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>
34 #include "got_error.h"
35 #include "got_object.h"
36 #include "got_repository.h"
37 #include "got_opentemp.h"
39 #include "got_lib_sha1.h"
40 #include "got_lib_pack.h"
41 #include "got_lib_path.h"
42 #include "got_lib_delta.h"
43 #include "got_lib_zbuf.h"
44 #include "got_lib_object.h"
45 #include "got_lib_repository.h"
48 #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
51 #define GOT_PACK_PREFIX "pack-"
52 #define GOT_PACKFILE_SUFFIX ".pack"
53 #define GOT_PACKIDX_SUFFIX ".idx"
54 #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
55 SHA1_DIGEST_STRING_LENGTH - 1 + \
56 strlen(GOT_PACKFILE_SUFFIX))
57 #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
58 SHA1_DIGEST_STRING_LENGTH - 1 + \
59 strlen(GOT_PACKIDX_SUFFIX))
62 #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
65 static const struct got_error *
66 verify_fanout_table(uint32_t *fanout_table)
70 for (i = 0; i < 0xff - 1; i++) {
71 if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
72 return got_error(GOT_ERR_BAD_PACKIDX);
78 static const struct got_error *
79 get_packfile_size(size_t *size, const char *path)
86 dot = strrchr(path, '.');
88 return got_error(GOT_ERR_BAD_PATH);
90 /* Path must point to a pack index or to a pack file. */
91 if (strcmp(dot, GOT_PACKIDX_SUFFIX) == 0) {
92 const struct got_error *err = NULL;
94 char base_path[PATH_MAX];
96 /* Convert pack index path to pack file path. */
97 if (strlcpy(base_path, path, PATH_MAX) > PATH_MAX)
98 return got_error(GOT_ERR_NO_SPACE);
99 dot = strrchr(base_path, '.');
101 return got_error(GOT_ERR_BAD_PATH);
103 if (asprintf(&path_pack, "%s.pack", base_path) == -1)
104 return got_error_from_errno();
106 if (stat(path_pack, &sb) != 0)
107 err = got_error_from_errno();
111 } else if (strcmp(dot, GOT_PACKFILE_SUFFIX) == 0) {
112 if (stat(path, &sb) != 0)
113 return got_error_from_errno();
115 return got_error(GOT_ERR_BAD_PATH);
121 const struct got_error *
122 got_packidx_open(struct got_packidx **packidx, const char *path)
124 struct got_packidx *p;
125 struct got_packidx_v2_hdr *h;
126 const struct got_error *err = NULL;
127 size_t nobj, len_fanout, len_ids, offset, remain;
129 uint8_t sha1[SHA1_DIGEST_LENGTH];
135 p = calloc(1, sizeof(*p));
137 return got_error_from_errno();
139 p->fd = open(path, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
141 return got_error_from_errno();
143 err = get_packfile_size(&p->len, path);
149 if (p->len < sizeof(p->hdr)) {
150 err = got_error(GOT_ERR_BAD_PACKIDX);
156 p->path_packidx = strdup(path);
157 if (p->path_packidx == NULL) {
158 err = got_error_from_errno();
162 p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
163 if (p->map == MAP_FAILED) {
164 err = got_error_from_errno();
171 if (remain < sizeof(*h->magic)) {
172 err = got_error(GOT_ERR_BAD_PACKIDX);
175 h->magic = (uint32_t *)(p->map + offset);
176 if (betoh32(*h->magic) != GOT_PACKIDX_V2_MAGIC) {
177 err = got_error(GOT_ERR_BAD_PACKIDX);
180 offset += sizeof(*h->magic);
181 remain -= sizeof(*h->magic);
183 SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
185 if (remain < sizeof(*h->version)) {
186 err = got_error(GOT_ERR_BAD_PACKIDX);
189 h->version = (uint32_t *)(p->map + offset);
190 if (betoh32(*h->version) != GOT_PACKIDX_VERSION) {
191 err = got_error(GOT_ERR_BAD_PACKIDX);
194 offset += sizeof(*h->version);
195 remain -= sizeof(*h->version);
197 SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
200 sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
201 if (remain < len_fanout) {
202 err = got_error(GOT_ERR_BAD_PACKIDX);
205 h->fanout_table = (uint32_t *)(p->map + offset);
206 err = verify_fanout_table(h->fanout_table);
209 SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
210 offset += len_fanout;
211 remain -= len_fanout;
213 nobj = betoh32(h->fanout_table[0xff]);
214 len_ids = nobj * sizeof(*h->sorted_ids);
215 if (len_ids <= nobj || len_ids > remain) {
216 err = got_error(GOT_ERR_BAD_PACKIDX);
