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 5822e79e 2023-02-23 op #include <sha2.h>
34 0a0a3048 2018-01-10 stsp #include <endian.h>
35 81a12da5 2020-09-09 naddy #include <unistd.h>
36 a1fd68d8 2018-01-12 stsp #include <zlib.h>
37 876c234b 2018-09-10 stsp #include <imsg.h>
39 0a0a3048 2018-01-10 stsp #include "got_error.h"
40 a1fd68d8 2018-01-12 stsp #include "got_object.h"
41 324d37e7 2019-05-11 stsp #include "got_path.h"
43 53bf0b54 2023-02-23 op #include "got_lib_hash.h"
44 718b3ab0 2018-03-17 stsp #include "got_lib_delta.h"
45 ab2f42e7 2019-11-10 stsp #include "got_lib_delta_cache.h"
46 63581804 2018-07-09 stsp #include "got_lib_inflate.h"
47 718b3ab0 2018-03-17 stsp #include "got_lib_object.h"
48 dd88155e 2019-06-29 stsp #include "got_lib_object_parse.h"
49 876c234b 2018-09-10 stsp #include "got_lib_privsep.h"
50 15a94983 2018-12-23 stsp #include "got_lib_pack.h"
52 79b11c62 2018-03-09 stsp #ifndef nitems
53 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
57 a1fd68d8 2018-01-12 stsp #define MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
60 0a0a3048 2018-01-10 stsp static const struct got_error *
61 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
65 0a0a3048 2018-01-10 stsp for (i = 0; i < 0xff - 1; i++) {
66 a1fd68d8 2018-01-12 stsp if (be32toh(fanout_table[i]) > be32toh(fanout_table[i + 1]))
67 0a0a3048 2018-01-10 stsp return got_error(GOT_ERR_BAD_PACKIDX);
70 0a0a3048 2018-01-10 stsp return NULL;
73 1510f469 2018-09-09 stsp const struct got_error *
74 c3564dfa 2021-07-15 stsp got_packidx_init_hdr(struct got_packidx *p, int verify, off_t packfile_size)
76 0a0a3048 2018-01-10 stsp const struct got_error *err = NULL;
77 ae25a666 2023-02-23 op enum got_hash_algorithm algo = GOT_HASH_SHA1;
78 817c5a18 2018-09-09 stsp struct got_packidx_v2_hdr *h;
79 ae25a666 2023-02-23 op struct got_hash ctx;
80 ae25a666 2023-02-23 op uint8_t hash[GOT_HASH_DIGEST_MAXLEN];
81 817c5a18 2018-09-09 stsp size_t nobj, len_fanout, len_ids, offset, remain;
85 ae25a666 2023-02-23 op got_hash_init(&ctx, algo);
87 57b35b75 2018-06-22 stsp h = &p->hdr;
89 57b35b75 2018-06-22 stsp remain = p->len;
91 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->magic)) {
92 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
96 fc79a48d 2018-07-09 stsp h->magic = (uint32_t *)(p->map + offset);
98 fc79a48d 2018-07-09 stsp h->magic = malloc(sizeof(*h->magic));
99 fc79a48d 2018-07-09 stsp if (h->magic == NULL) {
100 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
103 fc79a48d 2018-07-09 stsp n = read(p->fd, h->magic, sizeof(*h->magic));
104 b1317e77 2019-09-22 stsp if (n < 0) {
105 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
107 b1317e77 2019-09-22 stsp } else if (n != sizeof(*h->magic)) {
108 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
112 ac62b712 2021-03-30 stsp if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
113 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
116 57b35b75 2018-06-22 stsp offset += sizeof(*h->magic);
117 57b35b75 2018-06-22 stsp remain -= sizeof(*h->magic);
119 0cb74cf4 2018-07-08 stsp if (verify)
120 ae25a666 2023-02-23 op got_hash_update(&ctx, h->magic, sizeof(*h->magic));
122 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->version)) {
123 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
126 fc79a48d 2018-07-09 stsp if (p->map)
127 fc79a48d 2018-07-09 stsp h->version = (uint32_t *)(p->map + offset);
129 fc79a48d 2018-07-09 stsp h->version = malloc(sizeof(*h->version));
130 fc79a48d 2018-07-09 stsp if (h->version == NULL) {
131 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
134 fc79a48d 2018-07-09 stsp n = read(p->fd, h->version, sizeof(*h->version));
135 c6368c2e 2019-10-11 stsp if (n < 0) {
136 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
138 c6368c2e 2019-10-11 stsp } else if (n != sizeof(*h->version)) {
139 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
143 ac62b712 2021-03-30 stsp if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
144 0a0a3048 2018-01-10 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
147 57b35b75 2018-06-22 stsp offset += sizeof(*h->version);
148 57b35b75 2018-06-22 stsp remain -= sizeof(*h->version);
150 0cb74cf4 2018-07-08 stsp if (verify)
151 ae25a666 2023-02-23 op got_hash_update(&ctx, h->version, sizeof(*h->version));
153 57b35b75 2018-06-22 stsp len_fanout =
154 57b35b75 2018-06-22 stsp sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
155 57b35b75 2018-06-22 stsp if (remain < len_fanout) {
156 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
159 fc79a48d 2018-07-09 stsp if (p->map)
160 fc79a48d 2018-07-09 stsp h->fanout_table = (uint32_t *)(p->map + offset);
162 fc79a48d 2018-07-09 stsp h->fanout_table = malloc(len_fanout);
163 fc79a48d 2018-07-09 stsp if (h->fanout_table == NULL) {
164 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
167 fc79a48d 2018-07-09 stsp n = read(p->fd, h->fanout_table, len_fanout);
168 c6368c2e 2019-10-11 stsp if (n < 0) {
169 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
171 c6368c2e 2019-10-11 stsp } else if (n != len_fanout) {
172 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
176 c8262310 2018-06-04 stsp err = verify_fanout_table(h->fanout_table);
179 0cb74cf4 2018-07-08 stsp if (verify)
180 ae25a666 2023-02-23 op got_hash_update(&ctx, h->fanout_table, len_fanout);
181 57b35b75 2018-06-22 stsp offset += len_fanout;
182 57b35b75 2018-06-22 stsp remain -= len_fanout;
184 78fb0967 2020-09-09 naddy nobj = be32toh(h->fanout_table[0xff]);
185 57b35b75 2018-06-22 stsp len_ids = nobj * sizeof(*h->sorted_ids);
186 57b35b75 2018-06-22 stsp if (len_ids <= nobj || len_ids > remain) {
187 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
190 fc79a48d 2018-07-09 stsp if (p->map)
191 fc79a48d 2018-07-09 stsp h->sorted_ids =
192 fc79a48d 2018-07-09 stsp (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
194 fc79a48d 2018-07-09 stsp h->sorted_ids = malloc(len_ids);
195 fc79a48d 2018-07-09 stsp if (h->sorted_ids == NULL) {
196 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
199 fc79a48d 2018-07-09 stsp n = read(p->fd, h->sorted_ids, len_ids);
201 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
202 faaa1c0f 2018-09-15 stsp else if (n != len_ids) {
203 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
207 0cb74cf4 2018-07-08 stsp if (verify)
208 ae25a666 2023-02-23 op got_hash_update(&ctx, h->sorted_ids, len_ids);
209 57b35b75 2018-06-22 stsp offset += len_ids;
210 57b35b75 2018-06-22 stsp remain -= len_ids;
212 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->crc32)) {
213 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
216 fc79a48d 2018-07-09 stsp if (p->map)
217 fc79a48d 2018-07-09 stsp h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
219 fc79a48d 2018-07-09 stsp h->crc32 = malloc(nobj * sizeof(*h->crc32));
220 fc79a48d 2018-07-09 stsp if (h->crc32 == NULL) {
221 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
224 fc79a48d 2018-07-09 stsp n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
226 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
227 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->crc32)) {
228 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
232 0cb74cf4 2018-07-08 stsp if (verify)
233 ae25a666 2023-02-23 op got_hash_update(&ctx, h->crc32, nobj * sizeof(*h->crc32));
234 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->crc32);
235 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->crc32);
237 57b35b75 2018-06-22 stsp if (remain < nobj * sizeof(*h->offsets)) {
238 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
241 fc79a48d 2018-07-09 stsp if (p->map)
242 fc79a48d 2018-07-09 stsp h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
244 fc79a48d 2018-07-09 stsp h->offsets = malloc(nobj * sizeof(*h->offsets));
245 fc79a48d 2018-07-09 stsp if (h->offsets == NULL) {
246 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
249 fc79a48d 2018-07-09 stsp n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
251 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
252 faaa1c0f 2018-09-15 stsp else if (n != nobj * sizeof(*h->offsets)) {
253 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
257 0cb74cf4 2018-07-08 stsp if (verify)
258 ae25a666 2023-02-23 op got_hash_update(&ctx, h->offsets, nobj * sizeof(*h->offsets));
259 57b35b75 2018-06-22 stsp remain -= nobj * sizeof(*h->offsets);
260 57b35b75 2018-06-22 stsp offset += nobj * sizeof(*h->offsets);
262 0a0a3048 2018-01-10 stsp /* Large file offsets are contained only in files > 2GB. */
263 c3564dfa 2021-07-15 stsp if (verify || packfile_size > 0x7fffffff) {
264 c3564dfa 2021-07-15 stsp for (i = 0; i < nobj; i++) {
265 c3564dfa 2021-07-15 stsp uint32_t o = h->offsets[i];
266 c3564dfa 2021-07-15 stsp if (o & htobe32(GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX))
267 c3564dfa 2021-07-15 stsp p->nlargeobj++;
270 5e6be232 2019-11-08 stsp if (p->nlargeobj == 0)
271 0a0a3048 2018-01-10 stsp goto checksum;
272 c3564dfa 2021-07-15 stsp else if (packfile_size <= 0x7fffffff) {
273 c3564dfa 2021-07-15 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
277 5e6be232 2019-11-08 stsp if (remain < p->nlargeobj * sizeof(*h->large_offsets)) {
278 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
281 fc79a48d 2018-07-09 stsp if (p->map)
282 fc79a48d 2018-07-09 stsp h->large_offsets = (uint64_t *)((uint8_t*)(p->map + offset));
284 5e6be232 2019-11-08 stsp h->large_offsets = malloc(p->nlargeobj *
285 5e6be232 2019-11-08 stsp sizeof(*h->large_offsets));
286 de30857e 2019-08-23 stsp if (h->large_offsets == NULL) {
287 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
290 fc79a48d 2018-07-09 stsp n = read(p->fd, h->large_offsets,
291 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
293 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
294 5e6be232 2019-11-08 stsp else if (n != p->nlargeobj * sizeof(*h->large_offsets)) {
295 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
299 0cb74cf4 2018-07-08 stsp if (verify)
300 ae25a666 2023-02-23 op got_hash_update(&ctx, h->large_offsets,
301 5e6be232 2019-11-08 stsp p->nlargeobj * sizeof(*h->large_offsets));
302 5e6be232 2019-11-08 stsp remain -= p->nlargeobj * sizeof(*h->large_offsets);
303 5e6be232 2019-11-08 stsp offset += p->nlargeobj * sizeof(*h->large_offsets);
306 57b35b75 2018-06-22 stsp if (remain < sizeof(*h->trailer)) {
307 57b35b75 2018-06-22 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
310 fc79a48d 2018-07-09 stsp if (p->map)
311 fc79a48d 2018-07-09 stsp h->trailer =
312 fc79a48d 2018-07-09 stsp (struct got_packidx_trailer *)((uint8_t*)(p->map + offset));
314 fc79a48d 2018-07-09 stsp h->trailer = malloc(sizeof(*h->trailer));
315 fc79a48d 2018-07-09 stsp if (h->trailer == NULL) {
316 638f9024 2019-05-13 stsp err = got_error_from_errno("malloc");
319 fc79a48d 2018-07-09 stsp n = read(p->fd, h->trailer, sizeof(*h->trailer));
321 638f9024 2019-05-13 stsp err = got_error_from_errno("read");
322 faaa1c0f 2018-09-15 stsp else if (n != sizeof(*h->trailer)) {
323 fc79a48d 2018-07-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
327 0cb74cf4 2018-07-08 stsp if (verify) {
328 ae25a666 2023-02-23 op got_hash_update(&ctx, h->trailer->packfile_sha1,
329 ae25a666 2023-02-23 op SHA1_DIGEST_LENGTH);
330 ae25a666 2023-02-23 op got_hash_final(&ctx, hash);
331 ae25a666 2023-02-23 op if (got_hash_cmp(ctx.algo, hash, h->trailer->packidx_sha1)
333 0cb74cf4 2018-07-08 stsp err = got_error(GOT_ERR_PACKIDX_CSUM);
336 817c5a18 2018-09-09 stsp return err;
339 817c5a18 2018-09-09 stsp const struct got_error *
340 6d5a9006 2020-12-16 yzhong got_packidx_open(struct got_packidx **packidx,
341 6d5a9006 2020-12-16 yzhong int dir_fd, const char *relpath, int verify)
343 817c5a18 2018-09-09 stsp const struct got_error *err = NULL;
344 1124fe40 2021-07-07 stsp struct got_packidx *p = NULL;
345 1124fe40 2021-07-07 stsp char *pack_relpath;
346 c3564dfa 2021-07-15 stsp struct stat idx_sb, pack_sb;
348 817c5a18 2018-09-09 stsp *packidx = NULL;
350 1124fe40 2021-07-07 stsp err = got_packidx_get_packfile_path(&pack_relpath, relpath);
352 1124fe40 2021-07-07 stsp return err;
355 1124fe40 2021-07-07 stsp * Ensure that a corresponding pack file exists.
