1 0a0a3048 2018-01-10 stsp /*
   2 5aa81393 2020-01-06 stsp  * Copyright (c) 2018, 2019, 2020 Stefan Sperling <stsp@openbsd.org>
   3 0a0a3048 2018-01-10 stsp  *
   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.
   7 0a0a3048 2018-01-10 stsp  *
   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.
  15 0a0a3048 2018-01-10 stsp  */
  16 0a0a3048 2018-01-10 stsp 
  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>
  24 0a0a3048 2018-01-10 stsp 
  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 69c6accf 2023-02-04 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>
  38 0a0a3048 2018-01-10 stsp 
  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"
  42 0a0a3048 2018-01-10 stsp 
  43 1362b0e3 2023-02-04 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"
  51 79b11c62 2018-03-09 stsp 
  52 79b11c62 2018-03-09 stsp #ifndef nitems
  53 79b11c62 2018-03-09 stsp #define nitems(_a) (sizeof(_a) / sizeof((_a)[0]))
  54 79b11c62 2018-03-09 stsp #endif
  55 1411938b 2018-02-12 stsp 
  56 a1fd68d8 2018-01-12 stsp #ifndef MIN
  57 a1fd68d8 2018-01-12 stsp #define	MIN(_a,_b) ((_a) < (_b) ? (_a) : (_b))
  58 a1fd68d8 2018-01-12 stsp #endif
  59 a1fd68d8 2018-01-12 stsp 
  60 0a0a3048 2018-01-10 stsp static const struct got_error *
  61 0a0a3048 2018-01-10 stsp verify_fanout_table(uint32_t *fanout_table)
  62 0a0a3048 2018-01-10 stsp {
  63 0a0a3048 2018-01-10 stsp 	int i;
  64 0a0a3048 2018-01-10 stsp 
  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);
  68 0a0a3048 2018-01-10 stsp 	}
  69 0a0a3048 2018-01-10 stsp 
  70 0a0a3048 2018-01-10 stsp 	return NULL;
  71 0a0a3048 2018-01-10 stsp }
  72 0a0a3048 2018-01-10 stsp 
  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)
  75 0a0a3048 2018-01-10 stsp {
  76 0a0a3048 2018-01-10 stsp 	const struct got_error *err = NULL;
  77 817c5a18 2018-09-09 stsp 	struct got_packidx_v2_hdr *h;
  78 0ebaf008 2018-01-10 stsp 	SHA1_CTX ctx;
  79 3093e2d7 2023-02-04 op 	uint8_t hash[SHA1_DIGEST_LENGTH];
  80 817c5a18 2018-09-09 stsp 	size_t nobj, len_fanout, len_ids, offset, remain;
  81 817c5a18 2018-09-09 stsp 	ssize_t n;
  82 5e6be232 2019-11-08 stsp 	int i;
  83 0a0a3048 2018-01-10 stsp 
  84 0ebaf008 2018-01-10 stsp 	SHA1Init(&ctx);
  85 0ebaf008 2018-01-10 stsp 
  86 57b35b75 2018-06-22 stsp 	h = &p->hdr;
  87 57b35b75 2018-06-22 stsp 	offset = 0;
  88 57b35b75 2018-06-22 stsp 	remain = p->len;
  89 0a0a3048 2018-01-10 stsp 
  90 57b35b75 2018-06-22 stsp 	if (remain < sizeof(*h->magic)) {
  91 0a0a3048 2018-01-10 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
  92 0a0a3048 2018-01-10 stsp 		goto done;
  93 0a0a3048 2018-01-10 stsp 	}
  94 fc79a48d 2018-07-09 stsp 	if (p->map)
  95 fc79a48d 2018-07-09 stsp 		h->magic = (uint32_t *)(p->map + offset);
  96 fc79a48d 2018-07-09 stsp 	else {
  97 fc79a48d 2018-07-09 stsp 		h->magic = malloc(sizeof(*h->magic));
  98 fc79a48d 2018-07-09 stsp 		if (h->magic == NULL) {
  99 638f9024 2019-05-13 stsp 			err = got_error_from_errno("malloc");
 100 fc79a48d 2018-07-09 stsp 			goto done;
 101 fc79a48d 2018-07-09 stsp 		}
 102 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->magic, sizeof(*h->magic));
 103 b1317e77 2019-09-22 stsp 		if (n < 0) {
 104 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 105 b1317e77 2019-09-22 stsp 			goto done;
 106 b1317e77 2019-09-22 stsp 		} else if (n != sizeof(*h->magic)) {
 107 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 108 fc79a48d 2018-07-09 stsp 			goto done;
 109 fc79a48d 2018-07-09 stsp 		}
 110 fc79a48d 2018-07-09 stsp 	}
 111 ac62b712 2021-03-30 stsp 	if (*h->magic != htobe32(GOT_PACKIDX_V2_MAGIC)) {
 112 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 113 57b35b75 2018-06-22 stsp 		goto done;
 114 57b35b75 2018-06-22 stsp 	}
 115 57b35b75 2018-06-22 stsp 	offset += sizeof(*h->magic);
 116 57b35b75 2018-06-22 stsp 	remain -= sizeof(*h->magic);
 117 0a0a3048 2018-01-10 stsp 
 118 0cb74cf4 2018-07-08 stsp 	if (verify)
 119 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->magic, sizeof(*h->magic));
 120 0ebaf008 2018-01-10 stsp 
 121 57b35b75 2018-06-22 stsp 	if (remain < sizeof(*h->version)) {
 122 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 123 0a0a3048 2018-01-10 stsp 		goto done;
 124 0a0a3048 2018-01-10 stsp 	}
 125 fc79a48d 2018-07-09 stsp 	if (p->map)
 126 fc79a48d 2018-07-09 stsp 		h->version = (uint32_t *)(p->map + offset);
 127 fc79a48d 2018-07-09 stsp 	else {
 128 fc79a48d 2018-07-09 stsp 		h->version = malloc(sizeof(*h->version));
 129 fc79a48d 2018-07-09 stsp 		if (h->version == NULL) {
 130 638f9024 2019-05-13 stsp 			err = got_error_from_errno("malloc");
 131 fc79a48d 2018-07-09 stsp 			goto done;
 132 fc79a48d 2018-07-09 stsp 		}
 133 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->version, sizeof(*h->version));
 134 c6368c2e 2019-10-11 stsp 		if (n < 0) {
 135 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 136 c6368c2e 2019-10-11 stsp 			goto done;
 137 c6368c2e 2019-10-11 stsp 		} else if (n != sizeof(*h->version)) {
 138 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 139 fc79a48d 2018-07-09 stsp 			goto done;
 140 fc79a48d 2018-07-09 stsp 		}
 141 fc79a48d 2018-07-09 stsp 	}
 142 ac62b712 2021-03-30 stsp 	if (*h->version != htobe32(GOT_PACKIDX_VERSION)) {
 143 0a0a3048 2018-01-10 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 144 0a0a3048 2018-01-10 stsp 		goto done;
 145 0a0a3048 2018-01-10 stsp 	}
 146 57b35b75 2018-06-22 stsp 	offset += sizeof(*h->version);
 147 57b35b75 2018-06-22 stsp 	remain -= sizeof(*h->version);
 148 0a0a3048 2018-01-10 stsp 
 149 0cb74cf4 2018-07-08 stsp 	if (verify)
 150 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->version, sizeof(*h->version));
 151 0ebaf008 2018-01-10 stsp 
 152 57b35b75 2018-06-22 stsp 	len_fanout =
 153 57b35b75 2018-06-22 stsp 	    sizeof(*h->fanout_table) * GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS;
 154 57b35b75 2018-06-22 stsp 	if (remain < len_fanout) {
 155 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 156 0a0a3048 2018-01-10 stsp 		goto done;
 157 0a0a3048 2018-01-10 stsp 	}
 158 fc79a48d 2018-07-09 stsp 	if (p->map)
 159 fc79a48d 2018-07-09 stsp 		h->fanout_table = (uint32_t *)(p->map + offset);
 160 fc79a48d 2018-07-09 stsp 	else {
 161 fc79a48d 2018-07-09 stsp 		h->fanout_table = malloc(len_fanout);
 162 fc79a48d 2018-07-09 stsp 		if (h->fanout_table == NULL) {
 163 638f9024 2019-05-13 stsp 			err = got_error_from_errno("malloc");
 164 fc79a48d 2018-07-09 stsp 			goto done;
 165 fc79a48d 2018-07-09 stsp 		}
 166 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->fanout_table, len_fanout);
 167 c6368c2e 2019-10-11 stsp 		if (n < 0) {
 168 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 169 c6368c2e 2019-10-11 stsp 			goto done;
 170 c6368c2e 2019-10-11 stsp 		} else if (n != len_fanout) {
 171 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 172 fc79a48d 2018-07-09 stsp 			goto done;
 173 fc79a48d 2018-07-09 stsp 		}
 174 fc79a48d 2018-07-09 stsp 	}
 175 c8262310 2018-06-04 stsp 	err = verify_fanout_table(h->fanout_table);
 176 0a0a3048 2018-01-10 stsp 	if (err)
 177 0a0a3048 2018-01-10 stsp 		goto done;
 178 0cb74cf4 2018-07-08 stsp 	if (verify)
 179 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->fanout_table, len_fanout);
 180 57b35b75 2018-06-22 stsp 	offset += len_fanout;
 181 57b35b75 2018-06-22 stsp 	remain -= len_fanout;
 182 0a0a3048 2018-01-10 stsp 
 183 78fb0967 2020-09-09 naddy 	nobj = be32toh(h->fanout_table[0xff]);
 184 57b35b75 2018-06-22 stsp 	len_ids = nobj * sizeof(*h->sorted_ids);
 185 57b35b75 2018-06-22 stsp 	if (len_ids <= nobj || len_ids > remain) {
 186 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 187 0a0a3048 2018-01-10 stsp 		goto done;
 188 0a0a3048 2018-01-10 stsp 	}
 189 fc79a48d 2018-07-09 stsp 	if (p->map)
 190 fc79a48d 2018-07-09 stsp 		h->sorted_ids =
 191 fc79a48d 2018-07-09 stsp 		    (struct got_packidx_object_id *)((uint8_t*)(p->map + offset));
 192 fc79a48d 2018-07-09 stsp 	else {
 193 fc79a48d 2018-07-09 stsp 		h->sorted_ids = malloc(len_ids);
 194 fc79a48d 2018-07-09 stsp 		if (h->sorted_ids == NULL) {
 195 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 196 fc79a48d 2018-07-09 stsp 			goto done;
 197 fc79a48d 2018-07-09 stsp 		}
 198 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->sorted_ids, len_ids);
 199 faaa1c0f 2018-09-15 stsp 		if (n < 0)
 200 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 201 faaa1c0f 2018-09-15 stsp 		else if (n != len_ids) {
 202 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 203 fc79a48d 2018-07-09 stsp 			goto done;
 204 fc79a48d 2018-07-09 stsp 		}
 205 fc79a48d 2018-07-09 stsp 	}
 206 0cb74cf4 2018-07-08 stsp 	if (verify)
 207 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->sorted_ids, len_ids);
 208 57b35b75 2018-06-22 stsp 	offset += len_ids;
 209 57b35b75 2018-06-22 stsp 	remain -= len_ids;
 210 0a0a3048 2018-01-10 stsp 
 211 57b35b75 2018-06-22 stsp 	if (remain < nobj * sizeof(*h->crc32)) {
 212 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 213 0a0a3048 2018-01-10 stsp 		goto done;
 214 0a0a3048 2018-01-10 stsp 	}
 215 fc79a48d 2018-07-09 stsp 	if (p->map)
 216 fc79a48d 2018-07-09 stsp 		h->crc32 = (uint32_t *)((uint8_t*)(p->map + offset));
 217 fc79a48d 2018-07-09 stsp 	else {
 218 fc79a48d 2018-07-09 stsp 		h->crc32 = malloc(nobj * sizeof(*h->crc32));
 219 fc79a48d 2018-07-09 stsp 		if (h->crc32 == NULL) {
 220 638f9024 2019-05-13 stsp 			err = got_error_from_errno("malloc");
 221 fc79a48d 2018-07-09 stsp 			goto done;
 222 fc79a48d 2018-07-09 stsp 		}
 223 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->crc32, nobj * sizeof(*h->crc32));
 224 faaa1c0f 2018-09-15 stsp 		if (n < 0)
 225 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 226 faaa1c0f 2018-09-15 stsp 		else if (n != nobj * sizeof(*h->crc32)) {
 227 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 228 fc79a48d 2018-07-09 stsp 			goto done;
 229 fc79a48d 2018-07-09 stsp 		}
 230 fc79a48d 2018-07-09 stsp 	}
 231 0cb74cf4 2018-07-08 stsp 	if (verify)
 232 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->crc32, nobj * sizeof(*h->crc32));
 233 57b35b75 2018-06-22 stsp 	remain -= nobj * sizeof(*h->crc32);
 234 57b35b75 2018-06-22 stsp 	offset += nobj * sizeof(*h->crc32);
 235 0ebaf008 2018-01-10 stsp 
 236 57b35b75 2018-06-22 stsp 	if (remain < nobj * sizeof(*h->offsets)) {
 237 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 238 0a0a3048 2018-01-10 stsp 		goto done;
 239 0a0a3048 2018-01-10 stsp 	}
 240 fc79a48d 2018-07-09 stsp 	if (p->map)
 241 fc79a48d 2018-07-09 stsp 		h->offsets = (uint32_t *)((uint8_t*)(p->map + offset));
 242 fc79a48d 2018-07-09 stsp 	else {
 243 fc79a48d 2018-07-09 stsp 		h->offsets = malloc(nobj * sizeof(*h->offsets));
 244 fc79a48d 2018-07-09 stsp 		if (h->offsets == NULL) {
 245 638f9024 2019-05-13 stsp 			err = got_error_from_errno("malloc");
 246 fc79a48d 2018-07-09 stsp 			goto done;
 247 fc79a48d 2018-07-09 stsp 		}
 248 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->offsets, nobj * sizeof(*h->offsets));
 249 faaa1c0f 2018-09-15 stsp 		if (n < 0)
 250 638f9024 2019-05-13 stsp 			err = got_error_from_errno("read");
 251 faaa1c0f 2018-09-15 stsp 		else if (n != nobj * sizeof(*h->offsets)) {
 252 fc79a48d 2018-07-09 stsp 			err = got_error(GOT_ERR_BAD_PACKIDX);
 253 fc79a48d 2018-07-09 stsp 			goto done;
 254 fc79a48d 2018-07-09 stsp 		}
 255 fc79a48d 2018-07-09 stsp 	}
 256 0cb74cf4 2018-07-08 stsp 	if (verify)
 257 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t *)h->offsets,
 258 0cb74cf4 2018-07-08 stsp 		    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);
 261 0ebaf008 2018-01-10 stsp 
 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++;
 268 c3564dfa 2021-07-15 stsp 		}
 269 5e6be232 2019-11-08 stsp 	}
 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);
 274 c3564dfa 2021-07-15 stsp 		goto done;
 275 c3564dfa 2021-07-15 stsp 	}
 276 0a0a3048 2018-01-10 stsp 
 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);
 279 0a0a3048 2018-01-10 stsp 		goto done;
 280 0a0a3048 2018-01-10 stsp 	}
 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));
 283 fc79a48d 2018-07-09 stsp 	else {
 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");
 288 fc79a48d 2018-07-09 stsp 			goto done;
 289 fc79a48d 2018-07-09 stsp 		}
 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));
 292 faaa1c0f 2018-09-15 stsp 		if (n < 0)
 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);
 296 fc79a48d 2018-07-09 stsp 			goto done;
 297 fc79a48d 2018-07-09 stsp 		}
 298 fc79a48d 2018-07-09 stsp 	}
 299 0cb74cf4 2018-07-08 stsp 	if (verify)
 300 0cb74cf4 2018-07-08 stsp 		SHA1Update(&ctx, (uint8_t*)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);
 304 0ebaf008 2018-01-10 stsp 
 305 0a0a3048 2018-01-10 stsp checksum:
 306 57b35b75 2018-06-22 stsp 	if (remain < sizeof(*h->trailer)) {
 307 57b35b75 2018-06-22 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 308 0a0a3048 2018-01-10 stsp 		goto done;
 309 0a0a3048 2018-01-10 stsp 	}
 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));
 313 fc79a48d 2018-07-09 stsp 	else {
 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");
 317 fc79a48d 2018-07-09 stsp 			goto done;
 318 fc79a48d 2018-07-09 stsp 		}
 319 fc79a48d 2018-07-09 stsp 		n = read(p->fd, h->trailer, sizeof(*h->trailer));
 320 faaa1c0f 2018-09-15 stsp 		if (n < 0)
 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);
 324 fc79a48d 2018-07-09 stsp 			goto done;
 325 fc79a48d 2018-07-09 stsp 		}
 326 fc79a48d 2018-07-09 stsp 	}
 327 0cb74cf4 2018-07-08 stsp 	if (verify) {
 328 3093e2d7 2023-02-04 op 		SHA1Update(&ctx, h->trailer->packfile_hash, SHA1_DIGEST_LENGTH);
 329 3093e2d7 2023-02-04 op 		SHA1Final(hash, &ctx);
 330 3093e2d7 2023-02-04 op 		if (memcmp(h->trailer->packidx_hash, hash,
 331 0cb74cf4 2018-07-08 stsp 		    SHA1_DIGEST_LENGTH) != 0)
 332 0cb74cf4 2018-07-08 stsp 			err = got_error(GOT_ERR_PACKIDX_CSUM);
 333 0cb74cf4 2018-07-08 stsp 	}
 334 0a0a3048 2018-01-10 stsp done:
 335 817c5a18 2018-09-09 stsp 	return err;
 336 817c5a18 2018-09-09 stsp }
 337 817c5a18 2018-09-09 stsp 
 338 817c5a18 2018-09-09 stsp const struct got_error *
 339 6d5a9006 2020-12-16 yzhong got_packidx_open(struct got_packidx **packidx,
 340 6d5a9006 2020-12-16 yzhong     int dir_fd, const char *relpath, int verify)
 341 817c5a18 2018-09-09 stsp {
 342 817c5a18 2018-09-09 stsp 	const struct got_error *err = NULL;
 343 1124fe40 2021-07-07 stsp 	struct got_packidx *p = NULL;
 344 1124fe40 2021-07-07 stsp 	char *pack_relpath;
 345 c3564dfa 2021-07-15 stsp 	struct stat idx_sb, pack_sb;
 346 817c5a18 2018-09-09 stsp 
 347 817c5a18 2018-09-09 stsp 	*packidx = NULL;
 348 1124fe40 2021-07-07 stsp 
 349 1124fe40 2021-07-07 stsp 	err = got_packidx_get_packfile_path(&pack_relpath, relpath);
 350 1124fe40 2021-07-07 stsp 	if (err)
 351 1124fe40 2021-07-07 stsp 		return err;
 352 1124fe40 2021-07-07 stsp 
 353 1124fe40 2021-07-07 stsp 	/*
 354 1124fe40 2021-07-07 stsp 	 * Ensure that a corresponding pack file exists.
