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 "cache.h"
  10 #include "delta.h"
  11 #include "pack.h"
  12 #include "blob.h"
  13 #include "commit.h"
  14 #include "tag.h"
  15 #include "tree.h"
  16
  17 #ifndef O_NOATIME
  18 #if defined(__linux__) && (defined(__i386__) || defined(__PPC__))
  19 #define O_NOATIME 01000000
  20 #else
  21 #define O_NOATIME 0
  22 #endif
  23 #endif
  24
  25 #ifdef NO_C99_FORMAT
  26 #define SZ_FMT "lu"
  27 #else
  28 #define SZ_FMT "zu"
  29 #endif
  30
  31 const unsigned char null_sha1[20];
  32
  33 static unsigned int sha1_file_open_flag = O_NOATIME;
  34
  35 signed char hexval_table[256] = {
  36          -1, -1, -1, -1, -1, -1, -1, -1,                /* 00-07 */
  37          -1, -1, -1, -1, -1, -1, -1, -1,                /* 08-0f */
  38          -1, -1, -1, -1, -1, -1, -1, -1,                /* 10-17 */
  39          -1, -1, -1, -1, -1, -1, -1, -1,                /* 18-1f */
  40          -1, -1, -1, -1, -1, -1, -1, -1,                /* 20-27 */
  41          -1, -1, -1, -1, -1, -1, -1, -1,                /* 28-2f */
  42           0,  1,  2,  3,  4,  5,  6,  7,                /* 30-37 */
  43           8,  9, -1, -1, -1, -1, -1, -1,                /* 38-3f */
  44          -1, 10, 11, 12, 13, 14, 15, -1,                /* 40-47 */
  45          -1, -1, -1, -1, -1, -1, -1, -1,                /* 48-4f */
  46          -1, -1, -1, -1, -1, -1, -1, -1,                /* 50-57 */
  47          -1, -1, -1, -1, -1, -1, -1, -1,                /* 58-5f */
  48          -1, 10, 11, 12, 13, 14, 15, -1,                /* 60-67 */
  49          -1, -1, -1, -1, -1, -1, -1, -1,                /* 68-67 */
  50          -1, -1, -1, -1, -1, -1, -1, -1,                /* 70-77 */
  51          -1, -1, -1, -1, -1, -1, -1, -1,                /* 78-7f */
  52          -1, -1, -1, -1, -1, -1, -1, -1,                /* 80-87 */
  53          -1, -1, -1, -1, -1, -1, -1, -1,                /* 88-8f */
  54          -1, -1, -1, -1, -1, -1, -1, -1,                /* 90-97 */
  55          -1, -1, -1, -1, -1, -1, -1, -1,                /* 98-9f */
  56          -1, -1, -1, -1, -1, -1, -1, -1,                /* a0-a7 */
  57          -1, -1, -1, -1, -1, -1, -1, -1,                /* a8-af */
  58          -1, -1, -1, -1, -1, -1, -1, -1,                /* b0-b7 */
  59          -1, -1, -1, -1, -1, -1, -1, -1,                /* b8-bf */
  60          -1, -1, -1, -1, -1, -1, -1, -1,                /* c0-c7 */
  61          -1, -1, -1, -1, -1, -1, -1, -1,                /* c8-cf */
  62          -1, -1, -1, -1, -1, -1, -1, -1,                /* d0-d7 */
  63          -1, -1, -1, -1, -1, -1, -1, -1,                /* d8-df */
  64          -1, -1, -1, -1, -1, -1, -1, -1,                /* e0-e7 */
  65          -1, -1, -1, -1, -1, -1, -1, -1,                /* e8-ef */
  66          -1, -1, -1, -1, -1, -1, -1, -1,                /* f0-f7 */
  67          -1, -1, -1, -1, -1, -1, -1, -1,                /* f8-ff */
  68 };
  69
  70 int get_sha1_hex(const char *hex, unsigned char *sha1)
  71 {
  72         int i;
  73         for (i = 0; i < 20; i++) {
  74                 unsigned int val = (hexval(hex[0]) << 4) | hexval(hex[1]);
  75                 if (val & ~0xff)
  76                         return -1;
  77                 *sha1++ = val;
  78                 hex += 2;
  79         }
  80         return 0;
  81 }
  82
  83 int safe_create_leading_directories(char *path)
  84 {
  85         char *pos = path;
  86         struct stat st;
  87
  88         if (*pos == '/')
  89                 pos++;
  90
  91         while (pos) {
  92                 pos = strchr(pos, '/');
  93                 if (!pos)
  94                         break;
  95                 *pos = 0;
  96                 if (!stat(path, &st)) {
  97                         /* path exists */
  98                         if (!S_ISDIR(st.st_mode)) {
  99                                 *pos = '/';
 100                                 return -3;
 101                         }
 102                 }
 103                 else if (mkdir(path, 0777)) {
 104                         *pos = '/';
 105                         return -1;
 106                 }
 107                 else if (adjust_shared_perm(path)) {
 108                         *pos = '/';
 109                         return -2;
 110                 }
 111                 *pos++ = '/';
 112         }
 113         return 0;
 114 }
 115
 116 char * sha1_to_hex(const unsigned char *sha1)
 117 {
 118         static int bufno;
 119         static char hexbuffer[4][50];
 120         static const char hex[] = "0123456789abcdef";
 121         char *buffer = hexbuffer[3 & ++bufno], *buf = buffer;
 122         int i;
 123
 124         for (i = 0; i < 20; i++) {
 125                 unsigned int val = *sha1++;
 126                 *buf++ = hex[val >> 4];
 127                 *buf++ = hex[val & 0xf];
 128         }
 129         *buf = '\0';
 130
 131         return buffer;
 132 }
 133
 134 static void fill_sha1_path(char *pathbuf, const unsigned char *sha1)
 135 {
 136         int i;
 137         for (i = 0; i < 20; i++) {
 138                 static char hex[] = "0123456789abcdef";
 139                 unsigned int val = sha1[i];
 140                 char *pos = pathbuf + i*2 + (i > 0);
 141                 *pos++ = hex[val >> 4];
 142                 *pos = hex[val & 0xf];
 143         }
 144 }
 145
 146 /*
 147  * NOTE! This returns a statically allocated buffer, so you have to be
 148  * careful about using it. Do a "xstrdup()" if you need to save the
 149  * filename.
 150  *
 151  * Also note that this returns the location for creating.  Reading
 152  * SHA1 file can happen from any alternate directory listed in the
 153  * DB_ENVIRONMENT environment variable if it is not found in
 154  * the primary object database.
 155  */
 156 char *sha1_file_name(const unsigned char *sha1)
 157 {
 158         static char *name, *base;
 159
 160         if (!base) {
 161                 const char *sha1_file_directory = get_object_directory();
 162                 int len = strlen(sha1_file_directory);
 163                 base = xmalloc(len + 60);
 164                 memcpy(base, sha1_file_directory, len);
 165                 memset(base+len, 0, 60);
 166                 base[len] = '/';
 167                 base[len+3] = '/';
 168                 name = base + len + 1;
 169         }
 170         fill_sha1_path(name, sha1);
 171         return base;
 172 }
 173
 174 char *sha1_pack_name(const unsigned char *sha1)
 175 {
 176         static const char hex[] = "0123456789abcdef";
 177         static char *name, *base, *buf;
 178         int i;
 179
 180         if (!base) {
 181                 const char *sha1_file_directory = get_object_directory();
 182                 int len = strlen(sha1_file_directory);
 183                 base = xmalloc(len + 60);
 184                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.pack", sha1_file_directory);
 185                 name = base + len + 11;
 186         }
 187
 188         buf = name;
 189
 190         for (i = 0; i < 20; i++) {
 191                 unsigned int val = *sha1++;
 192                 *buf++ = hex[val >> 4];
 193                 *buf++ = hex[val & 0xf];
 194         }
 195
 196         return base;
 197 }
 198
 199 char *sha1_pack_index_name(const unsigned char *sha1)
 200 {
 201         static const char hex[] = "0123456789abcdef";
 202         static char *name, *base, *buf;
 203         int i;
 204
 205         if (!base) {
 206                 const char *sha1_file_directory = get_object_directory();
 207                 int len = strlen(sha1_file_directory);
 208                 base = xmalloc(len + 60);
 209                 sprintf(base, "%s/pack/pack-1234567890123456789012345678901234567890.idx", sha1_file_directory);
 210                 name = base + len + 11;
 211         }
 212
 213         buf = name;
 214
 215         for (i = 0; i < 20; i++) {
 216                 unsigned int val = *sha1++;
 217                 *buf++ = hex[val >> 4];
 218                 *buf++ = hex[val & 0xf];
 219         }
 220
 221         return base;
 222 }
 223
 224 struct alternate_object_database *alt_odb_list;
 225 static struct alternate_object_database **alt_odb_tail;
 226
 227 static void read_info_alternates(const char * alternates, int depth);
 228
 229 /*
 230  * Prepare alternate object database registry.
