2 5aa81393 2020-01-06 stsp * Copyright (c) 2018, 2019, 2020 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>
20 876c234b 2018-09-10 stsp #include <sys/uio.h>
21 57b35b75 2018-06-22 stsp #include <sys/mman.h>
22 3d589bee 2022-06-25 stsp #include <sys/resource.h>
23 3d589bee 2022-06-25 stsp #include <sys/socket.h>
25 7e656b93 2018-03-17 stsp #include <fcntl.h>
26 a1fd68d8 2018-01-12 stsp #include <errno.h>
27 0a0a3048 2018-01-10 stsp #include <stdio.h>
28 a1fd68d8 2018-01-12 stsp #include <stdint.h>
29 0a0a3048 2018-01-10 stsp #include <stdlib.h>
30 0a0a3048 2018-01-10 stsp #include <string.h>
31 0a0a3048 2018-01-10 stsp #include <limits.h>
32 0a0a3048 2018-01-10 stsp #include <sha1.h>
33 0a0a3048 2018-01-10 stsp #include <endian.h>
34 81a12da5 2020-09-09 naddy #include <unistd.h>
35 a1fd68d8 2018-01-12 stsp #include <zlib.h>
36 876c234b 2018-09-10 stsp #include <imsg.h>
38 0a0a3048 2018-01-10 stsp #include "got_error.h"
39 a1fd68d8 2018-01-12 stsp #include "got_object.h"
40 324d37e7 2019-05-11 stsp #include "got_path.h"
42 718b3ab0 2018-03-17 stsp #include "got_lib_sha1.h"
43 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
44 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
45 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
46 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
47 dd88155e 2019-06-29 stsp #include "got_lib_object_parse.h"
48 876c234b 2018-09-10 stsp #include "got_lib_privsep.h"
49 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
51 79b11c62 2018-03-09 stsp #ifndef nitems
52 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
56 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
59 0a0a3048 2018-01-10 stsp static const struct got_error *
60 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
64 0a0a3048 2018-01-10 stsp for (i = 0; i < 0xff - 1; i++) {
65 a1fd68d8 2018-01-12 stsp if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
66 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PACKIDX);
69 0a0a3048 2018-01-10 stsp return NULL;
72 1510f469 2018-09-09 stsp const struct got_error *
73 c3564dfa 2021-07-15 stsp got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
75 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
76 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
77 0ebaf008 2018-01-10 stsp SHA1_CTX ctx;
78 0ebaf008 2018-01-10 stsp uint8_t sha1[SHA1_DIGEST_LENGTH];
79 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
83 0ebaf008 2018-01-10 stsp SHA1Init(&ctx);
85 57b35b75 2018-06-22 stsp h = &p->hdr;
87 57b35b75 2018-06-22 stsp remain = p->len;
89 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
90 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
94 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
96 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
97 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
98 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
101 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
102 b1317e77 2019-09-22 stsp if (n < 0) {
103 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
105 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
106 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
110 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
111 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
114 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
115 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
117 0cb74cf4 2018-07-08 stsp if (verify)
118 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
120 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
121 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
124 fc79a48d 2018-07-09 stsp if (p->map)
125 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
127 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
128 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
129 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
132 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
133 c6368c2e 2019-10-11 stsp if (n < 0) {
134 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
136 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
137 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
141 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
142 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
145 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
146 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
148 0cb74cf4 2018-07-08 stsp if (verify)
149 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
151 57b35b75 2018-06-22 stsp len_fanout =
152 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
153 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
154 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
157 fc79a48d 2018-07-09 stsp if (p->map)
158 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
160 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
161 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
162 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
165 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
166 c6368c2e 2019-10-11 stsp if (n < 0) {
167 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
169 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
170 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
174 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
177 0cb74cf4 2018-07-08 stsp if (verify)
178 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
179 57b35b75 2018-06-22 stsp offset += len_fanout;
180 57b35b75 2018-06-22 stsp remain -= len_fanout;
182 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
183 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
184 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
185 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
188 fc79a48d 2018-07-09 stsp if (p->map)
189 fc79a48d 2018-07-09 stsp h->sorted_ids =
190 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
192 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
193 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
194 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
197 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
199 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
200 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
201 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
205 0cb74cf4 2018-07-08 stsp if (verify)
206 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
207 57b35b75 2018-06-22 stsp offset += len_ids;
208 57b35b75 2018-06-22 stsp remain -= len_ids;
210 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
211 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
214 fc79a48d 2018-07-09 stsp if (p->map)
215 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
217 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
218 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
219 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
222 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
224 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
225 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
226 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
230 0cb74cf4 2018-07-08 stsp if (verify)
231 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
232 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
233 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
235 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
236 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
239 fc79a48d 2018-07-09 stsp if (p->map)
240 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
242 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
243 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
244 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
247 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
249 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
250 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
251 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
255 0cb74cf4 2018-07-08 stsp if (verify)
256 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t *)h->offsets,
257 0cb74cf4 2018-07-08 stsp nobj * sizeof(*h->offsets));
258 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->offsets);
259 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->offsets);
261 0a0a3048 2018-01-10 stsp /* Large file offsets are contained only in files > 2GB. */
262 c3564dfa 2021-07-15 stsp if (verify || packfile_size > 0x7fffffff) {
263 c3564dfa 2021-07-15 stsp for (i = 0; i < nobj; i++) {
264 c3564dfa 2021-07-15 stsp uint32_t o = h->offsets[i];
265 c3564dfa 2021-07-15 stsp if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
266 c3564dfa 2021-07-15 stsp p->nlargeobj++;
269 5e6be232 2019-11-08 stsp if (p->nlargeobj == 0)
270 0a0a3048 2018-01-10 stsp goto checksum;
271 c3564dfa 2021-07-15 stsp else if (packfile_size <= 0x7fffffff) {
272 c3564dfa 2021-07-15 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
276 5e6be232 2019-11-08 stsp if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
277 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
280 fc79a48d 2018-07-09 stsp if (p->map)
281 fc79a48d 2018-07-09 stsp h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
283 5e6be232 2019-11-08 stsp h->large_offsets = malloc(p->nlargeobj *
284 5e6be232 2019-11-08 stsp sizeof(*h->large_offsets));
285 de30857e 2019-08-23 stsp if (h->large_offsets == NULL) {
286 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
289 fc79a48d 2018-07-09 stsp n = read(p->fd, h->large_offsets,
290 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
292 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
293 5e6be232 2019-11-08 stsp else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
294 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
298 0cb74cf4 2018-07-08 stsp if (verify)
299 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, (uint8_t*)h->large_offsets,
300 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
301 5e6be232 2019-11-08 stsp remain -= p->nlargeobj * sizeof(*h->large_offsets);
302 5e6be232 2019-11-08 stsp offset += p->nlargeobj * sizeof(*h->large_offsets);
305 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->trailer)) {
306 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
309 fc79a48d 2018-07-09 stsp if (p->map)
310 fc79a48d 2018-07-09 stsp h->trailer =
311 fc79a48d 2018-07-09 stsp (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
313 fc79a48d 2018-07-09 stsp h->trailer = malloc(sizeof(*h->trailer));
314 fc79a48d 2018-07-09 stsp if (h->trailer == NULL) {
315 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
318 fc79a48d 2018-07-09 stsp n = read(p->fd, h->trailer, sizeof(*h->trailer));
320 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
321 faaa1c0f 2018-09-15 stsp else if (n != sizeof(*h->trailer)) {
322 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
326 0cb74cf4 2018-07-08 stsp if (verify) {
327 0cb74cf4 2018-07-08 stsp SHA1Update(&ctx, h->trailer->packfile_sha1, SHA1_DIGEST_LENGTH);
328 0cb74cf4 2018-07-08 stsp SHA1Final(sha1, &ctx);
329 0cb74cf4 2018-07-08 stsp if (memcmp(h->trailer->packidx_sha1, sha1,
330 0cb74cf4 2018-07-08 stsp SHA1_DIGEST_LENGTH) != 0)
331 0cb74cf4 2018-07-08 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
334 817c5a18 2018-09-09 stsp return err;
337 817c5a18 2018-09-09 stsp const struct got_error *
338 6d5a9006 2020-12-16 yzhong got_packidx_open(struct got_packidx **packidx,
339 6d5a9006 2020-12-16 yzhong int dir_fd, const char *relpath, int verify)
341 817c5a18 2018-09-09 stsp const struct got_error *err = NULL;
342 1124fe40 2021-07-07 stsp struct got_packidx *p = NULL;
343 1124fe40 2021-07-07 stsp char *pack_relpath;
344 c3564dfa 2021-07-15 stsp struct stat idx_sb, pack_sb;
346 817c5a18 2018-09-09 stsp *packidx = NULL;
348 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath, relpath);
350 1124fe40 2021-07-07 stsp return err;
353 1124fe40 2021-07-07 stsp * Ensure that a corresponding pack file exists.
