2 * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
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.
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.
17 #include <sys/types.h>
19 #include <sys/queue.h>
33 #include "got_error.h"
34 #include "got_object.h"
35 #include "got_repository.h"
36 #include "got_opentemp.h"
38 #include "got_lib_sha1.h"
39 #include "got_lib_pack.h"
40 #include "got_lib_path.h"
41 #include "got_lib_delta.h"
42 #include "got_lib_zbuf.h"
43 #include "got_lib_object.h"
44 #include "got_lib_repository.h"
47 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
50 #define GOT_PACK_PREFIX "pack-"
51 #define GOT_PACKFILE_SUFFIX ".pack"
52 #define GOT_PACKIDX_SUFFIX ".idx"
53 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
54 SHA1_DIGEST_STRING_LENGTH - 1 + \
55 strlen(GOT_PACKFILE_SUFFIX))
56 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
57 SHA1_DIGEST_STRING_LENGTH - 1 + \
58 strlen(GOT_PACKIDX_SUFFIX))
61 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
64 static const struct got_error *
65 verify_fanout_table(uint32_t *fanout_table)
69 for (i = 0; i < 0xff - 1; i++) {
70 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
71 return got_error(GOT_ERR_BAD_PACKIDX);
77 static const struct got_error *
78 get_packfile_size(size_t *size, const char *path)
85 dot = strrchr(path, '.');
87 return got_error(GOT_ERR_BAD_PATH);
89 /* Path must point to a pack index or to a pack file. */
90 if (strcmp(dot, ".idx") == 0) {
91 const struct got_error *err = NULL;
93 char base_path[PATH_MAX];
95 /* Convert pack index path to pack file path. */
96 if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
97 return got_error(GOT_ERR_NO_SPACE);
98 dot = strrchr(base_path, '.');
100 return got_error(GOT_ERR_BAD_PATH);
102 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
103 return got_error_from_errno();
105 if (stat(path_pack, &sb) != 0)
106 err = got_error_from_errno();
110 } else if (strcmp(dot, ".pack") == 0) {
111 if (stat(path, &sb) != 0)
112 return got_error_from_errno();
114 return got_error(GOT_ERR_BAD_PATH);
120 const struct got_error *
121 got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
123 struct got_packidx_v2_hdr *p;
125 const struct got_error *err = NULL;
126 size_t n, nobj, packfile_size;
128 uint8_t sha1[SHA1_DIGEST_LENGTH];
132 f = fopen(path, "rb");
134 return got_error_from_errno();
136 err = get_packfile_size(&packfile_size, path);
140 p = calloc(1, sizeof(*p));
142 err = got_error_from_errno();
146 n = fread(&p->magic, sizeof(p->magic), 1, f);
148 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
152 if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
153 err = got_error(GOT_ERR_BAD_PACKIDX);
157 SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
159 n = fread(&p->version, sizeof(p->version), 1, f);
161 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
165 if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
166 err = got_error(GOT_ERR_BAD_PACKIDX);
170 SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
172 n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
174 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
178 err = verify_fanout_table(p->fanout_table);
182 SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
184 nobj = betoh32(p->fanout_table[0xff]);
186 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
187 if (p->sorted_ids == NULL) {
188 err = got_error_from_errno();
192 n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
194 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
198 SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
199 nobj * sizeof(*p->sorted_ids));
201 p->crc32 = calloc(nobj, sizeof(*p->crc32));
202 if (p->crc32 == NULL) {
203 err = got_error_from_errno();
207 n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
209 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
213 SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
215 p->offsets = calloc(nobj, sizeof(*p->offsets));
216 if (p->offsets == NULL) {
217 err = got_error_from_errno();
221 n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
223 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
227 SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
229 /* Large file offsets are contained only in files > 2GB. */
230 if (packfile_size <= 0x80000000)
233 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
234 if (p->large_offsets == NULL) {
235 err = got_error_from_errno();
239 n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
241 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
245 SHA1Update(&ctx, (uint8_t*)p->large_offsets,
246 nobj * sizeof(*p->large_offsets));
249 n = fread(&p->trailer, sizeof(p->trailer), 1, f);