356 1124fe40 2021-07-07 stsp * Some Git repositories have this problem. Git seems to ignore
357 1124fe40 2021-07-07 stsp * the existence of lonely pack index files but we do not.
359 c3564dfa 2021-07-15 stsp if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
360 1124fe40 2021-07-07 stsp if (errno == ENOENT) {
361 1124fe40 2021-07-07 stsp err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
362 1124fe40 2021-07-07 stsp "%s", relpath);
364 1124fe40 2021-07-07 stsp err = got_error_from_errno2("fstatat", pack_relpath);
368 817c5a18 2018-09-09 stsp p = calloc(1, sizeof(*p));
369 1124fe40 2021-07-07 stsp if (p == NULL) {
370 1124fe40 2021-07-07 stsp err = got_error_from_errno("calloc");
374 e7ae0baf 2021-12-31 stsp p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
375 6772cf22 2019-08-28 hiltjo if (p->fd == -1) {
376 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("openat", relpath);
380 c3564dfa 2021-07-15 stsp if (fstat(p->fd, &idx_sb) != 0) {
381 6d5a9006 2020-12-16 yzhong err = got_error_from_errno2("fstat", relpath);
384 c3564dfa 2021-07-15 stsp p->len = idx_sb.st_size;
385 817c5a18 2018-09-09 stsp if (p->len < sizeof(p->hdr)) {
386 817c5a18 2018-09-09 stsp err = got_error(GOT_ERR_BAD_PACKIDX);
390 6d5a9006 2020-12-16 yzhong p->path_packidx = strdup(relpath);
391 817c5a18 2018-09-09 stsp if (p->path_packidx == NULL) {
392 638f9024 2019-05-13 stsp err = got_error_from_errno("strdup");
396 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
397 1c28a361 2022-10-25 op if (p->len > 0 && p->len <= SIZE_MAX) {
398 1c28a361 2022-10-25 op p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
399 1c28a361 2022-10-25 op if (p->map == MAP_FAILED) {
400 1c28a361 2022-10-25 op if (errno != ENOMEM) {
401 1c28a361 2022-10-25 op err = got_error_from_errno("mmap");
404 1c28a361 2022-10-25 op p->map = NULL; /* fall back to read(2) */
409 c3564dfa 2021-07-15 stsp err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
413 1124fe40 2021-07-07 stsp got_packidx_close(p);
415 0a0a3048 2018-01-10 stsp *packidx = p;
416 1124fe40 2021-07-07 stsp free(pack_relpath);
417 0a0a3048 2018-01-10 stsp return err;
420 57b35b75 2018-06-22 stsp const struct got_error *
421 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
423 57b35b75 2018-06-22 stsp const struct got_error *err = NULL;
425 6fd11751 2018-06-04 stsp free(packidx->path_packidx);
426 fc79a48d 2018-07-09 stsp if (packidx->map) {
427 57b35b75 2018-06-22 stsp if (munmap(packidx->map, packidx->len) == -1)
428 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
430 fc79a48d 2018-07-09 stsp free(packidx->hdr.magic);
431 fc79a48d 2018-07-09 stsp free(packidx->hdr.version);
432 fc79a48d 2018-07-09 stsp free(packidx->hdr.fanout_table);
433 fc79a48d 2018-07-09 stsp free(packidx->hdr.sorted_ids);
434 fc79a48d 2018-07-09 stsp free(packidx->hdr.crc32);
435 fc79a48d 2018-07-09 stsp free(packidx->hdr.offsets);
436 fc79a48d 2018-07-09 stsp free(packidx->hdr.large_offsets);
437 fc79a48d 2018-07-09 stsp free(packidx->hdr.trailer);
439 08578a35 2021-01-22 stsp if (close(packidx->fd) == -1 && err == NULL)
440 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
441 67fd6849 2022-02-13 stsp free(packidx->sorted_offsets);
442 67fd6849 2022-02-13 stsp free(packidx->sorted_large_offsets);
443 0a0a3048 2018-01-10 stsp free(packidx);
445 57b35b75 2018-06-22 stsp return err;
448 509c9973 2021-04-10 stsp const struct got_error *
449 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
451 509c9973 2021-04-10 stsp size_t size;
453 509c9973 2021-04-10 stsp /* Packfile path contains ".pack" instead of ".idx", so add one byte. */
454 aea75d87 2021-07-06 stsp size = strlen(path_packidx) + 2;
455 509c9973 2021-04-10 stsp if (size < GOT_PACKFILE_NAMELEN + 1)
456 aea75d87 2021-07-06 stsp return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
458 509c9973 2021-04-10 stsp *path_packfile = malloc(size);
459 509c9973 2021-04-10 stsp if (*path_packfile == NULL)
460 509c9973 2021-04-10 stsp return got_error_from_errno("malloc");
462 509c9973 2021-04-10 stsp /* Copy up to and excluding ".idx". */
463 aea75d87 2021-07-06 stsp if (strlcpy(*path_packfile, path_packidx,
464 509c9973 2021-04-10 stsp size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
465 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
467 509c9973 2021-04-10 stsp if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
468 509c9973 2021-04-10 stsp return got_error(GOT_ERR_NO_SPACE);
470 509c9973 2021-04-10 stsp return NULL;
474 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
476 78fb0967 2020-09-09 naddy uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
477 a1fd68d8 2018-01-12 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
478 a1fd68d8 2018-01-12 stsp uint64_t loffset;
479 a1fd68d8 2018-01-12 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
480 5e6be232 2019-11-08 stsp if (idx < 0 || idx >= packidx->nlargeobj ||
481 6fd11751 2018-06-04 stsp packidx->hdr.large_offsets == NULL)
483 78fb0967 2020-09-09 naddy loffset = be64toh(packidx->hdr.large_offsets[idx]);
484 a1fd68d8 2018-01-12 stsp return (loffset > INT64_MAX ? -1 : (off_t)loffset);
486 a1fd68d8 2018-01-12 stsp return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
490 c0df5966 2021-12-31 stsp got_packidx_get_object_idx(struct got_packidx *packidx,
491 c0df5966 2021-12-31 stsp struct got_object_id *id)
493 00927983 2020-04-19 stsp u_int8_t id0 = id->sha1[0];
494 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
495 40aeb19c 2018-06-22 stsp int left = 0, right = totobj - 1;
497 a1fd68d8 2018-01-12 stsp if (id0 > 0)
498 78fb0967 2020-09-09 naddy left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
500 40aeb19c 2018-06-22 stsp while (left <= right) {
501 57b35b75 2018-06-22 stsp struct got_packidx_object_id *oid;
502 40aeb19c 2018-06-22 stsp int i, cmp;
504 40aeb19c 2018-06-22 stsp i = ((left + right) / 2);
505 40aeb19c 2018-06-22 stsp oid = &packidx->hdr.sorted_ids[i];
506 00927983 2020-04-19 stsp cmp = memcmp(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
507 16dcbf91 2018-04-01 stsp if (cmp == 0)
509 40aeb19c 2018-06-22 stsp else if (cmp > 0)
510 40aeb19c 2018-06-22 stsp left = i + 1;
511 40aeb19c 2018-06-22 stsp else if (cmp < 0)
512 40aeb19c 2018-06-22 stsp right = i - 1;
519 67fd6849 2022-02-13 stsp offset_cmp(const void *pa, const void *pb)
521 67fd6849 2022-02-13 stsp const struct got_pack_offset_index *a, *b;
523 67fd6849 2022-02-13 stsp a = (const struct got_pack_offset_index *)pa;
524 67fd6849 2022-02-13 stsp b = (const struct got_pack_offset_index *)pb;
526 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
528 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
535 67fd6849 2022-02-13 stsp large_offset_cmp(const void *pa, const void *pb)
537 67fd6849 2022-02-13 stsp const struct got_pack_large_offset_index *a, *b;
539 67fd6849 2022-02-13 stsp a = (const struct got_pack_large_offset_index *)pa;
540 67fd6849 2022-02-13 stsp b = (const struct got_pack_large_offset_index *)pb;
542 67fd6849 2022-02-13 stsp if (a->offset < b->offset)
544 67fd6849 2022-02-13 stsp else if (a->offset > b->offset)
550 67fd6849 2022-02-13 stsp static const struct got_error *
551 67fd6849 2022-02-13 stsp build_offset_index(struct got_packidx *p)
553 67fd6849 2022-02-13 stsp uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
554 67fd6849 2022-02-13 stsp unsigned int i, j, k;
556 67fd6849 2022-02-13 stsp p->sorted_offsets = calloc(nobj - p->nlargeobj,
557 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]));
558 67fd6849 2022-02-13 stsp if (p->sorted_offsets == NULL)
559 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
561 67fd6849 2022-02-13 stsp if (p->nlargeobj > 0) {
562 67fd6849 2022-02-13 stsp p->sorted_large_offsets = calloc(p->nlargeobj,
563 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]));
564 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets == NULL)
565 67fd6849 2022-02-13 stsp return got_error_from_errno("calloc");
570 67fd6849 2022-02-13 stsp for (i = 0; i < nobj; i++) {
571 67fd6849 2022-02-13 stsp uint32_t offset = be32toh(p->hdr.offsets[i]);
572 67fd6849 2022-02-13 stsp if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
573 67fd6849 2022-02-13 stsp uint64_t loffset;
574 67fd6849 2022-02-13 stsp uint32_t idx;
575 67fd6849 2022-02-13 stsp idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
576 67fd6849 2022-02-13 stsp if (idx >= p->nlargeobj ||
577 67fd6849 2022-02-13 stsp p->nlargeobj == 0 ||
578 67fd6849 2022-02-13 stsp p->hdr.large_offsets == NULL)
579 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
580 67fd6849 2022-02-13 stsp loffset = be64toh(p->hdr.large_offsets[idx]);
581 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].offset = loffset;
582 67fd6849 2022-02-13 stsp p->sorted_large_offsets[j].idx = i;
585 67fd6849 2022-02-13 stsp p->sorted_offsets[k].offset = offset;
586 67fd6849 2022-02-13 stsp p->sorted_offsets[k].idx = i;
590 67fd6849 2022-02-13 stsp if (j != p->nlargeobj || k != nobj - p->nlargeobj)
591 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
593 67fd6849 2022-02-13 stsp qsort(p->sorted_offsets, nobj - p->nlargeobj,
594 67fd6849 2022-02-13 stsp sizeof(p->sorted_offsets[0]), offset_cmp);
596 67fd6849 2022-02-13 stsp if (p->sorted_large_offsets)
597 67fd6849 2022-02-13 stsp qsort(p->sorted_large_offsets, p->nlargeobj,
598 67fd6849 2022-02-13 stsp sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
600 67fd6849 2022-02-13 stsp return NULL;
603 876c234b 2018-09-10 stsp const struct got_error *
604 67fd6849 2022-02-13 stsp got_packidx_get_offset_idx(int *idx, struct got_packidx *packidx, off_t offset)
