| File: | libexec/got-index-pack/got-index-pack.c |
| Warning: | line 722, column 10 Potential leak of memory pointed to by 'packidx.hdr.large_offsets' |
Press '?' to see keyboard shortcuts
Keyboard shortcuts:
| 1 | /* | |||
| 2 | * Copyright (c) 2019 Ori Bernstein <ori@openbsd.org> | |||
| 3 | * Copyright (c) 2020 Stefan Sperling <stsp@openbsd.org> | |||
| 4 | * | |||
| 5 | * Permission to use, copy, modify, and distribute this software for any | |||
| 6 | * purpose with or without fee is hereby granted, provided that the above | |||
| 7 | * copyright notice and this permission notice appear in all copies. | |||
| 8 | * | |||
| 9 | * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES | |||
| 10 | * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF | |||
| 11 | * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR | |||
| 12 | * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES | |||
| 13 | * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN | |||
| 14 | * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF | |||
| 15 | * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. | |||
| 16 | */ | |||
| 17 | ||||
| 18 | #include <sys/queue.h> | |||
| 19 | #include <sys/stat.h> | |||
| 20 | #include <sys/time.h> | |||
| 21 | #include <sys/types.h> | |||
| 22 | #include <sys/uio.h> | |||
| 23 | #include <sys/mman.h> | |||
| 24 | ||||
| 25 | #include <stdint.h> | |||
| 26 | #include <errno(*__errno()).h> | |||
| 27 | #include <imsg.h> | |||
| 28 | #include <limits.h> | |||
| 29 | #include <signal.h> | |||
| 30 | #include <stdio.h> | |||
| 31 | #include <stdlib.h> | |||
| 32 | #include <string.h> | |||
| 33 | #include <ctype.h> | |||
| 34 | #include <sha1.h> | |||
| 35 | #include <endian.h> | |||
| 36 | #include <fcntl.h> | |||
| 37 | #include <unistd.h> | |||
| 38 | #include <zlib.h> | |||
| 39 | #include <err.h> | |||
| 40 | #include <assert.h> | |||
| 41 | #include <dirent.h> | |||
| 42 | ||||
| 43 | #include "got_error.h" | |||
| 44 | #include "got_object.h" | |||
| 45 | ||||
| 46 | #include "got_lib_sha1.h" | |||
| 47 | #include "got_lib_delta.h" | |||
| 48 | #include "got_lib_inflate.h" | |||
| 49 | #include "got_lib_object.h" | |||
| 50 | #include "got_lib_object_parse.h" | |||
| 51 | #include "got_lib_object_idset.h" | |||
| 52 | #include "got_lib_privsep.h" | |||
| 53 | #include "got_lib_pack.h" | |||
| 54 | #include "got_lib_delta_cache.h" | |||
| 55 | ||||
| 56 | #ifndef nitems | |||
| 57 | #define nitems(_a)(sizeof((_a)) / sizeof((_a)[0])) (sizeof((_a)) / sizeof((_a)[0])) | |||
| 58 | #endif | |||
| 59 | ||||
| 60 | struct got_indexed_object { | |||
| 61 | struct got_object_id id; | |||
| 62 | ||||
| 63 | /* | |||
| 64 | * Has this object been fully resolved? | |||
| 65 | * If so, we know its ID, otherwise we don't and 'id' is invalid. | |||
| 66 | */ | |||
| 67 | int valid; | |||
| 68 | ||||
| 69 | /* Offset of type+size field for this object in pack file. */ | |||
| 70 | off_t off; | |||
| 71 | ||||
| 72 | /* Type+size values parsed from pack file. */ | |||
| 73 | uint8_t type; | |||
| 74 | uint64_t size; | |||
| 75 | ||||
| 76 | /* Length of on-disk type+size data. */ | |||
| 77 | size_t tslen; | |||
| 78 | ||||
| 79 | /* Length of object data following type+size. */ | |||
| 80 | size_t len; | |||
| 81 | ||||
| 82 | uint32_t crc; | |||
| 83 | ||||
| 84 | union { | |||
| 85 | struct { | |||
| 86 | /* For ref deltas. */ | |||
| 87 | struct got_object_id ref_id; | |||
| 88 | } ref; | |||
| 89 | struct { | |||
| 90 | /* For offset deltas. */ | |||
| 91 | off_t base_offset; | |||
| 92 | size_t base_offsetlen; | |||
| 93 | } ofs; | |||
| 94 | } delta; | |||
| 95 | }; | |||
| 96 | ||||
| 97 | static void | |||
| 98 | putbe32(char *b, uint32_t n) | |||
| 99 | { | |||
| 100 | b[0] = n >> 24; | |||
| 101 | b[1] = n >> 16; | |||
| 102 | b[2] = n >> 8; | |||
| 103 | b[3] = n >> 0; | |||
| 104 | } | |||
| 105 | ||||
| 106 | static const struct got_error * | |||
| 107 | get_obj_type_label(const char **label, int obj_type) | |||
| 108 | { | |||
| 109 | const struct got_error *err = NULL((void *)0); | |||
| 110 | ||||
| 111 | switch (obj_type) { | |||
| 112 | case GOT_OBJ_TYPE_BLOB3: | |||
| 113 | *label = GOT_OBJ_LABEL_BLOB"blob"; | |||
| 114 | break; | |||
| 115 | case GOT_OBJ_TYPE_TREE2: | |||
| 116 | *label = GOT_OBJ_LABEL_TREE"tree"; | |||
| 117 | break; | |||
| 118 | case GOT_OBJ_TYPE_COMMIT1: | |||
| 119 | *label = GOT_OBJ_LABEL_COMMIT"commit"; | |||
| 120 | break; | |||
| 121 | case GOT_OBJ_TYPE_TAG4: | |||
| 122 | *label = GOT_OBJ_LABEL_TAG"tag"; | |||
| 123 | break; | |||
| 124 | default: | |||
| 125 | *label = NULL((void *)0); | |||
| 126 | err = got_error(GOT_ERR_OBJ_TYPE11); | |||
| 127 | break; | |||
| 128 | } | |||
| 129 | ||||
| 130 | return err; | |||
| 131 | } | |||
| 132 | ||||
| 133 | static const struct got_error * | |||
| 134 | read_checksum(uint32_t *crc, SHA1_CTX *sha1_ctx, int fd, size_t len) | |||
| 135 | { | |||
| 136 | uint8_t buf[8192]; | |||
| 137 | size_t n; | |||
| 138 | ssize_t r; | |||
| 139 | ||||
| 140 | for (n = len; n > 0; n -= r){ | |||
| 141 | r = read(fd, buf, n > sizeof(buf) ? sizeof(buf) : n); | |||
| 142 | if (r == -1) | |||
| 143 | return got_error_from_errno("read"); | |||
| 144 | if (r == 0) | |||
| 145 | break; | |||
| 146 | if (crc) | |||
| 147 | *crc = crc32(*crc, buf, r); | |||
| 148 | if (sha1_ctx) | |||
| 149 | SHA1Update(sha1_ctx, buf, r); | |||
| 150 | } | |||
| 151 | ||||
| 152 | return NULL((void *)0); | |||
| 153 | } | |||
| 154 | ||||
| 155 | static const struct got_error * | |||
| 156 | read_file_sha1(SHA1_CTX *ctx, FILE *f, size_t len) | |||
| 157 | { | |||
| 158 | uint8_t buf[8192]; | |||
| 159 | size_t n, r; | |||
| 160 | ||||
| 161 | for (n = len; n > 0; n -= r) { | |||
| 162 | r = fread(buf, 1, n > sizeof(buf) ? sizeof(buf) : n, f); | |||
| 163 | if (r == 0) { | |||
| 164 | if (feof(f)(!__isthreaded ? (((f)->_flags & 0x0020) != 0) : (feof )(f))) | |||
| 165 | return NULL((void *)0); | |||
| 166 | return got_ferror(f, GOT_ERR_IO6); | |||
| 167 | } | |||
| 168 | SHA1Update(ctx, buf, r); | |||
| 169 | } | |||
| 170 | ||||
| 171 | return NULL((void *)0); | |||
| 172 | } | |||
| 173 | ||||
| 174 | static const struct got_error * | |||
| 175 | read_packed_object(struct got_pack *pack, struct got_indexed_object *obj, | |||