354 1124fe40 2021-07-07 stsp * Some Git repositories have this problem. Git seems to ignore
355 1124fe40 2021-07-07 stsp * the existence of lonely pack index files but we do not.
357 c3564dfa 2021-07-15 stsp if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
358 1124fe40 2021-07-07 stsp if (errno == ENOENT) {
359 1124fe40 2021-07-07 stsp err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
360 1124fe40 2021-07-07 stsp "%s", relpath);
362 1124fe40 2021-07-07 stsp err = got_error_from_errno2("fstatat", pack_relpath);
366 817c5a18 2018-09-09 stsp p = calloc(1, sizeof(*p));
367 1124fe40 2021-07-07 stsp if (p == NULL) {
368 1124fe40 2021-07-07 stsp err = got_error_from_errno("calloc");
372 e7ae0baf 2021-12-31 stsp p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
373 6772cf22 2019-08-28 hiltjo if (p->fd == -1) {
374 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("openat", relpath);
378 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
379 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
382 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
383 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
384 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
388 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
389 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
390 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
394 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
395 1c28a361 2022-10-25 op if (p->len > 0 && p->len <= SIZE_MAX) {
396 1c28a361 2022-10-25 op p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
397 1c28a361 2022-10-25 op if (p->map == MAP_FAILED) {
398 1c28a361 2022-10-25 op if (errno != ENOMEM) {
399 1c28a361 2022-10-25 op err = got_error_from_errno("mmap");
402 1c28a361 2022-10-25 op p->map = NULL; /* fall back to read(2) */
407 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
411 1124fe40 2021-07-07 stsp got_packidx_close(p);
413 0a0a3048 2018-01-10 stsp *packidx = p;
414 1124fe40 2021-07-07 stsp free(pack_relpath);
415 0a0a3048 2018-01-10 stsp return err;
418 57b35b75 2018-06-22 stsp const struct got_error *
419 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
421 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
423 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
424 fc79a48d 2018-07-09 stsp if (packidx->map) {
425 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
426 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
428 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
429 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
434 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
435 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
437 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
438 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
439 67fd6849 2022-02-13 stsp free(packidx->sorted_offsets);
440 67fd6849 2022-02-13 stsp free(packidx->sorted_large_offsets);
441 0a0a3048 2018-01-10 stsp free(packidx);
443 57b35b75 2018-06-22 stsp return err;
446 509c9973 2021-04-10 stsp const struct got_error *
447 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
449 509c9973 2021-04-10 stsp size_t size;
451 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
452 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
453 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
454 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
456 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
457 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
458 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
460 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
461 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
462 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
463 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
465 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
466 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
468 509c9973 2021-04-10 stsp return NULL;
472 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
474 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
475 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
476 a1fd68d8 2018-01-12 stsp uint64_t loffset;
477 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
478 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
479 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
481 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
482 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
484 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
488 c0df5966 2021-12-31 stsp got_packidx_get_object_idx(struct got_packidx *packidx,
489 c0df5966 2021-12-31 stsp struct got_object_id *id)
491 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
492 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
493 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
495 a1fd68d8 2018-01-12 stsp if (id0 > 0)
496 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
498 40aeb19c 2018-06-22 stsp while (left <= right) {
499 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
500 40aeb19c 2018-06-22 stsp int i, cmp;
502 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
503 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
504 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
505 16dcbf91 2018-04-01 stsp if (cmp == 0)
507 40aeb19c 2018-06-22 stsp else if (cmp > 0)
508 40aeb19c 2018-06-22 stsp left = i + 1;
509 40aeb19c 2018-06-22 stsp else if (cmp < 0)
510 40aeb19c 2018-06-22 stsp right = i - 1;
517 67fd6849 2022-02-13 stsp offset_cmp(const void *pa, const void *pb)
519 67fd6849 2022-02-13 stsp const struct got_pack_offset_index *a, *b;
521 67fd6849 2022-02-13 stsp a = (const struct got_pack_offset_index *)pa;
522 67fd6849 2022-02-13 stsp b = (const struct got_pack_offset_index *)pb;
524 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
526 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
533 67fd6849 2022-02-13 stsp large_offset_cmp(const void *pa, const void *pb)
535 67fd6849 2022-02-13 stsp const struct got_pack_large_offset_index *a, *b;
537 67fd6849 2022-02-13 stsp a = (const struct got_pack_large_offset_index *)pa;
538 67fd6849 2022-02-13 stsp b = (const struct got_pack_large_offset_index *)pb;
540 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
542 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
548 67fd6849 2022-02-13 stsp static const struct got_error *
549 67fd6849 2022-02-13 stsp build_offset_index(struct got_packidx *p)
551 67fd6849 2022-02-13 stsp uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
552 67fd6849 2022-02-13 stsp unsigned int i, j, k;
554 67fd6849 2022-02-13 stsp p->sorted_offsets = calloc(nobj - p->nlargeobj,
555 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]));
556 67fd6849 2022-02-13 stsp if (p->sorted_offsets == NULL)
557 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
559 67fd6849 2022-02-13 stsp if (p->nlargeobj > 0) {
560 67fd6849 2022-02-13 stsp p->sorted_large_offsets = calloc(p->nlargeobj,
561 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]));
562 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets == NULL)
563 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
568 67fd6849 2022-02-13 stsp for (i = 0; i < nobj; i++) {
569 67fd6849 2022-02-13 stsp uint32_t offset = be32toh(p->hdr.offsets[i]);
570 67fd6849 2022-02-13 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
571 67fd6849 2022-02-13 stsp uint64_t loffset;
572 67fd6849 2022-02-13 stsp uint32_t idx;
573 67fd6849 2022-02-13 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
574 67fd6849 2022-02-13 stsp if (idx >= p->nlargeobj ||
575 67fd6849 2022-02-13 stsp p->nlargeobj == 0 ||
576 67fd6849 2022-02-13 stsp p->hdr.large_offsets == NULL)
577 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
578 67fd6849 2022-02-13 stsp loffset = be64toh(p->hdr.large_offsets[idx]);
579 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].offset = loffset;
580 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].idx = i;
583 67fd6849 2022-02-13 stsp p->sorted_offsets[k].offset = offset;
584 67fd6849 2022-02-13 stsp p->sorted_offsets[k].idx = i;
588 67fd6849 2022-02-13 stsp if (j != p->nlargeobj || k != nobj - p->nlargeobj)
589 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
591 67fd6849 2022-02-13 stsp qsort(p->sorted_offsets, nobj - p->nlargeobj,
592 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]), offset_cmp);
594 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets)
595 67fd6849 2022-02-13 stsp qsort(p->sorted_large_offsets, p->nlargeobj,
596 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
598 67fd6849 2022-02-13 stsp return NULL;
601 876c234b 2018-09-10 stsp const struct got_error *
602 67fd6849 2022-02-13 stsp got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
604 67fd6849 2022-02-13 stsp const struct got_error *err;
605 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
606 67fd6849 2022-02-13 stsp int i, left, right;
