builtin/unpack-objects.c

   1 #include "builtin.h"
   2 #include "cache.h"
   3 #include "config.h"
   4 #include "object.h"
   5 #include "delta.h"
   6 #include "pack.h"
   7 #include "blob.h"
   8 #include "commit.h"
   9 #include "tag.h"
  10 #include "tree.h"
  11 #include "tree-walk.h"
  12 #include "progress.h"
  13 #include "decorate.h"
  14 #include "fsck.h"
  15
  16 static int dry_run, quiet, recover, has_errors, strict;
  17 static const char unpack_usage[] = "git unpack-objects [-n] [-q] [-r] [--strict]";
  18
  19 /* We always read in 4kB chunks. */
  20 static unsigned char buffer[4096];
  21 static unsigned int offset, len;
  22 static off_t consumed_bytes;
  23 static off_t max_input_size;
  24 static git_SHA_CTX ctx;
  25 static struct fsck_options fsck_options = FSCK_OPTIONS_STRICT;
  26
  27 /*
  28  * When running under --strict mode, objects whose reachability are
  29  * suspect are kept in core without getting written in the object
  30  * store.
  31  */
  32 struct obj_buffer {
  33         char *buffer;
  34         unsigned long size;
  35 };
  36
  37 static struct decoration obj_decorate;
  38
  39 static struct obj_buffer *lookup_object_buffer(struct object *base)
  40 {
  41         return lookup_decoration(&obj_decorate, base);
  42 }
  43
  44 static void add_object_buffer(struct object *object, char *buffer, unsigned long size)
  45 {
  46         struct obj_buffer *obj;
  47         obj = xcalloc(1, sizeof(struct obj_buffer));
  48         obj->buffer = buffer;
  49         obj->size = size;
  50         if (add_decoration(&obj_decorate, object, obj))
  51                 die("object %s tried to add buffer twice!", oid_to_hex(&object->oid));
  52 }
  53
  54 /*
  55  * Make sure at least "min" bytes are available in the buffer, and
  56  * return the pointer to the buffer.
  57  */
  58 static void *fill(int min)
  59 {
  60         if (min <= len)
  61                 return buffer + offset;
  62         if (min > sizeof(buffer))
  63                 die("cannot fill %d bytes", min);
  64         if (offset) {
  65                 git_SHA1_Update(&ctx, buffer, offset);
  66                 memmove(buffer, buffer + offset, len);
  67                 offset = 0;
  68         }
  69         do {
  70                 ssize_t ret = xread(0, buffer + len, sizeof(buffer) - len);
  71                 if (ret <= 0) {
  72                         if (!ret)
  73                                 die("early EOF");
  74                         die_errno("read error on input");
  75                 }
  76                 len += ret;
  77         } while (len < min);
  78         return buffer;
  79 }
  80
  81 static void use(int bytes)
  82 {
  83         if (bytes > len)
  84                 die("used more bytes than were available");
  85         len -= bytes;
  86         offset += bytes;
  87
  88         /* make sure off_t is sufficiently large not to wrap */
  89         if (signed_add_overflows(consumed_bytes, bytes))
  90                 die("pack too large for current definition of off_t");
  91         consumed_bytes += bytes;
  92         if (max_input_size && consumed_bytes > max_input_size)
  93                 die(_("pack exceeds maximum allowed size"));
  94 }
  95
  96 static void *get_data(unsigned long size)
  97 {
  98         git_zstream stream;
  99         void *buf = xmallocz(size);
 100
 101         memset(&stream, 0, sizeof(stream));
 102
 103         stream.next_out = buf;
 104         stream.avail_out = size;
 105         stream.next_in = fill(1);
 106         stream.avail_in = len;
 107         git_inflate_init(&stream);
 108
 109         for (;;) {
 110                 int ret = git_inflate(&stream, 0);
 111                 use(len - stream.avail_in);
 112                 if (stream.total_out == size && ret == Z_STREAM_END)
 113                         break;
 114                 if (ret != Z_OK) {
 115                         error("inflate returned %d", ret);
 116                         FREE_AND_NULL(buf);
 117                         if (!recover)
 118                                 exit(1);
