2 0a0a3048 2018-01-10 stsp * Copyright (c) 2018 Stefan Sperling <stsp@openbsd.org>
4 0a0a3048 2018-01-10 stsp * Permission to use, copy, modify, and distribute this software for any
5 0a0a3048 2018-01-10 stsp * purpose with or without fee is hereby granted, provided that the above
6 0a0a3048 2018-01-10 stsp * copyright notice and this permission notice appear in all copies.
8 0a0a3048 2018-01-10 stsp * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
9 0a0a3048 2018-01-10 stsp * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
10 0a0a3048 2018-01-10 stsp * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
11 0a0a3048 2018-01-10 stsp * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
12 0a0a3048 2018-01-10 stsp * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
13 0a0a3048 2018-01-10 stsp * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
14 0a0a3048 2018-01-10 stsp * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17 a1fd68d8 2018-01-12 stsp #include <sys/types.h>
18 0a0a3048 2018-01-10 stsp #include <sys/stat.h>
19 a1fd68d8 2018-01-12 stsp #include <sys/queue.h>
21 a1fd68d8 2018-01-12 stsp #include <dirent.h>
22 a1fd68d8 2018-01-12 stsp #include <errno.h>
23 0a0a3048 2018-01-10 stsp #include <stdio.h>
24 a1fd68d8 2018-01-12 stsp #include <stdint.h>
25 0a0a3048 2018-01-10 stsp #include <stdlib.h>
26 0a0a3048 2018-01-10 stsp #include <string.h>
27 0a0a3048 2018-01-10 stsp #include <limits.h>
28 0a0a3048 2018-01-10 stsp #include <sha1.h>
29 0a0a3048 2018-01-10 stsp #include <endian.h>
30 a1fd68d8 2018-01-12 stsp #include <zlib.h>
32 0a0a3048 2018-01-10 stsp #include "got_error.h"
33 a1fd68d8 2018-01-12 stsp #include "got_object.h"
34 a1fd68d8 2018-01-12 stsp #include "got_repository.h"
35 a1fd68d8 2018-01-12 stsp #include "got_sha1.h"
36 0a0a3048 2018-01-10 stsp #include "pack.h"
37 a1fd68d8 2018-01-12 stsp #include "path.h"
38 efd2a263 2018-01-19 stsp #include "delta.h"
39 eef6493a 2018-01-19 stsp #include "object.h"
41 a1fd68d8 2018-01-12 stsp #define GOT_PACK_PREFIX "pack-"
42 a1fd68d8 2018-01-12 stsp #define GOT_PACKFILE_SUFFIX ".pack"
43 a1fd68d8 2018-01-12 stsp #define GOT_PACKIDX_SUFFIX ".idx"
44 a1fd68d8 2018-01-12 stsp #define GOT_PACKFILE_NAMELEN (strlen(GOT_PACK_PREFIX) + \
45 a1fd68d8 2018-01-12 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
46 a1fd68d8 2018-01-12 stsp strlen(GOT_PACKFILE_SUFFIX))
47 a1fd68d8 2018-01-12 stsp #define GOT_PACKIDX_NAMELEN (strlen(GOT_PACK_PREFIX) + \
48 a1fd68d8 2018-01-12 stsp SHA1_DIGEST_STRING_LENGTH - 1 + \
49 a1fd68d8 2018-01-12 stsp strlen(GOT_PACKIDX_SUFFIX))
52 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
55 0a0a3048 2018-01-10 stsp static const struct got_error *
56 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
60 0a0a3048 2018-01-10 stsp for (i = 0; i < 0xff - 1; i++) {
61 a1fd68d8 2018-01-12 stsp if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
62 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PACKIDX);
65 0a0a3048 2018-01-10 stsp return NULL;
68 24541888 2018-01-10 stsp static const struct got_error *
69 0a0a3048 2018-01-10 stsp get_packfile_size(size_t *size, const char *path_idx)
71 0a0a3048 2018-01-10 stsp struct stat sb;
72 0a0a3048 2018-01-10 stsp char *path_pack;
73 0a0a3048 2018-01-10 stsp char base_path[PATH_MAX];
76 0a0a3048 2018-01-10 stsp if (strlcpy(base_path, path_idx, PATH_MAX) > PATH_MAX)
77 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_NO_SPACE);
79 0a0a3048 2018-01-10 stsp dot = strrchr(base_path, '.');
80 0a0a3048 2018-01-10 stsp if (dot == NULL)
81 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PATH);
82 0a0a3048 2018-01-10 stsp *dot = '\0';
83 0a0a3048 2018-01-10 stsp if (asprintf(&path_pack, "%s.pack", base_path) == -1)
84 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_NO_MEM);
86 0a0a3048 2018-01-10 stsp if (stat(path_pack, &sb) != 0) {
87 0a0a3048 2018-01-10 stsp free(path_pack);
88 8251fdbc 2018-01-12 stsp return got_error_from_errno();
91 0a0a3048 2018-01-10 stsp free(path_pack);
92 0a0a3048 2018-01-10 stsp *size = sb.st_size;
96 0a0a3048 2018-01-10 stsp const struct got_error *
97 0a0a3048 2018-01-10 stsp got_packidx_open(struct got_packidx_v2_hdr **packidx, const char *path)