610 67fd6849 2022-02-13 stsp if (packidx->sorted_offsets == NULL) {
611 67fd6849 2022-02-13 stsp err = build_offset_index(packidx);
613 67fd6849 2022-02-13 stsp return err;
616 67fd6849 2022-02-13 stsp if (offset >= 0x7fffffff) {
617 67fd6849 2022-02-13 stsp uint64_t lo;
618 67fd6849 2022-02-13 stsp left = 0, right = packidx->nlargeobj - 1;
619 67fd6849 2022-02-13 stsp while (left <= right) {
620 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
621 67fd6849 2022-02-13 stsp lo = packidx->sorted_large_offsets[i].offset;
622 67fd6849 2022-02-13 stsp if (lo == offset) {
623 67fd6849 2022-02-13 stsp *idx = packidx->sorted_large_offsets[i].idx;
625 67fd6849 2022-02-13 stsp } else if (offset > lo)
626 67fd6849 2022-02-13 stsp left = i + 1;
627 67fd6849 2022-02-13 stsp else if (offset < lo)
628 67fd6849 2022-02-13 stsp right = i - 1;
631 67fd6849 2022-02-13 stsp uint32_t o;
632 67fd6849 2022-02-13 stsp left = 0, right = totobj - packidx->nlargeobj - 1;
633 67fd6849 2022-02-13 stsp while (left <= right) {
634 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
635 67fd6849 2022-02-13 stsp o = packidx->sorted_offsets[i].offset;
636 67fd6849 2022-02-13 stsp if (o == offset) {
637 67fd6849 2022-02-13 stsp *idx = packidx->sorted_offsets[i].idx;
639 67fd6849 2022-02-13 stsp } else if (offset > o)
640 67fd6849 2022-02-13 stsp left = i + 1;
641 67fd6849 2022-02-13 stsp else if (offset < o)
642 67fd6849 2022-02-13 stsp right = i - 1;
646 67fd6849 2022-02-13 stsp return NULL;
649 67fd6849 2022-02-13 stsp const struct got_error *
650 67fd6849 2022-02-13 stsp got_packidx_get_object_id(struct got_object_id *id,
651 67fd6849 2022-02-13 stsp struct got_packidx *packidx, int idx)
653 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
654 67fd6849 2022-02-13 stsp struct got_packidx_object_id *oid;
656 67fd6849 2022-02-13 stsp if (idx < 0 || idx >= totobj)
657 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_NO_OBJ);
659 67fd6849 2022-02-13 stsp oid = &packidx->hdr.sorted_ids[idx];
660 67fd6849 2022-02-13 stsp memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
661 67fd6849 2022-02-13 stsp return NULL;
664 67fd6849 2022-02-13 stsp const struct got_error *
665 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
666 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
668 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
669 4277420a 2019-06-29 stsp u_int8_t id0;
670 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
671 4277420a 2019-06-29 stsp char hex[3];
672 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
673 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
674 404bde06 2022-01-03 stsp uint32_t i = 0;
676 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
678 4277420a 2019-06-29 stsp if (prefix_len < 2)
679 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
681 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
682 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
683 4277420a 2019-06-29 stsp hex[2] = '\0';
684 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
685 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
687 404bde06 2022-01-03 stsp if (id0 > 0)
688 404bde06 2022-01-03 stsp i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
689 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
690 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
691 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
692 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
695 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
696 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
698 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
699 4277420a 2019-06-29 stsp if (cmp < 0) {
700 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
702 4277420a 2019-06-29 stsp } else if (cmp > 0)
705 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
708 d7b5a0e8 2022-04-20 stsp memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
709 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
711 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
715 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
716 3d589bee 2022-06-25 stsp return err;
719 3d589bee 2022-06-25 stsp static void
720 3d589bee 2022-06-25 stsp set_max_datasize(void)
722 3d589bee 2022-06-25 stsp struct rlimit rl;
724 3d589bee 2022-06-25 stsp if (getrlimit(RLIMIT_DATA, &rl) != 0)
727 3d589bee 2022-06-25 stsp rl.rlim_cur = rl.rlim_max;
728 3d589bee 2022-06-25 stsp setrlimit(RLIMIT_DATA, &rl);
731 3d589bee 2022-06-25 stsp const struct got_error *
732 3d589bee 2022-06-25 stsp got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
734 3d589bee 2022-06-25 stsp const struct got_error *err = NULL;
735 3d589bee 2022-06-25 stsp int imsg_fds[2];
737 3d589bee 2022-06-25 stsp struct imsgbuf *ibuf;
739 3d589bee 2022-06-25 stsp ibuf = calloc(1, sizeof(*ibuf));
740 3d589bee 2022-06-25 stsp if (ibuf == NULL)
741 3d589bee 2022-06-25 stsp return got_error_from_errno("calloc");
743 3d589bee 2022-06-25 stsp pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
744 3d589bee 2022-06-25 stsp if (pack->privsep_child == NULL) {
745 3d589bee 2022-06-25 stsp err = got_error_from_errno("calloc");
746 3d589bee 2022-06-25 stsp free(ibuf);
747 3d589bee 2022-06-25 stsp return err;
749 3d589bee 2022-06-25 stsp pack->child_has_tempfiles = 0;
750 3d589bee 2022-06-25 stsp pack->child_has_delta_outfd = 0;
752 3d589bee 2022-06-25 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
753 3d589bee 2022-06-25 stsp err = got_error_from_errno("socketpair");
757 3d589bee 2022-06-25 stsp pid = fork();
758 3d589bee 2022-06-25 stsp if (pid == -1) {
759 3d589bee 2022-06-25 stsp err = got_error_from_errno("fork");
761 3d589bee 2022-06-25 stsp } else if (pid == 0) {
762 3d589bee 2022-06-25 stsp set_max_datasize();
763 3d589bee 2022-06-25 stsp got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
764 3d589bee 2022-06-25 stsp pack->path_packfile);
765 3d589bee 2022-06-25 stsp /* not reached */
768 3d589bee 2022-06-25 stsp if (close(imsg_fds[1]) == -1)
769 3d589bee 2022-06-25 stsp return got_error_from_errno("close");
770 3d589bee 2022-06-25 stsp pack->privsep_child->imsg_fd = imsg_fds[0];
771 3d589bee 2022-06-25 stsp pack->privsep_child->pid = pid;
772 3d589bee 2022-06-25 stsp imsg_init(ibuf, imsg_fds[0]);
773 3d589bee 2022-06-25 stsp pack->privsep_child->ibuf = ibuf;
775 3d589bee 2022-06-25 stsp err = got_privsep_init_pack_child(ibuf, pack, packidx);
777 3d589bee 2022-06-25 stsp const struct got_error *child_err;
778 3d589bee 2022-06-25 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
779 3d589bee 2022-06-25 stsp child_err = got_privsep_wait_for_child(
780 3d589bee 2022-06-25 stsp pack->privsep_child->pid);
781 3d589bee 2022-06-25 stsp if (child_err && err == NULL)
782 3d589bee 2022-06-25 stsp err = child_err;
786 3d589bee 2022-06-25 stsp free(ibuf);
787 3d589bee 2022-06-25 stsp free(pack->privsep_child);
788 3d589bee 2022-06-25 stsp pack->privsep_child = NULL;
790 dd88155e 2019-06-29 stsp return err;
793 b4f37570 2021-06-19 stsp static const struct got_error *
794 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
796 74ef8aae 2022-10-24 stsp const struct got_error *err = NULL, *close_err = NULL;
798 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
799 876c234b 2018-09-10 stsp return NULL;
801 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
803 96732e0b 2018-11-11 stsp return err;
804 74ef8aae 2022-10-24 stsp if (close(pack->privsep_child->imsg_fd) == -1)
805 74ef8aae 2022-10-24 stsp close_err = got_error_from_errno("close");
806 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
807 74ef8aae 2022-10-24 stsp if (close_err && err == NULL)
808 74ef8aae 2022-10-24 stsp err = close_err;
809 74ef8aae 2022-10-24 stsp imsg_clear(pack->privsep_child->ibuf);
810 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
811 876c234b 2018-09-10 stsp free(pack->privsep_child);
812 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
813 876c234b 2018-09-10 stsp return err;
816 d7464085 2018-07-09 stsp const struct got_error *
817 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
819 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
821 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
822 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
823 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
824 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
825 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
826 8b2180d4 2018-04-26 stsp pack->fd = -1;
827 4589e373 2018-03-17 stsp free(pack->path_packfile);
828 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
829 4589e373 2018-03-17 stsp pack->filesize = 0;
830 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
831 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
832 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
836 57160834 2022-05-31 stsp * Leave accumfd and basefd alone. They are managed by the
837 57160834 2022-05-31 stsp * repository layer and can be reused.