 231  *
 232  * The variable alt_odb_list points at the list of struct
 233  * alternate_object_database.  The elements on this list come from
 234  * non-empty elements from colon separated ALTERNATE_DB_ENVIRONMENT
 235  * environment variable, and $GIT_OBJECT_DIRECTORY/info/alternates,
 236  * whose contents is similar to that environment variable but can be
 237  * LF separated.  Its base points at a statically allocated buffer that
 238  * contains "/the/directory/corresponding/to/.git/objects/...", while
 239  * its name points just after the slash at the end of ".git/objects/"
 240  * in the example above, and has enough space to hold 40-byte hex
 241  * SHA1, an extra slash for the first level indirection, and the
 242  * terminating NUL.
 243  */
 244 static int link_alt_odb_entry(const char * entry, int len, const char * relative_base, int depth)
 245 {
 246         struct stat st;
 247         const char *objdir = get_object_directory();
 248         struct alternate_object_database *ent;
 249         struct alternate_object_database *alt;
 250         /* 43 = 40-byte + 2 '/' + terminating NUL */
 251         int pfxlen = len;
 252         int entlen = pfxlen + 43;
 253         int base_len = -1;
 254
 255         if (*entry != '/' && relative_base) {
 256                 /* Relative alt-odb */
 257                 if (base_len < 0)
 258                         base_len = strlen(relative_base) + 1;
 259                 entlen += base_len;
 260                 pfxlen += base_len;
 261         }
 262         ent = xmalloc(sizeof(*ent) + entlen);
 263
 264         if (*entry != '/' && relative_base) {
 265                 memcpy(ent->base, relative_base, base_len - 1);
 266                 ent->base[base_len - 1] = '/';
 267                 memcpy(ent->base + base_len, entry, len);
 268         }
 269         else
 270                 memcpy(ent->base, entry, pfxlen);
 271
 272         ent->name = ent->base + pfxlen + 1;
 273         ent->base[pfxlen + 3] = '/';
 274         ent->base[pfxlen] = ent->base[entlen-1] = 0;
 275
 276         /* Detect cases where alternate disappeared */
 277         if (stat(ent->base, &st) || !S_ISDIR(st.st_mode)) {
 278                 error("object directory %s does not exist; "
 279                       "check .git/objects/info/alternates.",
 280                       ent->base);
 281                 free(ent);
 282                 return -1;
 283         }
 284
 285         /* Prevent the common mistake of listing the same
 286          * thing twice, or object directory itself.
 287          */
 288         for (alt = alt_odb_list; alt; alt = alt->next) {
 289                 if (!memcmp(ent->base, alt->base, pfxlen)) {
 290                         free(ent);
 291                         return -1;
 292                 }
 293         }
 294         if (!memcmp(ent->base, objdir, pfxlen)) {
 295                 free(ent);
 296                 return -1;
 297         }
 298
 299         /* add the alternate entry */
 300         *alt_odb_tail = ent;
 301         alt_odb_tail = &(ent->next);
 302         ent->next = NULL;
 303
 304         /* recursively add alternates */
 305         read_info_alternates(ent->base, depth + 1);
 306
 307         ent->base[pfxlen] = '/';
 308
 309         return 0;
 310 }
 311
 312 static void link_alt_odb_entries(const char *alt, const char *ep, int sep,
 313                                  const char *relative_base, int depth)
 314 {
 315         const char *cp, *last;
 316
 317         if (depth > 5) {
 318                 error("%s: ignoring alternate object stores, nesting too deep.",
 319                                 relative_base);
 320                 return;
 321         }
 322
 323         last = alt;
 324         while (last < ep) {
 325                 cp = last;
 326                 if (cp < ep && *cp == '#') {
 327                         while (cp < ep && *cp != sep)
 328                                 cp++;
 329                         last = cp + 1;
 330                         continue;
 331                 }
 332                 while (cp < ep && *cp != sep)
 333                         cp++;
 334                 if (last != cp) {
 335                         if ((*last != '/') && depth) {
 336                                 error("%s: ignoring relative alternate object store %s",
 337                                                 relative_base, last);
 338                         } else {
 339                                 link_alt_odb_entry(last, cp - last,
 340                                                 relative_base, depth);
 341                         }
 342                 }
 343                 while (cp < ep && *cp == sep)
 344                         cp++;
 345                 last = cp;
 346         }
 347 }
 348
 349 static void read_info_alternates(const char * relative_base, int depth)
 350 {
 351         char *map;
 352         struct stat st;
 353         char path[PATH_MAX];
 354         int fd;
 355
 356         sprintf(path, "%s/info/alternates", relative_base);
 357         fd = open(path, O_RDONLY);
 358         if (fd < 0)
 359                 return;
 360         if (fstat(fd, &st) || (st.st_size == 0)) {
 361                 close(fd);
 362                 return;
 363         }
 364         map = xmmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 365         close(fd);
 366
 367         link_alt_odb_entries(map, map + st.st_size, '\n', relative_base, depth);
 368
 369         munmap(map, st.st_size);
 370 }
 371
 372 void prepare_alt_odb(void)
 373 {
 374         const char *alt;
 375
 376         alt = getenv(ALTERNATE_DB_ENVIRONMENT);
 377         if (!alt) alt = "";
 378
 379         if (alt_odb_tail)
 380                 return;
 381         alt_odb_tail = &alt_odb_list;
 382         link_alt_odb_entries(alt, alt + strlen(alt), ':', NULL, 0);
 383
 384         read_info_alternates(get_object_directory(), 0);
 385 }
 386
 387 static char *find_sha1_file(const unsigned char *sha1, struct stat *st)
 388 {
 389         char *name = sha1_file_name(sha1);
 390         struct alternate_object_database *alt;
 391
 392         if (!stat(name, st))
 393                 return name;
 394         prepare_alt_odb();
 395         for (alt = alt_odb_list; alt; alt = alt->next) {
 396                 name = alt->name;
 397                 fill_sha1_path(name, sha1);
 398                 if (!stat(alt->base, st))
 399                         return alt->base;
 400         }
 401         return NULL;
 402 }
 403
 404 static unsigned int pack_used_ctr;
 405 static unsigned int pack_mmap_calls;
 406 static unsigned int peak_pack_open_windows;
 407 static unsigned int pack_open_windows;
 408 static size_t peak_pack_mapped;
 409 static size_t pack_mapped;
 410 static size_t page_size;
 411 struct packed_git *packed_git;
 412
 413 void pack_report()
 414 {
 415         fprintf(stderr,
 416                 "pack_report: getpagesize()            = %10" SZ_FMT "\n"
 417                 "pack_report: core.packedGitWindowSize = %10" SZ_FMT "\n"
 418                 "pack_report: core.packedGitLimit      = %10" SZ_FMT "\n",
 419                 page_size,
 420                 packed_git_window_size,
 421                 packed_git_limit);
 422         fprintf(stderr,
 423                 "pack_report: pack_used_ctr            = %10u\n"
 424                 "pack_report: pack_mmap_calls          = %10u\n"
 425                 "pack_report: pack_open_windows        = %10u / %10u\n"
 426                 "pack_report: pack_mapped              = "
 427                         "%10" SZ_FMT " / %10" SZ_FMT "\n",
 428                 pack_used_ctr,
 429                 pack_mmap_calls,
 430                 pack_open_windows, peak_pack_open_windows,
 431                 pack_mapped, peak_pack_mapped);
 432 }
 433
 434 static int check_packed_git_idx(const char *path, unsigned long *idx_size_,
 435                                 void **idx_map_)
 436 {
 437         void *idx_map;
 438         uint32_t *index;
 439         unsigned long idx_size;
 440         int nr, i;
 441         int fd = open(path, O_RDONLY);
 442         struct stat st;
 443         if (fd < 0)
 444                 return -1;
 445         if (fstat(fd, &st)) {
 446                 close(fd);
 447                 return -1;
 448         }
 449         idx_size = st.st_size;
 450         idx_map = xmmap(NULL, idx_size, PROT_READ, MAP_PRIVATE, fd, 0);
 451         close(fd);
 452
 453         index = idx_map;
 454         *idx_map_ = idx_map;
 455         *idx_size_ = idx_size;
 456
 457         /* check index map */
 458         if (idx_size < 4*256 + 20 + 20)
 459                 return error("index file %s is too small", path);
 460
 461         /* a future index format would start with this, as older git
 462          * binaries would fail the non-monotonic index check below.
 463          * give a nicer warning to the user if we can.
