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_inflate.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, int verify)
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;
130 uint8_t sha1[SHA1_DIGEST_LENGTH];
136 p = calloc(1, sizeof(*p));
138 return got_error_from_errno();
140 p->fd = open(path, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
142 return got_error_from_errno();
144 err = get_packfile_size(&p->len, path);
150 if (p->len < sizeof(p->hdr)) {
151 err = got_error(GOT_ERR_BAD_PACKIDX);
157 p->path_packidx = strdup(path);
158 if (p->path_packidx == NULL) {
159 err = got_error_from_errno();
163 #ifndef GOT_PACK_NO_MMAP
164 p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
165 if (p->map == MAP_FAILED)
166 p->map = NULL; /* fall back to read(2) */
173 if (remain < sizeof(*h->magic)) {
174 err = got_error(GOT_ERR_BAD_PACKIDX);
178 h->magic = (uint32_t *)(p->map + offset);
180 h->magic = malloc(sizeof(*h->magic));
181 if (h->magic == NULL) {
182 err = got_error_from_errno();
185 n = read(p->fd, h->magic, sizeof(*h->magic));
186 if (n != sizeof(*h->magic)) {
187 err = got_error(GOT_ERR_BAD_PACKIDX);
191 if (betoh32(*h->magic) != GOT_PACKIDX_V2_MAGIC) {
192 err = got_error(GOT_ERR_BAD_PACKIDX);
195 offset += sizeof(*h->magic);
196 remain -= sizeof(*h->magic);
199 SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
201 if (remain < sizeof(*h->version)) {
202 err = got_error(GOT_ERR_BAD_PACKIDX);
206 h->version = (uint32_t *)(p->map + offset);
208 h->version = malloc(sizeof(*h->version));
209 if (h->version == NULL) {
210 err = got_error_from_errno();
213 n = read(p->fd, h->version, sizeof(*h->version));
214 if (n != sizeof(*h->version)) {
215 err = got_error(GOT_ERR_BAD_PACKIDX);
219 if (betoh32(*h->version) != GOT_PACKIDX_VERSION) {
220 err = got_error(GOT_ERR_BAD_PACKIDX);
223 offset += sizeof(*h->version);
224 remain -= sizeof(*h->version);
227 SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
230 sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
231 if (remain < len_fanout) {
232 err = got_error(GOT_ERR_BAD_PACKIDX);
236 h->fanout_table = (uint32_t *)(p->map + offset);
238 h->fanout_table = malloc(len_fanout);
239 if (h->fanout_table == NULL) {
240 err = got_error_from_errno();
243 n = read(p->fd, h->fanout_table, len_fanout);
244 if (n != len_fanout) {
245 err = got_error(GOT_ERR_BAD_PACKIDX);
249 err = verify_fanout_table(h->fanout_table);
253 SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
254 offset += len_fanout;
255 remain -= len_fanout;
257 nobj = betoh32(h->fanout_table[0xff]);
258 len_ids = nobj * sizeof(*h->sorted_ids);
259 if (len_ids <= nobj || len_ids > remain) {
260 err = got_error(GOT_ERR_BAD_PACKIDX);
265 (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
267 h->sorted_ids = malloc(len_ids);
268 if (h->sorted_ids == NULL) {
269 err = got_error(GOT_ERR_BAD_PACKIDX);
272 n = read(p->fd, h->sorted_ids, len_ids);
274 err = got_error(GOT_ERR_BAD_PACKIDX);
279 SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
283 if (remain < nobj * sizeof(*h->crc32)) {
284 err = got_error(GOT_ERR_BAD_PACKIDX);
288 h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
290 h->crc32 = malloc(nobj * sizeof(*h->crc32));
291 if (h->crc32 == NULL) {
292 err = got_error_from_errno();
295 n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
296 if (n != nobj * sizeof(*h->crc32)) {
297 err = got_error(GOT_ERR_BAD_PACKIDX);
302 SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
303 remain -= nobj * sizeof(*h->crc32);
304 offset += nobj * sizeof(*h->crc32);
306 if (remain < nobj * sizeof(*h->offsets)) {
307 err = got_error(GOT_ERR_BAD_PACKIDX);
311 h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
313 h->offsets = malloc(nobj * sizeof(*h->offsets));
314 if (h->offsets == NULL) {