840 c7fe698a 2018-01-23 stsp return err;
843 668a20f6 2020-03-18 stsp const struct got_error *
844 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
845 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
847 a487c1d0 2018-01-14 stsp uint8_t t = 0;
848 a487c1d0 2018-01-14 stsp uint64_t s = 0;
849 a487c1d0 2018-01-14 stsp uint8_t sizeN;
850 d7464085 2018-07-09 stsp size_t mapoff = 0;
855 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
856 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
858 d7464085 2018-07-09 stsp if (pack->map) {
859 ad4cc361 2022-10-27 op if (offset > SIZE_MAX) {
860 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
861 ad4cc361 2022-10-27 op "offset %lld overflows size_t",
862 ad4cc361 2022-10-27 op (long long)offset);
865 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
867 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
868 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
872 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
874 09ee8ded 2022-10-20 stsp return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
875 e082ed67 2022-10-24 naddy "packfile offset %lld", (long long)offset);
877 d7464085 2018-07-09 stsp if (pack->map) {
878 3976db15 2022-01-10 stsp if (mapoff + sizeof(sizeN) >= pack->filesize)
879 3976db15 2022-01-10 stsp return got_error(GOT_ERR_BAD_PACKFILE);
880 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
881 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
883 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
885 638f9024 2019-05-13 stsp return got_error_from_errno("read");
886 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
887 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
889 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
891 a1fd68d8 2018-01-12 stsp if (i == 0) {
892 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
893 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
894 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
896 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
897 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
900 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
904 a487c1d0 2018-01-14 stsp return NULL;
907 a1fd68d8 2018-01-12 stsp static const struct got_error *
908 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
909 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
911 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
912 6ccb713b 2018-01-19 stsp if (*obj == NULL)
913 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
915 6ccb713b 2018-01-19 stsp (*obj)->type = type;
916 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
917 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
918 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
919 6ccb713b 2018-01-19 stsp (*obj)->size = size;
920 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
921 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
923 b107e67f 2018-01-19 stsp return NULL;
926 b107e67f 2018-01-19 stsp static const struct got_error *
927 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
928 d7464085 2018-07-09 stsp off_t delta_offset)
930 b107e67f 2018-01-19 stsp int64_t o = 0;
931 b107e67f 2018-01-19 stsp uint8_t offN;
934 a0de39f3 2019-08-09 stsp *offset = 0;
938 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
940 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
942 d7464085 2018-07-09 stsp if (pack->map) {
943 d7464085 2018-07-09 stsp size_t mapoff;
945 ad4cc361 2022-10-27 op if (delta_offset + *len > SIZE_MAX) {
946 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
947 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
948 ad4cc361 2022-10-27 op (long long)delta_offset + *len);
951 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
952 3976db15 2022-01-10 stsp if (mapoff + sizeof(offN) >= pack->filesize)
953 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
954 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
957 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
959 638f9024 2019-05-13 stsp return got_error_from_errno("read");
960 d7464085 2018-07-09 stsp if (n != sizeof(offN))
961 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
963 d7464085 2018-07-09 stsp *len += sizeof(offN);
965 b107e67f 2018-01-19 stsp if (i == 0)
966 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
970 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
973 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
975 b107e67f 2018-01-19 stsp *offset = o;
976 6ccb713b 2018-01-19 stsp return NULL;
979 668a20f6 2020-03-18 stsp const struct got_error *
980 c0df5966 2021-12-31 stsp got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
981 24d916d2 2022-10-24 op struct got_pack *pack, off_t offset, size_t tslen)
983 96f5e8b3 2018-01-23 stsp const struct got_error *err;
984 b107e67f 2018-01-19 stsp int64_t negoffset;
985 b107e67f 2018-01-19 stsp size_t negofflen;
989 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
990 d7464085 2018-07-09 stsp offset + tslen);
992 b107e67f 2018-01-19 stsp return err;
994 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
995 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
996 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
997 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
999 d7464085 2018-07-09 stsp *len = negofflen;
1000 96f5e8b3 2018-01-23 stsp return NULL;
1003 0e22967e 2018-02-11 stsp static const struct got_error *
1004 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1005 2d9e6abf 2022-05-04 stsp size_t *delta_compressed_len, size_t delta_data_offset,
1006 2d9e6abf 2022-05-04 stsp struct got_pack *pack)
1008 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
1009 2d9e6abf 2022-05-04 stsp size_t consumed = 0;
1011 42c69117 2019-11-10 stsp if (pack->map) {
1012 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
1013 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1014 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1015 2d9e6abf 2022-05-04 stsp &consumed, NULL, pack->map, delta_data_offset,
1016 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
1018 2d9e6abf 2022-05-04 stsp return err;
1020 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1021 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
1022 2d9e6abf 2022-05-04 stsp err = got_inflate_to_mem_fd(delta_buf, delta_len,
1023 2d9e6abf 2022-05-04 stsp &consumed, NULL, 0, pack->fd);
1025 2d9e6abf 2022-05-04 stsp return err;
1028 2d9e6abf 2022-05-04 stsp if (delta_compressed_len)
1029 2d9e6abf 2022-05-04 stsp *delta_compressed_len = consumed;
1031 2d9e6abf 2022-05-04 stsp return NULL;
1034 96f5e8b3 2018-01-23 stsp static const struct got_error *
1035 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1036 e5792992 2022-10-28 op int delta_type, size_t delta_size, off_t delta_data_offset)
1038 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
1040 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1041 42c69117 2019-11-10 stsp delta_data_offset);
1042 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
1043 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
1044 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
1045 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
1046 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1047 bdd2fbb3 2018-02-11 stsp return NULL;
1050 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1051 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1052 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1053 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1055 a3500804 2018-01-23 stsp const struct got_error *err;
1056 c3703302 2018-01-23 stsp off_t base_offset;
1057 c3703302 2018-01-23 stsp uint8_t base_type;
1058 c3703302 2018-01-23 stsp uint64_t base_size;
1059 c3703302 2018-01-23 stsp size_t base_tslen;
1060 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1061 42c69117 2019-11-10 stsp size_t consumed;
1063 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1064 d7464085 2018-07-09 stsp delta_offset, tslen);
1066 a3500804 2018-01-23 stsp return err;
1068 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
1069 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
1070 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1072 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1073 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1074 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
1075 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1078 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1079 e5792992 2022-10-28 op delta_data_offset);
1081 1a35c1bc 2019-05-22 stsp return err;
1083 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
1084 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1085 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1087 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1088 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1090 1a35c1bc 2019-05-22 stsp return err;
