Blob


1 /*
2 * Copyright (c) 2018 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 */
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <sys/queue.h>
21 #include <dirent.h>
22 #include <fcntl.h>
23 #include <errno.h>
24 #include <stdio.h>
25 #include <stdint.h>
26 #include <stdlib.h>
27 #include <string.h>
28 #include <limits.h>
29 #include <sha1.h>
30 #include <endian.h>
31 #include <zlib.h>
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_repository.h"
37 #include "got_lib_sha1.h"
38 #include "got_lib_pack.h"
39 #include "got_lib_path.h"
40 #include "got_lib_delta.h"
41 #include "got_lib_zbuf.h"
42 #include "got_lib_object.h"
43 #include "got_lib_repository.h"
45 #ifndef nitems
46 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
47 #endif
49 #define GOT_PACK_PREFIX "pack-"
50 #define GOT_PACKFILE_SUFFIX ".pack"
51 #define GOT_PACKIDX_SUFFIX ".idx"
52 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
53 SHA1_DIGEST_STRING_LENGTH - 1 + \
54 strlen(GOT_PACKFILE_SUFFIX))
55 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
56 SHA1_DIGEST_STRING_LENGTH - 1 + \
57 strlen(GOT_PACKIDX_SUFFIX))
59 #ifndef MIN
60 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
61 #endif
63 static const struct got_error *
64 verify_fanout_table(uint32_t *fanout_table)
65 {
66 int i;
68 for (i = 0; i < 0xff - 1; i++) {
69 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
70 return got_error(GOT_ERR_BAD_PACKIDX);
71 }
73 return NULL;
74 }
76 static const struct got_error *
77 get_packfile_size(size_t *size, const char *path_idx)
78 {
79 struct stat sb;
80 char *path_pack;
81 char base_path[PATH_MAX];
82 char *dot;
84 if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
85 return got_error(GOT_ERR_NO_SPACE);
87 dot = strrchr(base_path, '.');
88 if (dot == NULL)
89 return got_error(GOT_ERR_BAD_PATH);
90 *dot = '\0';
91 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
92 return got_error_from_errno();
94 if (stat(path_pack, &sb) != 0) {
95 free(path_pack);
96 return got_error_from_errno();
97 }
99 free(path_pack);
100 *size = sb.st_size;
101 return 0;
104 const struct got_error *
105 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
107 struct got_packidx_v2_hdr *p;
108 FILE *f;
109 const struct got_error *err = NULL;
110 size_t n, nobj, packfile_size;
111 SHA1_CTX ctx;
112 uint8_t sha1[SHA1_DIGEST_LENGTH];
114 SHA1Init(&ctx);
116 f = fopen(path, "rb");
117 if (f == NULL)
118 return got_error_from_errno();
120 err = get_packfile_size(&packfile_size, path);
121 if (err)
122 return err;
124 p = calloc(1, sizeof(*p));
125 if (p == NULL) {
126 err = got_error_from_errno();
127 goto done;
130 n = fread(&p->magic, sizeof(p->magic), 1, f);
131 if (n != 1) {
132 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
133 goto done;
136 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
137 err = got_error(GOT_ERR_BAD_PACKIDX);
138 goto done;
141 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
143 n = fread(&p->version, sizeof(p->version), 1, f);
144 if (n != 1) {
145 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
146 goto done;
149 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
150 err = got_error(GOT_ERR_BAD_PACKIDX);
151 goto done;
154 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
156 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
157 if (n != 1) {
158 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
159 goto done;
162 err = verify_fanout_table(p->fanout_table);
163 if (err)
164 goto done;
166 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
168 nobj = betoh32(p->fanout_table[0xff]);
170 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
171 if (p->sorted_ids == NULL) {
172 err = got_error_from_errno();
173 goto done;
176 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
177 if (n != nobj) {
178 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
179 goto done;
182 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
183 nobj * sizeof(*p->sorted_ids));
185 p->crc32 = calloc(nobj, sizeof(*p->crc32));
186 if (p->crc32 == NULL) {
187 err = got_error_from_errno();
188 goto done;
191 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
