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 <sys/queue.h>
19 #include <sys/stat.h>
20 #include <sys/time.h>
21 #include <sys/types.h>
22 #include <sys/uio.h>
23 #include <sys/mman.h>
24 
25 #include <stdint.h>
26 #include <errno.h>
27 #include <imsg.h>
28 #include <limits.h>
29 #include <signal.h>
30 #include <stdio.h>
31 #include <stdlib.h>
32 #include <string.h>
33 #include <ctype.h>
34 #include <fcntl.h>
35 #include <unistd.h>
36 #include <zlib.h>
37 #include <err.h>
38 #include <assert.h>
39 #include <dirent.h>
40 
41 #include "got_error.h"
42 #include "got_object.h"
43 
44 #include "got_lib_sha1.h"
45 #include "got_lib_delta.h"
46 #include "got_lib_inflate.h"
47 #include "got_lib_object.h"
48 #include "got_lib_object_parse.h"
49 #include "got_lib_object_idset.h"
50 #include "got_lib_privsep.h"
51 #include "got_lib_pack.h"
52 #include "got_lib_ratelimit.h"
53 #include "got_lib_pack_index.h"
54 #include "got_lib_delta_cache.h"
55 
56 struct got_indexed_object {
57 	struct got_object_id id;
58 
59 	/*
60 	 * Has this object been fully resolved?
61 	 * If so, we know its ID, otherwise we don't and 'id' is invalid.
62 	 */
63 	int valid;
64 
65 	/* Offset of type+size field for this object in pack file. */
66 	off_t off;
67 
68 	/* Type+size values parsed from pack file. */
69 	uint8_t type;
70 	uint64_t size;
71 
72 	/* Length of on-disk type+size data. */
73 	size_t tslen;
74 
75 	/* Length of object data following type+size. */
76 	size_t len;
77 
78 	uint32_t crc;
79 
80 	union {
81 		struct {
82 			/* For ref deltas. */
83 			struct got_object_id ref_id;
84 		} ref;
85 		struct {
86 			/* For offset deltas. */
87 			off_t base_offset;
88 			size_t base_offsetlen;
89 		} ofs;
90 	} delta;
91 };
92 
93 static void
94 putbe32(char *b, uint32_t n)
95 {
96 	b[0] = n >> 24;
97 	b[1] = n >> 16;
98 	b[2] = n >> 8;
99 	b[3] = n >> 0;
100 }
101 
102 static const struct got_error *
103 get_obj_type_label(const char **label, int obj_type)
104 {
105 	const struct got_error *err = NULL;
106 
107 	switch (obj_type) {
108 	case GOT_OBJ_TYPE_BLOB:
109 		*label = GOT_OBJ_LABEL_BLOB;
110 		break;
111 	case GOT_OBJ_TYPE_TREE:
112 		*label = GOT_OBJ_LABEL_TREE;
113 		break;
114 	case GOT_OBJ_TYPE_COMMIT:
115 		*label = GOT_OBJ_LABEL_COMMIT;
116 		break;
117 	case GOT_OBJ_TYPE_TAG:
118 		*label = GOT_OBJ_LABEL_TAG;
119 		break;
120 	default:
121 		*label = NULL;
122 		err = got_error(GOT_ERR_OBJ_TYPE);
123 		break;
124 	}
125 
126 	return err;
127 }
128 
129 static const struct got_error *
130 read_checksum(uint32_t *crc, SHA1_CTX *sha1_ctx, int fd, size_t len)
131 {
132 	uint8_t buf[8192];
133 	size_t n;
134 	ssize_t r;
135 
136 	for (n = len; n > 0; n -= r){
137 		r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n);
138 		if (r == -1)
139 			return got_error_from_errno("read");
140 		if (r == 0)
141 			break;
142 		if (crc)
143 			*crc = crc32(*crc, buf, r);
144 		if (sha1_ctx)
145 			SHA1Update(sha1_ctx, buf, r);
146 	}
147 
148 	return NULL;
149 }
150 
151 static const struct got_error *
152 read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len)
153 {
154 	uint8_t buf[8192];
155 	size_t n, r;
156 
157 	for (n = len; n > 0; n -= r) {
158 		r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f);
159 		if (r == 0) {
160 			if (feof(f))
161 				return NULL;
162 			return got_ferror(f, GOT_ERR_IO);
163 		}
164 		SHA1Update(ctx, buf, r);
165 	}
166 
167 	return NULL;
168 }
169 
170 static const struct got_error *
171 read_packed_object(struct got_pack *pack, struct got_indexed_object *obj,
172     FILE *tmpfile, SHA1_CTX *pack_sha1_ctx)
173 {
174 	const struct got_error *err = NULL;
175 	SHA1_CTX ctx;
176 	uint8_t *data = NULL;
177 	size_t datalen = 0;
178 	ssize_t n;
179 	char *header;
180 	size_t headerlen;
181 	const char *obj_label;
182 	size_t mapoff = obj->off;
183 	struct got_inflate_checksum csum;
184 
185 	memset(&csum, 0, sizeof(csum));
186 	csum.input_sha1 = pack_sha1_ctx;
187 	csum.input_crc = &obj->crc;
188 
189 	err = got_pack_parse_object_type_and_size(&obj->type, &obj->size,
190 	    &obj->tslen, pack, obj->off);
