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 unuse_pack(struct pack_window **w_cursor)
 526 {
 527         struct pack_window *w = *w_cursor;
 528         if (w) {
 529                 w->inuse_cnt--;
 530                 *w_cursor = NULL;
 531         }
 532 }
 533
 534 static void open_packed_git(struct packed_git *p)
 535 {
 536         struct stat st;
 537         struct pack_header hdr;
 538         unsigned char sha1[20];
 539         unsigned char *idx_sha1;
 540
 541         p->pack_fd = open(p->pack_name, O_RDONLY);
 542         if (p->pack_fd < 0 || fstat(p->pack_fd, &st))
 543                 die("packfile %s cannot be opened", p->pack_name);
 544
 545         /* If we created the struct before we had the pack we lack size. */
 546         if (!p->pack_size) {
 547                 if (!S_ISREG(st.st_mode))
 548                         die("packfile %s not a regular file", p->pack_name);
 549                 p->pack_size = st.st_size;
 550         } else if (p->pack_size != st.st_size)
 551                 die("packfile %s size changed", p->pack_name);
 552
 553         /* Verify we recognize this pack file format. */
 554         read_or_die(p->pack_fd, &hdr, sizeof(hdr));
 555         if (hdr.hdr_signature != htonl(PACK_SIGNATURE))
 556                 die("file %s is not a GIT packfile", p->pack_name);
 557         if (!pack_version_ok(hdr.hdr_version))
 558                 die("packfile %s is version %u and not supported"
 559                         " (try upgrading GIT to a newer version)",
 560                         p->pack_name, ntohl(hdr.hdr_version));
 561
 562         /* Verify the pack matches its index. */
 563         if (num_packed_objects(p) != ntohl(hdr.hdr_entries))
 564                 die("packfile %s claims to have %u objects"
 565                         " while index size indicates %u objects",
 566                         p->pack_name, ntohl(hdr.hdr_entries),
 567                         num_packed_objects(p));
 568         if (lseek(p->pack_fd, p->pack_size - sizeof(sha1), SEEK_SET) == -1)
 569                 die("end of packfile %s is unavailable", p->pack_name);
 570         read_or_die(p->pack_fd, sha1, sizeof(sha1));
 571         idx_sha1 = ((unsigned char *)p->index_base) + p->index_size - 40;
 572         if (hashcmp(sha1, idx_sha1))
 573                 die("packfile %s does not match index", p->pack_name);
 574 }
 575
 576 static int in_window(struct pack_window *win, unsigned long offset)
 577 {
 578         /* We must promise at least 20 bytes (one hash) after the
 579          * offset is available from this window, otherwise the offset
 580          * is not actually in this window and a different window (which
 581          * has that one hash excess) must be used.  This is to support
 582          * the object header and delta base parsing routines below.
 583          */
 584         off_t win_off = win->offset;
 585         return win_off <= offset
 586                 && (offset + 20) <= (win_off + win->len);
 587 }
 588
 589 unsigned char* use_pack(struct packed_git *p,
 590                 struct pack_window **w_cursor,
 591                 unsigned long offset,
 592                 unsigned int *left)
 593 {
 594         struct pack_window *win = *w_cursor;
 595
 596         if (p->pack_fd == -1)
 597                 open_packed_git(p);
 598
 599         /* Since packfiles end in a hash of their content and its
 600          * pointless to ask for an offset into the middle of that
 601          * hash, and the in_window function above wouldn't match
 602          * don't allow an offset too close to the end of the file.