 355 1124fe40 2021-07-07 stsp 	 * Some Git repositories have this problem. Git seems to ignore
 356 1124fe40 2021-07-07 stsp 	 * the existence of lonely pack index files but we do not.
 357 1124fe40 2021-07-07 stsp 	 */
 358 c3564dfa 2021-07-15 stsp 	if (fstatat(dir_fd, pack_relpath, &pack_sb, 0) == -1) {
 359 1124fe40 2021-07-07 stsp 		if (errno == ENOENT) {
 360 1124fe40 2021-07-07 stsp 			err = got_error_fmt(GOT_ERR_LONELY_PACKIDX,
 361 1124fe40 2021-07-07 stsp 			    "%s", relpath);
 362 1124fe40 2021-07-07 stsp 		} else
 363 1124fe40 2021-07-07 stsp 			err = got_error_from_errno2("fstatat", pack_relpath);
 364 1124fe40 2021-07-07 stsp 		goto done;
 365 1124fe40 2021-07-07 stsp 	}
 366 817c5a18 2018-09-09 stsp 
 367 817c5a18 2018-09-09 stsp 	p = calloc(1, sizeof(*p));
 368 1124fe40 2021-07-07 stsp 	if (p == NULL) {
 369 1124fe40 2021-07-07 stsp 		err = got_error_from_errno("calloc");
 370 1124fe40 2021-07-07 stsp 		goto done;
 371 1124fe40 2021-07-07 stsp 	}
 372 817c5a18 2018-09-09 stsp 
 373 e7ae0baf 2021-12-31 stsp 	p->fd = openat(dir_fd, relpath, O_RDONLY | O_NOFOLLOW | O_CLOEXEC);
 374 6772cf22 2019-08-28 hiltjo 	if (p->fd == -1) {
 375 6d5a9006 2020-12-16 yzhong 		err = got_error_from_errno2("openat", relpath);
 376 1124fe40 2021-07-07 stsp 		goto done;
 377 6772cf22 2019-08-28 hiltjo 	}
 378 817c5a18 2018-09-09 stsp 
 379 c3564dfa 2021-07-15 stsp 	if (fstat(p->fd, &idx_sb) != 0) {
 380 6d5a9006 2020-12-16 yzhong 		err = got_error_from_errno2("fstat", relpath);
 381 1124fe40 2021-07-07 stsp 		goto done;
 382 817c5a18 2018-09-09 stsp 	}
 383 c3564dfa 2021-07-15 stsp 	p->len = idx_sb.st_size;
 384 817c5a18 2018-09-09 stsp 	if (p->len < sizeof(p->hdr)) {
 385 817c5a18 2018-09-09 stsp 		err = got_error(GOT_ERR_BAD_PACKIDX);
 386 1124fe40 2021-07-07 stsp 		goto done;
 387 817c5a18 2018-09-09 stsp 	}
 388 817c5a18 2018-09-09 stsp 
 389 6d5a9006 2020-12-16 yzhong 	p->path_packidx = strdup(relpath);
 390 817c5a18 2018-09-09 stsp 	if (p->path_packidx == NULL) {
 391 638f9024 2019-05-13 stsp 		err = got_error_from_errno("strdup");
 392 817c5a18 2018-09-09 stsp 		goto done;
 393 817c5a18 2018-09-09 stsp 	}
 394 817c5a18 2018-09-09 stsp 
 395 817c5a18 2018-09-09 stsp #ifndef GOT_PACK_NO_MMAP
 396 1c28a361 2022-10-25 op 	if (p->len > 0 && p->len <= SIZE_MAX) {
 397 1c28a361 2022-10-25 op 		p->map = mmap(NULL, p->len, PROT_READ, MAP_PRIVATE, p->fd, 0);
 398 1c28a361 2022-10-25 op 		if (p->map == MAP_FAILED) {
 399 1c28a361 2022-10-25 op 			if (errno != ENOMEM) {
 400 1c28a361 2022-10-25 op 				err = got_error_from_errno("mmap");
 401 1c28a361 2022-10-25 op 				goto done;
 402 1c28a361 2022-10-25 op 			}
 403 1c28a361 2022-10-25 op 			p->map = NULL; /* fall back to read(2) */
 404 3a11398b 2019-02-21 stsp 		}
 405 3a11398b 2019-02-21 stsp 	}
 406 817c5a18 2018-09-09 stsp #endif
 407 817c5a18 2018-09-09 stsp 
 408 c3564dfa 2021-07-15 stsp 	err = got_packidx_init_hdr(p, verify, pack_sb.st_size);
 409 817c5a18 2018-09-09 stsp done:
 410 1124fe40 2021-07-07 stsp 	if (err) {
 411 1124fe40 2021-07-07 stsp 		if (p)
 412 1124fe40 2021-07-07 stsp 			got_packidx_close(p);
 413 1124fe40 2021-07-07 stsp 	} else
 414 0a0a3048 2018-01-10 stsp 		*packidx = p;
 415 1124fe40 2021-07-07 stsp 	free(pack_relpath);
 416 0a0a3048 2018-01-10 stsp 	return err;
 417 0a0a3048 2018-01-10 stsp }
 418 0a0a3048 2018-01-10 stsp 
 419 57b35b75 2018-06-22 stsp const struct got_error *
 420 6fd11751 2018-06-04 stsp got_packidx_close(struct got_packidx *packidx)
 421 0a0a3048 2018-01-10 stsp {
 422 57b35b75 2018-06-22 stsp 	const struct got_error *err = NULL;
 423 57b35b75 2018-06-22 stsp 
 424 6fd11751 2018-06-04 stsp 	free(packidx->path_packidx);
 425 fc79a48d 2018-07-09 stsp 	if (packidx->map) {
 426 57b35b75 2018-06-22 stsp 		if (munmap(packidx->map, packidx->len) == -1)
 427 638f9024 2019-05-13 stsp 			err = got_error_from_errno("munmap");
 428 fc79a48d 2018-07-09 stsp 	} else {
 429 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.magic);
 430 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.version);
 431 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.fanout_table);
 432 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.sorted_ids);
 433 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.crc32);
 434 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.offsets);
 435 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.large_offsets);
 436 fc79a48d 2018-07-09 stsp 		free(packidx->hdr.trailer);
 437 57b35b75 2018-06-22 stsp 	}
 438 08578a35 2021-01-22 stsp 	if (close(packidx->fd) == -1 && err == NULL)
 439 638f9024 2019-05-13 stsp 		err = got_error_from_errno("close");
 440 67fd6849 2022-02-13 stsp 	free(packidx->sorted_offsets);
 441 67fd6849 2022-02-13 stsp 	free(packidx->sorted_large_offsets);
 442 0a0a3048 2018-01-10 stsp 	free(packidx);
 443 57b35b75 2018-06-22 stsp 
 444 57b35b75 2018-06-22 stsp 	return err;
 445 a1fd68d8 2018-01-12 stsp }
 446 509c9973 2021-04-10 stsp 
 447 509c9973 2021-04-10 stsp const struct got_error *
 448 aea75d87 2021-07-06 stsp got_packidx_get_packfile_path(char **path_packfile, const char *path_packidx)
 449 509c9973 2021-04-10 stsp {
 450 509c9973 2021-04-10 stsp 	size_t size;
 451 509c9973 2021-04-10 stsp 
 452 509c9973 2021-04-10 stsp 	/* Packfile path contains ".pack" instead of ".idx", so add one byte. */
 453 aea75d87 2021-07-06 stsp 	size = strlen(path_packidx) + 2;
 454 509c9973 2021-04-10 stsp 	if (size < GOT_PACKFILE_NAMELEN + 1)
 455 aea75d87 2021-07-06 stsp 		return got_error_path(path_packidx, GOT_ERR_BAD_PATH);
 456 a1fd68d8 2018-01-12 stsp 
 457 509c9973 2021-04-10 stsp 	*path_packfile = malloc(size);
 458 509c9973 2021-04-10 stsp 	if (*path_packfile == NULL)
 459 509c9973 2021-04-10 stsp 		return got_error_from_errno("malloc");
 460 509c9973 2021-04-10 stsp 
 461 509c9973 2021-04-10 stsp 	/* Copy up to and excluding ".idx". */
 462 aea75d87 2021-07-06 stsp 	if (strlcpy(*path_packfile, path_packidx,
 463 509c9973 2021-04-10 stsp 	    size - strlen(GOT_PACKIDX_SUFFIX) - 1) >= size)
 464 509c9973 2021-04-10 stsp 		return got_error(GOT_ERR_NO_SPACE);
 465 509c9973 2021-04-10 stsp 
 466 509c9973 2021-04-10 stsp 	if (strlcat(*path_packfile, GOT_PACKFILE_SUFFIX, size) >= size)
 467 509c9973 2021-04-10 stsp 		return got_error(GOT_ERR_NO_SPACE);
 468 509c9973 2021-04-10 stsp 
 469 509c9973 2021-04-10 stsp 	return NULL;
 470 509c9973 2021-04-10 stsp }
 471 509c9973 2021-04-10 stsp 
 472 02828bfd 2021-06-22 stsp off_t
 473 02828bfd 2021-06-22 stsp got_packidx_get_object_offset(struct got_packidx *packidx, int idx)
 474 a1fd68d8 2018-01-12 stsp {
 475 78fb0967 2020-09-09 naddy 	uint32_t offset = be32toh(packidx->hdr.offsets[idx]);
 476 a1fd68d8 2018-01-12 stsp 	if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
 477 a1fd68d8 2018-01-12 stsp 		uint64_t loffset;
 478 a1fd68d8 2018-01-12 stsp 		idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
 479 5e6be232 2019-11-08 stsp 		if (idx < 0 || idx >= packidx->nlargeobj ||
 480 6fd11751 2018-06-04 stsp 		    packidx->hdr.large_offsets == NULL)
 481 a1fd68d8 2018-01-12 stsp 			return -1;
 482 78fb0967 2020-09-09 naddy 		loffset = be64toh(packidx->hdr.large_offsets[idx]);
 483 a1fd68d8 2018-01-12 stsp 		return (loffset > INT64_MAX ? -1 : (off_t)loffset);
 484 a1fd68d8 2018-01-12 stsp 	}
 485 a1fd68d8 2018-01-12 stsp 	return (off_t)(offset & GOT_PACKIDX_OFFSET_VAL_MASK);
 486 a1fd68d8 2018-01-12 stsp }
 487 a1fd68d8 2018-01-12 stsp 
 488 1510f469 2018-09-09 stsp int
 489 c0df5966 2021-12-31 stsp got_packidx_get_object_idx(struct got_packidx *packidx,
 490 c0df5966 2021-12-31 stsp     struct got_object_id *id)
 491 a1fd68d8 2018-01-12 stsp {
 492 3093e2d7 2023-02-04 op 	u_int8_t id0 = id->hash[0];
 493 78fb0967 2020-09-09 naddy 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
 494 40aeb19c 2018-06-22 stsp 	int left = 0, right = totobj - 1;
 495 a1fd68d8 2018-01-12 stsp 
 496 a1fd68d8 2018-01-12 stsp 	if (id0 > 0)
 497 78fb0967 2020-09-09 naddy 		left = be32toh(packidx->hdr.fanout_table[id0 - 1]);
 498 a1fd68d8 2018-01-12 stsp 
 499 40aeb19c 2018-06-22 stsp 	while (left <= right) {
 500 57b35b75 2018-06-22 stsp 		struct got_packidx_object_id *oid;
 501 40aeb19c 2018-06-22 stsp 		int i, cmp;
 502 a1fd68d8 2018-01-12 stsp 
 503 40aeb19c 2018-06-22 stsp 		i = ((left + right) / 2);
 504 40aeb19c 2018-06-22 stsp 		oid = &packidx->hdr.sorted_ids[i];
 505 3093e2d7 2023-02-04 op 		cmp = memcmp(id->hash, oid->hash, SHA1_DIGEST_LENGTH);
 506 16dcbf91 2018-04-01 stsp 		if (cmp == 0)
 507 6c00b545 2018-01-17 stsp 			return i;
 508 40aeb19c 2018-06-22 stsp 		else if (cmp > 0)
 509 40aeb19c 2018-06-22 stsp 			left = i + 1;
 510 40aeb19c 2018-06-22 stsp 		else if (cmp < 0)
 511 40aeb19c 2018-06-22 stsp 			right = i - 1;
 512 a1fd68d8 2018-01-12 stsp 	}
 513 a1fd68d8 2018-01-12 stsp 
 514 a1fd68d8 2018-01-12 stsp 	return -1;
 515 67fd6849 2022-02-13 stsp }
 516 67fd6849 2022-02-13 stsp 
 517 67fd6849 2022-02-13 stsp static int
 518 67fd6849 2022-02-13 stsp offset_cmp(const void *pa, const void *pb)
 519 67fd6849 2022-02-13 stsp {
 520 67fd6849 2022-02-13 stsp 	const struct got_pack_offset_index *a, *b;
 521 67fd6849 2022-02-13 stsp 
 522 67fd6849 2022-02-13 stsp 	a = (const struct got_pack_offset_index *)pa;
 523 67fd6849 2022-02-13 stsp 	b = (const struct got_pack_offset_index *)pb;
 524 67fd6849 2022-02-13 stsp 
 525 67fd6849 2022-02-13 stsp 	if (a->offset < b->offset)
 526 67fd6849 2022-02-13 stsp 		return -1;
 527 67fd6849 2022-02-13 stsp 	else if (a->offset > b->offset)
 528 67fd6849 2022-02-13 stsp 		return 1;
 529 67fd6849 2022-02-13 stsp 
 530 67fd6849 2022-02-13 stsp 	return 0;
 531 876c234b 2018-09-10 stsp }
 532 876c234b 2018-09-10 stsp 
 533 67fd6849 2022-02-13 stsp static int
 534 67fd6849 2022-02-13 stsp large_offset_cmp(const void *pa, const void *pb)
 535 67fd6849 2022-02-13 stsp {
 536 67fd6849 2022-02-13 stsp 	const struct got_pack_large_offset_index *a, *b;
 537 67fd6849 2022-02-13 stsp 
 538 67fd6849 2022-02-13 stsp 	a = (const struct got_pack_large_offset_index *)pa;
 539 67fd6849 2022-02-13 stsp 	b = (const struct got_pack_large_offset_index *)pb;
 540 67fd6849 2022-02-13 stsp 
 541 67fd6849 2022-02-13 stsp 	if (a->offset < b->offset)
 542 67fd6849 2022-02-13 stsp 		return -1;
 543 67fd6849 2022-02-13 stsp 	else if (a->offset > b->offset)
 544 67fd6849 2022-02-13 stsp 		return 1;
 545 67fd6849 2022-02-13 stsp 
 546 67fd6849 2022-02-13 stsp 	return 0;
 547 67fd6849 2022-02-13 stsp }
 548 67fd6849 2022-02-13 stsp 
 549 67fd6849 2022-02-13 stsp static const struct got_error *
 550 67fd6849 2022-02-13 stsp build_offset_index(struct got_packidx *p)
 551 67fd6849 2022-02-13 stsp {
 552 67fd6849 2022-02-13 stsp 	uint32_t nobj = be32toh(p->hdr.fanout_table[0xff]);
 553 67fd6849 2022-02-13 stsp 	unsigned int i, j, k;
 554 67fd6849 2022-02-13 stsp 
 555 67fd6849 2022-02-13 stsp 	p->sorted_offsets = calloc(nobj - p->nlargeobj,
 556 67fd6849 2022-02-13 stsp 	    sizeof(p->sorted_offsets[0]));
 557 67fd6849 2022-02-13 stsp 	if (p->sorted_offsets == NULL)
 558 67fd6849 2022-02-13 stsp 		return got_error_from_errno("calloc");
 559 67fd6849 2022-02-13 stsp 
 560 67fd6849 2022-02-13 stsp 	if (p->nlargeobj > 0) {
 561 67fd6849 2022-02-13 stsp 		p->sorted_large_offsets = calloc(p->nlargeobj,
 562 67fd6849 2022-02-13 stsp 		    sizeof(p->sorted_large_offsets[0]));
 563 67fd6849 2022-02-13 stsp 		if (p->sorted_large_offsets == NULL)
 564 67fd6849 2022-02-13 stsp 			return got_error_from_errno("calloc");
 565 67fd6849 2022-02-13 stsp 	}
 566 67fd6849 2022-02-13 stsp 
 567 67fd6849 2022-02-13 stsp 	j = 0;
 568 67fd6849 2022-02-13 stsp 	k = 0;
 569 67fd6849 2022-02-13 stsp 	for (i = 0; i < nobj; i++) {
 570 67fd6849 2022-02-13 stsp 		uint32_t offset = be32toh(p->hdr.offsets[i]);
 571 67fd6849 2022-02-13 stsp 		if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX) {
 572 67fd6849 2022-02-13 stsp 			uint64_t loffset;
 573 67fd6849 2022-02-13 stsp 			uint32_t idx;
 574 67fd6849 2022-02-13 stsp 			idx = offset & GOT_PACKIDX_OFFSET_VAL_MASK;
 575 67fd6849 2022-02-13 stsp 			if (idx >= p->nlargeobj ||
 576 67fd6849 2022-02-13 stsp 			    p->nlargeobj == 0 ||
 577 67fd6849 2022-02-13 stsp 			    p->hdr.large_offsets == NULL)
 578 67fd6849 2022-02-13 stsp 				return got_error(GOT_ERR_BAD_PACKIDX);
 579 67fd6849 2022-02-13 stsp 			loffset = be64toh(p->hdr.large_offsets[idx]);
 580 67fd6849 2022-02-13 stsp 			p->sorted_large_offsets[j].offset = loffset;
 581 67fd6849 2022-02-13 stsp 			p->sorted_large_offsets[j].idx = i;
 582 67fd6849 2022-02-13 stsp 			j++;
 583 67fd6849 2022-02-13 stsp 		} else {
 584 67fd6849 2022-02-13 stsp 			p->sorted_offsets[k].offset = offset;
 585 67fd6849 2022-02-13 stsp 			p->sorted_offsets[k].idx = i;
 586 67fd6849 2022-02-13 stsp 			k++;
 587 67fd6849 2022-02-13 stsp 		}
 588 67fd6849 2022-02-13 stsp 	}
 589 67fd6849 2022-02-13 stsp 	if (j != p->nlargeobj || k != nobj - p->nlargeobj)
 590 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_BAD_PACKIDX);
 591 67fd6849 2022-02-13 stsp 
 592 67fd6849 2022-02-13 stsp 	qsort(p->sorted_offsets, nobj - p->nlargeobj,
 593 67fd6849 2022-02-13 stsp 	    sizeof(p->sorted_offsets[0]), offset_cmp);
 594 67fd6849 2022-02-13 stsp 