| 176 | FILE *tmpfile, SHA1_CTX *pack_sha1_ctx) | |||
| 177 | { | |||
| 178 | const struct got_error *err = NULL((void *)0); | |||
| 179 | SHA1_CTX ctx; | |||
| 180 | uint8_t *data = NULL((void *)0); | |||
| 181 | size_t datalen = 0; | |||
| 182 | ssize_t n; | |||
| 183 | char *header; | |||
| 184 | size_t headerlen; | |||
| 185 | const char *obj_label; | |||
| 186 | size_t mapoff = obj->off; | |||
| 187 | struct got_inflate_checksum csum; | |||
| 188 | ||||
| 189 | csum.input_sha1 = pack_sha1_ctx; | |||
| 190 | csum.input_crc = &obj->crc; | |||
| 191 | ||||
| 192 | err = got_pack_parse_object_type_and_size(&obj->type, &obj->size, | |||
| 193 | &obj->tslen, pack, obj->off); | |||
| 194 | if (err) | |||
| 195 | return err; | |||
| 196 | ||||
| 197 | if (pack->map) { | |||
| 198 | obj->crc = crc32(obj->crc, pack->map + mapoff, obj->tslen); | |||
| 199 | SHA1Update(pack_sha1_ctx, pack->map + mapoff, obj->tslen); | |||
| 200 | mapoff += obj->tslen; | |||
| 201 | } else { | |||
| 202 | /* XXX Seek back and get the CRC of on-disk type+size bytes. */ | |||
| 203 | if (lseek(pack->fd, obj->off, SEEK_SET0) == -1) | |||
| 204 | return got_error_from_errno("lseek"); | |||
| 205 | err = read_checksum(&obj->crc, pack_sha1_ctx, | |||
| 206 | pack->fd, obj->tslen); | |||
| 207 | if (err) | |||
| 208 | return err; | |||
| 209 | } | |||
| 210 | ||||
| 211 | switch (obj->type) { | |||
| 212 | case GOT_OBJ_TYPE_BLOB3: | |||
| 213 | case GOT_OBJ_TYPE_COMMIT1: | |||
| 214 | case GOT_OBJ_TYPE_TREE2: | |||
| 215 | case GOT_OBJ_TYPE_TAG4: | |||
| 216 | if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX(8 * 1024 * 1024)) { | |||
| 217 | if (fseek(tmpfile, 0L, SEEK_SET0) == -1) { | |||
| 218 | err = got_error_from_errno("fseek"); | |||
| 219 | break; | |||
| 220 | } | |||
| 221 | if (pack->map) { | |||
| 222 | err = got_inflate_to_file_mmap(&datalen, | |||
| 223 | &obj->len, &csum, pack->map, mapoff, | |||
| 224 | pack->filesize - mapoff, tmpfile); | |||
| 225 | } else { | |||
| 226 | err = got_inflate_to_file_fd(&datalen, | |||
| 227 | &obj->len, &csum, pack->fd, tmpfile); | |||
| 228 | } | |||
| 229 | } else { | |||
| 230 | if (pack->map) { | |||
| 231 | err = got_inflate_to_mem_mmap(&data, &datalen, | |||
| 232 | &obj->len, &csum, pack->map, mapoff, | |||
| 233 | pack->filesize - mapoff); | |||
| 234 | } else { | |||
| 235 | err = got_inflate_to_mem_fd(&data, &datalen, | |||
| 236 | &obj->len, &csum, obj->size, pack->fd); | |||
| 237 | } | |||
| 238 | } | |||
| 239 | if (err) | |||
| 240 | break; | |||
| 241 | SHA1Init(&ctx); | |||
| 242 | err = get_obj_type_label(&obj_label, obj->type); | |||
| 243 | if (err) { | |||
| 244 | free(data); | |||
| 245 | break; | |||
| 246 | } | |||
| 247 | if (asprintf(&header, "%s %lld", obj_label, | |||
| 248 | (long long)obj->size) == -1) { | |||
| 249 | err = got_error_from_errno("asprintf"); | |||
| 250 | free(data); | |||
| 251 | break; | |||
| 252 | } | |||
| 253 | headerlen = strlen(header) + 1; | |||
| 254 | SHA1Update(&ctx, header, headerlen); | |||
| 255 | if (obj->size > GOT_DELTA_RESULT_SIZE_CACHED_MAX(8 * 1024 * 1024)) { | |||
| 256 | err = read_file_sha1(&ctx, tmpfile, datalen); | |||
| 257 | if (err) | |||
| 258 | break; | |||
| 259 | } else | |||
| 260 | SHA1Update(&ctx, data, datalen); | |||
| 261 | SHA1Final(obj->id.sha1, &ctx); | |||
| 262 | free(header); | |||
| 263 | free(data); | |||
| 264 | break; | |||
| 265 | case GOT_OBJ_TYPE_REF_DELTA7: | |||
| 266 | memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH20); | |||
| 267 | if (pack->map) { | |||
| 268 | if (mapoff + SHA1_DIGEST_LENGTH20 >= pack->filesize) { | |||
| 269 | err = got_error(GOT_ERR_BAD_PACKFILE16); | |||
| 270 | break; | |||
| 271 | } | |||
| 272 | memcpy(obj->delta.ref.ref_id.sha1, pack->map + mapoff, | |||
| 273 | SHA1_DIGEST_LENGTH20); | |||
| 274 | obj->crc = crc32(obj->crc, pack->map + mapoff, | |||
| 275 | SHA1_DIGEST_LENGTH20); | |||
| 276 | SHA1Update(pack_sha1_ctx, pack->map + mapoff, | |||
| 277 | SHA1_DIGEST_LENGTH20); | |||
| 278 | mapoff += SHA1_DIGEST_LENGTH20; | |||
| 279 | err = got_inflate_to_mem_mmap(NULL((void *)0), &datalen, | |||
| 280 | &obj->len, &csum, pack->map, mapoff, | |||
| 281 | pack->filesize - mapoff); | |||
| 282 | if (err) | |||
| 283 | break; | |||
| 284 | } else { | |||
| 285 | n = read(pack->fd, obj->delta.ref.ref_id.sha1, | |||
| 286 | SHA1_DIGEST_LENGTH20); | |||
| 287 | if (n == -1) { | |||
| 288 | err = got_error_from_errno("read"); | |||
| 289 | break; | |||
| 290 | } | |||
| 291 | if (n < sizeof(obj->id)) { | |||
| 292 | err = got_error(GOT_ERR_BAD_PACKFILE16); | |||
| 293 | break; | |||
| 294 | } | |||
| 295 | obj->crc = crc32(obj->crc, obj->delta.ref.ref_id.sha1, | |||
| 296 | SHA1_DIGEST_LENGTH20); | |||
| 297 | SHA1Update(pack_sha1_ctx, obj->delta.ref.ref_id.sha1, | |||
| 298 | SHA1_DIGEST_LENGTH20); | |||
| 299 | err = got_inflate_to_mem_fd(NULL((void *)0), &datalen, &obj->len, | |||
| 300 | &csum, obj->size, pack->fd); | |||
| 301 | if (err) | |||
| 302 | break; | |||
| 303 | } | |||
| 304 | obj->len += SHA1_DIGEST_LENGTH20; | |||
| 305 | break; | |||
| 306 | case GOT_OBJ_TYPE_OFFSET_DELTA6: | |||
| 307 | memset(obj->id.sha1, 0xff, SHA1_DIGEST_LENGTH20); | |||
| 308 | err = got_pack_parse_offset_delta(&obj->delta.ofs.base_offset, | |||
| 309 | &obj->delta.ofs.base_offsetlen, pack, obj->off, | |||
| 310 | obj->tslen); | |||
| 311 | if (err) | |||
| 312 | break; | |||
| 313 | ||||
| 314 | if (pack->map) { | |||
| 315 | obj->crc = crc32(obj->crc, pack->map + mapoff, | |||
| 316 | obj->delta.ofs.base_offsetlen); | |||
| 317 | SHA1Update(pack_sha1_ctx, pack->map + mapoff, | |||
| 318 | obj->delta.ofs.base_offsetlen); | |||
| 319 | mapoff += obj->delta.ofs.base_offsetlen; | |||
| 320 | err = got_inflate_to_mem_mmap(NULL((void *)0), &datalen, | |||
| 321 | &obj->len, &csum, pack->map, mapoff, | |||
| 322 | pack->filesize - mapoff); | |||
| 323 | if (err) | |||
| 324 | break; | |||
| 325 | } else { | |||
| 326 | /* | |||
| 327 | * XXX Seek back and get CRC and SHA1 of on-disk | |||
| 328 | * offset bytes. | |||
| 329 | */ | |||
| 330 | if (lseek(pack->fd, obj->off + obj->tslen, SEEK_SET0) | |||
| 331 | == -1) { | |||
| 332 | err = got_error_from_errno("lseek"); | |||
| 333 | break; | |||
| 334 | } | |||
| 335 | err = read_checksum(&obj->crc, pack_sha1_ctx, | |||