606 67fd6849 2022-02-13 stsp const struct got_error *err;
607 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
608 67fd6849 2022-02-13 stsp int i, left, right;
612 67fd6849 2022-02-13 stsp if (packidx->sorted_offsets == NULL) {
613 67fd6849 2022-02-13 stsp err = build_offset_index(packidx);
615 67fd6849 2022-02-13 stsp return err;
618 67fd6849 2022-02-13 stsp if (offset >= 0x7fffffff) {
619 67fd6849 2022-02-13 stsp uint64_t lo;
620 67fd6849 2022-02-13 stsp left = 0, right = packidx->nlargeobj - 1;
621 67fd6849 2022-02-13 stsp while (left <= right) {
622 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
623 67fd6849 2022-02-13 stsp lo = packidx->sorted_large_offsets[i].offset;
624 67fd6849 2022-02-13 stsp if (lo == offset) {
625 67fd6849 2022-02-13 stsp *idx = packidx->sorted_large_offsets[i].idx;
627 67fd6849 2022-02-13 stsp } else if (offset > lo)
628 67fd6849 2022-02-13 stsp left = i + 1;
629 67fd6849 2022-02-13 stsp else if (offset < lo)
630 67fd6849 2022-02-13 stsp right = i - 1;
633 67fd6849 2022-02-13 stsp uint32_t o;
634 67fd6849 2022-02-13 stsp left = 0, right = totobj - packidx->nlargeobj - 1;
635 67fd6849 2022-02-13 stsp while (left <= right) {
636 67fd6849 2022-02-13 stsp i = ((left + right) / 2);
637 67fd6849 2022-02-13 stsp o = packidx->sorted_offsets[i].offset;
638 67fd6849 2022-02-13 stsp if (o == offset) {
639 67fd6849 2022-02-13 stsp *idx = packidx->sorted_offsets[i].idx;
641 67fd6849 2022-02-13 stsp } else if (offset > o)
642 67fd6849 2022-02-13 stsp left = i + 1;
643 67fd6849 2022-02-13 stsp else if (offset < o)
644 67fd6849 2022-02-13 stsp right = i - 1;
648 67fd6849 2022-02-13 stsp return NULL;
651 67fd6849 2022-02-13 stsp const struct got_error *
652 67fd6849 2022-02-13 stsp got_packidx_get_object_id(struct got_object_id *id,
653 67fd6849 2022-02-13 stsp struct got_packidx *packidx, int idx)
655 67fd6849 2022-02-13 stsp uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
656 67fd6849 2022-02-13 stsp struct got_packidx_object_id *oid;
658 67fd6849 2022-02-13 stsp if (idx < 0 || idx >= totobj)
659 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_NO_OBJ);
661 67fd6849 2022-02-13 stsp oid = &packidx->hdr.sorted_ids[idx];
662 67fd6849 2022-02-13 stsp memcpy(id->sha1, oid->sha1, SHA1_DIGEST_LENGTH);
663 67fd6849 2022-02-13 stsp return NULL;
666 67fd6849 2022-02-13 stsp const struct got_error *
667 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
668 4277420a 2019-06-29 stsp struct got_packidx *packidx, const char *id_str_prefix)
670 dd88155e 2019-06-29 stsp const struct got_error *err = NULL;
671 4277420a 2019-06-29 stsp u_int8_t id0;
672 78fb0967 2020-09-09 naddy uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
673 4277420a 2019-06-29 stsp char hex[3];
674 4277420a 2019-06-29 stsp size_t prefix_len = strlen(id_str_prefix);
675 e09a504c 2019-06-28 stsp struct got_packidx_object_id *oid;
676 404bde06 2022-01-03 stsp uint32_t i = 0;
678 dbdddfee 2021-06-23 naddy STAILQ_INIT(matched_ids);
680 4277420a 2019-06-29 stsp if (prefix_len < 2)
681 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
683 4277420a 2019-06-29 stsp hex[0] = id_str_prefix[0];
684 4277420a 2019-06-29 stsp hex[1] = id_str_prefix[1];
685 4277420a 2019-06-29 stsp hex[2] = '\0';
686 4277420a 2019-06-29 stsp if (!got_parse_xdigit(&id0, hex))
687 6dd1ece6 2019-11-10 stsp return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
689 404bde06 2022-01-03 stsp if (id0 > 0)
690 404bde06 2022-01-03 stsp i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
691 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[i];
692 4277420a 2019-06-29 stsp while (i < totobj && oid->sha1[0] == id0) {
693 4277420a 2019-06-29 stsp char id_str[SHA1_DIGEST_STRING_LENGTH];
694 dd88155e 2019-06-29 stsp struct got_object_qid *qid;
697 4277420a 2019-06-29 stsp if (!got_sha1_digest_to_str(oid->sha1, id_str, sizeof(id_str)))
698 4277420a 2019-06-29 stsp return got_error(GOT_ERR_NO_SPACE);
700 4277420a 2019-06-29 stsp cmp = strncmp(id_str, id_str_prefix, prefix_len);
701 4277420a 2019-06-29 stsp if (cmp < 0) {
702 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
704 4277420a 2019-06-29 stsp } else if (cmp > 0)
707 dd88155e 2019-06-29 stsp err = got_object_qid_alloc_partial(&qid);
710 d7b5a0e8 2022-04-20 stsp memcpy(qid->id.sha1, oid->sha1, SHA1_DIGEST_LENGTH);
711 dbdddfee 2021-06-23 naddy STAILQ_INSERT_TAIL(matched_ids, qid, entry);
713 4277420a 2019-06-29 stsp oid = &packidx->hdr.sorted_ids[++i];
717 0adc7bcc 2019-06-29 stsp got_object_id_queue_free(matched_ids);
718 3d589bee 2022-06-25 stsp return err;
721 3d589bee 2022-06-25 stsp static void
722 3d589bee 2022-06-25 stsp set_max_datasize(void)
724 3d589bee 2022-06-25 stsp struct rlimit rl;
726 3d589bee 2022-06-25 stsp if (getrlimit(RLIMIT_DATA, &rl) != 0)
729 3d589bee 2022-06-25 stsp rl.rlim_cur = rl.rlim_max;
730 3d589bee 2022-06-25 stsp setrlimit(RLIMIT_DATA, &rl);
733 3d589bee 2022-06-25 stsp const struct got_error *
734 3d589bee 2022-06-25 stsp got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
736 3d589bee 2022-06-25 stsp const struct got_error *err = NULL;
737 3d589bee 2022-06-25 stsp int imsg_fds[2];
739 3d589bee 2022-06-25 stsp struct imsgbuf *ibuf;
741 3d589bee 2022-06-25 stsp ibuf = calloc(1, sizeof(*ibuf));
742 3d589bee 2022-06-25 stsp if (ibuf == NULL)
743 3d589bee 2022-06-25 stsp return got_error_from_errno("calloc");
745 3d589bee 2022-06-25 stsp pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
746 3d589bee 2022-06-25 stsp if (pack->privsep_child == NULL) {
747 3d589bee 2022-06-25 stsp err = got_error_from_errno("calloc");
748 3d589bee 2022-06-25 stsp free(ibuf);
749 3d589bee 2022-06-25 stsp return err;
751 3d589bee 2022-06-25 stsp pack->child_has_tempfiles = 0;
752 3d589bee 2022-06-25 stsp pack->child_has_delta_outfd = 0;
754 3d589bee 2022-06-25 stsp if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
755 3d589bee 2022-06-25 stsp err = got_error_from_errno("socketpair");
759 3d589bee 2022-06-25 stsp pid = fork();
760 3d589bee 2022-06-25 stsp if (pid == -1) {
761 3d589bee 2022-06-25 stsp err = got_error_from_errno("fork");
763 3d589bee 2022-06-25 stsp } else if (pid == 0) {
764 3d589bee 2022-06-25 stsp set_max_datasize();
765 3d589bee 2022-06-25 stsp got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
766 3d589bee 2022-06-25 stsp pack->path_packfile);
767 3d589bee 2022-06-25 stsp /* not reached */
770 3d589bee 2022-06-25 stsp if (close(imsg_fds[1]) == -1)
771 3d589bee 2022-06-25 stsp return got_error_from_errno("close");
772 3d589bee 2022-06-25 stsp pack->privsep_child->imsg_fd = imsg_fds[0];
773 3d589bee 2022-06-25 stsp pack->privsep_child->pid = pid;
774 3d589bee 2022-06-25 stsp imsg_init(ibuf, imsg_fds[0]);
775 3d589bee 2022-06-25 stsp pack->privsep_child->ibuf = ibuf;
777 3d589bee 2022-06-25 stsp err = got_privsep_init_pack_child(ibuf, pack, packidx);
779 3d589bee 2022-06-25 stsp const struct got_error *child_err;
780 3d589bee 2022-06-25 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
781 3d589bee 2022-06-25 stsp child_err = got_privsep_wait_for_child(
782 3d589bee 2022-06-25 stsp pack->privsep_child->pid);
783 3d589bee 2022-06-25 stsp if (child_err && err == NULL)
784 3d589bee 2022-06-25 stsp err = child_err;
788 3d589bee 2022-06-25 stsp free(ibuf);
789 3d589bee 2022-06-25 stsp free(pack->privsep_child);
790 3d589bee 2022-06-25 stsp pack->privsep_child = NULL;
792 dd88155e 2019-06-29 stsp return err;
795 b4f37570 2021-06-19 stsp static const struct got_error *
796 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
798 74ef8aae 2022-10-24 stsp const struct got_error *err = NULL, *close_err = NULL;
800 876c234b 2018-09-10 stsp if (pack->privsep_child == NULL)
801 876c234b 2018-09-10 stsp return NULL;
803 876c234b 2018-09-10 stsp err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
805 96732e0b 2018-11-11 stsp return err;
806 74ef8aae 2022-10-24 stsp if (close(pack->privsep_child->imsg_fd) == -1)
807 74ef8aae 2022-10-24 stsp close_err = got_error_from_errno("close");
808 96732e0b 2018-11-11 stsp err = got_privsep_wait_for_child(pack->privsep_child->pid);
809 74ef8aae 2022-10-24 stsp if (close_err && err == NULL)
810 74ef8aae 2022-10-24 stsp err = close_err;
811 74ef8aae 2022-10-24 stsp imsg_clear(pack->privsep_child->ibuf);
812 cc2a8ef4 2021-06-19 tracey free(pack->privsep_child->ibuf);
813 876c234b 2018-09-10 stsp free(pack->privsep_child);
814 876c234b 2018-09-10 stsp pack->privsep_child = NULL;
815 876c234b 2018-09-10 stsp return err;
818 d7464085 2018-07-09 stsp const struct got_error *
819 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
821 d7464085 2018-07-09 stsp const struct got_error *err = NULL;
823 b4f37570 2021-06-19 stsp err = pack_stop_privsep_child(pack);
824 876c234b 2018-09-10 stsp if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
825 638f9024 2019-05-13 stsp err = got_error_from_errno("munmap");
826 08578a35 2021-01-22 stsp if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
827 638f9024 2019-05-13 stsp err = got_error_from_errno("close");
828 8b2180d4 2018-04-26 stsp pack->fd = -1;
829 4589e373 2018-03-17 stsp free(pack->path_packfile);
830 4589e373 2018-03-17 stsp pack->path_packfile = NULL;
831 4589e373 2018-03-17 stsp pack->filesize = 0;
832 ab2f42e7 2019-11-10 stsp if (pack->delta_cache) {
833 ab2f42e7 2019-11-10 stsp got_delta_cache_free(pack->delta_cache);
834 ab2f42e7 2019-11-10 stsp pack->delta_cache = NULL;
838 57160834 2022-05-31 stsp * Leave accumfd and basefd alone. They are managed by the
839 57160834 2022-05-31 stsp * repository layer and can be reused.