99 0a0a3048 2018-01-10 stsp struct got_packidx_v2_hdr *p;
101 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
102 0a0a3048 2018-01-10 stsp size_t n, nobj, packfile_size;
103 0ebaf008 2018-01-10 stsp SHA1_CTX ctx;
104 0ebaf008 2018-01-10 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
106 0ebaf008 2018-01-10 stsp SHA1Init(&ctx);
108 0a0a3048 2018-01-10 stsp f = fopen(path, "rb");
109 0a0a3048 2018-01-10 stsp if (f == NULL)
110 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PATH);
112 0a0a3048 2018-01-10 stsp err = get_packfile_size(&packfile_size, path);
114 0a0a3048 2018-01-10 stsp return err;
116 0a0a3048 2018-01-10 stsp p = calloc(1, sizeof(*p));
117 0a0a3048 2018-01-10 stsp if (p == NULL) {
118 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_NO_MEM);
122 0a0a3048 2018-01-10 stsp n = fread(&p->magic, sizeof(p->magic), 1, f);
123 0a0a3048 2018-01-10 stsp if (n != 1) {
124 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
128 0a0a3048 2018-01-10 stsp if (betoh32(p->magic) != GOT_PACKIDX_V2_MAGIC) {
129 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
133 0ebaf008 2018-01-10 stsp SHA1Update(&ctx, (uint8_t *)&p->magic, sizeof(p->magic));
135 0a0a3048 2018-01-10 stsp n = fread(&p->version, sizeof(p->version), 1, f);
136 0a0a3048 2018-01-10 stsp if (n != 1) {
137 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
141 0a0a3048 2018-01-10 stsp if (betoh32(p->version) != GOT_PACKIDX_VERSION) {
142 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
146 0ebaf008 2018-01-10 stsp SHA1Update(&ctx, (uint8_t *)&p->version, sizeof(p->version));
148 0a0a3048 2018-01-10 stsp n = fread(&p->fanout_table, sizeof(p->fanout_table), 1, f);
149 0a0a3048 2018-01-10 stsp if (n != 1) {
150 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
154 0a0a3048 2018-01-10 stsp err = verify_fanout_table(p->fanout_table);
158 0ebaf008 2018-01-10 stsp SHA1Update(&ctx, (uint8_t *)p->fanout_table, sizeof(p->fanout_table));
160 0a0a3048 2018-01-10 stsp nobj = betoh32(p->fanout_table[0xff]);
162 0a0a3048 2018-01-10 stsp p->sorted_ids = calloc(nobj, sizeof(*p->sorted_ids));
163 0a0a3048 2018-01-10 stsp if (p->sorted_ids == NULL) {
164 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_NO_MEM);
168 0a0a3048 2018-01-10 stsp n = fread(p->sorted_ids, sizeof(*p->sorted_ids), nobj, f);
169 0a0a3048 2018-01-10 stsp if (n != nobj) {
170 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
174 0ebaf008 2018-01-10 stsp SHA1Update(&ctx, (uint8_t *)p->sorted_ids,
175 0ebaf008 2018-01-10 stsp nobj * sizeof(*p->sorted_ids));
177 a1fd68d8 2018-01-12 stsp p->crc32 = calloc(nobj, sizeof(*p->crc32));
178 a1fd68d8 2018-01-12 stsp if (p->crc32 == NULL) {
179 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_NO_MEM);
183 a1fd68d8 2018-01-12 stsp n = fread(p->crc32, sizeof(*p->crc32), nobj, f);
184 0a0a3048 2018-01-10 stsp if (n != nobj) {
185 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
189 a1fd68d8 2018-01-12 stsp SHA1Update(&ctx, (uint8_t *)p->crc32, nobj * sizeof(*p->crc32));
191 a1fd68d8 2018-01-12 stsp p->offsets = calloc(nobj, sizeof(*p->offsets));
192 a1fd68d8 2018-01-12 stsp if (p->offsets == NULL) {
193 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_NO_MEM);
197 a1fd68d8 2018-01-12 stsp n = fread(p->offsets, sizeof(*p->offsets), nobj, f);
198 0a0a3048 2018-01-10 stsp if (n != nobj) {
199 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
203 a1fd68d8 2018-01-12 stsp SHA1Update(&ctx, (uint8_t *)p->offsets, nobj * sizeof(*p->offsets));
205 0a0a3048 2018-01-10 stsp /* Large file offsets are contained only in files > 2GB. */
206 b0517dd0 2018-01-10 stsp if (packfile_size <= 0x80000000)
207 0a0a3048 2018-01-10 stsp goto checksum;
209 0a0a3048 2018-01-10 stsp p->large_offsets = calloc(nobj, sizeof(*p->large_offsets));
210 0a0a3048 2018-01-10 stsp if (p->large_offsets == NULL) {
211 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_NO_MEM);
215 0a0a3048 2018-01-10 stsp n = fread(p->large_offsets, sizeof(*p->large_offsets), nobj, f);
216 0a0a3048 2018-01-10 stsp if (n != nobj) {
217 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
221 0ebaf008 2018-01-10 stsp SHA1Update(&ctx, (uint8_t*)p->large_offsets,
