sha1_file.c

   1 /*
   2  * GIT - The information manager from hell
   3  *
   4  * Copyright (C) Linus Torvalds, 2005
   5  *
   6  * This handles basic git sha1 object files - packing, unpacking,
   7  * creation etc.
   8  */
   9 #include <sys/types.h>
  10 #include <dirent.h>
  11 #include "cache.h"
  12 #include "delta.h"
  13 #include "pack.h"
  14
  15 #ifndef O_NOATIME
  16 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  17 #define O_NOATIME 01000000
  18 #else
  19 #define O_NOATIME 0
  20 #endif
  21 #endif
  22
  23 static unsigned int sha1_file_open_flag = O_NOATIME;
  24
  25 static unsigned hexval(char c)
  26 {
  27         if (c >= '0' && c <= '9')
  28                 return c - '0';
  29         if (c >= 'a' && c <= 'f')
  30                 return c - 'a' + 10;
  31         if (c >= 'A' && c <= 'F')
  32                 return c - 'A' + 10;
  33         return ~0;
  34 }
  35
  36 int get_sha1_hex(const char *hex, unsigned char *sha1)
  37 {
  38         int i;
  39         for (i = 0; i < 20; i++) {
  40                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  41                 if (val & ~0xff)
  42                         return -1;
  43                 *sha1++ = val;
  44                 hex += 2;
  45         }
  46         return 0;
  47 }
  48
  49 static int get_sha1_file(const char *path, unsigned char *result)
  50 {
  51         char buffer[60];
  52         int fd = open(path, O_RDONLY);
  53         int len;
  54
  55         if (fd < 0)
  56                 return -1;
  57         len = read(fd, buffer, sizeof(buffer));
  58         close(fd);
  59         if (len < 40)
  60                 return -1;
  61         return get_sha1_hex(buffer, result);
  62 }
  63
  64 static char *git_dir, *git_object_dir, *git_index_file, *git_refs_dir;
  65 static void setup_git_env(void)
  66 {
  67         git_dir = gitenv(GIT_DIR_ENVIRONMENT);
  68         if (!git_dir)
  69                 git_dir = DEFAULT_GIT_DIR_ENVIRONMENT;
  70         git_object_dir = gitenv(DB_ENVIRONMENT);
  71         if (!git_object_dir) {
  72                 git_object_dir = xmalloc(strlen(git_dir) + 9);
  73                 sprintf(git_object_dir, "%s/objects", git_dir);
  74         }
  75         git_refs_dir = xmalloc(strlen(git_dir) + 6);
  76         sprintf(git_refs_dir, "%s/refs", git_dir);
  77         git_index_file = gitenv(INDEX_ENVIRONMENT);
  78         if (!git_index_file) {
  79                 git_index_file = xmalloc(strlen(git_dir) + 7);
  80                 sprintf(git_index_file, "%s/index", git_dir);
  81         }
  82 }
  83
  84 char *get_object_directory(void)
  85 {
  86         if (!git_object_dir)
  87                 setup_git_env();
  88         return git_object_dir;
  89 }
  90
  91 char *get_refs_directory(void)
  92 {
  93         if (!git_refs_dir)
  94                 setup_git_env();
  95         return git_refs_dir;
  96 }
  97
  98 char *get_index_file(void)
  99 {
 100         if (!git_index_file)
 101                 setup_git_env();
 102         return git_index_file;
 103 }
 104
 105 char *git_path(const char *fmt, ...)
 106 {
 107         static char pathname[PATH_MAX], *ret;
 108         va_list args;
 109         int len;
 110
 111         if (!git_dir)
 112                 setup_git_env();
 113         len = strlen(git_dir);
 114         if (len > PATH_MAX-100)
 115                 return "pad-path";
 116         memcpy(pathname, git_dir, len);
 117         if (len && git_dir[len-1] != '/')
 118                 pathname[len++] = '/';
 119         va_start(args, fmt);
 120         vsnprintf(pathname + len, sizeof(pathname) - len, fmt, args);
 121         va_end(args);
 122         ret = pathname;
 123
 124         /* Clean it up */
 125         if (!memcmp(pathname, "./", 2)) {
 126                 ret += 2;
 127                 while (*ret == '/')
 128                         ret++;
 129         }
 130         return ret;
 131 }
 132
 133 int get_sha1(const char *str, unsigned char *sha1)
 134 {
 135         static const char *prefix[] = {
 136                 "",
 137                 "refs",
 138                 "refs/tags",
 139                 "refs/heads",
 140                 "refs/snap",
 141                 NULL
 142         };
 143         const char **p;
 144
 145         if (!get_sha1_hex(str, sha1))
 146                 return 0;
 147
 148         for (p = prefix; *p; p++) {
 149                 char * pathname = git_path("%s/%s", *p, str);
 150                 if (!get_sha1_file(pathname, sha1))
 151                         return 0;
 152         }
 153
 154         return -1;
 155 }
 156
 157 char * sha1_to_hex(const unsigned char *sha1)
 158 {
 159         static char buffer[50];
 160         static const char hex[] = "0123456789abcdef";
 161         char *buf = buffer;
 162         int i;
 163
 164         for (i = 0; i < 20; i++) {
 165                 unsigned int val = *sha1++;
 166                 *buf++ = hex[val >> 4];
 167                 *buf++ = hex[val & 0xf];
 168         }
 169         return buffer;
 170 }
 171
 172 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 173 {
 174         int i;
 175         for (i = 0; i < 20; i++) {
 176                 static char hex[] = "0123456789abcdef";
 177                 unsigned int val = sha1[i];
 178                 char *pos = pathbuf + i*2 + (i > 0);
 179                 *pos++ = hex[val >> 4];
 180                 *pos = hex[val & 0xf];
 181         }
 182 }
 183
 184 /*
 185  * NOTE! This returns a statically allocated buffer, so you have to be
 186  * careful about using it. Do a "strdup()" if you need to save the
 187  * filename.