191 	if (err)
192 		return err;
193 
194 	if (pack->map) {
195 		obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen);
196 		SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen);
197 		mapoff += obj->tslen;
198 	} else {
199 		/* XXX Seek back and get the CRC of on-disk type+size bytes. */
200 		if (lseek(pack->fd, obj->off, SEEK_SET) == -1)
201 			return got_error_from_errno("lseek");
202 		err = read_checksum(&obj->crc, pack_sha1_ctx,
203 		    pack->fd, obj->tslen);
204 		if (err)
205 			return err;
206 	}
207 
208 	switch (obj->type) {
209 	case GOT_OBJ_TYPE_BLOB:
210 	case GOT_OBJ_TYPE_COMMIT:
211 	case GOT_OBJ_TYPE_TREE:
212 	case GOT_OBJ_TYPE_TAG:
213 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
214 			if (fseek(tmpfile, 0L, SEEK_SET) == -1) {
215 				err = got_error_from_errno("fseek");
216 				break;
217 			}
218 			if (pack->map) {
219 				err = got_inflate_to_file_mmap(&datalen,
220 				    &obj->len, &csum, pack->map, mapoff,
221 				    pack->filesize - mapoff, tmpfile);
222 			} else {
223 				err = got_inflate_to_file_fd(&datalen,
224 				    &obj->len, &csum, pack->fd, tmpfile);
225 			}
226 		} else {
227 			if (pack->map) {
228 				err = got_inflate_to_mem_mmap(&data, &datalen,
229 				    &obj->len, &csum, pack->map, mapoff,
230 				    pack->filesize - mapoff);
231 			} else {
232 				err = got_inflate_to_mem_fd(&data, &datalen,
233 				    &obj->len, &csum, obj->size, pack->fd);
234 			}
235 		}
236 		if (err)
237 			break;
238 		SHA1Init(&ctx);
239 		err = get_obj_type_label(&obj_label, obj->type);
240 		if (err) {
241 			free(data);
242 			break;
243 		}
244 		if (asprintf(&header, "%s %lld", obj_label,
245 		    (long long)obj->size) == -1) {
246 			err = got_error_from_errno("asprintf");
247 			free(data);
248 			break;
249 		}
250 		headerlen = strlen(header) + 1;
251 		SHA1Update(&ctx, header, headerlen);
252 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
253 			err = read_file_sha1(&ctx, tmpfile, datalen);
254 			if (err) {
255 				free(header);
256 				free(data);
257 				break;
258 			}
259 		} else
260 			SHA1Update(&ctx, data, datalen);
261 		SHA1Final(obj->id.sha1, &ctx);
262 		free(header);
263 		free(data);
264 		break;
265 	case GOT_OBJ_TYPE_REF_DELTA:
266 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
267 		if (pack->map) {
268 			if (mapoff + SHA1_DIGEST_LENGTH >= pack->filesize) {
269 				err = got_error(GOT_ERR_BAD_PACKFILE);
270 				break;
271 			}
272 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
273 			    SHA1_DIGEST_LENGTH);
274 			obj->crc = crc32(obj->crc, pack->map + mapoff,
275 			    SHA1_DIGEST_LENGTH);
276 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
277 			    SHA1_DIGEST_LENGTH);
278 			mapoff += SHA1_DIGEST_LENGTH;
279 			err = got_inflate_to_mem_mmap(NULL, &datalen,
280 			    &obj->len, &csum, pack->map, mapoff,
281 			    pack->filesize - mapoff);
282 			if (err)
283 				break;
284 		} else {
285 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
286 			    SHA1_DIGEST_LENGTH);
287 			if (n == -1) {
288 				err = got_error_from_errno("read");
289 				break;
290 			}
291 			if (n < sizeof(obj->id)) {
292 				err = got_error(GOT_ERR_BAD_PACKFILE);
293 				break;
294 			}
295 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
296 			    SHA1_DIGEST_LENGTH);
297 			SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
298 			    SHA1_DIGEST_LENGTH);
299 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
300 			    &csum, obj->size, pack->fd);
301 			if (err)
302 				break;
303 		}
304 		obj->len += SHA1_DIGEST_LENGTH;
305 		break;
306 	case GOT_OBJ_TYPE_OFFSET_DELTA:
307 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
308 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
309 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
310 		    obj->tslen);
311 		if (err)
312 			break;
313 
314 		if (pack->map) {
315 			obj->crc = crc32(obj->crc, pack->map + mapoff,
316 			    obj->delta.ofs.base_offsetlen);
317 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
318 			    obj->delta.ofs.base_offsetlen);