| 336 | pack->fd, obj->delta.ofs.base_offsetlen); | |||
| 337 | if (err) | |||
| 338 | break; | |||
| 339 | ||||
| 340 | err = got_inflate_to_mem_fd(NULL((void *)0), &datalen, &obj->len, | |||
| 341 | &csum, obj->size, pack->fd); | |||
| 342 | if (err) | |||
| 343 | break; | |||
| 344 | } | |||
| 345 | obj->len += obj->delta.ofs.base_offsetlen; | |||
| 346 | break; | |||
| 347 | default: | |||
| 348 | err = got_error(GOT_ERR_OBJ_TYPE11); | |||
| 349 | break; | |||
| 350 | } | |||
| 351 | ||||
| 352 | return err; | |||
| 353 | } | |||
| 354 | ||||
| 355 | static const struct got_error * | |||
| 356 | hwrite(int fd, void *buf, int len, SHA1_CTX *ctx) | |||
| 357 | { | |||
| 358 | ssize_t w; | |||
| 359 | ||||
| 360 | SHA1Update(ctx, buf, len); | |||
| 361 | ||||
| 362 | w = write(fd, buf, len); | |||
| 363 | if (w == -1) | |||
| 364 | return got_error_from_errno("write"); | |||
| 365 | if (w != len) | |||
| 366 | return got_error(GOT_ERR_IO6); | |||
| 367 | ||||
| 368 | return NULL((void *)0); | |||
| 369 | } | |||
| 370 | ||||
| 371 | static const struct got_error * | |||
| 372 | resolve_deltified_object(struct got_pack *pack, struct got_packidx *packidx, | |||
| 373 | struct got_indexed_object *obj, FILE *tmpfile, FILE *delta_base_file, | |||
| 374 | FILE *delta_accum_file) | |||
| 375 | { | |||
| 376 | const struct got_error *err = NULL((void *)0); | |||
| 377 | struct got_delta_chain deltas; | |||
| 378 | struct got_delta *delta; | |||
| 379 | uint8_t *buf = NULL((void *)0); | |||
| 380 | size_t len = 0; | |||
| 381 | SHA1_CTX ctx; | |||
| 382 | char *header = NULL((void *)0); | |||
| 383 | size_t headerlen; | |||
| 384 | uint64_t max_size; | |||
| 385 | int base_obj_type; | |||
| 386 | const char *obj_label; | |||
| 387 | ||||
| 388 | deltas.nentries = 0; | |||
| 389 | SIMPLEQ_INIT(&deltas.entries)do { (&deltas.entries)->sqh_first = ((void *)0); (& deltas.entries)->sqh_last = &(&deltas.entries)-> sqh_first; } while (0); | |||
| 390 | ||||
| 391 | err = got_pack_resolve_delta_chain(&deltas, packidx, pack, | |||
| 392 | obj->off, obj->tslen, obj->type, obj->size, | |||
| 393 | GOT_DELTA_CHAIN_RECURSION_MAX500); | |||
| 394 | if (err) | |||
| 395 | goto done; | |||
| 396 | ||||
| 397 | err = got_pack_get_delta_chain_max_size(&max_size, &deltas, pack); | |||
| 398 | if (err) | |||
| 399 | goto done; | |||
| 400 | if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX(8 * 1024 * 1024)) { | |||
| 401 | rewind(tmpfile); | |||
| 402 | rewind(delta_base_file); | |||
| 403 | rewind(delta_accum_file); | |||
| 404 | err = got_pack_dump_delta_chain_to_file(&len, &deltas, | |||
| 405 | pack, tmpfile, delta_base_file, delta_accum_file); | |||
| 406 | if (err) | |||
| 407 | goto done; | |||
| 408 | } else { | |||
| 409 | err = got_pack_dump_delta_chain_to_mem(&buf, &len, | |||
| 410 | &deltas, pack); | |||
| 411 | } | |||
| 412 | if (err) | |||
| 413 | goto done; | |||
| 414 | ||||
| 415 | err = got_delta_chain_get_base_type(&base_obj_type, &deltas); | |||
| 416 | if (err) | |||
| 417 | goto done; | |||
| 418 | err = get_obj_type_label(&obj_label, base_obj_type); | |||
| 419 | if (err) | |||
| 420 | goto done; | |||
| 421 | if (asprintf(&header, "%s %zd", obj_label, len) == -1) { | |||
| 422 | err = got_error_from_errno("asprintf"); | |||
| 423 | goto done; | |||
| 424 | } | |||
| 425 | headerlen = strlen(header) + 1; | |||
| 426 | SHA1Init(&ctx); | |||
| 427 | SHA1Update(&ctx, header, headerlen); | |||
| 428 | if (max_size > GOT_DELTA_RESULT_SIZE_CACHED_MAX(8 * 1024 * 1024)) { | |||
| 429 | err = read_file_sha1(&ctx, tmpfile, len); | |||
| 430 | if (err) | |||
| 431 | goto done; | |||
| 432 | } else | |||
| 433 | SHA1Update(&ctx, buf, len); | |||
| 434 | SHA1Final(obj->id.sha1, &ctx); | |||
| 435 | done: | |||
| 436 | free(buf); | |||
| 437 | free(header); | |||
| 438 | while (!SIMPLEQ_EMPTY(&deltas.entries)(((&deltas.entries)->sqh_first) == ((void *)0))) { | |||
| 439 | delta = SIMPLEQ_FIRST(&deltas.entries)((&deltas.entries)->sqh_first); | |||
| 440 | SIMPLEQ_REMOVE_HEAD(&deltas.entries, entry)do { if (((&deltas.entries)->sqh_first = (&deltas. entries)->sqh_first->entry.sqe_next) == ((void *)0)) (& deltas.entries)->sqh_last = &(&deltas.entries)-> sqh_first; } while (0); | |||
| 441 | free(delta); | |||
| 442 | } | |||
| 443 | return err; | |||
| 444 | } | |||
| 445 | ||||
| 446 | /* Determine the slot in the pack index a given object ID should use. */ | |||
| 447 | static int | |||
| 448 | find_object_idx(struct got_packidx *packidx, uint8_t *sha1) | |||
| 449 | { | |||
| 450 | u_int8_t id0 = sha1[0]; | |||
| 451 | uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [0xff]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [0xff]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr .fanout_table[0xff]) & 0xff00) << 8 | ((__uint32_t) (packidx->hdr.fanout_table[0xff]) & 0xff0000) >> 8 | ((__uint32_t)(packidx->hdr.fanout_table[0xff]) & 0xff000000 ) >> 24) : __swap32md(packidx->hdr.fanout_table[0xff ])); | |||
| 452 | int left = 0, right = nindexed - 1; | |||
| 453 | int cmp = 0, i = 0; | |||
| 454 | ||||
| 455 | if (id0 > 0) | |||
| 456 | left = be32toh(packidx->hdr.fanout_table[id0 - 1])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [id0 - 1]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [id0 - 1]) & 0xff) << 24 | ((__uint32_t)(packidx-> hdr.fanout_table[id0 - 1]) & 0xff00) << 8 | ((__uint32_t )(packidx->hdr.fanout_table[id0 - 1]) & 0xff0000) >> 8 | ((__uint32_t)(packidx->hdr.fanout_table[id0 - 1]) & 0xff000000) >> 24) : __swap32md(packidx->hdr.fanout_table [id0 - 1])); | |||
| 457 | ||||
| 458 | while (left <= right) { | |||
| 459 | struct got_packidx_object_id *oid; | |||
| 460 | ||||
| 461 | i = ((left + right) / 2); | |||
| 462 | oid = &packidx->hdr.sorted_ids[i]; | |||
| 463 | ||||
| 464 | cmp = memcmp(sha1, oid->sha1, SHA1_DIGEST_LENGTH20); | |||
| 465 | if (cmp == 0) | |||
| 466 | return -1; /* object already indexed */ | |||
| 467 | else if (cmp > 0) | |||
| 468 | left = i + 1; | |||
| 469 | else if (cmp < 0) | |||
| 470 | right = i - 1; | |||
| 471 | } | |||
| 472 | ||||
| 473 | return left; | |||
| 474 | } | |||
| 475 | ||||
| 476 | #if 0 | |||
| 477 | static void | |||
| 478 | print_packidx(struct got_packidx *packidx) | |||
| 479 | { | |||