 595 67fd6849 2022-02-13 stsp 	if (p->sorted_large_offsets)
 596 67fd6849 2022-02-13 stsp 		qsort(p->sorted_large_offsets, p->nlargeobj,
 597 67fd6849 2022-02-13 stsp 		    sizeof(p->sorted_large_offsets[0]), large_offset_cmp);
 598 67fd6849 2022-02-13 stsp 
 599 67fd6849 2022-02-13 stsp 	return NULL;
 600 67fd6849 2022-02-13 stsp }
 601 67fd6849 2022-02-13 stsp 
 602 876c234b 2018-09-10 stsp const struct got_error *
 603 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 {
 605 67fd6849 2022-02-13 stsp 	const struct got_error *err;
 606 67fd6849 2022-02-13 stsp 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
 607 67fd6849 2022-02-13 stsp 	int i, left, right;
 608 67fd6849 2022-02-13 stsp 
 609 67fd6849 2022-02-13 stsp 	*idx = -1;
 610 67fd6849 2022-02-13 stsp 
 611 67fd6849 2022-02-13 stsp 	if (packidx->sorted_offsets == NULL) {
 612 67fd6849 2022-02-13 stsp 		err = build_offset_index(packidx);
 613 67fd6849 2022-02-13 stsp 		if (err)
 614 67fd6849 2022-02-13 stsp 			return err;
 615 67fd6849 2022-02-13 stsp 	}
 616 67fd6849 2022-02-13 stsp 
 617 67fd6849 2022-02-13 stsp 	if (offset >= 0x7fffffff) {
 618 67fd6849 2022-02-13 stsp 		uint64_t lo;
 619 67fd6849 2022-02-13 stsp 		left = 0, right = packidx->nlargeobj - 1;
 620 67fd6849 2022-02-13 stsp 		while (left <= right) {
 621 67fd6849 2022-02-13 stsp 			i = ((left + right) / 2);
 622 67fd6849 2022-02-13 stsp 			lo = packidx->sorted_large_offsets[i].offset;
 623 67fd6849 2022-02-13 stsp 			if (lo == offset) {
 624 67fd6849 2022-02-13 stsp 				*idx = packidx->sorted_large_offsets[i].idx;
 625 67fd6849 2022-02-13 stsp 				break;
 626 67fd6849 2022-02-13 stsp 			} else if (offset > lo)
 627 67fd6849 2022-02-13 stsp 				left = i + 1;
 628 67fd6849 2022-02-13 stsp 			else if (offset < lo)
 629 67fd6849 2022-02-13 stsp 				right = i - 1;
 630 67fd6849 2022-02-13 stsp 		}
 631 67fd6849 2022-02-13 stsp 	} else {
 632 67fd6849 2022-02-13 stsp 		uint32_t o;
 633 67fd6849 2022-02-13 stsp 		left = 0, right = totobj - packidx->nlargeobj - 1;
 634 67fd6849 2022-02-13 stsp 		while (left <= right) {
 635 67fd6849 2022-02-13 stsp 			i = ((left + right) / 2);
 636 67fd6849 2022-02-13 stsp 			o = packidx->sorted_offsets[i].offset;
 637 67fd6849 2022-02-13 stsp 			if (o == offset) {
 638 67fd6849 2022-02-13 stsp 				*idx = packidx->sorted_offsets[i].idx;
 639 67fd6849 2022-02-13 stsp 				break;
 640 67fd6849 2022-02-13 stsp 			} else if (offset > o)
 641 67fd6849 2022-02-13 stsp 				left = i + 1;
 642 67fd6849 2022-02-13 stsp 			else if (offset < o)
 643 67fd6849 2022-02-13 stsp 				right = i - 1;
 644 67fd6849 2022-02-13 stsp 		}
 645 67fd6849 2022-02-13 stsp 	}
 646 67fd6849 2022-02-13 stsp 
 647 67fd6849 2022-02-13 stsp 	return NULL;
 648 67fd6849 2022-02-13 stsp }
 649 67fd6849 2022-02-13 stsp 
 650 67fd6849 2022-02-13 stsp const struct got_error *
 651 67fd6849 2022-02-13 stsp got_packidx_get_object_id(struct got_object_id *id,
 652 67fd6849 2022-02-13 stsp     struct got_packidx *packidx, int idx)
 653 67fd6849 2022-02-13 stsp {
 654 67fd6849 2022-02-13 stsp 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
 655 67fd6849 2022-02-13 stsp 	struct got_packidx_object_id *oid;
 656 67fd6849 2022-02-13 stsp 
 657 67fd6849 2022-02-13 stsp 	if (idx < 0 || idx >= totobj)
 658 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_NO_OBJ);
 659 67fd6849 2022-02-13 stsp 
 660 67fd6849 2022-02-13 stsp 	oid = &packidx->hdr.sorted_ids[idx];
 661 3093e2d7 2023-02-04 op 	memcpy(id->hash, oid->hash, SHA1_DIGEST_LENGTH);
 662 67fd6849 2022-02-13 stsp 	return NULL;
 663 67fd6849 2022-02-13 stsp }
 664 67fd6849 2022-02-13 stsp 
 665 67fd6849 2022-02-13 stsp const struct got_error *
 666 dd88155e 2019-06-29 stsp got_packidx_match_id_str_prefix(struct got_object_id_queue *matched_ids,
 667 4277420a 2019-06-29 stsp     struct got_packidx *packidx, const char *id_str_prefix)
 668 e09a504c 2019-06-28 stsp {
 669 dd88155e 2019-06-29 stsp 	const struct got_error *err = NULL;
 670 4277420a 2019-06-29 stsp 	u_int8_t id0;
 671 78fb0967 2020-09-09 naddy 	uint32_t totobj = be32toh(packidx->hdr.fanout_table[0xff]);
 672 4277420a 2019-06-29 stsp 	char hex[3];
 673 4277420a 2019-06-29 stsp 	size_t prefix_len = strlen(id_str_prefix);
 674 e09a504c 2019-06-28 stsp 	struct got_packidx_object_id *oid;
 675 404bde06 2022-01-03 stsp 	uint32_t i = 0;
 676 e09a504c 2019-06-28 stsp 
 677 dbdddfee 2021-06-23 naddy 	STAILQ_INIT(matched_ids);
 678 4277420a 2019-06-29 stsp 
 679 4277420a 2019-06-29 stsp 	if (prefix_len < 2)
 680 6dd1ece6 2019-11-10 stsp 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
 681 4277420a 2019-06-29 stsp 
 682 4277420a 2019-06-29 stsp 	hex[0] = id_str_prefix[0];
 683 4277420a 2019-06-29 stsp 	hex[1] = id_str_prefix[1];
 684 4277420a 2019-06-29 stsp 	hex[2] = '\0';
 685 4277420a 2019-06-29 stsp 	if (!got_parse_xdigit(&id0, hex))
 686 6dd1ece6 2019-11-10 stsp 		return got_error_path(id_str_prefix, GOT_ERR_BAD_OBJ_ID_STR);
 687 4277420a 2019-06-29 stsp 
 688 404bde06 2022-01-03 stsp 	if (id0 > 0)
 689 404bde06 2022-01-03 stsp 		i = be32toh(packidx->hdr.fanout_table[id0 - 1]);
 690 4277420a 2019-06-29 stsp 	oid = &packidx->hdr.sorted_ids[i];
 691 3093e2d7 2023-02-04 op 	while (i < totobj && oid->hash[0] == id0) {
 692 4277420a 2019-06-29 stsp 		char id_str[SHA1_DIGEST_STRING_LENGTH];
 693 dd88155e 2019-06-29 stsp 		struct got_object_qid *qid;
 694 4277420a 2019-06-29 stsp 		int cmp;
 695 4277420a 2019-06-29 stsp 
 696 3093e2d7 2023-02-04 op 		if (!got_sha1_digest_to_str(oid->hash, id_str, sizeof(id_str)))
 697 4277420a 2019-06-29 stsp 		        return got_error(GOT_ERR_NO_SPACE);
 698 4277420a 2019-06-29 stsp 
 699 4277420a 2019-06-29 stsp 		cmp = strncmp(id_str, id_str_prefix, prefix_len);
 700 4277420a 2019-06-29 stsp 		if (cmp < 0) {
 701 4277420a 2019-06-29 stsp 			oid = &packidx->hdr.sorted_ids[++i];
 702 4277420a 2019-06-29 stsp 			continue;
 703 4277420a 2019-06-29 stsp 		} else if (cmp > 0)
 704 e09a504c 2019-06-28 stsp 			break;
 705 4277420a 2019-06-29 stsp 
 706 dd88155e 2019-06-29 stsp 		err = got_object_qid_alloc_partial(&qid);
 707 dd88155e 2019-06-29 stsp 		if (err)
 708 dd88155e 2019-06-29 stsp 			break;
 709 3093e2d7 2023-02-04 op 		memcpy(qid->id.hash, oid->hash, SHA1_DIGEST_LENGTH);
 710 dbdddfee 2021-06-23 naddy 		STAILQ_INSERT_TAIL(matched_ids, qid, entry);
 711 4277420a 2019-06-29 stsp 
 712 4277420a 2019-06-29 stsp 		oid = &packidx->hdr.sorted_ids[++i];
 713 e09a504c 2019-06-28 stsp 	}
 714 e09a504c 2019-06-28 stsp 
 715 0adc7bcc 2019-06-29 stsp 	if (err)
 716 0adc7bcc 2019-06-29 stsp 		got_object_id_queue_free(matched_ids);
 717 3d589bee 2022-06-25 stsp 	return err;
 718 3d589bee 2022-06-25 stsp }
 719 3d589bee 2022-06-25 stsp 
 720 3d589bee 2022-06-25 stsp static void
 721 3d589bee 2022-06-25 stsp set_max_datasize(void)
 722 3d589bee 2022-06-25 stsp {
 723 3d589bee 2022-06-25 stsp 	struct rlimit rl;
 724 3d589bee 2022-06-25 stsp 
 725 3d589bee 2022-06-25 stsp 	if (getrlimit(RLIMIT_DATA, &rl) != 0)
 726 3d589bee 2022-06-25 stsp 		return;
 727 3d589bee 2022-06-25 stsp 
 728 3d589bee 2022-06-25 stsp 	rl.rlim_cur = rl.rlim_max;
 729 3d589bee 2022-06-25 stsp 	setrlimit(RLIMIT_DATA, &rl);
 730 3d589bee 2022-06-25 stsp }
 731 3d589bee 2022-06-25 stsp 
 732 3d589bee 2022-06-25 stsp const struct got_error *
 733 3d589bee 2022-06-25 stsp got_pack_start_privsep_child(struct got_pack *pack, struct got_packidx *packidx)
 734 3d589bee 2022-06-25 stsp {
 735 3d589bee 2022-06-25 stsp 	const struct got_error *err = NULL;
 736 3d589bee 2022-06-25 stsp 	int imsg_fds[2];
 737 3d589bee 2022-06-25 stsp 	pid_t pid;
 738 3d589bee 2022-06-25 stsp 	struct imsgbuf *ibuf;
 739 3d589bee 2022-06-25 stsp 
 740 3d589bee 2022-06-25 stsp 	ibuf = calloc(1, sizeof(*ibuf));
 741 3d589bee 2022-06-25 stsp 	if (ibuf == NULL)
 742 3d589bee 2022-06-25 stsp 		return got_error_from_errno("calloc");
 743 3d589bee 2022-06-25 stsp 
 744 3d589bee 2022-06-25 stsp 	pack->privsep_child = calloc(1, sizeof(*pack->privsep_child));
 745 3d589bee 2022-06-25 stsp 	if (pack->privsep_child == NULL) {
 746 3d589bee 2022-06-25 stsp 		err = got_error_from_errno("calloc");
 747 3d589bee 2022-06-25 stsp 		free(ibuf);
 748 3d589bee 2022-06-25 stsp 		return err;
 749 3d589bee 2022-06-25 stsp 	}
 750 3d589bee 2022-06-25 stsp 	pack->child_has_tempfiles = 0;
 751 3d589bee 2022-06-25 stsp 	pack->child_has_delta_outfd = 0;
 752 3d589bee 2022-06-25 stsp 
 753 3d589bee 2022-06-25 stsp 	if (socketpair(AF_UNIX, SOCK_STREAM, PF_UNSPEC, imsg_fds) == -1) {
 754 3d589bee 2022-06-25 stsp 		err = got_error_from_errno("socketpair");
 755 3d589bee 2022-06-25 stsp 		goto done;
 756 3d589bee 2022-06-25 stsp 	}
 757 3d589bee 2022-06-25 stsp 
 758 3d589bee 2022-06-25 stsp 	pid = fork();
 759 3d589bee 2022-06-25 stsp 	if (pid == -1) {
 760 3d589bee 2022-06-25 stsp 		err = got_error_from_errno("fork");
 761 3d589bee 2022-06-25 stsp 		goto done;
 762 3d589bee 2022-06-25 stsp 	} else if (pid == 0) {
 763 3d589bee 2022-06-25 stsp 		set_max_datasize();
 764 3d589bee 2022-06-25 stsp 		got_privsep_exec_child(imsg_fds, GOT_PATH_PROG_READ_PACK,
 765 3d589bee 2022-06-25 stsp 		    pack->path_packfile);
 766 3d589bee 2022-06-25 stsp 		/* not reached */
 767 3d589bee 2022-06-25 stsp 	}
 768 3d589bee 2022-06-25 stsp 
 769 3d589bee 2022-06-25 stsp 	if (close(imsg_fds[1]) == -1)
 770 3d589bee 2022-06-25 stsp 		return got_error_from_errno("close");
 771 3d589bee 2022-06-25 stsp 	pack->privsep_child->imsg_fd = imsg_fds[0];
 772 3d589bee 2022-06-25 stsp 	pack->privsep_child->pid = pid;
 773 3d589bee 2022-06-25 stsp 	imsg_init(ibuf, imsg_fds[0]);
 774 3d589bee 2022-06-25 stsp 	pack->privsep_child->ibuf = ibuf;
 775 3d589bee 2022-06-25 stsp 
 776 3d589bee 2022-06-25 stsp 	err = got_privsep_init_pack_child(ibuf, pack, packidx);
 777 3d589bee 2022-06-25 stsp 	if (err) {
 778 3d589bee 2022-06-25 stsp 		const struct got_error *child_err;
 779 3d589bee 2022-06-25 stsp 		err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
 780 3d589bee 2022-06-25 stsp 		child_err = got_privsep_wait_for_child(
 781 3d589bee 2022-06-25 stsp 		    pack->privsep_child->pid);
 782 3d589bee 2022-06-25 stsp 		if (child_err && err == NULL)
 783 3d589bee 2022-06-25 stsp 			err = child_err;
 784 3d589bee 2022-06-25 stsp 	}
 785 3d589bee 2022-06-25 stsp done:
 786 3d589bee 2022-06-25 stsp 	if (err) {
 787 3d589bee 2022-06-25 stsp 		free(ibuf);
 788 3d589bee 2022-06-25 stsp 		free(pack->privsep_child);
 789 3d589bee 2022-06-25 stsp 		pack->privsep_child = NULL;
 790 3d589bee 2022-06-25 stsp 	}
 791 dd88155e 2019-06-29 stsp 	return err;
 792 e09a504c 2019-06-28 stsp }
 793 e09a504c 2019-06-28 stsp 
 794 b4f37570 2021-06-19 stsp static const struct got_error *
 795 b4f37570 2021-06-19 stsp pack_stop_privsep_child(struct got_pack *pack)
 796 876c234b 2018-09-10 stsp {
 797 74ef8aae 2022-10-24 stsp 	const struct got_error *err = NULL, *close_err = NULL;
 798 876c234b 2018-09-10 stsp 
 799 876c234b 2018-09-10 stsp 	if (pack->privsep_child == NULL)
 800 876c234b 2018-09-10 stsp 		return NULL;
 801 876c234b 2018-09-10 stsp 
 802 876c234b 2018-09-10 stsp 	err = got_privsep_send_stop(pack->privsep_child->imsg_fd);
 803 96732e0b 2018-11-11 stsp 	if (err)
 804 96732e0b 2018-11-11 stsp 		return err;
 805 74ef8aae 2022-10-24 stsp 	if (close(pack->privsep_child->imsg_fd) == -1)
 806 74ef8aae 2022-10-24 stsp 		close_err = got_error_from_errno("close");
 807 96732e0b 2018-11-11 stsp 	err = got_privsep_wait_for_child(pack->privsep_child->pid);
 808 74ef8aae 2022-10-24 stsp 	if (close_err && err == NULL)
 809 74ef8aae 2022-10-24 stsp 		err = close_err;
 810 74ef8aae 2022-10-24 stsp 	imsg_clear(pack->privsep_child->ibuf);
 811 cc2a8ef4 2021-06-19 tracey 	free(pack->privsep_child->ibuf);
 812 876c234b 2018-09-10 stsp 	free(pack->privsep_child);
 813 876c234b 2018-09-10 stsp 	pack->privsep_child = NULL;
 814 876c234b 2018-09-10 stsp 	return err;
 815 7e656b93 2018-03-17 stsp }
 816 7e656b93 2018-03-17 stsp 
 817 d7464085 2018-07-09 stsp const struct got_error *
 818 7e656b93 2018-03-17 stsp got_pack_close(struct got_pack *pack)
 819 7e656b93 2018-03-17 stsp {
 820 d7464085 2018-07-09 stsp 	const struct got_error *err = NULL;
 821 d7464085 2018-07-09 stsp 
 822 b4f37570 2021-06-19 stsp 	err = pack_stop_privsep_child(pack);
 823 876c234b 2018-09-10 stsp 	if (pack->map && munmap(pack->map, pack->filesize) == -1 && !err)
 824 638f9024 2019-05-13 stsp 		err = got_error_from_errno("munmap");
 825 08578a35 2021-01-22 stsp 	if (pack->fd != -1 && close(pack->fd) == -1 && err == NULL)
 826 638f9024 2019-05-13 stsp 		err = got_error_from_errno("close");
 827 8b2180d4 2018-04-26 stsp 	pack->fd = -1;
 828 4589e373 2018-03-17 stsp 	free(pack->path_packfile);
 829 4589e373 2018-03-17 stsp 	pack->path_packfile = NULL;
 830 4589e373 2018-03-17 stsp 	pack->filesize = 0;
 831 ab2f42e7 2019-11-10 stsp 	if (pack->delta_cache) {
 832 ab2f42e7 2019-11-10 stsp 		got_delta_cache_free(pack->delta_cache);
 833 ab2f42e7 2019-11-10 stsp 		pack->delta_cache = NULL;
 834 ab2f42e7 2019-11-10 stsp 	}
 835 7e656b93 2018-03-17 stsp 
 836 57160834 2022-05-31 stsp 	/*
 837 57160834 2022-05-31 stsp 	 * Leave accumfd and basefd alone. They are managed by the
 838 57160834 2022-05-31 stsp 	 * repository layer and can be reused.