 188  *
 189  * Also note that this returns the location for creating.  Reading
 190  * SHA1 file can happen from any alternate directory listed in the
 191  * DB_ENVIRONMENT environment variable if it is not found in
 192  * the primary object database.
 193  */
 194 char *sha1_file_name(const unsigned char *sha1)
 195 {
 196         static char *name, *base;
 197
 198         if (!base) {
 199                 const char *sha1_file_directory = get_object_directory();
 200                 int len = strlen(sha1_file_directory);
 201                 base = xmalloc(len + 60);
 202                 memcpy(base, sha1_file_directory, len);
 203                 memset(base+len, 0, 60);
 204                 base[len] = '/';
 205                 base[len+3] = '/';
 206                 name = base + len + 1;
 207         }
 208         fill_sha1_path(name, sha1);
 209         return base;
 210 }
 211
 212 struct alternate_object_database *alt_odb;
 213
 214 /*
 215  * Prepare alternate object database registry.
 216  * alt_odb points at an array of struct alternate_object_database.
 217  * This array is terminated with an element that has both its base
 218  * and name set to NULL.  alt_odb[n] comes from n'th non-empty
 219  * element from colon separated ALTERNATE_DB_ENVIRONMENT environment
 220  * variable, and its base points at a statically allocated buffer
 221  * that contains "/the/directory/corresponding/to/.git/objects/...",
 222  * while its name points just after the slash at the end of
 223  * ".git/objects/" in the example above, and has enough space to hold
 224  * 40-byte hex SHA1, an extra slash for the first level indirection,
 225  * and the terminating NUL.
 226  * This function allocates the alt_odb array and all the strings
 227  * pointed by base fields of the array elements with one xmalloc();
 228  * the string pool immediately follows the array.
 229  */
 230 void prepare_alt_odb(void)
 231 {
 232         int pass, totlen, i;
 233         const char *cp, *last;
 234         char *op = NULL;
 235         const char *alt = gitenv(ALTERNATE_DB_ENVIRONMENT) ? : "";
 236
 237         if (alt_odb)
 238                 return;
 239         /* The first pass counts how large an area to allocate to
 240          * hold the entire alt_odb structure, including array of
 241          * structs and path buffers for them.  The second pass fills
 242          * the structure and prepares the path buffers for use by
 243          * fill_sha1_path().
 244          */
 245         for (totlen = pass = 0; pass < 2; pass++) {
 246                 last = alt;
 247                 i = 0;
 248                 do {
 249                         cp = strchr(last, ':') ? : last + strlen(last);
 250                         if (last != cp) {
 251                                 /* 43 = 40-byte + 2 '/' + terminating NUL */
 252                                 int pfxlen = cp - last;
 253                                 int entlen = pfxlen + 43;
 254                                 if (pass == 0)
 255                                         totlen += entlen;
 256                                 else {
 257                                         alt_odb[i].base = op;
 258                                         alt_odb[i].name = op + pfxlen + 1;
 259                                         memcpy(op, last, pfxlen);
 260                                         op[pfxlen] = op[pfxlen + 3] = '/';
 261                                         op[entlen-1] = 0;
 262                                         op += entlen;
 263                                 }
 264                                 i++;
 265                         }
 266                         while (*cp && *cp == ':')
 267                                 cp++;
 268                         last = cp;
 269                 } while (*cp);
 270                 if (pass)
 271                         break;
 272                 alt_odb = xmalloc(sizeof(*alt_odb) * (i + 1) + totlen);
 273                 alt_odb[i].base = alt_odb[i].name = NULL;
 274                 op = (char*)(&alt_odb[i+1]);
 275         }
 276 }
 277
 278 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 279 {
 280         int i;
 281         char *name = sha1_file_name(sha1);
 282
 283         if (!stat(name, st))
 284                 return name;
 285         prepare_alt_odb();
 286         for (i = 0; (name = alt_odb[i].name) != NULL; i++) {
 287                 fill_sha1_path(name, sha1);
 288                 if (!stat(alt_odb[i].base, st))
 289                         return alt_odb[i].base;
 290         }
 291         return NULL;
 292 }
 293
 294 #define PACK_MAX_SZ (1<<26)
 295 static int pack_used_ctr;
 296 static unsigned long pack_mapped;
 297 struct packed_git *packed_git;
 298
 299 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 300                                 void **idx_map_)
 301 {
 302         void *idx_map;
 303         unsigned int *index;
 304         unsigned long idx_size;
 305         int nr, i;
 306         int fd = open(path, O_RDONLY);
 307         struct stat st;
 308         if (fd < 0)
 309                 return -1;
 310         if (fstat(fd, &st)) {
 311                 close(fd);
 312                 return -1;
 313         }
 314         idx_size = st.st_size;
 315         idx_map = mmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 316         close(fd);
 317         if (idx_map == MAP_FAILED)
 318                 return -1;
 319
 320         index = idx_map;
 321         *idx_map_ = idx_map;
 322         *idx_size_ = idx_size;
 323
 324         /* check index map */
 325         if (idx_size < 4*256 + 20 + 20)
 326                 return error("index file too small");
 327         nr = 0;
 328         for (i = 0; i < 256; i++) {
 329                 unsigned int n = ntohl(index[i]);
 330                 if (n < nr)
 331                         return error("non-monotonic index");
 332                 nr = n;
 333         }
 334
 335         /*
 336          * Total size:
 337          *  - 256 index entries 4 bytes each
 338          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 339          *  - 20-byte SHA1 of the packfile
 340          *  - 20-byte SHA1 file checksum
 341          */
 342         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 343                 return error("wrong index file size");
 344
 345         return 0;
 346 }
 347
 348 static int unuse_one_packed_git(void)