| 480 | uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [0xff]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [0xff]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr .fanout_table[0xff]) & 0xff00) << 8 | ((__uint32_t) (packidx->hdr.fanout_table[0xff]) & 0xff0000) >> 8 | ((__uint32_t)(packidx->hdr.fanout_table[0xff]) & 0xff000000 ) >> 24) : __swap32md(packidx->hdr.fanout_table[0xff ])); | |||
| 481 | int i; | |||
| 482 | ||||
| 483 | fprintf(stderr(&__sF[2]), "object IDs:\n"); | |||
| 484 | for (i = 0; i < nindexed; i++) { | |||
| 485 | char hex[SHA1_DIGEST_STRING_LENGTH(20 * 2 + 1)]; | |||
| 486 | got_sha1_digest_to_str(packidx->hdr.sorted_ids[i].sha1, | |||
| 487 | hex, sizeof(hex)); | |||
| 488 | fprintf(stderr(&__sF[2]), "%s\n", hex); | |||
| 489 | } | |||
| 490 | fprintf(stderr(&__sF[2]), "\n"); | |||
| 491 | ||||
| 492 | fprintf(stderr(&__sF[2]), "object offsets:\n"); | |||
| 493 | for (i = 0; i < nindexed; i++) { | |||
| 494 | uint32_t offset = be32toh(packidx->hdr.offsets[i])(__uint32_t)(__builtin_constant_p(packidx->hdr.offsets[i]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.offsets[i]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr.offsets[i] ) & 0xff00) << 8 | ((__uint32_t)(packidx->hdr.offsets [i]) & 0xff0000) >> 8 | ((__uint32_t)(packidx->hdr .offsets[i]) & 0xff000000) >> 24) : __swap32md(packidx ->hdr.offsets[i])); | |||
| 495 | if (offset & GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX0x80000000) { | |||
| 496 | int j = offset & GOT_PACKIDX_OFFSET_VAL_MASK0x7fffffff; | |||
| 497 | fprintf(stderr(&__sF[2]), "%u -> %llu\n", offset, | |||
| 498 | be64toh(packidx->hdr.large_offsets[j])(__uint64_t)(__builtin_constant_p(packidx->hdr.large_offsets [j]) ? (__uint64_t)((((__uint64_t)(packidx->hdr.large_offsets [j]) & 0xff) << 56) | ((__uint64_t)(packidx->hdr .large_offsets[j]) & 0xff00ULL) << 40 | ((__uint64_t )(packidx->hdr.large_offsets[j]) & 0xff0000ULL) << 24 | ((__uint64_t)(packidx->hdr.large_offsets[j]) & 0xff000000ULL ) << 8 | ((__uint64_t)(packidx->hdr.large_offsets[j] ) & 0xff00000000ULL) >> 8 | ((__uint64_t)(packidx-> hdr.large_offsets[j]) & 0xff0000000000ULL) >> 24 | ( (__uint64_t)(packidx->hdr.large_offsets[j]) & 0xff000000000000ULL ) >> 40 | ((__uint64_t)(packidx->hdr.large_offsets[j ]) & 0xff00000000000000ULL) >> 56) : __swap64md(packidx ->hdr.large_offsets[j]))); | |||
| 499 | } else | |||
| 500 | fprintf(stderr(&__sF[2]), "%u\n", offset); | |||
| 501 | } | |||
| 502 | fprintf(stderr(&__sF[2]), "\n"); | |||
| 503 | ||||
| 504 | fprintf(stderr(&__sF[2]), "fanout table:"); | |||
| 505 | for (i = 0; i <= 0xff; i++) | |||
| 506 | fprintf(stderr(&__sF[2]), " %u", be32toh(packidx->hdr.fanout_table[i])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [i]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [i]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr. fanout_table[i]) & 0xff00) << 8 | ((__uint32_t)(packidx ->hdr.fanout_table[i]) & 0xff0000) >> 8 | ((__uint32_t )(packidx->hdr.fanout_table[i]) & 0xff000000) >> 24) : __swap32md(packidx->hdr.fanout_table[i]))); | |||
| 507 | fprintf(stderr(&__sF[2]), "\n"); | |||
| 508 | } | |||
| 509 | #endif | |||
| 510 | ||||
| 511 | static void | |||
| 512 | add_indexed_object(struct got_packidx *packidx, uint32_t idx, | |||
| 513 | struct got_indexed_object *obj) | |||
| 514 | { | |||
| 515 | int i; | |||
| 516 | ||||
| 517 | memcpy(packidx->hdr.sorted_ids[idx].sha1, obj->id.sha1, | |||
| 518 | SHA1_DIGEST_LENGTH20); | |||
| 519 | packidx->hdr.crc32[idx] = htobe32(obj->crc)(__uint32_t)(__builtin_constant_p(obj->crc) ? (__uint32_t) (((__uint32_t)(obj->crc) & 0xff) << 24 | ((__uint32_t )(obj->crc) & 0xff00) << 8 | ((__uint32_t)(obj-> crc) & 0xff0000) >> 8 | ((__uint32_t)(obj->crc) & 0xff000000) >> 24) : __swap32md(obj->crc)); | |||
| 520 | if (obj->off < GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX0x80000000) | |||
| 521 | packidx->hdr.offsets[idx] = htobe32(obj->off)(__uint32_t)(__builtin_constant_p(obj->off) ? (__uint32_t) (((__uint32_t)(obj->off) & 0xff) << 24 | ((__uint32_t )(obj->off) & 0xff00) << 8 | ((__uint32_t)(obj-> off) & 0xff0000) >> 8 | ((__uint32_t)(obj->off) & 0xff000000) >> 24) : __swap32md(obj->off)); | |||
| 522 | else { | |||
| 523 | packidx->hdr.offsets[idx] = htobe32(packidx->nlargeobj |(__uint32_t)(__builtin_constant_p(packidx->nlargeobj | 0x80000000 ) ? (__uint32_t)(((__uint32_t)(packidx->nlargeobj | 0x80000000 ) & 0xff) << 24 | ((__uint32_t)(packidx->nlargeobj | 0x80000000) & 0xff00) << 8 | ((__uint32_t)(packidx ->nlargeobj | 0x80000000) & 0xff0000) >> 8 | ((__uint32_t )(packidx->nlargeobj | 0x80000000) & 0xff000000) >> 24) : __swap32md(packidx->nlargeobj | 0x80000000)) | |||
| 524 | GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX)(__uint32_t)(__builtin_constant_p(packidx->nlargeobj | 0x80000000 ) ? (__uint32_t)(((__uint32_t)(packidx->nlargeobj | 0x80000000 ) & 0xff) << 24 | ((__uint32_t)(packidx->nlargeobj | 0x80000000) & 0xff00) << 8 | ((__uint32_t)(packidx ->nlargeobj | 0x80000000) & 0xff0000) >> 8 | ((__uint32_t )(packidx->nlargeobj | 0x80000000) & 0xff000000) >> 24) : __swap32md(packidx->nlargeobj | 0x80000000)); | |||
| 525 | packidx->hdr.large_offsets[packidx->nlargeobj] = | |||
| 526 | htobe64(obj->off)(__uint64_t)(__builtin_constant_p(obj->off) ? (__uint64_t) ((((__uint64_t)(obj->off) & 0xff) << 56) | ((__uint64_t )(obj->off) & 0xff00ULL) << 40 | ((__uint64_t)(obj ->off) & 0xff0000ULL) << 24 | ((__uint64_t)(obj-> off) & 0xff000000ULL) << 8 | ((__uint64_t)(obj-> off) & 0xff00000000ULL) >> 8 | ((__uint64_t)(obj-> off) & 0xff0000000000ULL) >> 24 | ((__uint64_t)(obj ->off) & 0xff000000000000ULL) >> 40 | ((__uint64_t )(obj->off) & 0xff00000000000000ULL) >> 56) : __swap64md (obj->off)); | |||
| 527 | packidx->nlargeobj++; | |||
| 528 | } | |||
| 529 | ||||
| 530 | for (i = obj->id.sha1[0]; i <= 0xff; i++) { | |||
| 531 | uint32_t n = be32toh(packidx->hdr.fanout_table[i])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [i]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [i]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr. fanout_table[i]) & 0xff00) << 8 | ((__uint32_t)(packidx ->hdr.fanout_table[i]) & 0xff0000) >> 8 | ((__uint32_t )(packidx->hdr.fanout_table[i]) & 0xff000000) >> 24) : __swap32md(packidx->hdr.fanout_table[i])); | |||