1092 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1093 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
1096 c4330eff 2021-06-22 stsp const struct got_error *
1097 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1098 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
1100 c4330eff 2021-06-22 stsp if (pack->map) {
1101 ad4cc361 2022-10-27 op size_t mapoff;
1103 ad4cc361 2022-10-27 op if (delta_offset + tslen > SIZE_MAX) {
1104 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
1105 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
1106 ad4cc361 2022-10-27 op (long long)delta_offset + tslen);
1109 ad4cc361 2022-10-27 op mapoff = delta_offset + tslen;
1110 3976db15 2022-01-10 stsp if (mapoff + sizeof(*id) >= pack->filesize)
1111 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1112 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
1114 c4330eff 2021-06-22 stsp ssize_t n;
1115 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1116 c4330eff 2021-06-22 stsp if (n < 0)
1117 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1118 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1119 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1122 c4330eff 2021-06-22 stsp return NULL;
1125 c3703302 2018-01-23 stsp static const struct got_error *
1126 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1127 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1128 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1130 6b9c9673 2018-01-23 stsp const struct got_error *err;
1131 6b9c9673 2018-01-23 stsp struct got_object_id id;
1133 6b9c9673 2018-01-23 stsp off_t base_offset;
1134 6b9c9673 2018-01-23 stsp uint8_t base_type;
1135 6b9c9673 2018-01-23 stsp uint64_t base_size;
1136 6b9c9673 2018-01-23 stsp size_t base_tslen;
1137 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1139 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1140 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1142 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1144 c4330eff 2021-06-22 stsp return err;
1145 d7464085 2018-07-09 stsp if (pack->map) {
1146 c4330eff 2021-06-22 stsp delta_data_offset = delta_offset + tslen + sizeof(id);
1148 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1149 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1150 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1153 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1154 e5792992 2022-10-28 op delta_data_offset);
1156 1a35c1bc 2019-05-22 stsp return err;
1158 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1159 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1160 1a35c1bc 2019-05-22 stsp if (idx == -1)
1161 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1163 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1164 04aed155 2022-07-24 naddy if (base_offset == -1)
1165 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1167 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1168 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1170 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1171 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1173 1a35c1bc 2019-05-22 stsp return err;
1175 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1176 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1179 668a20f6 2020-03-18 stsp const struct got_error *
1180 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1181 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1182 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1184 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1186 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1187 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1189 c3703302 2018-01-23 stsp switch (delta_type) {
1190 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1191 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1192 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1193 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1194 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1195 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1196 42c69117 2019-11-10 stsp delta_size, 0);
1198 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1199 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1200 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1202 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1203 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1204 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1207 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1210 96f5e8b3 2018-01-23 stsp return err;
1213 96f5e8b3 2018-01-23 stsp static const struct got_error *
1214 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1215 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1216 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1218 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1219 96f5e8b3 2018-01-23 stsp int resolved_type;
1221 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1222 b107e67f 2018-01-19 stsp if (*obj == NULL)
1223 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1225 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1226 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1227 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1228 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1229 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1231 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1232 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1233 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1234 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1236 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1237 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1238 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1240 96f5e8b3 2018-01-23 stsp goto done;
1242 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1244 96f5e8b3 2018-01-23 stsp goto done;
1245 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1247 96f5e8b3 2018-01-23 stsp if (err) {
1248 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1249 96f5e8b3 2018-01-23 stsp *obj = NULL;
1251 96f5e8b3 2018-01-23 stsp return err;
1254 2090a03d 2018-09-09 stsp const struct got_error *
1255 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1256 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1258 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1259 a1fd68d8 2018-01-12 stsp off_t offset;
1260 6c00b545 2018-01-17 stsp uint8_t type;
1261 6c00b545 2018-01-17 stsp uint64_t size;
1262 3ee5fc21 2018-01-17 stsp size_t tslen;
1264 6c00b545 2018-01-17 stsp *obj = NULL;
1266 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1267 04aed155 2022-07-24 naddy if (offset == -1)
1268 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1270 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1271 668a20f6 2020-03-18 stsp pack, offset);
1273 35c73765 2018-09-09 stsp return err;
1275 6c00b545 2018-01-17 stsp switch (type) {
1276 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1277 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1278 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1279 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1280 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1281 5f25cc85 2019-11-26 stsp size, idx);
1283 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1284 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1285 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1286 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1289 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1293 3ee5fc21 2018-01-17 stsp return err;
1296 d582f26c 2020-03-18 stsp const struct got_error *
1297 c0df5966 2021-12-31 stsp got_pack_get_delta_chain_max_size(uint64_t *max_size,
1298 c0df5966 2021-12-31 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1300 22484865 2018-03-13 stsp struct got_delta *delta;
1301 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1303 22484865 2018-03-13 stsp *max_size = 0;
1304 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1305 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1306 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1307 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1308 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1309 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1310 22484865 2018-03-13 stsp const struct got_error *err;
1311 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1312 42c69117 2019-11-10 stsp size_t delta_len;
1313 ab2f42e7 2019-11-10 stsp int cached = 1;
1315 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1316 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1317 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1319 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1320 ab2f42e7 2019-11-10 stsp cached = 0;
