op public repos

Blob

Date:: Thu Oct 20 11:59:04 2022 UTC
Message:: let callers of got_pack_index() configure the rate limit of progress reporting
Actions:: History | Blame | Raw File
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 <sha1.h>
35 #include <endian.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_sha1.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, SHA1_CTX *sha1_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 (sha1_ctx)
147 			SHA1Update(sha1_ctx, buf, r);
148 	}
149 
150 	return NULL;
151 }
152 
153 static const struct got_error *
154 read_file_sha1(SHA1_CTX *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 		SHA1Update(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, SHA1_CTX *pack_sha1_ctx)
175 {
176 	const struct got_error *err = NULL;
177 	SHA1_CTX 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_sha1 = 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 		SHA1Update(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 		SHA1Init(&ctx);
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 		SHA1Update(&ctx, header, headerlen);
254 		if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
255 			err = read_file_sha1(&ctx, tmpfile, datalen);
256 			if (err) {
257 				free(header);
258 				free(data);
259 				break;
260 			}
261 		} else
262 			SHA1Update(&ctx, data, datalen);
263 		SHA1Final(obj->id.sha1, &ctx);
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 			memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff,
275 			    SHA1_DIGEST_LENGTH);
276 			obj->crc = crc32(obj->crc, pack->map + mapoff,
277 			    SHA1_DIGEST_LENGTH);
278 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
279 			    SHA1_DIGEST_LENGTH);
280 			mapoff += SHA1_DIGEST_LENGTH;
281 			err = got_inflate_to_mem_mmap(NULL, &datalen,
282 			    &obj->len, &csum, pack->map, mapoff,
283 			    pack->filesize - mapoff);
284 			if (err)
285 				break;
286 		} else {
287 			n = read(pack->fd, obj->delta.ref.ref_id.sha1,
288 			    SHA1_DIGEST_LENGTH);
289 			if (n == -1) {
290 				err = got_error_from_errno("read");
291 				break;
292 			}
293 			if (n < sizeof(obj->id)) {
294 				err = got_error(GOT_ERR_BAD_PACKFILE);
295 				break;
296 			}
297 			obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1,
298 			    SHA1_DIGEST_LENGTH);
299 			SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1,
300 			    SHA1_DIGEST_LENGTH);
301 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
302 			    &csum, obj->size, pack->fd);
303 			if (err)
304 				break;
305 		}
306 		obj->len += SHA1_DIGEST_LENGTH;
307 		break;
308 	case GOT_OBJ_TYPE_OFFSET_DELTA:
309 		memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH);
310 		err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset,
311 		    &obj->delta.ofs.base_offsetlen, pack, obj->off,
312 		    obj->tslen);
313 		if (err)
314 			break;
315 
316 		if (pack->map) {
317 			obj->crc = crc32(obj->crc, pack->map + mapoff,
318 			    obj->delta.ofs.base_offsetlen);
319 			SHA1Update(pack_sha1_ctx, pack->map + mapoff,
320 			    obj->delta.ofs.base_offsetlen);
321 			mapoff += obj->delta.ofs.base_offsetlen;
322 			err = got_inflate_to_mem_mmap(NULL, &datalen,
323 			    &obj->len, &csum, pack->map, mapoff,
324 			    pack->filesize - mapoff);
325 			if (err)
326 				break;
327 		} else {
328 			/*
329 			 * XXX Seek back and get CRC and SHA1 of on-disk
330 			 * offset bytes.