842 c7fe698a 2018-01-23 stsp return err;
845 668a20f6 2020-03-18 stsp const struct got_error *
846 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
847 d7464085 2018-07-09 stsp struct got_pack *pack, off_t offset)
849 a487c1d0 2018-01-14 stsp uint8_t t = 0;
850 a487c1d0 2018-01-14 stsp uint64_t s = 0;
851 a487c1d0 2018-01-14 stsp uint8_t sizeN;
852 d7464085 2018-07-09 stsp size_t mapoff = 0;
857 d7464085 2018-07-09 stsp if (offset >= pack->filesize)
858 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
860 d7464085 2018-07-09 stsp if (pack->map) {
861 ad4cc361 2022-10-27 op if (offset > SIZE_MAX) {
862 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
863 ad4cc361 2022-10-27 op "offset %lld overflows size_t",
864 ad4cc361 2022-10-27 op (long long)offset);
867 d7464085 2018-07-09 stsp mapoff = (size_t)offset;
869 d7464085 2018-07-09 stsp if (lseek(pack->fd, offset, SEEK_SET) == -1)
870 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
874 a1fd68d8 2018-01-12 stsp /* We do not support size values which don't fit in 64 bit. */
876 09ee8ded 2022-10-20 stsp return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
877 e082ed67 2022-10-24 naddy "packfile offset %lld", (long long)offset);
879 d7464085 2018-07-09 stsp if (pack->map) {
880 3976db15 2022-01-10 stsp if (mapoff + sizeof(sizeN) >= pack->filesize)
881 3976db15 2022-01-10 stsp return got_error(GOT_ERR_BAD_PACKFILE);
882 d7464085 2018-07-09 stsp sizeN = *(pack->map + mapoff);
883 d7464085 2018-07-09 stsp mapoff += sizeof(sizeN);
885 d7464085 2018-07-09 stsp ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
887 638f9024 2019-05-13 stsp return got_error_from_errno("read");
888 d7464085 2018-07-09 stsp if (n != sizeof(sizeN))
889 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
891 d7464085 2018-07-09 stsp *len += sizeof(sizeN);
893 a1fd68d8 2018-01-12 stsp if (i == 0) {
894 a487c1d0 2018-01-14 stsp t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
895 a1fd68d8 2018-01-12 stsp GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
896 a487c1d0 2018-01-14 stsp s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
898 a1fd68d8 2018-01-12 stsp size_t shift = 4 + 7 * (i - 1);
899 a487c1d0 2018-01-14 stsp s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
902 a1fd68d8 2018-01-12 stsp } while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
906 a487c1d0 2018-01-14 stsp return NULL;
909 a1fd68d8 2018-01-12 stsp static const struct got_error *
910 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
911 5f25cc85 2019-11-26 stsp uint8_t type, off_t offset, size_t size, int idx)
913 6ccb713b 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
914 6ccb713b 2018-01-19 stsp if (*obj == NULL)
915 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
917 6ccb713b 2018-01-19 stsp (*obj)->type = type;
918 6ccb713b 2018-01-19 stsp (*obj)->flags = GOT_OBJ_FLAG_PACKED;
919 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
920 6ccb713b 2018-01-19 stsp (*obj)->hdrlen = 0;
921 6ccb713b 2018-01-19 stsp (*obj)->size = size;
922 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
923 6ccb713b 2018-01-19 stsp (*obj)->pack_offset = offset;
925 b107e67f 2018-01-19 stsp return NULL;
928 b107e67f 2018-01-19 stsp static const struct got_error *
929 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
930 d7464085 2018-07-09 stsp off_t delta_offset)
932 b107e67f 2018-01-19 stsp int64_t o = 0;
933 b107e67f 2018-01-19 stsp uint8_t offN;
936 a0de39f3 2019-08-09 stsp *offset = 0;
940 b107e67f 2018-01-19 stsp /* We do not support offset values which don't fit in 64 bit. */
942 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_NO_SPACE);
944 d7464085 2018-07-09 stsp if (pack->map) {
945 d7464085 2018-07-09 stsp size_t mapoff;
947 ad4cc361 2022-10-27 op if (delta_offset + *len > SIZE_MAX) {
948 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
949 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
950 ad4cc361 2022-10-27 op (long long)delta_offset + *len);
953 d7464085 2018-07-09 stsp mapoff = (size_t)delta_offset + *len;
954 3976db15 2022-01-10 stsp if (mapoff + sizeof(offN) >= pack->filesize)
955 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
956 d7464085 2018-07-09 stsp offN = *(pack->map + mapoff);
959 d7464085 2018-07-09 stsp n = read(pack->fd, &offN, sizeof(offN));
961 638f9024 2019-05-13 stsp return got_error_from_errno("read");
962 d7464085 2018-07-09 stsp if (n != sizeof(offN))
963 d7464085 2018-07-09 stsp return got_error(GOT_ERR_BAD_PACKFILE);
965 d7464085 2018-07-09 stsp *len += sizeof(offN);
967 b107e67f 2018-01-19 stsp if (i == 0)
968 b107e67f 2018-01-19 stsp o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
972 cecc778e 2018-01-23 stsp o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
975 b107e67f 2018-01-19 stsp } while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
977 b107e67f 2018-01-19 stsp *offset = o;
978 6ccb713b 2018-01-19 stsp return NULL;
981 668a20f6 2020-03-18 stsp const struct got_error *
982 c0df5966 2021-12-31 stsp got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
983 24d916d2 2022-10-24 op struct got_pack *pack, off_t offset, size_t tslen)
985 96f5e8b3 2018-01-23 stsp const struct got_error *err;
986 b107e67f 2018-01-19 stsp int64_t negoffset;
987 b107e67f 2018-01-19 stsp size_t negofflen;
991 d7464085 2018-07-09 stsp err = parse_negative_offset(&negoffset, &negofflen, pack,
992 d7464085 2018-07-09 stsp offset + tslen);
994 b107e67f 2018-01-19 stsp return err;
996 b107e67f 2018-01-19 stsp /* Compute the base object's offset (must be in the same pack file). */
997 96f5e8b3 2018-01-23 stsp *base_offset = (offset - negoffset);
998 96f5e8b3 2018-01-23 stsp if (*base_offset <= 0)
999 b107e67f 2018-01-19 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1001 d7464085 2018-07-09 stsp *len = negofflen;
1002 96f5e8b3 2018-01-23 stsp return NULL;
1005 0e22967e 2018-02-11 stsp static const struct got_error *
1006 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1007 2d9e6abf 2022-05-04 stsp size_t *delta_compressed_len, size_t delta_data_offset,
1008 2d9e6abf 2022-05-04 stsp struct got_pack *pack)
1010 42c69117 2019-11-10 stsp const struct got_error *err = NULL;
1011 2d9e6abf 2022-05-04 stsp size_t consumed = 0;
1013 42c69117 2019-11-10 stsp if (pack->map) {
1014 42c69117 2019-11-10 stsp if (delta_data_offset >= pack->filesize)
1015 42c69117 2019-11-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1016 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1017 2d9e6abf 2022-05-04 stsp &consumed, NULL, pack->map, delta_data_offset,
1018 2e5a6fad 2020-03-18 stsp pack->filesize - delta_data_offset);
1020 2d9e6abf 2022-05-04 stsp return err;
1022 42c69117 2019-11-10 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1023 42c69117 2019-11-10 stsp return got_error_from_errno("lseek");
1024 2d9e6abf 2022-05-04 stsp err = got_inflate_to_mem_fd(delta_buf, delta_len,
1025 2d9e6abf 2022-05-04 stsp &consumed, NULL, 0, pack->fd);
1027 2d9e6abf 2022-05-04 stsp return err;
1030 2d9e6abf 2022-05-04 stsp if (delta_compressed_len)
1031 2d9e6abf 2022-05-04 stsp *delta_compressed_len = consumed;
1033 2d9e6abf 2022-05-04 stsp return NULL;
1036 96f5e8b3 2018-01-23 stsp static const struct got_error *
1037 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1038 e5792992 2022-10-28 op int delta_type, size_t delta_size, off_t delta_data_offset)
1040 bdd2fbb3 2018-02-11 stsp struct got_delta *delta;
1042 c336f889 2018-07-23 stsp delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1043 42c69117 2019-11-10 stsp delta_data_offset);
1044 bdd2fbb3 2018-02-11 stsp if (delta == NULL)
1045 638f9024 2019-05-13 stsp return got_error_from_errno("got_delta_open");
1046 bdd2fbb3 2018-02-11 stsp /* delta is freed in got_object_close() */
1047 bdd2fbb3 2018-02-11 stsp deltas->nentries++;
1048 dbdddfee 2021-06-23 naddy STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1049 bdd2fbb3 2018-02-11 stsp return NULL;
1052 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1053 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1054 c8ecd499 2018-09-09 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1055 c8ecd499 2018-09-09 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1057 a3500804 2018-01-23 stsp const struct got_error *err;
1058 c3703302 2018-01-23 stsp off_t base_offset;
1059 c3703302 2018-01-23 stsp uint8_t base_type;
1060 c3703302 2018-01-23 stsp uint64_t base_size;
1061 c3703302 2018-01-23 stsp size_t base_tslen;
1062 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1063 42c69117 2019-11-10 stsp size_t consumed;
1065 668a20f6 2020-03-18 stsp err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1066 d7464085 2018-07-09 stsp delta_offset, tslen);
1068 a3500804 2018-01-23 stsp return err;
1070 d7464085 2018-07-09 stsp delta_data_offset = delta_offset + tslen + consumed;
1071 d7464085 2018-07-09 stsp if (delta_data_offset >= pack->filesize)
1072 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1074 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1075 d7464085 2018-07-09 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1076 d7464085 2018-07-09 stsp if (delta_data_offset == -1)
1077 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1080 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1081 e5792992 2022-10-28 op delta_data_offset);
1083 1a35c1bc 2019-05-22 stsp return err;
1085 c3703302 2018-01-23 stsp /* An offset delta must be in the same packfile. */