| 532 | packidx->hdr.fanout_table[i] = htobe32(n + 1)(__uint32_t)(__builtin_constant_p(n + 1) ? (__uint32_t)(((__uint32_t )(n + 1) & 0xff) << 24 | ((__uint32_t)(n + 1) & 0xff00) << 8 | ((__uint32_t)(n + 1) & 0xff0000) >> 8 | ((__uint32_t)(n + 1) & 0xff000000) >> 24) : __swap32md (n + 1)); | |||
| 533 | } | |||
| 534 | } | |||
| 535 | ||||
| 536 | static int | |||
| 537 | indexed_obj_cmp(const void *pa, const void *pb) | |||
| 538 | { | |||
| 539 | struct got_indexed_object *a, *b; | |||
| 540 | ||||
| 541 | a = (struct got_indexed_object *)pa; | |||
| 542 | b = (struct got_indexed_object *)pb; | |||
| 543 | return got_object_id_cmp(&a->id, &b->id); | |||
| 544 | } | |||
| 545 | ||||
| 546 | static void | |||
| 547 | make_packidx(struct got_packidx *packidx, int nobj, | |||
| 548 | struct got_indexed_object *objects) | |||
| 549 | { | |||
| 550 | struct got_indexed_object *obj; | |||
| 551 | int i; | |||
| 552 | uint32_t idx = 0; | |||
| 553 | ||||
| 554 | qsort(objects, nobj, sizeof(struct got_indexed_object), | |||
| 555 | indexed_obj_cmp); | |||
| 556 | ||||
| 557 | memset(packidx->hdr.fanout_table, 0, | |||
| 558 | GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS(0xff + 1) * sizeof(uint32_t)); | |||
| 559 | packidx->nlargeobj = 0; | |||
| 560 | ||||
| 561 | for (i = 0; i < nobj; i++) { | |||
| 562 | obj = &objects[i]; | |||
| 563 | if (obj->valid) | |||
| 564 | add_indexed_object(packidx, idx++, obj); | |||
| 565 | } | |||
| 566 | } | |||
| 567 | ||||
| 568 | static void | |||
| 569 | update_packidx(struct got_packidx *packidx, int nobj, | |||
| 570 | struct got_indexed_object *obj) | |||
| 571 | { | |||
| 572 | int idx; | |||
| 573 | uint32_t nindexed = be32toh(packidx->hdr.fanout_table[0xff])(__uint32_t)(__builtin_constant_p(packidx->hdr.fanout_table [0xff]) ? (__uint32_t)(((__uint32_t)(packidx->hdr.fanout_table [0xff]) & 0xff) << 24 | ((__uint32_t)(packidx->hdr .fanout_table[0xff]) & 0xff00) << 8 | ((__uint32_t) (packidx->hdr.fanout_table[0xff]) & 0xff0000) >> 8 | ((__uint32_t)(packidx->hdr.fanout_table[0xff]) & 0xff000000 ) >> 24) : __swap32md(packidx->hdr.fanout_table[0xff ])); | |||
| 574 | ||||
| 575 | idx = find_object_idx(packidx, obj->id.sha1); | |||
| 576 | if (idx == -1) { | |||
| 577 | char hex[SHA1_DIGEST_STRING_LENGTH(20 * 2 + 1)]; | |||
| 578 | got_sha1_digest_to_str(obj->id.sha1, hex, sizeof(hex)); | |||
| 579 | return; /* object already indexed */ | |||
| 580 | } | |||
| 581 | ||||
| 582 | memmove(&packidx->hdr.sorted_ids[idx + 1], | |||
| 583 | &packidx->hdr.sorted_ids[idx], | |||
| 584 | sizeof(struct got_packidx_object_id) * (nindexed - idx)); | |||
| 585 | memmove(&packidx->hdr.offsets[idx + 1], &packidx->hdr.offsets[idx], | |||
| 586 | sizeof(uint32_t) * (nindexed - idx)); | |||
| 587 | ||||
| 588 | add_indexed_object(packidx, idx, obj); | |||
| 589 | } | |||
| 590 | ||||
| 591 | static const struct got_error * | |||
| 592 | send_index_pack_progress(struct imsgbuf *ibuf, int nobj_total, | |||
| 593 | int nobj_indexed, int nobj_loose, int nobj_resolved) | |||
| 594 | { | |||
| 595 | struct got_imsg_index_pack_progress iprogress; | |||
| 596 | ||||
| 597 | iprogress.nobj_total = nobj_total; | |||
| 598 | iprogress.nobj_indexed = nobj_indexed; | |||
| 599 | iprogress.nobj_loose = nobj_loose; | |||
| 600 | iprogress.nobj_resolved = nobj_resolved; | |||
| 601 | ||||
| 602 | if (imsg_compose(ibuf, GOT_IMSG_IDXPACK_PROGRESS, 0, 0, -1, | |||
| 603 | &iprogress, sizeof(iprogress)) == -1) | |||
| 604 | return got_error_from_errno("imsg_compose IDXPACK_PROGRESS"); | |||
| 605 | ||||
| 606 | return got_privsep_flush_imsg(ibuf); | |||
| 607 | } | |||
| 608 | ||||
| 609 | static const struct got_error * | |||
| 610 | send_index_pack_done(struct imsgbuf *ibuf) | |||
| 611 | { | |||
| 612 | if (imsg_compose(ibuf, GOT_IMSG_IDXPACK_DONE, 0, 0, -1, NULL((void *)0), 0) == -1) | |||
| 613 | return got_error_from_errno("imsg_compose FETCH"); | |||
| 614 | return got_privsep_flush_imsg(ibuf); | |||
| 615 | } | |||
| 616 | ||||
| 617 | ||||
| 618 | static const struct got_error * | |||
| 619 | index_pack(struct got_pack *pack, int idxfd, FILE *tmpfile, | |||
| 620 | FILE *delta_base_file, FILE *delta_accum_file, uint8_t *pack_sha1_expected, | |||
| 621 | struct imsgbuf *ibuf) | |||
| 622 | { | |||
| 623 | const struct got_error *err; | |||
| 624 | struct got_packfile_hdr hdr; | |||
| 625 | struct got_packidx packidx; | |||
| 626 | char buf[8]; | |||
| 627 | char pack_sha1[SHA1_DIGEST_LENGTH20]; | |||
| 628 | int nobj, nvalid, nloose, nresolved = 0, i; | |||
| 629 | struct got_indexed_object *objects = NULL((void *)0), *obj; | |||
| 630 | SHA1_CTX ctx; | |||
| 631 | uint8_t packidx_hash[SHA1_DIGEST_LENGTH20]; | |||
| 632 | ssize_t r, w; | |||
| 633 | int pass, have_ref_deltas = 0, first_delta_idx = -1; | |||
| 634 | size_t mapoff = 0; | |||
| 635 | int p_indexed = 0, last_p_indexed = -1; | |||
| 636 | int p_resolved = 0, last_p_resolved = -1; | |||
| 637 | ||||
| 638 | /* Require that pack file header and SHA1 trailer are present. */ | |||
| 639 | if (pack->filesize < sizeof(hdr) + SHA1_DIGEST_LENGTH20) | |||
| ||||
| 640 | return got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 641 | "short pack file"); | |||
| 642 | ||||
| 643 | if (pack->map) { | |||
| 644 | memcpy(&hdr, pack->map, sizeof(hdr)); | |||
| 645 | mapoff += sizeof(hdr); | |||
| 646 | } else { | |||
| 647 | r = read(pack->fd, &hdr, sizeof(hdr)); | |||
| 648 | if (r == -1) | |||
| 649 | return got_error_from_errno("read"); | |||
| 650 | if (r < sizeof(hdr)) | |||
| 651 | return got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 652 | "short pack file"); | |||
| 653 | } | |||
| 654 | ||||
| 655 | if (hdr.signature != htobe32(GOT_PACKFILE_SIGNATURE)(__uint32_t)(__builtin_constant_p(0x5041434b) ? (__uint32_t)( ((__uint32_t)(0x5041434b) & 0xff) << 24 | ((__uint32_t )(0x5041434b) & 0xff00) << 8 | ((__uint32_t)(0x5041434b ) & 0xff0000) >> 8 | ((__uint32_t)(0x5041434b) & 0xff000000) >> 24) : __swap32md(0x5041434b))) | |||
| 656 | return got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 657 | "bad packfile signature"); | |||
| 658 | if (hdr.version != htobe32(GOT_PACKFILE_VERSION)(__uint32_t)(__builtin_constant_p(2) ? (__uint32_t)(((__uint32_t )(2) & 0xff) << 24 | ((__uint32_t)(2) & 0xff00) << 8 | ((__uint32_t)(2) & 0xff0000) >> 8 | ( (__uint32_t)(2) & 0xff000000) >> 24) : __swap32md(2 ))) | |||