192 if (n != nobj) {
193 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
194 goto done;
197 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
199 p->offsets = calloc(nobj, sizeof(*p->offsets));
200 if (p->offsets == NULL) {
201 err = got_error_from_errno();
202 goto done;
205 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
206 if (n != nobj) {
207 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
208 goto done;
211 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
213 /* Large file offsets are contained only in files > 2GB. */
214 if (packfile_size <= 0x80000000)
215 goto checksum;
217 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
218 if (p->large_offsets == NULL) {
219 err = got_error_from_errno();
220 goto done;
223 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
224 if (n != nobj) {
225 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
226 goto done;
229 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
230 nobj * sizeof(*p->large_offsets));
232 checksum:
233 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
234 if (n != 1) {
235 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
236 goto done;
239 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
240 SHA1Final(sha1, &ctx);
241 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
242 err = got_error(GOT_ERR_PACKIDX_CSUM);
243 done:
244 fclose(f);
245 if (err)
246 got_packidx_close(p);
247 else
248 *packidx = p;
249 return err;
252 void
253 got_packidx_close(struct got_packidx_v2_hdr *packidx)
255 free(packidx->sorted_ids);
256 free(packidx->offsets);
257 free(packidx->crc32);
258 free(packidx->large_offsets);
259 free(packidx);
262 static int
263 is_packidx_filename(const char *name, size_t len)
265 if (len != GOT_PACKIDX_NAMELEN)
266 return 0;
268 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
269 return 0;
271 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
272 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
273 return 0;
275 return 1;
278 static off_t
279 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
281 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
282 uint32_t offset = betoh32(packidx->offsets[idx]);
283 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
284 uint64_t loffset;
285 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
286 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
287 return -1;
288 loffset = betoh64(packidx->large_offsets[idx]);
289 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
291 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
294 static const struct got_error *
295 get_packfile_path(char **path_packfile, struct got_repository *repo,
296 struct got_packidx_v2_hdr *packidx);
298 static int
299 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id, struct got_repository *repo)
301 u_int8_t id0 = id->sha1[0];
302 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
303 int i = 0;
305 if (id0 > 0)
306 i = betoh32(packidx->fanout_table[id0 - 1]);
308 while (i < totobj) {
309 struct got_object_id *oid = &packidx->sorted_ids[i];
310 int cmp = got_object_id_cmp(id, oid);
312 if (cmp == 0)
313 return i;
314 i++;
317 return -1;
320 static struct got_packidx_v2_hdr *
321 dup_packidx(struct got_packidx_v2_hdr *packidx)
323 struct got_packidx_v2_hdr *p;
324 size_t nobj;
326 p = calloc(1, sizeof(*p));
327 if (p == NULL)
328 return NULL;
330 memcpy(p, packidx, sizeof(*p));
331 p->sorted_ids = NULL;
332 p->crc32 = NULL;
333 p->offsets = NULL;
334 p->large_offsets = NULL;
336 nobj = betoh32(p->fanout_table[0xff]);
338 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
339 if (p->sorted_ids == NULL)
340 goto err;
341 memcpy(p->sorted_ids, packidx->sorted_ids, nobj * sizeof(*p->sorted_ids));
343 p->crc32 = calloc(nobj, sizeof(*p->crc32));
344 if (p->crc32 == NULL)
345 goto err;
346 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
348 p->offsets = calloc(nobj, sizeof(*p->offsets));
349 if (p->offsets == NULL)
350 goto err;
351 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
353 if (p->large_offsets) {
354 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
355 if (p->large_offsets == NULL)