1086 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1087 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1089 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1090 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1092 1a35c1bc 2019-05-22 stsp return err;
1094 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1095 b0e2201a 2018-06-22 stsp base_tslen, base_type, base_size, recursion - 1);
1098 c4330eff 2021-06-22 stsp const struct got_error *
1099 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1100 c4330eff 2021-06-22 stsp struct got_pack *pack, off_t delta_offset, int tslen)
1102 c4330eff 2021-06-22 stsp if (pack->map) {
1103 ad4cc361 2022-10-27 op size_t mapoff;
1105 ad4cc361 2022-10-27 op if (delta_offset + tslen > SIZE_MAX) {
1106 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_PACK_OFFSET,
1107 ad4cc361 2022-10-27 op "mapoff %lld would overflow size_t",
1108 ad4cc361 2022-10-27 op (long long)delta_offset + tslen);
1111 ad4cc361 2022-10-27 op mapoff = delta_offset + tslen;
1112 3976db15 2022-01-10 stsp if (mapoff + sizeof(*id) >= pack->filesize)
1113 3976db15 2022-01-10 stsp return got_error(GOT_ERR_PACK_OFFSET);
1114 c4330eff 2021-06-22 stsp memcpy(id, pack->map + mapoff, sizeof(*id));
1116 c4330eff 2021-06-22 stsp ssize_t n;
1117 c4330eff 2021-06-22 stsp n = read(pack->fd, id, sizeof(*id));
1118 c4330eff 2021-06-22 stsp if (n < 0)
1119 c4330eff 2021-06-22 stsp return got_error_from_errno("read");
1120 c4330eff 2021-06-22 stsp if (n != sizeof(*id))
1121 c4330eff 2021-06-22 stsp return got_error(GOT_ERR_BAD_PACKFILE);
1124 c4330eff 2021-06-22 stsp return NULL;
1127 c3703302 2018-01-23 stsp static const struct got_error *
1128 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1129 c8ecd499 2018-09-09 stsp struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1130 c8ecd499 2018-09-09 stsp size_t delta_size, unsigned int recursion)
1132 6b9c9673 2018-01-23 stsp const struct got_error *err;
1133 6b9c9673 2018-01-23 stsp struct got_object_id id;
1135 6b9c9673 2018-01-23 stsp off_t base_offset;
1136 6b9c9673 2018-01-23 stsp uint8_t base_type;
1137 6b9c9673 2018-01-23 stsp uint64_t base_size;
1138 6b9c9673 2018-01-23 stsp size_t base_tslen;
1139 bdd2fbb3 2018-02-11 stsp off_t delta_data_offset;
1141 42c69117 2019-11-10 stsp if (delta_offset + tslen >= pack->filesize)
1142 d7464085 2018-07-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1144 c4330eff 2021-06-22 stsp err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1146 c4330eff 2021-06-22 stsp return err;
1147 d7464085 2018-07-09 stsp if (pack->map) {
1148 f990756a 2023-02-18 op delta_data_offset = delta_offset + tslen + SHA1_DIGEST_LENGTH;
1150 e40b19ed 2020-01-06 stsp delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1151 e40b19ed 2020-01-06 stsp if (delta_data_offset == -1)
1152 e40b19ed 2020-01-06 stsp return got_error_from_errno("lseek");
1155 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1156 e5792992 2022-10-28 op delta_data_offset);
1158 1a35c1bc 2019-05-22 stsp return err;
1160 652b2703 2018-06-22 stsp /* Delta base must be in the same pack file. */
1161 1510f469 2018-09-09 stsp idx = got_packidx_get_object_idx(packidx, &id);
1162 1a35c1bc 2019-05-22 stsp if (idx == -1)
1163 668a20f6 2020-03-18 stsp return got_error(GOT_ERR_NO_OBJ);
1165 02828bfd 2021-06-22 stsp base_offset = got_packidx_get_object_offset(packidx, idx);
1166 04aed155 2022-07-24 naddy if (base_offset == -1)
1167 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1169 1a35c1bc 2019-05-22 stsp if (base_offset >= pack->filesize)
1170 1a35c1bc 2019-05-22 stsp return got_error(GOT_ERR_PACK_OFFSET);
1172 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1173 668a20f6 2020-03-18 stsp &base_tslen, pack, base_offset);
1175 1a35c1bc 2019-05-22 stsp return err;
1177 668a20f6 2020-03-18 stsp return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1178 0c048b15 2018-04-27 stsp base_tslen, base_type, base_size, recursion - 1);
1181 668a20f6 2020-03-18 stsp const struct got_error *
1182 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1183 668a20f6 2020-03-18 stsp struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1184 668a20f6 2020-03-18 stsp size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1186 c3703302 2018-01-23 stsp const struct got_error *err = NULL;
1188 5b7e13a7 2018-04-02 stsp if (--recursion == 0)
1189 5b7e13a7 2018-04-02 stsp return got_error(GOT_ERR_RECURSION);
1191 c3703302 2018-01-23 stsp switch (delta_type) {
1192 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_COMMIT:
1193 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TREE:
1194 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_BLOB:
1195 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_TAG:
1196 c3703302 2018-01-23 stsp /* Plain types are the final delta base. Recursion ends. */
1197 c336f889 2018-07-23 stsp err = add_delta(deltas, delta_offset, tslen, delta_type,
1198 42c69117 2019-11-10 stsp delta_size, 0);
1200 a3500804 2018-01-23 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1201 c8ecd499 2018-09-09 stsp err = resolve_offset_delta(deltas, packidx, pack,
1202 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1204 4e8cda55 2018-01-19 stsp case GOT_OBJ_TYPE_REF_DELTA:
1205 c8ecd499 2018-09-09 stsp err = resolve_ref_delta(deltas, packidx, pack,
1206 b0e2201a 2018-06-22 stsp delta_offset, tslen, delta_type, delta_size, recursion - 1);
1209 4810de4a 2018-04-01 stsp return got_error(GOT_ERR_OBJ_TYPE);
1212 96f5e8b3 2018-01-23 stsp return err;
1215 96f5e8b3 2018-01-23 stsp static const struct got_error *
1216 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1217 a98c62b1 2018-09-09 stsp struct got_pack *pack, struct got_object_id *id, off_t offset,
1218 c59b3346 2018-09-11 stsp size_t tslen, int delta_type, size_t delta_size, int idx)
1220 96f5e8b3 2018-01-23 stsp const struct got_error *err = NULL;
1221 96f5e8b3 2018-01-23 stsp int resolved_type;
1223 b107e67f 2018-01-19 stsp *obj = calloc(1, sizeof(**obj));
1224 b107e67f 2018-01-19 stsp if (*obj == NULL)
1225 638f9024 2019-05-13 stsp return got_error_from_errno("calloc");
1227 96f5e8b3 2018-01-23 stsp (*obj)->flags = 0;
1228 b107e67f 2018-01-19 stsp (*obj)->hdrlen = 0;
1229 39e73dc9 2018-03-03 stsp (*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1230 106807b4 2018-09-15 stsp memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1231 cecc778e 2018-01-23 stsp (*obj)->pack_offset = offset + tslen;
1233 dbdddfee 2021-06-23 naddy STAILQ_INIT(&(*obj)->deltas.entries);
1234 1a35c1bc 2019-05-22 stsp (*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1235 96f5e8b3 2018-01-23 stsp (*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1236 c59b3346 2018-09-11 stsp (*obj)->pack_idx = idx;
1238 668a20f6 2020-03-18 stsp err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1239 668a20f6 2020-03-18 stsp offset, tslen, delta_type, delta_size,
1240 668a20f6 2020-03-18 stsp GOT_DELTA_CHAIN_RECURSION_MAX);
1242 96f5e8b3 2018-01-23 stsp goto done;
1244 96f5e8b3 2018-01-23 stsp err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1246 96f5e8b3 2018-01-23 stsp goto done;
1247 96f5e8b3 2018-01-23 stsp (*obj)->type = resolved_type;
1249 96f5e8b3 2018-01-23 stsp if (err) {
1250 96f5e8b3 2018-01-23 stsp got_object_close(*obj);
1251 96f5e8b3 2018-01-23 stsp *obj = NULL;
1253 96f5e8b3 2018-01-23 stsp return err;
1256 2090a03d 2018-09-09 stsp const struct got_error *
1257 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1258 6fd11751 2018-06-04 stsp struct got_packidx *packidx, int idx, struct got_object_id *id)
1260 a1fd68d8 2018-01-12 stsp const struct got_error *err = NULL;
1261 a1fd68d8 2018-01-12 stsp off_t offset;
1262 6c00b545 2018-01-17 stsp uint8_t type;
1263 6c00b545 2018-01-17 stsp uint64_t size;
1264 3ee5fc21 2018-01-17 stsp size_t tslen;
1266 6c00b545 2018-01-17 stsp *obj = NULL;
1268 02828bfd 2021-06-22 stsp offset = got_packidx_get_object_offset(packidx, idx);
1269 04aed155 2022-07-24 naddy if (offset == -1)
1270 a1fd68d8 2018-01-12 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1272 668a20f6 2020-03-18 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1273 668a20f6 2020-03-18 stsp pack, offset);
1275 35c73765 2018-09-09 stsp return err;
1277 6c00b545 2018-01-17 stsp switch (type) {
1278 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_COMMIT:
1279 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_TREE:
1280 6c00b545 2018-01-17 stsp case GOT_OBJ_TYPE_BLOB:
1281 d33fc9ef 2018-01-23 stsp case GOT_OBJ_TYPE_TAG:
1282 5f25cc85 2019-11-26 stsp err = open_plain_object(obj, id, type, offset + tslen,
1283 5f25cc85 2019-11-26 stsp size, idx);
1285 b107e67f 2018-01-19 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1286 a48db7e5 2018-01-23 stsp case GOT_OBJ_TYPE_REF_DELTA:
1287 a98c62b1 2018-09-09 stsp err = open_delta_object(obj, packidx, pack, id, offset,
1288 c59b3346 2018-09-11 stsp tslen, type, size, idx);
1291 4810de4a 2018-04-01 stsp err = got_error(GOT_ERR_OBJ_TYPE);
1295 3ee5fc21 2018-01-17 stsp return err;
1298 d582f26c 2020-03-18 stsp const struct got_error *
1299 c0df5966 2021-12-31 stsp got_pack_get_delta_chain_max_size(uint64_t *max_size,
1300 c0df5966 2021-12-31 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1302 22484865 2018-03-13 stsp struct got_delta *delta;
1303 22484865 2018-03-13 stsp uint64_t base_size = 0, result_size = 0;
1305 22484865 2018-03-13 stsp *max_size = 0;
1306 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1307 22484865 2018-03-13 stsp /* Plain object types are the delta base. */