251 err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
255 SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
256 SHA1Final(sha1, &ctx);
257 if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
258 err = got_error(GOT_ERR_PACKIDX_CSUM);
262 got_packidx_close(p);
269 got_packidx_close(struct got_packidx_v2_hdr *packidx)
271 free(packidx->sorted_ids);
272 free(packidx->offsets);
273 free(packidx->crc32);
274 free(packidx->large_offsets);
279 is_packidx_filename(const char *name, size_t len)
281 if (len != GOT_PACKIDX_NAMELEN)
284 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
287 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
288 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
295 get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
297 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
298 uint32_t offset = betoh32(packidx->offsets[idx]);
299 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
301 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
302 if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
304 loffset = betoh64(packidx->large_offsets[idx]);
305 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
307 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
310 static const struct got_error *
311 get_packfile_path(char **path_packfile, struct got_repository *repo,
312 struct got_packidx_v2_hdr *packidx);
315 get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id, struct got_repository *repo)
317 u_int8_t id0 = id->sha1[0];
318 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
322 i = betoh32(packidx->fanout_table[id0 - 1]);
325 struct got_object_id *oid = &packidx->sorted_ids[i];
326 int cmp = got_object_id_cmp(id, oid);
336 static struct got_packidx_v2_hdr *
337 dup_packidx(struct got_packidx_v2_hdr *packidx)
339 struct got_packidx_v2_hdr *p;
342 p = calloc(1, sizeof(*p));
346 memcpy(p, packidx, sizeof(*p));
347 p->sorted_ids = NULL;
350 p->large_offsets = NULL;
352 nobj = betoh32(p->fanout_table[0xff]);
354 p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
355 if (p->sorted_ids == NULL)
357 memcpy(p->sorted_ids, packidx->sorted_ids,
358 nobj * sizeof(*p->sorted_ids));
360 p->crc32 = calloc(nobj, sizeof(*p->crc32));
361 if (p->crc32 == NULL)
363 memcpy(p->crc32, packidx->crc32, nobj * sizeof(*p->crc32));
365 p->offsets = calloc(nobj, sizeof(*p->offsets));
366 if (p->offsets == NULL)
368 memcpy(p->offsets, packidx->offsets, nobj * sizeof(*p->offsets));
370 if (p->large_offsets) {
371 p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
372 if (p->large_offsets == NULL)
374 memcpy(p->large_offsets, packidx->large_offsets,
375 nobj * sizeof(*p->large_offsets));
381 free(p->large_offsets);
390 cache_packidx(struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
394 for (i = 0; i < nitems(repo->packidx_cache); i++) {
395 if (repo->packidx_cache[i] == NULL)
399 if (i == nitems(repo->packidx_cache)) {
400 got_packidx_close(repo->packidx_cache[i - 1]);
401 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
402 sizeof(repo->packidx_cache) -
403 sizeof(repo->packidx_cache[0]));
407 repo->packidx_cache[i] = dup_packidx(packidx);
410 static const struct got_error *
411 search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
412 struct got_repository *repo, struct got_object_id *id)
414 const struct got_error *err;
421 /* Search pack index cache. */
422 for (i = 0; i < nitems(repo->packidx_cache); i++) {
423 if (repo->packidx_cache[i] == NULL)
425 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
427 *packidx = repo->packidx_cache[i];
431 /* No luck. Search the filesystem. */
433 path_packdir = got_repo_get_path_objects_pack(repo);
434 if (path_packdir == NULL)
435 return got_error_from_errno();
437 packdir = opendir(path_packdir);
438 if (packdir == NULL) {
439 err = got_error_from_errno();
443 while ((dent = readdir(packdir)) != NULL) {
444 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
447 if (asprintf(&path_packidx, "%s/%s", path_packdir,
448 dent->d_name) == -1) {
449 err = got_error_from_errno();
453 err = got_packidx_open(packidx, path_packidx);
458 *idx = get_object_idx(*packidx, id, repo);
460 err = NULL; /* found the object */
461 cache_packidx(*packidx, repo);
465 got_packidx_close(*packidx);
469 err = got_error(GOT_ERR_NO_OBJ);
472 if (packdir && closedir(packdir) != 0 && err == 0)
473 err = got_error_from_errno();
477 static const struct got_error *
478 get_packfile_path(char **path_packfile, struct got_repository *repo,
479 struct got_packidx_v2_hdr *packidx)
482 char hex[SHA1_DIGEST_STRING_LENGTH];
485 path_packdir = got_repo_get_path_objects_pack(repo);
486 if (path_packdir == NULL)