 603          */
 604         if (offset > (p->pack_size - 20))
 605                 die("offset beyond end of packfile (truncated pack?)");
 606
 607         if (!win || !in_window(win, offset)) {
 608                 if (win)
 609                         win->inuse_cnt--;
 610                 for (win = p->windows; win; win = win->next) {
 611                         if (in_window(win, offset))
 612                                 break;
 613                 }
 614                 if (!win) {
 615                         if (!page_size)
 616                                 page_size = getpagesize();
 617                         win = xcalloc(1, sizeof(*win));
 618                         win->offset = (offset / page_size) * page_size;
 619                         win->len = p->pack_size - win->offset;
 620                         if (win->len > packed_git_window_size)
 621                                 win->len = packed_git_window_size;
 622                         pack_mapped += win->len;
 623                         while (packed_git_limit < pack_mapped
 624                                 && unuse_one_window(p))
 625                                 ; /* nothing */
 626                         win->base = mmap(NULL, win->len,
 627                                 PROT_READ, MAP_PRIVATE,
 628                                 p->pack_fd, win->offset);
 629                         if (win->base == MAP_FAILED)
 630                                 die("packfile %s cannot be mapped: %s",
 631                                         p->pack_name,
 632                                         strerror(errno));
 633                         pack_mmap_calls++;
 634                         pack_open_windows++;
 635                         if (pack_mapped > peak_pack_mapped)
 636                                 peak_pack_mapped = pack_mapped;
 637                         if (pack_open_windows > peak_pack_open_windows)
 638                                 peak_pack_open_windows = pack_open_windows;
 639                         win->next = p->windows;
 640                         p->windows = win;
 641                 }
 642         }
 643         if (win != *w_cursor) {
 644                 win->last_used = pack_used_ctr++;
 645                 win->inuse_cnt++;
 646                 *w_cursor = win;
 647         }
 648         offset -= win->offset;
 649         if (left)
 650                 *left = win->len - offset;
 651         return win->base + offset;
 652 }
 653
 654 struct packed_git *add_packed_git(char *path, int path_len, int local)
 655 {
 656         struct stat st;
 657         struct packed_git *p;
 658         unsigned long idx_size;
 659         void *idx_map;
 660         unsigned char sha1[20];
 661
 662         if (check_packed_git_idx(path, &idx_size, &idx_map))
 663                 return NULL;
 664
 665         /* do we have a corresponding .pack file? */
 666         strcpy(path + path_len - 4, ".pack");
 667         if (stat(path, &st) || !S_ISREG(st.st_mode)) {
 668                 munmap(idx_map, idx_size);
 669                 return NULL;
 670         }
 671         /* ok, it looks sane as far as we can check without
 672          * actually mapping the pack file.
 673          */
 674         p = xmalloc(sizeof(*p) + path_len + 2);
 675         strcpy(p->pack_name, path);
 676         p->index_size = idx_size;
 677         p->pack_size = st.st_size;
 678         p->index_base = idx_map;
 679         p->next = NULL;
 680         p->windows = NULL;
 681         p->pack_fd = -1;
 682         p->pack_local = local;
 683         if ((path_len > 44) && !get_sha1_hex(path + path_len - 44, sha1))
 684                 hashcpy(p->sha1, sha1);
 685         return p;
 686 }
 687
 688 struct packed_git *parse_pack_index(unsigned char *sha1)
 689 {
 690         char *path = sha1_pack_index_name(sha1);
 691         return parse_pack_index_file(sha1, path);
 692 }
 693
 694 struct packed_git *parse_pack_index_file(const unsigned char *sha1, char *idx_path)
 695 {
 696         struct packed_git *p;
 697         unsigned long idx_size;
 698         void *idx_map;
 699         char *path;
 700
 701         if (check_packed_git_idx(idx_path, &idx_size, &idx_map))
 702                 return NULL;
 703
 704         path = sha1_pack_name(sha1);
 705
 706         p = xmalloc(sizeof(*p) + strlen(path) + 2);
 707         strcpy(p->pack_name, path);
 708         p->index_size = idx_size;
 709         p->pack_size = 0;
 710         p->index_base = idx_map;
 711         p->next = NULL;
 712         p->windows = NULL;
 713         p->pack_fd = -1;
 714         hashcpy(p->sha1, sha1);
 715         return p;
 716 }
 717
 718 void install_packed_git(struct packed_git *pack)
 719 {
 720         pack->next = packed_git;
 721         packed_git = pack;
 722 }
 723
 724 static void prepare_packed_git_one(char *objdir, int local)
 725 {
 726         char path[PATH_MAX];
 727         int len;
 728         DIR *dir;
 729         struct dirent *de;
 730
 731         sprintf(path, "%s/pack", objdir);
 732         len = strlen(path);
 733         dir = opendir(path);
 734         if (!dir) {
 735                 if (errno != ENOENT)
 736                         error("unable to open object pack directory: %s: %s",
 737                               path, strerror(errno));
 738                 return;
 739         }
 740         path[len++] = '/';
 741         while ((de = readdir(dir)) != NULL) {
 742                 int namelen = strlen(de->d_name);
 743                 struct packed_git *p;
 744
 745                 if (!has_extension(de->d_name, ".idx"))
 746                         continue;
 747
 748                 /* we have .idx.  Is it a file we can map? */
 749                 strcpy(path + len, de->d_name);
 750                 for (p = packed_git; p; p = p->next) {
 751                         if (!memcmp(path, p->pack_name, len + namelen - 4))
 752                                 break;
 753                 }
 754                 if (p)
 755                         continue;
 756                 p = add_packed_git(path, len + namelen, local);
 757                 if (!p)