222 0ebaf008 2018-01-10 stsp nobj * sizeof(*p->large_offsets));
225 0a0a3048 2018-01-10 stsp n = fread(&p->trailer, sizeof(p->trailer), 1, f);
226 0a0a3048 2018-01-10 stsp if (n != 1) {
227 8251fdbc 2018-01-12 stsp err = got_ferror(f, GOT_ERR_BAD_PACKIDX);
231 a1fd68d8 2018-01-12 stsp SHA1Update(&ctx, p->trailer.packfile_sha1, SHA1_DIGEST_LENGTH);
232 0ebaf008 2018-01-10 stsp SHA1Final(sha1, &ctx);
233 a1fd68d8 2018-01-12 stsp if (memcmp(p->trailer.packidx_sha1, sha1, SHA1_DIGEST_LENGTH) != 0)
234 0ebaf008 2018-01-10 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
238 0a0a3048 2018-01-10 stsp got_packidx_close(p);
240 0a0a3048 2018-01-10 stsp *packidx = p;
241 0a0a3048 2018-01-10 stsp return err;
245 0a0a3048 2018-01-10 stsp got_packidx_close(struct got_packidx_v2_hdr *packidx)
247 0a0a3048 2018-01-10 stsp free(packidx->sorted_ids);
248 0a0a3048 2018-01-10 stsp free(packidx->offsets);
249 0a0a3048 2018-01-10 stsp free(packidx->crc32);
250 0a0a3048 2018-01-10 stsp free(packidx->large_offsets);
251 0a0a3048 2018-01-10 stsp free(packidx);
255 a1fd68d8 2018-01-12 stsp is_packidx_filename(const char *name, size_t len)
257 a1fd68d8 2018-01-12 stsp if (len != GOT_PACKIDX_NAMELEN)
260 a1fd68d8 2018-01-12 stsp if (strncmp(name, GOT_PACK_PREFIX, strlen(GOT_PACK_PREFIX)) != 0)
263 a1fd68d8 2018-01-12 stsp if (strcmp(name + strlen(GOT_PACK_PREFIX) +
264 a1fd68d8 2018-01-12 stsp SHA1_DIGEST_STRING_LENGTH - 1, GOT_PACKIDX_SUFFIX) != 0)
270 a1fd68d8 2018-01-12 stsp static off_t
271 a1fd68d8 2018-01-12 stsp get_object_offset(struct got_packidx_v2_hdr *packidx, int idx)
273 a1fd68d8 2018-01-12 stsp uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
274 a1fd68d8 2018-01-12 stsp uint32_t offset = betoh32(packidx->offsets[idx]);
275 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
276 a1fd68d8 2018-01-12 stsp uint64_t loffset;
277 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
278 a1fd68d8 2018-01-12 stsp if (idx < 0 || idx > totobj || packidx->large_offsets == NULL)
280 a1fd68d8 2018-01-12 stsp loffset = betoh64(packidx->large_offsets[idx]);
281 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
283 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
287 a1fd68d8 2018-01-12 stsp get_object_idx(struct got_packidx_v2_hdr *packidx, struct got_object_id *id)
289 a1fd68d8 2018-01-12 stsp u_int8_t id0 = id->sha1[0];
290 a1fd68d8 2018-01-12 stsp uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
293 a1fd68d8 2018-01-12 stsp if (id0 > 0)
294 a1fd68d8 2018-01-12 stsp i = betoh32(packidx->fanout_table[id0 - 1]);
296 a1fd68d8 2018-01-12 stsp while (i < totobj) {
297 6c00b545 2018-01-17 stsp struct got_object_id *oid = &packidx->sorted_ids[i];
298 a1fd68d8 2018-01-12 stsp uint32_t offset;
299 2b2ca9f0 2018-01-13 stsp int cmp = got_object_id_cmp(id, oid);
301 6c00b545 2018-01-17 stsp if (cmp == 0)
303 6c00b545 2018-01-17 stsp else if (cmp > 0)
311 6b9c9673 2018-01-23 stsp static const struct got_error *
312 6b9c9673 2018-01-23 stsp search_packidx(struct got_packidx_v2_hdr **packidx, int *idx,
313 6b9c9673 2018-01-23 stsp struct got_repository *repo, struct got_object_id *id)
315 6b9c9673 2018-01-23 stsp const struct got_error *err;
316 6b9c9673 2018-01-23 stsp char *path_packdir;
317 6b9c9673 2018-01-23 stsp DIR *packdir;
318 6b9c9673 2018-01-23 stsp struct dirent *dent;
319 6b9c9673 2018-01-23 stsp char *path_packidx;
321 6b9c9673 2018-01-23 stsp path_packdir = got_repo_get_path_objects_pack(repo);
322 6b9c9673 2018-01-23 stsp if (path_packdir == NULL)
323 6b9c9673 2018-01-23 stsp return got_error(GOT_ERR_NO_MEM);
325 6b9c9673 2018-01-23 stsp packdir = opendir(path_packdir);
326 6b9c9673 2018-01-23 stsp if (packdir == NULL) {
327 6b9c9673 2018-01-23 stsp err = got_error_from_errno();
331 6b9c9673 2018-01-23 stsp while ((dent = readdir(packdir)) != NULL) {
332 6b9c9673 2018-01-23 stsp if (!is_packidx_filename(dent->d_name, dent->d_namlen))
335 6b9c9673 2018-01-23 stsp if (asprintf(&path_packidx, "%s/%s", path_packdir,
336 6b9c9673 2018-01-23 stsp dent->d_name) == -1) {
337 6b9c9673 2018-01-23 stsp err = got_error(GOT_ERR_NO_MEM);
341 6b9c9673 2018-01-23 stsp err = got_packidx_open(packidx, path_packidx);
342 6b9c9673 2018-01-23 stsp free(path_packidx);