 349 {
 350         struct packed_git *p, *lru = NULL;
 351
 352         for (p = packed_git; p; p = p->next) {
 353                 if (p->pack_use_cnt || !p->pack_base)
 354                         continue;
 355                 if (!lru || p->pack_last_used < lru->pack_last_used)
 356                         lru = p;
 357         }
 358         if (!lru)
 359                 return 0;
 360         munmap(lru->pack_base, lru->pack_size);
 361         lru->pack_base = NULL;
 362         return 1;
 363 }
 364
 365 void unuse_packed_git(struct packed_git *p)
 366 {
 367         p->pack_use_cnt--;
 368 }
 369
 370 int use_packed_git(struct packed_git *p)
 371 {
 372         if (!p->pack_base) {
 373                 int fd;
 374                 struct stat st;
 375                 void *map;
 376
 377                 pack_mapped += p->pack_size;
 378                 while (PACK_MAX_SZ < pack_mapped && unuse_one_packed_git())
 379                         ; /* nothing */
 380                 fd = open(p->pack_name, O_RDONLY);
 381                 if (fd < 0)
 382                         die("packfile %s cannot be opened", p->pack_name);
 383                 if (fstat(fd, &st)) {
 384                         close(fd);
 385                         die("packfile %s cannot be opened", p->pack_name);
 386                 }
 387                 if (st.st_size != p->pack_size)
 388                         die("packfile %s size mismatch.", p->pack_name);
 389                 map = mmap(NULL, p->pack_size, PROT_READ, MAP_PRIVATE, fd, 0);
 390                 close(fd);
 391                 if (map == MAP_FAILED)
 392                         die("packfile %s cannot be mapped.", p->pack_name);
 393                 p->pack_base = map;
 394
 395                 /* Check if the pack file matches with the index file.
 396                  * this is cheap.
 397                  */
 398                 if (memcmp((char*)(p->index_base) + p->index_size - 40,
 399                            p->pack_base + p->pack_size - 20, 20))
 400                         die("packfile %s does not match index.", p->pack_name);
 401         }
 402         p->pack_last_used = pack_used_ctr++;
 403         p->pack_use_cnt++;
 404         return 0;
 405 }
 406
 407 struct packed_git *add_packed_git(char *path, int path_len)
 408 {
 409         struct stat st;
 410         struct packed_git *p;
 411         unsigned long idx_size;
 412         void *idx_map;
 413
 414         if (check_packed_git_idx(path, &idx_size, &idx_map))
 415                 return NULL;
 416
 417         /* do we have a corresponding .pack file? */
 418         strcpy(path + path_len - 4, ".pack");
 419         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 420                 munmap(idx_map, idx_size);
 421                 return NULL;
 422         }
 423         /* ok, it looks sane as far as we can check without
 424          * actually mapping the pack file.
 425          */
 426         p = xmalloc(sizeof(*p) + path_len + 2);
 427         strcpy(p->pack_name, path);
 428         p->index_size = idx_size;
 429         p->pack_size = st.st_size;
 430         p->index_base = idx_map;
 431         p->next = NULL;
 432         p->pack_base = NULL;
 433         p->pack_last_used = 0;
 434         p->pack_use_cnt = 0;
 435         return p;
 436 }
 437
 438 static void prepare_packed_git_one(char *objdir)
 439 {
 440         char path[PATH_MAX];
 441         int len;
 442         DIR *dir;
 443         struct dirent *de;
 444
 445         sprintf(path, "%s/pack", objdir);
 446         len = strlen(path);
 447         dir = opendir(path);
 448         if (!dir)
 449                 return;
 450         path[len++] = '/';
 451         while ((de = readdir(dir)) != NULL) {
 452                 int namelen = strlen(de->d_name);
 453                 struct packed_git *p;
 454
 455                 if (strcmp(de->d_name + namelen - 4, ".idx"))
 456                         continue;
 457
 458                 /* we have .idx.  Is it a file we can map? */
 459                 strcpy(path + len, de->d_name);
 460                 p = add_packed_git(path, len + namelen);
 461                 if (!p)
 462                         continue;
 463                 p->next = packed_git;
 464                 packed_git = p;
 465         }
 466 }
 467
 468 void prepare_packed_git(void)
 469 {
 470         int i;
 471         static int run_once = 0;
 472
 473         if (run_once++)
 474                 return;
 475
 476         prepare_packed_git_one(get_object_directory());
 477         prepare_alt_odb();
 478         for (i = 0; alt_odb[i].base != NULL; i++) {
 479                 alt_odb[i].name[0] = 0;
 480                 prepare_packed_git_one(alt_odb[i].base);
 481         }
 482 }
 483
 484 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 485 {
 486         char header[100];
 487         unsigned char real_sha1[20];
 488         SHA_CTX c;
 489
 490         SHA1_Init(&c);
 491         SHA1_Update(&c, header, 1+sprintf(header, "%s %lu", type, size));
 492         SHA1_Update(&c, map, size);
 493         SHA1_Final(real_sha1, &c);
 494         return memcmp(sha1, real_sha1, 20) ? -1 : 0;
 495 }
 496
 497 static void *map_sha1_file_internal(const unsigned char *sha1,
 498                                     unsigned long *size,
 499                                     int say_error)
 500 {
 501         struct stat st;
 502         void *map;
 503         int fd;
 504         char *filename = find_sha1_file(sha1, &st);
 505
 506         if (!filename) {
 507                 if (say_error)
 508                         error("cannot map sha1 file %s", sha1_to_hex(sha1));
 509                 return NULL;
 510         }
 511
 512         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 513         if (fd < 0) {
 514                 /* See if it works without O_NOATIME */
 515                 switch (sha1_file_open_flag) {
 516                 default:
 517                         fd = open(filename, O_RDONLY);
 518                         if (fd >= 0)
 519                                 break;
 520                 /* Fallthrough */
 521                 case 0:
 522                         if (say_error)
 523                                 perror(filename);
 524                         return NULL;
 525                 }
 526
 527                 /* If it failed once, it will probably fail again.