 758                         continue;
 759                 p->next = packed_git;
 760                 packed_git = p;
 761         }
 762         closedir(dir);
 763 }
 764
 765 static int prepare_packed_git_run_once = 0;
 766 void prepare_packed_git(void)
 767 {
 768         struct alternate_object_database *alt;
 769
 770         if (prepare_packed_git_run_once)
 771                 return;
 772         prepare_packed_git_one(get_object_directory(), 1);
 773         prepare_alt_odb();
 774         for (alt = alt_odb_list; alt; alt = alt->next) {
 775                 alt->name[-1] = 0;
 776                 prepare_packed_git_one(alt->base, 0);
 777                 alt->name[-1] = '/';
 778         }
 779         prepare_packed_git_run_once = 1;
 780 }
 781
 782 void reprepare_packed_git(void)
 783 {
 784         prepare_packed_git_run_once = 0;
 785         prepare_packed_git();
 786 }
 787
 788 int check_sha1_signature(const unsigned char *sha1, void *map, unsigned long size, const char *type)
 789 {
 790         unsigned char real_sha1[20];
 791         hash_sha1_file(map, size, type, real_sha1);
 792         return hashcmp(sha1, real_sha1) ? -1 : 0;
 793 }
 794
 795 void *map_sha1_file(const unsigned char *sha1, unsigned long *size)
 796 {
 797         struct stat st;
 798         void *map;
 799         int fd;
 800         char *filename = find_sha1_file(sha1, &st);
 801
 802         if (!filename) {
 803                 return NULL;
 804         }
 805
 806         fd = open(filename, O_RDONLY | sha1_file_open_flag);
 807         if (fd < 0) {
 808                 /* See if it works without O_NOATIME */
 809                 switch (sha1_file_open_flag) {
 810                 default:
 811                         fd = open(filename, O_RDONLY);
 812                         if (fd >= 0)
 813                                 break;
 814                 /* Fallthrough */
 815                 case 0:
 816                         return NULL;
 817                 }
 818
 819                 /* If it failed once, it will probably fail again.
 820                  * Stop using O_NOATIME
 821                  */
 822                 sha1_file_open_flag = 0;
 823         }
 824         map = mmap(NULL, st.st_size, PROT_READ, MAP_PRIVATE, fd, 0);
 825         close(fd);
 826         if (map == MAP_FAILED)
 827                 return NULL;
 828         *size = st.st_size;
 829         return map;
 830 }
 831
 832 int legacy_loose_object(unsigned char *map)
 833 {
 834         unsigned int word;
 835
 836         /*
 837          * Is it a zlib-compressed buffer? If so, the first byte
 838          * must be 0x78 (15-bit window size, deflated), and the
 839          * first 16-bit word is evenly divisible by 31
 840          */
 841         word = (map[0] << 8) + map[1];
 842         if (map[0] == 0x78 && !(word % 31))
 843                 return 1;
 844         else
 845                 return 0;
 846 }
 847
 848 unsigned long unpack_object_header_gently(const unsigned char *buf, unsigned long len, enum object_type *type, unsigned long *sizep)
 849 {
 850         unsigned shift;
 851         unsigned char c;
 852         unsigned long size;
 853         unsigned long used = 0;
 854
 855         c = buf[used++];
 856         *type = (c >> 4) & 7;
 857         size = c & 15;
 858         shift = 4;
 859         while (c & 0x80) {
 860                 if (len <= used)
 861                         return 0;
 862                 if (sizeof(long) * 8 <= shift)
 863                         return 0;
 864                 c = buf[used++];
 865                 size += (c & 0x7f) << shift;
 866                 shift += 7;
 867         }
 868         *sizep = size;
 869         return used;
 870 }
 871
 872 static int unpack_sha1_header(z_stream *stream, unsigned char *map, unsigned long mapsize, void *buffer, unsigned long bufsiz)
 873 {
 874         unsigned long size, used;
 875         static const char valid_loose_object_type[8] = {
 876                 0, /* OBJ_EXT */
 877                 1, 1, 1, 1, /* "commit", "tree", "blob", "tag" */
 878                 0, /* "delta" and others are invalid in a loose object */
 879         };
 880         enum object_type type;
 881
 882         /* Get the data stream */
 883         memset(stream, 0, sizeof(*stream));
 884         stream->next_in = map;
 885         stream->avail_in = mapsize;
 886         stream->next_out = buffer;
 887         stream->avail_out = bufsiz;
 888
 889         if (legacy_loose_object(map)) {
 890                 inflateInit(stream);
 891                 return inflate(stream, 0);
 892         }
 893
 894         used = unpack_object_header_gently(map, mapsize, &type, &size);
 895         if (!used || !valid_loose_object_type[type])
 896                 return -1;
 897         map += used;
 898         mapsize -= used;
 899
 900         /* Set up the stream for the rest.. */
 901         stream->next_in = map;
 902         stream->avail_in = mapsize;
 903         inflateInit(stream);
 904
 905         /* And generate the fake traditional header */
 906         stream->total_out = 1 + snprintf(buffer, bufsiz, "%s %lu",
 907                                          type_names[type], size);
 908         return 0;
 909 }
 910
 911 static void *unpack_sha1_rest(z_stream *stream, void *buffer, unsigned long size)
 912 {
 913         int bytes = strlen(buffer) + 1;
 914         unsigned char *buf = xmalloc(1+size);
 915         unsigned long n;
 916
 917         n = stream->total_out - bytes;
 918         if (n > size)
 919                 n = size;
 920         memcpy(buf, (char *) buffer + bytes, n);
 921         bytes = n;
 922         if (bytes < size) {
 923                 stream->next_out = buf + bytes;
 924                 stream->avail_out = size - bytes;
 925                 while (inflate(stream, Z_FINISH) == Z_OK)
 926                         /* nothing */;
 927         }
 928         buf[size] = 0;
 929         inflateEnd(stream);
 930         return buf;
 931 }
 932
 933 /*
 934  * We used to just use "sscanf()", but that's actually way
 935  * too permissive for what we want to check. So do an anal
 936  * object header parse by hand.