1321 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1322 2d9e6abf 2022-05-04 stsp NULL, delta->data_offset, pack);
1324 ab2f42e7 2019-11-10 stsp return err;
1326 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1327 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1328 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1329 ab2f42e7 2019-11-10 stsp if (err == NULL)
1330 ab2f42e7 2019-11-10 stsp cached = 1;
1331 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1332 ab2f42e7 2019-11-10 stsp free(delta_buf);
1333 ab2f42e7 2019-11-10 stsp return err;
1336 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1337 42c69117 2019-11-10 stsp delta_buf, delta_len);
1338 ab2f42e7 2019-11-10 stsp if (!cached)
1339 ab2f42e7 2019-11-10 stsp free(delta_buf);
1341 22484865 2018-03-13 stsp return err;
1343 22484865 2018-03-13 stsp base_size = delta->size;
1344 22484865 2018-03-13 stsp if (base_size > *max_size)
1345 22484865 2018-03-13 stsp *max_size = base_size;
1346 22484865 2018-03-13 stsp if (result_size > *max_size)
1347 22484865 2018-03-13 stsp *max_size = result_size;
1350 9feb4ff2 2018-03-14 stsp return NULL;
1353 ac544f8c 2019-01-13 stsp const struct got_error *
1354 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1355 42c69117 2019-11-10 stsp struct got_pack *pack)
1357 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1358 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1360 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1363 668a20f6 2020-03-18 stsp const struct got_error *
1364 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1365 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1366 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1368 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1369 6691714a 2018-01-23 stsp struct got_delta *delta;
1370 aabb25f8 2022-10-17 stsp uint8_t *base_buf = NULL, *accum_buf = NULL;
1371 6395114c 2022-05-31 stsp size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1372 6395114c 2022-05-31 stsp /* We process small enough files entirely in memory for speed. */
1373 6395114c 2022-05-31 stsp const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1374 6395114c 2022-05-31 stsp uint64_t max_size = 0;
1375 6691714a 2018-01-23 stsp int n = 0;
1377 b29656e2 2018-03-16 stsp *result_size = 0;
1379 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1380 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1382 6395114c 2022-05-31 stsp if (fseeko(base_file, 0L, SEEK_SET) == -1)
1383 6395114c 2022-05-31 stsp return got_error_from_errno("fseeko");
1384 6395114c 2022-05-31 stsp if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1385 6395114c 2022-05-31 stsp return got_error_from_errno("fseeko");
1387 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1388 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1389 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1390 6395114c 2022-05-31 stsp uint64_t base_size, result_size = 0;
1391 ab2f42e7 2019-11-10 stsp int cached = 1;
1392 a6b158cc 2018-02-11 stsp if (n == 0) {
1393 34fca9c3 2018-11-11 stsp size_t mapoff;
1394 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1396 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1397 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1398 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1399 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1400 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1401 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1402 72eb3431 2018-04-01 stsp goto done;
1405 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1406 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1407 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1408 0c048b15 2018-04-27 stsp goto done;
1410 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1411 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1413 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1414 d7464085 2018-07-09 stsp goto done;
1417 6395114c 2022-05-31 stsp if (delta->size > max_size)
1418 6395114c 2022-05-31 stsp max_size = delta->size;
1419 6395114c 2022-05-31 stsp if (max_size > max_bufsize) {
1420 d7464085 2018-07-09 stsp if (pack->map) {
1421 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1422 ad4cc361 2022-10-27 op return got_error_fmt(
1423 ad4cc361 2022-10-27 op GOT_ERR_RANGE,
1424 ad4cc361 2022-10-27 op "delta offset %lld "
1425 ad4cc361 2022-10-27 op "overflows size_t",
1427 ad4cc361 2022-10-27 op delta_data_offset);
1430 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1431 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1432 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1433 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1434 4788f1ce 2020-03-18 stsp base_file);
1436 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1437 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1438 4788f1ce 2020-03-18 stsp base_file);
1440 6395114c 2022-05-31 stsp accum_buf = malloc(max_size);
1441 6395114c 2022-05-31 stsp if (accum_buf == NULL) {
1442 6395114c 2022-05-31 stsp err = got_error_from_errno("malloc");
1443 6395114c 2022-05-31 stsp goto done;
1445 6395114c 2022-05-31 stsp accum_bufsz = max_size;
1446 d7464085 2018-07-09 stsp if (pack->map) {
1447 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1448 ad4cc361 2022-10-27 op return got_error_fmt(
1449 ad4cc361 2022-10-27 op GOT_ERR_RANGE,
1450 ad4cc361 2022-10-27 op "delta offset %lld "
1451 ad4cc361 2022-10-27 op "overflows size_t",
1453 ad4cc361 2022-10-27 op delta_data_offset);
1456 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1457 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1458 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1459 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1460 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1462 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1463 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1464 55fdd257 2020-03-18 stsp pack->fd);
1467 a6b158cc 2018-02-11 stsp goto done;
1469 6395114c 2022-05-31 stsp if (base_buf == NULL)
1470 8628c62d 2018-03-15 stsp rewind(base_file);
1474 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1475 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1476 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1478 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1479 ab2f42e7 2019-11-10 stsp cached = 0;
1480 2d9e6abf 2022-05-04 stsp err = read_delta_data(&delta_buf, &delta_len, NULL,
1481 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1483 ab2f42e7 2019-11-10 stsp goto done;
1485 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1486 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1487 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1488 ab2f42e7 2019-11-10 stsp if (err == NULL)
1489 ab2f42e7 2019-11-10 stsp cached = 1;
1490 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1491 ab2f42e7 2019-11-10 stsp free(delta_buf);
1492 ab2f42e7 2019-11-10 stsp goto done;
1496 6395114c 2022-05-31 stsp err = got_delta_get_sizes(&base_size, &result_size,
1497 6395114c 2022-05-31 stsp delta_buf, delta_len);
1498 e62fc520 2022-11-08 stsp if (err) {
1499 e62fc520 2022-11-08 stsp if (!cached)
1500 e62fc520 2022-11-08 stsp free(delta_buf);
1501 6395114c 2022-05-31 stsp goto done;
1503 6395114c 2022-05-31 stsp if (base_size > max_size)
1504 6395114c 2022-05-31 stsp max_size = base_size;
1505 6395114c 2022-05-31 stsp if (result_size > max_size)
1506 6395114c 2022-05-31 stsp max_size = result_size;
1508 6395114c 2022-05-31 stsp if (base_buf && max_size > max_bufsize) {
1509 6395114c 2022-05-31 stsp /* Switch from buffers to temporary files. */
1510 6395114c 2022-05-31 stsp size_t w = fwrite(base_buf, 1, base_bufsz,
1511 6395114c 2022-05-31 stsp base_file);
1512 6395114c 2022-05-31 stsp if (w != base_bufsz) {
1513 6395114c 2022-05-31 stsp err = got_ferror(outfile, GOT_ERR_IO);
1514 e62fc520 2022-11-08 stsp if (!cached)
1515 e62fc520 2022-11-08 stsp free(delta_buf);
1516 6395114c 2022-05-31 stsp goto done;
1518 6395114c 2022-05-31 stsp free(base_buf);
1519 6395114c 2022-05-31 stsp base_buf = NULL;
1520 6395114c 2022-05-31 stsp free(accum_buf);
1521 6395114c 2022-05-31 stsp accum_buf = NULL;
1524 6395114c 2022-05-31 stsp if (base_buf && max_size > base_bufsz) {
1525 6395114c 2022-05-31 stsp uint8_t *p = realloc(base_buf, max_size);
1526 6395114c 2022-05-31 stsp if (p == NULL) {
1527 6395114c 2022-05-31 stsp err = got_error_from_errno("realloc");
1528 e62fc520 2022-11-08 stsp if (!cached)
1529 e62fc520 2022-11-08 stsp free(delta_buf);
1530 6395114c 2022-05-31 stsp goto done;
1532 6395114c 2022-05-31 stsp base_buf = p;
1533 6395114c 2022-05-31 stsp base_bufsz = max_size;
1536 6395114c 2022-05-31 stsp if (accum_buf && max_size > accum_bufsz) {
1537 6395114c 2022-05-31 stsp uint8_t *p = realloc(accum_buf, max_size);