356 goto err;
357 memcpy(p->large_offsets, packidx->large_offsets,
358 nobj * sizeof(*p->large_offsets));
361 return p;
363 err:
364 free(p->large_offsets);
365 free(p->offsets);
366 free(p->crc32);
367 free(p->sorted_ids);
368 free(p);
369 return NULL;
372 static void
373 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
375 int i;
377 for (i = 0; i < nitems(repo->packidx_cache); i++) {
378 if (repo->packidx_cache[i] == NULL)
379 break;
382 if (i == nitems(repo->packidx_cache)) {
383 got_packidx_close(repo->packidx_cache[i - 1]);
384 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
385 sizeof(repo->packidx_cache) -
386 sizeof(repo->packidx_cache[0]));
387 i = 0;
390 repo->packidx_cache[i] = dup_packidx(packidx);
393 static const struct got_error *
394 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
395 struct got_repository *repo, struct got_object_id *id)
397 const struct got_error *err;
398 char *path_packdir;
399 DIR *packdir;
400 struct dirent *dent;
401 char *path_packidx;
402 int i;
404 /* Search pack index cache. */
405 for (i = 0; i < nitems(repo->packidx_cache); i++) {
406 if (repo->packidx_cache[i] == NULL)
407 break;
408 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
409 if (*idx != -1) {
410 *packidx = repo->packidx_cache[i];
411 return NULL;
414 /* No luck. Search the filesystem. */
416 path_packdir = got_repo_get_path_objects_pack(repo);
417 if (path_packdir == NULL)
418 return got_error_from_errno();
420 packdir = opendir(path_packdir);
421 if (packdir == NULL) {
422 err = got_error_from_errno();
423 goto done;
426 while ((dent = readdir(packdir)) != NULL) {
427 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
428 continue;
430 if (asprintf(&path_packidx, "%s/%s", path_packdir,
431 dent->d_name) == -1) {
432 err = got_error_from_errno();
433 goto done;
436 err = got_packidx_open(packidx, path_packidx);
437 free(path_packidx);
438 if (err)
439 goto done;
441 *idx = get_object_idx(*packidx, id, repo);
442 if (*idx != -1) {
443 err = NULL; /* found the object */
444 cache_packidx(*packidx, repo);
445 goto done;
448 got_packidx_close(*packidx);
449 *packidx = NULL;
452 err = got_error(GOT_ERR_NO_OBJ);
453 done:
454 free(path_packdir);
455 if (packdir && closedir(packdir) != 0 && err == 0)
456 err = got_error_from_errno();
457 return err;
460 static const struct got_error *
461 get_packfile_path(char **path_packfile, struct got_repository *repo,
462 struct got_packidx_v2_hdr *packidx)
464 char *path_packdir;
465 char hex[SHA1_DIGEST_STRING_LENGTH];
466 char *sha1str;
468 path_packdir = got_repo_get_path_objects_pack(repo);
469 if (path_packdir == NULL)
470 return got_error_from_errno();
472 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
473 hex, sizeof(hex));
474 if (sha1str == NULL)
475 return got_error(GOT_ERR_PACKIDX_CSUM);
477 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
478 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
479 *path_packfile = NULL;
480 return got_error_from_errno();
483 return NULL;
486 const struct got_error *
487 read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
489 const struct got_error *err = NULL;
490 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
491 struct got_packfile_hdr hdr;
492 size_t n;
494 n = fread(&hdr, sizeof(hdr), 1, f);
495 if (n != 1)
496 return got_ferror(f, GOT_ERR_BAD_PACKIDX);
498 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
499 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
500 betoh32(hdr.nobjects) != totobj)
501 err = got_error(GOT_ERR_BAD_PACKFILE);
503 return err;
506 static const struct got_error *
507 open_packfile(FILE **packfile, const char *path_packfile,
508 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
510 const struct got_error *err = NULL;
512 *packfile = NULL;
514 *packfile = fopen(path_packfile, "rb");
515 if (*packfile == NULL) {
516 err = got_error_from_errno();
517 return err;
520 if (packidx) {
521 err = read_packfile_hdr(*packfile, packidx);
522 if (err) {
523 fclose(*packfile);
524 *packfile = NULL;
527 return err;
530 void
531 got_pack_close(struct got_pack *pack)
533 fclose(pack->packfile);
534 pack->packfile = NULL;