331 			 */
332 			if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET)
333 			    == -1) {
334 				err = got_error_from_errno("lseek");
335 				break;
336 			}
337 			err = read_checksum(&obj->crc, pack_sha1_ctx,
338 			    pack->fd, obj->delta.ofs.base_offsetlen);
339 			if (err)
340 				break;
341 
342 			err = got_inflate_to_mem_fd(NULL, &datalen, &obj->len,
343 			    &csum, obj->size, pack->fd);
344 			if (err)
345 				break;
346 		}
347 		obj->len += obj->delta.ofs.base_offsetlen;
348 		break;
349 	default:
350 		err = got_error(GOT_ERR_OBJ_TYPE);
351 		break;
352 	}
353 
354 	return err;
355 }
356 
357 static const struct got_error *
358 hwrite(int fd, void *buf, int len, SHA1_CTX *ctx)
359 {
360 	ssize_t w;
361 
362 	SHA1Update(ctx, buf, len);
363 
364 	w = write(fd, buf, len);
365 	if (w == -1)
366 		return got_error_from_errno("write");
367 	if (w != len)
368 		return got_error(GOT_ERR_IO);
369 
370 	return NULL;
371 }
372 
373 static const struct got_error *
374 resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx,
375     struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file,
376     FILE *delta_accum_file)
377 {
378 	const struct got_error *err = NULL;
379 	struct got_delta_chain deltas;
380 	struct got_delta *delta;
381 	uint8_t *buf = NULL;
382 	size_t len = 0;
383 	SHA1_CTX ctx;
384 	char *header = NULL;
385 	size_t headerlen;
386 	uint64_t max_size;
387 	int base_obj_type;
388 	const char *obj_label;
389 
390 	deltas.nentries = 0;
391 	STAILQ_INIT(&deltas.entries);
392 
393 	err = got_pack_resolve_delta_chain(&deltas, packidx, pack,
394 	    obj->off, obj->tslen, obj->type, obj->size,
395 	    GOT_DELTA_CHAIN_RECURSION_MAX);
396 	if (err)
397 		goto done;
398 
399 	err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack);
400 	if (err)
401 		goto done;
402 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
403 		rewind(tmpfile);
404 		rewind(delta_base_file);
405 		rewind(delta_accum_file);
406 		err = got_pack_dump_delta_chain_to_file(&len, &deltas,
407 		    pack, tmpfile, delta_base_file, delta_accum_file);
408 		if (err)
409 			goto done;
410 	} else {
411 		err = got_pack_dump_delta_chain_to_mem(&buf, &len,
412 		    &deltas, pack);
413 	}
414 	if (err)
415 		goto done;
416 
417 	err = got_delta_chain_get_base_type(&base_obj_type, &deltas);
418 	if (err)
419 		goto done;
420 	err = get_obj_type_label(&obj_label, base_obj_type);
421 	if (err)
422 		goto done;
423 	if (asprintf(&header, "%s %zd", obj_label, len) == -1) {
424 		err = got_error_from_errno("asprintf");
425 		goto done;
426 	}
427 	headerlen = strlen(header) + 1;
428 	SHA1Init(&ctx);
429 	SHA1Update(&ctx, header, headerlen);
430 	if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
431 		err = read_file_sha1(&ctx, tmpfile, len);
432 		if (err)
433 			goto done;
434 	} else
435 		SHA1Update(&ctx, buf, len);
436 	SHA1Final(obj->id.sha1, &ctx);
437 done:
438 	free(buf);
439 	free(header);
440 	while (!STAILQ_EMPTY(&deltas.entries)) {
441 		delta = STAILQ_FIRST(&deltas.entries);
442 		STAILQ_REMOVE_HEAD(&deltas.entries, entry);
443 		free(delta);
444 	}
445 	return err;
446 }
447 
448 /* Determine the slot in the pack index a given object ID should use. */
449 static int
450 find_object_idx(struct got_packidx *packidx, uint8_t *sha1)
451 {
452 	u_int8_t id0 = sha1[0];
453 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
454 	int left = 0, right = nindexed - 1;
455 	int cmp = 0, i = 0;
456 
457 	if (id0 > 0)
458 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
459 
460 	while (left <= right) {
461 		struct got_packidx_object_id *oid;
462 
463 		i = ((left + right) / 2);
464 		oid = &packidx->hdr.sorted_ids[i];
465 
466 		cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH);
467 		if (cmp == 0)
468 			return -1; /* object already indexed */
469 		else if (cmp > 0)
470 			left = i + 1;
471 		else if (cmp < 0)
472 			right = i - 1;
473 	}
474 
475 	return left;
476 }
477 
478 #if 0
479 static void
480 print_packidx(struct got_packidx *packidx)
481 {
482 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
483 	int i;
484 
