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