535 free(pack->path_packfile);
536 pack->path_packfile = NULL;
537 pack->filesize = 0;
540 static const struct got_error *
541 cache_pack(struct got_pack **packp, const char *path_packfile,
542 struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
544 const struct got_error *err = NULL;
545 struct got_pack *pack = NULL;
546 int i;
548 if (packp)
549 *packp = NULL;
551 for (i = 0; i < nitems(repo->packs); i++) {
552 pack = &repo->packs[i];
553 if (pack->path_packfile == NULL)
554 break;
555 if (strcmp(pack->path_packfile, path_packfile) == 0)
556 return NULL;
559 if (i == nitems(repo->packs) - 1) {
560 got_pack_close(&repo->packs[i - 1]);
561 memmove(&repo->packs[1], &repo->packs[0],
562 sizeof(repo->packs) - sizeof(repo->packs[0]));
563 i = 0;
566 pack = &repo->packs[i];
568 pack->path_packfile = strdup(path_packfile);
569 if (pack->path_packfile == NULL) {
570 err = got_error_from_errno();
571 goto done;
574 err = open_packfile(&pack->packfile, path_packfile, repo, packidx);
575 if (err)
576 goto done;
578 err = get_packfile_size(&pack->filesize, path_packfile);
579 done:
580 if (err) {
581 if (pack)
582 free(pack->path_packfile);
583 free(pack);
584 } else if (packp)
585 *packp = pack;
586 return err;
589 struct got_pack *
590 get_cached_pack(const char *path_packfile, struct got_repository *repo)
592 struct got_pack *pack = NULL;
593 int i;
595 for (i = 0; i < nitems(repo->packs); i++) {
596 pack = &repo->packs[i];
597 if (pack->path_packfile == NULL)
598 break;
599 if (strcmp(pack->path_packfile, path_packfile) == 0)
600 return pack;
603 return NULL;
606 static const struct got_error *
607 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
608 FILE *packfile)
610 uint8_t t = 0;
611 uint64_t s = 0;
612 uint8_t sizeN;
613 size_t n;
614 int i = 0;
616 do {
617 /* We do not support size values which don't fit in 64 bit. */
618 if (i > 9)
619 return got_error(GOT_ERR_NO_SPACE);
621 n = fread(&sizeN, sizeof(sizeN), 1, packfile);
622 if (n != 1)
623 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
625 if (i == 0) {
626 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
627 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
628 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
629 } else {
630 size_t shift = 4 + 7 * (i - 1);
631 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
633 i++;
634 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
636 *type = t;
637 *size = s;
638 *len = i * sizeof(sizeN);
639 return NULL;
642 static const struct got_error *
643 open_plain_object(struct got_object **obj, const char *path_packfile,
644 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
646 *obj = calloc(1, sizeof(**obj));
647 if (*obj == NULL)
648 return got_error_from_errno();
650 (*obj)->path_packfile = strdup(path_packfile);
651 if ((*obj)->path_packfile == NULL) {
652 const struct got_error *err = got_error_from_errno();
653 free(*obj);
654 *obj = NULL;
655 return err;
658 (*obj)->type = type;
659 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
660 (*obj)->hdrlen = 0;
661 (*obj)->size = size;
662 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
663 (*obj)->pack_offset = offset;
665 return NULL;
668 static const struct got_error *
669 parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
671 int64_t o = 0;
672 uint8_t offN;
673 size_t n;
674 int i = 0;
676 do {
677 /* We do not support offset values which don't fit in 64 bit. */
678 if (i > 8)
679 return got_error(GOT_ERR_NO_SPACE);
681 n = fread(&offN, sizeof(offN), 1, packfile);
682 if (n != 1)
683 return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
685 if (i == 0)
686 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
687 else {
688 o++;
689 o <<= 7;
690 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
692 i++;
693 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
695 *offset = o;
696 *len = i * sizeof(offN);
697 return NULL;
700 static const struct got_error *
701 parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
703 const struct got_error *err;
704 int64_t negoffset;
705 size_t negofflen;
707 err = parse_negative_offset(&negoffset, &negofflen, packfile);
708 if (err)
709 return err;