485 	fprintf(stderr, "object IDs:\n");
486 	for (i = 0; i < nindexed; i++) {
487 		char hex[SHA1_DIGEST_STRING_LENGTH];
488 		got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1,
489 		    hex, sizeof(hex));
490 		fprintf(stderr, "%s\n", hex);
491 	}
492 	fprintf(stderr, "\n");
493 
494 	fprintf(stderr, "object offsets:\n");
495 	for (i = 0; i < nindexed; i++) {
496 		uint32_t offset = be32toh(packidx->hdr.offsets[i]);
497 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
498 			int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
499 			fprintf(stderr, "%u -> %llu\n", offset,
500 			    be64toh(packidx->hdr.large_offsets[j]));
501 		} else
502 			fprintf(stderr, "%u\n", offset);
503 	}
504 	fprintf(stderr, "\n");
505 
506 	fprintf(stderr, "fanout table:");
507 	for (i = 0; i <= 0xff; i++)
508 		fprintf(stderr, " %u", be32toh(packidx->hdr.fanout_table[i]));
509 	fprintf(stderr, "\n");
510 }
511 #endif
512 
513 static void
514 add_indexed_object(struct got_packidx *packidx, uint32_t idx,
515     struct got_indexed_object *obj)
516 {
517 	int i;
518 
519 	memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1,
520 	    SHA1_DIGEST_LENGTH);
521 	packidx->hdr.crc32[idx] = htobe32(obj->crc);
522 	if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)
523 		packidx->hdr.offsets[idx] = htobe32(obj->off);
524 	else {
525 		packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |
526 		    GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX);
527 		packidx->hdr.large_offsets[packidx->nlargeobj] =
528 		    htobe64(obj->off);
529 		packidx->nlargeobj++;
530 	}
531 
532 	for (i = obj->id.sha1[0]; i <= 0xff; i++) {
533 		uint32_t n = be32toh(packidx->hdr.fanout_table[i]);
534 		packidx->hdr.fanout_table[i] = htobe32(n + 1);
535 	}
536 }
537 
538 static int
539 indexed_obj_cmp(const void *pa, const void *pb)
540 {
541 	struct got_indexed_object *a, *b;
542 
543 	a = (struct got_indexed_object *)pa;
544 	b = (struct got_indexed_object *)pb;
545 	return got_object_id_cmp(&a->id, &b->id);
546 }
547 
548 static void
549 make_packidx(struct got_packidx *packidx, int nobj,
550     struct got_indexed_object *objects)
551 {
552 	struct got_indexed_object *obj;
553 	int i;
554 	uint32_t idx = 0;
555 
556 	qsort(objects, nobj, sizeof(struct got_indexed_object),
557 	    indexed_obj_cmp);
558 
559 	memset(packidx->hdr.fanout_table, 0,
560 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t));
561 	packidx->nlargeobj = 0;
562 
563 	for (i = 0; i < nobj; i++) {
564 		obj = &objects[i];
565 		if (obj->valid)
566 			add_indexed_object(packidx, idx++, obj);
567 	}
568 }
569 
570 static void
571 update_packidx(struct got_packidx *packidx, int nobj,
572     struct got_indexed_object *obj)
573 {
574 	int idx;
575 	uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff]);
576 
577 	idx = find_object_idx(packidx, obj->id.sha1);
578 	if (idx == -1) {
579 		char hex[SHA1_DIGEST_STRING_LENGTH];
580 		got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex));
581 		return; /* object already indexed */
582 	}
583 
584 	memmove(&packidx->hdr.sorted_ids[idx + 1],
585 	    &packidx->hdr.sorted_ids[idx],
586 	    sizeof(struct got_packidx_object_id) * (nindexed - idx));
587 	memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx],
588 	    sizeof(uint32_t) * (nindexed - idx));
589 
590 	add_indexed_object(packidx, idx, obj);
591 }
592 
593 static const struct got_error *
594 report_progress(int nobj_total, int nobj_indexed, int nobj_loose,
595     int nobj_resolved, struct got_ratelimit *rl,
596     got_pack_index_progress_cb progress_cb, void *progress_arg)
597 {
598 	const struct got_error *err;
599 	int elapsed = 0;
600 
601 	if (rl) {
602 		err = got_ratelimit_check(&elapsed, rl);
603 		if (err || !elapsed)
604 			return err;
605 	}
606 
607 	return progress_cb(progress_arg, nobj_total, nobj_indexed, nobj_loose,
608 	    nobj_resolved);
609 }
610 
611 const struct got_error *
612 got_pack_index(struct got_pack *pack, int idxfd, FILE *tmpfile,
613     FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected,
614     got_pack_index_progress_cb progress_cb, void *progress_arg,