220 (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
221 SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
225 if (remain < nobj * sizeof(*h->crc32)) {
226 err = got_error(GOT_ERR_BAD_PACKIDX);
229 h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
230 SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
231 remain -= nobj * sizeof(*h->crc32);
232 offset += nobj * sizeof(*h->crc32);
234 if (remain < nobj * sizeof(*h->offsets)) {
235 err = got_error(GOT_ERR_BAD_PACKIDX);
238 h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
239 SHA1Update(&ctx, (uint8_t *)h->offsets, nobj * sizeof(*h->offsets));
240 remain -= nobj * sizeof(*h->offsets);
241 offset += nobj * sizeof(*h->offsets);
243 /* Large file offsets are contained only in files > 2GB. */
244 if (p->len <= 0x80000000)
247 if (remain < nobj * sizeof(*h->large_offsets)) {
248 err = got_error(GOT_ERR_BAD_PACKIDX);
251 h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
252 SHA1Update(&ctx, (uint8_t*)h->large_offsets,
253 nobj * sizeof(*h->large_offsets));
254 remain -= nobj * sizeof(*h->large_offsets);
255 offset += nobj * sizeof(*h->large_offsets);
258 if (remain < sizeof(*h->trailer)) {
259 err = got_error(GOT_ERR_BAD_PACKIDX);
263 (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
264 SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
265 SHA1Final(sha1, &ctx);
266 if (memcmp(h->trailer->packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
267 err = got_error(GOT_ERR_PACKIDX_CSUM);
270 got_packidx_close(p);
276 const struct got_error *
277 got_packidx_close(struct got_packidx *packidx)
279 const struct got_error *err = NULL;
281 free(packidx->path_packidx);
282 if (packidx->map != NULL && packidx->map != MAP_FAILED) {
283 if (munmap(packidx->map, packidx->len) == -1)
284 err = got_error_from_errno();
293 is_packidx_filename(const char *name, size_t len)
295 if (len != GOT_PACKIDX_NAMELEN)
298 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
301 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
302 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
309 get_object_offset(struct got_packidx *packidx, int idx)
311 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
312 uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
313 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
315 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
316 if (idx < 0 || idx > totobj ||
317 packidx->hdr.large_offsets == NULL)
319 loffset = betoh64(packidx->hdr.large_offsets[idx]);
320 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
322 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
326 get_object_idx(struct got_packidx *packidx, struct got_object_id *id,
327 struct got_repository *repo)
329 u_int8_t id0 = id->sha1[0];
330 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
331 int left = 0, right = totobj - 1;
334 left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
336 while (left <= right) {
337 struct got_packidx_object_id *oid;
340 i = ((left + right) / 2);
341 oid = &packidx->hdr.sorted_ids[i];
342 cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
354 static const struct got_error *
355 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
357 const struct got_error *err = NULL;
360 for (i = 0; i < nitems(repo->packidx_cache); i++) {
361 if (repo->packidx_cache[i] == NULL)
365 if (i == nitems(repo->packidx_cache)) {
366 err = got_packidx_close(repo->packidx_cache[i - 1]);
369 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
370 sizeof(repo->packidx_cache) -
371 sizeof(repo->packidx_cache[0]));
375 repo->packidx_cache[i] = packidx;
379 static const struct got_error *
380 search_packidx(struct got_packidx **packidx, int *idx,
381 struct got_repository *repo, struct got_object_id *id)
383 const struct got_error *err;
390 /* Search pack index cache. */
391 for (i = 0; i < nitems(repo->packidx_cache); i++) {
392 if (repo->packidx_cache[i] == NULL)
394 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