315 err = got_error_from_errno();
318 n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
319 if (n != nobj * sizeof(*h->offsets)) {
320 err = got_error(GOT_ERR_BAD_PACKIDX);
325 SHA1Update(&ctx, (uint8_t *)h->offsets,
326 nobj * sizeof(*h->offsets));
327 remain -= nobj * sizeof(*h->offsets);
328 offset += nobj * sizeof(*h->offsets);
330 /* Large file offsets are contained only in files > 2GB. */
331 if (p->len <= 0x80000000)
334 if (remain < nobj * sizeof(*h->large_offsets)) {
335 err = got_error(GOT_ERR_BAD_PACKIDX);
339 h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
341 h->offsets = malloc(nobj * sizeof(*h->large_offsets));
342 if (h->offsets == NULL) {
343 err = got_error_from_errno();
346 n = read(p->fd, h->large_offsets,
347 nobj * sizeof(*h->large_offsets));
348 if (n != nobj * sizeof(*h->large_offsets)) {
349 err = got_error(GOT_ERR_BAD_PACKIDX);
354 SHA1Update(&ctx, (uint8_t*)h->large_offsets,
355 nobj * sizeof(*h->large_offsets));
356 remain -= nobj * sizeof(*h->large_offsets);
357 offset += nobj * sizeof(*h->large_offsets);
360 if (remain < sizeof(*h->trailer)) {
361 err = got_error(GOT_ERR_BAD_PACKIDX);
366 (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
368 h->trailer = malloc(sizeof(*h->trailer));
369 if (h->trailer == NULL) {
370 err = got_error_from_errno();
373 n = read(p->fd, h->trailer, sizeof(*h->trailer));
374 if (n != sizeof(*h->trailer)) {
375 err = got_error(GOT_ERR_BAD_PACKIDX);
380 SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
381 SHA1Final(sha1, &ctx);
382 if (memcmp(h->trailer->packidx_sha1, sha1,
383 SHA1_DIGEST_LENGTH) != 0)
384 err = got_error(GOT_ERR_PACKIDX_CSUM);
388 got_packidx_close(p);
394 const struct got_error *
395 got_packidx_close(struct got_packidx *packidx)
397 const struct got_error *err = NULL;
399 free(packidx->path_packidx);
401 if (munmap(packidx->map, packidx->len) == -1)
402 err = got_error_from_errno();
404 free(packidx->hdr.magic);
405 free(packidx->hdr.version);
406 free(packidx->hdr.fanout_table);
407 free(packidx->hdr.sorted_ids);
408 free(packidx->hdr.crc32);
409 free(packidx->hdr.offsets);
410 free(packidx->hdr.large_offsets);
411 free(packidx->hdr.trailer);
420 is_packidx_filename(const char *name, size_t len)
422 if (len != GOT_PACKIDX_NAMELEN)
425 if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
428 if (strcmp(name + strlen(GOT_PACK_PREFIX) +
429 SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
436 get_object_offset(struct got_packidx *packidx, int idx)
438 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
439 uint32_t offset = betoh32(packidx->hdr.offsets[idx]);
440 if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
442 idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
443 if (idx < 0 || idx > totobj ||
444 packidx->hdr.large_offsets == NULL)
446 loffset = betoh64(packidx->hdr.large_offsets[idx]);
447 return (loffset > INT64_MAX ? -1 : (off_t)loffset);
449 return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
453 get_object_idx(struct got_packidx *packidx, struct got_object_id *id,
454 struct got_repository *repo)
456 u_int8_t id0 = id->sha1[0];
457 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
458 int left = 0, right = totobj - 1;
461 left = betoh32(packidx->hdr.fanout_table[id0 - 1]);
463 while (left <= right) {
464 struct got_packidx_object_id *oid;
467 i = ((left + right) / 2);
468 oid = &packidx->hdr.sorted_ids[i];
469 cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
481 static const struct got_error *
482 cache_packidx(struct got_packidx *packidx, struct got_repository *repo)
484 const struct got_error *err = NULL;
487 for (i = 0; i < nitems(repo->packidx_cache); i++) {
488 if (repo->packidx_cache[i] == NULL)
492 if (i == nitems(repo->packidx_cache)) {
493 err = got_packidx_close(repo->packidx_cache[i - 1]);