615     struct got_ratelimit *rl)
616 {
617 	const struct got_error *err;
618 	struct got_packfile_hdr hdr;
619 	struct got_packidx packidx;
620 	char buf[8];
621 	char pack_sha1[SHA1_DIGEST_LENGTH];
622 	int nobj, nvalid, nloose, nresolved = 0, i;
623 	struct got_indexed_object *objects = NULL, *obj;
624 	SHA1_CTX ctx;
625 	uint8_t packidx_hash[SHA1_DIGEST_LENGTH];
626 	ssize_t r, w;
627 	int pass, have_ref_deltas = 0, first_delta_idx = -1;
628 	size_t mapoff = 0;
629 	int p_indexed = 0, last_p_indexed = -1;
630 	int p_resolved = 0, last_p_resolved = -1;
631 
632 	/* Require that pack file header and SHA1 trailer are present. */
633 	if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH)
634 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
635 		    "short pack file");
636 
637 	if (pack->map) {
638 		memcpy(&hdr, pack->map, sizeof(hdr));
639 		mapoff += sizeof(hdr);
640 	} else {
641 		r = read(pack->fd, &hdr, sizeof(hdr));
642 		if (r == -1)
643 			return got_error_from_errno("read");
644 		if (r < sizeof(hdr))
645 			return got_error_msg(GOT_ERR_BAD_PACKFILE,
646 			    "short pack file");
647 	}
648 
649 	if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE))
650 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
651 		    "bad packfile signature");
652 	if (hdr.version != htobe32(GOT_PACKFILE_VERSION))
653 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
654 		    "bad packfile version");
655 	nobj = be32toh(hdr.nobjects);
656 	if (nobj == 0)
657 		return got_error_msg(GOT_ERR_BAD_PACKFILE,
658 		    "bad packfile with zero objects");
659 
660 	/* We compute the SHA1 of pack file contents and verify later on. */
661 	SHA1Init(&ctx);
662 	SHA1Update(&ctx, (void *)&hdr, sizeof(hdr));
663 
664 	/*
665 	 * Create an in-memory pack index which will grow as objects
666 	 * IDs in the pack file are discovered. Only fields used to
667 	 * read deltified objects will be needed by the pack.c library
668 	 * code, so setting up just a pack index header is sufficient.
669 	 */
670 	memset(&packidx, 0, sizeof(packidx));
671 	packidx.hdr.magic = malloc(sizeof(uint32_t));
672 	if (packidx.hdr.magic == NULL)
673 		return got_error_from_errno("malloc");
674 	*packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC);
675 	packidx.hdr.version = malloc(sizeof(uint32_t));
676 	if (packidx.hdr.version == NULL) {
677 		err = got_error_from_errno("malloc");
678 		goto done;
679 	}
680 	*packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION);
681 	packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS,
682 	    sizeof(uint32_t));
683 	if (packidx.hdr.fanout_table == NULL) {
684 		err = got_error_from_errno("calloc");
685 		goto done;
686 	}
687 	packidx.hdr.sorted_ids = calloc(nobj,
688 	    sizeof(struct got_packidx_object_id));
689 	if (packidx.hdr.sorted_ids == NULL) {
690 		err = got_error_from_errno("calloc");
691 		goto done;
692 	}
693 	packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t));
694 	if (packidx.hdr.crc32 == NULL) {
695 		err = got_error_from_errno("calloc");
696 		goto done;
697 	}
698 	packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t));
699 	if (packidx.hdr.offsets == NULL) {
700 		err = got_error_from_errno("calloc");
701 		goto done;
702 	}
703 	/* Large offsets table is empty for pack files < 2 GB. */
704 	if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
705 		packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t));
706 		if (packidx.hdr.large_offsets == NULL) {
707 			err = got_error_from_errno("calloc");
708 			goto done;
709 		}
710 	}
711 
712 	nvalid = 0;
713 	nloose = 0;
714 	objects = calloc(nobj, sizeof(struct got_indexed_object));
715 	if (objects == NULL)
716 		return got_error_from_errno("calloc");
717 
718 	/*
719 	 * First pass: locate all objects and identify un-deltified objects.
720 	 *
721 	 * When this pass has completed we will know offset, type, size, and
722 	 * CRC information for all objects in this pack file. We won't know
723 	 * any of the actual object IDs of deltified objects yet since we
724 	 * will not yet attempt to combine deltas.