319 			mapoff += obj->delta.ofs.base_offsetlen;
320 			err = got_inflate_to_mem_mmap(NULL, &datalen,
321 			    &obj->len, &csum, pack->map, mapoff,
322 			    pack->filesize - mapoff);
323 			if (err)
324 				break;
325 		} else {
326 			/*
327 			 * XXX Seek back and get CRC and SHA1 of on-disk
328 			 * offset bytes.
329 			 */
330 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
331 			    == -1) {
332 				err = got_error_from_errno("lseek");
333 				break;
334 			}
335 			err = read_checksum(&obj->crc, pack_sha1_ctx,
336 			    pack->fd, obj->delta.ofs.base_offsetlen);
337 			if (err)
338 				break;
339 
340 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
341 			    &csum, obj->size, pack->fd);
342 			if (err)
343 				break;
344 		}
345 		obj->len += obj->delta.ofs.base_offsetlen;
346 		break;
347 	default:
348 		err = got_error(GOT_ERR_OBJ_TYPE);
349 		break;
350 	}
351 
352 	return err;
353 }
354 
355 const struct got_error *
356 got_pack_hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
357 {
358 	ssize_t w;
359 
360 	SHA1Update(ctx, buf, len);
361 
362 	w = write(fd, buf, len);
363 	if (w == -1)
364 		return got_error_from_errno("write");
365 	if (w != len)
366 		return got_error(GOT_ERR_IO);
367 
368 	return NULL;
369 }
370 
371 static const struct got_error *
372 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
373     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
374     FILE *delta_accum_file)
375 {
376 	const struct got_error *err = NULL;
377 	struct got_delta_chain deltas;
378 	struct got_delta *delta;
379 	uint8_t *buf = NULL;
380 	size_t len = 0;
381 	SHA1_CTX ctx;
382 	char *header = NULL;
383 	size_t headerlen;
384 	uint64_t max_size;
385 	int base_obj_type;
386 	const char *obj_label;
387 
388 	deltas.nentries = 0;
389 	STAILQ_INIT(&deltas.entries);
390 
391 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
392 	    obj->off, obj->tslen, obj->type, obj->size,
393 	    GOT_DELTA_CHAIN_RECURSION_MAX);
394 	if (err)
395 		goto done;
396 
397 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
398 	if (err)
399 		goto done;
400 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
401 		rewind(tmpfile);
402 		rewind(delta_base_file);
403 		rewind(delta_accum_file);
404 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
405 		    pack, tmpfile, delta_base_file, delta_accum_file);
406 		if (err)
407 			goto done;
408 	} else {
409 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
410 		    &deltas, pack);
411 	}
412 	if (err)
413 		goto done;
414 
415 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
416 	if (err)
417 		goto done;
418 	err = get_obj_type_label(&obj_label, base_obj_type);
419 	if (err)
420 		goto done;
421 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
422 		err = got_error_from_errno("asprintf");
423 		goto done;
424 	}
425 	headerlen = strlen(header) + 1;
426 	SHA1Init(&ctx);
427 	SHA1Update(&ctx, header, headerlen);
428 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
429 		err = read_file_sha1(&ctx, tmpfile, len);
430 		if (err)
431 			goto done;
432 	} else
433 		SHA1Update(&ctx, buf, len);
434 	SHA1Final(obj->id.sha1, &ctx);
435 done:
436 	free(buf);
437 	free(header);
438 	while (!STAILQ_EMPTY(&deltas.entries)) {
439 		delta = STAILQ_FIRST(&deltas.entries);
440 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
441 		free(delta);
442 	}
443 	return err;
444 }
445 
446 /* Determine the slot in the pack index a given object ID should use. */
447 static int
448 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
449 {
450 	u_int8_t id0 = sha1[0];
451 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
452 	int left = 0, right = nindexed - 1;
453 	int cmp = 0, i = 0;
454 
455 	if (id0 > 0)
456 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
457 
458 	while (left <= right) {
459 		struct got_packidx_object_id *oid;
460 
461 		i = ((left + right) / 2);
462 		oid = &packidx->hdr.sorted_ids[i];
463 
464 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
465 		if (cmp == 0)
466 			return -1; /* object already indexed */