487 return got_error_from_errno();
489 sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
492 return got_error(GOT_ERR_PACKIDX_CSUM);
494 if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
495 GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
496 *path_packfile = NULL;
497 return got_error_from_errno();
503 const struct got_error *
504 read_packfile_hdr(int fd, struct got_packidx_v2_hdr *packidx)
506 const struct got_error *err = NULL;
507 uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
508 struct got_packfile_hdr hdr;
511 n = read(fd, &hdr, sizeof(hdr));
513 return got_error_from_errno();
514 if (n != sizeof(hdr))
515 return got_error(GOT_ERR_BAD_PACKFILE);
517 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
518 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
519 betoh32(hdr.nobjects) != totobj)
520 err = got_error(GOT_ERR_BAD_PACKFILE);
525 static const struct got_error *
526 open_packfile(int *fd, const char *path_packfile,
527 struct got_repository *repo, struct got_packidx_v2_hdr *packidx)
529 const struct got_error *err = NULL;
531 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
533 return got_error_from_errno();
536 err = read_packfile_hdr(*fd, packidx);
546 got_pack_close(struct got_pack *pack)
550 free(pack->path_packfile);
551 pack->path_packfile = NULL;
555 static const struct got_error *
556 cache_pack(struct got_pack **packp, const char *path_packfile,
557 struct got_packidx_v2_hdr *packidx, struct got_repository *repo)
559 const struct got_error *err = NULL;
560 struct got_pack *pack = NULL;
566 for (i = 0; i < nitems(repo->packs); i++) {
567 pack = &repo->packs[i];
568 if (pack->path_packfile == NULL)
570 if (strcmp(pack->path_packfile, path_packfile) == 0)
574 if (i == nitems(repo->packs) - 1) {
575 got_pack_close(&repo->packs[i - 1]);
576 memmove(&repo->packs[1], &repo->packs[0],
577 sizeof(repo->packs) - sizeof(repo->packs[0]));
581 pack = &repo->packs[i];
583 pack->path_packfile = strdup(path_packfile);
584 if (pack->path_packfile == NULL) {
585 err = got_error_from_errno();
589 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
593 err = get_packfile_size(&pack->filesize, path_packfile);
597 free(pack->path_packfile);
598 memset(pack, 0, sizeof(*pack));
606 get_cached_pack(const char *path_packfile, struct got_repository *repo)
608 struct got_pack *pack = NULL;
611 for (i = 0; i < nitems(repo->packs); i++) {
612 pack = &repo->packs[i];
613 if (pack->path_packfile == NULL)
615 if (strcmp(pack->path_packfile, path_packfile) == 0)
622 static const struct got_error *
623 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len, int fd)
632 /* We do not support size values which don't fit in 64 bit. */
634 return got_error(GOT_ERR_NO_SPACE);
636 n = read(fd, &sizeN, sizeof(sizeN));
638 return got_error_from_errno();
639 if (n != sizeof(sizeN))
640 return got_error(GOT_ERR_BAD_PACKFILE);
643 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
644 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
645 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
647 size_t shift = 4 + 7 * (i - 1);
648 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
651 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
655 *len = i * sizeof(sizeN);
659 static const struct got_error *
660 open_plain_object(struct got_object **obj, const char *path_packfile,
661 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
663 *obj = calloc(1, sizeof(**obj));
665 return got_error_from_errno();
667 (*obj)->path_packfile = strdup(path_packfile);
668 if ((*obj)->path_packfile == NULL) {
669 const struct got_error *err = got_error_from_errno();
676 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
679 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
680 (*obj)->pack_offset = offset;
685 static const struct got_error *
686 parse_negative_offset(int64_t *offset, size_t *len, int fd)
694 /* We do not support offset values which don't fit in 64 bit. */
696 return got_error(GOT_ERR_NO_SPACE);
698 n = read(fd, &offN, sizeof(offN));
700 return got_error_from_errno();
701 if (n != sizeof(offN))
702 return got_error(GOT_ERR_BAD_PACKFILE);
705 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
709 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
712 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
715 *len = i * sizeof(offN);
719 static const struct got_error *
720 parse_offset_delta(off_t *base_offset, int fd, off_t offset)
722 const struct got_error *err;
726 err = parse_negative_offset(&negoffset, &negofflen, fd);
730 /* Compute the base object's offset (must be in the same pack file). */
731 *base_offset = (offset - negoffset);
732 if (*base_offset <= 0)