 464          */
 465         if (index[0] == htonl(PACK_IDX_SIGNATURE))
 466                 return error("index file %s is a newer version"
 467                         " and is not supported by this binary"
 468                         " (try upgrading GIT to a newer version)",
 469                         path);
 470
 471         nr = 0;
 472         for (i = 0; i < 256; i++) {
 473                 unsigned int n = ntohl(index[i]);
 474                 if (n < nr)
 475                         return error("non-monotonic index %s", path);
 476                 nr = n;
 477         }
 478
 479         /*
 480          * Total size:
 481          *  - 256 index entries 4 bytes each
 482          *  - 24-byte entries * nr (20-byte sha1 + 4-byte offset)
 483          *  - 20-byte SHA1 of the packfile
 484          *  - 20-byte SHA1 file checksum
 485          */
 486         if (idx_size != 4*256 + nr * 24 + 20 + 20)
 487                 return error("wrong index file size in %s", path);
 488
 489         return 0;
 490 }
 491
 492 static void scan_windows(struct packed_git *p,
 493         struct packed_git **lru_p,
 494         struct pack_window **lru_w,
 495         struct pack_window **lru_l)
 496 {
 497         struct pack_window *w, *w_l;
 498
 499         for (w_l = NULL, w = p->windows; w; w = w->next) {
 500                 if (!w->inuse_cnt) {
 501                         if (!*lru_w || w->last_used < (*lru_w)->last_used) {
 502                                 *lru_p = p;
 503                                 *lru_w = w;
 504                                 *lru_l = w_l;
 505                         }
 506                 }
 507                 w_l = w;
 508         }
 509 }
 510
 511 static int unuse_one_window(struct packed_git *current)
 512 {
 513         struct packed_git *p, *lru_p = NULL;
 514         struct pack_window *lru_w = NULL, *lru_l = NULL;
 515
 516         if (current)
 517                 scan_windows(current, &lru_p, &lru_w, &lru_l);
 518         for (p = packed_git; p; p = p->next)
 519                 scan_windows(p, &lru_p, &lru_w, &lru_l);
 520         if (lru_p) {
 521                 munmap(lru_w->base, lru_w->len);
 522                 pack_mapped -= lru_w->len;
 523                 if (lru_l)
 524                         lru_l->next = lru_w->next;
 525                 else {
 526                         lru_p->windows = lru_w->next;
 527                         if (!lru_p->windows && lru_p != current) {
 528                                 close(lru_p->pack_fd);
 529                                 lru_p->pack_fd = -1;
 530                         }
 531                 }
 532                 free(lru_w);
 533                 pack_open_windows--;
 534                 return 1;
 535         }
 536         return 0;
 537 }
 538
 539 void release_pack_memory(size_t need)
 540 {
 541         size_t cur = pack_mapped;
 542         while (need >= (cur - pack_mapped) && unuse_one_window(NULL))
 543                 ; /* nothing */
 544 }
 545
 546 void unuse_pack(struct pack_window **w_cursor)
 547 {
 548         struct pack_window *w = *w_cursor;
 549         if (w) {
 550                 w->inuse_cnt--;
 551                 *w_cursor = NULL;
 552         }
 553 }
 554
 555 static void open_packed_git(struct packed_git *p)
 556 {
 557         struct stat st;
 558         struct pack_header hdr;
 559         unsigned char sha1[20];
 560         unsigned char *idx_sha1;
 561         long fd_flag;
 562
 563         p->pack_fd = open(p->pack_name, O_RDONLY);
 564         if (p->pack_fd < 0 || fstat(p->pack_fd, &st))
 565                 die("packfile %s cannot be opened", p->pack_name);
 566
 567         /* If we created the struct before we had the pack we lack size. */
 568         if (!p->pack_size) {
 569                 if (!S_ISREG(st.st_mode))
 570                         die("packfile %s not a regular file", p->pack_name);
 571                 p->pack_size = st.st_size;
 572         } else if (p->pack_size != st.st_size)
 573                 die("packfile %s size changed", p->pack_name);
 574
 575         /* We leave these file descriptors open with sliding mmap;
 576          * there is no point keeping them open across exec(), though.
 577          */
 578         fd_flag = fcntl(p->pack_fd, F_GETFD, 0);
 579         if (fd_flag < 0)
 580                 die("cannot determine file descriptor flags");
 581         fd_flag |= FD_CLOEXEC;
 582         if (fcntl(p->pack_fd, F_SETFD, fd_flag) == -1)
 583                 die("cannot set FD_CLOEXEC");
 584
 585         /* Verify we recognize this pack file format. */
 586         if (read_in_full(p->pack_fd, &hdr, sizeof(hdr)) != sizeof(hdr))
 587                 die("file %s is far too short to be a packfile", p->pack_name);
 588         if (hdr.hdr_signature != htonl(PACK_SIGNATURE))
 589                 die("file %s is not a GIT packfile", p->pack_name);
 590         if (!pack_version_ok(hdr.hdr_version))
 591                 die("packfile %s is version %u and not supported"
 592                         " (try upgrading GIT to a newer version)",
 593                         p->pack_name, ntohl(hdr.hdr_version));
 594
 595         /* Verify the pack matches its index. */
 596         if (num_packed_objects(p) != ntohl(hdr.hdr_entries))
 597                 die("packfile %s claims to have %u objects"
 598                         " while index size indicates %u objects",
 599                         p->pack_name, ntohl(hdr.hdr_entries),
 600                         num_packed_objects(p));
 601         if (lseek(p->pack_fd, p->pack_size - sizeof(sha1), SEEK_SET) == -1)
 602                 die("end of packfile %s is unavailable", p->pack_name);
 603         if (read_in_full(p->pack_fd, sha1, sizeof(sha1)) != sizeof(sha1))
 604                 die("packfile %s signature is unavailable", p->pack_name);
 605         idx_sha1 = ((unsigned char *)p->index_base) + p->index_size - 40;
 606         if (hashcmp(sha1, idx_sha1))
 607                 die("packfile %s does not match index", p->pack_name);
 608 }
 609
 610 static int in_window(struct pack_window *win, unsigned long offset)
 611 {
 612         /* We must promise at least 20 bytes (one hash) after the
 613          * offset is available from this window, otherwise the offset
 614          * is not actually in this window and a different window (which
 615          * has that one hash excess) must be used.  This is to support
 616          * the object header and delta base parsing routines below.
 617          */
 618         off_t win_off = win->offset;
 619         return win_off <= offset
 620                 && (offset + 20) <= (win_off + win->len);
 621 }
 622
 623 unsigned char* use_pack(struct packed_git *p,
 624                 struct pack_window **w_cursor,
 625                 unsigned long offset,
 626                 unsigned int *left)
 627 {
 628         struct pack_window *win = *w_cursor;
 629
 630         if (p->pack_fd == -1)
 631                 open_packed_git(p);
 632
 633         /* Since packfiles end in a hash of their content and its
 634          * pointless to ask for an offset into the middle of that
 635          * hash, and the in_window function above wouldn't match
 636          * don't allow an offset too close to the end of the file.
 637          */
 638         if (offset > (p->pack_size - 20))
 639                 die("offset beyond end of packfile (truncated pack?)");
 640
 641         if (!win || !in_window(win, offset)) {
 642                 if (win)
 643                         win->inuse_cnt--;
 644                 for (win = p->windows; win; win = win->next) {
 645                         if (in_window(win, offset))
 646                                 break;
 647                 }
 648                 if (!win) {
 649                         if (!page_size)
 650                                 page_size = getpagesize();
 651                         win = xcalloc(1, sizeof(*win));
 652                         win->offset = (offset / page_size) * page_size;
 653                         win->len = p->pack_size - win->offset;
 654                         if (win->len > packed_git_window_size)
 655                                 win->len = packed_git_window_size;
 656                         pack_mapped += win->len;
 657                         while (packed_git_limit < pack_mapped
 658                                 && unuse_one_window(p))
 659                                 ; /* nothing */
 660                         win->base = xmmap(NULL, win->len,
 661                                 PROT_READ, MAP_PRIVATE,
 662                                 p->pack_fd, win->offset);
 663                         if (win->base == MAP_FAILED)
 664                                 die("packfile %s cannot be mapped: %s",
 665                                         p->pack_name,
 666                                         strerror(errno));
 667                         pack_mmap_calls++;
 668                         pack_open_windows++;
 669                         if (pack_mapped > peak_pack_mapped)
 670                                 peak_pack_mapped = pack_mapped;
 671                         if (pack_open_windows > peak_pack_open_windows)
 672                                 peak_pack_open_windows = pack_open_windows;
 673                         win->next = p->windows;
 674                         p->windows = win;
 675                 }
 676         }
 677         if (win != *w_cursor) {
 678                 win->last_used = pack_used_ctr++;
 679                 win->inuse_cnt++;
 680                 *w_cursor = win;
 681         }
 682         offset -= win->offset;
 683         if (left)
 684                 *left = win->len - offset;
 685         return win->base + offset;
 686 }
 687
 688 struct packed_git *add_packed_git(char *path, int path_len, int local)
 689 {
 690         struct stat st;
 691         struct packed_git *p;
 692         unsigned long idx_size;
 693         void *idx_map;
 694         unsigned char sha1[20];
 695
 696         if (check_packed_git_idx(path, &idx_size, &idx_map))
 697                 return NULL;
 698
 699         /* do we have a corresponding .pack file? */
 700         strcpy(path + path_len - 4, ".pack");
 701         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 702                 munmap(idx_map, idx_size);
 703                 return NULL;
 704         }
 705         /* ok, it looks sane as far as we can check without
 706          * actually mapping the pack file.