| 659 | return got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 660 | "bad packfile version"); | |||
| 661 | nobj = be32toh(hdr.nobjects)(__uint32_t)(__builtin_constant_p(hdr.nobjects) ? (__uint32_t )(((__uint32_t)(hdr.nobjects) & 0xff) << 24 | ((__uint32_t )(hdr.nobjects) & 0xff00) << 8 | ((__uint32_t)(hdr. nobjects) & 0xff0000) >> 8 | ((__uint32_t)(hdr.nobjects ) & 0xff000000) >> 24) : __swap32md(hdr.nobjects)); | |||
| 662 | if (nobj == 0) | |||
| 663 | return got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 664 | "bad packfile with zero objects"); | |||
| 665 | ||||
| 666 | /* We compute the SHA1 of pack file contents and verify later on. */ | |||
| 667 | SHA1Init(&ctx); | |||
| 668 | SHA1Update(&ctx, (void *)&hdr, sizeof(hdr)); | |||
| 669 | ||||
| 670 | /* | |||
| 671 | * Create an in-memory pack index which will grow as objects | |||
| 672 | * IDs in the pack file are discovered. Only fields used to | |||
| 673 | * read deltified objects will be needed by the pack.c library | |||
| 674 | * code, so setting up just a pack index header is sufficient. | |||
| 675 | */ | |||
| 676 | memset(&packidx, 0, sizeof(packidx)); | |||
| 677 | packidx.hdr.magic = malloc(sizeof(uint32_t)); | |||
| 678 | if (packidx.hdr.magic == NULL((void *)0)) | |||
| 679 | return got_error_from_errno("calloc"); | |||
| 680 | *packidx.hdr.magic = htobe32(GOT_PACKIDX_V2_MAGIC)(__uint32_t)(__builtin_constant_p(0xff744f63) ? (__uint32_t)( ((__uint32_t)(0xff744f63) & 0xff) << 24 | ((__uint32_t )(0xff744f63) & 0xff00) << 8 | ((__uint32_t)(0xff744f63 ) & 0xff0000) >> 8 | ((__uint32_t)(0xff744f63) & 0xff000000) >> 24) : __swap32md(0xff744f63)); | |||
| 681 | packidx.hdr.version = malloc(sizeof(uint32_t)); | |||
| 682 | if (packidx.hdr.version == NULL((void *)0)) { | |||
| 683 | err = got_error_from_errno("malloc"); | |||
| 684 | goto done; | |||
| 685 | } | |||
| 686 | *packidx.hdr.version = htobe32(GOT_PACKIDX_VERSION)(__uint32_t)(__builtin_constant_p(2) ? (__uint32_t)(((__uint32_t )(2) & 0xff) << 24 | ((__uint32_t)(2) & 0xff00) << 8 | ((__uint32_t)(2) & 0xff0000) >> 8 | ( (__uint32_t)(2) & 0xff000000) >> 24) : __swap32md(2 )); | |||
| 687 | packidx.hdr.fanout_table = calloc(GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS(0xff + 1), | |||
| 688 | sizeof(uint32_t)); | |||
| 689 | if (packidx.hdr.fanout_table == NULL((void *)0)) { | |||
| 690 | err = got_error_from_errno("calloc"); | |||
| 691 | goto done; | |||
| 692 | } | |||
| 693 | packidx.hdr.sorted_ids = calloc(nobj, | |||
| 694 | sizeof(struct got_packidx_object_id)); | |||
| 695 | if (packidx.hdr.sorted_ids == NULL((void *)0)) { | |||
| 696 | err = got_error_from_errno("calloc"); | |||
| 697 | goto done; | |||
| 698 | } | |||
| 699 | packidx.hdr.crc32 = calloc(nobj, sizeof(uint32_t)); | |||
| 700 | if (packidx.hdr.crc32 == NULL((void *)0)) { | |||
| 701 | err = got_error_from_errno("calloc"); | |||
| 702 | goto done; | |||
| 703 | } | |||
| 704 | packidx.hdr.offsets = calloc(nobj, sizeof(uint32_t)); | |||
| 705 | if (packidx.hdr.offsets == NULL((void *)0)) { | |||
| 706 | err = got_error_from_errno("calloc"); | |||
| 707 | goto done; | |||
| 708 | } | |||
| 709 | /* Large offsets table is empty for pack files < 2 GB. */ | |||
| 710 | if (pack->filesize >= GOT_PACKIDX_OFFSET_VAL_IS_LARGE_IDX0x80000000) { | |||
| 711 | packidx.hdr.large_offsets = calloc(nobj, sizeof(uint64_t)); | |||
| 712 | if (packidx.hdr.large_offsets == NULL((void *)0)) { | |||
| 713 | err = got_error_from_errno("calloc"); | |||
| 714 | goto done; | |||
| 715 | } | |||
| 716 | } | |||
| 717 | ||||
| 718 | nvalid = 0; | |||
| 719 | nloose = 0; | |||
| 720 | objects = calloc(nobj, sizeof(struct got_indexed_object)); | |||
| 721 | if (objects == NULL((void *)0)) | |||
| 722 | return got_error_from_errno("calloc"); | |||
| ||||
| 723 | ||||
| 724 | /* | |||
| 725 | * First pass: locate all objects and identify un-deltified objects. | |||
| 726 | * | |||
| 727 | * When this pass has completed we will know offset, type, size, and | |||
| 728 | * CRC information for all objects in this pack file. We won't know | |||
| 729 | * any of the actual object IDs of deltified objects yet since we | |||
| 730 | * will not yet attempt to combine deltas. | |||
| 731 | */ | |||
| 732 | pass = 1; | |||
| 733 | for (i = 0; i < nobj; i++) { | |||
| 734 | /* Don't send too many progress privsep messages. */ | |||
| 735 | p_indexed = ((i + 1) * 100) / nobj; | |||
| 736 | if (p_indexed != last_p_indexed) { | |||
| 737 | err = send_index_pack_progress(ibuf, nobj, i + 1, | |||
| 738 | nloose, 0); | |||
| 739 | if (err) | |||
| 740 | goto done; | |||
| 741 | last_p_indexed = p_indexed; | |||
| 742 | } | |||
| 743 | ||||
| 744 | obj = &objects[i]; | |||
| 745 | obj->crc = crc32(0L, NULL((void *)0), 0); | |||
| 746 | ||||
| 747 | /* Store offset to type+size information for this object. */ | |||
| 748 | if (pack->map) { | |||
| 749 | obj->off = mapoff; | |||
| 750 | } else { | |||
| 751 | obj->off = lseek(pack->fd, 0, SEEK_CUR1); | |||
| 752 | if (obj->off == -1) { | |||
| 753 | err = got_error_from_errno("lseek"); | |||
| 754 | goto done; | |||
| 755 | } | |||
| 756 | } | |||
| 757 | ||||
| 758 | err = read_packed_object(pack, obj, tmpfile, &ctx); | |||
| 759 | if (err) | |||
| 760 | goto done; | |||
| 761 | ||||
| 762 | if (pack->map) { | |||
| 763 | mapoff += obj->tslen + obj->len; | |||
| 764 | } else { | |||
| 765 | if (lseek(pack->fd, obj->off + obj->tslen + obj->len, | |||
| 766 | SEEK_SET0) == -1) { | |||
| 767 | err = got_error_from_errno("lseek"); | |||
| 768 | goto done; | |||
| 769 | } | |||
| 770 | } | |||
| 771 | ||||
| 772 | if (obj->type == GOT_OBJ_TYPE_BLOB3 || | |||
| 773 | obj->type == GOT_OBJ_TYPE_TREE2 || | |||
| 774 | obj->type == GOT_OBJ_TYPE_COMMIT1 || | |||
| 775 | obj->type == GOT_OBJ_TYPE_TAG4) { | |||
| 776 | obj->valid = 1; | |||
| 777 | nloose++; | |||
| 778 | } else { | |||
| 779 | if (first_delta_idx == -1) | |||
| 780 | first_delta_idx = i; | |||
| 781 | if (obj->type == GOT_OBJ_TYPE_REF_DELTA7) | |||
| 782 | have_ref_deltas = 1; | |||
| 783 | } | |||
| 784 | } | |||
| 785 | nvalid = nloose; | |||
| 786 | ||||
| 787 | /* | |||
| 788 | * Having done a full pass over the pack file and can now | |||
| 789 | * verify its checksum. | |||
| 790 | */ | |||
| 791 | SHA1Final(pack_sha1, &ctx); | |||