733 return got_error(GOT_ERR_BAD_PACKFILE);
738 static const struct got_error *
739 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
740 int, size_t, const char *, off_t, size_t, int, size_t, unsigned int);
742 static const struct got_error *
743 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
744 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
745 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
747 struct got_delta *delta;
749 delta = got_delta_open(path_packfile, delta_offset, tslen,
750 delta_type, delta_size, delta_data_offset, delta_buf,
753 return got_error_from_errno();
754 /* delta is freed in got_object_close() */
756 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
760 static const struct got_error *
761 resolve_offset_delta(struct got_delta_chain *deltas,
762 struct got_repository *repo, int fd, size_t packfile_size,
763 const char *path_packfile, off_t delta_offset, size_t tslen,
764 int delta_type, size_t delta_size, unsigned int recursion)
767 const struct got_error *err;
772 off_t delta_data_offset;
776 err = parse_offset_delta(&base_offset, fd, delta_offset);
780 delta_data_offset = lseek(fd, 0, SEEK_CUR);
781 if (delta_data_offset == -1)
782 return got_error_from_errno();
784 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
788 err = add_delta(deltas, path_packfile, delta_offset, tslen,
789 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
793 /* An offset delta must be in the same packfile. */
794 if (base_offset >= packfile_size)
795 return got_error(GOT_ERR_PACK_OFFSET);
796 if (lseek(fd, base_offset, SEEK_SET) == -1)
797 return got_error_from_errno();
799 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
804 return resolve_delta_chain(deltas, repo, fd, packfile_size,
805 path_packfile, base_offset, base_tslen, base_type, base_size,
809 static const struct got_error *
810 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
811 int fd, const char *path_packfile, off_t delta_offset,
812 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
814 const struct got_error *err;
815 struct got_object_id id;
816 struct got_packidx_v2_hdr *packidx;
823 char *path_base_packfile;
824 struct got_pack *base_pack;
825 off_t delta_data_offset;
829 n = read(fd, &id, sizeof(id));
831 return got_error_from_errno();
833 return got_error(GOT_ERR_BAD_PACKFILE);
835 delta_data_offset = lseek(fd, 0, SEEK_CUR);
836 if (delta_data_offset == -1)
837 return got_error_from_errno();
839 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
843 err = add_delta(deltas, path_packfile, delta_offset, tslen,
844 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
848 err = search_packidx(&packidx, &idx, repo, &id);
852 base_offset = get_object_offset(packidx, idx);
853 if (base_offset == (uint64_t)-1) {
854 return got_error(GOT_ERR_BAD_PACKIDX);
857 err = get_packfile_path(&path_base_packfile, repo, packidx);
861 base_pack = get_cached_pack(path_base_packfile, repo);
862 if (base_pack == NULL) {
863 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
868 if (base_offset >= base_pack->filesize) {
869 err = got_error(GOT_ERR_PACK_OFFSET);
872 if (lseek(base_pack->fd, base_offset, SEEK_SET) == -1) {
873 err = got_error_from_errno();
877 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
882 err = resolve_delta_chain(deltas, repo, base_pack->fd,
883 base_pack->filesize, path_base_packfile, base_offset,
884 base_tslen, base_type, base_size, recursion - 1);
886 free(path_base_packfile);
890 static const struct got_error *
891 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
892 int fd, size_t packfile_size, const char *path_packfile, off_t delta_offset,
893 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
895 const struct got_error *err = NULL;
897 if (--recursion == 0)
898 return got_error(GOT_ERR_RECURSION);
900 switch (delta_type) {
901 case GOT_OBJ_TYPE_COMMIT:
902 case GOT_OBJ_TYPE_TREE:
903 case GOT_OBJ_TYPE_BLOB:
904 case GOT_OBJ_TYPE_TAG:
905 /* Plain types are the final delta base. Recursion ends. */
906 err = add_delta(deltas, path_packfile, delta_offset, tslen,
907 delta_type, delta_size, 0, NULL, 0);
909 case GOT_OBJ_TYPE_OFFSET_DELTA:
910 err = resolve_offset_delta(deltas, repo, fd, packfile_size,
911 path_packfile, delta_offset, tslen, delta_type, delta_size,
914 case GOT_OBJ_TYPE_REF_DELTA:
915 err = resolve_ref_delta(deltas, repo, fd, path_packfile,
916 delta_offset, tslen, delta_type, delta_size, recursion - 1);
919 return got_error(GOT_ERR_OBJ_TYPE);
925 static const struct got_error *
926 open_delta_object(struct got_object **obj, struct got_repository *repo,
927 struct got_packidx_v2_hdr *packidx, const char *path_packfile,
928 int fd, size_t packfile_size, struct got_object_id *id, off_t offset,