 707          */
 708         p = xmalloc(sizeof(*p) + path_len + 2);
 709         strcpy(p->pack_name, path);
 710         p->index_size = idx_size;
 711         p->pack_size = st.st_size;
 712         p->index_base = idx_map;
 713         p->next = NULL;
 714         p->windows = NULL;
 715         p->pack_fd = -1;
 716         p->pack_local = local;
 717         if ((path_len > 44) && !get_sha1_hex(path + path_len - 44, sha1))
 718                 hashcpy(p->sha1, sha1);
 719         return p;
 720 }
 721
 722 struct packed_git *parse_pack_index(unsigned char *sha1)
 723 {
 724         char *path = sha1_pack_index_name(sha1);
 725         return parse_pack_index_file(sha1, path);
 726 }
 727
 728 struct packed_git *parse_pack_index_file(const unsigned char *sha1, char *idx_path)
 729 {
 730         struct packed_git *p;
 731         unsigned long idx_size;
 732         void *idx_map;
 733         char *path;
 734
 735         if (check_packed_git_idx(idx_path, &idx_size, &idx_map))
 736                 return NULL;
 737
 738         path = sha1_pack_name(sha1);
 739
 740         p = xmalloc(sizeof(*p) + strlen(path) + 2);
 741         strcpy(p->pack_name, path);
 742         p->index_size = idx_size;
 743         p->pack_size = 0;
 744         p->index_base = idx_map;
 745         p->next = NULL;
 746         p->windows = NULL;
 747         p->pack_fd = -1;
 748         hashcpy(p->sha1, sha1);
 749         return p;
 750 }
 751
 752 void install_packed_git(struct packed_git *pack)
 753 {
 754         pack->next = packed_git;
 755         packed_git = pack;
 756 }
 757
 758 static void prepare_packed_git_one(char *objdir, int local)
 759 {
 760         char path[PATH_MAX];
 761         int len;
 762         DIR *dir;
 763         struct dirent *de;
 764
 765         sprintf(path, "%s/pack", objdir);
 766         len = strlen(path);
 767         dir = opendir(path);
 768         if (!dir) {
 769                 if (errno != ENOENT)
 770                         error("unable to open object pack directory: %s: %s",
 771                               path, strerror(errno));
 772                 return;
 773         }
 774         path[len++] = '/';
 775         while ((de = readdir(dir)) != NULL) {
 776                 int namelen = strlen(de->d_name);
 777                 struct packed_git *p;
 778
 779                 if (!has_extension(de->d_name, ".idx"))
 780                         continue;
 781
 782                 /* Don't reopen a pack we already have. */
 783                 strcpy(path + len, de->d_name);
 784                 for (p = packed_git; p; p = p->next) {
 785                         if (!memcmp(path, p->pack_name, len + namelen - 4))
 786                                 break;
 787                 }
 788                 if (p)
 789                         continue;
 790                 /* See if it really is a valid .idx file with corresponding
 791                  * .pack file that we can map.
 792                  */
 793                 p = add_packed_git(path, len + namelen, local);
 794                 if (!p)
 795                         continue;
 796                 install_packed_git(p);
 797         }
 798         closedir(dir);
 799 }
 800
 801 static int prepare_packed_git_run_once = 0;
 802 void prepare_packed_git(void)
 803 {
 804         struct alternate_object_database *alt;
 805
 806         if (prepare_packed_git_run_once)
 807                 return;
 808         prepare_packed_git_one(get_object_directory(), 1);
 809         prepare_alt_odb();
 810         for (alt = alt_odb_list; alt; alt = alt->next) {
 811                 alt->name[-1] = 0;
 812                 prepare_packed_git_one(alt->base, 0);
 813                 alt->name[-1] = '/';
 814         }
 815         prepare_packed_git_run_once = 1;
 816 }
 817
 818 void reprepare_packed_git(void)
 819 {
 820         prepare_packed_git_run_once = 0;
 821         prepare_packed_git();
 822 }
 823
 824 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 825 {
 826         unsigned char real_sha1[20];
 827         hash_sha1_file(map, size, type, real_sha1);
 828         return hashcmp(sha1, real_sha1) ? -1 : 0;
 829 }
 830
 831 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 832 {
 833         struct stat st;
 834         void *map;
 835         int fd;
 836         char *filename = find_sha1_file(sha1, &st);
 837
 838         if (!filename) {
 839                 return NULL;
 840         }
 841
 842         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 843         if (fd < 0) {
 844                 /* See if it works without O_NOATIME */
 845                 switch (sha1_file_open_flag) {
 846                 default:
 847                         fd = open(filename, O_RDONLY);
 848                         if (fd >= 0)
 849                                 break;
 850                 /* Fallthrough */
 851                 case 0:
 852                         return NULL;
 853                 }
 854
 855                 /* If it failed once, it will probably fail again.
 856                  * Stop using O_NOATIME
 857                  */
 858                 sha1_file_open_flag = 0;
 859         }
 860         map = xmmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 861         close(fd);
 862         *size = st.st_size;
 863         return map;
 864 }
 865
 866 int legacy_loose_object(unsigned char *map)
 867 {
 868         unsigned int word;
 869
 870         /*
 871          * Is it a zlib-compressed buffer? If so, the first byte
 872          * must be 0x78 (15-bit window size, deflated), and the
 873          * first 16-bit word is evenly divisible by 31
 874          */
 875         word = (map[0] << 8) + map[1];
 876         if (map[0] == 0x78 && !(word % 31))
 877                 return 1;
 878         else
 879                 return 0;
 880 }
 881
 882 unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep)
 883 {
 884         unsigned shift;
 885         unsigned char c;
 886         unsigned long size;
 887         unsigned long used = 0;
 888
 889         c = buf[used++];
 890         *type = (c >> 4) & 7;
 891         size = c & 15;
 892         shift = 4;
 893         while (c & 0x80) {
 894                 if (len <= used)
 895                         return 0;
 896                 if (sizeof(long) * 8 <= shift)
 897                         return 0;
 898                 c = buf[used++];
 899                 size += (c & 0x7f) << shift;
 900                 shift += 7;
 901         }
 902         *sizep = size;
 903         return used;
 904 }
 905
 906 static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
 907 {
 908         unsigned long size, used;
 909         static const char valid_loose_object_type[8] = {
 910                 0, /* OBJ_EXT */
 911                 1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
 912                 0, /* "delta" and others are invalid in a loose object */
 913         };
 914         enum object_type type;
 915
 916         /* Get the data stream */
 917         memset(stream, 0, sizeof(*stream));
 918         stream->next_in = map;
 919         stream->avail_in = mapsize;
 920         stream->next_out = buffer;
 921         stream->avail_out = bufsiz;
 922
 923         if (legacy_loose_object(map)) {
 924                 inflateInit(stream);
 925                 return inflate(stream, 0);
 926         }
 927
 928         used = unpack_object_header_gently(map, mapsize, &type, &size);
 929         if (!used || !valid_loose_object_type[type])
 930                 return -1;
 931         map += used;
 932         mapsize -= used;
 933
 934         /* Set up the stream for the rest.. */
 935         stream->next_in = map;
 936         stream->avail_in = mapsize;
 937         inflateInit(stream);
 938
 939         /* And generate the fake traditional header */
 940         stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
 941                                          type_names[type], size);
 942         return 0;
 943 }
 944
 945 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 946 {
 947         int bytes = strlen(buffer) + 1;
 948         unsigned char *buf = xmalloc(1+size);
 949         unsigned long n;
 950
 951         n = stream->total_out - bytes;
 952         if (n > size)
 953                 n = size;
 954         memcpy(buf, (char *) buffer + bytes, n);
 955         bytes = n;
 956         if (bytes < size) {
 957                 stream->next_out = buf + bytes;
 958                 stream->avail_out = size - bytes;
 959                 while (inflate(stream, Z_FINISH) == Z_OK)
 960                         /* nothing */;
 961         }
 962         buf[size] = 0;
 963         inflateEnd(stream);
 964         return buf;
 965 }
 966
 967 /*
 968  * We used to just use "sscanf()", but that's actually way
 969  * too permissive for what we want to check. So do an anal
 970  * object header parse by hand.
 971  */
 972 static int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 973 {
 974         int i;
 975         unsigned long size;
 976
 977         /*
 978          * The type can be at most ten bytes (including the
 979          * terminating '\0' that we add), and is followed by
 980          * a space.