346 6b9c9673 2018-01-23 stsp *idx = get_object_idx(*packidx, id);
347 6b9c9673 2018-01-23 stsp if (*idx != -1) {
348 6b9c9673 2018-01-23 stsp err = NULL; /* found the object */
352 6b9c9673 2018-01-23 stsp got_packidx_close(*packidx);
353 6b9c9673 2018-01-23 stsp *packidx = NULL;
356 6b9c9673 2018-01-23 stsp err = got_error(GOT_ERR_NO_OBJ);
358 6b9c9673 2018-01-23 stsp free(path_packdir);
359 6b9c9673 2018-01-23 stsp if (closedir(packdir) != 0 && err == 0)
360 6b9c9673 2018-01-23 stsp err = got_error_from_errno();
361 6b9c9673 2018-01-23 stsp return err;
364 6b9c9673 2018-01-23 stsp const struct got_error *
365 6b9c9673 2018-01-23 stsp get_packfile_path(char **path_packfile, struct got_repository *repo,
366 6b9c9673 2018-01-23 stsp struct got_packidx_v2_hdr *packidx)
368 6b9c9673 2018-01-23 stsp char *path_packdir;
369 6b9c9673 2018-01-23 stsp char hex[SHA1_DIGEST_STRING_LENGTH];
370 6b9c9673 2018-01-23 stsp char *sha1str;
371 6b9c9673 2018-01-23 stsp char *path_packidx;
373 6b9c9673 2018-01-23 stsp *path_packfile = NULL;
375 6b9c9673 2018-01-23 stsp path_packdir = got_repo_get_path_objects_pack(repo);
376 6b9c9673 2018-01-23 stsp if (path_packdir == NULL)
377 6b9c9673 2018-01-23 stsp return got_error(GOT_ERR_NO_MEM);
379 6b9c9673 2018-01-23 stsp sha1str = got_sha1_digest_to_str(packidx->trailer.packfile_sha1,
380 6b9c9673 2018-01-23 stsp hex, sizeof(hex));
381 6b9c9673 2018-01-23 stsp if (sha1str == NULL)
382 6b9c9673 2018-01-23 stsp return got_error(GOT_ERR_PACKIDX_CSUM);
384 6b9c9673 2018-01-23 stsp if (asprintf(path_packfile, "%s/%s%s%s", path_packdir,
385 6b9c9673 2018-01-23 stsp GOT_PACK_PREFIX, sha1str, GOT_PACKFILE_SUFFIX) == -1) {
386 6b9c9673 2018-01-23 stsp *path_packfile = NULL;
387 6b9c9673 2018-01-23 stsp return got_error(GOT_ERR_NO_MEM);
390 6b9c9673 2018-01-23 stsp return NULL;
393 a1fd68d8 2018-01-12 stsp const struct got_error *
394 a1fd68d8 2018-01-12 stsp read_packfile_hdr(FILE *f, struct got_packidx_v2_hdr *packidx)
396 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
397 a1fd68d8 2018-01-12 stsp uint32_t totobj = betoh32(packidx->fanout_table[0xff]);
398 a1fd68d8 2018-01-12 stsp struct got_packfile_hdr hdr;
401 a1fd68d8 2018-01-12 stsp n = fread(&hdr, sizeof(hdr), 1, f);
402 a1fd68d8 2018-01-12 stsp if (n != 1)
403 8251fdbc 2018-01-12 stsp return got_ferror(f, GOT_ERR_BAD_PACKIDX);
405 a1fd68d8 2018-01-12 stsp if (betoh32(hdr.signature) != GOT_PACKFILE_SIGNATURE ||
406 a1fd68d8 2018-01-12 stsp betoh32(hdr.version) != GOT_PACKFILE_VERSION ||
407 a1fd68d8 2018-01-12 stsp betoh32(hdr.nobjects) != totobj)
408 a1fd68d8 2018-01-12 stsp err = got_error(GOT_ERR_BAD_PACKFILE);
410 a1fd68d8 2018-01-12 stsp return err;
413 a487c1d0 2018-01-14 stsp static const struct got_error *
414 348f621c 2018-01-23 stsp parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
415 c3703302 2018-01-23 stsp FILE *packfile)
417 a487c1d0 2018-01-14 stsp uint8_t t = 0;
418 a487c1d0 2018-01-14 stsp uint64_t s = 0;
419 a487c1d0 2018-01-14 stsp uint8_t sizeN;
424 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
426 a487c1d0 2018-01-14 stsp return got_error(GOT_ERR_NO_SPACE);
428 a1fd68d8 2018-01-12 stsp n = fread(&sizeN, sizeof(sizeN), 1, packfile);
429 a487c1d0 2018-01-14 stsp if (n != 1)
430 a487c1d0 2018-01-14 stsp return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
432 a1fd68d8 2018-01-12 stsp if (i == 0) {
433 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
434 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
435 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
437 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
438 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
441 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
445 3ee5fc21 2018-01-17 stsp *len = i * sizeof(sizeN);
446 a487c1d0 2018-01-14 stsp return NULL;
449 a1fd68d8 2018-01-12 stsp static const struct got_error *
450 9710aac2 2018-01-19 stsp open_plain_object(struct got_object **obj, const char *path_packfile,
451 6ccb713b 2018-01-19 stsp struct got_object_id *id, uint8_t type, off_t offset, size_t size)
453 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
454 6ccb713b 2018-01-19 stsp if (*obj == NULL)
455 6ccb713b 2018-01-19 stsp return got_error(GOT_ERR_NO_MEM);