1538 6395114c 2022-05-31 stsp if (p == NULL) {
1539 6395114c 2022-05-31 stsp err = got_error_from_errno("realloc");
1540 e62fc520 2022-11-08 stsp if (!cached)
1541 e62fc520 2022-11-08 stsp free(delta_buf);
1542 6395114c 2022-05-31 stsp goto done;
1544 6395114c 2022-05-31 stsp accum_buf = p;
1545 6395114c 2022-05-31 stsp accum_bufsz = max_size;
1548 8628c62d 2018-03-15 stsp if (base_buf) {
1549 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1550 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1551 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1554 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1555 42c69117 2019-11-10 stsp delta_len,
1556 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1557 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1558 b29656e2 2018-03-16 stsp &accum_size);
1560 ab2f42e7 2019-11-10 stsp if (!cached)
1561 ab2f42e7 2019-11-10 stsp free(delta_buf);
1563 6691714a 2018-01-23 stsp goto done;
1565 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1566 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1567 8628c62d 2018-03-15 stsp if (base_buf) {
1568 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1569 6395114c 2022-05-31 stsp size_t tmp_size = accum_bufsz;
1570 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1571 6395114c 2022-05-31 stsp accum_bufsz = base_bufsz;
1572 8628c62d 2018-03-15 stsp base_buf = tmp;
1573 6395114c 2022-05-31 stsp base_bufsz = tmp_size;
1575 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1576 8628c62d 2018-03-15 stsp accum_file = base_file;
1577 8628c62d 2018-03-15 stsp base_file = tmp;
1578 8628c62d 2018-03-15 stsp rewind(base_file);
1579 8628c62d 2018-03-15 stsp rewind(accum_file);
1585 8628c62d 2018-03-15 stsp free(base_buf);
1586 8628c62d 2018-03-15 stsp if (accum_buf) {
1587 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1588 8628c62d 2018-03-15 stsp free(accum_buf);
1589 8628c62d 2018-03-15 stsp if (len != accum_size)
1590 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1592 6691714a 2018-01-23 stsp rewind(outfile);
1593 b29656e2 2018-03-16 stsp if (err == NULL)
1594 b29656e2 2018-03-16 stsp *result_size = accum_size;
1595 e0ab43e7 2018-03-16 stsp return err;
1598 668a20f6 2020-03-18 stsp const struct got_error *
1599 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1600 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1602 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1603 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1604 aabb25f8 2022-10-17 stsp uint8_t *base_buf = NULL, *accum_buf = NULL;
1605 f18c433a 2022-05-31 stsp size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1606 f18c433a 2022-05-31 stsp uint64_t max_size = 0;
1607 e0ab43e7 2018-03-16 stsp int n = 0;
1609 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1610 e0ab43e7 2018-03-16 stsp *outlen = 0;
1612 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1613 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1615 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1616 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1617 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1618 f18c433a 2022-05-31 stsp uint64_t base_size, result_size = 0;
1619 ab2f42e7 2019-11-10 stsp int cached = 1;
1620 e0ab43e7 2018-03-16 stsp if (n == 0) {
1621 ad4cc361 2022-10-27 op off_t delta_data_offset;
1623 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1624 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1625 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1626 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1627 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1628 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1629 72eb3431 2018-04-01 stsp goto done;
1632 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1633 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1634 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1635 0c048b15 2018-04-27 stsp goto done;
1638 f18c433a 2022-05-31 stsp if (delta->size > max_size)
1639 f18c433a 2022-05-31 stsp max_size = delta->size;
1641 d7464085 2018-07-09 stsp if (pack->map) {
1642 ad4cc361 2022-10-27 op size_t mapoff;
1644 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1645 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1646 ad4cc361 2022-10-27 op "delta %lld offset would "
1647 ad4cc361 2022-10-27 op "overflow size_t",
1648 ad4cc361 2022-10-27 op (long long)delta_data_offset);
1651 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1652 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1653 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1654 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1656 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1658 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1659 d7464085 2018-07-09 stsp goto done;
1661 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1662 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1663 55fdd257 2020-03-18 stsp pack->fd);
1666 d7464085 2018-07-09 stsp goto done;
1671 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1672 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1673 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1675 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1676 ab2f42e7 2019-11-10 stsp cached = 0;
1677 2d9e6abf 2022-05-04 stsp err = read_delta_data(&delta_buf, &delta_len, NULL,
1678 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1680 ab2f42e7 2019-11-10 stsp goto done;
1682 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1683 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1684 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1685 ab2f42e7 2019-11-10 stsp if (err == NULL)
1686 ab2f42e7 2019-11-10 stsp cached = 1;
1687 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1688 ab2f42e7 2019-11-10 stsp free(delta_buf);
1689 ab2f42e7 2019-11-10 stsp goto done;
1693 f18c433a 2022-05-31 stsp err = got_delta_get_sizes(&base_size, &result_size,
1694 f18c433a 2022-05-31 stsp delta_buf, delta_len);
1695 e62fc520 2022-11-08 stsp if (err) {
1696 e62fc520 2022-11-08 stsp if (!cached)
1697 e62fc520 2022-11-08 stsp free(delta_buf);
1698 f18c433a 2022-05-31 stsp goto done;
1700 f18c433a 2022-05-31 stsp if (base_size > max_size)
1701 f18c433a 2022-05-31 stsp max_size = base_size;
1702 f18c433a 2022-05-31 stsp if (result_size > max_size)
1703 f18c433a 2022-05-31 stsp max_size = result_size;
1705 f18c433a 2022-05-31 stsp if (max_size > base_bufsz) {
1706 f18c433a 2022-05-31 stsp uint8_t *p = realloc(base_buf, max_size);
1707 f18c433a 2022-05-31 stsp if (p == NULL) {
1708 f18c433a 2022-05-31 stsp err = got_error_from_errno("realloc");
1709 e62fc520 2022-11-08 stsp if (!cached)
1710 e62fc520 2022-11-08 stsp free(delta_buf);
1711 f18c433a 2022-05-31 stsp goto done;
1713 f18c433a 2022-05-31 stsp base_buf = p;
1714 f18c433a 2022-05-31 stsp base_bufsz = max_size;
1717 f18c433a 2022-05-31 stsp if (max_size > accum_bufsz) {
1718 f18c433a 2022-05-31 stsp uint8_t *p = realloc(accum_buf, max_size);
1719 f18c433a 2022-05-31 stsp if (p == NULL) {
1720 f18c433a 2022-05-31 stsp err = got_error_from_errno("realloc");
1721 e62fc520 2022-11-08 stsp if (!cached)
1722 e62fc520 2022-11-08 stsp free(delta_buf);
1723 f18c433a 2022-05-31 stsp goto done;
1725 f18c433a 2022-05-31 stsp accum_buf = p;
1726 f18c433a 2022-05-31 stsp accum_bufsz = max_size;
1729 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1730 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1731 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1732 ab2f42e7 2019-11-10 stsp if (!cached)
1733 ab2f42e7 2019-11-10 stsp free(delta_buf);
1736 e0ab43e7 2018-03-16 stsp goto done;
1738 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1739 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1740 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1741 f18c433a 2022-05-31 stsp size_t tmp_size = accum_bufsz;
1742 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1743 f18c433a 2022-05-31 stsp accum_bufsz = base_bufsz;
1744 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1745 f18c433a 2022-05-31 stsp base_bufsz = tmp_size;
1750 e0ab43e7 2018-03-16 stsp free(base_buf);
1751 e0ab43e7 2018-03-16 stsp if (err) {
1752 e0ab43e7 2018-03-16 stsp free(accum_buf);
1753 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1754 e0ab43e7 2018-03-16 stsp *outlen = 0;
1756 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1757 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1759 efd2a263 2018-01-19 stsp return err;
1762 3ee5fc21 2018-01-17 stsp const struct got_error *
1763 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1764 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1766 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1768 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1769 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1771 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1772 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1773 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1775 d7464085 2018-07-09 stsp if (pack->map) {