496 memmove(&repo->packidx_cache[1], &repo->packidx_cache[0],
497 sizeof(repo->packidx_cache) -
498 sizeof(repo->packidx_cache[0]));
502 repo->packidx_cache[i] = packidx;
506 static const struct got_error *
507 search_packidx(struct got_packidx **packidx, int *idx,
508 struct got_repository *repo, struct got_object_id *id)
510 const struct got_error *err;
517 /* Search pack index cache. */
518 for (i = 0; i < nitems(repo->packidx_cache); i++) {
519 if (repo->packidx_cache[i] == NULL)
521 *idx = get_object_idx(repo->packidx_cache[i], id, repo);
523 *packidx = repo->packidx_cache[i];
527 /* No luck. Search the filesystem. */
529 path_packdir = got_repo_get_path_objects_pack(repo);
530 if (path_packdir == NULL)
531 return got_error_from_errno();
533 packdir = opendir(path_packdir);
534 if (packdir == NULL) {
535 err = got_error_from_errno();
539 while ((dent = readdir(packdir)) != NULL) {
540 if (!is_packidx_filename(dent->d_name, dent->d_namlen))
543 if (asprintf(&path_packidx, "%s/%s", path_packdir,
544 dent->d_name) == -1) {
545 err = got_error_from_errno();
549 err = got_packidx_open(packidx, path_packidx, 0);
554 *idx = get_object_idx(*packidx, id, repo);
556 err = NULL; /* found the object */
557 err = cache_packidx(*packidx, repo);
561 err = got_packidx_close(*packidx);
567 err = got_error(GOT_ERR_NO_OBJ);
570 if (packdir && closedir(packdir) != 0 && err == 0)
571 err = got_error_from_errno();
575 static const struct got_error *
576 get_packfile_path(char **path_packfile, struct got_repository *repo,
577 struct got_packidx *packidx)
581 /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
582 size = strlen(packidx->path_packidx) + 2;
583 if (size < GOT_PACKFILE_NAMELEN + 1)
584 return got_error(GOT_ERR_BAD_PATH);
586 *path_packfile = calloc(size, sizeof(**path_packfile));
587 if (*path_packfile == NULL)
588 return got_error_from_errno();
590 /* Copy up to and excluding ".idx". */
591 if (strlcpy(*path_packfile, packidx->path_packidx,
592 size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
593 return got_error(GOT_ERR_NO_SPACE);
595 if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
596 return got_error(GOT_ERR_NO_SPACE);
601 static const struct got_error *
602 read_packfile_hdr(int fd, struct got_packidx *packidx)
604 const struct got_error *err = NULL;
605 uint32_t totobj = betoh32(packidx->hdr.fanout_table[0xff]);
606 struct got_packfile_hdr hdr;
609 n = read(fd, &hdr, sizeof(hdr));
611 return got_error_from_errno();
612 if (n != sizeof(hdr))
613 return got_error(GOT_ERR_BAD_PACKFILE);
615 if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
616 betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
617 betoh32(hdr.nobjects) != totobj)
618 err = got_error(GOT_ERR_BAD_PACKFILE);
623 static const struct got_error *
624 open_packfile(int *fd, const char *path_packfile,
625 struct got_repository *repo, struct got_packidx *packidx)
627 const struct got_error *err = NULL;
629 *fd = open(path_packfile, O_RDONLY | O_NOFOLLOW, GOT_DEFAULT_FILE_MODE);
631 return got_error_from_errno();
634 err = read_packfile_hdr(*fd, packidx);
643 const struct got_error *
644 got_pack_close(struct got_pack *pack)
646 const struct got_error *err = NULL;
648 if (pack->map && munmap(pack->map, pack->filesize) == -1)
649 err = got_error_from_errno();
652 free(pack->path_packfile);
653 pack->path_packfile = NULL;
659 static const struct got_error *
660 cache_pack(struct got_pack **packp, const char *path_packfile,
661 struct got_packidx *packidx, struct got_repository *repo)
663 const struct got_error *err = NULL;
664 struct got_pack *pack = NULL;
670 for (i = 0; i < nitems(repo->packs); i++) {
671 pack = &repo->packs[i];
672 if (pack->path_packfile == NULL)
674 if (strcmp(pack->path_packfile, path_packfile) == 0)
678 if (i == nitems(repo->packs) - 1) {