457 6ccb713b 2018-01-19 stsp (*obj)->path_packfile = strdup(path_packfile);
458 6ccb713b 2018-01-19 stsp if ((*obj)->path_packfile == NULL) {
459 6ccb713b 2018-01-19 stsp free(*obj);
460 6ccb713b 2018-01-19 stsp *obj = NULL;
461 6ccb713b 2018-01-19 stsp return got_error(GOT_ERR_NO_MEM);
464 6ccb713b 2018-01-19 stsp (*obj)->type = type;
465 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
466 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
467 6ccb713b 2018-01-19 stsp (*obj)->size = size;
468 6ccb713b 2018-01-19 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
469 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
471 b107e67f 2018-01-19 stsp return NULL;
474 b107e67f 2018-01-19 stsp static const struct got_error *
475 348f621c 2018-01-23 stsp parse_negative_offset(int64_t *offset, size_t *len, FILE *packfile)
477 b107e67f 2018-01-19 stsp int64_t o = 0;
478 b107e67f 2018-01-19 stsp uint8_t offN;
483 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
485 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
487 b107e67f 2018-01-19 stsp n = fread(&offN, sizeof(offN), 1, packfile);
488 b107e67f 2018-01-19 stsp if (n != 1)
489 b107e67f 2018-01-19 stsp return got_ferror(packfile, GOT_ERR_BAD_PACKIDX);
491 b107e67f 2018-01-19 stsp if (i == 0)
492 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
496 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
499 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
501 b107e67f 2018-01-19 stsp *offset = o;
502 b107e67f 2018-01-19 stsp *len = i * sizeof(offN);
503 6ccb713b 2018-01-19 stsp return NULL;
506 6ccb713b 2018-01-19 stsp static const struct got_error *
507 96f5e8b3 2018-01-23 stsp parse_offset_delta(off_t *base_offset, FILE *packfile, off_t offset)
509 96f5e8b3 2018-01-23 stsp const struct got_error *err;
510 b107e67f 2018-01-19 stsp int64_t negoffset;
511 b107e67f 2018-01-19 stsp size_t negofflen;
513 348f621c 2018-01-23 stsp err = parse_negative_offset(&negoffset, &negofflen, packfile);
515 b107e67f 2018-01-19 stsp return err;
517 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
518 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
519 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
520 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
522 96f5e8b3 2018-01-23 stsp return NULL;
525 a3500804 2018-01-23 stsp static const struct got_error *resolve_delta_chain(struct got_delta_chain *,
526 6b9c9673 2018-01-23 stsp struct got_repository *repo, FILE *, const char *, int, off_t, size_t);
528 96f5e8b3 2018-01-23 stsp static const struct got_error *
529 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
530 6b9c9673 2018-01-23 stsp struct got_repository *repo, FILE *packfile, const char *path_packfile,
531 6b9c9673 2018-01-23 stsp off_t delta_offset)
533 a3500804 2018-01-23 stsp const struct got_error *err;
534 c3703302 2018-01-23 stsp off_t base_offset;
535 c3703302 2018-01-23 stsp uint8_t base_type;
536 c3703302 2018-01-23 stsp uint64_t base_size;
537 c3703302 2018-01-23 stsp size_t base_tslen;
539 c3703302 2018-01-23 stsp err = parse_offset_delta(&base_offset, packfile, delta_offset);
541 a3500804 2018-01-23 stsp return err;
543 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
544 c3703302 2018-01-23 stsp if (fseeko(packfile, base_offset, SEEK_SET) != 0)
545 b107e67f 2018-01-19 stsp return got_error_from_errno();
547 348f621c 2018-01-23 stsp err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
550 b107e67f 2018-01-19 stsp return err;
552 6b9c9673 2018-01-23 stsp return resolve_delta_chain(deltas, repo, packfile, path_packfile,
553 c3703302 2018-01-23 stsp base_type, base_offset + base_tslen, base_size);
556 c3703302 2018-01-23 stsp static const struct got_error *
557 6b9c9673 2018-01-23 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_repository *repo,
558 6b9c9673 2018-01-23 stsp FILE *packfile, const char *path_packfile, off_t delta_offset)
560 6b9c9673 2018-01-23 stsp const struct got_error *err;
561 6b9c9673 2018-01-23 stsp struct got_object_id id;
562 6b9c9673 2018-01-23 stsp struct got_packidx_v2_hdr *packidx;
564 6b9c9673 2018-01-23 stsp off_t base_offset;
565 6b9c9673 2018-01-23 stsp uint8_t base_type;