 528                  * Stop using O_NOATIME
 529                  */
 530                 sha1_file_open_flag = 0;
 531         }
 532         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 533         close(fd);
 534         if (-1 == (int)(long)map)
 535                 return NULL;
 536         *size = st.st_size;
 537         return map;
 538 }
 539
 540 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 541 {
 542         return map_sha1_file_internal(sha1, size, 1);
 543 }
 544
 545 int unpack_sha1_header(z_stream *stream, void *map, unsigned long mapsize, void *buffer, unsigned long size)
 546 {
 547         /* Get the data stream */
 548         memset(stream, 0, sizeof(*stream));
 549         stream->next_in = map;
 550         stream->avail_in = mapsize;
 551         stream->next_out = buffer;
 552         stream->avail_out = size;
 553
 554         inflateInit(stream);
 555         return inflate(stream, 0);
 556 }
 557
 558 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 559 {
 560         int bytes = strlen(buffer) + 1;
 561         unsigned char *buf = xmalloc(1+size);
 562
 563         memcpy(buf, buffer + bytes, stream->total_out - bytes);
 564         bytes = stream->total_out - bytes;
 565         if (bytes < size) {
 566                 stream->next_out = buf + bytes;
 567                 stream->avail_out = size - bytes;
 568                 while (inflate(stream, Z_FINISH) == Z_OK)
 569                         /* nothing */;
 570         }
 571         buf[size] = 0;
 572         inflateEnd(stream);
 573         return buf;
 574 }
 575
 576 /*
 577  * We used to just use "sscanf()", but that's actually way
 578  * too permissive for what we want to check. So do an anal
 579  * object header parse by hand.
 580  */
 581 int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 582 {
 583         int i;
 584         unsigned long size;
 585
 586         /*
 587          * The type can be at most ten bytes (including the
 588          * terminating '\0' that we add), and is followed by
 589          * a space.
 590          */
 591         i = 10;
 592         for (;;) {
 593                 char c = *hdr++;
 594                 if (c == ' ')
 595                         break;
 596                 if (!--i)
 597                         return -1;
 598                 *type++ = c;
 599         }
 600         *type = 0;
 601
 602         /*
 603          * The length must follow immediately, and be in canonical
 604          * decimal format (ie "010" is not valid).
 605          */
 606         size = *hdr++ - '0';
 607         if (size > 9)
 608                 return -1;
 609         if (size) {
 610                 for (;;) {
 611                         unsigned long c = *hdr - '0';
 612                         if (c > 9)
 613                                 break;
 614                         hdr++;
 615                         size = size * 10 + c;
 616                 }
 617         }
 618         *sizep = size;
 619
 620         /*
 621          * The length must be followed by a zero byte
 622          */
 623         return *hdr ? -1 : 0;
 624 }
 625
 626 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 627 {
 628         int ret;
 629         z_stream stream;
 630         char hdr[8192];
 631
 632         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 633         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 634                 return NULL;
 635
 636         return unpack_sha1_rest(&stream, hdr, *size);
 637 }
 638
 639 /* forward declaration for a mutually recursive function */
 640 static int packed_object_info(struct pack_entry *entry,
 641                               char *type, unsigned long *sizep);
 642
 643 static int packed_delta_info(unsigned char *base_sha1,
 644                              unsigned long delta_size,
 645                              unsigned long left,
 646                              char *type,
 647                              unsigned long *sizep,
 648                              struct packed_git *p)
 649 {
 650         struct pack_entry base_ent;
 651
 652         if (left < 20)
 653                 die("truncated pack file");
 654
 655         /* The base entry _must_ be in the same pack */
 656         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 657                 die("failed to find delta-pack base object %s",
 658                     sha1_to_hex(base_sha1));
 659
 660         /* We choose to only get the type of the base object and
 661          * ignore potentially corrupt pack file that expects the delta
 662          * based on a base with a wrong size.  This saves tons of
 663          * inflate() calls.