396 *packidx = repo->packidx_cache[i];
400 /* No luck. Search the filesystem. */
402 path_packdir = got_repo_get_path_objects_pack(repo);
403 if (path_packdir == NULL)
404 return got_error_from_errno();
406 packdir = opendir(path_packdir);
407 if (packdir == NULL) {
408 err = got_error_from_errno();
412 while ((dent = readdir(packdir)) != NULL) {
413 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
416 if (asprintf(&path_packidx, "%s/%s", path_packdir,
417 dent->d_name) == -1) {
418 err = got_error_from_errno();
422 err = got_packidx_open(packidx, path_packidx);
427 *idx = get_object_idx(*packidx, id, repo);
429 err = NULL; /* found the object */
430 err = cache_packidx(*packidx, repo);
434 err = got_packidx_close(*packidx);
440 err = got_error(GOT_ERR_NO_OBJ);
443 if (packdir && closedir(packdir) != 0 && err == 0)
444 err = got_error_from_errno();
448 static const struct got_error *
449 get_packfile_path(char **path_packfile, struct got_repository *repo,
450 struct got_packidx *packidx)
454 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
455 size = strlen(packidx->path_packidx) + 2;
456 if (size < GOT_PACKFILE_NAMELEN + 1)
457 return got_error(GOT_ERR_BAD_PATH);
459 *path_packfile = calloc(size, sizeof(**path_packfile));
460 if (*path_packfile == NULL)
461 return got_error_from_errno();
463 /* Copy up to and excluding ".idx". */
464 if (strlcpy(*path_packfile, packidx->path_packidx,
465 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
466 return got_error(GOT_ERR_NO_SPACE);
468 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
469 return got_error(GOT_ERR_NO_SPACE);
474 const struct got_error *
475 read_packfile_hdr(int fd, struct got_packidx *packidx)
477 const struct got_error *err = NULL;
478 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
479 struct got_packfile_hdr hdr;
482 n = read(fd, &hdr, sizeof(hdr));
484 return got_error_from_errno();
485 if (n != sizeof(hdr))
486 return got_error(GOT_ERR_BAD_PACKFILE);
488 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
489 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
490 betoh32(hdr.nobjects) != totobj)
491 err = got_error(GOT_ERR_BAD_PACKFILE);
496 static const struct got_error *
497 open_packfile(int *fd, const char *path_packfile,
498 struct got_repository *repo, struct got_packidx *packidx)
500 const struct got_error *err = NULL;
502 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
504 return got_error_from_errno();
507 err = read_packfile_hdr(*fd, packidx);
517 got_pack_close(struct got_pack *pack)
521 free(pack->path_packfile);
522 pack->path_packfile = NULL;
526 static const struct got_error *
527 cache_pack(struct got_pack **packp, const char *path_packfile,
528 struct got_packidx *packidx, struct got_repository *repo)
530 const struct got_error *err = NULL;
531 struct got_pack *pack = NULL;
537 for (i = 0; i < nitems(repo->packs); i++) {
538 pack = &repo->packs[i];
539 if (pack->path_packfile == NULL)
541 if (strcmp(pack->path_packfile, path_packfile) == 0)
545 if (i == nitems(repo->packs) - 1) {
546 got_pack_close(&repo->packs[i - 1]);
547 memmove(&repo->packs[1], &repo->packs[0],
548 sizeof(repo->packs) - sizeof(repo->packs[0]));
552 pack = &repo->packs[i];
554 pack->path_packfile = strdup(path_packfile);
555 if (pack->path_packfile == NULL) {
556 err = got_error_from_errno();
560 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
564 err = get_packfile_size(&pack->filesize, path_packfile);
568 free(pack->path_packfile);
569 memset(pack, 0, sizeof(*pack));
577 get_cached_pack(const char *path_packfile, struct got_repository *repo)
579 struct got_pack *pack = NULL;
582 for (i = 0; i < nitems(repo->packs); i++) {
583 pack = &repo->packs[i];
584 if (pack->path_packfile == NULL)
586 if (strcmp(pack->path_packfile, path_packfile) == 0)
593 static const struct got_error *
594 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len, int fd)
603 /* We do not support size values which don't fit in 64 bit. */
605 return got_error(GOT_ERR_NO_SPACE);