 839 57160834 2022-05-31 stsp 	 */
 840 57160834 2022-05-31 stsp 
 841 c7fe698a 2018-01-23 stsp 	return err;
 842 c7fe698a 2018-01-23 stsp }
 843 c7fe698a 2018-01-23 stsp 
 844 668a20f6 2020-03-18 stsp const struct got_error *
 845 668a20f6 2020-03-18 stsp got_pack_parse_object_type_and_size(uint8_t *type, uint64_t *size, size_t *len,
 846 d7464085 2018-07-09 stsp     struct got_pack *pack, off_t offset)
 847 a487c1d0 2018-01-14 stsp {
 848 a487c1d0 2018-01-14 stsp 	uint8_t t = 0;
 849 a487c1d0 2018-01-14 stsp 	uint64_t s = 0;
 850 a487c1d0 2018-01-14 stsp 	uint8_t sizeN;
 851 d7464085 2018-07-09 stsp 	size_t mapoff = 0;
 852 a487c1d0 2018-01-14 stsp 	int i = 0;
 853 d7464085 2018-07-09 stsp 
 854 d7464085 2018-07-09 stsp 	*len = 0;
 855 d7464085 2018-07-09 stsp 
 856 d7464085 2018-07-09 stsp 	if (offset >= pack->filesize)
 857 d7464085 2018-07-09 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
 858 d7464085 2018-07-09 stsp 
 859 d7464085 2018-07-09 stsp 	if (pack->map) {
 860 ad4cc361 2022-10-27 op 		if (offset > SIZE_MAX) {
 861 ad4cc361 2022-10-27 op 			return got_error_fmt(GOT_ERR_PACK_OFFSET,
 862 ad4cc361 2022-10-27 op 			    "offset %lld overflows size_t",
 863 ad4cc361 2022-10-27 op 			    (long long)offset);
 864 ad4cc361 2022-10-27 op 		}
 865 ad4cc361 2022-10-27 op 
 866 d7464085 2018-07-09 stsp 		mapoff = (size_t)offset;
 867 d7464085 2018-07-09 stsp 	} else {
 868 d7464085 2018-07-09 stsp 		if (lseek(pack->fd, offset, SEEK_SET) == -1)
 869 638f9024 2019-05-13 stsp 			return got_error_from_errno("lseek");
 870 d7464085 2018-07-09 stsp 	}
 871 a487c1d0 2018-01-14 stsp 
 872 a1fd68d8 2018-01-12 stsp 	do {
 873 a1fd68d8 2018-01-12 stsp 		/* We do not support size values which don't fit in 64 bit. */
 874 a487c1d0 2018-01-14 stsp 		if (i > 9)
 875 09ee8ded 2022-10-20 stsp 			return got_error_fmt(GOT_ERR_OBJ_TOO_LARGE,
 876 e082ed67 2022-10-24 naddy 			    "packfile offset %lld", (long long)offset);
 877 a1fd68d8 2018-01-12 stsp 
 878 d7464085 2018-07-09 stsp 		if (pack->map) {
 879 3976db15 2022-01-10 stsp 			if (mapoff + sizeof(sizeN) >= pack->filesize)
 880 3976db15 2022-01-10 stsp 				return got_error(GOT_ERR_BAD_PACKFILE);
 881 d7464085 2018-07-09 stsp 			sizeN = *(pack->map + mapoff);
 882 d7464085 2018-07-09 stsp 			mapoff += sizeof(sizeN);
 883 d7464085 2018-07-09 stsp 		} else {
 884 d7464085 2018-07-09 stsp 			ssize_t n = read(pack->fd, &sizeN, sizeof(sizeN));
 885 d7464085 2018-07-09 stsp 			if (n < 0)
 886 638f9024 2019-05-13 stsp 				return got_error_from_errno("read");
 887 d7464085 2018-07-09 stsp 			if (n != sizeof(sizeN))
 888 d7464085 2018-07-09 stsp 				return got_error(GOT_ERR_BAD_PACKFILE);
 889 d7464085 2018-07-09 stsp 		}
 890 d7464085 2018-07-09 stsp 		*len += sizeof(sizeN);
 891 8251fdbc 2018-01-12 stsp 
 892 a1fd68d8 2018-01-12 stsp 		if (i == 0) {
 893 a487c1d0 2018-01-14 stsp 			t = (sizeN & GOT_PACK_OBJ_SIZE0_TYPE_MASK) >>
 894 a1fd68d8 2018-01-12 stsp 			    GOT_PACK_OBJ_SIZE0_TYPE_MASK_SHIFT;
 895 a487c1d0 2018-01-14 stsp 			s = (sizeN & GOT_PACK_OBJ_SIZE0_VAL_MASK);
 896 a1fd68d8 2018-01-12 stsp 		} else {
 897 a1fd68d8 2018-01-12 stsp 			size_t shift = 4 + 7 * (i - 1);
 898 a487c1d0 2018-01-14 stsp 			s |= ((sizeN & GOT_PACK_OBJ_SIZE_VAL_MASK) << shift);
 899 a1fd68d8 2018-01-12 stsp 		}
 900 a1fd68d8 2018-01-12 stsp 		i++;
 901 a1fd68d8 2018-01-12 stsp 	} while (sizeN & GOT_PACK_OBJ_SIZE_MORE);
 902 a1fd68d8 2018-01-12 stsp 
 903 a487c1d0 2018-01-14 stsp 	*type = t;
 904 a487c1d0 2018-01-14 stsp 	*size = s;
 905 a487c1d0 2018-01-14 stsp 	return NULL;
 906 0a0a3048 2018-01-10 stsp }
 907 c54542a0 2018-01-13 stsp 
 908 a1fd68d8 2018-01-12 stsp static const struct got_error *
 909 5f25cc85 2019-11-26 stsp open_plain_object(struct got_object **obj, struct got_object_id *id,
 910 5f25cc85 2019-11-26 stsp     uint8_t type, off_t offset, size_t size, int idx)
 911 6ccb713b 2018-01-19 stsp {
 912 6ccb713b 2018-01-19 stsp 	*obj = calloc(1, sizeof(**obj));
 913 6ccb713b 2018-01-19 stsp 	if (*obj == NULL)
 914 638f9024 2019-05-13 stsp 		return got_error_from_errno("calloc");
 915 6ccb713b 2018-01-19 stsp 
 916 6ccb713b 2018-01-19 stsp 	(*obj)->type = type;
 917 6ccb713b 2018-01-19 stsp 	(*obj)->flags = GOT_OBJ_FLAG_PACKED;
 918 c59b3346 2018-09-11 stsp 	(*obj)->pack_idx = idx;
 919 6ccb713b 2018-01-19 stsp 	(*obj)->hdrlen = 0;
 920 6ccb713b 2018-01-19 stsp 	(*obj)->size = size;
 921 106807b4 2018-09-15 stsp 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
 922 6ccb713b 2018-01-19 stsp 	(*obj)->pack_offset = offset;
 923 b107e67f 2018-01-19 stsp 
 924 b107e67f 2018-01-19 stsp 	return NULL;
 925 b107e67f 2018-01-19 stsp }
 926 b107e67f 2018-01-19 stsp 
 927 b107e67f 2018-01-19 stsp static const struct got_error *
 928 d7464085 2018-07-09 stsp parse_negative_offset(int64_t *offset, size_t *len, struct got_pack *pack,
 929 d7464085 2018-07-09 stsp     off_t delta_offset)
 930 b107e67f 2018-01-19 stsp {
 931 b107e67f 2018-01-19 stsp 	int64_t o = 0;
 932 b107e67f 2018-01-19 stsp 	uint8_t offN;
 933 b107e67f 2018-01-19 stsp 	int i = 0;
 934 b107e67f 2018-01-19 stsp 
 935 a0de39f3 2019-08-09 stsp 	*offset = 0;
 936 d7464085 2018-07-09 stsp 	*len = 0;
 937 d7464085 2018-07-09 stsp 
 938 b107e67f 2018-01-19 stsp 	do {
 939 b107e67f 2018-01-19 stsp 		/* We do not support offset values which don't fit in 64 bit. */
 940 b107e67f 2018-01-19 stsp 		if (i > 8)
 941 b107e67f 2018-01-19 stsp 			return got_error(GOT_ERR_NO_SPACE);
 942 b107e67f 2018-01-19 stsp 
 943 d7464085 2018-07-09 stsp 		if (pack->map) {
 944 d7464085 2018-07-09 stsp 			size_t mapoff;
 945 ad4cc361 2022-10-27 op 
 946 ad4cc361 2022-10-27 op 			if (delta_offset + *len > SIZE_MAX) {
 947 ad4cc361 2022-10-27 op 				return got_error_fmt(GOT_ERR_PACK_OFFSET,
 948 ad4cc361 2022-10-27 op 				    "mapoff %lld would overflow size_t",
 949 ad4cc361 2022-10-27 op 				    (long long)delta_offset + *len);
 950 ad4cc361 2022-10-27 op 			}
 951 ad4cc361 2022-10-27 op 
 952 d7464085 2018-07-09 stsp 			mapoff = (size_t)delta_offset + *len;
 953 3976db15 2022-01-10 stsp 			if (mapoff + sizeof(offN) >= pack->filesize)
 954 3976db15 2022-01-10 stsp 				return got_error(GOT_ERR_PACK_OFFSET);
 955 d7464085 2018-07-09 stsp 			offN = *(pack->map + mapoff);
 956 d7464085 2018-07-09 stsp 		} else {
 957 d7464085 2018-07-09 stsp 			ssize_t n;
 958 d7464085 2018-07-09 stsp 			n = read(pack->fd, &offN, sizeof(offN));
 959 d7464085 2018-07-09 stsp 			if (n < 0)
 960 638f9024 2019-05-13 stsp 				return got_error_from_errno("read");
 961 d7464085 2018-07-09 stsp 			if (n != sizeof(offN))
 962 d7464085 2018-07-09 stsp 				return got_error(GOT_ERR_BAD_PACKFILE);
 963 d7464085 2018-07-09 stsp 		}
 964 d7464085 2018-07-09 stsp 		*len += sizeof(offN);
 965 b107e67f 2018-01-19 stsp 
 966 b107e67f 2018-01-19 stsp 		if (i == 0)
 967 b107e67f 2018-01-19 stsp 			o = (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
 968 b107e67f 2018-01-19 stsp 		else {
 969 cecc778e 2018-01-23 stsp 			o++;
 970 b107e67f 2018-01-19 stsp 			o <<= 7;
 971 cecc778e 2018-01-23 stsp 			o += (offN & GOT_PACK_OBJ_DELTA_OFF_VAL_MASK);
 972 b107e67f 2018-01-19 stsp 		}
 973 b107e67f 2018-01-19 stsp 		i++;
 974 b107e67f 2018-01-19 stsp 	} while (offN & GOT_PACK_OBJ_DELTA_OFF_MORE);
 975 6ccb713b 2018-01-19 stsp 
 976 b107e67f 2018-01-19 stsp 	*offset = o;
 977 6ccb713b 2018-01-19 stsp 	return NULL;
 978 6ccb713b 2018-01-19 stsp }
 979 6ccb713b 2018-01-19 stsp 
 980 668a20f6 2020-03-18 stsp const struct got_error *
 981 c0df5966 2021-12-31 stsp got_pack_parse_offset_delta(off_t *base_offset, size_t *len,
 982 24d916d2 2022-10-24 op     struct got_pack *pack, off_t offset, size_t tslen)
 983 b107e67f 2018-01-19 stsp {
 984 96f5e8b3 2018-01-23 stsp 	const struct got_error *err;
 985 b107e67f 2018-01-19 stsp 	int64_t negoffset;
 986 b107e67f 2018-01-19 stsp 	size_t negofflen;
 987 b107e67f 2018-01-19 stsp 
 988 d7464085 2018-07-09 stsp 	*len = 0;
 989 d7464085 2018-07-09 stsp 
 990 d7464085 2018-07-09 stsp 	err = parse_negative_offset(&negoffset, &negofflen, pack,
 991 d7464085 2018-07-09 stsp 	    offset + tslen);
 992 b107e67f 2018-01-19 stsp 	if (err)
 993 b107e67f 2018-01-19 stsp 		return err;
 994 b107e67f 2018-01-19 stsp 
 995 b107e67f 2018-01-19 stsp 	/* Compute the base object's offset (must be in the same pack file). */
 996 96f5e8b3 2018-01-23 stsp 	*base_offset = (offset - negoffset);
 997 96f5e8b3 2018-01-23 stsp 	if (*base_offset <= 0)
 998 b107e67f 2018-01-19 stsp 		return got_error(GOT_ERR_BAD_PACKFILE);
 999 b107e67f 2018-01-19 stsp 
1000 d7464085 2018-07-09 stsp 	*len = negofflen;
1001 96f5e8b3 2018-01-23 stsp 	return NULL;
1002 96f5e8b3 2018-01-23 stsp }
1003 96f5e8b3 2018-01-23 stsp 
1004 0e22967e 2018-02-11 stsp static const struct got_error *
1005 42c69117 2019-11-10 stsp read_delta_data(uint8_t **delta_buf, size_t *delta_len,
1006 2d9e6abf 2022-05-04 stsp     size_t *delta_compressed_len, size_t delta_data_offset,
1007 2d9e6abf 2022-05-04 stsp     struct got_pack *pack)
1008 42c69117 2019-11-10 stsp {
1009 42c69117 2019-11-10 stsp 	const struct got_error *err = NULL;
1010 2d9e6abf 2022-05-04 stsp 	size_t consumed = 0;
1011 42c69117 2019-11-10 stsp 
1012 42c69117 2019-11-10 stsp 	if (pack->map) {
1013 42c69117 2019-11-10 stsp 		if (delta_data_offset >= pack->filesize)
1014 42c69117 2019-11-10 stsp 			return got_error(GOT_ERR_PACK_OFFSET);
1015 2e5a6fad 2020-03-18 stsp 		err = got_inflate_to_mem_mmap(delta_buf, delta_len,
1016 2d9e6abf 2022-05-04 stsp 		    &consumed, NULL, pack->map, delta_data_offset,
1017 2e5a6fad 2020-03-18 stsp 		    pack->filesize - delta_data_offset);
1018 2d9e6abf 2022-05-04 stsp 		if (err)
1019 2d9e6abf 2022-05-04 stsp 			return err;
1020 42c69117 2019-11-10 stsp 	} else {
1021 42c69117 2019-11-10 stsp 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1)
1022 42c69117 2019-11-10 stsp 			return got_error_from_errno("lseek");
1023 2d9e6abf 2022-05-04 stsp 		err = got_inflate_to_mem_fd(delta_buf, delta_len,
1024 2d9e6abf 2022-05-04 stsp 		    &consumed, NULL, 0, pack->fd);
1025 2d9e6abf 2022-05-04 stsp 		if (err)
1026 2d9e6abf 2022-05-04 stsp 			return err;
1027 42c69117 2019-11-10 stsp 	}
1028 2d9e6abf 2022-05-04 stsp 
1029 2d9e6abf 2022-05-04 stsp 	if (delta_compressed_len)
1030 2d9e6abf 2022-05-04 stsp 		*delta_compressed_len = consumed;
1031 2d9e6abf 2022-05-04 stsp 
1032 2d9e6abf 2022-05-04 stsp 	return NULL;
1033 42c69117 2019-11-10 stsp }
1034 a3500804 2018-01-23 stsp 
1035 96f5e8b3 2018-01-23 stsp static const struct got_error *
1036 c336f889 2018-07-23 stsp add_delta(struct got_delta_chain *deltas, off_t delta_offset, size_t tslen,
1037 e5792992 2022-10-28 op     int delta_type, size_t delta_size, off_t delta_data_offset)
1038 bdd2fbb3 2018-02-11 stsp {
1039 bdd2fbb3 2018-02-11 stsp 	struct got_delta *delta;
1040 bdd2fbb3 2018-02-11 stsp 
1041 c336f889 2018-07-23 stsp 	delta = got_delta_open(delta_offset, tslen, delta_type, delta_size,
1042 42c69117 2019-11-10 stsp 	    delta_data_offset);
1043 bdd2fbb3 2018-02-11 stsp 	if (delta == NULL)
1044 638f9024 2019-05-13 stsp 		return got_error_from_errno("got_delta_open");
1045 bdd2fbb3 2018-02-11 stsp 	/* delta is freed in got_object_close() */
1046 bdd2fbb3 2018-02-11 stsp 	deltas->nentries++;
1047 dbdddfee 2021-06-23 naddy 	STAILQ_INSERT_HEAD(&deltas->entries, delta, entry);
1048 bdd2fbb3 2018-02-11 stsp 	return NULL;
1049 bdd2fbb3 2018-02-11 stsp }
1050 bdd2fbb3 2018-02-11 stsp 
1051 bdd2fbb3 2018-02-11 stsp static const struct got_error *
1052 6b9c9673 2018-01-23 stsp resolve_offset_delta(struct got_delta_chain *deltas,
1053 c8ecd499 2018-09-09 stsp     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1054 c8ecd499 2018-09-09 stsp     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1055 a3500804 2018-01-23 stsp {
1056 a3500804 2018-01-23 stsp 	const struct got_error *err;
1057 c3703302 2018-01-23 stsp 	off_t base_offset;
1058 c3703302 2018-01-23 stsp 	uint8_t base_type;
1059 c3703302 2018-01-23 stsp 	uint64_t base_size;
1060 c3703302 2018-01-23 stsp 	size_t base_tslen;
1061 bdd2fbb3 2018-02-11 stsp 	off_t delta_data_offset;
1062 42c69117 2019-11-10 stsp 	size_t consumed;
1063 a3500804 2018-01-23 stsp 
1064 668a20f6 2020-03-18 stsp 	err = got_pack_parse_offset_delta(&base_offset, &consumed, pack,
1065 d7464085 2018-07-09 stsp 	    delta_offset, tslen);
1066 a3500804 2018-01-23 stsp 	if (err)
1067 a3500804 2018-01-23 stsp 		return err;
1068 a3500804 2018-01-23 stsp 
1069 d7464085 2018-07-09 stsp 	delta_data_offset = delta_offset + tslen + consumed;
1070 d7464085 2018-07-09 stsp 	if (delta_data_offset >= pack->filesize)
1071 d7464085 2018-07-09 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1072 a53d2f13 2018-03-17 stsp 