679 err = got_pack_close(&repo->packs[i - 1]);
682 memmove(&repo->packs[1], &repo->packs[0],
683 sizeof(repo->packs) - sizeof(repo->packs[0]));
687 pack = &repo->packs[i];
689 pack->path_packfile = strdup(path_packfile);
690 if (pack->path_packfile == NULL) {
691 err = got_error_from_errno();
695 err = open_packfile(&pack->fd, path_packfile, repo, packidx);
699 err = get_packfile_size(&pack->filesize, path_packfile);
703 #ifndef GOT_PACK_NO_MMAP
704 pack->map = mmap(NULL, pack->filesize, PROT_READ, MAP_PRIVATE,
706 if (pack->map == MAP_FAILED)
707 pack->map = NULL; /* fall back to read(2) */
712 free(pack->path_packfile);
713 memset(pack, 0, sizeof(*pack));
721 get_cached_pack(const char *path_packfile, struct got_repository *repo)
723 struct got_pack *pack = NULL;
726 for (i = 0; i < nitems(repo->packs); i++) {
727 pack = &repo->packs[i];
728 if (pack->path_packfile == NULL)
730 if (strcmp(pack->path_packfile, path_packfile) == 0)
737 static const struct got_error *
738 parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
739 struct got_pack *pack, off_t offset)
749 if (offset >= pack->filesize)
750 return got_error(GOT_ERR_PACK_OFFSET);
753 mapoff = (size_t)offset;
755 if (lseek(pack->fd, offset, SEEK_SET) == -1)
756 return got_error_from_errno();
760 /* We do not support size values which don't fit in 64 bit. */
762 return got_error(GOT_ERR_NO_SPACE);
765 sizeN = *(pack->map + mapoff);
766 mapoff += sizeof(sizeN);
768 ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
770 return got_error_from_errno();
771 if (n != sizeof(sizeN))
772 return got_error(GOT_ERR_BAD_PACKFILE);
774 *len += sizeof(sizeN);
777 t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
778 GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
779 s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
781 size_t shift = 4 + 7 * (i - 1);
782 s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
785 } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
792 static const struct got_error *
793 open_plain_object(struct got_object **obj, const char *path_packfile,
794 struct got_object_id *id, uint8_t type, off_t offset, size_t size)
796 *obj = calloc(1, sizeof(**obj));
798 return got_error_from_errno();
800 (*obj)->path_packfile = strdup(path_packfile);
801 if ((*obj)->path_packfile == NULL) {
802 const struct got_error *err = got_error_from_errno();
809 (*obj)->flags = GOT_OBJ_FLAG_PACKED;
812 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
813 (*obj)->pack_offset = offset;
818 static const struct got_error *
819 parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
829 /* We do not support offset values which don't fit in 64 bit. */
831 return got_error(GOT_ERR_NO_SPACE);
835 if (delta_offset >= pack->filesize)
836 return got_error(GOT_ERR_PACK_OFFSET);
837 mapoff = (size_t)delta_offset + *len;
838 offN = *(pack->map + mapoff);
841 n = read(pack->fd, &offN, sizeof(offN));
843 return got_error_from_errno();
844 if (n != sizeof(offN))
845 return got_error(GOT_ERR_BAD_PACKFILE);
847 *len += sizeof(offN);
850 o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
854 o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
857 } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
863 static const struct got_error *
864 parse_offset_delta(off_t *base_offset, size_t *len, struct got_pack *pack,
865 off_t offset, int tslen)
867 const struct got_error *err;
873 err = parse_negative_offset(&negoffset, &negofflen, pack,
878 /* Compute the base object's offset (must be in the same pack file). */
879 *base_offset = (offset - negoffset);
880 if (*base_offset <= 0)
881 return got_error(GOT_ERR_BAD_PACKFILE);
887 static const struct got_error *
888 resolve_delta_chain(struct got_delta_chain *, struct got_repository *,
889 struct got_packidx *, struct got_pack *, off_t, size_t, int,