467 		else if (cmp > 0)
468 			left = i + 1;
469 		else if (cmp < 0)
470 			right = i - 1;
471 	}
472 
473 	return left;
474 }
475 
476 #if 0
477 static void
478 print_packidx(struct got_packidx *packidx)
479 {
480 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
481 	int i;
482 
483 	fprintf(stderr, "object IDs:\n");
484 	for (i = 0; i < nindexed; i++) {
485 		char hex[SHA1_DIGEST_STRING_LENGTH];
486 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
487 		    hex, sizeof(hex));
488 		fprintf(stderr, "%s\n", hex);
489 	}
490 	fprintf(stderr, "\n");
491 
492 	fprintf(stderr, "object offsets:\n");
493 	for (i = 0; i < nindexed; i++) {
494 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
495 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
496 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
497 			fprintf(stderr, "%u -> %llu\n", offset,
498 			    be64toh(packidx->hdr.large_offsets[j]));
499 		} else
500 			fprintf(stderr, "%u\n", offset);
501 	}
502 	fprintf(stderr, "\n");
503 
504 	fprintf(stderr, "fanout table:");
505 	for (i = 0; i <= 0xff; i++)
506 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
507 	fprintf(stderr, "\n");
508 }
509 #endif
510 
511 static void
512 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
513     struct got_indexed_object *obj)
514 {
515 	int i;
516 
517 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
518 	    SHA1_DIGEST_LENGTH);
519 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
520 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
521 		packidx->hdr.offsets[idx] = htobe32(obj->off);
522 	else {
523 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
524 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
525 		packidx->hdr.large_offsets[packidx->nlargeobj] =
526 		    htobe64(obj->off);
527 		packidx->nlargeobj++;
528 	}
529 
530 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
531 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
532 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
533 	}
534 }
535 
536 static int
537 indexed_obj_cmp(const void *pa, const void *pb)
538 {
539 	struct got_indexed_object *a, *b;
540 
541 	a = (struct got_indexed_object *)pa;
542 	b = (struct got_indexed_object *)pb;
543 	return got_object_id_cmp(&a->id, &b->id);
544 }
545 
546 static void
547 make_packidx(struct got_packidx *packidx, uint32_t nobj,
548     struct got_indexed_object *objects)
549 {
550 	struct got_indexed_object *obj;
551 	int i;
552 	uint32_t idx = 0;
553 
554 	qsort(objects, nobj, sizeof(struct got_indexed_object),
555 	    indexed_obj_cmp);
556 
557 	memset(packidx->hdr.fanout_table, 0,
558 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
559 	packidx->nlargeobj = 0;
560 
561 	for (i = 0; i < nobj; i++) {
562 		obj = &objects[i];
563 		if (obj->valid)
564 			add_indexed_object(packidx, idx++, obj);
565 	}
566 }
567 
568 static void
569 update_packidx(struct got_packidx *packidx, uint32_t nobj,
570     struct got_indexed_object *obj)
571 {
572 	int idx;
573 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
574 
575 	idx = find_object_idx(packidx, obj->id.sha1);
576 	if (idx == -1) {
577 		char hex[SHA1_DIGEST_STRING_LENGTH];
578 		got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
579 		return; /* object already indexed */
580 	}
581 
582 	memmove(&packidx->hdr.sorted_ids[idx + 1],
583 	    &packidx->hdr.sorted_ids[idx],
584 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
585 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
586 	    sizeof(uint32_t) * (nindexed - idx));
587 
588 	add_indexed_object(packidx, idx, obj);
589 }
590 
591 static const struct got_error *
592 report_progress(uint32_t nobj_total, uint32_t nobj_indexed, uint32_t nobj_loose,
593     uint32_t nobj_resolved, struct got_ratelimit *rl,
594     got_pack_index_progress_cb progress_cb, void *progress_arg)
595 {
596 	const struct got_error *err;
597 	int elapsed = 0;
598 
599 	if (rl) {
600 		err = got_ratelimit_check(&elapsed, rl);
601 		if (err || !elapsed)
602 			return err;
603 	}
604 