 981          */
 982         i = 10;
 983         for (;;) {
 984                 char c = *hdr++;
 985                 if (c == ' ')
 986                         break;
 987                 if (!--i)
 988                         return -1;
 989                 *type++ = c;
 990         }
 991         *type = 0;
 992
 993         /*
 994          * The length must follow immediately, and be in canonical
 995          * decimal format (ie "010" is not valid).
 996          */
 997         size = *hdr++ - '0';
 998         if (size > 9)
 999                 return -1;
1000         if (size) {
1001                 for (;;) {
1002                         unsigned long c = *hdr - '0';
1003                         if (c > 9)
1004                                 break;
1005                         hdr++;
1006                         size = size * 10 + c;
1007                 }
1008         }
1009         *sizep = size;
1010
1011         /*
1012          * The length must be followed by a zero byte
1013          */
1014         return *hdr ? -1 : 0;
1015 }
1016
1017 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
1018 {
1019         int ret;
1020         z_stream stream;
1021         char hdr[8192];
1022
1023         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
1024         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
1025                 return NULL;
1026
1027         return unpack_sha1_rest(&stream, hdr, *size);
1028 }
1029
1030 static unsigned long get_delta_base(struct packed_git *p,
1031                                     struct pack_window **w_curs,
1032                                     unsigned long offset,
1033                                     enum object_type kind,
1034                                     unsigned long delta_obj_offset,
1035                                     unsigned long *base_obj_offset)
1036 {
1037         unsigned char *base_info = use_pack(p, w_curs, offset, NULL);
1038         unsigned long base_offset;
1039
1040         /* use_pack() assured us we have [base_info, base_info + 20)
1041          * as a range that we can look at without walking off the
1042          * end of the mapped window.  Its actually the hash size
1043          * that is assured.  An OFS_DELTA longer than the hash size
1044          * is stupid, as then a REF_DELTA would be smaller to store.
1045          */
1046         if (kind == OBJ_OFS_DELTA) {
1047                 unsigned used = 0;
1048                 unsigned char c = base_info[used++];
1049                 base_offset = c & 127;
1050                 while (c & 128) {
1051                         base_offset += 1;
1052                         if (!base_offset || base_offset & ~(~0UL >> 7))
1053                                 die("offset value overflow for delta base object");
1054                         c = base_info[used++];
1055                         base_offset = (base_offset << 7) + (c & 127);
1056                 }
1057                 base_offset = delta_obj_offset - base_offset;
1058                 if (base_offset >= delta_obj_offset)
1059                         die("delta base offset out of bound");
1060                 offset += used;
1061         } else if (kind == OBJ_REF_DELTA) {
1062                 /* The base entry _must_ be in the same pack */
1063                 base_offset = find_pack_entry_one(base_info, p);
1064                 if (!base_offset)
1065                         die("failed to find delta-pack base object %s",
1066                                 sha1_to_hex(base_info));
1067                 offset += 20;
1068         } else
1069                 die("I am totally screwed");
1070         *base_obj_offset = base_offset;
1071         return offset;
1072 }
1073
1074 /* forward declaration for a mutually recursive function */
1075 static int packed_object_info(struct packed_git *p, unsigned long offset,
1076                               char *type, unsigned long *sizep);
1077
1078 static int packed_delta_info(struct packed_git *p,
1079                              struct pack_window **w_curs,
1080                              unsigned long offset,
1081                              enum object_type kind,
1082                              unsigned long obj_offset,
1083                              char *type,
1084                              unsigned long *sizep)
1085 {
1086         unsigned long base_offset;
1087
1088         offset = get_delta_base(p, w_curs, offset, kind,
1089                 obj_offset, &base_offset);
1090
1091         /* We choose to only get the type of the base object and
1092          * ignore potentially corrupt pack file that expects the delta
1093          * based on a base with a wrong size.  This saves tons of
1094          * inflate() calls.
1095          */
1096         if (packed_object_info(p, base_offset, type, NULL))
1097                 die("cannot get info for delta-pack base");
1098
1099         if (sizep) {
1100                 const unsigned char *data;
1101                 unsigned char delta_head[20], *in;
1102                 unsigned long result_size;
1103                 z_stream stream;
1104                 int st;
1105
1106                 memset(&stream, 0, sizeof(stream));
1107                 stream.next_out = delta_head;
1108                 stream.avail_out = sizeof(delta_head);
1109
1110                 inflateInit(&stream);
1111                 do {
1112                         in = use_pack(p, w_curs, offset, &stream.avail_in);
1113                         stream.next_in = in;
1114                         st = inflate(&stream, Z_FINISH);
1115                         offset += stream.next_in - in;
1116                 } while ((st == Z_OK || st == Z_BUF_ERROR)
1117                         && stream.total_out < sizeof(delta_head));
1118                 inflateEnd(&stream);
1119                 if ((st != Z_STREAM_END) &&
1120                     stream.total_out != sizeof(delta_head))
1121                         die("delta data unpack-initial failed");
1122
1123                 /* Examine the initial part of the delta to figure out
1124                  * the result size.
1125                  */
1126                 data = delta_head;
1127
1128                 /* ignore base size */
1129                 get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
1130
1131                 /* Read the result size */
1132                 result_size = get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
1133                 *sizep = result_size;
1134         }
1135         return 0;
1136 }
1137
1138 static unsigned long unpack_object_header(struct packed_git *p,
1139                 struct pack_window **w_curs,
1140                 unsigned long offset,
1141                 enum object_type *type,
1142                 unsigned long *sizep)
1143 {
1144         unsigned char *base;
1145         unsigned int left;
1146         unsigned long used;
1147
1148         /* use_pack() assures us we have [base, base + 20) available
1149          * as a range that we can look at at.  (Its actually the hash
1150          * size that is assurred.)  With our object header encoding
1151          * the maximum deflated object size is 2^137, which is just
1152          * insane, so we know won't exceed what we have been given.
1153          */
1154         base = use_pack(p, w_curs, offset, &left);
1155         used = unpack_object_header_gently(base, left, type, sizep);
1156         if (!used)
1157                 die("object offset outside of pack file");
1158
1159         return offset + used;
1160 }
1161
1162 void packed_object_info_detail(struct packed_git *p,
1163                                unsigned long offset,
1164                                char *type,
1165                                unsigned long *size,
1166                                unsigned long *store_size,
1167                                unsigned int *delta_chain_length,
1168                                unsigned char *base_sha1)
1169 {
1170         struct pack_window *w_curs = NULL;
1171         unsigned long obj_offset, val;
1172         unsigned char *next_sha1;
1173         enum object_type kind;
1174
1175         *delta_chain_length = 0;
1176         obj_offset = offset;
1177         offset = unpack_object_header(p, &w_curs, offset, &kind, size);
1178
1179         for (;;) {
1180                 switch (kind) {
1181                 default:
1182                         die("pack %s contains unknown object type %d",
1183                             p->pack_name, kind);
1184                 case OBJ_COMMIT:
1185                 case OBJ_TREE:
1186                 case OBJ_BLOB:
1187                 case OBJ_TAG:
1188                         strcpy(type, type_names[kind]);
1189                         *store_size = 0; /* notyet */
1190                         unuse_pack(&w_curs);
1191                         return;
1192                 case OBJ_OFS_DELTA:
1193                         get_delta_base(p, &w_curs, offset, kind,
1194                                 obj_offset, &offset);
1195                         if (*delta_chain_length == 0) {
1196                                 /* TODO: find base_sha1 as pointed by offset */
1197                         }
1198                         break;
1199                 case OBJ_REF_DELTA:
1200                         next_sha1 = use_pack(p, &w_curs, offset, NULL);
1201                         if (*delta_chain_length == 0)
1202                                 hashcpy(base_sha1, next_sha1);
1203                         offset = find_pack_entry_one(next_sha1, p);
1204                         break;
1205                 }
1206                 obj_offset = offset;
1207                 offset = unpack_object_header(p, &w_curs, offset, &kind, &val);
1208                 (*delta_chain_length)++;
1209         }
1210 }
1211
1212 static int packed_object_info(struct packed_git *p, unsigned long offset,
1213                               char *type, unsigned long *sizep)
1214 {
1215         struct pack_window *w_curs = NULL;
1216         unsigned long size, obj_offset = offset;
1217         enum object_type kind;
1218         int r;
1219
1220         offset = unpack_object_header(p, &w_curs, offset, &kind, &size);
1221
1222         switch (kind) {
1223         case OBJ_OFS_DELTA:
1224         case OBJ_REF_DELTA:
1225                 r = packed_delta_info(p, &w_curs, offset, kind,
1226                         obj_offset, type, sizep);
1227                 unuse_pack(&w_curs);
1228                 return r;
1229         case OBJ_COMMIT:
1230         case OBJ_TREE:
1231         case OBJ_BLOB:
1232         case OBJ_TAG:
1233                 strcpy(type, type_names[kind]);
1234                 unuse_pack(&w_curs);
1235                 break;
1236         default:
1237                 die("pack %s contains unknown object type %d",
1238                     p->pack_name, kind);
1239         }
1240         if (sizep)
1241                 *sizep = size;
1242         return 0;
1243 }
1244
1245 static void *unpack_compressed_entry(struct packed_git *p,
1246                                     struct pack_window **w_curs,
1247                                     unsigned long offset,
1248                                     unsigned long size)
1249 {
1250         int st;
1251         z_stream stream;
1252         unsigned char *buffer, *in;