890 size_t, unsigned int);
892 static const struct got_error *
893 add_delta(struct got_delta_chain *deltas, const char *path_packfile,
894 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
895 size_t delta_data_offset, uint8_t *delta_buf, size_t delta_len)
897 struct got_delta *delta;
899 delta = got_delta_open(path_packfile, delta_offset, tslen,
900 delta_type, delta_size, delta_data_offset, delta_buf,
903 return got_error_from_errno();
904 /* delta is freed in got_object_close() */
906 SIMPLEQ_INSERT_HEAD(&deltas->entries, delta, entry);
910 static const struct got_error *
911 resolve_offset_delta(struct got_delta_chain *deltas,
912 struct got_repository *repo, struct got_packidx *packidx,
913 struct got_pack *pack, off_t delta_offset, size_t tslen,
914 int delta_type, size_t delta_size, unsigned int recursion)
917 const struct got_error *err;
922 off_t delta_data_offset;
924 size_t delta_len, consumed;
926 err = parse_offset_delta(&base_offset, &consumed, pack,
927 delta_offset, tslen);
931 delta_data_offset = delta_offset + tslen + consumed;
932 if (delta_data_offset >= pack->filesize)
933 return got_error(GOT_ERR_PACK_OFFSET);
935 if (pack->map == NULL) {
936 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
937 if (delta_data_offset == -1)
938 return got_error_from_errno();
942 size_t mapoff = (size_t)delta_data_offset;
943 err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
944 mapoff, pack->filesize - mapoff);
949 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
954 err = add_delta(deltas, pack->path_packfile, delta_offset, tslen,
955 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
959 /* An offset delta must be in the same packfile. */
960 if (base_offset >= pack->filesize)
961 return got_error(GOT_ERR_PACK_OFFSET);
963 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
968 return resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
969 base_tslen, base_type, base_size, recursion - 1);
972 static const struct got_error *
973 resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
974 struct got_packidx *packidx, struct got_pack *pack,
975 off_t delta_offset, size_t tslen, int delta_type, size_t delta_size,
976 unsigned int recursion)
978 const struct got_error *err;
979 struct got_object_id id;
985 off_t delta_data_offset;
986 uint8_t *delta_buf = NULL;
989 if (delta_offset >= pack->filesize)
990 return got_error(GOT_ERR_PACK_OFFSET);
991 delta_data_offset = delta_offset + tslen;
992 if (delta_data_offset >= pack->filesize)
993 return got_error(GOT_ERR_PACK_OFFSET);
995 if (pack->map == NULL) {
996 delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
997 if (delta_data_offset == -1)
998 return got_error_from_errno();
1003 size_t mapoff = (size_t)delta_data_offset;
1004 memcpy(&id, pack->map + mapoff, sizeof(id));
1005 mapoff += sizeof(id);
1006 err = got_inflate_to_mem_mmap(&delta_buf, &delta_len, pack->map,
1007 mapoff, pack->filesize - mapoff);
1011 ssize_t n = read(pack->fd, &id, sizeof(id));
1013 return got_error_from_errno();
1014 if (n != sizeof(id))
1015 return got_error(GOT_ERR_BAD_PACKFILE);
1016 err = got_inflate_to_mem_fd(&delta_buf, &delta_len, pack->fd);
1021 err = add_delta(deltas, pack->path_packfile, delta_offset, tslen,
1022 delta_type, delta_size, delta_data_offset, delta_buf, delta_len);
1026 /* Delta base must be in the same pack file. */
1027 idx = get_object_idx(packidx, &id, repo);
1029 err = got_error(GOT_ERR_BAD_PACKFILE);
1033 base_offset = get_object_offset(packidx, idx);
1034 if (base_offset == (uint64_t)-1) {
1035 err = got_error(GOT_ERR_BAD_PACKIDX);
1039 if (base_offset >= pack->filesize) {
1040 err = got_error(GOT_ERR_PACK_OFFSET);
1044 err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
1049 err = resolve_delta_chain(deltas, repo, packidx, pack, base_offset,