 119                         has_errors = 1;
 120                         break;
 121                 }
 122                 stream.next_in = fill(1);
 123                 stream.avail_in = len;
 124         }
 125         git_inflate_end(&stream);
 126         return buf;
 127 }
 128
 129 struct delta_info {
 130         struct object_id base_oid;
 131         unsigned nr;
 132         off_t base_offset;
 133         unsigned long size;
 134         void *delta;
 135         struct delta_info *next;
 136 };
 137
 138 static struct delta_info *delta_list;
 139
 140 static void add_delta_to_list(unsigned nr, const struct object_id *base_oid,
 141                               off_t base_offset,
 142                               void *delta, unsigned long size)
 143 {
 144         struct delta_info *info = xmalloc(sizeof(*info));
 145
 146         oidcpy(&info->base_oid, base_oid);
 147         info->base_offset = base_offset;
 148         info->size = size;
 149         info->delta = delta;
 150         info->nr = nr;
 151         info->next = delta_list;
 152         delta_list = info;
 153 }
 154
 155 struct obj_info {
 156         off_t offset;
 157         struct object_id oid;
 158         struct object *obj;
 159 };
 160
 161 /* Remember to update object flag allocation in object.h */
 162 #define FLAG_OPEN (1u<<20)
 163 #define FLAG_WRITTEN (1u<<21)
 164
 165 static struct obj_info *obj_list;
 166 static unsigned nr_objects;
 167
 168 /*
 169  * Called only from check_object() after it verified this object
 170  * is Ok.
 171  */
 172 static void write_cached_object(struct object *obj, struct obj_buffer *obj_buf)
 173 {
 174         struct object_id oid;
 175
 176         if (write_sha1_file(obj_buf->buffer, obj_buf->size, typename(obj->type), oid.hash) < 0)
 177                 die("failed to write object %s", oid_to_hex(&obj->oid));
 178         obj->flags |= FLAG_WRITTEN;
 179 }
 180
 181 /*
 182  * At the very end of the processing, write_rest() scans the objects
 183  * that have reachability requirements and calls this function.
 184  * Verify its reachability and validity recursively and write it out.
 185  */
 186 static int check_object(struct object *obj, int type, void *data, struct fsck_options *options)
 187 {
 188         struct obj_buffer *obj_buf;
 189
 190         if (!obj)
 191                 return 1;
 192
 193         if (obj->flags & FLAG_WRITTEN)
 194                 return 0;
 195
 196         if (type != OBJ_ANY && obj->type != type)
 197                 die("object type mismatch");
 198
 199         if (!(obj->flags & FLAG_OPEN)) {
 200                 unsigned long size;
 201                 int type = sha1_object_info(obj->oid.hash, &size);
 202                 if (type != obj->type || type <= 0)
 203                         die("object of unexpected type");
 204                 obj->flags |= FLAG_WRITTEN;
 205                 return 0;
 206         }
 207
 208         obj_buf = lookup_object_buffer(obj);
 209         if (!obj_buf)
 210                 die("Whoops! Cannot find object '%s'", oid_to_hex(&obj->oid));
 211         if (fsck_object(obj, obj_buf->buffer, obj_buf->size, &fsck_options))
 212                 die("Error in object");
 213         fsck_options.walk = check_object;
 214         if (fsck_walk(obj, NULL, &fsck_options))
 215                 die("Error on reachable objects of %s", oid_to_hex(&obj->oid));
 216         write_cached_object(obj, obj_buf);
 217         return 0;
 218 }
 219
 220 static void write_rest(void)
 221 {
 222         unsigned i;
 223         for (i = 0; i < nr_objects; i++) {
 224                 if (obj_list[i].obj)
 225                         check_object(obj_list[i].obj, OBJ_ANY, NULL, NULL);
 226         }
 227 }
 228
 229 static void added_object(unsigned nr, enum object_type type,
 230                          void *data, unsigned long size);
 231
 232 /*
 233  * Write out nr-th object from the list, now we know the contents
 234  * of it.  Under --strict, this buffers structured objects in-core,
 235  * to be checked at the end.