711 /* Compute the base object's offset (must be in the same pack file). */
712 *base_offset = (offset - negoffset);
713 if (*base_offset <= 0)
714 return got_error(GOT_ERR_BAD_PACKFILE);
716 return NULL;
719 static const struct got_error *
720 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
721 FILE *, const char *, off_t, size_t, int, size_t);
723 static const struct got_error *
724 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
725 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
726 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
728 struct got_delta *delta;
730 delta = got_delta_open(path_packfile, delta_offset, tslen,
731 delta_type, delta_size, delta_data_offset, delta_buf,
732 delta_len);
733 if (delta == NULL)
734 return got_error_from_errno();
735 /* delta is freed in got_object_close() */
736 deltas->nentries++;
737 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
738 return NULL;
741 static const struct got_error *
742 resolve_offset_delta(struct got_delta_chain *deltas,
743 struct got_repository *repo, FILE *packfile, const char *path_packfile,
744 off_t delta_offset,size_t tslen, int delta_type, size_t delta_size)
747 const struct got_error *err;
748 off_t base_offset;
749 uint8_t base_type;
750 uint64_t base_size;
751 size_t base_tslen;
752 off_t delta_data_offset;
753 uint8_t *delta_buf;
754 size_t delta_len;
756 err = parse_offset_delta(&base_offset, packfile, delta_offset);
757 if (err)
758 return err;
760 delta_data_offset = ftello(packfile);
761 if (delta_data_offset == -1)
762 return got_error_from_errno();
764 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
765 if (err)
766 return err;
768 err = add_delta(deltas, path_packfile, delta_offset, tslen,
769 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
770 if (err)
771 return err;
773 /* An offset delta must be in the same packfile. */
774 if (fseeko(packfile, base_offset, SEEK_SET) != 0)
775 return got_error_from_errno();
777 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
778 packfile);
779 if (err)
780 return err;
782 return resolve_delta_chain(deltas, repo, packfile, path_packfile,
783 base_offset, base_tslen, base_type, base_size);
786 static const struct got_error *
787 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
788 FILE *packfile, const char *path_packfile, off_t delta_offset,
789 size_t tslen, int delta_type, size_t delta_size)
791 const struct got_error *err;
792 struct got_object_id id;
793 struct got_packidx_v2_hdr *packidx;
794 int idx;
795 off_t base_offset;
796 uint8_t base_type;
797 uint64_t base_size;
798 size_t base_tslen;
799 size_t n;
800 char *path_base_packfile;
801 struct got_pack *base_pack;
802 off_t delta_data_offset;
803 uint8_t *delta_buf;
804 size_t delta_len;
806 n = fread(&id, sizeof(id), 1, packfile);
807 if (n != 1)
808 return got_ferror(packfile, GOT_ERR_IO);
810 delta_data_offset = ftello(packfile);
811 if (delta_data_offset == -1)
812 return got_error_from_errno();
814 err = got_inflate_to_mem(&delta_buf, &delta_len, packfile);
815 if (err)
816 return err;
818 err = add_delta(deltas, path_packfile, delta_offset, tslen,
819 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
820 if (err)
821 return err;
823 err = search_packidx(&packidx, &idx, repo, &id);
824 if (err)
825 return err;
827 base_offset = get_object_offset(packidx, idx);
828 if (base_offset == (uint64_t)-1) {
829 return got_error(GOT_ERR_BAD_PACKIDX);
832 err = get_packfile_path(&path_base_packfile, repo, packidx);
833 if (err)
834 return err;
836 base_pack = get_cached_pack(path_base_packfile, repo);
837 if (base_pack == NULL) {
838 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
839 if (err)
840 goto done;
843 if (fseeko(base_pack->packfile, base_offset, SEEK_SET) != 0) {
844 err = got_error_from_errno();
845 goto done;
848 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
849 base_pack->packfile);
850 if (err)
851 goto done;
853 err = resolve_delta_chain(deltas, repo, base_pack->packfile,
854 path_base_packfile, base_offset, base_tslen, base_type,
855 base_size);
856 done:
857 free(path_base_packfile);
858 return err;
861 static const struct got_error *