1073 d7464085 2018-07-09 stsp 	if (pack->map == NULL) {
1074 d7464085 2018-07-09 stsp 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1075 d7464085 2018-07-09 stsp 		if (delta_data_offset == -1)
1076 638f9024 2019-05-13 stsp 			return got_error_from_errno("lseek");
1077 d7464085 2018-07-09 stsp 	}
1078 bdd2fbb3 2018-02-11 stsp 
1079 c336f889 2018-07-23 stsp 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1080 e5792992 2022-10-28 op 	    delta_data_offset);
1081 bdd2fbb3 2018-02-11 stsp 	if (err)
1082 1a35c1bc 2019-05-22 stsp 		return err;
1083 bdd2fbb3 2018-02-11 stsp 
1084 c3703302 2018-01-23 stsp 	/* An offset delta must be in the same packfile. */
1085 1a35c1bc 2019-05-22 stsp 	if (base_offset >= pack->filesize)
1086 1a35c1bc 2019-05-22 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1087 b107e67f 2018-01-19 stsp 
1088 668a20f6 2020-03-18 stsp 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1089 668a20f6 2020-03-18 stsp 	    &base_tslen, pack, base_offset);
1090 b107e67f 2018-01-19 stsp 	if (err)
1091 1a35c1bc 2019-05-22 stsp 		return err;
1092 b107e67f 2018-01-19 stsp 
1093 668a20f6 2020-03-18 stsp 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1094 b0e2201a 2018-06-22 stsp 	    base_tslen, base_type, base_size, recursion - 1);
1095 c3703302 2018-01-23 stsp }
1096 c4330eff 2021-06-22 stsp 
1097 c4330eff 2021-06-22 stsp const struct got_error *
1098 c4330eff 2021-06-22 stsp got_pack_parse_ref_delta(struct got_object_id *id,
1099 c4330eff 2021-06-22 stsp     struct got_pack *pack, off_t delta_offset, int tslen)
1100 c4330eff 2021-06-22 stsp {
1101 c4330eff 2021-06-22 stsp 	if (pack->map) {
1102 ad4cc361 2022-10-27 op 		size_t mapoff;
1103 ad4cc361 2022-10-27 op 
1104 ad4cc361 2022-10-27 op 		if (delta_offset + tslen > SIZE_MAX) {
1105 ad4cc361 2022-10-27 op 			return got_error_fmt(GOT_ERR_PACK_OFFSET,
1106 ad4cc361 2022-10-27 op 			    "mapoff %lld would overflow size_t",
1107 ad4cc361 2022-10-27 op 			    (long long)delta_offset + tslen);
1108 ad4cc361 2022-10-27 op 		}
1109 ad4cc361 2022-10-27 op 
1110 ad4cc361 2022-10-27 op 		mapoff = delta_offset + tslen;
1111 3976db15 2022-01-10 stsp 		if (mapoff + sizeof(*id) >= pack->filesize)
1112 3976db15 2022-01-10 stsp 			return got_error(GOT_ERR_PACK_OFFSET);
1113 c4330eff 2021-06-22 stsp 		memcpy(id, pack->map + mapoff, sizeof(*id));
1114 c4330eff 2021-06-22 stsp 	} else {
1115 c4330eff 2021-06-22 stsp 		ssize_t n;
1116 c4330eff 2021-06-22 stsp 		n = read(pack->fd, id, sizeof(*id));
1117 c4330eff 2021-06-22 stsp 		if (n < 0)
1118 c4330eff 2021-06-22 stsp 			return got_error_from_errno("read");
1119 c4330eff 2021-06-22 stsp 		if (n != sizeof(*id))
1120 c4330eff 2021-06-22 stsp 			return got_error(GOT_ERR_BAD_PACKFILE);
1121 c4330eff 2021-06-22 stsp 	}
1122 c3703302 2018-01-23 stsp 
1123 c4330eff 2021-06-22 stsp 	return NULL;
1124 c4330eff 2021-06-22 stsp }
1125 c4330eff 2021-06-22 stsp 
1126 c3703302 2018-01-23 stsp static const struct got_error *
1127 c8ecd499 2018-09-09 stsp resolve_ref_delta(struct got_delta_chain *deltas, struct got_packidx *packidx,
1128 c8ecd499 2018-09-09 stsp     struct got_pack *pack, off_t delta_offset, size_t tslen, int delta_type,
1129 c8ecd499 2018-09-09 stsp     size_t delta_size, unsigned int recursion)
1130 c3703302 2018-01-23 stsp {
1131 6b9c9673 2018-01-23 stsp 	const struct got_error *err;
1132 6b9c9673 2018-01-23 stsp 	struct got_object_id id;
1133 6b9c9673 2018-01-23 stsp 	int idx;
1134 6b9c9673 2018-01-23 stsp 	off_t base_offset;
1135 6b9c9673 2018-01-23 stsp 	uint8_t base_type;
1136 6b9c9673 2018-01-23 stsp 	uint64_t base_size;
1137 6b9c9673 2018-01-23 stsp 	size_t base_tslen;
1138 bdd2fbb3 2018-02-11 stsp 	off_t delta_data_offset;
1139 6b9c9673 2018-01-23 stsp 
1140 42c69117 2019-11-10 stsp 	if (delta_offset + tslen >= pack->filesize)
1141 d7464085 2018-07-09 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1142 a53d2f13 2018-03-17 stsp 
1143 c4330eff 2021-06-22 stsp 	err = got_pack_parse_ref_delta(&id, pack, delta_offset, tslen);
1144 c4330eff 2021-06-22 stsp 	if (err)
1145 c4330eff 2021-06-22 stsp 		return err;
1146 d7464085 2018-07-09 stsp 	if (pack->map) {
1147 c4330eff 2021-06-22 stsp 		delta_data_offset = delta_offset + tslen + sizeof(id);
1148 d7464085 2018-07-09 stsp 	} else {
1149 e40b19ed 2020-01-06 stsp 		delta_data_offset = lseek(pack->fd, 0, SEEK_CUR);
1150 e40b19ed 2020-01-06 stsp 		if (delta_data_offset == -1)
1151 e40b19ed 2020-01-06 stsp 			return got_error_from_errno("lseek");
1152 d7464085 2018-07-09 stsp 	}
1153 d7464085 2018-07-09 stsp 
1154 c336f889 2018-07-23 stsp 	err = add_delta(deltas, delta_offset, tslen, delta_type, delta_size,
1155 e5792992 2022-10-28 op 	    delta_data_offset);
1156 bdd2fbb3 2018-02-11 stsp 	if (err)
1157 1a35c1bc 2019-05-22 stsp 		return err;
1158 bdd2fbb3 2018-02-11 stsp 
1159 652b2703 2018-06-22 stsp 	/* Delta base must be in the same pack file. */
1160 1510f469 2018-09-09 stsp 	idx = got_packidx_get_object_idx(packidx, &id);
1161 1a35c1bc 2019-05-22 stsp 	if (idx == -1)
1162 668a20f6 2020-03-18 stsp 		return got_error(GOT_ERR_NO_OBJ);
1163 6b9c9673 2018-01-23 stsp 
1164 02828bfd 2021-06-22 stsp 	base_offset = got_packidx_get_object_offset(packidx, idx);
1165 04aed155 2022-07-24 naddy 	if (base_offset == -1)
1166 1a35c1bc 2019-05-22 stsp 		return got_error(GOT_ERR_BAD_PACKIDX);
1167 6b9c9673 2018-01-23 stsp 
1168 1a35c1bc 2019-05-22 stsp 	if (base_offset >= pack->filesize)
1169 1a35c1bc 2019-05-22 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1170 6b9c9673 2018-01-23 stsp 
1171 668a20f6 2020-03-18 stsp 	err = got_pack_parse_object_type_and_size(&base_type, &base_size,
1172 668a20f6 2020-03-18 stsp 	    &base_tslen, pack, base_offset);
1173 6b9c9673 2018-01-23 stsp 	if (err)
1174 1a35c1bc 2019-05-22 stsp 		return err;
1175 6b9c9673 2018-01-23 stsp 
1176 668a20f6 2020-03-18 stsp 	return got_pack_resolve_delta_chain(deltas, packidx, pack, base_offset,
1177 0c048b15 2018-04-27 stsp 	    base_tslen, base_type, base_size, recursion - 1);
1178 6b9c9673 2018-01-23 stsp }
1179 6b9c9673 2018-01-23 stsp 
1180 668a20f6 2020-03-18 stsp const struct got_error *
1181 668a20f6 2020-03-18 stsp got_pack_resolve_delta_chain(struct got_delta_chain *deltas,
1182 668a20f6 2020-03-18 stsp     struct got_packidx *packidx, struct got_pack *pack, off_t delta_offset,
1183 668a20f6 2020-03-18 stsp     size_t tslen, int delta_type, size_t delta_size, unsigned int recursion)
1184 6b9c9673 2018-01-23 stsp {
1185 c3703302 2018-01-23 stsp 	const struct got_error *err = NULL;
1186 c3703302 2018-01-23 stsp 
1187 5b7e13a7 2018-04-02 stsp 	if (--recursion == 0)
1188 5b7e13a7 2018-04-02 stsp 		return got_error(GOT_ERR_RECURSION);
1189 5b7e13a7 2018-04-02 stsp 
1190 c3703302 2018-01-23 stsp 	switch (delta_type) {
1191 4e8cda55 2018-01-19 stsp 	case GOT_OBJ_TYPE_COMMIT:
1192 4e8cda55 2018-01-19 stsp 	case GOT_OBJ_TYPE_TREE:
1193 4e8cda55 2018-01-19 stsp 	case GOT_OBJ_TYPE_BLOB:
1194 4e8cda55 2018-01-19 stsp 	case GOT_OBJ_TYPE_TAG:
1195 c3703302 2018-01-23 stsp 		/* Plain types are the final delta base. Recursion ends. */
1196 c336f889 2018-07-23 stsp 		err = add_delta(deltas, delta_offset, tslen, delta_type,
1197 42c69117 2019-11-10 stsp 		    delta_size, 0);
1198 4e8cda55 2018-01-19 stsp 		break;
1199 a3500804 2018-01-23 stsp 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1200 c8ecd499 2018-09-09 stsp 		err = resolve_offset_delta(deltas, packidx, pack,
1201 b0e2201a 2018-06-22 stsp 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1202 96f5e8b3 2018-01-23 stsp 		break;
1203 4e8cda55 2018-01-19 stsp 	case GOT_OBJ_TYPE_REF_DELTA:
1204 c8ecd499 2018-09-09 stsp 		err = resolve_ref_delta(deltas, packidx, pack,
1205 b0e2201a 2018-06-22 stsp 		    delta_offset, tslen, delta_type, delta_size, recursion - 1);
1206 6b9c9673 2018-01-23 stsp 		break;
1207 4e8cda55 2018-01-19 stsp 	default:
1208 4810de4a 2018-04-01 stsp 		return got_error(GOT_ERR_OBJ_TYPE);
1209 4e8cda55 2018-01-19 stsp 	}
1210 96f5e8b3 2018-01-23 stsp 
1211 96f5e8b3 2018-01-23 stsp 	return err;
1212 96f5e8b3 2018-01-23 stsp }
1213 96f5e8b3 2018-01-23 stsp 
1214 96f5e8b3 2018-01-23 stsp static const struct got_error *
1215 a98c62b1 2018-09-09 stsp open_delta_object(struct got_object **obj, struct got_packidx *packidx,
1216 a98c62b1 2018-09-09 stsp     struct got_pack *pack, struct got_object_id *id, off_t offset,
1217 c59b3346 2018-09-11 stsp     size_t tslen, int delta_type, size_t delta_size, int idx)
1218 96f5e8b3 2018-01-23 stsp {
1219 96f5e8b3 2018-01-23 stsp 	const struct got_error *err = NULL;
1220 96f5e8b3 2018-01-23 stsp 	int resolved_type;
1221 4e8cda55 2018-01-19 stsp 
1222 b107e67f 2018-01-19 stsp 	*obj = calloc(1, sizeof(**obj));
1223 b107e67f 2018-01-19 stsp 	if (*obj == NULL)
1224 638f9024 2019-05-13 stsp 		return got_error_from_errno("calloc");
1225 b107e67f 2018-01-19 stsp 
1226 96f5e8b3 2018-01-23 stsp 	(*obj)->flags = 0;
1227 b107e67f 2018-01-19 stsp 	(*obj)->hdrlen = 0;
1228 39e73dc9 2018-03-03 stsp 	(*obj)->size = 0; /* Not known because deltas aren't applied yet. */
1229 106807b4 2018-09-15 stsp 	memcpy(&(*obj)->id, id, sizeof((*obj)->id));
1230 cecc778e 2018-01-23 stsp 	(*obj)->pack_offset = offset + tslen;
1231 1a35c1bc 2019-05-22 stsp 
1232 dbdddfee 2021-06-23 naddy 	STAILQ_INIT(&(*obj)->deltas.entries);
1233 1a35c1bc 2019-05-22 stsp 	(*obj)->flags |= GOT_OBJ_FLAG_DELTIFIED;
1234 96f5e8b3 2018-01-23 stsp 	(*obj)->flags |= GOT_OBJ_FLAG_PACKED;
1235 c59b3346 2018-09-11 stsp 	(*obj)->pack_idx = idx;
1236 96f5e8b3 2018-01-23 stsp 
1237 668a20f6 2020-03-18 stsp 	err = got_pack_resolve_delta_chain(&(*obj)->deltas, packidx, pack,
1238 668a20f6 2020-03-18 stsp 	    offset, tslen, delta_type, delta_size,
1239 668a20f6 2020-03-18 stsp 	    GOT_DELTA_CHAIN_RECURSION_MAX);
1240 96f5e8b3 2018-01-23 stsp 	if (err)
1241 96f5e8b3 2018-01-23 stsp 		goto done;
1242 96f5e8b3 2018-01-23 stsp 
1243 96f5e8b3 2018-01-23 stsp 	err = got_delta_chain_get_base_type(&resolved_type, &(*obj)->deltas);
1244 96f5e8b3 2018-01-23 stsp 	if (err)
1245 96f5e8b3 2018-01-23 stsp 		goto done;
1246 96f5e8b3 2018-01-23 stsp 	(*obj)->type = resolved_type;
1247 96f5e8b3 2018-01-23 stsp done:
1248 96f5e8b3 2018-01-23 stsp 	if (err) {
1249 96f5e8b3 2018-01-23 stsp 		got_object_close(*obj);
1250 96f5e8b3 2018-01-23 stsp 		*obj = NULL;
1251 96f5e8b3 2018-01-23 stsp 	}
1252 96f5e8b3 2018-01-23 stsp 	return err;
1253 b107e67f 2018-01-19 stsp }
1254 b107e67f 2018-01-19 stsp 
1255 2090a03d 2018-09-09 stsp const struct got_error *
1256 2090a03d 2018-09-09 stsp got_packfile_open_object(struct got_object **obj, struct got_pack *pack,
1257 6fd11751 2018-06-04 stsp     struct got_packidx *packidx, int idx, struct got_object_id *id)
1258 a1fd68d8 2018-01-12 stsp {
1259 a1fd68d8 2018-01-12 stsp 	const struct got_error *err = NULL;
1260 a1fd68d8 2018-01-12 stsp 	off_t offset;
1261 6c00b545 2018-01-17 stsp 	uint8_t type;
1262 6c00b545 2018-01-17 stsp 	uint64_t size;
1263 3ee5fc21 2018-01-17 stsp 	size_t tslen;
1264 a1fd68d8 2018-01-12 stsp 
1265 6c00b545 2018-01-17 stsp 	*obj = NULL;
1266 a1fd68d8 2018-01-12 stsp 
1267 02828bfd 2021-06-22 stsp 	offset = got_packidx_get_object_offset(packidx, idx);
1268 04aed155 2022-07-24 naddy 	if (offset == -1)
1269 a1fd68d8 2018-01-12 stsp 		return got_error(GOT_ERR_BAD_PACKIDX);
1270 a1fd68d8 2018-01-12 stsp 
1271 668a20f6 2020-03-18 stsp 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1272 668a20f6 2020-03-18 stsp 	    pack, offset);
1273 a1fd68d8 2018-01-12 stsp 	if (err)
1274 35c73765 2018-09-09 stsp 		return err;
1275 a1fd68d8 2018-01-12 stsp 
1276 6c00b545 2018-01-17 stsp 	switch (type) {
1277 6c00b545 2018-01-17 stsp 	case GOT_OBJ_TYPE_COMMIT:
1278 6c00b545 2018-01-17 stsp 	case GOT_OBJ_TYPE_TREE:
1279 6c00b545 2018-01-17 stsp 	case GOT_OBJ_TYPE_BLOB:
1280 d33fc9ef 2018-01-23 stsp 	case GOT_OBJ_TYPE_TAG:
1281 5f25cc85 2019-11-26 stsp 		err = open_plain_object(obj, id, type, offset + tslen,
1282 5f25cc85 2019-11-26 stsp 		    size, idx);
1283 6c00b545 2018-01-17 stsp 		break;
1284 b107e67f 2018-01-19 stsp 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1285 a48db7e5 2018-01-23 stsp 	case GOT_OBJ_TYPE_REF_DELTA:
1286 a98c62b1 2018-09-09 stsp 		err = open_delta_object(obj, packidx, pack, id, offset,
1287 c59b3346 2018-09-11 stsp 		    tslen, type, size, idx);
1288 b107e67f 2018-01-19 stsp 		break;
1289 6c00b545 2018-01-17 stsp 	default:
1290 4810de4a 2018-04-01 stsp 		err = got_error(GOT_ERR_OBJ_TYPE);
1291 35c73765 2018-09-09 stsp 		break;
1292 35c73765 2018-09-09 stsp 	}
1293 35c73765 2018-09-09 stsp 
1294 3ee5fc21 2018-01-17 stsp 	return err;
1295 3ee5fc21 2018-01-17 stsp }
1296 efd2a263 2018-01-19 stsp 