 937  */
 938 static int parse_sha1_header(char *hdr, char *type, unsigned long *sizep)
 939 {
 940         int i;
 941         unsigned long size;
 942
 943         /*
 944          * The type can be at most ten bytes (including the
 945          * terminating '\0' that we add), and is followed by
 946          * a space.
 947          */
 948         i = 10;
 949         for (;;) {
 950                 char c = *hdr++;
 951                 if (c == ' ')
 952                         break;
 953                 if (!--i)
 954                         return -1;
 955                 *type++ = c;
 956         }
 957         *type = 0;
 958
 959         /*
 960          * The length must follow immediately, and be in canonical
 961          * decimal format (ie "010" is not valid).
 962          */
 963         size = *hdr++ - '0';
 964         if (size > 9)
 965                 return -1;
 966         if (size) {
 967                 for (;;) {
 968                         unsigned long c = *hdr - '0';
 969                         if (c > 9)
 970                                 break;
 971                         hdr++;
 972                         size = size * 10 + c;
 973                 }
 974         }
 975         *sizep = size;
 976
 977         /*
 978          * The length must be followed by a zero byte
 979          */
 980         return *hdr ? -1 : 0;
 981 }
 982
 983 void * unpack_sha1_file(void *map, unsigned long mapsize, char *type, unsigned long *size)
 984 {
 985         int ret;
 986         z_stream stream;
 987         char hdr[8192];
 988
 989         ret = unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr));
 990         if (ret < Z_OK || parse_sha1_header(hdr, type, size) < 0)
 991                 return NULL;
 992
 993         return unpack_sha1_rest(&stream, hdr, *size);
 994 }
 995
 996 static unsigned long get_delta_base(struct packed_git *p,
 997                                     struct pack_window **w_curs,
 998                                     unsigned long offset,
 999                                     enum object_type kind,
1000                                     unsigned long delta_obj_offset,
1001                                     unsigned long *base_obj_offset)
1002 {
1003         unsigned char *base_info = use_pack(p, w_curs, offset, NULL);
1004         unsigned long base_offset;
1005
1006         /* use_pack() assured us we have [base_info, base_info + 20)
1007          * as a range that we can look at without walking off the
1008          * end of the mapped window.  Its actually the hash size
1009          * that is assured.  An OFS_DELTA longer than the hash size
1010          * is stupid, as then a REF_DELTA would be smaller to store.