605 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
606 	    nobj_resolved);
607 }
608 
609 const struct got_error *
610 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
611     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
612     got_pack_index_progress_cb progress_cb, void *progress_arg,
613     struct got_ratelimit *rl)
614 {
615 	const struct got_error *err;
616 	struct got_packfile_hdr hdr;
617 	struct got_packidx packidx;
618 	char buf[8];
619 	char pack_sha1[SHA1_DIGEST_LENGTH];
620 	uint32_t nobj, nvalid, nloose, nresolved = 0, i;
621 	struct got_indexed_object *objects = NULL, *obj;
622 	SHA1_CTX ctx;
623 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
624 	ssize_t r, w;
625 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
626 	size_t mapoff = 0;
627 	int p_indexed = 0, last_p_indexed = -1;
628 	int p_resolved = 0, last_p_resolved = -1;
629 
630 	/* Require that pack file header and SHA1 trailer are present. */
631 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
632 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
633 		    "short pack file");
634 
635 	if (pack->map) {
636 		memcpy(&hdr, pack->map, sizeof(hdr));
637 		mapoff += sizeof(hdr);
638 	} else {
639 		r = read(pack->fd, &hdr, sizeof(hdr));
640 		if (r == -1)
641 			return got_error_from_errno("read");
642 		if (r < sizeof(hdr))
643 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
644 			    "short pack file");
645 	}
646 
647 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
648 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
649 		    "bad packfile signature");
650 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
651 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
652 		    "bad packfile version");
653 	nobj = be32toh(hdr.nobjects);
654 	if (nobj == 0)
655 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
656 		    "bad packfile with zero objects");
657 
658 	/* We compute the SHA1 of pack file contents and verify later on. */
659 	SHA1Init(&ctx);
660 	SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
661 
662 	/*
663 	 * Create an in-memory pack index which will grow as objects
664 	 * IDs in the pack file are discovered. Only fields used to
665 	 * read deltified objects will be needed by the pack.c library
666 	 * code, so setting up just a pack index header is sufficient.
667 	 */
668 	memset(&packidx, 0, sizeof(packidx));
669 	packidx.hdr.magic = malloc(sizeof(uint32_t));
670 	if (packidx.hdr.magic == NULL)
671 		return got_error_from_errno("malloc");
672 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
673 	packidx.hdr.version = malloc(sizeof(uint32_t));
674 	if (packidx.hdr.version == NULL) {
675 		err = got_error_from_errno("malloc");
676 		goto done;
677 	}
678 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
679 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
680 	    sizeof(uint32_t));
681 	if (packidx.hdr.fanout_table == NULL) {
682 		err = got_error_from_errno("calloc");
683 		goto done;
684 	}
685 	packidx.hdr.sorted_ids = calloc(nobj,
686 	    sizeof(struct got_packidx_object_id));
687 	if (packidx.hdr.sorted_ids == NULL) {
688 		err = got_error_from_errno("calloc");
689 		goto done;
690 	}
691 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
692 	if (packidx.hdr.crc32 == NULL) {
693 		err = got_error_from_errno("calloc");
694 		goto done;
695 	}
696 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
697 	if (packidx.hdr.offsets == NULL) {
698 		err = got_error_from_errno("calloc");
699 		goto done;
700 	}
701 	/* Large offsets table is empty for pack files < 2 GB. */
702 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
703 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
704 		if (packidx.hdr.large_offsets == NULL) {
705 			err = got_error_from_errno("calloc");
706 			goto done;
707 		}
708 	}
709 
710 	nvalid = 0;
711 	nloose = 0;
712 	objects = calloc(nobj, sizeof(struct got_indexed_object));
713 	if (objects == NULL)
714 		return got_error_from_errno("calloc");
715 
716 	/*
717 	 * First pass: locate all objects and identify un-deltified objects.