1253
1254         buffer = xmalloc(size + 1);
1255         buffer[size] = 0;
1256         memset(&stream, 0, sizeof(stream));
1257         stream.next_out = buffer;
1258         stream.avail_out = size;
1259
1260         inflateInit(&stream);
1261         do {
1262                 in = use_pack(p, w_curs, offset, &stream.avail_in);
1263                 stream.next_in = in;
1264                 st = inflate(&stream, Z_FINISH);
1265                 offset += stream.next_in - in;
1266         } while (st == Z_OK || st == Z_BUF_ERROR);
1267         inflateEnd(&stream);
1268         if ((st != Z_STREAM_END) || stream.total_out != size) {
1269                 free(buffer);
1270                 return NULL;
1271         }
1272
1273         return buffer;
1274 }
1275
1276 static void *unpack_delta_entry(struct packed_git *p,
1277                                 struct pack_window **w_curs,
1278                                 unsigned long offset,
1279                                 unsigned long delta_size,
1280                                 enum object_type kind,
1281                                 unsigned long obj_offset,
1282                                 char *type,
1283                                 unsigned long *sizep)
1284 {
1285         void *delta_data, *result, *base;
1286         unsigned long result_size, base_size, base_offset;
1287
1288         offset = get_delta_base(p, w_curs, offset, kind,
1289                 obj_offset, &base_offset);
1290         base = unpack_entry(p, base_offset, type, &base_size);
1291         if (!base)
1292                 die("failed to read delta base object at %lu from %s",
1293                     base_offset, p->pack_name);
1294
1295         delta_data = unpack_compressed_entry(p, w_curs, offset, delta_size);
1296         result = patch_delta(base, base_size,
1297                              delta_data, delta_size,
1298                              &result_size);
1299         if (!result)
1300                 die("failed to apply delta");
1301         free(delta_data);
1302         free(base);
1303         *sizep = result_size;
1304         return result;
1305 }
1306
1307 void *unpack_entry(struct packed_git *p, unsigned long offset,
1308                           char *type, unsigned long *sizep)
1309 {
1310         struct pack_window *w_curs = NULL;
1311         unsigned long size, obj_offset = offset;
1312         enum object_type kind;
1313         void *retval;
1314
1315         offset = unpack_object_header(p, &w_curs, offset, &kind, &size);
1316         switch (kind) {
1317         case OBJ_OFS_DELTA:
1318         case OBJ_REF_DELTA:
1319                 retval = unpack_delta_entry(p, &w_curs, offset, size,
1320                         kind, obj_offset, type, sizep);
1321                 break;
1322         case OBJ_COMMIT:
1323         case OBJ_TREE:
1324         case OBJ_BLOB:
1325         case OBJ_TAG:
1326                 strcpy(type, type_names[kind]);
1327                 *sizep = size;
1328                 retval = unpack_compressed_entry(p, &w_curs, offset, size);
1329                 break;
1330         default:
1331                 die("unknown object type %i in %s", kind, p->pack_name);
1332         }
1333         unuse_pack(&w_curs);
1334         return retval;
1335 }
1336
1337 int num_packed_objects(const struct packed_git *p)
1338 {
1339         /* See check_packed_git_idx() */
1340         return (p->index_size - 20 - 20 - 4*256) / 24;
1341 }
1342
1343 int nth_packed_object_sha1(const struct packed_git *p, int n,
1344                            unsigned char* sha1)
1345 {
1346         void *index = p->index_base + 256;
1347         if (n < 0 || num_packed_objects(p) <= n)
1348                 return -1;
1349         hashcpy(sha1, (unsigned char *) index + (24 * n) + 4);
1350         return 0;
1351 }
1352
1353 unsigned long find_pack_entry_one(const unsigned char *sha1,
1354                                   struct packed_git *p)
1355 {
1356         uint32_t *level1_ofs = p->index_base;
1357         int hi = ntohl(level1_ofs[*sha1]);
1358         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1359         void *index = p->index_base + 256;
1360
1361         do {
1362                 int mi = (lo + hi) / 2;
1363                 int cmp = hashcmp((unsigned char *)index + (24 * mi) + 4, sha1);
1364                 if (!cmp)
1365                         return ntohl(*((uint32_t *)((char *)index + (24 * mi))));
1366                 if (cmp > 0)
1367                         hi = mi;
1368                 else
1369                         lo = mi+1;
1370         } while (lo < hi);
1371         return 0;
1372 }
1373
1374 static int matches_pack_name(struct packed_git *p, const char *ig)
1375 {
1376         const char *last_c, *c;
1377
1378         if (!strcmp(p->pack_name, ig))
1379                 return 0;
1380
1381         for (c = p->pack_name, last_c = c; *c;)
1382                 if (*c == '/')
1383                         last_c = ++c;
1384                 else
1385                         ++c;
1386         if (!strcmp(last_c, ig))
1387                 return 0;
1388
1389         return 1;
1390 }
1391
1392 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e, const char **ignore_packed)
1393 {
1394         struct packed_git *p;
1395         unsigned long offset;
1396
1397         prepare_packed_git();
1398
1399         for (p = packed_git; p; p = p->next) {
1400                 if (ignore_packed) {
1401                         const char **ig;
1402                         for (ig = ignore_packed; *ig; ig++)
1403                                 if (!matches_pack_name(p, *ig))
1404                                         break;
1405                         if (*ig)
1406                                 continue;
1407                 }
1408                 offset = find_pack_entry_one(sha1, p);
1409                 if (offset) {
1410                         e->offset = offset;
1411                         e->p = p;
1412                         hashcpy(e->sha1, sha1);
1413                         return 1;
1414                 }
1415         }
1416         return 0;
1417 }
1418
1419 struct packed_git *find_sha1_pack(const unsigned char *sha1,
1420                                   struct packed_git *packs)
1421 {
1422         struct packed_git *p;
1423
1424         for (p = packs; p; p = p->next) {
1425                 if (find_pack_entry_one(sha1, p))
1426                         return p;
1427         }
1428         return NULL;
1429
1430 }
1431
1432 static int sha1_loose_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1433 {
1434         int status;
1435         unsigned long mapsize, size;
1436         void *map;
1437         z_stream stream;
1438         char hdr[128];
1439
1440         map = map_sha1_file(sha1, &mapsize);
1441         if (!map)
1442                 return error("unable to find %s", sha1_to_hex(sha1));
1443         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1444                 status = error("unable to unpack %s header",
1445                                sha1_to_hex(sha1));
1446         if (parse_sha1_header(hdr, type, &size) < 0)
1447                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1448         else {
1449                 status = 0;
1450                 if (sizep)
1451                         *sizep = size;
1452         }
1453         inflateEnd(&stream);
1454         munmap(map, mapsize);
1455         return status;
1456 }
1457
1458 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1459 {
1460         struct pack_entry e;
1461
1462         if (!find_pack_entry(sha1, &e, NULL)) {
1463                 reprepare_packed_git();
1464                 if (!find_pack_entry(sha1, &e, NULL))
1465                         return sha1_loose_object_info(sha1, type, sizep);
1466         }
1467         return packed_object_info(e.p, e.offset, type, sizep);
1468 }
1469
1470 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1471 {
1472         struct pack_entry e;
1473
1474         if (!find_pack_entry(sha1, &e, NULL))
1475                 return NULL;
1476         else
1477                 return unpack_entry(e.p, e.offset, type, size);
1478 }
1479
1480 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1481 {
1482         unsigned long mapsize;
1483         void *map, *buf;
1484
1485         buf = read_packed_sha1(sha1, type, size);
1486         if (buf)
1487                 return buf;
1488         map = map_sha1_file(sha1, &mapsize);
1489         if (map) {
1490                 buf = unpack_sha1_file(map, mapsize, type, size);
1491                 munmap(map, mapsize);
1492                 return buf;
1493         }
1494         reprepare_packed_git();
1495         return read_packed_sha1(sha1, type, size);
1496 }
1497
1498 void *read_object_with_reference(const unsigned char *sha1,
1499                                  const char *required_type,
1500                                  unsigned long *size,
1501                                  unsigned char *actual_sha1_return)
1502 {
1503         char type[20];
1504         void *buffer;
1505         unsigned long isize;
1506         unsigned char actual_sha1[20];
1507
1508         hashcpy(actual_sha1, sha1);
1509         while (1) {
1510                 int ref_length = -1;
1511                 const char *ref_type = NULL;
1512
1513                 buffer = read_sha1_file(actual_sha1, type, &isize);
1514                 if (!buffer)
1515                         return NULL;
1516                 if (!strcmp(type, required_type)) {
1517                         *size = isize;
1518                         if (actual_sha1_return)
1519                                 hashcpy(actual_sha1_return, actual_sha1);
1520                         return buffer;
1521                 }
1522                 /* Handle references */
1523                 else if (!strcmp(type, commit_type))
1524                         ref_type = "tree ";
1525                 else if (!strcmp(type, tag_type))
1526                         ref_type = "object ";
1527                 else {
1528                         free(buffer);
1529                         return NULL;
1530                 }
1531                 ref_length = strlen(ref_type);
1532
1533                 if (memcmp(buffer, ref_type, ref_length) ||
1534                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1535                         free(buffer);
1536                         return NULL;
1537                 }
1538                 free(buffer);
1539                 /* Now we have the ID of the referred-to object in
1540                  * actual_sha1.  Check again. */
1541         }
1542 }
1543
1544 static void write_sha1_file_prepare(void *buf, unsigned long len,
1545                                     const char *type, unsigned char *sha1,
1546                                     unsigned char *hdr, int *hdrlen)
1547 {
1548         SHA_CTX c;
1549
1550         /* Generate the header */
1551         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1552
1553         /* Sha1.. */
1554         SHA1_Init(&c);
1555         SHA1_Update(&c, hdr, *hdrlen);
1556         SHA1_Update(&c, buf, len);
1557         SHA1_Final(sha1, &c);
1558 }
1559
1560 /*
1561  * Link the tempfile to the final place, possibly creating the
1562  * last directory level as you do so.