862 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
863 FILE *packfile, const char *path_packfile, off_t delta_offset, size_t tslen,
864 int delta_type, size_t delta_size)
866 const struct got_error *err = NULL;
868 switch (delta_type) {
869 case GOT_OBJ_TYPE_COMMIT:
870 case GOT_OBJ_TYPE_TREE:
871 case GOT_OBJ_TYPE_BLOB:
872 case GOT_OBJ_TYPE_TAG:
873 /* Plain types are the final delta base. Recursion ends. */
874 err = add_delta(deltas, path_packfile, delta_offset, tslen,
875 delta_type, delta_size, 0, NULL, 0);
876 break;
877 case GOT_OBJ_TYPE_OFFSET_DELTA:
878 err = resolve_offset_delta(deltas, repo, packfile,
879 path_packfile, delta_offset, tslen, delta_type,
880 delta_size);
881 break;
882 case GOT_OBJ_TYPE_REF_DELTA:
883 err = resolve_ref_delta(deltas, repo, packfile,
884 path_packfile, delta_offset, tslen, delta_type,
885 delta_size);
886 break;
887 default:
888 return got_error(GOT_ERR_NOT_IMPL);
891 return err;
894 static const struct got_error *
895 open_delta_object(struct got_object **obj, struct got_repository *repo,
896 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
897 FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
898 int delta_type, size_t delta_size)
900 const struct got_error *err = NULL;
901 int resolved_type;
903 *obj = calloc(1, sizeof(**obj));
904 if (*obj == NULL)
905 return got_error_from_errno();
907 (*obj)->flags = 0;
908 (*obj)->hdrlen = 0;
909 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
910 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
911 (*obj)->pack_offset = offset + tslen;
913 (*obj)->path_packfile = strdup(path_packfile);
914 if ((*obj)->path_packfile == NULL) {
915 err = got_error_from_errno();
916 goto done;
918 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
920 SIMPLEQ_INIT(&(*obj)->deltas.entries);
921 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
923 err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
924 path_packfile, offset, tslen, delta_type, delta_size);
925 if (err)
926 goto done;
928 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
929 if (err)
930 goto done;
931 (*obj)->type = resolved_type;
933 done:
934 if (err) {
935 got_object_close(*obj);
936 *obj = NULL;
938 return err;
941 static const struct got_error *
942 open_packed_object(struct got_object **obj, struct got_repository *repo,
943 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
945 const struct got_error *err = NULL;
946 off_t offset;
947 char *path_packfile;
948 struct got_pack *pack;
949 uint8_t type;
950 uint64_t size;
951 size_t tslen;
953 *obj = NULL;
955 offset = get_object_offset(packidx, idx);
956 if (offset == (uint64_t)-1)
957 return got_error(GOT_ERR_BAD_PACKIDX);
959 err = get_packfile_path(&path_packfile, repo, packidx);
960 if (err)
961 return err;
963 pack = get_cached_pack(path_packfile, repo);
964 if (pack == NULL) {
965 err = cache_pack(&pack, path_packfile, packidx, repo);
966 if (err)
967 goto done;
970 if (fseeko(pack->packfile, offset, SEEK_SET) != 0) {
971 err = got_error_from_errno();
972 goto done;
975 err = parse_object_type_and_size(&type, &size, &tslen, pack->packfile);
976 if (err)
977 goto done;
979 switch (type) {
980 case GOT_OBJ_TYPE_COMMIT:
981 case GOT_OBJ_TYPE_TREE:
982 case GOT_OBJ_TYPE_BLOB:
983 case GOT_OBJ_TYPE_TAG:
984 err = open_plain_object(obj, path_packfile, id, type,
985 offset + tslen, size);
986 break;
988 case GOT_OBJ_TYPE_OFFSET_DELTA:
989 case GOT_OBJ_TYPE_REF_DELTA:
990 err = open_delta_object(obj, repo, packidx, path_packfile,
991 pack->packfile, id, offset, tslen, type, size);
992 break;
994 default:
995 err = got_error(GOT_ERR_NOT_IMPL);
996 goto done;
998 done:
999 free(path_packfile);
1000 return err;
1003 const struct got_error *
1004 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1005 struct got_repository *repo)
1007 const struct got_error *err = NULL;
1008 struct got_packidx_v2_hdr *packidx = NULL;
1009 int idx;
1011 err = search_packidx(&packidx, &idx, repo, id);
1012 if (err)
1013 return err;
1015 err = open_packed_object(obj, repo, packidx, idx, id);
1016 if (err)
1017 return err;
1019 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1020 return err;