718 	 *
719 	 * When this pass has completed we will know offset, type, size, and
720 	 * CRC information for all objects in this pack file. We won't know
721 	 * any of the actual object IDs of deltified objects yet since we
722 	 * will not yet attempt to combine deltas.
723 	 */
724 	pass = 1;
725 	for (i = 0; i < nobj; i++) {
726 		/* Don't send too many progress privsep messages. */
727 		p_indexed = ((i + 1) * 100) / nobj;
728 		if (p_indexed != last_p_indexed) {
729 			err = report_progress(nobj, i + 1, nloose, 0,
730 			    rl, progress_cb, progress_arg);
731 			if (err)
732 				goto done;
733 			last_p_indexed = p_indexed;
734 		}
735 
736 		obj = &objects[i];
737 		obj->crc = crc32(0L, NULL, 0);
738 
739 		/* Store offset to type+size information for this object. */
740 		if (pack->map) {
741 			obj->off = mapoff;
742 		} else {
743 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
744 			if (obj->off == -1) {
745 				err = got_error_from_errno("lseek");
746 				goto done;
747 			}
748 		}
749 
750 		err = read_packed_object(pack, obj, tmpfile, &ctx);
751 		if (err)
752 			goto done;
753 
754 		if (pack->map) {
755 			mapoff += obj->tslen + obj->len;
756 		} else {
757 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
758 			    SEEK_SET) == -1) {
759 				err = got_error_from_errno("lseek");
760 				goto done;
761 			}
762 		}
763 
764 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
765 		    obj->type == GOT_OBJ_TYPE_TREE ||
766 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
767 		    obj->type == GOT_OBJ_TYPE_TAG) {
768 			obj->valid = 1;
769 			nloose++;
770 		} else {
771 			if (first_delta_idx == -1)
772 				first_delta_idx = i;
773 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
774 				have_ref_deltas = 1;
775 		}
776 	}
777 	nvalid = nloose;
778 
779 	/*
780 	 * Having done a full pass over the pack file and can now
781 	 * verify its checksum.
782 	 */
783 	SHA1Final(pack_sha1, &ctx);
784 
785 	if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
786 		err = got_error(GOT_ERR_PACKFILE_CSUM);
787 		goto done;
788 	}
789 
790 	/* Verify the SHA1 checksum stored at the end of the pack file. */
791 	if (pack->map) {
792 		memcpy(pack_sha1_expected, pack->map +
793 		    pack->filesize - SHA1_DIGEST_LENGTH,
794 		    SHA1_DIGEST_LENGTH);
795 	} else {
796 		ssize_t n;
797 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
798 			err = got_error_from_errno("lseek");
799 			goto done;
800 		}
801 		n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
802 		if (n == -1) {
803 			err = got_error_from_errno("read");
804 			goto done;
805 		}
806 		if (n != SHA1_DIGEST_LENGTH) {
807 			err = got_error(GOT_ERR_IO);
808 			goto done;
809 		}
810 	}
811 	if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
812 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
813 		    "bad checksum in pack file trailer");
814 		goto done;
815 	}
816 
817 	if (first_delta_idx == -1)
818 		first_delta_idx = 0;
819 
820 	/* In order to resolve ref deltas we need an in-progress pack index. */
821 	if (have_ref_deltas)
822 		make_packidx(&packidx, nobj, objects);
823 
824 	/*
825 	 * Second pass: We can now resolve deltas to compute the IDs of
826 	 * objects which appear in deltified form. Because deltas can be
827 	 * chained this pass may require a couple of iterations until all
828 	 * IDs of deltified objects have been discovered.