 664          */
 665
 666         if (packed_object_info(&base_ent, type, NULL))
 667                 die("cannot get info for delta-pack base");
 668
 669         if (sizep) {
 670                 const unsigned char *data;
 671                 unsigned char delta_head[64];
 672                 unsigned long result_size;
 673                 z_stream stream;
 674                 int st;
 675
 676                 memset(&stream, 0, sizeof(stream));
 677
 678                 data = stream.next_in = base_sha1 + 20;
 679                 stream.avail_in = left - 20;
 680                 stream.next_out = delta_head;
 681                 stream.avail_out = sizeof(delta_head);
 682
 683                 inflateInit(&stream);
 684                 st = inflate(&stream, Z_FINISH);
 685                 inflateEnd(&stream);
 686                 if ((st != Z_STREAM_END) &&
 687                     stream.total_out != sizeof(delta_head))
 688                         die("delta data unpack-initial failed");
 689
 690                 /* Examine the initial part of the delta to figure out
 691                  * the result size.
 692                  */
 693                 data = delta_head;
 694                 get_delta_hdr_size(&data); /* ignore base size */
 695
 696                 /* Read the result size */
 697                 result_size = get_delta_hdr_size(&data);
 698                 *sizep = result_size;
 699         }
 700         return 0;
 701 }
 702
 703 static unsigned long unpack_object_header(struct packed_git *p, unsigned long offset,
 704         enum object_type *type, unsigned long *sizep)
 705 {
 706         unsigned shift;
 707         unsigned char *pack, c;
 708         unsigned long size;
 709
 710         if (offset >= p->pack_size)
 711                 die("object offset outside of pack file");
 712
 713         pack =  p->pack_base + offset;
 714         c = *pack++;
 715         offset++;
 716         *type = (c >> 4) & 7;
 717         size = c & 15;
 718         shift = 4;
 719         while (c & 0x80) {
 720                 if (offset >= p->pack_size)
 721                         die("object offset outside of pack file");
 722                 c = *pack++;
 723                 offset++;
 724                 size += (c & 0x7f) << shift;
 725                 shift += 7;
 726         }
 727         *sizep = size;
 728         return offset;
 729 }
 730
 731 void packed_object_info_detail(struct pack_entry *e,
 732                                char *type,
 733                                unsigned long *size,
 734                                unsigned long *store_size,
 735                                int *delta_chain_length,
 736                                unsigned char *base_sha1)
 737 {
 738         struct packed_git *p = e->p;
 739         unsigned long offset, left;
 740         unsigned char *pack;
 741         enum object_type kind;
 742
 743         offset = unpack_object_header(p, e->offset, &kind, size);
 744         pack = p->pack_base + offset;
 745         left = p->pack_size - offset;
 746         if (kind != OBJ_DELTA)
 747                 *delta_chain_length = 0;
 748         else {
 749                 int chain_length = 0;
 750                 memcpy(base_sha1, pack, 20);
 751                 do {
 752                         struct pack_entry base_ent;
 753                         unsigned long junk;
 754
 755                         find_pack_entry_one(pack, &base_ent, p);
 756                         offset = unpack_object_header(p, base_ent.offset,
 757                                                       &kind, &junk);
 758                         pack = p->pack_base + offset;
 759                         chain_length++;
 760                 } while (kind == OBJ_DELTA);
 761                 *delta_chain_length = chain_length;
 762         }
 763         switch (kind) {
 764         case OBJ_COMMIT:
 765                 strcpy(type, "commit");
 766                 break;
 767         case OBJ_TREE:
 768                 strcpy(type, "tree");
 769                 break;
 770         case OBJ_BLOB:
 771                 strcpy(type, "blob");
 772                 break;
 773         case OBJ_TAG:
 774                 strcpy(type, "tag");
 775                 break;
 776         default:
 777                 die("corrupted pack file");
 778         }
 779         *store_size = 0; /* notyet */
 780 }
 781
 782 static int packed_object_info(struct pack_entry *entry,
 783                               char *type, unsigned long *sizep)
 784 {
 785         struct packed_git *p = entry->p;
 786         unsigned long offset, size, left;
 787         unsigned char *pack;
 788         enum object_type kind;
 789         int retval;
 790
 791         if (use_packed_git(p))
 792                 die("cannot map packed file");
 793
 794         offset = unpack_object_header(p, entry->offset, &kind, &size);
 795         pack = p->pack_base + offset;
 796         left = p->pack_size - offset;
 797
 798         switch (kind) {
 799         case OBJ_DELTA:
 800                 retval = packed_delta_info(pack, size, left, type, sizep, p);
 801                 unuse_packed_git(p);
 802                 return retval;
 803         case OBJ_COMMIT:
 804                 strcpy(type, "commit");
 805                 break;
 806         case OBJ_TREE:
 807                 strcpy(type, "tree");
 808                 break;
 809         case OBJ_BLOB:
 810                 strcpy(type, "blob");
 811                 break;
 812         case OBJ_TAG:
 813                 strcpy(type, "tag");
 814                 break;
 815         default:
 816                 die("corrupted pack file");
 817         }
 818         if (sizep)
 819                 *sizep = size;
 820         unuse_packed_git(p);
 821         return 0;
 822 }
 823
 824 /* forward declaration for a mutually recursive function */
 825 static void *unpack_entry(struct pack_entry *, char *, unsigned long *);
 826
 827 static void *unpack_delta_entry(unsigned char *base_sha1,