1308 22484865 2018-03-13 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1309 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1310 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1311 22484865 2018-03-13 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1312 22484865 2018-03-13 stsp const struct got_error *err;
1313 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1314 42c69117 2019-11-10 stsp size_t delta_len;
1315 ab2f42e7 2019-11-10 stsp int cached = 1;
1317 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1318 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1319 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1321 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1322 ab2f42e7 2019-11-10 stsp cached = 0;
1323 ab2f42e7 2019-11-10 stsp err = read_delta_data(&delta_buf, &delta_len,
1324 2d9e6abf 2022-05-04 stsp NULL, delta->data_offset, pack);
1326 ab2f42e7 2019-11-10 stsp return err;
1328 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1329 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1330 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1331 ab2f42e7 2019-11-10 stsp if (err == NULL)
1332 ab2f42e7 2019-11-10 stsp cached = 1;
1333 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1334 ab2f42e7 2019-11-10 stsp free(delta_buf);
1335 ab2f42e7 2019-11-10 stsp return err;
1338 a53d2f13 2018-03-17 stsp err = got_delta_get_sizes(&base_size, &result_size,
1339 42c69117 2019-11-10 stsp delta_buf, delta_len);
1340 ab2f42e7 2019-11-10 stsp if (!cached)
1341 ab2f42e7 2019-11-10 stsp free(delta_buf);
1343 22484865 2018-03-13 stsp return err;
1345 22484865 2018-03-13 stsp base_size = delta->size;
1346 22484865 2018-03-13 stsp if (base_size > *max_size)
1347 22484865 2018-03-13 stsp *max_size = base_size;
1348 22484865 2018-03-13 stsp if (result_size > *max_size)
1349 22484865 2018-03-13 stsp *max_size = result_size;
1352 9feb4ff2 2018-03-14 stsp return NULL;
1355 ac544f8c 2019-01-13 stsp const struct got_error *
1356 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1357 42c69117 2019-11-10 stsp struct got_pack *pack)
1359 85a703fa 2019-01-13 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1360 85a703fa 2019-01-13 stsp return got_error(GOT_ERR_OBJ_TYPE);
1362 d582f26c 2020-03-18 stsp return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1365 668a20f6 2020-03-18 stsp const struct got_error *
1366 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1367 4788f1ce 2020-03-18 stsp struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1368 4788f1ce 2020-03-18 stsp FILE *base_file, FILE *accum_file)
1370 efd2a263 2018-01-19 stsp const struct got_error *err = NULL;
1371 6691714a 2018-01-23 stsp struct got_delta *delta;
1372 aabb25f8 2022-10-17 stsp uint8_t *base_buf = NULL, *accum_buf = NULL;
1373 6395114c 2022-05-31 stsp size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1374 6395114c 2022-05-31 stsp /* We process small enough files entirely in memory for speed. */
1375 6395114c 2022-05-31 stsp const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1376 6395114c 2022-05-31 stsp uint64_t max_size = 0;
1377 6691714a 2018-01-23 stsp int n = 0;
1379 b29656e2 2018-03-16 stsp *result_size = 0;
1381 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1382 6691714a 2018-01-23 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1384 6395114c 2022-05-31 stsp if (fseeko(base_file, 0L, SEEK_SET) == -1)
1385 6395114c 2022-05-31 stsp return got_error_from_errno("fseeko");
1386 6395114c 2022-05-31 stsp if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1387 6395114c 2022-05-31 stsp return got_error_from_errno("fseeko");
1389 6691714a 2018-01-23 stsp /* Deltas are ordered in ascending order. */
1390 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1391 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1392 6395114c 2022-05-31 stsp uint64_t base_size, result_size = 0;
1393 ab2f42e7 2019-11-10 stsp int cached = 1;
1394 a6b158cc 2018-02-11 stsp if (n == 0) {
1395 34fca9c3 2018-11-11 stsp size_t mapoff;
1396 0c048b15 2018-04-27 stsp off_t delta_data_offset;
1398 a6b158cc 2018-02-11 stsp /* Plain object types are the delta base. */
1399 a6b158cc 2018-02-11 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1400 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1401 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1402 a6b158cc 2018-02-11 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1403 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1404 72eb3431 2018-04-01 stsp goto done;
1407 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1408 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1409 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1410 0c048b15 2018-04-27 stsp goto done;
1412 d7464085 2018-07-09 stsp if (pack->map == NULL) {
1413 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1415 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1416 d7464085 2018-07-09 stsp goto done;
1419 6395114c 2022-05-31 stsp if (delta->size > max_size)
1420 6395114c 2022-05-31 stsp max_size = delta->size;
1421 6395114c 2022-05-31 stsp if (max_size > max_bufsize) {
1422 d7464085 2018-07-09 stsp if (pack->map) {
1423 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1424 ad4cc361 2022-10-27 op return got_error_fmt(
1425 ad4cc361 2022-10-27 op GOT_ERR_RANGE,
1426 ad4cc361 2022-10-27 op "delta offset %lld "
1427 ad4cc361 2022-10-27 op "overflows size_t",
1429 ad4cc361 2022-10-27 op delta_data_offset);
1432 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1433 d7464085 2018-07-09 stsp err = got_inflate_to_file_mmap(
1434 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1435 4788f1ce 2020-03-18 stsp mapoff, pack->filesize - mapoff,
1436 4788f1ce 2020-03-18 stsp base_file);
1438 34fca9c3 2018-11-11 stsp err = got_inflate_to_file_fd(
1439 4788f1ce 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->fd,
1440 4788f1ce 2020-03-18 stsp base_file);
1442 6395114c 2022-05-31 stsp accum_buf = malloc(max_size);
1443 6395114c 2022-05-31 stsp if (accum_buf == NULL) {
1444 6395114c 2022-05-31 stsp err = got_error_from_errno("malloc");
1445 6395114c 2022-05-31 stsp goto done;
1447 6395114c 2022-05-31 stsp accum_bufsz = max_size;
1448 d7464085 2018-07-09 stsp if (pack->map) {
1449 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1450 ad4cc361 2022-10-27 op return got_error_fmt(
1451 ad4cc361 2022-10-27 op GOT_ERR_RANGE,
1452 ad4cc361 2022-10-27 op "delta offset %lld "
1453 ad4cc361 2022-10-27 op "overflows size_t",
1455 ad4cc361 2022-10-27 op delta_data_offset);
1458 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1459 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1460 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL,
1461 2e5a6fad 2020-03-18 stsp pack->map, mapoff,
1462 d7464085 2018-07-09 stsp pack->filesize - mapoff);
1464 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1465 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1466 55fdd257 2020-03-18 stsp pack->fd);
1469 a6b158cc 2018-02-11 stsp goto done;
1471 6395114c 2022-05-31 stsp if (base_buf == NULL)
1472 8628c62d 2018-03-15 stsp rewind(base_file);
1476 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1477 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1478 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1480 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1481 ab2f42e7 2019-11-10 stsp cached = 0;
1482 2d9e6abf 2022-05-04 stsp err = read_delta_data(&delta_buf, &delta_len, NULL,
1483 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1485 ab2f42e7 2019-11-10 stsp goto done;
1487 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1488 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1489 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1490 ab2f42e7 2019-11-10 stsp if (err == NULL)
1491 ab2f42e7 2019-11-10 stsp cached = 1;
1492 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1493 ab2f42e7 2019-11-10 stsp free(delta_buf);
1494 ab2f42e7 2019-11-10 stsp goto done;
1498 6395114c 2022-05-31 stsp err = got_delta_get_sizes(&base_size, &result_size,
1499 6395114c 2022-05-31 stsp delta_buf, delta_len);
1500 e62fc520 2022-11-08 stsp if (err) {
1501 e62fc520 2022-11-08 stsp if (!cached)
1502 e62fc520 2022-11-08 stsp free(delta_buf);
1503 6395114c 2022-05-31 stsp goto done;
1505 6395114c 2022-05-31 stsp if (base_size > max_size)
1506 6395114c 2022-05-31 stsp max_size = base_size;
1507 6395114c 2022-05-31 stsp if (result_size > max_size)
1508 6395114c 2022-05-31 stsp max_size = result_size;
1510 6395114c 2022-05-31 stsp if (base_buf && max_size > max_bufsize) {
1511 6395114c 2022-05-31 stsp /* Switch from buffers to temporary files. */
1512 6395114c 2022-05-31 stsp size_t w = fwrite(base_buf, 1, base_bufsz,
1513 6395114c 2022-05-31 stsp base_file);
1514 6395114c 2022-05-31 stsp if (w != base_bufsz) {
1515 6395114c 2022-05-31 stsp err = got_ferror(outfile, GOT_ERR_IO);
1516 e62fc520 2022-11-08 stsp if (!cached)
1517 e62fc520 2022-11-08 stsp free(delta_buf);
1518 6395114c 2022-05-31 stsp goto done;
1520 6395114c 2022-05-31 stsp free(base_buf);
1521 6395114c 2022-05-31 stsp base_buf = NULL;
1522 6395114c 2022-05-31 stsp free(accum_buf);
1523 6395114c 2022-05-31 stsp accum_buf = NULL;
1526 6395114c 2022-05-31 stsp if (base_buf && max_size > base_bufsz) {
1527 6395114c 2022-05-31 stsp uint8_t *p = realloc(base_buf, max_size);
1528 6395114c 2022-05-31 stsp if (p == NULL) {
1529 6395114c 2022-05-31 stsp err = got_error_from_errno("realloc");
1530 e62fc520 2022-11-08 stsp if (!cached)