1011          */
1012         if (kind == OBJ_OFS_DELTA) {
1013                 unsigned used = 0;
1014                 unsigned char c = base_info[used++];
1015                 base_offset = c & 127;
1016                 while (c & 128) {
1017                         base_offset += 1;
1018                         if (!base_offset || base_offset & ~(~0UL >> 7))
1019                                 die("offset value overflow for delta base object");
1020                         c = base_info[used++];
1021                         base_offset = (base_offset << 7) + (c & 127);
1022                 }
1023                 base_offset = delta_obj_offset - base_offset;
1024                 if (base_offset >= delta_obj_offset)
1025                         die("delta base offset out of bound");
1026                 offset += used;
1027         } else if (kind == OBJ_REF_DELTA) {
1028                 /* The base entry _must_ be in the same pack */
1029                 base_offset = find_pack_entry_one(base_info, p);
1030                 if (!base_offset)
1031                         die("failed to find delta-pack base object %s",
1032                                 sha1_to_hex(base_info));
1033                 offset += 20;
1034         } else
1035                 die("I am totally screwed");
1036         *base_obj_offset = base_offset;
1037         return offset;
1038 }
1039
1040 /* forward declaration for a mutually recursive function */
1041 static int packed_object_info(struct packed_git *p, unsigned long offset,
1042                               char *type, unsigned long *sizep);
1043
1044 static int packed_delta_info(struct packed_git *p,
1045                              struct pack_window **w_curs,
1046                              unsigned long offset,
1047                              enum object_type kind,
1048                              unsigned long obj_offset,
1049                              char *type,
1050                              unsigned long *sizep)
1051 {
1052         unsigned long base_offset;
1053
1054         offset = get_delta_base(p, w_curs, offset, kind,
1055                 obj_offset, &base_offset);
1056
1057         /* We choose to only get the type of the base object and
1058          * ignore potentially corrupt pack file that expects the delta
1059          * based on a base with a wrong size.  This saves tons of
1060          * inflate() calls.
1061          */
1062         if (packed_object_info(p, base_offset, type, NULL))
1063                 die("cannot get info for delta-pack base");
1064
1065         if (sizep) {
1066                 const unsigned char *data;
1067                 unsigned char delta_head[20], *in;
1068                 unsigned long result_size;
1069                 z_stream stream;
1070                 int st;
1071
1072                 memset(&stream, 0, sizeof(stream));
1073                 stream.next_out = delta_head;
1074                 stream.avail_out = sizeof(delta_head);
1075
1076                 inflateInit(&stream);
1077                 do {
1078                         in = use_pack(p, w_curs, offset, &stream.avail_in);
1079                         stream.next_in = in;
1080                         st = inflate(&stream, Z_FINISH);
1081                         offset += stream.next_in - in;
1082                 } while ((st == Z_OK || st == Z_BUF_ERROR)
1083                         && stream.total_out < sizeof(delta_head));
1084                 inflateEnd(&stream);
1085                 if ((st != Z_STREAM_END) &&
1086                     stream.total_out != sizeof(delta_head))
1087                         die("delta data unpack-initial failed");
1088
1089                 /* Examine the initial part of the delta to figure out
1090                  * the result size.
1091                  */
1092                 data = delta_head;
1093
1094                 /* ignore base size */
1095                 get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
1096
1097                 /* Read the result size */
1098                 result_size = get_delta_hdr_size(&data, delta_head+sizeof(delta_head));
1099                 *sizep = result_size;
1100         }
1101         return 0;
1102 }
1103
1104 static unsigned long unpack_object_header(struct packed_git *p,
1105                 struct pack_window **w_curs,
1106                 unsigned long offset,
1107                 enum object_type *type,
1108                 unsigned long *sizep)
1109 {
1110         unsigned char *base;
1111         unsigned int left;
1112         unsigned long used;
1113
1114         /* use_pack() assures us we have [base, base + 20) available
1115          * as a range that we can look at at.  (Its actually the hash
1116          * size that is assurred.)  With our object header encoding
1117          * the maximum deflated object size is 2^137, which is just
1118          * insane, so we know won't exceed what we have been given.
1119          */
1120         base = use_pack(p, w_curs, offset, &left);
1121         used = unpack_object_header_gently(base, left, type, sizep);
1122         if (!used)
1123                 die("object offset outside of pack file");
1124
1125         return offset + used;
1126 }
1127
1128 void packed_object_info_detail(struct packed_git *p,
1129                                unsigned long offset,