607 n = read(fd, &sizeN, sizeof(sizeN));
609 return got_error_from_errno();
610 if (n != sizeof(sizeN))
611 return got_error(GOT_ERR_BAD_PACKFILE);
614 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
615 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
616 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
618 size_t shift = 4 + 7 * (i - 1);
619 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
622 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
626 *len = i * sizeof(sizeN);
630 static const struct got_error *
631 open_plain_object(struct got_object **obj, const char *path_packfile,
632 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
634 *obj = calloc(1, sizeof(**obj));
636 return got_error_from_errno();
638 (*obj)->path_packfile = strdup(path_packfile);
639 if ((*obj)->path_packfile == NULL) {
640 const struct got_error *err = got_error_from_errno();
647 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
650 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
651 (*obj)->pack_offset = offset;
656 static const struct got_error *
657 parse_negative_offset(int64_t *offset, size_t *len, int fd)
665 /* We do not support offset values which don't fit in 64 bit. */
667 return got_error(GOT_ERR_NO_SPACE);
669 n = read(fd, &offN, sizeof(offN));
671 return got_error_from_errno();
672 if (n != sizeof(offN))
673 return got_error(GOT_ERR_BAD_PACKFILE);
676 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
680 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
683 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
686 *len = i * sizeof(offN);
690 static const struct got_error *
691 parse_offset_delta(off_t *base_offset, int fd, off_t offset)
693 const struct got_error *err;
697 err = parse_negative_offset(&negoffset, &negofflen, fd);
701 /* Compute the base object's offset (must be in the same pack file). */
702 *base_offset = (offset - negoffset);
703 if (*base_offset <= 0)
704 return got_error(GOT_ERR_BAD_PACKFILE);
709 static const struct got_error *
710 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
711 int, size_t, const char *, off_t, size_t, int, size_t, unsigned int);
713 static const struct got_error *
714 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
715 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
716 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
718 struct got_delta *delta;
720 delta = got_delta_open(path_packfile, delta_offset, tslen,
721 delta_type, delta_size, delta_data_offset, delta_buf,
724 return got_error_from_errno();
725 /* delta is freed in got_object_close() */
727 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
731 static const struct got_error *
732 resolve_offset_delta(struct got_delta_chain *deltas,
733 struct got_repository *repo, int fd, size_t packfile_size,
734 const char *path_packfile, off_t delta_offset, size_t tslen,
735 int delta_type, size_t delta_size, unsigned int recursion)
738 const struct got_error *err;
743 off_t delta_data_offset;
747 err = parse_offset_delta(&base_offset, fd, delta_offset);
751 delta_data_offset = lseek(fd, 0, SEEK_CUR);
752 if (delta_data_offset == -1)
753 return got_error_from_errno();
755 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
759 err = add_delta(deltas, path_packfile, delta_offset, tslen,
760 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
764 /* An offset delta must be in the same packfile. */
765 if (base_offset >= packfile_size)
766 return got_error(GOT_ERR_PACK_OFFSET);
767 if (lseek(fd, base_offset, SEEK_SET) == -1)
768 return got_error_from_errno();
770 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
775 return resolve_delta_chain(deltas, repo, fd, packfile_size,
776 path_packfile, base_offset, base_tslen, base_type, base_size,
780 static const struct got_error *
781 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
782 int fd, const char *path_packfile, off_t delta_offset,
783 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
785 const struct got_error *err;