1023 static const struct got_error *
1024 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1026 struct got_delta *delta;
1027 uint64_t base_size = 0, result_size = 0;
1029 *max_size = 0;
1030 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1031 /* Plain object types are the delta base. */
1032 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1033 delta->type != GOT_OBJ_TYPE_TREE &&
1034 delta->type != GOT_OBJ_TYPE_BLOB &&
1035 delta->type != GOT_OBJ_TYPE_TAG) {
1036 const struct got_error *err;
1037 err = got_delta_get_sizes(&base_size, &result_size,
1038 delta->delta_buf, delta->delta_len);
1039 if (err)
1040 return err;
1041 } else
1042 base_size = delta->size;
1043 if (base_size > *max_size)
1044 *max_size = base_size;
1045 if (result_size > *max_size)
1046 *max_size = result_size;
1049 return NULL;
1052 static const struct got_error *
1053 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1054 FILE *outfile, struct got_repository *repo)
1056 const struct got_error *err = NULL;
1057 struct got_delta *delta;
1058 FILE *base_file = NULL, *accum_file = NULL;
1059 uint8_t *base_buf = NULL, *accum_buf = NULL;
1060 size_t accum_size = 0;
1061 uint64_t max_size;
1062 int n = 0;
1064 *result_size = 0;
1066 if (SIMPLEQ_EMPTY(&deltas->entries))
1067 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1069 /* We process small enough files entirely in memory for speed. */
1070 err = get_delta_chain_max_size(&max_size, deltas);
1071 if (err)
1072 return err;
1073 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1074 accum_buf = malloc(max_size);
1075 if (accum_buf == NULL)
1076 return got_error_from_errno();
1077 } else {
1078 base_file = got_opentemp();
1079 if (base_file == NULL)
1080 return got_error_from_errno();
1082 accum_file = got_opentemp();
1083 if (accum_file == NULL) {
1084 err = got_error_from_errno();
1085 fclose(base_file);
1086 return err;
1090 /* Deltas are ordered in ascending order. */
1091 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1092 if (n == 0) {
1093 struct got_pack *pack;
1094 size_t base_len;
1096 /* Plain object types are the delta base. */
1097 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1098 delta->type != GOT_OBJ_TYPE_TREE &&
1099 delta->type != GOT_OBJ_TYPE_BLOB &&
1100 delta->type != GOT_OBJ_TYPE_TAG) {
1101 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1102 goto done;
1105 pack = get_cached_pack(delta->path_packfile, repo);
1106 if (pack == NULL) {
1107 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1108 goto done;
1111 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1112 SEEK_SET) != 0) {
1113 err = got_error_from_errno();
1114 goto done;
1116 if (base_file)
1117 err = got_inflate_to_file(&base_len,
1118 pack->packfile, base_file);
1119 else {
1120 err = got_inflate_to_mem(&base_buf, &base_len,
1121 pack->packfile);
1122 if (base_len < max_size) {
1123 uint8_t *p;
1124 p = reallocarray(base_buf, 1, max_size);
1125 if (p == NULL) {
1126 err = got_error_from_errno();
1127 goto done;
1129 base_buf = p;
1132 if (err)
1133 goto done;
1134 n++;
1135 if (base_file)
1136 rewind(base_file);
1137 continue;
1140 if (base_buf) {
1141 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1142 delta->delta_len, accum_buf, &accum_size);
1143 n++;
1144 } else {
1145 err = got_delta_apply(base_file, delta->delta_buf,
1146 delta->delta_len,
1147 /* Final delta application writes to output file. */
1148 ++n < deltas->nentries ? accum_file : outfile,
1149 &accum_size);
1151 if (err)
1152 goto done;
1154 if (n < deltas->nentries) {
1155 /* Accumulated delta becomes the new base. */
1156 if (base_buf) {
1157 uint8_t *tmp = accum_buf;
1158 accum_buf = base_buf;
1159 base_buf = tmp;
1160 } else {
1161 FILE *tmp = accum_file;
1162 accum_file = base_file;
1163 base_file = tmp;
1164 rewind(base_file);
1165 rewind(accum_file);
1170 done:
1171 free(base_buf);
1172 if (accum_buf) {
1173 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1174 free(accum_buf);
1175 if (len != accum_size)
1176 return got_ferror(outfile, GOT_ERR_IO);
1178 if (base_file)
1179 fclose(base_file);
1180 if (accum_file)
1181 fclose(accum_file);
1182 rewind(outfile);