566 6b9c9673 2018-01-23 stsp uint64_t base_size;
567 6b9c9673 2018-01-23 stsp size_t base_tslen;
569 6b9c9673 2018-01-23 stsp FILE *base_packfile;
570 6b9c9673 2018-01-23 stsp char *path_base_packfile;
572 6b9c9673 2018-01-23 stsp n = fread(&id, sizeof(id), 1, packfile);
573 6b9c9673 2018-01-23 stsp if (n != 1)
574 6b9c9673 2018-01-23 stsp return got_ferror(packfile, GOT_ERR_IO);
576 6b9c9673 2018-01-23 stsp err = search_packidx(&packidx, &idx, repo, &id);
578 6b9c9673 2018-01-23 stsp return err;
580 6b9c9673 2018-01-23 stsp base_offset = get_object_offset(packidx, idx);
581 6b9c9673 2018-01-23 stsp if (base_offset == (uint64_t)-1) {
582 6b9c9673 2018-01-23 stsp got_packidx_close(packidx);
583 6b9c9673 2018-01-23 stsp return got_error(GOT_ERR_BAD_PACKIDX);
586 6b9c9673 2018-01-23 stsp err = get_packfile_path(&path_base_packfile, repo, packidx);
587 6b9c9673 2018-01-23 stsp got_packidx_close(packidx);
589 6b9c9673 2018-01-23 stsp return err;
591 6b9c9673 2018-01-23 stsp base_packfile = fopen(path_base_packfile, "rb");
592 6b9c9673 2018-01-23 stsp if (base_packfile == NULL) {
593 6b9c9673 2018-01-23 stsp err = got_error_from_errno();
597 6b9c9673 2018-01-23 stsp if (fseeko(base_packfile, base_offset, SEEK_SET) != 0) {
598 6b9c9673 2018-01-23 stsp err = got_error_from_errno();
602 6b9c9673 2018-01-23 stsp err = parse_object_type_and_size(&base_type, &base_size, &base_tslen,
603 6b9c9673 2018-01-23 stsp base_packfile);
607 6b9c9673 2018-01-23 stsp err = resolve_delta_chain(deltas, repo, base_packfile,
608 6b9c9673 2018-01-23 stsp path_base_packfile, base_type, base_offset + base_tslen, base_size);
610 6b9c9673 2018-01-23 stsp free(path_base_packfile);
611 6b9c9673 2018-01-23 stsp if (base_packfile && fclose(base_packfile) == -1 && err == 0)
612 6b9c9673 2018-01-23 stsp err = got_error_from_errno();
613 6b9c9673 2018-01-23 stsp return err;
616 6b9c9673 2018-01-23 stsp static const struct got_error *
617 6b9c9673 2018-01-23 stsp resolve_delta_chain(struct got_delta_chain *deltas, struct got_repository *repo,
618 6b9c9673 2018-01-23 stsp FILE *packfile, const char *path_packfile, int delta_type,
619 6b9c9673 2018-01-23 stsp off_t delta_offset, size_t delta_size)
621 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
622 c3703302 2018-01-23 stsp struct got_delta *delta;
624 c3703302 2018-01-23 stsp delta = got_delta_open(path_packfile, delta_type, delta_offset,
625 96f5e8b3 2018-01-23 stsp delta_size);
626 c3703302 2018-01-23 stsp if (delta == NULL)
627 96f5e8b3 2018-01-23 stsp return got_error(GOT_ERR_NO_MEM);
628 96f5e8b3 2018-01-23 stsp deltas->nentries++;
629 c3703302 2018-01-23 stsp SIMPLEQ_INSERT_TAIL(&deltas->entries, delta, entry);
630 c3703302 2018-01-23 stsp /* In case of error below, delta is freed in got_object_close(). */
632 c3703302 2018-01-23 stsp switch (delta_type) {
633 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
634 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
635 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
636 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
637 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
639 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
640 6b9c9673 2018-01-23 stsp err = resolve_offset_delta(deltas, repo, packfile,
641 6b9c9673 2018-01-23 stsp path_packfile, delta_offset);
643 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
644 6b9c9673 2018-01-23 stsp err = resolve_ref_delta(deltas, repo, packfile, path_packfile,
645 6b9c9673 2018-01-23 stsp delta_offset);
648 4e8cda55 2018-01-19 stsp return got_error(GOT_ERR_NOT_IMPL);
651 96f5e8b3 2018-01-23 stsp return err;
654 96f5e8b3 2018-01-23 stsp static const struct got_error *
655 a48db7e5 2018-01-23 stsp open_delta_object(struct got_object **obj, struct got_repository *repo,
656 a48db7e5 2018-01-23 stsp struct got_packidx_v2_hdr *packidx, const char *path_packfile,
657 a48db7e5 2018-01-23 stsp FILE *packfile, struct got_object_id *id, off_t offset, size_t tslen,
658 a48db7e5 2018-01-23 stsp int delta_type, size_t delta_size)
660 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
661 96f5e8b3 2018-01-23 stsp struct got_object_id base_id;
662 96f5e8b3 2018-01-23 stsp uint8_t base_type;
663 96f5e8b3 2018-01-23 stsp int resolved_type;