786 struct got_object_id id;
787 struct got_packidx *packidx;
794 char *path_base_packfile;
795 struct got_pack *base_pack;
796 off_t delta_data_offset;
800 n = read(fd, &id, sizeof(id));
802 return got_error_from_errno();
804 return got_error(GOT_ERR_BAD_PACKFILE);
806 delta_data_offset = lseek(fd, 0, SEEK_CUR);
807 if (delta_data_offset == -1)
808 return got_error_from_errno();
810 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, fd);
814 err = add_delta(deltas, path_packfile, delta_offset, tslen,
815 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
819 err = search_packidx(&packidx, &idx, repo, &id);
823 base_offset = get_object_offset(packidx, idx);
824 if (base_offset == (uint64_t)-1) {
825 return got_error(GOT_ERR_BAD_PACKIDX);
828 err = get_packfile_path(&path_base_packfile, repo, packidx);
832 base_pack = get_cached_pack(path_base_packfile, repo);
833 if (base_pack == NULL) {
834 err = cache_pack(&base_pack, path_base_packfile, NULL, repo);
839 if (base_offset >= base_pack->filesize) {
840 err = got_error(GOT_ERR_PACK_OFFSET);
843 if (lseek(base_pack->fd, base_offset, SEEK_SET) == -1) {
844 err = got_error_from_errno();
848 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
853 err = resolve_delta_chain(deltas, repo, base_pack->fd,
854 base_pack->filesize, path_base_packfile, base_offset,
855 base_tslen, base_type, base_size, recursion - 1);
857 free(path_base_packfile);
861 static const struct got_error *
862 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
863 int fd, size_t packfile_size, const char *path_packfile, off_t delta_offset,
864 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
866 const struct got_error *err = NULL;
868 if (--recursion == 0)
869 return got_error(GOT_ERR_RECURSION);
871 switch (delta_type) {
872 case GOT_OBJ_TYPE_COMMIT:
873 case GOT_OBJ_TYPE_TREE:
874 case GOT_OBJ_TYPE_BLOB:
875 case GOT_OBJ_TYPE_TAG:
876 /* Plain types are the final delta base. Recursion ends. */
877 err = add_delta(deltas, path_packfile, delta_offset, tslen,
878 delta_type, delta_size, 0, NULL, 0);
880 case GOT_OBJ_TYPE_OFFSET_DELTA:
881 err = resolve_offset_delta(deltas, repo, fd, packfile_size,
882 path_packfile, delta_offset, tslen, delta_type, delta_size,
885 case GOT_OBJ_TYPE_REF_DELTA:
886 err = resolve_ref_delta(deltas, repo, fd, path_packfile,
887 delta_offset, tslen, delta_type, delta_size, recursion - 1);
890 return got_error(GOT_ERR_OBJ_TYPE);
896 static const struct got_error *
897 open_delta_object(struct got_object **obj, struct got_repository *repo,
898 const char *path_packfile, int fd, size_t packfile_size,
899 struct got_object_id *id, off_t offset, size_t tslen,
900 int delta_type, size_t delta_size)
902 const struct got_error *err = NULL;
905 *obj = calloc(1, sizeof(**obj));
907 return got_error_from_errno();
911 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
912 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
913 (*obj)->pack_offset = offset + tslen;
915 (*obj)->path_packfile = strdup(path_packfile);
916 if ((*obj)->path_packfile == NULL) {
917 err = got_error_from_errno();
920 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
922 SIMPLEQ_INIT(&(*obj)->deltas.entries);
923 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
925 err = resolve_delta_chain(&(*obj)->deltas, repo, fd, packfile_size,
926 path_packfile, offset, tslen, delta_type, delta_size,
927 GOT_DELTA_CHAIN_RECURSION_MAX);
931 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
934 (*obj)->type = resolved_type;
938 got_object_close(*obj);
944 static const struct got_error *
945 open_packed_object(struct got_object **obj, struct got_repository *repo,
946 struct got_packidx *packidx, int idx, struct got_object_id *id)
948 const struct got_error *err = NULL;
951 struct got_pack *pack;
958 offset = get_object_offset(packidx, idx);
959 if (offset == (uint64_t)-1)
960 return got_error(GOT_ERR_BAD_PACKIDX);
962 err = get_packfile_path(&path_packfile, repo, packidx);