1050 base_tslen, base_type, base_size, recursion - 1);
1057 static const struct got_error *
1058 resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
1059 struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1060 size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1062 const struct got_error *err = NULL;
1064 if (--recursion == 0)
1065 return got_error(GOT_ERR_RECURSION);
1067 switch (delta_type) {
1068 case GOT_OBJ_TYPE_COMMIT:
1069 case GOT_OBJ_TYPE_TREE:
1070 case GOT_OBJ_TYPE_BLOB:
1071 case GOT_OBJ_TYPE_TAG:
1072 /* Plain types are the final delta base. Recursion ends. */
1073 err = add_delta(deltas, pack->path_packfile, delta_offset,
1074 tslen, delta_type, delta_size, 0, NULL, 0);
1076 case GOT_OBJ_TYPE_OFFSET_DELTA:
1077 err = resolve_offset_delta(deltas, repo, packidx, pack,
1078 delta_offset, tslen, delta_type, delta_size, recursion - 1);
1080 case GOT_OBJ_TYPE_REF_DELTA:
1081 err = resolve_ref_delta(deltas, repo, packidx, pack,
1082 delta_offset, tslen, delta_type, delta_size, recursion - 1);
1085 return got_error(GOT_ERR_OBJ_TYPE);
1091 static const struct got_error *
1092 open_delta_object(struct got_object **obj, struct got_repository *repo,
1093 struct got_packidx *packidx, struct got_pack *pack,
1094 struct got_object_id *id, off_t offset, size_t tslen,
1095 int delta_type, size_t delta_size)
1097 const struct got_error *err = NULL;
1100 *obj = calloc(1, sizeof(**obj));
1102 return got_error_from_errno();
1106 (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1107 memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1108 (*obj)->pack_offset = offset + tslen;
1110 (*obj)->path_packfile = strdup(pack->path_packfile);
1111 if ((*obj)->path_packfile == NULL) {
1112 err = got_error_from_errno();
1115 (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1117 SIMPLEQ_INIT(&(*obj)->deltas.entries);
1118 (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1120 err = resolve_delta_chain(&(*obj)->deltas, repo, packidx, pack, offset,
1121 tslen, delta_type, delta_size, GOT_DELTA_CHAIN_RECURSION_MAX);
1125 err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1128 (*obj)->type = resolved_type;
1132 got_object_close(*obj);
1138 static const struct got_error *
1139 open_packed_object(struct got_object **obj, struct got_repository *repo,
1140 struct got_packidx *packidx, int idx, struct got_object_id *id)
1142 const struct got_error *err = NULL;
1144 char *path_packfile;
1145 struct got_pack *pack;
1152 offset = get_object_offset(packidx, idx);
1153 if (offset == (uint64_t)-1)
1154 return got_error(GOT_ERR_BAD_PACKIDX);
1156 err = get_packfile_path(&path_packfile, repo, packidx);
1160 pack = get_cached_pack(path_packfile, repo);
1162 err = cache_pack(&pack, path_packfile, packidx, repo);
1167 err = parse_object_type_and_size(&type, &size, &tslen, pack, offset);
1172 case GOT_OBJ_TYPE_COMMIT:
1173 case GOT_OBJ_TYPE_TREE:
1174 case GOT_OBJ_TYPE_BLOB:
1175 case GOT_OBJ_TYPE_TAG:
1176 err = open_plain_object(obj, path_packfile, id, type,
1177 offset + tslen, size);
1180 case GOT_OBJ_TYPE_OFFSET_DELTA:
1181 case GOT_OBJ_TYPE_REF_DELTA:
1182 err = open_delta_object(obj, repo, packidx, pack, id, offset,
1187 err = got_error(GOT_ERR_OBJ_TYPE);
1191 free(path_packfile);
1195 const struct got_error *
1196 got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
1197 struct got_repository *repo)
1199 const struct got_error *err = NULL;
1200 struct got_packidx *packidx = NULL;
1203 err = search_packidx(&packidx, &idx, repo, id);
1207 err = open_packed_object(obj, repo, packidx, idx, id);
1211 err = cache_pack(NULL, (*obj)->path_packfile, packidx, repo);
1215 static const struct got_error *
1216 get_delta_chain_max_size(uint64_t *max_size, struct got_delta_chain *deltas)
1218 struct got_delta *delta;