1183 if (err == NULL)
1184 *result_size = accum_size;
1185 return err;
1188 static const struct got_error *
1189 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1190 struct got_delta_chain *deltas, struct got_repository *repo)
1192 const struct got_error *err = NULL;
1193 struct got_delta *delta;
1194 uint8_t *base_buf = NULL, *accum_buf = NULL;
1195 size_t accum_size;
1196 uint64_t max_size;
1197 int n = 0;
1199 *outbuf = NULL;
1200 *outlen = 0;
1202 if (SIMPLEQ_EMPTY(&deltas->entries))
1203 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1205 err = get_delta_chain_max_size(&max_size, deltas);
1206 if (err)
1207 return err;
1208 accum_buf = malloc(max_size);
1209 if (accum_buf == NULL)
1210 return got_error_from_errno();
1212 /* Deltas are ordered in ascending order. */
1213 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1214 if (n == 0) {
1215 struct got_pack *pack;
1216 size_t base_len;
1218 /* Plain object types are the delta base. */
1219 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1220 delta->type != GOT_OBJ_TYPE_TREE &&
1221 delta->type != GOT_OBJ_TYPE_BLOB &&
1222 delta->type != GOT_OBJ_TYPE_TAG) {
1223 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1224 goto done;
1227 pack = get_cached_pack(delta->path_packfile, repo);
1228 if (pack == NULL) {
1229 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1230 goto done;
1233 if (fseeko(pack->packfile, delta->offset + delta->tslen,
1234 SEEK_SET) != 0) {
1235 err = got_error_from_errno();
1236 goto done;
1238 err = got_inflate_to_mem(&base_buf, &base_len,
1239 pack->packfile);
1240 if (base_len < max_size) {
1241 uint8_t *p;
1242 p = reallocarray(base_buf, 1, max_size);
1243 if (p == NULL) {
1244 err = got_error_from_errno();
1245 goto done;
1247 base_buf = p;
1249 if (err)
1250 goto done;
1251 n++;
1252 continue;
1255 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1256 delta->delta_len, accum_buf, &accum_size);
1257 n++;
1258 if (err)
1259 goto done;
1261 if (n < deltas->nentries) {
1262 /* Accumulated delta becomes the new base. */
1263 uint8_t *tmp = accum_buf;
1264 accum_buf = base_buf;
1265 base_buf = tmp;
1269 done:
1270 free(base_buf);
1271 if (err) {
1272 free(accum_buf);
1273 *outbuf = NULL;
1274 *outlen = 0;
1275 } else {
1276 *outbuf = accum_buf;
1277 *outlen = accum_size;
1279 return err;
1282 const struct got_error *
1283 got_packfile_extract_object(FILE **f, struct got_object *obj,
1284 struct got_repository *repo)
1286 const struct got_error *err = NULL;
1288 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1289 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1291 *f = got_opentemp();
1292 if (*f == NULL) {
1293 err = got_error(GOT_ERR_FILE_OPEN);
1294 goto done;
1297 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1298 struct got_pack *pack;
1300 pack = get_cached_pack(obj->path_packfile, repo);
1301 if (pack == NULL) {
1302 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1303 if (err)
1304 goto done;
1307 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1308 err = got_error_from_errno();
1309 goto done;
1312 err = got_inflate_to_file(&obj->size, pack->packfile, *f);
1313 } else
1314 err = dump_delta_chain_to_file(&obj->size, &obj->deltas, *f,
1315 repo);
1316 done:
1317 if (err && *f) {
1318 fclose(*f);
1319 *f = NULL;
1321 return err;
1324 const struct got_error *
1325 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1326 struct got_object *obj, struct got_repository *repo)
1328 const struct got_error *err = NULL;
1330 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1331 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1333 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1334 struct got_pack *pack;
1336 pack = get_cached_pack(obj->path_packfile, repo);
1337 if (pack == NULL) {
1338 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1339 if (err)
1340 goto done;
1343 if (fseeko(pack->packfile, obj->pack_offset, SEEK_SET) != 0) {
1344 err = got_error_from_errno();
1345 goto done;
1348 err = got_inflate_to_mem(buf, len, pack->packfile);
1349 } else
1350 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);
1351 done:
1352 return err;