const struct got_error *
629 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
630     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
631     got_pack_index_progress_cb progress_cb, void *progress_arg,
632     struct got_ratelimit *rl)
633 {
634 	const struct got_error *err;
635 	struct got_packfile_hdr hdr;
636 	struct got_packidx packidx;
637 	char buf[8];
638 	char pack_sha1[SHA1_DIGEST_LENGTH];
639 	uint32_t nobj, nvalid, nloose, nresolved = 0, i;
640 	struct got_indexed_object *objects = NULL, *obj;
641 	struct got_hash ctx;
642 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
643 	ssize_t r, w;
644 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
645 	size_t mapoff = 0;
646 	int p_indexed = 0, last_p_indexed = -1;
647 	int p_resolved = 0, last_p_resolved = -1;
648 
649 	/* Require that pack file header and SHA1 trailer are present. */
650 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
651 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
652 		    "short pack file");
653 
654 	if (pack->map) {
655 		memcpy(&hdr, pack->map, sizeof(hdr));
656 		mapoff += sizeof(hdr);
657 	} else {
658 		r = read(pack->fd, &hdr, sizeof(hdr));
659 		if (r == -1)
660 			return got_error_from_errno("read");
661 		if (r < sizeof(hdr))
662 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
663 			    "short pack file");
664 	}
665 
666 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
667 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
668 		    "bad packfile signature");
669 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
670 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
671 		    "bad packfile version");
672 	nobj = be32toh(hdr.nobjects);
673 	if (nobj == 0)
674 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
675 		    "bad packfile with zero objects");
676 
677 	/* We compute the SHA1 of pack file contents and verify later on. */
678 	got_hash_init(&ctx, GOT_HASH_SHA1);
679 	got_hash_update(&ctx, &hdr, sizeof(hdr));
680 
681 	/*
682 	 * Create an in-memory pack index which will grow as objects
683 	 * IDs in the pack file are discovered. Only fields used to
684 	 * read deltified objects will be needed by the pack.c library
685 	 * code, so setting up just a pack index header is sufficient.
686 	 */
687 	memset(&packidx, 0, sizeof(packidx));
688 	packidx.hdr.magic = malloc(sizeof(uint32_t));
689 	if (packidx.hdr.magic == NULL)
690 		return got_error_from_errno("malloc");
691 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
692 	packidx.hdr.version = malloc(sizeof(uint32_t));
693 	if (packidx.hdr.version == NULL) {
694 		err = got_error_from_errno("malloc");
695 		goto done;
696 	}
697 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
698 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
699 	    sizeof(uint32_t));
700 	if (packidx.hdr.fanout_table == NULL) {
701 		err = got_error_from_errno("calloc");
702 		goto done;
703 	}
704 	packidx.hdr.sorted_ids = calloc(nobj,
705 	    sizeof(struct got_packidx_object_id));
706 	if (packidx.hdr.sorted_ids == NULL) {
707 		err = got_error_from_errno("calloc");
708 		goto done;
709 	}
710 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
711 	if (packidx.hdr.crc32 == NULL) {
712 		err = got_error_from_errno("calloc");
713 		goto done;
714 	}
715 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
716 	if (packidx.hdr.offsets == NULL) {
717 		err = got_error_from_errno("calloc");
718 		goto done;
719 	}
720 	/* Large offsets table is empty for pack files < 2 GB. */
721 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
722 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
723 		if (packidx.hdr.large_offsets == NULL) {
724 			err = got_error_from_errno("calloc");
725 			goto done;
726 		}
727 	}
728 
729 	nvalid = 0;
730 	nloose = 0;
731 	objects = calloc(nobj, sizeof(struct got_indexed_object));
732 	if (objects == NULL)
733 		return got_error_from_errno("calloc");
734 
735 	/*
736 	 * First pass: locate all objects and identify un-deltified objects.
737 	 *
738 	 * When this pass has completed we will know offset, type, size, and
739 	 * CRC information for all objects in this pack file. We won't know
740 	 * any of the actual object IDs of deltified objects yet since we
741 	 * will not yet attempt to combine deltas.
742 	 */
743 	pass = 1;
744 	for (i = 0; i < nobj; i++) {
745 		/* Don't send too many progress privsep messages. */
746 		p_indexed = ((i + 1) * 100) / nobj;
747 		if (p_indexed != last_p_indexed) {
748 			err = report_progress(nobj, i + 1, nloose, 0,
749 			    rl, progress_cb, progress_arg);
750 			if (err)
751 				goto done;
752 			last_p_indexed = p_indexed;
753 		}
754 
755 		obj = &objects[i];
756 		obj->crc = crc32(0L, NULL, 0);
757 
758 		/* Store offset to type+size information for this object. */
759 		if (pack->map) {
760 			obj->off = mapoff;
761 		} else {
762 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
763 			if (obj->off == -1) {
764 				err = got_error_from_errno("lseek");
765 				goto done;
766 			}
767 		}
768 
769 		err = read_packed_object(pack, obj, tmpfile, &ctx);
770 		if (err)
771 			goto done;
772 
773 		if (pack->map) {
774 			mapoff += obj->tslen + obj->len;
775 		} else {
776 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
777 			    SEEK_SET) == -1) {
778 				err = got_error_from_errno("lseek");
779 				goto done;
780 			}
781 		}
782 
783 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
784 		    obj->type == GOT_OBJ_TYPE_TREE ||
785 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
786 		    obj->type == GOT_OBJ_TYPE_TAG) {
787 			obj->valid = 1;
788 			nloose++;
789 		} else {
790 			if (first_delta_idx == -1)
791 				first_delta_idx = i;
792 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
793 				have_ref_deltas = 1;
794 		}
795 	}
796 	nvalid = nloose;
797 
798 	/*
799 	 * Having done a full pass over the pack file and can now
800 	 * verify its checksum.