1563  *
1564  * Returns the errno on failure, 0 on success.
1565  */
1566 static int link_temp_to_file(const char *tmpfile, const char *filename)
1567 {
1568         int ret;
1569         char *dir;
1570
1571         if (!link(tmpfile, filename))
1572                 return 0;
1573
1574         /*
1575          * Try to mkdir the last path component if that failed.
1576          *
1577          * Re-try the "link()" regardless of whether the mkdir
1578          * succeeds, since a race might mean that somebody
1579          * else succeeded.
1580          */
1581         ret = errno;
1582         dir = strrchr(filename, '/');
1583         if (dir) {
1584                 *dir = 0;
1585                 if (!mkdir(filename, 0777) && adjust_shared_perm(filename)) {
1586                         *dir = '/';
1587                         return -2;
1588                 }
1589                 *dir = '/';
1590                 if (!link(tmpfile, filename))
1591                         return 0;
1592                 ret = errno;
1593         }
1594         return ret;
1595 }
1596
1597 /*
1598  * Move the just written object into its final resting place
1599  */
1600 int move_temp_to_file(const char *tmpfile, const char *filename)
1601 {
1602         int ret = link_temp_to_file(tmpfile, filename);
1603
1604         /*
1605          * Coda hack - coda doesn't like cross-directory links,
1606          * so we fall back to a rename, which will mean that it
1607          * won't be able to check collisions, but that's not a
1608          * big deal.
1609          *
1610          * The same holds for FAT formatted media.
1611          *
1612          * When this succeeds, we just return 0. We have nothing
1613          * left to unlink.
1614          */
1615         if (ret && ret != EEXIST) {
1616                 if (!rename(tmpfile, filename))
1617                         return 0;
1618                 ret = errno;
1619         }
1620         unlink(tmpfile);
1621         if (ret) {
1622                 if (ret != EEXIST) {
1623                         return error("unable to write sha1 filename %s: %s\n", filename, strerror(ret));
1624                 }
1625                 /* FIXME!!! Collision check here ? */
1626         }
1627
1628         return 0;
1629 }
1630
1631 static int write_buffer(int fd, const void *buf, size_t len)
1632 {
1633         if (write_in_full(fd, buf, len) < 0)
1634                 return error("file write error (%s)", strerror(errno));
1635         return 0;
1636 }
1637
1638 static int write_binary_header(unsigned char *hdr, enum object_type type, unsigned long len)
1639 {
1640         int hdr_len;
1641         unsigned char c;
1642
1643         c = (type << 4) | (len & 15);
1644         len >>= 4;
1645         hdr_len = 1;
1646         while (len) {
1647                 *hdr++ = c | 0x80;
1648                 hdr_len++;
1649                 c = (len & 0x7f);
1650                 len >>= 7;
1651         }
1652         *hdr = c;
1653         return hdr_len;
1654 }
1655
1656 static void setup_object_header(z_stream *stream, const char *type, unsigned long len)
1657 {
1658         int obj_type, hdr;
1659
1660         if (use_legacy_headers) {
1661                 while (deflate(stream, 0) == Z_OK)
1662                         /* nothing */;
1663                 return;
1664         }
1665         if (!strcmp(type, blob_type))
1666                 obj_type = OBJ_BLOB;
1667         else if (!strcmp(type, tree_type))
1668                 obj_type = OBJ_TREE;
1669         else if (!strcmp(type, commit_type))
1670                 obj_type = OBJ_COMMIT;
1671         else if (!strcmp(type, tag_type))
1672                 obj_type = OBJ_TAG;
1673         else
1674                 die("trying to generate bogus object of type '%s'", type);
1675         hdr = write_binary_header(stream->next_out, obj_type, len);
1676         stream->total_out = hdr;
1677         stream->next_out += hdr;
1678         stream->avail_out -= hdr;
1679 }
1680
1681 int hash_sha1_file(void *buf, unsigned long len, const char *type,
1682                    unsigned char *sha1)
1683 {
1684         unsigned char hdr[50];
1685         int hdrlen;
1686         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1687         return 0;
1688 }
1689
1690 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1691 {
1692         int size;
1693         unsigned char *compressed;
1694         z_stream stream;
1695         unsigned char sha1[20];
1696         char *filename;
1697         static char tmpfile[PATH_MAX];
1698         unsigned char hdr[50];
1699         int fd, hdrlen;
1700
1701         /* Normally if we have it in the pack then we do not bother writing
1702          * it out into .git/objects/??/?{38} file.
1703          */
1704         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1705         filename = sha1_file_name(sha1);
1706         if (returnsha1)
1707                 hashcpy(returnsha1, sha1);
1708         if (has_sha1_file(sha1))
1709                 return 0;
1710         fd = open(filename, O_RDONLY);
1711         if (fd >= 0) {
1712                 /*
1713                  * FIXME!!! We might do collision checking here, but we'd
1714                  * need to uncompress the old file and check it. Later.
1715                  */
1716                 close(fd);
1717                 return 0;
1718         }
1719
1720         if (errno != ENOENT) {
1721                 return error("sha1 file %s: %s\n", filename, strerror(errno));
1722         }
1723
1724         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1725
1726         fd = mkstemp(tmpfile);
1727         if (fd < 0) {
1728                 if (errno == EPERM)
1729                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1730                 else
1731                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1732         }
1733
1734         /* Set it up */
1735         memset(&stream, 0, sizeof(stream));
1736         deflateInit(&stream, zlib_compression_level);
1737         size = 8 + deflateBound(&stream, len+hdrlen);
1738         compressed = xmalloc(size);
1739
1740         /* Compress it */
1741         stream.next_out = compressed;
1742         stream.avail_out = size;
1743
1744         /* First header.. */
1745         stream.next_in = hdr;
1746         stream.avail_in = hdrlen;
1747         setup_object_header(&stream, type, len);
1748
1749         /* Then the data itself.. */
1750         stream.next_in = buf;
1751         stream.avail_in = len;
1752         while (deflate(&stream, Z_FINISH) == Z_OK)
1753                 /* nothing */;
1754         deflateEnd(&stream);
1755         size = stream.total_out;
1756
1757         if (write_buffer(fd, compressed, size) < 0)
1758                 die("unable to write sha1 file");
1759         fchmod(fd, 0444);
1760         close(fd);
1761         free(compressed);
1762
1763         return move_temp_to_file(tmpfile, filename);
1764 }
1765
1766 /*
1767  * We need to unpack and recompress the object for writing
1768  * it out to a different file.