664 96f5e8b3 2018-01-23 stsp uint64_t base_size;
665 96f5e8b3 2018-01-23 stsp size_t base_tslen;
667 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
668 b107e67f 2018-01-19 stsp if (*obj == NULL)
669 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_MEM);
671 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
672 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
673 96f5e8b3 2018-01-23 stsp (*obj)->size = 0; /* Not yet known because deltas aren't combined. */
674 b107e67f 2018-01-19 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
675 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
677 96f5e8b3 2018-01-23 stsp (*obj)->path_packfile = strdup(path_packfile);
678 96f5e8b3 2018-01-23 stsp if ((*obj)->path_packfile == NULL) {
679 96f5e8b3 2018-01-23 stsp err = got_error(GOT_ERR_NO_MEM);
682 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
684 96f5e8b3 2018-01-23 stsp SIMPLEQ_INIT(&(*obj)->deltas.entries);
685 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
687 6b9c9673 2018-01-23 stsp err = resolve_delta_chain(&(*obj)->deltas, repo, packfile,
688 a48db7e5 2018-01-23 stsp path_packfile, delta_type, offset, delta_size);
692 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
695 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
699 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
700 96f5e8b3 2018-01-23 stsp *obj = NULL;
702 96f5e8b3 2018-01-23 stsp return err;
705 b107e67f 2018-01-19 stsp static const struct got_error *
706 6c00b545 2018-01-17 stsp open_packed_object(struct got_object **obj, struct got_repository *repo,
707 6b9c9673 2018-01-23 stsp struct got_packidx_v2_hdr *packidx, int idx, struct got_object_id *id)
709 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
710 a1fd68d8 2018-01-12 stsp off_t offset;
711 6c00b545 2018-01-17 stsp char *path_packfile;
712 6c00b545 2018-01-17 stsp FILE *packfile;
713 6c00b545 2018-01-17 stsp uint8_t type;
714 6c00b545 2018-01-17 stsp uint64_t size;
715 3ee5fc21 2018-01-17 stsp size_t tslen;
717 6c00b545 2018-01-17 stsp *obj = NULL;
719 a1fd68d8 2018-01-12 stsp offset = get_object_offset(packidx, idx);
720 a1fd68d8 2018-01-12 stsp if (offset == (uint64_t)-1)
721 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
723 6b9c9673 2018-01-23 stsp err = get_packfile_path(&path_packfile, repo, packidx);
725 6b9c9673 2018-01-23 stsp return err;
727 a1fd68d8 2018-01-12 stsp packfile = fopen(path_packfile, "rb");
728 a1fd68d8 2018-01-12 stsp if (packfile == NULL) {
729 f334529e 2018-01-12 stsp err = got_error_from_errno();
733 a1fd68d8 2018-01-12 stsp err = read_packfile_hdr(packfile, packidx);
737 6c00b545 2018-01-17 stsp if (fseeko(packfile, offset, SEEK_SET) != 0) {
738 6c00b545 2018-01-17 stsp err = got_error_from_errno();
742 348f621c 2018-01-23 stsp err = parse_object_type_and_size(&type, &size, &tslen, packfile);
746 6c00b545 2018-01-17 stsp switch (type) {
747 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
748 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
749 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
750 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
751 9710aac2 2018-01-19 stsp err = open_plain_object(obj, path_packfile, id, type,
752 6ccb713b 2018-01-19 stsp offset + tslen, size);
755 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
756 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
757 a48db7e5 2018-01-23 stsp err = open_delta_object(obj, repo, packidx, path_packfile,
758 a48db7e5 2018-01-23 stsp packfile, id, offset, tslen, type, size);
762 6c00b545 2018-01-17 stsp err = got_error(GOT_ERR_NOT_IMPL);
766 a1fd68d8 2018-01-12 stsp free(path_packfile);
767 f334529e 2018-01-12 stsp if (packfile && fclose(packfile) == -1 && err == 0)
768 f334529e 2018-01-12 stsp err = got_error_from_errno();
769 a1fd68d8 2018-01-12 stsp return err;
772 a1fd68d8 2018-01-12 stsp const struct got_error *
773 6c00b545 2018-01-17 stsp got_packfile_open_object(struct got_object **obj, struct got_object_id *id,
774 a1fd68d8 2018-01-12 stsp struct got_repository *repo)
776 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
777 6b9c9673 2018-01-23 stsp struct got_packidx_v2_hdr *packidx = NULL;