1130                                char *type,
1131                                unsigned long *size,
1132                                unsigned long *store_size,
1133                                unsigned int *delta_chain_length,
1134                                unsigned char *base_sha1)
1135 {
1136         struct pack_window *w_curs = NULL;
1137         unsigned long obj_offset, val;
1138         unsigned char *next_sha1;
1139         enum object_type kind;
1140
1141         *delta_chain_length = 0;
1142         obj_offset = offset;
1143         offset = unpack_object_header(p, &w_curs, offset, &kind, size);
1144
1145         for (;;) {
1146                 switch (kind) {
1147                 default:
1148                         die("pack %s contains unknown object type %d",
1149                             p->pack_name, kind);
1150                 case OBJ_COMMIT:
1151                 case OBJ_TREE:
1152                 case OBJ_BLOB:
1153                 case OBJ_TAG:
1154                         strcpy(type, type_names[kind]);
1155                         *store_size = 0; /* notyet */
1156                         unuse_pack(&w_curs);
1157                         return;
1158                 case OBJ_OFS_DELTA:
1159                         get_delta_base(p, &w_curs, offset, kind,
1160                                 obj_offset, &offset);
1161                         if (*delta_chain_length == 0) {
1162                                 /* TODO: find base_sha1 as pointed by offset */
1163                         }
1164                         break;
1165                 case OBJ_REF_DELTA:
1166                         next_sha1 = use_pack(p, &w_curs, offset, NULL);
1167                         if (*delta_chain_length == 0)
1168                                 hashcpy(base_sha1, next_sha1);
1169                         offset = find_pack_entry_one(next_sha1, p);
1170                         break;
1171                 }
1172                 obj_offset = offset;
1173                 offset = unpack_object_header(p, &w_curs, offset, &kind, &val);
1174                 (*delta_chain_length)++;
1175         }
1176 }
1177
1178 static int packed_object_info(struct packed_git *p, unsigned long offset,
1179                               char *type, unsigned long *sizep)
1180 {
1181         struct pack_window *w_curs = NULL;
1182         unsigned long size, obj_offset = offset;
1183         enum object_type kind;
1184         int r;
1185
1186         offset = unpack_object_header(p, &w_curs, offset, &kind, &size);
1187
1188         switch (kind) {
1189         case OBJ_OFS_DELTA:
1190         case OBJ_REF_DELTA:
1191                 r = packed_delta_info(p, &w_curs, offset, kind,
1192                         obj_offset, type, sizep);
1193                 unuse_pack(&w_curs);
1194                 return r;
1195         case OBJ_COMMIT:
1196         case OBJ_TREE:
1197         case OBJ_BLOB:
1198         case OBJ_TAG:
1199                 strcpy(type, type_names[kind]);
1200                 unuse_pack(&w_curs);
1201                 break;
1202         default:
1203                 die("pack %s contains unknown object type %d",
1204                     p->pack_name, kind);
1205         }
1206         if (sizep)
1207                 *sizep = size;
1208         return 0;
1209 }
1210
1211 static void *unpack_compressed_entry(struct packed_git *p,
1212                                     struct pack_window **w_curs,
1213                                     unsigned long offset,
1214                                     unsigned long size)
1215 {
1216         int st;
1217         z_stream stream;
1218         unsigned char *buffer, *in;
1219
1220         buffer = xmalloc(size + 1);
1221         buffer[size] = 0;
1222         memset(&stream, 0, sizeof(stream));
1223         stream.next_out = buffer;
1224         stream.avail_out = size;
1225
1226         inflateInit(&stream);
1227         do {
1228                 in = use_pack(p, w_curs, offset, &stream.avail_in);
1229                 stream.next_in = in;
1230                 st = inflate(&stream, Z_FINISH);
1231                 offset += stream.next_in - in;
1232         } while (st == Z_OK || st == Z_BUF_ERROR);
1233         inflateEnd(&stream);
1234         if ((st != Z_STREAM_END) || stream.total_out != size) {
1235                 free(buffer);
1236                 return NULL;
1237         }
1238
1239         return buffer;
1240 }
1241
1242 static void *unpack_delta_entry(struct packed_git *p,
1243                                 struct pack_window **w_curs,
1244                                 unsigned long offset,
1245                                 unsigned long delta_size,
1246                                 enum object_type kind,
1247                                 unsigned long obj_offset,
1248                                 char *type,
1249                                 unsigned long *sizep)
1250 {
1251         void *delta_data, *result, *base;
1252         unsigned long result_size, base_size, base_offset;
1253
1254         offset = get_delta_base(p, w_curs, offset, kind,
1255                 obj_offset, &base_offset);
1256         base = unpack_entry(p, base_offset, type, &base_size);
1257         if (!base)
1258                 die("failed to read delta base object at %lu from %s",