 236  */
 237 static void write_object(unsigned nr, enum object_type type,
 238                          void *buf, unsigned long size)
 239 {
 240         if (!strict) {
 241                 if (write_sha1_file(buf, size, typename(type), obj_list[nr].oid.hash) < 0)
 242                         die("failed to write object");
 243                 added_object(nr, type, buf, size);
 244                 free(buf);
 245                 obj_list[nr].obj = NULL;
 246         } else if (type == OBJ_BLOB) {
 247                 struct blob *blob;
 248                 if (write_sha1_file(buf, size, typename(type), obj_list[nr].oid.hash) < 0)
 249                         die("failed to write object");
 250                 added_object(nr, type, buf, size);
 251                 free(buf);
 252
 253                 blob = lookup_blob(&obj_list[nr].oid);
 254                 if (blob)
 255                         blob->object.flags |= FLAG_WRITTEN;
 256                 else
 257                         die("invalid blob object");
 258                 obj_list[nr].obj = NULL;
 259         } else {
 260                 struct object *obj;
 261                 int eaten;
 262                 hash_sha1_file(buf, size, typename(type), obj_list[nr].oid.hash);
 263                 added_object(nr, type, buf, size);
 264                 obj = parse_object_buffer(&obj_list[nr].oid, type, size, buf,
 265                                           &eaten);
 266                 if (!obj)
 267                         die("invalid %s", typename(type));
 268                 add_object_buffer(obj, buf, size);
 269                 obj->flags |= FLAG_OPEN;
 270                 obj_list[nr].obj = obj;
 271         }
 272 }
 273
 274 static void resolve_delta(unsigned nr, enum object_type type,
 275                           void *base, unsigned long base_size,
 276                           void *delta, unsigned long delta_size)
 277 {
 278         void *result;
 279         unsigned long result_size;
 280
 281         result = patch_delta(base, base_size,
 282                              delta, delta_size,
 283                              &result_size);
 284         if (!result)
 285                 die("failed to apply delta");
 286         free(delta);
 287         write_object(nr, type, result, result_size);
 288 }
 289
 290 /*
 291  * We now know the contents of an object (which is nr-th in the pack);
 292  * resolve all the deltified objects that are based on it.
 293  */
 294 static void added_object(unsigned nr, enum object_type type,
 295                          void *data, unsigned long size)
 296 {
 297         struct delta_info **p = &delta_list;
 298         struct delta_info *info;
 299
 300         while ((info = *p) != NULL) {
 301                 if (!oidcmp(&info->base_oid, &obj_list[nr].oid) ||
 302                     info->base_offset == obj_list[nr].offset) {
 303                         *p = info->next;
 304                         p = &delta_list;
 305                         resolve_delta(info->nr, type, data, size,
 306                                       info->delta, info->size);
 307                         free(info);
 308                         continue;
 309                 }
 310                 p = &info->next;
 311         }
 312 }
 313
 314 static void unpack_non_delta_entry(enum object_type type, unsigned long size,
 315                                    unsigned nr)
 316 {
 317         void *buf = get_data(size);
 318
 319         if (!dry_run && buf)
 320                 write_object(nr, type, buf, size);
 321         else
 322                 free(buf);
 323 }
 324
 325 static int resolve_against_held(unsigned nr, const struct object_id *base,
 326                                 void *delta_data, unsigned long delta_size)
 327 {
 328         struct object *obj;
 329         struct obj_buffer *obj_buffer;
 330         obj = lookup_object(base->hash);
 331         if (!obj)
 332                 return 0;
 333         obj_buffer = lookup_object_buffer(obj);
 334         if (!obj_buffer)
 335                 return 0;
 336         resolve_delta(nr, obj->type, obj_buffer->buffer,
 337                       obj_buffer->size, delta_data, delta_size);
 338         return 1;
 339 }
 340
 341 static void unpack_delta_entry(enum object_type type, unsigned long delta_size,
 342                                unsigned nr)
 343 {
 344         void *delta_data, *base;
 345         unsigned long base_size;
 346         struct object_id base_oid;
 347
 348         if (type == OBJ_REF_DELTA) {
 349                 hashcpy(base_oid.hash, fill(GIT_SHA1_RAWSZ));
 350                 use(GIT_SHA1_RAWSZ);
 351                 delta_data = get_data(delta_size);
 352                 if (dry_run || !delta_data) {