1297 d582f26c 2020-03-18 stsp const struct got_error *
1298 c0df5966 2021-12-31 stsp got_pack_get_delta_chain_max_size(uint64_t *max_size,
1299 c0df5966 2021-12-31 stsp     struct got_delta_chain *deltas, struct got_pack *pack)
1300 22484865 2018-03-13 stsp {
1301 22484865 2018-03-13 stsp 	struct got_delta *delta;
1302 22484865 2018-03-13 stsp 	uint64_t base_size = 0, result_size = 0;
1303 22484865 2018-03-13 stsp 
1304 22484865 2018-03-13 stsp 	*max_size = 0;
1305 dbdddfee 2021-06-23 naddy 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1306 22484865 2018-03-13 stsp 		/* Plain object types are the delta base. */
1307 22484865 2018-03-13 stsp 		if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1308 22484865 2018-03-13 stsp 		    delta->type != GOT_OBJ_TYPE_TREE &&
1309 22484865 2018-03-13 stsp 		    delta->type != GOT_OBJ_TYPE_BLOB &&
1310 22484865 2018-03-13 stsp 		    delta->type != GOT_OBJ_TYPE_TAG) {
1311 22484865 2018-03-13 stsp 			const struct got_error *err;
1312 aabb25f8 2022-10-17 stsp 			uint8_t *delta_buf = NULL;
1313 42c69117 2019-11-10 stsp 			size_t delta_len;
1314 ab2f42e7 2019-11-10 stsp 			int cached = 1;
1315 42c69117 2019-11-10 stsp 
1316 aabb25f8 2022-10-17 stsp 			if (pack->delta_cache) {
1317 aabb25f8 2022-10-17 stsp 				got_delta_cache_get(&delta_buf, &delta_len,
1318 aabb25f8 2022-10-17 stsp 				    pack->delta_cache, delta->data_offset);
1319 aabb25f8 2022-10-17 stsp 			}
1320 ab2f42e7 2019-11-10 stsp 			if (delta_buf == NULL) {
1321 ab2f42e7 2019-11-10 stsp 				cached = 0;
1322 ab2f42e7 2019-11-10 stsp 				err = read_delta_data(&delta_buf, &delta_len,
1323 2d9e6abf 2022-05-04 stsp 				    NULL, delta->data_offset, pack);
1324 ab2f42e7 2019-11-10 stsp 				if (err)
1325 ab2f42e7 2019-11-10 stsp 					return err;
1326 aabb25f8 2022-10-17 stsp 			}
1327 aabb25f8 2022-10-17 stsp 			if (pack->delta_cache && !cached) {
1328 ab2f42e7 2019-11-10 stsp 				err = got_delta_cache_add(pack->delta_cache,
1329 ab2f42e7 2019-11-10 stsp 				    delta->data_offset, delta_buf, delta_len);
1330 ab2f42e7 2019-11-10 stsp 				if (err == NULL)
1331 ab2f42e7 2019-11-10 stsp 					cached = 1;
1332 ab2f42e7 2019-11-10 stsp 				else if (err->code != GOT_ERR_NO_SPACE) {
1333 ab2f42e7 2019-11-10 stsp 					free(delta_buf);
1334 ab2f42e7 2019-11-10 stsp 					return err;
1335 ab2f42e7 2019-11-10 stsp 				}
1336 ab2f42e7 2019-11-10 stsp 			}
1337 a53d2f13 2018-03-17 stsp 			err = got_delta_get_sizes(&base_size, &result_size,
1338 42c69117 2019-11-10 stsp 			    delta_buf, delta_len);
1339 ab2f42e7 2019-11-10 stsp 			if (!cached)
1340 ab2f42e7 2019-11-10 stsp 				free(delta_buf);
1341 22484865 2018-03-13 stsp 			if (err)
1342 22484865 2018-03-13 stsp 				return err;
1343 22484865 2018-03-13 stsp 		} else
1344 22484865 2018-03-13 stsp 			base_size = delta->size;
1345 22484865 2018-03-13 stsp 		if (base_size > *max_size)
1346 22484865 2018-03-13 stsp 			*max_size = base_size;
1347 22484865 2018-03-13 stsp 		if (result_size > *max_size)
1348 22484865 2018-03-13 stsp 			*max_size = result_size;
1349 22484865 2018-03-13 stsp 	}
1350 bd1223b9 2018-03-14 stsp 
1351 9feb4ff2 2018-03-14 stsp 	return NULL;
1352 ac544f8c 2019-01-13 stsp }
1353 ac544f8c 2019-01-13 stsp 
1354 ac544f8c 2019-01-13 stsp const struct got_error *
1355 42c69117 2019-11-10 stsp got_pack_get_max_delta_object_size(uint64_t *size, struct got_object *obj,
1356 42c69117 2019-11-10 stsp     struct got_pack *pack)
1357 ac544f8c 2019-01-13 stsp {
1358 85a703fa 2019-01-13 stsp 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0)
1359 85a703fa 2019-01-13 stsp 		return got_error(GOT_ERR_OBJ_TYPE);
1360 85a703fa 2019-01-13 stsp 
1361 d582f26c 2020-03-18 stsp 	return got_pack_get_delta_chain_max_size(size, &obj->deltas, pack);
1362 22484865 2018-03-13 stsp }
1363 22484865 2018-03-13 stsp 
1364 668a20f6 2020-03-18 stsp const struct got_error *
1365 4788f1ce 2020-03-18 stsp got_pack_dump_delta_chain_to_file(size_t *result_size,
1366 4788f1ce 2020-03-18 stsp     struct got_delta_chain *deltas, struct got_pack *pack, FILE *outfile,
1367 4788f1ce 2020-03-18 stsp     FILE *base_file, FILE *accum_file)
1368 efd2a263 2018-01-19 stsp {
1369 efd2a263 2018-01-19 stsp 	const struct got_error *err = NULL;
1370 6691714a 2018-01-23 stsp 	struct got_delta *delta;
1371 aabb25f8 2022-10-17 stsp 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1372 6395114c 2022-05-31 stsp 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1373 6395114c 2022-05-31 stsp 	/* We process small enough files entirely in memory for speed. */
1374 6395114c 2022-05-31 stsp 	const size_t max_bufsize = GOT_DELTA_RESULT_SIZE_CACHED_MAX;
1375 6395114c 2022-05-31 stsp 	uint64_t max_size = 0;
1376 6691714a 2018-01-23 stsp 	int n = 0;
1377 b29656e2 2018-03-16 stsp 
1378 b29656e2 2018-03-16 stsp 	*result_size = 0;
1379 3ee5fc21 2018-01-17 stsp 
1380 dbdddfee 2021-06-23 naddy 	if (STAILQ_EMPTY(&deltas->entries))
1381 6691714a 2018-01-23 stsp 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1382 efd2a263 2018-01-19 stsp 
1383 6395114c 2022-05-31 stsp 	if (fseeko(base_file, 0L, SEEK_SET) == -1)
1384 6395114c 2022-05-31 stsp 		return got_error_from_errno("fseeko");
1385 6395114c 2022-05-31 stsp 	if (fseeko(accum_file, 0L, SEEK_SET) == -1)
1386 6395114c 2022-05-31 stsp 		return got_error_from_errno("fseeko");
1387 efd2a263 2018-01-19 stsp 
1388 6691714a 2018-01-23 stsp 	/* Deltas are ordered in ascending order. */
1389 dbdddfee 2021-06-23 naddy 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1390 aabb25f8 2022-10-17 stsp 		uint8_t *delta_buf = NULL;
1391 6395114c 2022-05-31 stsp 		uint64_t base_size, result_size = 0;
1392 ab2f42e7 2019-11-10 stsp 		int cached = 1;
1393 a6b158cc 2018-02-11 stsp 		if (n == 0) {
1394 34fca9c3 2018-11-11 stsp 			size_t mapoff;
1395 0c048b15 2018-04-27 stsp 			off_t delta_data_offset;
1396 9db65d41 2018-03-14 stsp 
1397 a6b158cc 2018-02-11 stsp 			/* Plain object types are the delta base. */
1398 a6b158cc 2018-02-11 stsp 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1399 a6b158cc 2018-02-11 stsp 			    delta->type != GOT_OBJ_TYPE_TREE &&
1400 a6b158cc 2018-02-11 stsp 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1401 a6b158cc 2018-02-11 stsp 			    delta->type != GOT_OBJ_TYPE_TAG) {
1402 72eb3431 2018-04-01 stsp 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1403 72eb3431 2018-04-01 stsp 				goto done;
1404 72eb3431 2018-04-01 stsp 			}
1405 72eb3431 2018-04-01 stsp 
1406 0c048b15 2018-04-27 stsp 			delta_data_offset = delta->offset + delta->tslen;
1407 0c048b15 2018-04-27 stsp 			if (delta_data_offset >= pack->filesize) {
1408 0c048b15 2018-04-27 stsp 				err = got_error(GOT_ERR_PACK_OFFSET);
1409 0c048b15 2018-04-27 stsp 				goto done;
1410 0c048b15 2018-04-27 stsp 			}
1411 d7464085 2018-07-09 stsp 			if (pack->map == NULL) {
1412 d7464085 2018-07-09 stsp 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1413 d7464085 2018-07-09 stsp 				    == -1) {
1414 638f9024 2019-05-13 stsp 					err = got_error_from_errno("lseek");
1415 d7464085 2018-07-09 stsp 					goto done;
1416 d7464085 2018-07-09 stsp 				}
1417 bdd2fbb3 2018-02-11 stsp 			}
1418 6395114c 2022-05-31 stsp 			if (delta->size > max_size)
1419 6395114c 2022-05-31 stsp 				max_size = delta->size;
1420 6395114c 2022-05-31 stsp 			if (max_size > max_bufsize) {
1421 d7464085 2018-07-09 stsp 				if (pack->map) {
1422 ad4cc361 2022-10-27 op 					if (delta_data_offset > SIZE_MAX) {
1423 ad4cc361 2022-10-27 op 						return got_error_fmt(
1424 ad4cc361 2022-10-27 op 						    GOT_ERR_RANGE,
1425 ad4cc361 2022-10-27 op 						    "delta offset %lld "
1426 ad4cc361 2022-10-27 op 						    "overflows size_t",
1427 ad4cc361 2022-10-27 op 						    (long long)
1428 ad4cc361 2022-10-27 op 						    delta_data_offset);
1429 ad4cc361 2022-10-27 op 					}
1430 ad4cc361 2022-10-27 op 
1431 ad4cc361 2022-10-27 op 					mapoff = delta_data_offset;
1432 d7464085 2018-07-09 stsp 					err = got_inflate_to_file_mmap(
1433 4788f1ce 2020-03-18 stsp 					    &base_bufsz, NULL, NULL, pack->map,
1434 4788f1ce 2020-03-18 stsp 					    mapoff, pack->filesize - mapoff,
1435 4788f1ce 2020-03-18 stsp 					    base_file);
1436 d7464085 2018-07-09 stsp 				} else
1437 34fca9c3 2018-11-11 stsp 					err = got_inflate_to_file_fd(
1438 4788f1ce 2020-03-18 stsp 					    &base_bufsz, NULL, NULL, pack->fd,
1439 4788f1ce 2020-03-18 stsp 					    base_file);
1440 d7464085 2018-07-09 stsp 			} else {
1441 6395114c 2022-05-31 stsp 				accum_buf = malloc(max_size);
1442 6395114c 2022-05-31 stsp 				if (accum_buf == NULL) {
1443 6395114c 2022-05-31 stsp 					err = got_error_from_errno("malloc");
1444 6395114c 2022-05-31 stsp 					goto done;
1445 6395114c 2022-05-31 stsp 				}
1446 6395114c 2022-05-31 stsp 				accum_bufsz = max_size;
1447 d7464085 2018-07-09 stsp 				if (pack->map) {
1448 ad4cc361 2022-10-27 op 					if (delta_data_offset > SIZE_MAX) {
1449 ad4cc361 2022-10-27 op 						return got_error_fmt(
1450 ad4cc361 2022-10-27 op 						    GOT_ERR_RANGE,
1451 ad4cc361 2022-10-27 op 						    "delta offset %lld "
1452 ad4cc361 2022-10-27 op 						    "overflows size_t",
1453 ad4cc361 2022-10-27 op 						    (long long)
1454 ad4cc361 2022-10-27 op 						    delta_data_offset);
1455 ad4cc361 2022-10-27 op 					}
1456 ad4cc361 2022-10-27 op 
1457 ad4cc361 2022-10-27 op 					mapoff = delta_data_offset;
1458 d7464085 2018-07-09 stsp 					err = got_inflate_to_mem_mmap(&base_buf,
1459 2e5a6fad 2020-03-18 stsp 					    &base_bufsz, NULL, NULL,
1460 2e5a6fad 2020-03-18 stsp 					    pack->map, mapoff,
1461 d7464085 2018-07-09 stsp 					    pack->filesize - mapoff);
1462 d7464085 2018-07-09 stsp 				} else
1463 d7464085 2018-07-09 stsp 					err = got_inflate_to_mem_fd(&base_buf,
1464 55fdd257 2020-03-18 stsp 					    &base_bufsz, NULL, NULL, max_size,
1465 55fdd257 2020-03-18 stsp 					    pack->fd);
1466 8628c62d 2018-03-15 stsp 			}
1467 a6b158cc 2018-02-11 stsp 			if (err)
1468 a6b158cc 2018-02-11 stsp 				goto done;
1469 a6b158cc 2018-02-11 stsp 			n++;
1470 6395114c 2022-05-31 stsp 			if (base_buf == NULL)
1471 8628c62d 2018-03-15 stsp 				rewind(base_file);
1472 a6b158cc 2018-02-11 stsp 			continue;
1473 9db65d41 2018-03-14 stsp 		}
1474 885d3e02 2018-01-27 stsp 
1475 aabb25f8 2022-10-17 stsp 		if (pack->delta_cache) {
1476 aabb25f8 2022-10-17 stsp 			got_delta_cache_get(&delta_buf, &delta_len,
1477 aabb25f8 2022-10-17 stsp 			    pack->delta_cache, delta->data_offset);
1478 aabb25f8 2022-10-17 stsp 		}
1479 ab2f42e7 2019-11-10 stsp 		if (delta_buf == NULL) {
1480 ab2f42e7 2019-11-10 stsp 			cached = 0;
1481 2d9e6abf 2022-05-04 stsp 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1482 ab2f42e7 2019-11-10 stsp 			    delta->data_offset, pack);
1483 ab2f42e7 2019-11-10 stsp 			if (err)
1484 ab2f42e7 2019-11-10 stsp 				goto done;
1485 aabb25f8 2022-10-17 stsp 		}
1486 aabb25f8 2022-10-17 stsp 		if (pack->delta_cache && !cached) {
1487 ab2f42e7 2019-11-10 stsp 			err = got_delta_cache_add(pack->delta_cache,
1488 ab2f42e7 2019-11-10 stsp 			    delta->data_offset, delta_buf, delta_len);
1489 ab2f42e7 2019-11-10 stsp 			if (err == NULL)
1490 ab2f42e7 2019-11-10 stsp 				cached = 1;
1491 ab2f42e7 2019-11-10 stsp 			else if (err->code != GOT_ERR_NO_SPACE) {
1492 ab2f42e7 2019-11-10 stsp 				free(delta_buf);
1493 ab2f42e7 2019-11-10 stsp 				goto done;
1494 ab2f42e7 2019-11-10 stsp 			}
1495 ab2f42e7 2019-11-10 stsp 		}
1496 6395114c 2022-05-31 stsp 
1497 6395114c 2022-05-31 stsp 		err = got_delta_get_sizes(&base_size, &result_size,
1498 6395114c 2022-05-31 stsp 		    delta_buf, delta_len);
1499 e62fc520 2022-11-08 stsp 		if (err) {
1500 e62fc520 2022-11-08 stsp 			if (!cached)
1501 e62fc520 2022-11-08 stsp 				free(delta_buf);
1502 6395114c 2022-05-31 stsp 			goto done;
1503 e62fc520 2022-11-08 stsp 		}
1504 6395114c 2022-05-31 stsp 		if (base_size > max_size)
1505 6395114c 2022-05-31 stsp 			max_size = base_size;
1506 6395114c 2022-05-31 stsp 		if (result_size > max_size)
1507 6395114c 2022-05-31 stsp 			max_size = result_size;
1508 6395114c 2022-05-31 stsp 
1509 6395114c 2022-05-31 stsp 		if (base_buf && max_size > max_bufsize) {
1510 6395114c 2022-05-31 stsp 			/* Switch from buffers to temporary files. */
1511 6395114c 2022-05-31 stsp 			size_t w = fwrite(base_buf, 1, base_bufsz,
1512 6395114c 2022-05-31 stsp 			    base_file);
1513 6395114c 2022-05-31 stsp 			if (w != base_bufsz) {
1514 6395114c 2022-05-31 stsp 				err = got_ferror(outfile, GOT_ERR_IO);
1515 e62fc520 2022-11-08 stsp 				if (!cached)
1516 e62fc520 2022-11-08 stsp 					free(delta_buf);
1517 6395114c 2022-05-31 stsp 				goto done;
1518 6395114c 2022-05-31 stsp 			}
1519 6395114c 2022-05-31 stsp 			free(base_buf);