 828                                 unsigned long delta_size,
 829                                 unsigned long left,
 830                                 char *type,
 831                                 unsigned long *sizep,
 832                                 struct packed_git *p)
 833 {
 834         struct pack_entry base_ent;
 835         void *data, *delta_data, *result, *base;
 836         unsigned long data_size, result_size, base_size;
 837         z_stream stream;
 838         int st;
 839
 840         if (left < 20)
 841                 die("truncated pack file");
 842         data = base_sha1 + 20;
 843         data_size = left - 20;
 844         delta_data = xmalloc(delta_size);
 845
 846         memset(&stream, 0, sizeof(stream));
 847
 848         stream.next_in = data;
 849         stream.avail_in = data_size;
 850         stream.next_out = delta_data;
 851         stream.avail_out = delta_size;
 852
 853         inflateInit(&stream);
 854         st = inflate(&stream, Z_FINISH);
 855         inflateEnd(&stream);
 856         if ((st != Z_STREAM_END) || stream.total_out != delta_size)
 857                 die("delta data unpack failed");
 858
 859         /* The base entry _must_ be in the same pack */
 860         if (!find_pack_entry_one(base_sha1, &base_ent, p))
 861                 die("failed to find delta-pack base object %s",
 862                     sha1_to_hex(base_sha1));
 863         base = unpack_entry_gently(&base_ent, type, &base_size);
 864         if (!base)
 865                 die("failed to read delta-pack base object %s",
 866                     sha1_to_hex(base_sha1));
 867         result = patch_delta(base, base_size,
 868                              delta_data, delta_size,
 869                              &result_size);
 870         if (!result)
 871                 die("failed to apply delta");
 872         free(delta_data);
 873         free(base);
 874         *sizep = result_size;
 875         return result;
 876 }
 877
 878 static void *unpack_non_delta_entry(unsigned char *data,
 879                                     unsigned long size,
 880                                     unsigned long left)
 881 {
 882         int st;
 883         z_stream stream;
 884         unsigned char *buffer;
 885
 886         buffer = xmalloc(size + 1);
 887         buffer[size] = 0;
 888         memset(&stream, 0, sizeof(stream));
 889         stream.next_in = data;
 890         stream.avail_in = left;
 891         stream.next_out = buffer;
 892         stream.avail_out = size;
 893
 894         inflateInit(&stream);
 895         st = inflate(&stream, Z_FINISH);
 896         inflateEnd(&stream);
 897         if ((st != Z_STREAM_END) || stream.total_out != size) {
 898                 free(buffer);
 899                 return NULL;
 900         }
 901
 902         return buffer;
 903 }
 904
 905 static void *unpack_entry(struct pack_entry *entry,
 906                           char *type, unsigned long *sizep)
 907 {
 908         struct packed_git *p = entry->p;
 909         void *retval;
 910
 911         if (use_packed_git(p))
 912                 die("cannot map packed file");
 913         retval = unpack_entry_gently(entry, type, sizep);
 914         unuse_packed_git(p);
 915         if (!retval)
 916                 die("corrupted pack file");
 917         return retval;
 918 }
 919
 920 /* The caller is responsible for use_packed_git()/unuse_packed_git() pair */
 921 void *unpack_entry_gently(struct pack_entry *entry,
 922                           char *type, unsigned long *sizep)
 923 {
 924         struct packed_git *p = entry->p;
 925         unsigned long offset, size, left;
 926         unsigned char *pack;
 927         enum object_type kind;
 928         void *retval;
 929
 930         offset = unpack_object_header(p, entry->offset, &kind, &size);
 931         pack = p->pack_base + offset;
 932         left = p->pack_size - offset;
 933         switch (kind) {
 934         case OBJ_DELTA:
 935                 retval = unpack_delta_entry(pack, size, left, type, sizep, p);
 936                 return retval;
 937         case OBJ_COMMIT:
 938                 strcpy(type, "commit");
 939                 break;
 940         case OBJ_TREE:
 941                 strcpy(type, "tree");
 942                 break;
 943         case OBJ_BLOB:
 944                 strcpy(type, "blob");
 945                 break;
 946         case OBJ_TAG:
 947                 strcpy(type, "tag");
 948                 break;
 949         default:
 950                 return NULL;
 951         }
 952         *sizep = size;
 953         retval = unpack_non_delta_entry(pack, size, left);
 954         return retval;
 955 }
 956
 957 int num_packed_objects(const struct packed_git *p)
 958 {
 959         /* See check_packed_git_idx() */
 960         return (p->index_size - 20 - 20 - 4*256) / 24;
 961 }
 962
 963 int nth_packed_object_sha1(const struct packed_git *p, int n,
 964                            unsigned char* sha1)
 965 {
 966         void *index = p->index_base + 256;
 967         if (n < 0 || num_packed_objects(p) <= n)
 968                 return -1;
 969         memcpy(sha1, (index + 24 * n + 4), 20);
 970         return 0;
 971 }
 972
 973 int find_pack_entry_one(const unsigned char *sha1,
 974                         struct pack_entry *e, struct packed_git *p)
 975 {
 976         unsigned int *level1_ofs = p->index_base;
 977         int hi = ntohl(level1_ofs[*sha1]);
 978         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
 979         void *index = p->index_base + 256;
 980
 981         do {
 982                 int mi = (lo + hi) / 2;
 983                 int cmp = memcmp(index + 24 * mi + 4, sha1, 20);
 984                 if (!cmp) {
 985                         e->offset = ntohl(*((int*)(index + 24 * mi)));
 986                         memcpy(e->sha1, sha1, 20);
 987                         e->p = p;
 988                         return 1;
 989                 }