801 	 */
802 	got_hash_final(&ctx, pack_sha1);
803 
804 	if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
805 		err = got_error(GOT_ERR_PACKFILE_CSUM);
806 		goto done;
807 	}
808 
809 	/* Verify the SHA1 checksum stored at the end of the pack file. */
810 	if (pack->map) {
811 		if (pack->filesize > SIZE_MAX) {
812 			err = got_error_fmt(GOT_ERR_RANGE,
813 			    "filesize %lld overflows size_t",
814 			    (long long)pack->filesize);
815 			goto done;
816 		}
817 
818 		memcpy(pack_sha1_expected, pack->map +
819 		    pack->filesize - SHA1_DIGEST_LENGTH,
820 		    SHA1_DIGEST_LENGTH);
821 	} else {
822 		ssize_t n;
823 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
824 			err = got_error_from_errno("lseek");
825 			goto done;
826 		}
827 		n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
828 		if (n == -1) {
829 			err = got_error_from_errno("read");
830 			goto done;
831 		}
832 		if (n != SHA1_DIGEST_LENGTH) {
833 			err = got_error(GOT_ERR_IO);
834 			goto done;
835 		}
836 	}
837 	if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
838 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
839 		    "bad checksum in pack file trailer");
840 		goto done;
841 	}
842 
843 	if (first_delta_idx == -1)
844 		first_delta_idx = 0;
845 
846 	/* In order to resolve ref deltas we need an in-progress pack index. */
847 	if (have_ref_deltas)
848 		make_packidx(&packidx, nobj, objects);
849 
850 	/*
851 	 * Second pass: We can now resolve deltas to compute the IDs of
852 	 * objects which appear in deltified form. Because deltas can be
853 	 * chained this pass may require a couple of iterations until all
854 	 * IDs of deltified objects have been discovered.
855 	 */
856 	pass++;
857 	while (nvalid != nobj) {
858 		int n = 0;
859 		/*
860 		 * This loop will only run once unless the pack file
861 		 * contains ref deltas which refer to objects located
862 		 * later in the pack file, which is unusual.
863 		 * Offset deltas can always be resolved in one pass
864 		 * unless the packfile is corrupt.
865 		 */
866 		for (i = first_delta_idx; i < nobj; i++) {
867 			obj = &objects[i];
868 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
869 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
870 				continue;
871 
872 			if (obj->valid)
873 				continue;
874 
875 			if (pack->map == NULL && lseek(pack->fd,
876 			    obj->off + obj->tslen, SEEK_SET) == -1) {
877 				err = got_error_from_errno("lseek");
878 				goto done;
879 			}
880 
881 			err = resolve_deltified_object(pack, &packidx, obj,
882 			    tmpfile, delta_base_file, delta_accum_file);
883 			if (err) {
884 				if (err->code != GOT_ERR_NO_OBJ)
885 					goto done;
886 				/*
887 				 * We cannot resolve this object yet because
888 				 * a delta base is unknown. Try again later.
889 				 */
890 				continue;
891 			}
892 
893 			obj->valid = 1;
894 			n++;
895 			if (have_ref_deltas)
896 				update_packidx(&packidx, nobj, obj);
897 			/* Don't send too many progress privsep messages. */
898 			p_resolved = ((nresolved + n) * 100) / nobj;
899 			if (p_resolved != last_p_resolved) {
900 				err = report_progress(nobj, nobj,
901 				    nloose, nresolved + n, rl,
902 				    progress_cb, progress_arg);
903 				if (err)
904 					goto done;
905 				last_p_resolved = p_resolved;
906 			}
907 
908 		}
909 		if (pass++ > 3 && n == 0) {
910 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
911 			    "could not resolve any of deltas; packfile could "
912 			    "be corrupt");
913 			goto done;
914 		}
915 		nresolved += n;
916 		nvalid += n;
917 	}
918 
919 	if (nloose + nresolved != nobj) {
920 		static char msg[64];
921 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
922 		    nloose + nresolved, nobj);
923 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
924 		goto done;
925 	}
926 
927 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
928 	    progress_cb, progress_arg);
929 	if (err)
930 		goto done;
931 
932 	make_packidx(&packidx, nobj, objects);
933 
934 	free(objects);
935 	objects = NULL;
936 
937 	got_hash_init(&ctx, GOT_HASH_SHA1);
938 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
939 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
940 	err = got_pack_hwrite(idxfd, buf, 8, &ctx);
941 	if (err)
942 		goto done;
943 	err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
944 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
945 	if (err)
946 		goto done;
947 	err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
948 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
949 	if (err)
950 		goto done;
951 	err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
952 	    nobj * sizeof(uint32_t), &ctx);
953 	if (err)
954 		goto done;
955 	err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
956 	    nobj * sizeof(uint32_t), &ctx);
957 	if (err)
958 		goto done;
959 	if (packidx.nlargeobj > 0) {
960 		err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
961 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
962 		if (err)
963 			goto done;
964 	}
965 	err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
966 	if (err)
967 		goto done;
968 
969 	got_hash_final(&ctx, packidx_hash);
970 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
971 	if (w == -1) {
972 		err = got_error_from_errno("write");
973 		goto done;
974 	}
975 	if (w != sizeof(packidx_hash)) {
976 		err = got_error(GOT_ERR_IO);
977 		goto done;
978 	}
979 done:
980 	free(objects);
981 	free(packidx.hdr.magic);
982 	free(packidx.hdr.version);
983 	free(packidx.hdr.fanout_table);
984 	free(packidx.hdr.sorted_ids);
985 	free(packidx.hdr.offsets);
986 	free(packidx.hdr.large_offsets);
987 	return err;
988 }