1520 6395114c 2022-05-31 stsp 			base_buf = NULL;
1521 6395114c 2022-05-31 stsp 			free(accum_buf);
1522 6395114c 2022-05-31 stsp 			accum_buf = NULL;
1523 6395114c 2022-05-31 stsp 		}
1524 6395114c 2022-05-31 stsp 
1525 6395114c 2022-05-31 stsp 		if (base_buf && max_size > base_bufsz) {
1526 6395114c 2022-05-31 stsp 			uint8_t *p = realloc(base_buf, max_size);
1527 6395114c 2022-05-31 stsp 			if (p == NULL) {
1528 6395114c 2022-05-31 stsp 				err = got_error_from_errno("realloc");
1529 e62fc520 2022-11-08 stsp 				if (!cached)
1530 e62fc520 2022-11-08 stsp 					free(delta_buf);
1531 6395114c 2022-05-31 stsp 				goto done;
1532 6395114c 2022-05-31 stsp 			}
1533 6395114c 2022-05-31 stsp 			base_buf = p;
1534 6395114c 2022-05-31 stsp 			base_bufsz = max_size;
1535 6395114c 2022-05-31 stsp 		}
1536 6395114c 2022-05-31 stsp 
1537 6395114c 2022-05-31 stsp 		if (accum_buf && max_size > accum_bufsz) {
1538 6395114c 2022-05-31 stsp 			uint8_t *p = realloc(accum_buf, max_size);
1539 6395114c 2022-05-31 stsp 			if (p == NULL) {
1540 6395114c 2022-05-31 stsp 				err = got_error_from_errno("realloc");
1541 e62fc520 2022-11-08 stsp 				if (!cached)
1542 e62fc520 2022-11-08 stsp 					free(delta_buf);
1543 6395114c 2022-05-31 stsp 				goto done;
1544 6395114c 2022-05-31 stsp 			}
1545 6395114c 2022-05-31 stsp 			accum_buf = p;
1546 6395114c 2022-05-31 stsp 			accum_bufsz = max_size;
1547 6395114c 2022-05-31 stsp 		}
1548 6395114c 2022-05-31 stsp 
1549 8628c62d 2018-03-15 stsp 		if (base_buf) {
1550 34fca9c3 2018-11-11 stsp 			err = got_delta_apply_in_mem(base_buf, base_bufsz,
1551 42c69117 2019-11-10 stsp 			    delta_buf, delta_len, accum_buf,
1552 34fca9c3 2018-11-11 stsp 			    &accum_size, max_size);
1553 8628c62d 2018-03-15 stsp 			n++;
1554 8628c62d 2018-03-15 stsp 		} else {
1555 42c69117 2019-11-10 stsp 			err = got_delta_apply(base_file, delta_buf,
1556 42c69117 2019-11-10 stsp 			    delta_len,
1557 8628c62d 2018-03-15 stsp 			    /* Final delta application writes to output file. */
1558 b29656e2 2018-03-16 stsp 			    ++n < deltas->nentries ? accum_file : outfile,
1559 b29656e2 2018-03-16 stsp 			    &accum_size);
1560 8628c62d 2018-03-15 stsp 		}
1561 ab2f42e7 2019-11-10 stsp 		if (!cached)
1562 ab2f42e7 2019-11-10 stsp 			free(delta_buf);
1563 6691714a 2018-01-23 stsp 		if (err)
1564 6691714a 2018-01-23 stsp 			goto done;
1565 6691714a 2018-01-23 stsp 
1566 710bb5ca 2018-01-23 stsp 		if (n < deltas->nentries) {
1567 6691714a 2018-01-23 stsp 			/* Accumulated delta becomes the new base. */
1568 8628c62d 2018-03-15 stsp 			if (base_buf) {
1569 8628c62d 2018-03-15 stsp 				uint8_t *tmp = accum_buf;
1570 6395114c 2022-05-31 stsp 				size_t tmp_size = accum_bufsz;
1571 8628c62d 2018-03-15 stsp 				accum_buf = base_buf;
1572 6395114c 2022-05-31 stsp 				accum_bufsz = base_bufsz;
1573 8628c62d 2018-03-15 stsp 				base_buf = tmp;
1574 6395114c 2022-05-31 stsp 				base_bufsz = tmp_size;
1575 8628c62d 2018-03-15 stsp 			} else {
1576 8628c62d 2018-03-15 stsp 				FILE *tmp = accum_file;
1577 8628c62d 2018-03-15 stsp 				accum_file = base_file;
1578 8628c62d 2018-03-15 stsp 				base_file = tmp;
1579 8628c62d 2018-03-15 stsp 				rewind(base_file);
1580 8628c62d 2018-03-15 stsp 				rewind(accum_file);
1581 8628c62d 2018-03-15 stsp 			}
1582 6691714a 2018-01-23 stsp 		}
1583 6691714a 2018-01-23 stsp 	}
1584 6691714a 2018-01-23 stsp 
1585 6691714a 2018-01-23 stsp done:
1586 8628c62d 2018-03-15 stsp 	free(base_buf);
1587 8628c62d 2018-03-15 stsp 	if (accum_buf) {
1588 8628c62d 2018-03-15 stsp 		size_t len = fwrite(accum_buf, 1, accum_size, outfile);
1589 8628c62d 2018-03-15 stsp 		free(accum_buf);
1590 8628c62d 2018-03-15 stsp 		if (len != accum_size)
1591 673702af 2018-07-23 stsp 			err = got_ferror(outfile, GOT_ERR_IO);
1592 8628c62d 2018-03-15 stsp 	}
1593 6691714a 2018-01-23 stsp 	rewind(outfile);
1594 b29656e2 2018-03-16 stsp 	if (err == NULL)
1595 b29656e2 2018-03-16 stsp 		*result_size = accum_size;
1596 e0ab43e7 2018-03-16 stsp 	return err;
1597 e0ab43e7 2018-03-16 stsp }
1598 e0ab43e7 2018-03-16 stsp 
1599 668a20f6 2020-03-18 stsp const struct got_error *
1600 668a20f6 2020-03-18 stsp got_pack_dump_delta_chain_to_mem(uint8_t **outbuf, size_t *outlen,
1601 48095039 2018-09-09 stsp     struct got_delta_chain *deltas, struct got_pack *pack)
1602 e0ab43e7 2018-03-16 stsp {
1603 e0ab43e7 2018-03-16 stsp 	const struct got_error *err = NULL;
1604 e0ab43e7 2018-03-16 stsp 	struct got_delta *delta;
1605 aabb25f8 2022-10-17 stsp 	uint8_t *base_buf = NULL, *accum_buf = NULL;
1606 f18c433a 2022-05-31 stsp 	size_t base_bufsz = 0, accum_bufsz = 0, accum_size = 0, delta_len;
1607 f18c433a 2022-05-31 stsp 	uint64_t max_size = 0;
1608 e0ab43e7 2018-03-16 stsp 	int n = 0;
1609 e0ab43e7 2018-03-16 stsp 
1610 e0ab43e7 2018-03-16 stsp 	*outbuf = NULL;
1611 e0ab43e7 2018-03-16 stsp 	*outlen = 0;
1612 e0ab43e7 2018-03-16 stsp 
1613 dbdddfee 2021-06-23 naddy 	if (STAILQ_EMPTY(&deltas->entries))
1614 e0ab43e7 2018-03-16 stsp 		return got_error(GOT_ERR_BAD_DELTA_CHAIN);
1615 e0ab43e7 2018-03-16 stsp 
1616 e0ab43e7 2018-03-16 stsp 	/* Deltas are ordered in ascending order. */
1617 dbdddfee 2021-06-23 naddy 	STAILQ_FOREACH(delta, &deltas->entries, entry) {
1618 aabb25f8 2022-10-17 stsp 		uint8_t *delta_buf = NULL;
1619 f18c433a 2022-05-31 stsp 		uint64_t base_size, result_size = 0;
1620 ab2f42e7 2019-11-10 stsp 		int cached = 1;
1621 e0ab43e7 2018-03-16 stsp 		if (n == 0) {
1622 ad4cc361 2022-10-27 op 			off_t delta_data_offset;
1623 e0ab43e7 2018-03-16 stsp 
1624 e0ab43e7 2018-03-16 stsp 			/* Plain object types are the delta base. */
1625 e0ab43e7 2018-03-16 stsp 			if (delta->type != GOT_OBJ_TYPE_COMMIT &&
1626 e0ab43e7 2018-03-16 stsp 			    delta->type != GOT_OBJ_TYPE_TREE &&
1627 e0ab43e7 2018-03-16 stsp 			    delta->type != GOT_OBJ_TYPE_BLOB &&
1628 e0ab43e7 2018-03-16 stsp 			    delta->type != GOT_OBJ_TYPE_TAG) {
1629 72eb3431 2018-04-01 stsp 				err = got_error(GOT_ERR_BAD_DELTA_CHAIN);
1630 72eb3431 2018-04-01 stsp 				goto done;
1631 72eb3431 2018-04-01 stsp 			}
1632 72eb3431 2018-04-01 stsp 
1633 0c048b15 2018-04-27 stsp 			delta_data_offset = delta->offset + delta->tslen;
1634 0c048b15 2018-04-27 stsp 			if (delta_data_offset >= pack->filesize) {
1635 0c048b15 2018-04-27 stsp 				err = got_error(GOT_ERR_PACK_OFFSET);
1636 0c048b15 2018-04-27 stsp 				goto done;
1637 0c048b15 2018-04-27 stsp 			}
1638 f18c433a 2022-05-31 stsp 
1639 f18c433a 2022-05-31 stsp 			if (delta->size > max_size)
1640 f18c433a 2022-05-31 stsp 				max_size = delta->size;
1641 f18c433a 2022-05-31 stsp 
1642 d7464085 2018-07-09 stsp 			if (pack->map) {
1643 ad4cc361 2022-10-27 op 				size_t mapoff;
1644 ad4cc361 2022-10-27 op 
1645 ad4cc361 2022-10-27 op 				if (delta_data_offset > SIZE_MAX) {
1646 ad4cc361 2022-10-27 op 					return got_error_fmt(GOT_ERR_RANGE,
1647 ad4cc361 2022-10-27 op 					    "delta %lld offset would "
1648 ad4cc361 2022-10-27 op 					    "overflow size_t",
1649 ad4cc361 2022-10-27 op 					    (long long)delta_data_offset);
1650 ad4cc361 2022-10-27 op 				}
1651 ad4cc361 2022-10-27 op 
1652 ad4cc361 2022-10-27 op 				mapoff = delta_data_offset;
1653 d7464085 2018-07-09 stsp 				err = got_inflate_to_mem_mmap(&base_buf,
1654 2e5a6fad 2020-03-18 stsp 				    &base_bufsz, NULL, NULL, pack->map,
1655 2e5a6fad 2020-03-18 stsp 				    mapoff, pack->filesize - mapoff);
1656 d7464085 2018-07-09 stsp 			} else {
1657 d7464085 2018-07-09 stsp 				if (lseek(pack->fd, delta_data_offset, SEEK_SET)
1658 d7464085 2018-07-09 stsp 				    == -1) {
1659 638f9024 2019-05-13 stsp 					err = got_error_from_errno("lseek");
1660 d7464085 2018-07-09 stsp 					goto done;
1661 d7464085 2018-07-09 stsp 				}
1662 d7464085 2018-07-09 stsp 				err = got_inflate_to_mem_fd(&base_buf,
1663 55fdd257 2020-03-18 stsp 				    &base_bufsz, NULL, NULL, max_size,
1664 55fdd257 2020-03-18 stsp 				    pack->fd);
1665 e0ab43e7 2018-03-16 stsp 			}
1666 d7464085 2018-07-09 stsp 			if (err)
1667 d7464085 2018-07-09 stsp 				goto done;
1668 e0ab43e7 2018-03-16 stsp 			n++;
1669 e0ab43e7 2018-03-16 stsp 			continue;
1670 e0ab43e7 2018-03-16 stsp 		}
1671 e0ab43e7 2018-03-16 stsp 
1672 aabb25f8 2022-10-17 stsp 		if (pack->delta_cache) {
1673 aabb25f8 2022-10-17 stsp 			got_delta_cache_get(&delta_buf, &delta_len,
1674 aabb25f8 2022-10-17 stsp 			    pack->delta_cache, delta->data_offset);
1675 aabb25f8 2022-10-17 stsp 		}
1676 ab2f42e7 2019-11-10 stsp 		if (delta_buf == NULL) {
1677 ab2f42e7 2019-11-10 stsp 			cached = 0;
1678 2d9e6abf 2022-05-04 stsp 			err = read_delta_data(&delta_buf, &delta_len, NULL,
1679 ab2f42e7 2019-11-10 stsp 			    delta->data_offset, pack);
1680 ab2f42e7 2019-11-10 stsp 			if (err)
1681 ab2f42e7 2019-11-10 stsp 				goto done;
1682 aabb25f8 2022-10-17 stsp 		}
1683 aabb25f8 2022-10-17 stsp 		if (pack->delta_cache && !cached) {
1684 ab2f42e7 2019-11-10 stsp 			err = got_delta_cache_add(pack->delta_cache,
1685 ab2f42e7 2019-11-10 stsp 			    delta->data_offset, delta_buf, delta_len);
1686 ab2f42e7 2019-11-10 stsp 			if (err == NULL)
1687 ab2f42e7 2019-11-10 stsp 				cached = 1;
1688 ab2f42e7 2019-11-10 stsp 			else if (err->code != GOT_ERR_NO_SPACE) {
1689 ab2f42e7 2019-11-10 stsp 				free(delta_buf);
1690 ab2f42e7 2019-11-10 stsp 				goto done;
1691 ab2f42e7 2019-11-10 stsp 			}
1692 ab2f42e7 2019-11-10 stsp 		}
1693 f18c433a 2022-05-31 stsp 
1694 f18c433a 2022-05-31 stsp 		err = got_delta_get_sizes(&base_size, &result_size,
1695 f18c433a 2022-05-31 stsp 		    delta_buf, delta_len);
1696 e62fc520 2022-11-08 stsp 		if (err) {
1697 e62fc520 2022-11-08 stsp 			if (!cached)
1698 e62fc520 2022-11-08 stsp 				free(delta_buf);
1699 f18c433a 2022-05-31 stsp 			goto done;
1700 e62fc520 2022-11-08 stsp 		}
1701 f18c433a 2022-05-31 stsp 		if (base_size > max_size)
1702 f18c433a 2022-05-31 stsp 			max_size = base_size;
1703 f18c433a 2022-05-31 stsp 		if (result_size > max_size)
1704 f18c433a 2022-05-31 stsp 			max_size = result_size;
1705 f18c433a 2022-05-31 stsp 
1706 f18c433a 2022-05-31 stsp 		if (max_size > base_bufsz) {
1707 f18c433a 2022-05-31 stsp 			uint8_t *p = realloc(base_buf, max_size);
1708 f18c433a 2022-05-31 stsp 			if (p == NULL) {
1709 f18c433a 2022-05-31 stsp 				err = got_error_from_errno("realloc");
1710 e62fc520 2022-11-08 stsp 				if (!cached)
1711 e62fc520 2022-11-08 stsp 					free(delta_buf);
1712 f18c433a 2022-05-31 stsp 				goto done;
1713 f18c433a 2022-05-31 stsp 			}
1714 f18c433a 2022-05-31 stsp 			base_buf = p;
1715 f18c433a 2022-05-31 stsp 			base_bufsz = max_size;
1716 f18c433a 2022-05-31 stsp 		}
1717 f18c433a 2022-05-31 stsp 
1718 f18c433a 2022-05-31 stsp 		if (max_size > accum_bufsz) {
1719 f18c433a 2022-05-31 stsp 			uint8_t *p = realloc(accum_buf, max_size);
1720 f18c433a 2022-05-31 stsp 			if (p == NULL) {
1721 f18c433a 2022-05-31 stsp 				err = got_error_from_errno("realloc");
1722 e62fc520 2022-11-08 stsp 				if (!cached)
1723 e62fc520 2022-11-08 stsp 					free(delta_buf);
1724 f18c433a 2022-05-31 stsp 				goto done;
1725 f18c433a 2022-05-31 stsp 			}
1726 f18c433a 2022-05-31 stsp 			accum_buf = p;
1727 f18c433a 2022-05-31 stsp 			accum_bufsz = max_size;
1728 f18c433a 2022-05-31 stsp 		}
1729 f18c433a 2022-05-31 stsp 
1730 34fca9c3 2018-11-11 stsp 		err = got_delta_apply_in_mem(base_buf, base_bufsz,
1731 42c69117 2019-11-10 stsp 		    delta_buf, delta_len, accum_buf,
1732 34fca9c3 2018-11-11 stsp 		    &accum_size, max_size);
1733 ab2f42e7 2019-11-10 stsp 		if (!cached)
1734 ab2f42e7 2019-11-10 stsp 			free(delta_buf);
1735 e0ab43e7 2018-03-16 stsp 		n++;
1736 e0ab43e7 2018-03-16 stsp 		if (err)
1737 e0ab43e7 2018-03-16 stsp 			goto done;
1738 e0ab43e7 2018-03-16 stsp 
1739 e0ab43e7 2018-03-16 stsp 		if (n < deltas->nentries) {
1740 e0ab43e7 2018-03-16 stsp 			/* Accumulated delta becomes the new base. */
1741 e0ab43e7 2018-03-16 stsp 			uint8_t *tmp = accum_buf;
1742 f18c433a 2022-05-31 stsp 			size_t tmp_size = accum_bufsz;
1743 e0ab43e7 2018-03-16 stsp 			accum_buf = base_buf;
1744 f18c433a 2022-05-31 stsp 			accum_bufsz = base_bufsz;
1745 e0ab43e7 2018-03-16 stsp 			base_buf = tmp;
1746 f18c433a 2022-05-31 stsp 			base_bufsz = tmp_size;
1747 e0ab43e7 2018-03-16 stsp 		}
1748 e0ab43e7 2018-03-16 stsp 	}
1749 e0ab43e7 2018-03-16 stsp 
1750 e0ab43e7 2018-03-16 stsp done:
1751 e0ab43e7 2018-03-16 stsp 	free(base_buf);
1752 e0ab43e7 2018-03-16 stsp 	if (err) {
1753 e0ab43e7 2018-03-16 stsp 		free(accum_buf);
1754 e0ab43e7 2018-03-16 stsp 		*outbuf = NULL;
1755 e0ab43e7 2018-03-16 stsp 		*outlen = 0;
1756 e0ab43e7 2018-03-16 stsp 	} else {
1757 e0ab43e7 2018-03-16 stsp 		*outbuf = accum_buf;
1758 e0ab43e7 2018-03-16 stsp 		*outlen = accum_size;
1759 e0ab43e7 2018-03-16 stsp 	}