1219 uint64_t base_size = 0, result_size = 0;
1222 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1223 /* Plain object types are the delta base. */
1224 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1225 delta->type != GOT_OBJ_TYPE_TREE &&
1226 delta->type != GOT_OBJ_TYPE_BLOB &&
1227 delta->type != GOT_OBJ_TYPE_TAG) {
1228 const struct got_error *err;
1229 err = got_delta_get_sizes(&base_size, &result_size,
1230 delta->delta_buf, delta->delta_len);
1234 base_size = delta->size;
1235 if (base_size > *max_size)
1236 *max_size = base_size;
1237 if (result_size > *max_size)
1238 *max_size = result_size;
1244 static const struct got_error *
1245 dump_delta_chain_to_file(size_t *result_size, struct got_delta_chain *deltas,
1246 FILE *outfile, struct got_repository *repo)
1248 const struct got_error *err = NULL;
1249 struct got_delta *delta;
1250 FILE *base_file = NULL, *accum_file = NULL;
1251 uint8_t *base_buf = NULL, *accum_buf = NULL;
1252 size_t accum_size = 0;
1258 if (SIMPLEQ_EMPTY(&deltas->entries))
1259 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1261 /* We process small enough files entirely in memory for speed. */
1262 err = get_delta_chain_max_size(&max_size, deltas);
1265 if (max_size < GOT_DELTA_RESULT_SIZE_CACHED_MAX) {
1266 accum_buf = malloc(max_size);
1267 if (accum_buf == NULL)
1268 return got_error_from_errno();
1270 base_file = got_opentemp();
1271 if (base_file == NULL)
1272 return got_error_from_errno();
1274 accum_file = got_opentemp();
1275 if (accum_file == NULL) {
1276 err = got_error_from_errno();
1282 /* Deltas are ordered in ascending order. */
1283 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1285 struct got_pack *pack;
1286 size_t base_len, mapoff;
1287 off_t delta_data_offset;
1289 /* Plain object types are the delta base. */
1290 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1291 delta->type != GOT_OBJ_TYPE_TREE &&
1292 delta->type != GOT_OBJ_TYPE_BLOB &&
1293 delta->type != GOT_OBJ_TYPE_TAG) {
1294 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1298 pack = get_cached_pack(delta->path_packfile, repo);
1300 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1304 delta_data_offset = delta->offset + delta->tslen;
1305 if (delta_data_offset >= pack->filesize) {
1306 err = got_error(GOT_ERR_PACK_OFFSET);
1309 if (pack->map == NULL) {
1310 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1312 err = got_error_from_errno();
1318 mapoff = (size_t)delta_data_offset;
1319 err = got_inflate_to_file_mmap(
1320 &base_len, pack->map, mapoff,
1321 pack->filesize - mapoff, base_file);
1323 err = got_inflate_to_file_fd(&base_len,
1324 pack->fd, base_file);
1327 mapoff = (size_t)delta_data_offset;
1328 err = got_inflate_to_mem_mmap(&base_buf,
1329 &base_len, pack->map, mapoff,
1330 pack->filesize - mapoff);
1332 err = got_inflate_to_mem_fd(&base_buf,
1333 &base_len, pack->fd);
1334 if (base_len < max_size) {
1336 p = reallocarray(base_buf, 1, max_size);
1338 err = got_error_from_errno();
1353 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1354 delta->delta_len, accum_buf, &accum_size);
1357 err = got_delta_apply(base_file, delta->delta_buf,
1359 /* Final delta application writes to output file. */
1360 ++n < deltas->nentries ? accum_file : outfile,
1366 if (n < deltas->nentries) {
1367 /* Accumulated delta becomes the new base. */
1369 uint8_t *tmp = accum_buf;
1370 accum_buf = base_buf;
1373 FILE *tmp = accum_file;
1374 accum_file = base_file;
1385 size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1387 if (len != accum_size)
1388 return got_ferror(outfile, GOT_ERR_IO);
1396 *result_size = accum_size;
1400 static const struct got_error *
1401 dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1402 struct got_delta_chain *deltas, struct got_repository *repo)