| 792 | if (memcmp(pack_sha1_expected, pack_sha1, SHA1_DIGEST_LENGTH20) != 0) { | |||
| 793 | err = got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 794 | "pack file checksum mismatch"); | |||
| 795 | goto done; | |||
| 796 | } | |||
| 797 | ||||
| 798 | /* Verify the SHA1 checksum stored at the end of the pack file. */ | |||
| 799 | if (pack->map) { | |||
| 800 | memcpy(pack_sha1_expected, pack->map + | |||
| 801 | pack->filesize - SHA1_DIGEST_LENGTH20, | |||
| 802 | SHA1_DIGEST_LENGTH20); | |||
| 803 | } else { | |||
| 804 | ssize_t n; | |||
| 805 | if (lseek(pack->fd, -SHA1_DIGEST_LENGTH20, SEEK_END2) == -1) { | |||
| 806 | err = got_error_from_errno("lseek"); | |||
| 807 | goto done; | |||
| 808 | } | |||
| 809 | n = read(pack->fd, pack_sha1_expected, SHA1_DIGEST_LENGTH20); | |||
| 810 | if (n == -1) { | |||
| 811 | err = got_error_from_errno("read"); | |||
| 812 | goto done; | |||
| 813 | } | |||
| 814 | if (n != SHA1_DIGEST_LENGTH20) { | |||
| 815 | err = got_error(GOT_ERR_IO6); | |||
| 816 | goto done; | |||
| 817 | } | |||
| 818 | } | |||
| 819 | if (memcmp(pack_sha1, pack_sha1_expected, SHA1_DIGEST_LENGTH20) != 0) { | |||
| 820 | err = got_error_msg(GOT_ERR_BAD_PACKFILE16, | |||
| 821 | "bad checksum in pack file trailer"); | |||
| 822 | goto done; | |||
| 823 | } | |||
| 824 | ||||
| 825 | if (first_delta_idx == -1) | |||
| 826 | first_delta_idx = 0; | |||
| 827 | ||||
| 828 | /* In order to resolve ref deltas we need an in-progress pack index. */ | |||
| 829 | if (have_ref_deltas) | |||
| 830 | make_packidx(&packidx, nobj, objects); | |||
| 831 | ||||
| 832 | /* | |||
| 833 | * Second pass: We can now resolve deltas to compute the IDs of | |||
| 834 | * objects which appear in deltified form. Because deltas can be | |||
| 835 | * chained this pass may require a couple of iterations until all | |||
| 836 | * IDs of deltified objects have been discovered. | |||
| 837 | */ | |||
| 838 | pass++; | |||
| 839 | while (nvalid != nobj) { | |||
| 840 | int n = 0; | |||
| 841 | /* | |||
| 842 | * This loop will only run once unless the pack file | |||
| 843 | * contains ref deltas which refer to objects located | |||
| 844 | * later in the pack file, which is unusual. | |||
| 845 | * Offset deltas can always be resolved in one pass | |||
| 846 | * unless the packfile is corrupt. | |||
| 847 | */ | |||
| 848 | for (i = first_delta_idx; i < nobj; i++) { | |||
| 849 | obj = &objects[i]; | |||
| 850 | if (obj->type != GOT_OBJ_TYPE_REF_DELTA7 && | |||
| 851 | obj->type != GOT_OBJ_TYPE_OFFSET_DELTA6) | |||
| 852 | continue; | |||
| 853 | ||||
| 854 | if (obj->valid) | |||
| 855 | continue; | |||
| 856 | ||||
| 857 | if (pack->map == NULL((void *)0) && lseek(pack->fd, | |||
| 858 | obj->off + obj->tslen, SEEK_SET0) == -1) { | |||
| 859 | err = got_error_from_errno("lseek"); | |||
| 860 | goto done; | |||
| 861 | } | |||
| 862 | ||||
| 863 | err = resolve_deltified_object(pack, &packidx, obj, | |||
| 864 | tmpfile, delta_base_file, delta_accum_file); | |||
| 865 | if (err) { | |||
| 866 | if (err->code != GOT_ERR_NO_OBJ17) | |||
| 867 | goto done; | |||
| 868 | /* | |||
| 869 | * We cannot resolve this object yet because | |||
| 870 | * a delta base is unknown. Try again later. | |||
| 871 | */ | |||
| 872 | continue; | |||
| 873 | } | |||
| 874 | ||||
| 875 | obj->valid = 1; | |||
| 876 | n++; | |||
| 877 | if (have_ref_deltas) | |||
| 878 | update_packidx(&packidx, nobj, obj); | |||
| 879 | /* Don't send too many progress privsep messages. */ | |||
| 880 | p_resolved = ((nresolved + n) * 100) / nobj; | |||
| 881 | if (p_resolved != last_p_resolved) { | |||
| 882 | err = send_index_pack_progress(ibuf, nobj, | |||
| 883 | nobj, nloose, nresolved + n); | |||
| 884 | if (err) | |||
| 885 | goto done; | |||
| 886 | last_p_resolved = p_resolved; | |||
| 887 | } | |||
| 888 | ||||
| 889 | } | |||
| 890 | if (pass++ > 3 && n == 0) { | |||
| 891 | static char msg[64]; | |||
| 892 | snprintf(msg, sizeof(msg), "could not resolve " | |||
| 893 | "any of deltas; packfile could be corrupt"); | |||
| 894 | err = got_error_msg(GOT_ERR_BAD_PACKFILE16, msg); | |||
| 895 | goto done; | |||
| 896 | ||||
| 897 | } | |||
| 898 | nresolved += n; | |||
| 899 | nvalid += nresolved; | |||
| 900 | } | |||
| 901 | ||||
| 902 | if (nloose + nresolved != nobj) { | |||
| 903 | static char msg[64]; | |||
| 904 | snprintf(msg, sizeof(msg), "discovered only %d of %d objects", | |||
| 905 | nloose + nresolved, nobj); | |||
| 906 | err = got_error_msg(GOT_ERR_BAD_PACKFILE16, msg); | |||
| 907 | goto done; | |||
| 908 | } | |||
| 909 | ||||
| 910 | err = send_index_pack_progress(ibuf, nobj, nobj, nloose, nresolved); | |||
| 911 | if (err) | |||
| 912 | goto done; | |||
| 913 | ||||
| 914 | make_packidx(&packidx, nobj, objects); | |||
| 915 | ||||
| 916 | free(objects); | |||
| 917 | objects = NULL((void *)0); | |||
| 918 | ||||
| 919 | SHA1Init(&ctx); | |||
| 920 | putbe32(buf, GOT_PACKIDX_V2_MAGIC0xff744f63); | |||
| 921 | putbe32(buf + 4, GOT_PACKIDX_VERSION2); | |||
| 922 | err = hwrite(idxfd, buf, 8, &ctx); | |||
| 923 | if (err) | |||
| 924 | goto done; | |||
| 925 | err = hwrite(idxfd, packidx.hdr.fanout_table, | |||
| 926 | GOT_PACKIDX_V2_FANOUT_TABLE_ITEMS(0xff + 1) * sizeof(uint32_t), &ctx); | |||
| 927 | if (err) | |||
| 928 | goto done; | |||
| 929 | err = hwrite(idxfd, packidx.hdr.sorted_ids, | |||
| 930 | nobj * SHA1_DIGEST_LENGTH20, &ctx); | |||
| 931 | if (err) | |||
| 932 | goto done; | |||
| 933 | err = hwrite(idxfd, packidx.hdr.crc32, nobj * sizeof(uint32_t), &ctx); | |||
| 934 | if (err) | |||
| 935 | goto done; | |||
| 936 | err = hwrite(idxfd, packidx.hdr.offsets, nobj * sizeof(uint32_t), | |||
| 937 | &ctx); | |||
| 938 | if (err) | |||
| 939 | goto done; | |||
| 940 | if (packidx.nlargeobj > 0) { | |||
| 941 | err = hwrite(idxfd, packidx.hdr.large_offsets, | |||
| 942 | packidx.nlargeobj * sizeof(uint64_t), &ctx); | |||
| 943 | if (err) | |||
| 944 | goto done; | |||
| 945 | } | |||
| 946 | err = hwrite(idxfd, pack_sha1, SHA1_DIGEST_LENGTH20, &ctx); | |||
| 947 | if (err) | |||
| 948 | goto done; | |||
| 949 | ||||
| 950 | SHA1Final(packidx_hash, &ctx); | |||
| 951 | w = write(idxfd, packidx_hash, sizeof(packidx_hash)); | |||
| 952 | if (w == -1) { | |||
| 953 | err = got_error_from_errno("write"); | |||