966 pack = get_cached_pack(path_packfile, repo);
968 err = cache_pack(&pack, path_packfile, packidx, repo);
973 if (offset >= pack->filesize) {
974 err = got_error(GOT_ERR_PACK_OFFSET);
977 if (lseek(pack->fd, offset, SEEK_SET) == -1) {
978 err = got_error_from_errno();
982 err = parse_object_type_and_size(&type, &size, &tslen, pack->fd);
987 case GOT_OBJ_TYPE_COMMIT:
988 case GOT_OBJ_TYPE_TREE:
989 case GOT_OBJ_TYPE_BLOB:
990 case GOT_OBJ_TYPE_TAG:
991 err = open_plain_object(obj, path_packfile, id, type,
992 offset + tslen, size);
995 case GOT_OBJ_TYPE_OFFSET_DELTA:
996 case GOT_OBJ_TYPE_REF_DELTA:
997 err = open_delta_object(obj, repo, path_packfile, pack->fd,
998 pack->filesize, id, offset, tslen, type, size);
1002 err = got_error(GOT_ERR_OBJ_TYPE);
1006 free(path_packfile);
1010 const struct got_error *
1011 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1012 struct got_repository *repo)
1014 const struct got_error *err = NULL;
1015 struct got_packidx *packidx = NULL;
1018 err = search_packidx(&packidx, &idx, repo, id);
1022 err = open_packed_object(obj, repo, packidx, idx, id);
1026 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1030 static const struct got_error *
1031 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1033 struct got_delta *delta;
1034 uint64_t base_size = 0, result_size = 0;
1037 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1038 /* Plain object types are the delta base. */
1039 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1040 delta->type != GOT_OBJ_TYPE_TREE &&
1041 delta->type != GOT_OBJ_TYPE_BLOB &&
1042 delta->type != GOT_OBJ_TYPE_TAG) {
1043 const struct got_error *err;
1044 err = got_delta_get_sizes(&base_size, &result_size,
1045 delta->delta_buf, delta->delta_len);
1049 base_size = delta->size;
1050 if (base_size > *max_size)
1051 *max_size = base_size;
1052 if (result_size > *max_size)
1053 *max_size = result_size;
1059 static const struct got_error *
1060 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1061 FILE *outfile, struct got_repository *repo)
1063 const struct got_error *err = NULL;
1064 struct got_delta *delta;
1065 FILE *base_file = NULL, *accum_file = NULL;
1066 uint8_t *base_buf = NULL, *accum_buf = NULL;
1067 size_t accum_size = 0;
1073 if (SIMPLEQ_EMPTY(&deltas->entries))
1074 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1076 /* We process small enough files entirely in memory for speed. */
1077 err = get_delta_chain_max_size(&max_size, deltas);
1080 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1081 accum_buf = malloc(max_size);
1082 if (accum_buf == NULL)
1083 return got_error_from_errno();
1085 base_file = got_opentemp();
1086 if (base_file == NULL)
1087 return got_error_from_errno();
1089 accum_file = got_opentemp();
1090 if (accum_file == NULL) {
1091 err = got_error_from_errno();
1097 /* Deltas are ordered in ascending order. */
1098 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1100 struct got_pack *pack;
1102 off_t delta_data_offset;
1104 /* Plain object types are the delta base. */
1105 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1106 delta->type != GOT_OBJ_TYPE_TREE &&
1107 delta->type != GOT_OBJ_TYPE_BLOB &&
1108 delta->type != GOT_OBJ_TYPE_TAG) {
1109 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1113 pack = get_cached_pack(delta->path_packfile, repo);
1115 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1119 delta_data_offset = delta->offset + delta->tslen;
1120 if (delta_data_offset >= pack->filesize) {
1121 err = got_error(GOT_ERR_PACK_OFFSET);
1124 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1125 err = got_error_from_errno();
1129 err = got_inflate_to_file_fd(&base_len,
1130 pack->fd, base_file);
1132 err = got_inflate_to_mem_fd(&base_buf,
1133 &base_len, pack->fd);
1134 if (base_len < max_size) {
1136 p = reallocarray(base_buf, 1, max_size);
1138 err = got_error_from_errno();
1153 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1154 delta->delta_len, accum_buf, &accum_size);