1259                     base_offset, p->pack_name);
1260
1261         delta_data = unpack_compressed_entry(p, w_curs, offset, delta_size);
1262         result = patch_delta(base, base_size,
1263                              delta_data, delta_size,
1264                              &result_size);
1265         if (!result)
1266                 die("failed to apply delta");
1267         free(delta_data);
1268         free(base);
1269         *sizep = result_size;
1270         return result;
1271 }
1272
1273 void *unpack_entry(struct packed_git *p, unsigned long offset,
1274                           char *type, unsigned long *sizep)
1275 {
1276         struct pack_window *w_curs = NULL;
1277         unsigned long size, obj_offset = offset;
1278         enum object_type kind;
1279         void *retval;
1280
1281         offset = unpack_object_header(p, &w_curs, offset, &kind, &size);
1282         switch (kind) {
1283         case OBJ_OFS_DELTA:
1284         case OBJ_REF_DELTA:
1285                 retval = unpack_delta_entry(p, &w_curs, offset, size,
1286                         kind, obj_offset, type, sizep);
1287                 break;
1288         case OBJ_COMMIT:
1289         case OBJ_TREE:
1290         case OBJ_BLOB:
1291         case OBJ_TAG:
1292                 strcpy(type, type_names[kind]);
1293                 *sizep = size;
1294                 retval = unpack_compressed_entry(p, &w_curs, offset, size);
1295                 break;
1296         default:
1297                 die("unknown object type %i in %s", kind, p->pack_name);
1298         }
1299         unuse_pack(&w_curs);
1300         return retval;
1301 }
1302
1303 int num_packed_objects(const struct packed_git *p)
1304 {
1305         /* See check_packed_git_idx() */
1306         return (p->index_size - 20 - 20 - 4*256) / 24;
1307 }
1308
1309 int nth_packed_object_sha1(const struct packed_git *p, int n,
1310                            unsigned char* sha1)
1311 {
1312         void *index = p->index_base + 256;
1313         if (n < 0 || num_packed_objects(p) <= n)
1314                 return -1;
1315         hashcpy(sha1, (unsigned char *) index + (24 * n) + 4);
1316         return 0;
1317 }
1318
1319 unsigned long find_pack_entry_one(const unsigned char *sha1,
1320                                   struct packed_git *p)
1321 {
1322         unsigned int *level1_ofs = p->index_base;
1323         int hi = ntohl(level1_ofs[*sha1]);
1324         int lo = ((*sha1 == 0x0) ? 0 : ntohl(level1_ofs[*sha1 - 1]));
1325         void *index = p->index_base + 256;
1326
1327         do {
1328                 int mi = (lo + hi) / 2;
1329                 int cmp = hashcmp((unsigned char *)index + (24 * mi) + 4, sha1);
1330                 if (!cmp)
1331                         return ntohl(*((unsigned int *) ((char *) index + (24 * mi))));
1332                 if (cmp > 0)
1333                         hi = mi;
1334                 else
1335                         lo = mi+1;
1336         } while (lo < hi);
1337         return 0;
1338 }
1339
1340 static int matches_pack_name(struct packed_git *p, const char *ig)
1341 {
1342         const char *last_c, *c;
1343
1344         if (!strcmp(p->pack_name, ig))
1345                 return 0;
1346
1347         for (c = p->pack_name, last_c = c; *c;)
1348                 if (*c == '/')
1349                         last_c = ++c;
1350                 else
1351                         ++c;
1352         if (!strcmp(last_c, ig))
1353                 return 0;
1354
1355         return 1;
1356 }
1357
1358 static int find_pack_entry(const unsigned char *sha1, struct pack_entry *e, const char **ignore_packed)
1359 {
1360         struct packed_git *p;
1361         unsigned long offset;
1362
1363         prepare_packed_git();
1364
1365         for (p = packed_git; p; p = p->next) {
1366                 if (ignore_packed) {
1367                         const char **ig;
1368                         for (ig = ignore_packed; *ig; ig++)
1369                                 if (!matches_pack_name(p, *ig))
1370                                         break;
1371                         if (*ig)
1372                                 continue;
1373                 }
1374                 offset = find_pack_entry_one(sha1, p);
1375                 if (offset) {
1376                         e->offset = offset;
1377                         e->p = p;
1378                         hashcpy(e->sha1, sha1);
1379                         return 1;
1380                 }
1381         }
1382         return 0;
1383 }
1384
1385 struct packed_git *find_sha1_pack(const unsigned char *sha1,
1386                                   struct packed_git *packs)
1387 {
1388         struct packed_git *p;
1389
1390         for (p = packs; p; p = p->next) {
1391                 if (find_pack_entry_one(sha1, p))
1392                         return p;
1393         }
1394         return NULL;
1395
1396 }
1397
1398 static int sha1_loose_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1399 {
1400         int status;
1401         unsigned long mapsize, size;
1402         void *map;
1403         z_stream stream;
1404         char hdr[128];
1405
1406         map = map_sha1_file(sha1, &mapsize);
1407         if (!map)
1408                 return error("unable to find %s", sha1_to_hex(sha1));