1760 efd2a263 2018-01-19 stsp 	return err;
1761 efd2a263 2018-01-19 stsp }
1762 efd2a263 2018-01-19 stsp 
1763 3ee5fc21 2018-01-17 stsp const struct got_error *
1764 24140570 2018-09-09 stsp got_packfile_extract_object(struct got_pack *pack, struct got_object *obj,
1765 3840f4c9 2018-09-12 stsp     FILE *outfile, FILE *base_file, FILE *accum_file)
1766 3ee5fc21 2018-01-17 stsp {
1767 3ee5fc21 2018-01-17 stsp 	const struct got_error *err = NULL;
1768 3ee5fc21 2018-01-17 stsp 
1769 3ee5fc21 2018-01-17 stsp 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1770 3ee5fc21 2018-01-17 stsp 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1771 2ce68b2f 2018-09-09 stsp 
1772 ef2bccd9 2018-03-16 stsp 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1773 24140570 2018-09-09 stsp 		if (obj->pack_offset >= pack->filesize)
1774 24140570 2018-09-09 stsp 			return got_error(GOT_ERR_PACK_OFFSET);
1775 72eb3431 2018-04-01 stsp 
1776 d7464085 2018-07-09 stsp 		if (pack->map) {
1777 ad4cc361 2022-10-27 op 			size_t mapoff;
1778 ad4cc361 2022-10-27 op 
1779 ad4cc361 2022-10-27 op 			if (obj->pack_offset > SIZE_MAX) {
1780 ad4cc361 2022-10-27 op 				return got_error_fmt(GOT_ERR_RANGE,
1781 ad4cc361 2022-10-27 op 				    "pack offset %lld would overflow size_t",
1782 ad4cc361 2022-10-27 op 				    (long long)obj->pack_offset);
1783 ad4cc361 2022-10-27 op 			}
1784 ad4cc361 2022-10-27 op 
1785 ad4cc361 2022-10-27 op 			mapoff = obj->pack_offset;
1786 4788f1ce 2020-03-18 stsp 			err = got_inflate_to_file_mmap(&obj->size, NULL, NULL,
1787 4788f1ce 2020-03-18 stsp 			    pack->map, mapoff, pack->filesize - mapoff,
1788 4788f1ce 2020-03-18 stsp 			    outfile);
1789 d7464085 2018-07-09 stsp 		} else {
1790 24140570 2018-09-09 stsp 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1791 638f9024 2019-05-13 stsp 				return got_error_from_errno("lseek");
1792 4788f1ce 2020-03-18 stsp 			err = got_inflate_to_file_fd(&obj->size, NULL, NULL,
1793 4788f1ce 2020-03-18 stsp 			    pack->fd, outfile);
1794 d7464085 2018-07-09 stsp 		}
1795 040bf4a1 2018-04-01 stsp 	} else
1796 4788f1ce 2020-03-18 stsp 		err = got_pack_dump_delta_chain_to_file(&obj->size,
1797 4788f1ce 2020-03-18 stsp 		    &obj->deltas, pack, outfile, base_file, accum_file);
1798 24140570 2018-09-09 stsp 
1799 a1fd68d8 2018-01-12 stsp 	return err;
1800 a1fd68d8 2018-01-12 stsp }
1801 e0ab43e7 2018-03-16 stsp 
1802 e0ab43e7 2018-03-16 stsp const struct got_error *
1803 e0ab43e7 2018-03-16 stsp got_packfile_extract_object_to_mem(uint8_t **buf, size_t *len,
1804 7e212e3d 2018-09-09 stsp     struct got_object *obj, struct got_pack *pack)
1805 e0ab43e7 2018-03-16 stsp {
1806 e0ab43e7 2018-03-16 stsp 	const struct got_error *err = NULL;
1807 e0ab43e7 2018-03-16 stsp 
1808 e0ab43e7 2018-03-16 stsp 	if ((obj->flags & GOT_OBJ_FLAG_PACKED) == 0)
1809 e0ab43e7 2018-03-16 stsp 		return got_error(GOT_ERR_OBJ_NOT_PACKED);
1810 72eb3431 2018-04-01 stsp 
1811 48095039 2018-09-09 stsp 	if ((obj->flags & GOT_OBJ_FLAG_DELTIFIED) == 0) {
1812 7e212e3d 2018-09-09 stsp 		if (obj->pack_offset >= pack->filesize)
1813 7e212e3d 2018-09-09 stsp 			return got_error(GOT_ERR_PACK_OFFSET);
1814 d7464085 2018-07-09 stsp 		if (pack->map) {
1815 ad4cc361 2022-10-27 op 			size_t mapoff;
1816 ad4cc361 2022-10-27 op 
1817 ad4cc361 2022-10-27 op 			if (obj->pack_offset > SIZE_MAX) {
1818 ad4cc361 2022-10-27 op 				return got_error_fmt(GOT_ERR_RANGE,
1819 ad4cc361 2022-10-27 op 				    "pack offset %lld would overflow size_t",
1820 ad4cc361 2022-10-27 op 				    (long long)obj->pack_offset);
1821 ad4cc361 2022-10-27 op 			}
1822 ad4cc361 2022-10-27 op 
1823 ad4cc361 2022-10-27 op 			mapoff = obj->pack_offset;
1824 2e5a6fad 2020-03-18 stsp 			err = got_inflate_to_mem_mmap(buf, len, NULL, NULL,
1825 2e5a6fad 2020-03-18 stsp 			    pack->map, mapoff, pack->filesize - mapoff);
1826 d7464085 2018-07-09 stsp 		} else {
1827 7e212e3d 2018-09-09 stsp 			if (lseek(pack->fd, obj->pack_offset, SEEK_SET) == -1)
1828 638f9024 2019-05-13 stsp 				return got_error_from_errno("lseek");
1829 1e87a3c3 2020-03-18 stsp 			err = got_inflate_to_mem_fd(buf, len, NULL, NULL,
1830 55fdd257 2020-03-18 stsp 			    obj->size, pack->fd);
1831 e0ab43e7 2018-03-16 stsp 		}
1832 e0ab43e7 2018-03-16 stsp 	} else
1833 668a20f6 2020-03-18 stsp 		err = got_pack_dump_delta_chain_to_mem(buf, len, &obj->deltas,
1834 668a20f6 2020-03-18 stsp 		    pack);
1835 7e212e3d 2018-09-09 stsp 
1836 e0ab43e7 2018-03-16 stsp 	return err;
1837 e0ab43e7 2018-03-16 stsp }
1838 67fd6849 2022-02-13 stsp 
1839 2d9e6abf 2022-05-04 stsp static const struct got_error *
1840 2d9e6abf 2022-05-04 stsp read_raw_delta_data(uint8_t **delta_buf, size_t *delta_len,
1841 2d9e6abf 2022-05-04 stsp     size_t *delta_len_compressed, uint64_t *base_size, uint64_t *result_size,
1842 2d9e6abf 2022-05-04 stsp     off_t delta_data_offset, struct got_pack *pack, struct got_packidx *packidx)
1843 2d9e6abf 2022-05-04 stsp {
1844 2d9e6abf 2022-05-04 stsp 	const struct got_error *err = NULL;
1845 2d9e6abf 2022-05-04 stsp 
1846 2d9e6abf 2022-05-04 stsp 	/* Validate decompression and obtain the decompressed size. */
1847 2d9e6abf 2022-05-04 stsp 	err = read_delta_data(delta_buf, delta_len, delta_len_compressed,
1848 2d9e6abf 2022-05-04 stsp 	    delta_data_offset, pack);
1849 2d9e6abf 2022-05-04 stsp 	if (err)
1850 2d9e6abf 2022-05-04 stsp 		return err;
1851 2d9e6abf 2022-05-04 stsp 
1852 2d9e6abf 2022-05-04 stsp 	/* Read delta base/result sizes from head of delta stream. */
1853 2d9e6abf 2022-05-04 stsp 	err = got_delta_get_sizes(base_size, result_size,
1854 2d9e6abf 2022-05-04 stsp 	    *delta_buf, *delta_len);
1855 2d9e6abf 2022-05-04 stsp 	if (err)
1856 2d9e6abf 2022-05-04 stsp 		goto done;
1857 2d9e6abf 2022-05-04 stsp 
1858 2d9e6abf 2022-05-04 stsp 	/* Discard decompressed delta and read it again in compressed form. */
1859 2d9e6abf 2022-05-04 stsp 	free(*delta_buf);
1860 2d9e6abf 2022-05-04 stsp 	*delta_buf = malloc(*delta_len_compressed);
1861 2d9e6abf 2022-05-04 stsp 	if (*delta_buf == NULL) {
1862 2d9e6abf 2022-05-04 stsp 		err = got_error_from_errno("malloc");
1863 2d9e6abf 2022-05-04 stsp 		goto done;
1864 2d9e6abf 2022-05-04 stsp 	}
1865 2d9e6abf 2022-05-04 stsp 	if (pack->map) {
1866 82031ac8 2022-10-24 op 		if (delta_data_offset >= pack->filesize) {
1867 2d9e6abf 2022-05-04 stsp 			err = got_error(GOT_ERR_PACK_OFFSET);
1868 82031ac8 2022-10-24 op 			goto done;
1869 82031ac8 2022-10-24 op 		}
1870 2d9e6abf 2022-05-04 stsp 		memcpy(*delta_buf, pack->map + delta_data_offset,
1871 2d9e6abf 2022-05-04 stsp 		    *delta_len_compressed);
1872 2d9e6abf 2022-05-04 stsp 	} else {
1873 2d9e6abf 2022-05-04 stsp 		ssize_t n;
1874 2d9e6abf 2022-05-04 stsp 		if (lseek(pack->fd, delta_data_offset, SEEK_SET) == -1) {
1875 2d9e6abf 2022-05-04 stsp 			err = got_error_from_errno("lseek");
1876 2d9e6abf 2022-05-04 stsp 			goto done;
1877 2d9e6abf 2022-05-04 stsp 		}
1878 2d9e6abf 2022-05-04 stsp 		n = read(pack->fd, *delta_buf, *delta_len_compressed);
1879 2d9e6abf 2022-05-04 stsp 		if (n < 0) {
1880 2d9e6abf 2022-05-04 stsp 			err = got_error_from_errno("read");
1881 2d9e6abf 2022-05-04 stsp 			goto done;
1882 2d9e6abf 2022-05-04 stsp 		} else if (n != *delta_len_compressed) {
1883 2d9e6abf 2022-05-04 stsp 			err = got_error(GOT_ERR_IO);
1884 2d9e6abf 2022-05-04 stsp 			goto done;
1885 2d9e6abf 2022-05-04 stsp 		}
1886 2d9e6abf 2022-05-04 stsp 	}
1887 2d9e6abf 2022-05-04 stsp done:
1888 2d9e6abf 2022-05-04 stsp 	if (err) {
1889 2d9e6abf 2022-05-04 stsp 		free(*delta_buf);
1890 2d9e6abf 2022-05-04 stsp 		*delta_buf = NULL;
1891 2d9e6abf 2022-05-04 stsp 		*delta_len = 0;
1892 2d9e6abf 2022-05-04 stsp 		*delta_len_compressed = 0;
1893 2d9e6abf 2022-05-04 stsp 		*base_size = 0;
1894 2d9e6abf 2022-05-04 stsp 		*result_size = 0;
1895 2d9e6abf 2022-05-04 stsp 	}
1896 2d9e6abf 2022-05-04 stsp 	return err;
1897 2d9e6abf 2022-05-04 stsp }
1898 2d9e6abf 2022-05-04 stsp 
1899 67fd6849 2022-02-13 stsp const struct got_error *
1900 67fd6849 2022-02-13 stsp got_packfile_extract_raw_delta(uint8_t **delta_buf, size_t *delta_size,
1901 24b7de1c 2022-12-03 stsp     size_t *delta_compressed_size, off_t *delta_offset,
1902 24b7de1c 2022-12-03 stsp     off_t *delta_data_offset, off_t *base_offset,
1903 2d9e6abf 2022-05-04 stsp     struct got_object_id *base_id, uint64_t *base_size, uint64_t *result_size,
1904 2d9e6abf 2022-05-04 stsp     struct got_pack *pack, struct got_packidx *packidx, int idx)
1905 67fd6849 2022-02-13 stsp {
1906 67fd6849 2022-02-13 stsp 	const struct got_error *err = NULL;
1907 67fd6849 2022-02-13 stsp 	off_t offset;
1908 67fd6849 2022-02-13 stsp 	uint8_t type;
1909 67fd6849 2022-02-13 stsp 	uint64_t size;
1910 67fd6849 2022-02-13 stsp 	size_t tslen, delta_hdrlen;
1911 67fd6849 2022-02-13 stsp 
1912 67fd6849 2022-02-13 stsp 	*delta_buf = NULL;
1913 67fd6849 2022-02-13 stsp 	*delta_size = 0;
1914 2d9e6abf 2022-05-04 stsp 	*delta_compressed_size = 0;
1915 67fd6849 2022-02-13 stsp 	*delta_offset = 0;
1916 24b7de1c 2022-12-03 stsp 	*delta_data_offset = 0;
1917 67fd6849 2022-02-13 stsp 	*base_offset = 0;
1918 67fd6849 2022-02-13 stsp 	*base_size = 0;
1919 67fd6849 2022-02-13 stsp 	*result_size = 0;
1920 67fd6849 2022-02-13 stsp 
1921 67fd6849 2022-02-13 stsp 	offset = got_packidx_get_object_offset(packidx, idx);
1922 04aed155 2022-07-24 naddy 	if (offset == -1)
1923 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_BAD_PACKIDX);
1924 67fd6849 2022-02-13 stsp 
1925 67fd6849 2022-02-13 stsp 	if (offset >= pack->filesize)
1926 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1927 67fd6849 2022-02-13 stsp 
1928 67fd6849 2022-02-13 stsp 	err = got_pack_parse_object_type_and_size(&type, &size, &tslen,
1929 67fd6849 2022-02-13 stsp 	    pack, offset);
1930 67fd6849 2022-02-13 stsp 	if (err)
1931 67fd6849 2022-02-13 stsp 		return err;
1932 67fd6849 2022-02-13 stsp 
1933 67fd6849 2022-02-13 stsp 	if (tslen + size < tslen || offset + size < size ||
1934 67fd6849 2022-02-13 stsp 	    tslen + offset < tslen)
1935 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_PACK_OFFSET);
1936 67fd6849 2022-02-13 stsp 
1937 67fd6849 2022-02-13 stsp 	switch (type) {
1938 67fd6849 2022-02-13 stsp 	case GOT_OBJ_TYPE_OFFSET_DELTA:
1939 67fd6849 2022-02-13 stsp 		err = got_pack_parse_offset_delta(base_offset, &delta_hdrlen,
1940 67fd6849 2022-02-13 stsp 		    pack, offset, tslen);
1941 67fd6849 2022-02-13 stsp 		if (err)
1942 67fd6849 2022-02-13 stsp 			return err;
1943 67fd6849 2022-02-13 stsp 		break;
1944 67fd6849 2022-02-13 stsp 	case GOT_OBJ_TYPE_REF_DELTA:
1945 67fd6849 2022-02-13 stsp 		err = got_pack_parse_ref_delta(base_id, pack, offset, tslen);
1946 67fd6849 2022-02-13 stsp 		if (err)
1947 67fd6849 2022-02-13 stsp 			return err;
1948 67fd6849 2022-02-13 stsp 		delta_hdrlen = SHA1_DIGEST_LENGTH;
1949 67fd6849 2022-02-13 stsp 		break;
1950 67fd6849 2022-02-13 stsp 	default:
1951 67fd6849 2022-02-13 stsp 		return got_error_fmt(GOT_ERR_OBJ_TYPE,
1952 04aed155 2022-07-24 naddy 		    "non-delta object type %d found at offset %lld",
1953 04aed155 2022-07-24 naddy 		    type, (long long)offset);
1954 67fd6849 2022-02-13 stsp 	}
1955 67fd6849 2022-02-13 stsp 
1956 67fd6849 2022-02-13 stsp 	if (tslen + delta_hdrlen < delta_hdrlen ||
1957 67fd6849 2022-02-13 stsp 	    offset + delta_hdrlen < delta_hdrlen)
1958 67fd6849 2022-02-13 stsp 		return got_error(GOT_ERR_BAD_DELTA);
1959 67fd6849 2022-02-13 stsp 
1960 24b7de1c 2022-12-03 stsp 	*delta_data_offset = offset + tslen + delta_hdrlen;
1961 2d9e6abf 2022-05-04 stsp 	err = read_raw_delta_data(delta_buf, delta_size, delta_compressed_size,
1962 24b7de1c 2022-12-03 stsp 	    base_size, result_size, *delta_data_offset, pack, packidx);
1963 67fd6849 2022-02-13 stsp 	if (err)
1964 67fd6849 2022-02-13 stsp 		return err;
1965 67fd6849 2022-02-13 stsp 
1966 67fd6849 2022-02-13 stsp 	if (*delta_size != size) {
1967 67fd6849 2022-02-13 stsp 		err = got_error(GOT_ERR_BAD_DELTA);
1968 67fd6849 2022-02-13 stsp 		goto done;
1969 67fd6849 2022-02-13 stsp 	}
1970 67fd6849 2022-02-13 stsp 
1971 67fd6849 2022-02-13 stsp 	*delta_offset = offset;
1972 67fd6849 2022-02-13 stsp done:
1973 67fd6849 2022-02-13 stsp 	if (err) {
1974 67fd6849 2022-02-13 stsp 		free(*delta_buf);
1975 67fd6849 2022-02-13 stsp 		*delta_buf = NULL;
1976 2d9e6abf 2022-05-04 stsp 		*delta_size = 0;
1977 2d9e6abf 2022-05-04 stsp 		*delta_compressed_size = 0;
1978 2d9e6abf 2022-05-04 stsp 		*delta_offset = 0;
1979 2d9e6abf 2022-05-04 stsp 		*base_offset = 0;
1980 2d9e6abf 2022-05-04 stsp 		*base_size = 0;
1981 2d9e6abf 2022-05-04 stsp 		*result_size = 0;
1982 67fd6849 2022-02-13 stsp 	}
1983 67fd6849 2022-02-13 stsp 	return err;
1984 67fd6849 2022-02-13 stsp }