1531 e62fc520 2022-11-08 stsp free(delta_buf);
1532 6395114c 2022-05-31 stsp goto done;
1534 6395114c 2022-05-31 stsp base_buf = p;
1535 6395114c 2022-05-31 stsp base_bufsz = max_size;
1538 6395114c 2022-05-31 stsp if (accum_buf && max_size > accum_bufsz) {
1539 6395114c 2022-05-31 stsp uint8_t *p = realloc(accum_buf, max_size);
1540 6395114c 2022-05-31 stsp if (p == NULL) {
1541 6395114c 2022-05-31 stsp err = got_error_from_errno("realloc");
1542 e62fc520 2022-11-08 stsp if (!cached)
1543 e62fc520 2022-11-08 stsp free(delta_buf);
1544 6395114c 2022-05-31 stsp goto done;
1546 6395114c 2022-05-31 stsp accum_buf = p;
1547 6395114c 2022-05-31 stsp accum_bufsz = max_size;
1550 8628c62d 2018-03-15 stsp if (base_buf) {
1551 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1552 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1553 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1556 42c69117 2019-11-10 stsp err = got_delta_apply(base_file, delta_buf,
1557 42c69117 2019-11-10 stsp delta_len,
1558 8628c62d 2018-03-15 stsp /* Final delta application writes to output file. */
1559 b29656e2 2018-03-16 stsp ++n < deltas->nentries ? accum_file : outfile,
1560 b29656e2 2018-03-16 stsp &accum_size);
1562 ab2f42e7 2019-11-10 stsp if (!cached)
1563 ab2f42e7 2019-11-10 stsp free(delta_buf);
1565 6691714a 2018-01-23 stsp goto done;
1567 710bb5ca 2018-01-23 stsp if (n < deltas->nentries) {
1568 6691714a 2018-01-23 stsp /* Accumulated delta becomes the new base. */
1569 8628c62d 2018-03-15 stsp if (base_buf) {
1570 8628c62d 2018-03-15 stsp uint8_t *tmp = accum_buf;
1571 6395114c 2022-05-31 stsp size_t tmp_size = accum_bufsz;
1572 8628c62d 2018-03-15 stsp accum_buf = base_buf;
1573 6395114c 2022-05-31 stsp accum_bufsz = base_bufsz;
1574 8628c62d 2018-03-15 stsp base_buf = tmp;
1575 6395114c 2022-05-31 stsp base_bufsz = tmp_size;
1577 8628c62d 2018-03-15 stsp FILE *tmp = accum_file;
1578 8628c62d 2018-03-15 stsp accum_file = base_file;
1579 8628c62d 2018-03-15 stsp base_file = tmp;
1580 8628c62d 2018-03-15 stsp rewind(base_file);
1581 8628c62d 2018-03-15 stsp rewind(accum_file);
1587 8628c62d 2018-03-15 stsp free(base_buf);
1588 8628c62d 2018-03-15 stsp if (accum_buf) {
1589 8628c62d 2018-03-15 stsp size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1590 8628c62d 2018-03-15 stsp free(accum_buf);
1591 8628c62d 2018-03-15 stsp if (len != accum_size)
1592 673702af 2018-07-23 stsp err = got_ferror(outfile, GOT_ERR_IO);
1594 6691714a 2018-01-23 stsp rewind(outfile);
1595 b29656e2 2018-03-16 stsp if (err == NULL)
1596 b29656e2 2018-03-16 stsp *result_size = accum_size;
1597 e0ab43e7 2018-03-16 stsp return err;
1600 668a20f6 2020-03-18 stsp const struct got_error *
1601 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1602 48095039 2018-09-09 stsp struct got_delta_chain *deltas, struct got_pack *pack)
1604 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1605 e0ab43e7 2018-03-16 stsp struct got_delta *delta;
1606 aabb25f8 2022-10-17 stsp uint8_t *base_buf = NULL, *accum_buf = NULL;
1607 f18c433a 2022-05-31 stsp size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1608 f18c433a 2022-05-31 stsp uint64_t max_size = 0;
1609 e0ab43e7 2018-03-16 stsp int n = 0;
1611 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1612 e0ab43e7 2018-03-16 stsp *outlen = 0;
1614 dbdddfee 2021-06-23 naddy if (STAILQ_EMPTY(&deltas->entries))
1615 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1617 e0ab43e7 2018-03-16 stsp /* Deltas are ordered in ascending order. */
1618 dbdddfee 2021-06-23 naddy STAILQ_FOREACH(delta, &deltas->entries, entry) {
1619 aabb25f8 2022-10-17 stsp uint8_t *delta_buf = NULL;
1620 f18c433a 2022-05-31 stsp uint64_t base_size, result_size = 0;
1621 ab2f42e7 2019-11-10 stsp int cached = 1;
1622 e0ab43e7 2018-03-16 stsp if (n == 0) {
1623 ad4cc361 2022-10-27 op off_t delta_data_offset;
1625 e0ab43e7 2018-03-16 stsp /* Plain object types are the delta base. */
1626 e0ab43e7 2018-03-16 stsp if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1627 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TREE &&
1628 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_BLOB &&
1629 e0ab43e7 2018-03-16 stsp delta->type != GOT_OBJ_TYPE_TAG) {
1630 72eb3431 2018-04-01 stsp err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1631 72eb3431 2018-04-01 stsp goto done;
1634 0c048b15 2018-04-27 stsp delta_data_offset = delta->offset + delta->tslen;
1635 0c048b15 2018-04-27 stsp if (delta_data_offset >= pack->filesize) {
1636 0c048b15 2018-04-27 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1637 0c048b15 2018-04-27 stsp goto done;
1640 f18c433a 2022-05-31 stsp if (delta->size > max_size)
1641 f18c433a 2022-05-31 stsp max_size = delta->size;
1643 d7464085 2018-07-09 stsp if (pack->map) {
1644 ad4cc361 2022-10-27 op size_t mapoff;
1646 ad4cc361 2022-10-27 op if (delta_data_offset > SIZE_MAX) {
1647 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1648 ad4cc361 2022-10-27 op "delta %lld offset would "
1649 ad4cc361 2022-10-27 op "overflow size_t",
1650 ad4cc361 2022-10-27 op (long long)delta_data_offset);
1653 ad4cc361 2022-10-27 op mapoff = delta_data_offset;
1654 d7464085 2018-07-09 stsp err = got_inflate_to_mem_mmap(&base_buf,
1655 2e5a6fad 2020-03-18 stsp &base_bufsz, NULL, NULL, pack->map,
1656 2e5a6fad 2020-03-18 stsp mapoff, pack->filesize - mapoff);
1658 d7464085 2018-07-09 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1660 638f9024 2019-05-13 stsp err = got_error_from_errno("lseek");
1661 d7464085 2018-07-09 stsp goto done;
1663 d7464085 2018-07-09 stsp err = got_inflate_to_mem_fd(&base_buf,
1664 55fdd257 2020-03-18 stsp &base_bufsz, NULL, NULL, max_size,
1665 55fdd257 2020-03-18 stsp pack->fd);
1668 d7464085 2018-07-09 stsp goto done;
1673 aabb25f8 2022-10-17 stsp if (pack->delta_cache) {
1674 aabb25f8 2022-10-17 stsp got_delta_cache_get(&delta_buf, &delta_len,
1675 aabb25f8 2022-10-17 stsp pack->delta_cache, delta->data_offset);
1677 ab2f42e7 2019-11-10 stsp if (delta_buf == NULL) {
1678 ab2f42e7 2019-11-10 stsp cached = 0;
1679 2d9e6abf 2022-05-04 stsp err = read_delta_data(&delta_buf, &delta_len, NULL,
1680 ab2f42e7 2019-11-10 stsp delta->data_offset, pack);
1682 ab2f42e7 2019-11-10 stsp goto done;
1684 aabb25f8 2022-10-17 stsp if (pack->delta_cache && !cached) {
1685 ab2f42e7 2019-11-10 stsp err = got_delta_cache_add(pack->delta_cache,
1686 ab2f42e7 2019-11-10 stsp delta->data_offset, delta_buf, delta_len);
1687 ab2f42e7 2019-11-10 stsp if (err == NULL)
1688 ab2f42e7 2019-11-10 stsp cached = 1;
1689 ab2f42e7 2019-11-10 stsp else if (err->code != GOT_ERR_NO_SPACE) {
1690 ab2f42e7 2019-11-10 stsp free(delta_buf);
1691 ab2f42e7 2019-11-10 stsp goto done;
1695 f18c433a 2022-05-31 stsp err = got_delta_get_sizes(&base_size, &result_size,
1696 f18c433a 2022-05-31 stsp delta_buf, delta_len);
1697 e62fc520 2022-11-08 stsp if (err) {
1698 e62fc520 2022-11-08 stsp if (!cached)
1699 e62fc520 2022-11-08 stsp free(delta_buf);
1700 f18c433a 2022-05-31 stsp goto done;
1702 f18c433a 2022-05-31 stsp if (base_size > max_size)
1703 f18c433a 2022-05-31 stsp max_size = base_size;
1704 f18c433a 2022-05-31 stsp if (result_size > max_size)
1705 f18c433a 2022-05-31 stsp max_size = result_size;
1707 f18c433a 2022-05-31 stsp if (max_size > base_bufsz) {
1708 f18c433a 2022-05-31 stsp uint8_t *p = realloc(base_buf, max_size);
1709 f18c433a 2022-05-31 stsp if (p == NULL) {
1710 f18c433a 2022-05-31 stsp err = got_error_from_errno("realloc");
1711 e62fc520 2022-11-08 stsp if (!cached)
1712 e62fc520 2022-11-08 stsp free(delta_buf);
1713 f18c433a 2022-05-31 stsp goto done;
1715 f18c433a 2022-05-31 stsp base_buf = p;
1716 f18c433a 2022-05-31 stsp base_bufsz = max_size;
1719 f18c433a 2022-05-31 stsp if (max_size > accum_bufsz) {
1720 f18c433a 2022-05-31 stsp uint8_t *p = realloc(accum_buf, max_size);
1721 f18c433a 2022-05-31 stsp if (p == NULL) {
1722 f18c433a 2022-05-31 stsp err = got_error_from_errno("realloc");
1723 e62fc520 2022-11-08 stsp if (!cached)
1724 e62fc520 2022-11-08 stsp free(delta_buf);
1725 f18c433a 2022-05-31 stsp goto done;
1727 f18c433a 2022-05-31 stsp accum_buf = p;
1728 f18c433a 2022-05-31 stsp accum_bufsz = max_size;
1731 34fca9c3 2018-11-11 stsp err = got_delta_apply_in_mem(base_buf, base_bufsz,
1732 42c69117 2019-11-10 stsp delta_buf, delta_len, accum_buf,
1733 34fca9c3 2018-11-11 stsp &accum_size, max_size);
1734 ab2f42e7 2019-11-10 stsp if (!cached)
1735 ab2f42e7 2019-11-10 stsp free(delta_buf);
1738 e0ab43e7 2018-03-16 stsp goto done;
1740 e0ab43e7 2018-03-16 stsp if (n < deltas->nentries) {
1741 e0ab43e7 2018-03-16 stsp /* Accumulated delta becomes the new base. */
1742 e0ab43e7 2018-03-16 stsp uint8_t *tmp = accum_buf;
1743 f18c433a 2022-05-31 stsp size_t tmp_size = accum_bufsz;
1744 e0ab43e7 2018-03-16 stsp accum_buf = base_buf;
1745 f18c433a 2022-05-31 stsp accum_bufsz = base_bufsz;
1746 e0ab43e7 2018-03-16 stsp base_buf = tmp;
1747 f18c433a 2022-05-31 stsp base_bufsz = tmp_size;
1752 e0ab43e7 2018-03-16 stsp free(base_buf);
1753 e0ab43e7 2018-03-16 stsp if (err) {
1754 e0ab43e7 2018-03-16 stsp free(accum_buf);
1755 e0ab43e7 2018-03-16 stsp *outbuf = NULL;
1756 e0ab43e7 2018-03-16 stsp *outlen = 0;
1758 e0ab43e7 2018-03-16 stsp *outbuf = accum_buf;
1759 e0ab43e7 2018-03-16 stsp *outlen = accum_size;
1761 efd2a263 2018-01-19 stsp return err;
1764 3ee5fc21 2018-01-17 stsp const struct got_error *
1765 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1766 3840f4c9 2018-09-12 stsp FILE *outfile, FILE *base_file, FILE *accum_file)