1157 err = got_delta_apply(base_file, delta->delta_buf,
1159 /* Final delta application writes to output file. */
1160 ++n < deltas->nentries ? accum_file : outfile,
1166 if (n < deltas->nentries) {
1167 /* Accumulated delta becomes the new base. */
1169 uint8_t *tmp = accum_buf;
1170 accum_buf = base_buf;
1173 FILE *tmp = accum_file;
1174 accum_file = base_file;
1185 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1187 if (len != accum_size)
1188 return got_ferror(outfile, GOT_ERR_IO);
1196 *result_size = accum_size;
1200 static const struct got_error *
1201 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1202 struct got_delta_chain *deltas, struct got_repository *repo)
1204 const struct got_error *err = NULL;
1205 struct got_delta *delta;
1206 uint8_t *base_buf = NULL, *accum_buf = NULL;
1214 if (SIMPLEQ_EMPTY(&deltas->entries))
1215 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1217 err = get_delta_chain_max_size(&max_size, deltas);
1220 accum_buf = malloc(max_size);
1221 if (accum_buf == NULL)
1222 return got_error_from_errno();
1224 /* Deltas are ordered in ascending order. */
1225 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1227 struct got_pack *pack;
1229 size_t delta_data_offset;
1231 /* Plain object types are the delta base. */
1232 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1233 delta->type != GOT_OBJ_TYPE_TREE &&
1234 delta->type != GOT_OBJ_TYPE_BLOB &&
1235 delta->type != GOT_OBJ_TYPE_TAG) {
1236 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1240 pack = get_cached_pack(delta->path_packfile, repo);
1242 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1246 delta_data_offset = delta->offset + delta->tslen;
1247 if (delta_data_offset >= pack->filesize) {
1248 err = got_error(GOT_ERR_PACK_OFFSET);
1251 if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1252 err = got_error_from_errno();
1255 err = got_inflate_to_mem_fd(&base_buf, &base_len,
1257 if (base_len < max_size) {
1259 p = reallocarray(base_buf, 1, max_size);
1261 err = got_error_from_errno();
1272 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1273 delta->delta_len, accum_buf, &accum_size);
1278 if (n < deltas->nentries) {
1279 /* Accumulated delta becomes the new base. */
1280 uint8_t *tmp = accum_buf;
1281 accum_buf = base_buf;
1293 *outbuf = accum_buf;
1294 *outlen = accum_size;
1299 const struct got_error *
1300 got_packfile_extract_object(FILE **f, struct got_object *obj,
1301 struct got_repository *repo)
1303 const struct got_error *err = NULL;
1307 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1308 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1310 *f = got_opentemp();
1312 err = got_error_from_errno();
1316 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1317 struct got_pack *pack;
1319 pack = get_cached_pack(obj->path_packfile, repo);
1321 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1326 if (obj->pack_offset >= pack->filesize) {
1327 err = got_error(GOT_ERR_PACK_OFFSET);
1330 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1331 err = got_error_from_errno();
1335 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1337 err = dump_delta_chain_to_file(&obj->size,
1338 &obj->deltas, *f, repo);
1347 const struct got_error *
1348 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1349 struct got_object *obj, struct got_repository *repo)
1351 const struct got_error *err = NULL;
1353 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1354 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1356 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1357 struct got_pack *pack;
1359 pack = get_cached_pack(obj->path_packfile, repo);
1361 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1366 if (obj->pack_offset >= pack->filesize) {
1367 err = got_error(GOT_ERR_PACK_OFFSET);
1370 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1371 err = got_error_from_errno();
1375 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1377 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);