725 	 */
726 	pass = 1;
727 	for (i = 0; i < nobj; i++) {
728 		/* Don't send too many progress privsep messages. */
729 		p_indexed = ((i + 1) * 100) / nobj;
730 		if (p_indexed != last_p_indexed) {
731 			err = report_progress(nobj, i + 1, nloose, 0,
732 			    rl, progress_cb, progress_arg);
733 			if (err)
734 				goto done;
735 			last_p_indexed = p_indexed;
736 		}
737 
738 		obj = &objects[i];
739 		obj->crc = crc32(0L, NULL, 0);
740 
741 		/* Store offset to type+size information for this object. */
742 		if (pack->map) {
743 			obj->off = mapoff;
744 		} else {
745 			obj->off = lseek(pack->fd, 0, SEEK_CUR);
746 			if (obj->off == -1) {
747 				err = got_error_from_errno("lseek");
748 				goto done;
749 			}
750 		}
751 
752 		err = read_packed_object(pack, obj, tmpfile, &ctx);
753 		if (err)
754 			goto done;
755 
756 		if (pack->map) {
757 			mapoff += obj->tslen + obj->len;
758 		} else {
759 			if (lseek(pack->fd, obj->off + obj->tslen + obj->len,
760 			    SEEK_SET) == -1) {
761 				err = got_error_from_errno("lseek");
762 				goto done;
763 			}
764 		}
765 
766 		if (obj->type == GOT_OBJ_TYPE_BLOB ||
767 		    obj->type == GOT_OBJ_TYPE_TREE ||
768 		    obj->type == GOT_OBJ_TYPE_COMMIT ||
769 		    obj->type == GOT_OBJ_TYPE_TAG) {
770 			obj->valid = 1;
771 			nloose++;
772 		} else {
773 			if (first_delta_idx == -1)
774 				first_delta_idx = i;
775 			if (obj->type == GOT_OBJ_TYPE_REF_DELTA)
776 				have_ref_deltas = 1;
777 		}
778 	}
779 	nvalid = nloose;
780 
781 	/*
782 	 * Having done a full pass over the pack file and can now
783 	 * verify its checksum.
784 	 */
785 	SHA1Final(pack_sha1, &ctx);
786 	if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH) != 0) {
787 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
788 		    "pack file checksum mismatch");
789 		goto done;
790 	}
791 
792 	/* Verify the SHA1 checksum stored at the end of the pack file. */
793 	if (pack->map) {
794 		memcpy(pack_sha1_expected, pack->map +
795 		    pack->filesize - SHA1_DIGEST_LENGTH,
796 		    SHA1_DIGEST_LENGTH);
797 	} else {
798 		ssize_t n;
799 		if (lseek(pack->fd, -SHA1_DIGEST_LENGTH, SEEK_END) == -1) {
800 			err = got_error_from_errno("lseek");
801 			goto done;
802 		}
803 		n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH);
804 		if (n == -1) {
805 			err = got_error_from_errno("read");
806 			goto done;
807 		}
808 		if (n != SHA1_DIGEST_LENGTH) {
809 			err = got_error(GOT_ERR_IO);
810 			goto done;
811 		}
812 	}
813 	if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH) != 0) {
814 		err = got_error_msg(GOT_ERR_BAD_PACKFILE,
815 		    "bad checksum in pack file trailer");
816 		goto done;
817 	}
818 
819 	if (first_delta_idx == -1)
820 		first_delta_idx = 0;
821 
822 	/* In order to resolve ref deltas we need an in-progress pack index. */
823 	if (have_ref_deltas)
824 		make_packidx(&packidx, nobj, objects);
825 
826 	/*
827 	 * Second pass: We can now resolve deltas to compute the IDs of
828 	 * objects which appear in deltified form. Because deltas can be
829 	 * chained this pass may require a couple of iterations until all
830 	 * IDs of deltified objects have been discovered.
831 	 */
832 	pass++;
833 	while (nvalid != nobj) {
834 		int n = 0;
835 		/*
836 		 * This loop will only run once unless the pack file
837 		 * contains ref deltas which refer to objects located
838 		 * later in the pack file, which is unusual.
839 		 * Offset deltas can always be resolved in one pass
840 		 * unless the packfile is corrupt.