829 	 */
830 	pass++;
831 	while (nvalid != nobj) {
832 		int n = 0;
833 		/*
834 		 * This loop will only run once unless the pack file
835 		 * contains ref deltas which refer to objects located
836 		 * later in the pack file, which is unusual.
837 		 * Offset deltas can always be resolved in one pass
838 		 * unless the packfile is corrupt.
839 		 */
840 		for (i = first_delta_idx; i < nobj; i++) {
841 			obj = &objects[i];
842 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
843 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
844 				continue;
845 
846 			if (obj->valid)
847 				continue;
848 
849 			if (pack->map == NULL && lseek(pack->fd,
850 			    obj->off + obj->tslen, SEEK_SET) == -1) {
851 				err = got_error_from_errno("lseek");
852 				goto done;
853 			}
854 
855 			err = resolve_deltified_object(pack, &packidx, obj,
856 			    tmpfile, delta_base_file, delta_accum_file);
857 			if (err) {
858 				if (err->code != GOT_ERR_NO_OBJ)
859 					goto done;
860 				/*
861 				 * We cannot resolve this object yet because
862 				 * a delta base is unknown. Try again later.
863 				 */
864 				continue;
865 			}
866 
867 			obj->valid = 1;
868 			n++;
869 			if (have_ref_deltas)
870 				update_packidx(&packidx, nobj, obj);
871 			/* Don't send too many progress privsep messages. */
872 			p_resolved = ((nresolved + n) * 100) / nobj;
873 			if (p_resolved != last_p_resolved) {
874 				err = report_progress(nobj, nobj,
875 				    nloose, nresolved + n, rl,
876 				    progress_cb, progress_arg);
877 				if (err)
878 					goto done;
879 				last_p_resolved = p_resolved;
880 			}
881 
882 		}
883 		if (pass++ > 3 && n == 0) {
884 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
885 			    "could not resolve any of deltas; packfile could "
886 			    "be corrupt");
887 			goto done;
888 		}
889 		nresolved += n;
890 		nvalid += nresolved;
891 	}
892 
893 	if (nloose + nresolved != nobj) {
894 		static char msg[64];
895 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
896 		    nloose + nresolved, nobj);
897 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
898 		goto done;
899 	}
900 
901 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
902 	    progress_cb, progress_arg);
903 	if (err)
904 		goto done;
905 
906 	make_packidx(&packidx, nobj, objects);
907 
908 	free(objects);
909 	objects = NULL;
910 
911 	SHA1Init(&ctx);
912 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
913 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
914 	err = got_pack_hwrite(idxfd, buf, 8, &ctx);
915 	if (err)
916 		goto done;
917 	err = got_pack_hwrite(idxfd, packidx.hdr.fanout_table,
918 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
919 	if (err)
920 		goto done;
921 	err = got_pack_hwrite(idxfd, packidx.hdr.sorted_ids,
922 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
923 	if (err)
924 		goto done;
925 	err = got_pack_hwrite(idxfd, packidx.hdr.crc32,
926 	    nobj * sizeof(uint32_t), &ctx);
927 	if (err)
928 		goto done;
929 	err = got_pack_hwrite(idxfd, packidx.hdr.offsets,
930 	    nobj * sizeof(uint32_t), &ctx);
931 	if (err)
932 		goto done;
933 	if (packidx.nlargeobj > 0) {
934 		err = got_pack_hwrite(idxfd, packidx.hdr.large_offsets,
935 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
936 		if (err)
937 			goto done;
938 	}
939 	err = got_pack_hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
940 	if (err)
941 		goto done;
942 
943 	SHA1Final(packidx_hash, &ctx);
944 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
945 	if (w == -1) {
946 		err = got_error_from_errno("write");
947 		goto done;
948 	}
949 	if (w != sizeof(packidx_hash)) {
950 		err = got_error(GOT_ERR_IO);
951 		goto done;
952 	}
953 done:
954 	free(objects);
955 	free(packidx.hdr.magic);
956 	free(packidx.hdr.version);
957 	free(packidx.hdr.fanout_table);
958 	free(packidx.hdr.sorted_ids);
959 	free(packidx.hdr.offsets);
960 	free(packidx.hdr.large_offsets);
961 	return err;
962 }