929 size_t tslen, int delta_type, size_t delta_size)
931 const struct got_error *err = NULL;
934 *obj = calloc(1, sizeof(**obj));
936 return got_error_from_errno();
940 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
941 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
942 (*obj)->pack_offset = offset + tslen;
944 (*obj)->path_packfile = strdup(path_packfile);
945 if ((*obj)->path_packfile == NULL) {
946 err = got_error_from_errno();
949 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
951 SIMPLEQ_INIT(&(*obj)->deltas.entries);
952 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
954 err = resolve_delta_chain(&(*obj)->deltas, repo, fd, packfile_size,
955 path_packfile, offset, tslen, delta_type, delta_size,
956 GOT_DELTA_CHAIN_RECURSION_MAX);
960 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
963 (*obj)->type = resolved_type;
967 got_object_close(*obj);
973 static const struct got_error *
974 open_packed_object(struct got_object **obj, struct got_repository *repo,
975 struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
977 const struct got_error *err = NULL;
980 struct got_pack *pack;
987 offset = get_object_offset(packidx, idx);
988 if (offset == (uint64_t)-1)
989 return got_error(GOT_ERR_BAD_PACKIDX);
991 err = get_packfile_path(&path_packfile, repo, packidx);
995 pack = get_cached_pack(path_packfile, repo);
997 err = cache_pack(&pack, path_packfile, packidx, repo);
1002 if (offset >= pack->filesize) {
1003 err = got_error(GOT_ERR_PACK_OFFSET);
1006 if (lseek(pack->fd, offset, SEEK_SET) == -1) {
1007 err = got_error_from_errno();
1011 err = parse_object_type_and_size(&type, &size, &tslen, pack->fd);
1016 case GOT_OBJ_TYPE_COMMIT:
1017 case GOT_OBJ_TYPE_TREE:
1018 case GOT_OBJ_TYPE_BLOB:
1019 case GOT_OBJ_TYPE_TAG:
1020 err = open_plain_object(obj, path_packfile, id, type,
1021 offset + tslen, size);
1024 case GOT_OBJ_TYPE_OFFSET_DELTA:
1025 case GOT_OBJ_TYPE_REF_DELTA:
1026 err = open_delta_object(obj, repo, packidx, path_packfile,
1027 pack->fd, pack->filesize, id, offset, tslen, type, size);
1031 err = got_error(GOT_ERR_OBJ_TYPE);
1035 free(path_packfile);
1039 const struct got_error *
1040 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1041 struct got_repository *repo)
1043 const struct got_error *err = NULL;
1044 struct got_packidx_v2_hdr *packidx = NULL;
1047 err = search_packidx(&packidx, &idx, repo, id);
1051 err = open_packed_object(obj, repo, packidx, idx, id);
1055 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1059 static const struct got_error *
1060 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1062 struct got_delta *delta;
1063 uint64_t base_size = 0, result_size = 0;
1066 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1067 /* Plain object types are the delta base. */
1068 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1069 delta->type != GOT_OBJ_TYPE_TREE &&
1070 delta->type != GOT_OBJ_TYPE_BLOB &&
1071 delta->type != GOT_OBJ_TYPE_TAG) {
1072 const struct got_error *err;
1073 err = got_delta_get_sizes(&base_size, &result_size,
1074 delta->delta_buf, delta->delta_len);
1078 base_size = delta->size;
1079 if (base_size > *max_size)
1080 *max_size = base_size;
1081 if (result_size > *max_size)
1082 *max_size = result_size;
1088 static const struct got_error *
1089 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1090 FILE *outfile, struct got_repository *repo)
1092 const struct got_error *err = NULL;
1093 struct got_delta *delta;
1094 FILE *base_file = NULL, *accum_file = NULL;
1095 uint8_t *base_buf = NULL, *accum_buf = NULL;
1096 size_t accum_size = 0;
1102 if (SIMPLEQ_EMPTY(&deltas->entries))
1103 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1105 /* We process small enough files entirely in memory for speed. */
1106 err = get_delta_chain_max_size(&max_size, deltas);
1109 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1110 accum_buf = malloc(max_size);
1111 if (accum_buf == NULL)
1112 return got_error_from_errno();
1114 base_file = got_opentemp();
1115 if (base_file == NULL)
1116 return got_error_from_errno();
1118 accum_file = got_opentemp();
1119 if (accum_file == NULL) {
1120 err = got_error_from_errno();
1126 /* Deltas are ordered in ascending order. */
1127 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1129 struct got_pack *pack;
1131 off_t delta_data_offset;
1133 /* Plain object types are the delta base. */
1134 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1135 delta->type != GOT_OBJ_TYPE_TREE &&
1136 delta->type != GOT_OBJ_TYPE_BLOB &&
1137 delta->type != GOT_OBJ_TYPE_TAG) {
1138 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1142 pack = get_cached_pack(delta->path_packfile, repo);