1768 3ee5fc21 2018-01-17 stsp const struct got_error *err = NULL;
1770 3ee5fc21 2018-01-17 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1771 3ee5fc21 2018-01-17 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1773 ef2bccd9 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1774 24140570 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1775 24140570 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1777 d7464085 2018-07-09 stsp if (pack->map) {
1778 ad4cc361 2022-10-27 op size_t mapoff;
1780 ad4cc361 2022-10-27 op if (obj->pack_offset > SIZE_MAX) {
1781 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1782 ad4cc361 2022-10-27 op "pack offset %lld would overflow size_t",
1783 ad4cc361 2022-10-27 op (long long)obj->pack_offset);
1786 ad4cc361 2022-10-27 op mapoff = obj->pack_offset;
1787 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1788 4788f1ce 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff,
1791 24140570 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1792 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1793 4788f1ce 2020-03-18 stsp err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1794 4788f1ce 2020-03-18 stsp pack->fd, outfile);
1797 4788f1ce 2020-03-18 stsp err = got_pack_dump_delta_chain_to_file(&obj->size,
1798 4788f1ce 2020-03-18 stsp &obj->deltas, pack, outfile, base_file, accum_file);
1800 a1fd68d8 2018-01-12 stsp return err;
1803 e0ab43e7 2018-03-16 stsp const struct got_error *
1804 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1805 7e212e3d 2018-09-09 stsp struct got_object *obj, struct got_pack *pack)
1807 e0ab43e7 2018-03-16 stsp const struct got_error *err = NULL;
1809 e0ab43e7 2018-03-16 stsp if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1810 e0ab43e7 2018-03-16 stsp return got_error(GOT_ERR_OBJ_NOT_PACKED);
1812 48095039 2018-09-09 stsp if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1813 7e212e3d 2018-09-09 stsp if (obj->pack_offset >= pack->filesize)
1814 7e212e3d 2018-09-09 stsp return got_error(GOT_ERR_PACK_OFFSET);
1815 d7464085 2018-07-09 stsp if (pack->map) {
1816 ad4cc361 2022-10-27 op size_t mapoff;
1818 ad4cc361 2022-10-27 op if (obj->pack_offset > SIZE_MAX) {
1819 ad4cc361 2022-10-27 op return got_error_fmt(GOT_ERR_RANGE,
1820 ad4cc361 2022-10-27 op "pack offset %lld would overflow size_t",
1821 ad4cc361 2022-10-27 op (long long)obj->pack_offset);
1824 ad4cc361 2022-10-27 op mapoff = obj->pack_offset;
1825 2e5a6fad 2020-03-18 stsp err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1826 2e5a6fad 2020-03-18 stsp pack->map, mapoff, pack->filesize - mapoff);
1828 7e212e3d 2018-09-09 stsp if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1829 638f9024 2019-05-13 stsp return got_error_from_errno("lseek");
1830 1e87a3c3 2020-03-18 stsp err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1831 55fdd257 2020-03-18 stsp obj->size, pack->fd);
1834 668a20f6 2020-03-18 stsp err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1837 e0ab43e7 2018-03-16 stsp return err;
1840 2d9e6abf 2022-05-04 stsp static const struct got_error *
1841 2d9e6abf 2022-05-04 stsp read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1842 2d9e6abf 2022-05-04 stsp size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1843 2d9e6abf 2022-05-04 stsp off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1845 2d9e6abf 2022-05-04 stsp const struct got_error *err = NULL;
1847 2d9e6abf 2022-05-04 stsp /* Validate decompression and obtain the decompressed size. */
1848 2d9e6abf 2022-05-04 stsp err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1849 2d9e6abf 2022-05-04 stsp delta_data_offset, pack);
1851 2d9e6abf 2022-05-04 stsp return err;
1853 2d9e6abf 2022-05-04 stsp /* Read delta base/result sizes from head of delta stream. */
1854 2d9e6abf 2022-05-04 stsp err = got_delta_get_sizes(base_size, result_size,
1855 2d9e6abf 2022-05-04 stsp *delta_buf, *delta_len);
1857 2d9e6abf 2022-05-04 stsp goto done;
1859 2d9e6abf 2022-05-04 stsp /* Discard decompressed delta and read it again in compressed form. */
1860 2d9e6abf 2022-05-04 stsp free(*delta_buf);
1861 2d9e6abf 2022-05-04 stsp *delta_buf = malloc(*delta_len_compressed);
1862 2d9e6abf 2022-05-04 stsp if (*delta_buf == NULL) {
1863 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("malloc");
1864 2d9e6abf 2022-05-04 stsp goto done;
1866 2d9e6abf 2022-05-04 stsp if (pack->map) {
1867 82031ac8 2022-10-24 op if (delta_data_offset >= pack->filesize) {
1868 2d9e6abf 2022-05-04 stsp err = got_error(GOT_ERR_PACK_OFFSET);
1871 2d9e6abf 2022-05-04 stsp memcpy(*delta_buf, pack->map + delta_data_offset,
1872 2d9e6abf 2022-05-04 stsp *delta_len_compressed);
1874 2d9e6abf 2022-05-04 stsp ssize_t n;
1875 2d9e6abf 2022-05-04 stsp if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1876 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("lseek");
1877 2d9e6abf 2022-05-04 stsp goto done;
1879 2d9e6abf 2022-05-04 stsp n = read(pack->fd, *delta_buf, *delta_len_compressed);
1880 2d9e6abf 2022-05-04 stsp if (n < 0) {
1881 2d9e6abf 2022-05-04 stsp err = got_error_from_errno("read");
1882 2d9e6abf 2022-05-04 stsp goto done;
1883 2d9e6abf 2022-05-04 stsp } else if (n != *delta_len_compressed) {
1884 2d9e6abf 2022-05-04 stsp err = got_error(GOT_ERR_IO);
1885 2d9e6abf 2022-05-04 stsp goto done;
1889 2d9e6abf 2022-05-04 stsp if (err) {
1890 2d9e6abf 2022-05-04 stsp free(*delta_buf);
1891 2d9e6abf 2022-05-04 stsp *delta_buf = NULL;
1892 2d9e6abf 2022-05-04 stsp *delta_len = 0;
1893 2d9e6abf 2022-05-04 stsp *delta_len_compressed = 0;
1894 2d9e6abf 2022-05-04 stsp *base_size = 0;
1895 2d9e6abf 2022-05-04 stsp *result_size = 0;
1897 2d9e6abf 2022-05-04 stsp return err;
1900 67fd6849 2022-02-13 stsp const struct got_error *
1901 67fd6849 2022-02-13 stsp got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1902 24b7de1c 2022-12-03 stsp size_t *delta_compressed_size, off_t *delta_offset,
1903 24b7de1c 2022-12-03 stsp off_t *delta_data_offset, off_t *base_offset,
1904 2d9e6abf 2022-05-04 stsp struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1905 2d9e6abf 2022-05-04 stsp struct got_pack *pack, struct got_packidx *packidx, int idx)
1907 67fd6849 2022-02-13 stsp const struct got_error *err = NULL;
1908 67fd6849 2022-02-13 stsp off_t offset;
1909 67fd6849 2022-02-13 stsp uint8_t type;
1910 67fd6849 2022-02-13 stsp uint64_t size;
1911 67fd6849 2022-02-13 stsp size_t tslen, delta_hdrlen;
1913 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1914 67fd6849 2022-02-13 stsp *delta_size = 0;
1915 2d9e6abf 2022-05-04 stsp *delta_compressed_size = 0;
1916 67fd6849 2022-02-13 stsp *delta_offset = 0;
1917 24b7de1c 2022-12-03 stsp *delta_data_offset = 0;
1918 67fd6849 2022-02-13 stsp *base_offset = 0;
1919 67fd6849 2022-02-13 stsp *base_size = 0;
1920 67fd6849 2022-02-13 stsp *result_size = 0;
1922 67fd6849 2022-02-13 stsp offset = got_packidx_get_object_offset(packidx, idx);
1923 04aed155 2022-07-24 naddy if (offset == -1)
1924 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_PACKIDX);
1926 67fd6849 2022-02-13 stsp if (offset >= pack->filesize)
1927 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1929 67fd6849 2022-02-13 stsp err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1930 67fd6849 2022-02-13 stsp pack, offset);
1932 67fd6849 2022-02-13 stsp return err;
1934 67fd6849 2022-02-13 stsp if (tslen + size < tslen || offset + size < size ||
1935 67fd6849 2022-02-13 stsp tslen + offset < tslen)
1936 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_PACK_OFFSET);
1938 67fd6849 2022-02-13 stsp switch (type) {
1939 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_OFFSET_DELTA:
1940 67fd6849 2022-02-13 stsp err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1941 67fd6849 2022-02-13 stsp pack, offset, tslen);
1943 67fd6849 2022-02-13 stsp return err;
1945 67fd6849 2022-02-13 stsp case GOT_OBJ_TYPE_REF_DELTA:
1946 67fd6849 2022-02-13 stsp err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1948 67fd6849 2022-02-13 stsp return err;
1949 67fd6849 2022-02-13 stsp delta_hdrlen = SHA1_DIGEST_LENGTH;
1952 67fd6849 2022-02-13 stsp return got_error_fmt(GOT_ERR_OBJ_TYPE,
1953 04aed155 2022-07-24 naddy "non-delta object type %d found at offset %lld",
1954 04aed155 2022-07-24 naddy type, (long long)offset);
1957 67fd6849 2022-02-13 stsp if (tslen + delta_hdrlen < delta_hdrlen ||
1958 67fd6849 2022-02-13 stsp offset + delta_hdrlen < delta_hdrlen)
1959 67fd6849 2022-02-13 stsp return got_error(GOT_ERR_BAD_DELTA);
1961 24b7de1c 2022-12-03 stsp *delta_data_offset = offset + tslen + delta_hdrlen;
1962 2d9e6abf 2022-05-04 stsp err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1963 24b7de1c 2022-12-03 stsp base_size, result_size, *delta_data_offset, pack, packidx);
1965 67fd6849 2022-02-13 stsp return err;
1967 67fd6849 2022-02-13 stsp if (*delta_size != size) {
1968 67fd6849 2022-02-13 stsp err = got_error(GOT_ERR_BAD_DELTA);
1969 67fd6849 2022-02-13 stsp goto done;
1972 67fd6849 2022-02-13 stsp *delta_offset = offset;
1974 67fd6849 2022-02-13 stsp if (err) {
1975 67fd6849 2022-02-13 stsp free(*delta_buf);
1976 67fd6849 2022-02-13 stsp *delta_buf = NULL;
1977 2d9e6abf 2022-05-04 stsp *delta_size = 0;
1978 2d9e6abf 2022-05-04 stsp *delta_compressed_size = 0;
1979 2d9e6abf 2022-05-04 stsp *delta_offset = 0;
1980 2d9e6abf 2022-05-04 stsp *base_offset = 0;
1981 2d9e6abf 2022-05-04 stsp *base_size = 0;
1982 2d9e6abf 2022-05-04 stsp *result_size = 0;
1984 67fd6849 2022-02-13 stsp return err;