1769  */
1770 static void *repack_object(const unsigned char *sha1, unsigned long *objsize)
1771 {
1772         size_t size;
1773         z_stream stream;
1774         unsigned char *unpacked;
1775         unsigned long len;
1776         char type[20];
1777         char hdr[50];
1778         int hdrlen;
1779         void *buf;
1780
1781         /* need to unpack and recompress it by itself */
1782         unpacked = read_packed_sha1(sha1, type, &len);
1783         if (!unpacked)
1784                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1785
1786         hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1787
1788         /* Set it up */
1789         memset(&stream, 0, sizeof(stream));
1790         deflateInit(&stream, zlib_compression_level);
1791         size = deflateBound(&stream, len + hdrlen);
1792         buf = xmalloc(size);
1793
1794         /* Compress it */
1795         stream.next_out = buf;
1796         stream.avail_out = size;
1797
1798         /* First header.. */
1799         stream.next_in = (void *)hdr;
1800         stream.avail_in = hdrlen;
1801         while (deflate(&stream, 0) == Z_OK)
1802                 /* nothing */;
1803
1804         /* Then the data itself.. */
1805         stream.next_in = unpacked;
1806         stream.avail_in = len;
1807         while (deflate(&stream, Z_FINISH) == Z_OK)
1808                 /* nothing */;
1809         deflateEnd(&stream);
1810         free(unpacked);
1811
1812         *objsize = stream.total_out;
1813         return buf;
1814 }
1815
1816 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1817 {
1818         int retval;
1819         unsigned long objsize;
1820         void *buf = map_sha1_file(sha1, &objsize);
1821
1822         if (buf) {
1823                 retval = write_buffer(fd, buf, objsize);
1824                 munmap(buf, objsize);
1825                 return retval;
1826         }
1827
1828         buf = repack_object(sha1, &objsize);
1829         retval = write_buffer(fd, buf, objsize);
1830         free(buf);
1831         return retval;
1832 }
1833
1834 int write_sha1_from_fd(const unsigned char *sha1, int fd, char *buffer,
1835                        size_t bufsize, size_t *bufposn)
1836 {
1837         char tmpfile[PATH_MAX];
1838         int local;
1839         z_stream stream;
1840         unsigned char real_sha1[20];
1841         unsigned char discard[4096];
1842         int ret;
1843         SHA_CTX c;
1844
1845         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1846
1847         local = mkstemp(tmpfile);
1848         if (local < 0) {
1849                 if (errno == EPERM)
1850                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1851                 else
1852                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1853         }
1854
1855         memset(&stream, 0, sizeof(stream));
1856
1857         inflateInit(&stream);
1858
1859         SHA1_Init(&c);
1860
1861         do {
1862                 ssize_t size;
1863                 if (*bufposn) {
1864                         stream.avail_in = *bufposn;
1865                         stream.next_in = (unsigned char *) buffer;
1866                         do {
1867                                 stream.next_out = discard;
1868                                 stream.avail_out = sizeof(discard);
1869                                 ret = inflate(&stream, Z_SYNC_FLUSH);
1870                                 SHA1_Update(&c, discard, sizeof(discard) -
1871                                             stream.avail_out);
1872                         } while (stream.avail_in && ret == Z_OK);
1873                         if (write_buffer(local, buffer, *bufposn - stream.avail_in) < 0)
1874                                 die("unable to write sha1 file");
1875                         memmove(buffer, buffer + *bufposn - stream.avail_in,
1876                                 stream.avail_in);
1877                         *bufposn = stream.avail_in;
1878                         if (ret != Z_OK)
1879                                 break;
1880                 }
1881                 size = xread(fd, buffer + *bufposn, bufsize - *bufposn);
1882                 if (size <= 0) {
1883                         close(local);
1884                         unlink(tmpfile);
1885                         if (!size)
1886                                 return error("Connection closed?");
1887                         perror("Reading from connection");
1888                         return -1;
1889                 }
1890                 *bufposn += size;
1891         } while (1);
1892         inflateEnd(&stream);
1893
1894         close(local);
1895         SHA1_Final(real_sha1, &c);
1896         if (ret != Z_STREAM_END) {
1897                 unlink(tmpfile);
1898                 return error("File %s corrupted", sha1_to_hex(sha1));
1899         }
1900         if (hashcmp(sha1, real_sha1)) {
1901                 unlink(tmpfile);
1902                 return error("File %s has bad hash", sha1_to_hex(sha1));
1903         }
1904
1905         return move_temp_to_file(tmpfile, sha1_file_name(sha1));
1906 }
1907
1908 int has_pack_index(const unsigned char *sha1)
1909 {
1910         struct stat st;
1911         if (stat(sha1_pack_index_name(sha1), &st))
1912                 return 0;
1913         return 1;
1914 }
1915
1916 int has_pack_file(const unsigned char *sha1)
1917 {
1918         struct stat st;
1919         if (stat(sha1_pack_name(sha1), &st))
1920                 return 0;
1921         return 1;
1922 }
1923
1924 int has_sha1_pack(const unsigned char *sha1, const char **ignore_packed)
1925 {
1926         struct pack_entry e;
1927         return find_pack_entry(sha1, &e, ignore_packed);
1928 }
1929
1930 int has_sha1_file(const unsigned char *sha1)
1931 {
1932         struct stat st;
1933         struct pack_entry e;
1934
1935         if (find_pack_entry(sha1, &e, NULL))
1936                 return 1;
1937         return find_sha1_file(sha1, &st) ? 1 : 0;
1938 }
1939
1940 /*
1941  * reads from fd as long as possible into a supplied buffer of size bytes.
1942  * If necessary the buffer's size is increased using realloc()
1943  *
1944  * returns 0 if anything went fine and -1 otherwise
1945  *
1946  * NOTE: both buf and size may change, but even when -1 is returned
1947  * you still have to free() it yourself.
1948  */
1949 int read_pipe(int fd, char** return_buf, unsigned long* return_size)
1950 {
1951         char* buf = *return_buf;
1952         unsigned long size = *return_size;
1953         int iret;
1954         unsigned long off = 0;
1955
1956         do {
1957                 iret = xread(fd, buf + off, size - off);
1958                 if (iret > 0) {
1959                         off += iret;
1960                         if (off == size) {
1961                                 size *= 2;
1962                                 buf = xrealloc(buf, size);
1963                         }
1964                 }
1965         } while (iret > 0);
1966
1967         *return_buf = buf;
1968         *return_size = off;
1969
1970         if (iret < 0)
1971                 return -1;
1972         return 0;
1973 }
1974
1975 int index_pipe(unsigned char *sha1, int fd, const char *type, int write_object)
1976 {
1977         unsigned long size = 4096;
1978         char *buf = xmalloc(size);
1979         int ret;
1980
1981         if (read_pipe(fd, &buf, &size)) {
1982                 free(buf);
1983                 return -1;
1984         }
1985
1986         if (!type)
1987                 type = blob_type;
1988         if (write_object)
1989                 ret = write_sha1_file(buf, size, type, sha1);
1990         else
1991                 ret = hash_sha1_file(buf, size, type, sha1);
1992         free(buf);
1993         return ret;
1994 }
1995
1996 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1997 {
1998         unsigned long size = st->st_size;
1999         void *buf;
2000         int ret;
2001
2002         buf = "";
2003         if (size)
2004                 buf = xmmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
2005         close(fd);
2006
2007         if (!type)
2008                 type = blob_type;
2009         if (write_object)
2010                 ret = write_sha1_file(buf, size, type, sha1);
2011         else
2012                 ret = hash_sha1_file(buf, size, type, sha1);
2013         if (size)
2014                 munmap(buf, size);
2015         return ret;
2016 }
2017
2018 int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
2019 {
2020         int fd;
2021         char *target;
2022
2023         switch (st->st_mode & S_IFMT) {
2024         case S_IFREG:
2025                 fd = open(path, O_RDONLY);
2026                 if (fd < 0)
2027                         return error("open(\"%s\"): %s", path,
2028                                      strerror(errno));
2029                 if (index_fd(sha1, fd, st, write_object, NULL) < 0)
2030                         return error("%s: failed to insert into database",
2031                                      path);
2032                 break;
2033         case S_IFLNK:
2034                 target = xmalloc(st->st_size+1);
2035                 if (readlink(path, target, st->st_size+1) != st->st_size) {
2036                         char *errstr = strerror(errno);
2037                         free(target);
2038                         return error("readlink(\"%s\"): %s", path,
2039                                      errstr);
2040                 }
2041                 if (!write_object)
2042                         hash_sha1_file(target, st->st_size, blob_type, sha1);
2043                 else if (write_sha1_file(target, st->st_size, blob_type, sha1))
2044                         return error("%s: failed to insert into database",
2045                                      path);
2046                 free(target);
2047                 break;
2048         default:
2049                 return error("%s: unsupported file type", path);
2050         }
2051         return 0;
2052 }
2053
2054 int read_pack_header(int fd, struct pack_header *header)
2055 {
2056         char *c = (char*)header;
2057         ssize_t remaining = sizeof(struct pack_header);
2058         do {
2059                 ssize_t r = xread(fd, c, remaining);
2060                 if (r <= 0)
2061                         /* "eof before pack header was fully read" */
2062                         return PH_ERROR_EOF;
2063                 remaining -= r;
2064                 c += r;
2065         } while (remaining > 0);
2066         if (header->hdr_signature != htonl(PACK_SIGNATURE))
2067                 /* "protocol error (pack signature mismatch detected)" */
2068                 return PH_ERROR_PACK_SIGNATURE;
2069         if (!pack_version_ok(header->hdr_version))
2070                 /* "protocol error (pack version unsupported)" */
2071                 return PH_ERROR_PROTOCOL;
2072         return 0;
2073 }