 353                         free(delta_data);
 354                         return;
 355                 }
 356                 if (has_object_file(&base_oid))
 357                         ; /* Ok we have this one */
 358                 else if (resolve_against_held(nr, &base_oid,
 359                                               delta_data, delta_size))
 360                         return; /* we are done */
 361                 else {
 362                         /* cannot resolve yet --- queue it */
 363                         oidclr(&obj_list[nr].oid);
 364                         add_delta_to_list(nr, &base_oid, 0, delta_data, delta_size);
 365                         return;
 366                 }
 367         } else {
 368                 unsigned base_found = 0;
 369                 unsigned char *pack, c;
 370                 off_t base_offset;
 371                 unsigned lo, mid, hi;
 372
 373                 pack = fill(1);
 374                 c = *pack;
 375                 use(1);
 376                 base_offset = c & 127;
 377                 while (c & 128) {
 378                         base_offset += 1;
 379                         if (!base_offset || MSB(base_offset, 7))
 380                                 die("offset value overflow for delta base object");
 381                         pack = fill(1);
 382                         c = *pack;
 383                         use(1);
 384                         base_offset = (base_offset << 7) + (c & 127);
 385                 }
 386                 base_offset = obj_list[nr].offset - base_offset;
 387                 if (base_offset <= 0 || base_offset >= obj_list[nr].offset)
 388                         die("offset value out of bound for delta base object");
 389
 390                 delta_data = get_data(delta_size);
 391                 if (dry_run || !delta_data) {
 392                         free(delta_data);
 393                         return;
 394                 }
 395                 lo = 0;
 396                 hi = nr;
 397                 while (lo < hi) {
 398                         mid = lo + (hi - lo) / 2;
 399                         if (base_offset < obj_list[mid].offset) {
 400                                 hi = mid;
 401                         } else if (base_offset > obj_list[mid].offset) {
 402                                 lo = mid + 1;
 403                         } else {
 404                                 oidcpy(&base_oid, &obj_list[mid].oid);
 405                                 base_found = !is_null_oid(&base_oid);
 406                                 break;
 407                         }
 408                 }
 409                 if (!base_found) {
 410                         /*
 411                          * The delta base object is itself a delta that
 412                          * has not been resolved yet.
 413                          */
 414                         oidclr(&obj_list[nr].oid);
 415                         add_delta_to_list(nr, &null_oid, base_offset, delta_data, delta_size);
 416                         return;
 417                 }
 418         }
 419
 420         if (resolve_against_held(nr, &base_oid, delta_data, delta_size))
 421                 return;
 422
 423         base = read_sha1_file(base_oid.hash, &type, &base_size);
 424         if (!base) {
 425                 error("failed to read delta-pack base object %s",
 426                       oid_to_hex(&base_oid));
 427                 if (!recover)
 428                         exit(1);
 429                 has_errors = 1;
 430                 return;
 431         }
 432         resolve_delta(nr, type, base, base_size, delta_data, delta_size);
 433         free(base);
 434 }
 435
 436 static void unpack_one(unsigned nr)
 437 {
 438         unsigned shift;
 439         unsigned char *pack;
 440         unsigned long size, c;
 441         enum object_type type;
 442
 443         obj_list[nr].offset = consumed_bytes;
 444
 445         pack = fill(1);
 446         c = *pack;
 447         use(1);
 448         type = (c >> 4) & 7;
 449         size = (c & 15);
 450         shift = 4;
 451         while (c & 0x80) {
 452                 pack = fill(1);
 453                 c = *pack;
 454                 use(1);
 455                 size += (c & 0x7f) << shift;
 456                 shift += 7;
 457         }
 458
 459         switch (type) {
 460         case OBJ_COMMIT:
 461         case OBJ_TREE:
 462         case OBJ_BLOB:
 463         case OBJ_TAG:
 464                 unpack_non_delta_entry(type, size, nr);
 465                 return;
 466         case OBJ_REF_DELTA:
 467         case OBJ_OFS_DELTA:
 468                 unpack_delta_entry(type, size, nr);
 469                 return;
 470         default:
 471                 error("bad object type %d", type);
 472                 has_errors = 1;
 473                 if (recover)
 474                         return;
 475                 exit(1);
 476         }
 477 }
 478
 479 static void unpack_all(void)
 480 {
 481         int i;
 482         struct progress *progress = NULL;
 483         struct pack_header *hdr = fill(sizeof(struct pack_header));
 484
 485         nr_objects = ntohl(hdr->hdr_entries);
 486
 487         if (ntohl(hdr->hdr_signature) != PACK_SIGNATURE)
 488                 die("bad pack file");
 489         if (!pack_version_ok(hdr->hdr_version))
 490                 die("unknown pack file version %"PRIu32,
 491                         ntohl(hdr->hdr_version));
 492         use(sizeof(struct pack_header));
 493
 494         if (!quiet)
 495                 progress = start_progress(_("Unpacking objects"), nr_objects);
 496         obj_list = xcalloc(nr_objects, sizeof(*obj_list));
 497         for (i = 0; i < nr_objects; i++) {
 498                 unpack_one(i);
 499                 display_progress(progress, i + 1);
 500         }
 501         stop_progress(&progress);
 502
 503         if (delta_list)
 504                 die("unresolved deltas left after unpacking");
 505 }
 506
 507 int cmd_unpack_objects(int argc, const char **argv, const char *prefix)
 508 {
 509         int i;
 510         struct object_id oid;
 511
 512         check_replace_refs = 0;
 513
 514         git_config(git_default_config, NULL);
 515
 516         quiet = !isatty(2);
 517
 518         for (i = 1 ; i < argc; i++) {
 519                 const char *arg = argv[i];
 520
 521                 if (*arg == '-') {
 522                         if (!strcmp(arg, "-n")) {
 523                                 dry_run = 1;
 524                                 continue;
 525                         }
 526                         if (!strcmp(arg, "-q")) {
 527                                 quiet = 1;
 528                                 continue;
 529                         }
 530                         if (!strcmp(arg, "-r")) {
 531                                 recover = 1;
 532                                 continue;
 533                         }
 534                         if (!strcmp(arg, "--strict")) {
 535                                 strict = 1;
 536                                 continue;
 537                         }
 538                         if (skip_prefix(arg, "--strict=", &arg)) {
 539                                 strict = 1;
 540                                 fsck_set_msg_types(&fsck_options, arg);
 541                                 continue;
 542                         }
 543                         if (starts_with(arg, "--pack_header=")) {
 544                                 struct pack_header *hdr;
 545                                 char *c;
 546
 547                                 hdr = (struct pack_header *)buffer;
 548                                 hdr->hdr_signature = htonl(PACK_SIGNATURE);
 549                                 hdr->hdr_version = htonl(strtoul(arg + 14, &c, 10));
 550                                 if (*c != ',')
 551                                         die("bad %s", arg);
 552                                 hdr->hdr_entries = htonl(strtoul(c + 1, &c, 10));
 553                                 if (*c)
 554                                         die("bad %s", arg);
 555                                 len = sizeof(*hdr);
 556                                 continue;
 557                         }
 558                         if (skip_prefix(arg, "--max-input-size=", &arg)) {
 559                                 max_input_size = strtoumax(arg, NULL, 10);
 560                                 continue;
 561                         }
 562                         usage(unpack_usage);
 563                 }
 564
 565                 /* We don't take any non-flag arguments now.. Maybe some day */
 566                 usage(unpack_usage);
 567         }
 568         git_SHA1_Init(&ctx);
 569         unpack_all();
 570         git_SHA1_Update(&ctx, buffer, offset);
 571         git_SHA1_Final(oid.hash, &ctx);
 572         if (strict)
 573                 write_rest();
 574         if (hashcmp(fill(GIT_SHA1_RAWSZ), oid.hash))
 575                 die("final sha1 did not match");
 576         use(GIT_SHA1_RAWSZ);
 577
 578         /* Write the last part of the buffer to stdout */
 579         while (len) {
 580                 int ret = xwrite(1, buffer + offset, len);
 581                 if (ret <= 0)
 582                         break;
 583                 len -= ret;
 584                 offset += ret;
 585         }
 586
 587         /* All done */
 588         return has_errors;
 589 }