| 954 | goto done; | |||
| 955 | } | |||
| 956 | if (w != sizeof(packidx_hash)) { | |||
| 957 | err = got_error(GOT_ERR_IO6); | |||
| 958 | goto done; | |||
| 959 | } | |||
| 960 | done: | |||
| 961 | free(objects); | |||
| 962 | free(packidx.hdr.magic); | |||
| 963 | free(packidx.hdr.version); | |||
| 964 | free(packidx.hdr.fanout_table); | |||
| 965 | free(packidx.hdr.sorted_ids); | |||
| 966 | free(packidx.hdr.offsets); | |||
| 967 | free(packidx.hdr.large_offsets); | |||
| 968 | return err; | |||
| 969 | } | |||
| 970 | ||||
| 971 | int | |||
| 972 | main(int argc, char **argv) | |||
| 973 | { | |||
| 974 | const struct got_error *err = NULL((void *)0), *close_err; | |||
| 975 | struct imsgbuf ibuf; | |||
| 976 | struct imsg imsg; | |||
| 977 | size_t i; | |||
| 978 | int idxfd = -1, tmpfd = -1; | |||
| 979 | FILE *tmpfiles[3]; | |||
| 980 | struct got_pack pack; | |||
| 981 | uint8_t pack_hash[SHA1_DIGEST_LENGTH20]; | |||
| 982 | off_t packfile_size; | |||
| 983 | #if 0 | |||
| 984 | static int attached; | |||
| 985 | while (!attached) | |||
| 986 | sleep(1); | |||
| 987 | #endif | |||
| 988 | ||||
| 989 | for (i = 0; i < nitems(tmpfiles)(sizeof((tmpfiles)) / sizeof((tmpfiles)[0])); i++) | |||
| 990 | tmpfiles[i] = NULL((void *)0); | |||
| 991 | ||||
| 992 | memset(&pack, 0, sizeof(pack)); | |||
| 993 | pack.fd = -1; | |||
| 994 | pack.delta_cache = got_delta_cache_alloc(500, | |||
| 995 | GOT_DELTA_RESULT_SIZE_CACHED_MAX(8 * 1024 * 1024)); | |||
| 996 | if (pack.delta_cache == NULL((void *)0)) { | |||
| 997 | err = got_error_from_errno("got_delta_cache_alloc"); | |||
| 998 | goto done; | |||
| 999 | } | |||
| 1000 | ||||
| 1001 | imsg_init(&ibuf, GOT_IMSG_FD_CHILD(2 + 1)); | |||
| 1002 | #ifndef PROFILE | |||
| 1003 | /* revoke access to most system calls */ | |||
| 1004 | if (pledge("stdio recvfd", NULL((void *)0)) == -1) { | |||
| 1005 | err = got_error_from_errno("pledge"); | |||
| 1006 | got_privsep_send_error(&ibuf, err); | |||
| 1007 | return 1; | |||
| 1008 | } | |||
| 1009 | #endif | |||
| 1010 | err = got_privsep_recv_imsg(&imsg, &ibuf, 0); | |||
| 1011 | if (err) | |||
| 1012 | goto done; | |||
| 1013 | if (imsg.hdr.type == GOT_IMSG_STOP) | |||
| 1014 | goto done; | |||
| 1015 | if (imsg.hdr.type != GOT_IMSG_IDXPACK_REQUEST) { | |||
| 1016 | err = got_error(GOT_ERR_PRIVSEP_MSG39); | |||
| 1017 | goto done; | |||
| 1018 | } | |||
| 1019 | if (imsg.hdr.len - IMSG_HEADER_SIZEsizeof(struct imsg_hdr) != sizeof(pack_hash)) { | |||
| 1020 | err = got_error(GOT_ERR_PRIVSEP_LEN36); | |||
| 1021 | goto done; | |||
| 1022 | } | |||
| 1023 | memcpy(pack_hash, imsg.data, sizeof(pack_hash)); | |||
| 1024 | pack.fd = imsg.fd; | |||
| 1025 | ||||
| 1026 | err = got_privsep_recv_imsg(&imsg, &ibuf, 0); | |||
| 1027 | if (err) | |||
| 1028 | goto done; | |||
| 1029 | if (imsg.hdr.type == GOT_IMSG_STOP) | |||
| 1030 | goto done; | |||
| 1031 | if (imsg.hdr.type != GOT_IMSG_IDXPACK_OUTFD) { | |||
| 1032 | err = got_error(GOT_ERR_PRIVSEP_MSG39); | |||
| 1033 | goto done; | |||
| 1034 | } | |||
| 1035 | if (imsg.hdr.len - IMSG_HEADER_SIZEsizeof(struct imsg_hdr) != 0) { | |||
| 1036 | err = got_error(GOT_ERR_PRIVSEP_LEN36); | |||
| 1037 | goto done; | |||
| 1038 | } | |||
| 1039 | idxfd = imsg.fd; | |||
| 1040 | ||||
| 1041 | for (i = 0; i < nitems(tmpfiles)(sizeof((tmpfiles)) / sizeof((tmpfiles)[0])); i++) { | |||
| 1042 | err = got_privsep_recv_imsg(&imsg, &ibuf, 0); | |||
| 1043 | if (err) | |||
| 1044 | goto done; | |||
| 1045 | if (imsg.hdr.type == GOT_IMSG_STOP) | |||
| 1046 | goto done; | |||
| 1047 | if (imsg.hdr.type != GOT_IMSG_TMPFD) { | |||
| 1048 | err = got_error(GOT_ERR_PRIVSEP_MSG39); | |||
| 1049 | goto done; | |||
| 1050 | } | |||
| 1051 | if (imsg.hdr.len - IMSG_HEADER_SIZEsizeof(struct imsg_hdr) != 0) { | |||
| 1052 | err = got_error(GOT_ERR_PRIVSEP_LEN36); | |||
| 1053 | goto done; | |||
| 1054 | } | |||
| 1055 | tmpfd = imsg.fd; | |||
| 1056 | tmpfiles[i] = fdopen(tmpfd, "w+"); | |||
| 1057 | if (tmpfiles[i] == NULL((void *)0)) { | |||
| 1058 | err = got_error_from_errno("fdopen"); | |||
| 1059 | goto done; | |||
| 1060 | } | |||
| 1061 | tmpfd = -1; | |||
| 1062 | } | |||
| 1063 | ||||
| 1064 | if (lseek(pack.fd, 0, SEEK_END2) == -1) { | |||
| 1065 | err = got_error_from_errno("lseek"); | |||
| 1066 | goto done; | |||
| 1067 | } | |||
| 1068 | packfile_size = lseek(pack.fd, 0, SEEK_CUR1); | |||
| 1069 | if (packfile_size == -1) { | |||
| 1070 | err = got_error_from_errno("lseek"); | |||
| 1071 | goto done; | |||
| 1072 | } | |||
| 1073 | pack.filesize = packfile_size; /* XXX off_t vs size_t */ | |||
| 1074 | ||||
| 1075 | if (lseek(pack.fd, 0, SEEK_SET0) == -1) { | |||
| 1076 | err = got_error_from_errno("lseek"); | |||
| 1077 | goto done; | |||
| 1078 | } | |||
| 1079 | ||||
| 1080 | #ifndef GOT_PACK_NO_MMAP | |||
| 1081 | pack.map = mmap(NULL((void *)0), pack.filesize, PROT_READ0x01, MAP_PRIVATE0x0002, | |||
| 1082 | pack.fd, 0); | |||
| 1083 | if (pack.map == MAP_FAILED((void *)-1)) | |||
| 1084 | pack.map = NULL((void *)0); /* fall back to read(2) */ | |||
| 1085 | #endif | |||
| 1086 | err = index_pack(&pack, idxfd, tmpfiles[0], tmpfiles[1], tmpfiles[2], | |||
| 1087 | pack_hash, &ibuf); | |||
| 1088 | done: | |||
| 1089 | close_err = got_pack_close(&pack); | |||
| 1090 | if (close_err && err == NULL((void *)0)) | |||
| 1091 | err = close_err; | |||
| 1092 | if (idxfd != -1 && close(idxfd) == -1 && err == NULL((void *)0)) | |||
| 1093 | err = got_error_from_errno("close"); | |||
| 1094 | if (tmpfd != -1 && close(tmpfd) == -1 && err == NULL((void *)0)) | |||
| 1095 | err = got_error_from_errno("close"); | |||
| 1096 | for (i = 0; i < nitems(tmpfiles)(sizeof((tmpfiles)) / sizeof((tmpfiles)[0])); i++) { | |||
| 1097 | if (tmpfiles[i] != NULL((void *)0) && fclose(tmpfiles[i]) == EOF(-1) && | |||
| 1098 | err == NULL((void *)0)) | |||
| 1099 | err = got_error_from_errno("close"); | |||
| 1100 | } | |||
| 1101 | ||||
| 1102 | if (err == NULL((void *)0)) | |||
| 1103 | err = send_index_pack_done(&ibuf); | |||
| 1104 | if (err) { | |||
| 1105 | got_privsep_send_error(&ibuf, err); | |||
| 1106 | fprintf(stderr(&__sF[2]), "%s: %s\n", getprogname(), err->msg); | |||
| 1107 | got_privsep_send_error(&ibuf, err); | |||
| 1108 | exit(1); | |||
| 1109 | } | |||
| 1110 | ||||
| 1111 | exit(0); | |||
| 1112 | } |