1404 const struct got_error *err = NULL;
1405 struct got_delta *delta;
1406 uint8_t *base_buf = NULL, *accum_buf = NULL;
1414 if (SIMPLEQ_EMPTY(&deltas->entries))
1415 return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1417 err = get_delta_chain_max_size(&max_size, deltas);
1420 accum_buf = malloc(max_size);
1421 if (accum_buf == NULL)
1422 return got_error_from_errno();
1424 /* Deltas are ordered in ascending order. */
1425 SIMPLEQ_FOREACH(delta, &deltas->entries, entry) {
1427 struct got_pack *pack;
1429 size_t delta_data_offset;
1431 /* Plain object types are the delta base. */
1432 if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1433 delta->type != GOT_OBJ_TYPE_TREE &&
1434 delta->type != GOT_OBJ_TYPE_BLOB &&
1435 delta->type != GOT_OBJ_TYPE_TAG) {
1436 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1440 pack = get_cached_pack(delta->path_packfile, repo);
1442 err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1446 delta_data_offset = delta->offset + delta->tslen;
1447 if (delta_data_offset >= pack->filesize) {
1448 err = got_error(GOT_ERR_PACK_OFFSET);
1452 size_t mapoff = (size_t)delta_data_offset;
1453 err = got_inflate_to_mem_mmap(&base_buf,
1454 &base_len, pack->map, mapoff,
1455 pack->filesize - mapoff);
1457 if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1459 err = got_error_from_errno();
1462 err = got_inflate_to_mem_fd(&base_buf,
1463 &base_len, pack->fd);
1467 if (base_len < max_size) {
1469 p = reallocarray(base_buf, 1, max_size);
1471 err = got_error_from_errno();
1480 err = got_delta_apply_in_mem(base_buf, delta->delta_buf,
1481 delta->delta_len, accum_buf, &accum_size);
1486 if (n < deltas->nentries) {
1487 /* Accumulated delta becomes the new base. */
1488 uint8_t *tmp = accum_buf;
1489 accum_buf = base_buf;
1501 *outbuf = accum_buf;
1502 *outlen = accum_size;
1507 const struct got_error *
1508 got_packfile_extract_object(FILE **f, struct got_object *obj,
1509 struct got_repository *repo)
1511 const struct got_error *err = NULL;
1515 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1516 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1518 *f = got_opentemp();
1520 err = got_error_from_errno();
1524 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1525 struct got_pack *pack;
1527 pack = get_cached_pack(obj->path_packfile, repo);
1529 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1534 if (obj->pack_offset >= pack->filesize) {
1535 err = got_error(GOT_ERR_PACK_OFFSET);
1540 size_t mapoff = (size_t)obj->pack_offset;
1541 err = got_inflate_to_file_mmap(&obj->size, pack->map,
1542 mapoff, pack->filesize - mapoff, *f);
1544 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1545 err = got_error_from_errno();
1548 err = got_inflate_to_file_fd(&obj->size, pack->fd, *f);
1551 err = dump_delta_chain_to_file(&obj->size,
1552 &obj->deltas, *f, repo);
1561 const struct got_error *
1562 got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1563 struct got_object *obj, struct got_repository *repo)
1565 const struct got_error *err = NULL;
1567 if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1568 return got_error(GOT_ERR_OBJ_NOT_PACKED);
1570 if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1571 struct got_pack *pack;
1573 pack = get_cached_pack(obj->path_packfile, repo);
1575 err = cache_pack(&pack, obj->path_packfile, NULL, repo);
1580 if (obj->pack_offset >= pack->filesize) {
1581 err = got_error(GOT_ERR_PACK_OFFSET);
1585 size_t mapoff = (size_t)obj->pack_offset;
1586 err = got_inflate_to_mem_mmap(buf, len, pack->map,
1587 mapoff, pack->filesize - mapoff);
1589 if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1) {
1590 err = got_error_from_errno();
1593 err = got_inflate_to_mem_fd(buf, len, pack->fd);
1596 err = dump_delta_chain_to_mem(buf, len, &obj->deltas, repo);