1776 ad4cc361 2022-10-27 op size_t mapoff;
1778 ad4cc361 2022-10-27 op if (obj->pack_offset > SIZE_MAX) {
1779 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1780 ad4cc361 2022-10-27 op "pack offset %lld would overflow size_t",
1781 ad4cc361 2022-10-27 op (long long)obj->pack_offset);
1784 ad4cc361 2022-10-27 op mapoff = obj->pack_offset;
1785 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1786 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1789 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1790 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1791 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1792 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1795 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1796 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1798 a1fd68d8 2018-01-12 stsp return err;
1801 e0ab43e7 2018-03-16 stsp const struct got_error *
1802 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1803 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1805 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1807 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1808 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1810 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1811 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1812 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1813 d7464085 2018-07-09 stsp if (pack->map) {
1814 ad4cc361 2022-10-27 op size_t mapoff;
1816 ad4cc361 2022-10-27 op if (obj->pack_offset > SIZE_MAX) {
1817 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1818 ad4cc361 2022-10-27 op "pack offset %lld would overflow size_t",
1819 ad4cc361 2022-10-27 op (long long)obj->pack_offset);
1822 ad4cc361 2022-10-27 op mapoff = obj->pack_offset;
1823 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1824 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1826 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1827 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1828 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1829 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1832 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1835 e0ab43e7 2018-03-16 stsp return err;
1838 2d9e6abf 2022-05-04 stsp static const struct got_error *
1839 2d9e6abf 2022-05-04 stsp read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1840 2d9e6abf 2022-05-04 stsp size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1841 2d9e6abf 2022-05-04 stsp off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1843 2d9e6abf 2022-05-04 stsp const struct got_error *err = NULL;
1845 2d9e6abf 2022-05-04 stsp /* Validate decompression and obtain the decompressed size. */
1846 2d9e6abf 2022-05-04 stsp err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1847 2d9e6abf 2022-05-04 stsp delta_data_offset, pack);
1849 2d9e6abf 2022-05-04 stsp return err;
1851 2d9e6abf 2022-05-04 stsp /* Read delta base/result sizes from head of delta stream. */
1852 2d9e6abf 2022-05-04 stsp err = got_delta_get_sizes(base_size, result_size,
1853 2d9e6abf 2022-05-04 stsp *delta_buf, *delta_len);
1855 2d9e6abf 2022-05-04 stsp goto done;
1857 2d9e6abf 2022-05-04 stsp /* Discard decompressed delta and read it again in compressed form. */
1858 2d9e6abf 2022-05-04 stsp free(*delta_buf);
1859 2d9e6abf 2022-05-04 stsp *delta_buf = malloc(*delta_len_compressed);
1860 2d9e6abf 2022-05-04 stsp if (*delta_buf == NULL) {
1861 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("malloc");
1862 2d9e6abf 2022-05-04 stsp goto done;
1864 2d9e6abf 2022-05-04 stsp if (pack->map) {
1865 82031ac8 2022-10-24 op if (delta_data_offset >= pack->filesize) {
1866 2d9e6abf 2022-05-04 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1869 2d9e6abf 2022-05-04 stsp memcpy(*delta_buf, pack->map + delta_data_offset,
1870 2d9e6abf 2022-05-04 stsp *delta_len_compressed);
1872 2d9e6abf 2022-05-04 stsp ssize_t n;
1873 2d9e6abf 2022-05-04 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1874 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("lseek");
1875 2d9e6abf 2022-05-04 stsp goto done;
1877 2d9e6abf 2022-05-04 stsp n = read(pack->fd, *delta_buf, *delta_len_compressed);
1878 2d9e6abf 2022-05-04 stsp if (n < 0) {
1879 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("read");
1880 2d9e6abf 2022-05-04 stsp goto done;
1881 2d9e6abf 2022-05-04 stsp } else if (n != *delta_len_compressed) {
1882 2d9e6abf 2022-05-04 stsp err = got_error(GOT_ERR_IO);
1883 2d9e6abf 2022-05-04 stsp goto done;
1887 2d9e6abf 2022-05-04 stsp if (err) {
1888 2d9e6abf 2022-05-04 stsp free(*delta_buf);
1889 2d9e6abf 2022-05-04 stsp *delta_buf = NULL;
1890 2d9e6abf 2022-05-04 stsp *delta_len = 0;
1891 2d9e6abf 2022-05-04 stsp *delta_len_compressed = 0;
1892 2d9e6abf 2022-05-04 stsp *base_size = 0;
1893 2d9e6abf 2022-05-04 stsp *result_size = 0;
1895 2d9e6abf 2022-05-04 stsp return err;
1898 67fd6849 2022-02-13 stsp const struct got_error *
1899 67fd6849 2022-02-13 stsp got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1900 24b7de1c 2022-12-03 stsp size_t *delta_compressed_size, off_t *delta_offset,
1901 24b7de1c 2022-12-03 stsp off_t *delta_data_offset, off_t *base_offset,
1902 2d9e6abf 2022-05-04 stsp struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1903 2d9e6abf 2022-05-04 stsp struct got_pack *pack, struct got_packidx *packidx, int idx)
1905 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
1906 67fd6849 2022-02-13 stsp off_t offset;
1907 67fd6849 2022-02-13 stsp uint8_t type;
1908 67fd6849 2022-02-13 stsp uint64_t size;
1909 67fd6849 2022-02-13 stsp size_t tslen, delta_hdrlen;
1911 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1912 67fd6849 2022-02-13 stsp *delta_size = 0;
1913 2d9e6abf 2022-05-04 stsp *delta_compressed_size = 0;
1914 67fd6849 2022-02-13 stsp *delta_offset = 0;
1915 24b7de1c 2022-12-03 stsp *delta_data_offset = 0;
1916 67fd6849 2022-02-13 stsp *base_offset = 0;
1917 67fd6849 2022-02-13 stsp *base_size = 0;
1918 67fd6849 2022-02-13 stsp *result_size = 0;
1920 67fd6849 2022-02-13 stsp offset = got_packidx_get_object_offset(packidx, idx);
1921 04aed155 2022-07-24 naddy if (offset == -1)
1922 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1924 67fd6849 2022-02-13 stsp if (offset >= pack->filesize)
1925 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1927 67fd6849 2022-02-13 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1928 67fd6849 2022-02-13 stsp pack, offset);
1930 67fd6849 2022-02-13 stsp return err;
1932 67fd6849 2022-02-13 stsp if (tslen + size < tslen || offset + size < size ||
1933 67fd6849 2022-02-13 stsp tslen + offset < tslen)
1934 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1936 67fd6849 2022-02-13 stsp switch (type) {
1937 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1938 67fd6849 2022-02-13 stsp err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1939 67fd6849 2022-02-13 stsp pack, offset, tslen);
1941 67fd6849 2022-02-13 stsp return err;
1943 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_REF_DELTA:
1944 67fd6849 2022-02-13 stsp err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1946 67fd6849 2022-02-13 stsp return err;
1947 67fd6849 2022-02-13 stsp delta_hdrlen = SHA1_DIGEST_LENGTH;
1950 67fd6849 2022-02-13 stsp return got_error_fmt(GOT_ERR_OBJ_TYPE,
1951 04aed155 2022-07-24 naddy "non-delta object type %d found at offset %lld",
1952 04aed155 2022-07-24 naddy type, (long long)offset);
1955 67fd6849 2022-02-13 stsp if (tslen + delta_hdrlen < delta_hdrlen ||
1956 67fd6849 2022-02-13 stsp offset + delta_hdrlen < delta_hdrlen)
1957 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_DELTA);
1959 24b7de1c 2022-12-03 stsp *delta_data_offset = offset + tslen + delta_hdrlen;
1960 2d9e6abf 2022-05-04 stsp err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1961 24b7de1c 2022-12-03 stsp base_size, result_size, *delta_data_offset, pack, packidx);
1963 67fd6849 2022-02-13 stsp return err;
1965 67fd6849 2022-02-13 stsp if (*delta_size != size) {
1966 67fd6849 2022-02-13 stsp err = got_error(GOT_ERR_BAD_DELTA);
1967 67fd6849 2022-02-13 stsp goto done;
1970 67fd6849 2022-02-13 stsp *delta_offset = offset;
1972 67fd6849 2022-02-13 stsp if (err) {
1973 67fd6849 2022-02-13 stsp free(*delta_buf);
1974 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1975 2d9e6abf 2022-05-04 stsp *delta_size = 0;
1976 2d9e6abf 2022-05-04 stsp *delta_compressed_size = 0;
1977 2d9e6abf 2022-05-04 stsp *delta_offset = 0;
1978 2d9e6abf 2022-05-04 stsp *base_offset = 0;
1979 2d9e6abf 2022-05-04 stsp *base_size = 0;
1980 2d9e6abf 2022-05-04 stsp *result_size = 0;
1982 67fd6849 2022-02-13 stsp return err;