1144 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1148 delta_data_offset = delta->offset + delta->tslen;
1149 if (delta_data_offset >= pack->filesize) {
1150 err = got_error(GOT_ERR_PACK_OFFSET);
1153 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1154 err = got_error_from_errno();
1158 err = got_inflate_to_file_fd(&base_len,
1159 pack->fd, base_file);
1161 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1163 if (base_len < max_size) {
1165 p = reallocarray(base_buf, 1, max_size);
1167 err = got_error_from_errno();
1182 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1183 delta->delta_len, accum_buf, &accum_size);
1186 err = got_delta_apply(base_file, delta->delta_buf,
1188 /* Final delta application writes to output file. */
1189 ++n < deltas->nentries ? accum_file : outfile,
1195 if (n < deltas->nentries) {
1196 /* Accumulated delta becomes the new base. */
1198 uint8_t *tmp = accum_buf;
1199 accum_buf = base_buf;
1202 FILE *tmp = accum_file;
1203 accum_file = base_file;
1214 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1216 if (len != accum_size)
1217 return got_ferror(outfile, GOT_ERR_IO);
1225 *result_size = accum_size;
1229 static const struct got_error *
1230 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1231 struct got_delta_chain *deltas, struct got_repository *repo)
1233 const struct got_error *err = NULL;
1234 struct got_delta *delta;
1235 uint8_t *base_buf = NULL, *accum_buf = NULL;
1243 if (SIMPLEQ_EMPTY(&deltas->entries))
1244 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1246 err = get_delta_chain_max_size(&max_size, deltas);
1249 accum_buf = malloc(max_size);
1250 if (accum_buf == NULL)
1251 return got_error_from_errno();
1253 /* Deltas are ordered in ascending order. */
1254 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1256 struct got_pack *pack;
1258 size_t delta_data_offset;
1260 /* Plain object types are the delta base. */
1261 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1262 delta->type != GOT_OBJ_TYPE_TREE &&
1263 delta->type != GOT_OBJ_TYPE_BLOB &&
1264 delta->type != GOT_OBJ_TYPE_TAG) {
1265 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1269 pack = get_cached_pack(delta->path_packfile, repo);
1271 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1275 delta_data_offset = delta->offset + delta->tslen;
1276 if (delta_data_offset >= pack->filesize) {
1277 err = got_error(GOT_ERR_PACK_OFFSET);
1280 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1281 err = got_error_from_errno();
1284 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1286 if (base_len < max_size) {
1288 p = reallocarray(base_buf, 1, max_size);
1290 err = got_error_from_errno();
1301 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1302 delta->delta_len, accum_buf, &accum_size);
1307 if (n < deltas->nentries) {
1308 /* Accumulated delta becomes the new base. */
1309 uint8_t *tmp = accum_buf;
1310 accum_buf = base_buf;
1322 *outbuf = accum_buf;
1323 *outlen = accum_size;
1328 const struct got_error *
1329 got_packfile_extract_object(FILE **f, struct got_object *obj,
1330 struct got_repository *repo)
1332 const struct got_error *err = NULL;
1336 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1337 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1339 *f = got_opentemp();
1341 err = got_error_from_errno();
1345 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1346 struct got_pack *pack;
1348 pack = get_cached_pack(obj->path_packfile, repo);
1350 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1355 if (obj->pack_offset >= pack->filesize) {
1356 err = got_error(GOT_ERR_PACK_OFFSET);
1359 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1360 err = got_error_from_errno();
1364 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1366 err = dump_delta_chain_to_file(&obj->size,
1367 &obj->deltas, *f, repo);
1376 const struct got_error *
1377 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1378 struct got_object *obj, struct got_repository *repo)
1380 const struct got_error *err = NULL;
1382 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1383 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1385 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1386 struct got_pack *pack;
1388 pack = get_cached_pack(obj->path_packfile, repo);
1390 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1395 if (obj->pack_offset >= pack->filesize) {
1396 err = got_error(GOT_ERR_PACK_OFFSET);
1399 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1400 err = got_error_from_errno();
1404 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1406 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);