 990                 if (cmp > 0)
 991                         hi = mi;
 992                 else
 993                         lo = mi+1;
 994         } while (lo < hi);
 995         return 0;
 996 }
 997
 998 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e)
 999 {
1000         struct packed_git *p;
1001         prepare_packed_git();
1002
1003         for (p = packed_git; p; p = p->next) {
1004                 if (find_pack_entry_one(sha1, e, p))
1005                         return 1;
1006         }
1007         return 0;
1008 }
1009
1010 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1011 {
1012         int status;
1013         unsigned long mapsize, size;
1014         void *map;
1015         z_stream stream;
1016         char hdr[128];
1017
1018         map = map_sha1_file_internal(sha1, &mapsize, 0);
1019         if (!map) {
1020                 struct pack_entry e;
1021
1022                 if (!find_pack_entry(sha1, &e))
1023                         return error("unable to find %s", sha1_to_hex(sha1));
1024                 return packed_object_info(&e, type, sizep);
1025         }
1026         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1027                 status = error("unable to unpack %s header",
1028                                sha1_to_hex(sha1));
1029         if (parse_sha1_header(hdr, type, &size) < 0)
1030                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1031         else {
1032                 status = 0;
1033                 if (sizep)
1034                         *sizep = size;
1035         }
1036         inflateEnd(&stream);
1037         munmap(map, mapsize);
1038         return status;
1039 }
1040
1041 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1042 {
1043         struct pack_entry e;
1044
1045         if (!find_pack_entry(sha1, &e)) {
1046                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1047                 return NULL;
1048         }
1049         return unpack_entry(&e, type, size);
1050 }
1051
1052 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1053 {
1054         unsigned long mapsize;
1055         void *map, *buf;
1056
1057         map = map_sha1_file_internal(sha1, &mapsize, 0);
1058         if (map) {
1059                 buf = unpack_sha1_file(map, mapsize, type, size);
1060                 munmap(map, mapsize);
1061                 return buf;
1062         }
1063         return read_packed_sha1(sha1, type, size);
1064 }
1065
1066 void *read_object_with_reference(const unsigned char *sha1,
1067                                  const char *required_type,
1068                                  unsigned long *size,
1069                                  unsigned char *actual_sha1_return)
1070 {
1071         char type[20];
1072         void *buffer;
1073         unsigned long isize;
1074         unsigned char actual_sha1[20];
1075
1076         memcpy(actual_sha1, sha1, 20);
1077         while (1) {
1078                 int ref_length = -1;
1079                 const char *ref_type = NULL;
1080
1081                 buffer = read_sha1_file(actual_sha1, type, &isize);
1082                 if (!buffer)
1083                         return NULL;
1084                 if (!strcmp(type, required_type)) {
1085                         *size = isize;
1086                         if (actual_sha1_return)
1087                                 memcpy(actual_sha1_return, actual_sha1, 20);
1088                         return buffer;
1089                 }
1090                 /* Handle references */
1091                 else if (!strcmp(type, "commit"))
1092                         ref_type = "tree ";
1093                 else if (!strcmp(type, "tag"))
1094                         ref_type = "object ";
1095                 else {
1096                         free(buffer);
1097                         return NULL;
1098                 }
1099                 ref_length = strlen(ref_type);
1100
1101                 if (memcmp(buffer, ref_type, ref_length) ||
1102                     get_sha1_hex(buffer + ref_length, actual_sha1)) {
1103                         free(buffer);
1104                         return NULL;
1105                 }
1106                 /* Now we have the ID of the referred-to object in
1107                  * actual_sha1.  Check again. */
1108         }
1109 }
1110
1111 static char *write_sha1_file_prepare(void *buf,
1112                                      unsigned long len,
1113                                      const char *type,
1114                                      unsigned char *sha1,
1115                                      unsigned char *hdr,
1116                                      int *hdrlen)
1117 {
1118         SHA_CTX c;
1119
1120         /* Generate the header */
1121         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1122
1123         /* Sha1.. */
1124         SHA1_Init(&c);
1125         SHA1_Update(&c, hdr, *hdrlen);
1126         SHA1_Update(&c, buf, len);
1127         SHA1_Final(sha1, &c);
1128
1129         return sha1_file_name(sha1);
1130 }
1131
1132 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1133 {
1134         int size;
1135         unsigned char *compressed;
1136         z_stream stream;
1137         unsigned char sha1[20];
1138         char *filename;
1139         static char tmpfile[PATH_MAX];
1140         unsigned char hdr[50];
1141         int fd, hdrlen, ret;
1142
1143         /* Normally if we have it in the pack then we do not bother writing
1144          * it out into .git/objects/??/?{38} file.
1145          */
1146         filename = write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1147         if (returnsha1)
1148                 memcpy(returnsha1, sha1, 20);
1149         if (has_sha1_file(sha1))
1150                 return 0;
1151         fd = open(filename, O_RDONLY);
1152         if (fd >= 0) {
1153                 /*
1154                  * FIXME!!! We might do collision checking here, but we'd
1155                  * need to uncompress the old file and check it. Later.