841 		 */
842 		for (i = first_delta_idx; i < nobj; i++) {
843 			obj = &objects[i];
844 			if (obj->type != GOT_OBJ_TYPE_REF_DELTA &&
845 			    obj->type != GOT_OBJ_TYPE_OFFSET_DELTA)
846 				continue;
847 
848 			if (obj->valid)
849 				continue;
850 
851 			if (pack->map == NULL && lseek(pack->fd,
852 			    obj->off + obj->tslen, SEEK_SET) == -1) {
853 				err = got_error_from_errno("lseek");
854 				goto done;
855 			}
856 
857 			err = resolve_deltified_object(pack, &packidx, obj,
858 			    tmpfile, delta_base_file, delta_accum_file);
859 			if (err) {
860 				if (err->code != GOT_ERR_NO_OBJ)
861 					goto done;
862 				/*
863 				 * We cannot resolve this object yet because
864 				 * a delta base is unknown. Try again later.
865 				 */
866 				continue;
867 			}
868 
869 			obj->valid = 1;
870 			n++;
871 			if (have_ref_deltas)
872 				update_packidx(&packidx, nobj, obj);
873 			/* Don't send too many progress privsep messages. */
874 			p_resolved = ((nresolved + n) * 100) / nobj;
875 			if (p_resolved != last_p_resolved) {
876 				err = report_progress(nobj, nobj,
877 				    nloose, nresolved + n, rl,
878 				    progress_cb, progress_arg);
879 				if (err)
880 					goto done;
881 				last_p_resolved = p_resolved;
882 			}
883 
884 		}
885 		if (pass++ > 3 && n == 0) {
886 			err = got_error_msg(GOT_ERR_BAD_PACKFILE,
887 			    "could not resolve any of deltas; packfile could "
888 			    "be corrupt");
889 			goto done;
890 		}
891 		nresolved += n;
892 		nvalid += nresolved;
893 	}
894 
895 	if (nloose + nresolved != nobj) {
896 		static char msg[64];
897 		snprintf(msg, sizeof(msg), "discovered only %d of %d objects",
898 		    nloose + nresolved, nobj);
899 		err = got_error_msg(GOT_ERR_BAD_PACKFILE, msg);
900 		goto done;
901 	}
902 
903 	err = report_progress(nobj, nobj, nloose, nresolved, NULL,
904 	    progress_cb, progress_arg);
905 	if (err)
906 		goto done;
907 
908 	make_packidx(&packidx, nobj, objects);
909 
910 	free(objects);
911 	objects = NULL;
912 
913 	SHA1Init(&ctx);
914 	putbe32(buf, GOT_PACKIDX_V2_MAGIC);
915 	putbe32(buf + 4, GOT_PACKIDX_VERSION);
916 	err = hwrite(idxfd, buf, 8, &ctx);
917 	if (err)
918 		goto done;
919 	err = hwrite(idxfd, packidx.hdr.fanout_table,
920 	    GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS * sizeof(uint32_t), &ctx);
921 	if (err)
922 		goto done;
923 	err = hwrite(idxfd, packidx.hdr.sorted_ids,
924 	    nobj * SHA1_DIGEST_LENGTH, &ctx);
925 	if (err)
926 		goto done;
927 	err = hwrite(idxfd, packidx.hdr.crc32, nobj * sizeof(uint32_t), &ctx);
928 	if (err)
929 		goto done;
930 	err = hwrite(idxfd, packidx.hdr.offsets, nobj * sizeof(uint32_t),
931 	    &ctx);
932 	if (err)
933 		goto done;
934 	if (packidx.nlargeobj > 0) {
935 		err = hwrite(idxfd, packidx.hdr.large_offsets,
936 		    packidx.nlargeobj * sizeof(uint64_t), &ctx);
937 		if (err)
938 			goto done;
939 	}
940 	err = hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH, &ctx);
941 	if (err)
942 		goto done;
943 
944 	SHA1Final(packidx_hash, &ctx);
945 	w = write(idxfd, packidx_hash, sizeof(packidx_hash));
946 	if (w == -1) {
947 		err = got_error_from_errno("write");
948 		goto done;
949 	}
950 	if (w != sizeof(packidx_hash)) {
951 		err = got_error(GOT_ERR_IO);
952 		goto done;
953 	}
954 done:
955 	free(objects);
956 	free(packidx.hdr.magic);
957 	free(packidx.hdr.version);
958 	free(packidx.hdr.fanout_table);
959 	free(packidx.hdr.sorted_ids);
960 	free(packidx.hdr.offsets);
961 	free(packidx.hdr.large_offsets);
962 	return err;
963 }