780 6b9c9673 2018-01-23 stsp err = search_packidx(&packidx, &idx, repo, id);
782 6b9c9673 2018-01-23 stsp return err;
784 6b9c9673 2018-01-23 stsp err = open_packed_object(obj, repo, packidx, idx, id);
785 6b9c9673 2018-01-23 stsp got_packidx_close(packidx);
786 3ee5fc21 2018-01-17 stsp return err;
789 3ee5fc21 2018-01-17 stsp static const struct got_error *
790 3ee5fc21 2018-01-17 stsp dump_plain_object(FILE *infile, uint8_t type, size_t size, FILE *outfile)
794 3ee5fc21 2018-01-17 stsp while (size > 0) {
795 3ee5fc21 2018-01-17 stsp uint8_t data[2048];
796 3ee5fc21 2018-01-17 stsp size_t len = MIN(size, sizeof(data));
798 3ee5fc21 2018-01-17 stsp n = fread(data, len, 1, infile);
799 3ee5fc21 2018-01-17 stsp if (n != 1)
800 efd2a263 2018-01-19 stsp return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
802 3ee5fc21 2018-01-17 stsp n = fwrite(data, len, 1, outfile);
803 3ee5fc21 2018-01-17 stsp if (n != 1)
804 efd2a263 2018-01-19 stsp return got_ferror(outfile, GOT_ERR_IO);
806 3ee5fc21 2018-01-17 stsp size -= len;
809 3ee5fc21 2018-01-17 stsp rewind(outfile);
810 3ee5fc21 2018-01-17 stsp return NULL;
813 efd2a263 2018-01-19 stsp static const struct got_error *
814 efd2a263 2018-01-19 stsp dump_ref_delta_object(struct got_repository *repo, FILE *infile, uint8_t type,
815 efd2a263 2018-01-19 stsp size_t size, FILE *outfile)
817 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
818 efd2a263 2018-01-19 stsp struct got_object_id base_id;
819 efd2a263 2018-01-19 stsp struct got_object *base_obj;
822 efd2a263 2018-01-19 stsp if (size < sizeof(base_id))
823 efd2a263 2018-01-19 stsp return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
825 efd2a263 2018-01-19 stsp n = fread(&base_id, sizeof(base_id), 1, infile);
826 efd2a263 2018-01-19 stsp if (n != 1)
827 efd2a263 2018-01-19 stsp return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
829 efd2a263 2018-01-19 stsp size -= sizeof(base_id);
830 efd2a263 2018-01-19 stsp if (size <= 0)
831 efd2a263 2018-01-19 stsp return got_ferror(infile, GOT_ERR_BAD_PACKFILE);
833 efd2a263 2018-01-19 stsp err = got_object_open(&base_obj, repo, &base_id);
835 efd2a263 2018-01-19 stsp return err;
837 efd2a263 2018-01-19 stsp err = got_delta_apply(repo, infile, size, base_obj, outfile);
838 efd2a263 2018-01-19 stsp got_object_close(base_obj);
839 efd2a263 2018-01-19 stsp return err;
842 3ee5fc21 2018-01-17 stsp const struct got_error *
843 3ee5fc21 2018-01-17 stsp got_packfile_extract_object(FILE **f, struct got_object *obj,
844 3ee5fc21 2018-01-17 stsp struct got_repository *repo)
846 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
847 3ee5fc21 2018-01-17 stsp FILE *packfile = NULL;
849 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
850 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
852 3ee5fc21 2018-01-17 stsp *f = got_opentemp();
853 3ee5fc21 2018-01-17 stsp if (*f == NULL) {
854 3ee5fc21 2018-01-17 stsp err = got_error(GOT_ERR_FILE_OPEN);
858 3ee5fc21 2018-01-17 stsp packfile = fopen(obj->path_packfile, "rb");
859 3ee5fc21 2018-01-17 stsp if (packfile == NULL) {
860 f334529e 2018-01-12 stsp err = got_error_from_errno();
864 3ee5fc21 2018-01-17 stsp if (fseeko(packfile, obj->pack_offset, SEEK_SET) != 0) {
865 3ee5fc21 2018-01-17 stsp err = got_error_from_errno();
869 3ee5fc21 2018-01-17 stsp switch (obj->type) {
870 3ee5fc21 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
871 3ee5fc21 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
872 3ee5fc21 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
873 b432fb3b 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
874 3ee5fc21 2018-01-17 stsp err = dump_plain_object(packfile, obj->type, obj->size, *f);
876 3ee5fc21 2018-01-17 stsp case GOT_OBJ_TYPE_REF_DELTA:
877 efd2a263 2018-01-19 stsp err = dump_ref_delta_object(repo, packfile, obj->type,
878 efd2a263 2018-01-19 stsp obj->size, *f);
880 3ee5fc21 2018-01-17 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
882 3ee5fc21 2018-01-17 stsp err = got_error(GOT_ERR_NOT_IMPL);
886 3ee5fc21 2018-01-17 stsp if (packfile)
887 3ee5fc21 2018-01-17 stsp fclose(packfile);
888 3ee5fc21 2018-01-17 stsp if (err && *f)
889 3ee5fc21 2018-01-17 stsp fclose(*f);
890 a1fd68d8 2018-01-12 stsp return err;