1156                  */
1157                 close(fd);
1158                 return 0;
1159         }
1160
1161         if (errno != ENOENT) {
1162                 fprintf(stderr, "sha1 file %s: %s", filename, strerror(errno));
1163                 return -1;
1164         }
1165
1166         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1167
1168         fd = mkstemp(tmpfile);
1169         if (fd < 0) {
1170                 fprintf(stderr, "unable to create temporary sha1 filename %s: %s", tmpfile, strerror(errno));
1171                 return -1;
1172         }
1173
1174         /* Set it up */
1175         memset(&stream, 0, sizeof(stream));
1176         deflateInit(&stream, Z_BEST_COMPRESSION);
1177         size = deflateBound(&stream, len+hdrlen);
1178         compressed = xmalloc(size);
1179
1180         /* Compress it */
1181         stream.next_out = compressed;
1182         stream.avail_out = size;
1183
1184         /* First header.. */
1185         stream.next_in = hdr;
1186         stream.avail_in = hdrlen;
1187         while (deflate(&stream, 0) == Z_OK)
1188                 /* nothing */;
1189
1190         /* Then the data itself.. */
1191         stream.next_in = buf;
1192         stream.avail_in = len;
1193         while (deflate(&stream, Z_FINISH) == Z_OK)
1194                 /* nothing */;
1195         deflateEnd(&stream);
1196         size = stream.total_out;
1197
1198         if (write(fd, compressed, size) != size)
1199                 die("unable to write file");
1200         fchmod(fd, 0444);
1201         close(fd);
1202         free(compressed);
1203
1204         ret = link(tmpfile, filename);
1205         if (ret < 0) {
1206                 ret = errno;
1207
1208                 /*
1209                  * Coda hack - coda doesn't like cross-directory links,
1210                  * so we fall back to a rename, which will mean that it
1211                  * won't be able to check collisions, but that's not a
1212                  * big deal.
1213                  *
1214                  * When this succeeds, we just return 0. We have nothing
1215                  * left to unlink.
1216                  */
1217                 if (ret == EXDEV && !rename(tmpfile, filename))
1218                         return 0;
1219         }
1220         unlink(tmpfile);
1221         if (ret) {
1222                 if (ret != EEXIST) {
1223                         fprintf(stderr, "unable to write sha1 filename %s: %s", filename, strerror(ret));
1224                         return -1;
1225                 }
1226                 /* FIXME!!! Collision check here ? */
1227         }
1228
1229         return 0;
1230 }
1231
1232 int write_sha1_from_fd(const unsigned char *sha1, int fd)
1233 {
1234         char *filename = sha1_file_name(sha1);
1235
1236         int local;
1237         z_stream stream;
1238         unsigned char real_sha1[20];
1239         unsigned char buf[4096];
1240         unsigned char discard[4096];
1241         int ret;
1242         SHA_CTX c;
1243
1244         local = open(filename, O_WRONLY | O_CREAT | O_EXCL, 0666);
1245
1246         if (local < 0)
1247                 return error("Couldn't open %s\n", filename);
1248
1249         memset(&stream, 0, sizeof(stream));
1250
1251         inflateInit(&stream);
1252
1253         SHA1_Init(&c);
1254
1255         do {
1256                 ssize_t size;
1257                 size = read(fd, buf, 4096);
1258                 if (size <= 0) {
1259                         close(local);
1260                         unlink(filename);
1261                         if (!size)
1262                                 return error("Connection closed?");
1263                         perror("Reading from connection");
1264                         return -1;
1265                 }
1266                 write(local, buf, size);
1267                 stream.avail_in = size;
1268                 stream.next_in = buf;
1269                 do {
1270                         stream.next_out = discard;
1271                         stream.avail_out = sizeof(discard);
1272                         ret = inflate(&stream, Z_SYNC_FLUSH);
1273                         SHA1_Update(&c, discard, sizeof(discard) -
1274                                     stream.avail_out);
1275                 } while (stream.avail_in && ret == Z_OK);
1276
1277         } while (ret == Z_OK);
1278         inflateEnd(&stream);
1279
1280         close(local);
1281         SHA1_Final(real_sha1, &c);
1282         if (ret != Z_STREAM_END) {
1283                 unlink(filename);
1284                 return error("File %s corrupted", sha1_to_hex(sha1));
1285         }
1286         if (memcmp(sha1, real_sha1, 20)) {
1287                 unlink(filename);
1288                 return error("File %s has bad hash\n", sha1_to_hex(sha1));
1289         }
1290
1291         return 0;
1292 }
1293
1294 int has_sha1_pack(const unsigned char *sha1)
1295 {
1296         struct pack_entry e;
1297         return find_pack_entry(sha1, &e);
1298 }
1299
1300 int has_sha1_file(const unsigned char *sha1)
1301 {
1302         struct stat st;
1303         struct pack_entry e;
1304
1305         if (find_sha1_file(sha1, &st))
1306                 return 1;
1307         return find_pack_entry(sha1, &e);
1308 }
1309
1310 int index_fd(unsigned char *sha1, int fd, struct stat *st)
1311 {
1312         unsigned long size = st->st_size;
1313         void *buf;
1314         int ret;
1315
1316         buf = "";
1317         if (size)
1318                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1319         close(fd);
1320         if ((int)(long)buf == -1)
1321                 return -1;
1322
1323         ret = write_sha1_file(buf, size, "blob", sha1);
1324         if (size)
1325                 munmap(buf, size);
1326         return ret;
1327 }