1409         if (unpack_sha1_header(&stream, map, mapsize, hdr, sizeof(hdr)) < 0)
1410                 status = error("unable to unpack %s header",
1411                                sha1_to_hex(sha1));
1412         if (parse_sha1_header(hdr, type, &size) < 0)
1413                 status = error("unable to parse %s header", sha1_to_hex(sha1));
1414         else {
1415                 status = 0;
1416                 if (sizep)
1417                         *sizep = size;
1418         }
1419         inflateEnd(&stream);
1420         munmap(map, mapsize);
1421         return status;
1422 }
1423
1424 int sha1_object_info(const unsigned char *sha1, char *type, unsigned long *sizep)
1425 {
1426         struct pack_entry e;
1427
1428         if (!find_pack_entry(sha1, &e, NULL)) {
1429                 reprepare_packed_git();
1430                 if (!find_pack_entry(sha1, &e, NULL))
1431                         return sha1_loose_object_info(sha1, type, sizep);
1432         }
1433         return packed_object_info(e.p, e.offset, type, sizep);
1434 }
1435
1436 static void *read_packed_sha1(const unsigned char *sha1, char *type, unsigned long *size)
1437 {
1438         struct pack_entry e;
1439
1440         if (!find_pack_entry(sha1, &e, NULL)) {
1441                 error("cannot read sha1_file for %s", sha1_to_hex(sha1));
1442                 return NULL;
1443         }
1444         return unpack_entry(e.p, e.offset, type, size);
1445 }
1446
1447 void * read_sha1_file(const unsigned char *sha1, char *type, unsigned long *size)
1448 {
1449         unsigned long mapsize;
1450         void *map, *buf;
1451         struct pack_entry e;
1452
1453         if (find_pack_entry(sha1, &e, NULL))
1454                 return read_packed_sha1(sha1, type, size);
1455         map = map_sha1_file(sha1, &mapsize);
1456         if (map) {
1457                 buf = unpack_sha1_file(map, mapsize, type, size);
1458                 munmap(map, mapsize);
1459                 return buf;
1460         }
1461         reprepare_packed_git();
1462         if (find_pack_entry(sha1, &e, NULL))
1463                 return read_packed_sha1(sha1, type, size);
1464         return NULL;
1465 }
1466
1467 void *read_object_with_reference(const unsigned char *sha1,
1468                                  const char *required_type,
1469                                  unsigned long *size,
1470                                  unsigned char *actual_sha1_return)
1471 {
1472         char type[20];
1473         void *buffer;
1474         unsigned long isize;
1475         unsigned char actual_sha1[20];
1476
1477         hashcpy(actual_sha1, sha1);
1478         while (1) {
1479                 int ref_length = -1;
1480                 const char *ref_type = NULL;
1481
1482                 buffer = read_sha1_file(actual_sha1, type, &isize);
1483                 if (!buffer)
1484                         return NULL;
1485                 if (!strcmp(type, required_type)) {
1486                         *size = isize;
1487                         if (actual_sha1_return)
1488                                 hashcpy(actual_sha1_return, actual_sha1);
1489                         return buffer;
1490                 }
1491                 /* Handle references */
1492                 else if (!strcmp(type, commit_type))
1493                         ref_type = "tree ";
1494                 else if (!strcmp(type, tag_type))
1495                         ref_type = "object ";
1496                 else {
1497                         free(buffer);
1498                         return NULL;
1499                 }
1500                 ref_length = strlen(ref_type);
1501
1502                 if (memcmp(buffer, ref_type, ref_length) ||
1503                     get_sha1_hex((char *) buffer + ref_length, actual_sha1)) {
1504                         free(buffer);
1505                         return NULL;
1506                 }
1507                 free(buffer);
1508                 /* Now we have the ID of the referred-to object in
1509                  * actual_sha1.  Check again. */
1510         }
1511 }
1512
1513 static void write_sha1_file_prepare(void *buf, unsigned long len,
1514                                     const char *type, unsigned char *sha1,
1515                                     unsigned char *hdr, int *hdrlen)
1516 {
1517         SHA_CTX c;
1518
1519         /* Generate the header */
1520         *hdrlen = sprintf((char *)hdr, "%s %lu", type, len)+1;
1521
1522         /* Sha1.. */
1523         SHA1_Init(&c);
1524         SHA1_Update(&c, hdr, *hdrlen);
1525         SHA1_Update(&c, buf, len);
1526         SHA1_Final(sha1, &c);
1527 }
1528
1529 /*
1530  * Link the tempfile to the final place, possibly creating the
1531  * last directory level as you do so.
1532  *
1533  * Returns the errno on failure, 0 on success.
1534  */
1535 static int link_temp_to_file(const char *tmpfile, const char *filename)
1536 {
1537         int ret;
1538         char *dir;
1539
1540         if (!link(tmpfile, filename))
1541                 return 0;
1542
1543         /*
1544          * Try to mkdir the last path component if that failed.
1545          *
1546          * Re-try the "link()" regardless of whether the mkdir
1547          * succeeds, since a race might mean that somebody
1548          * else succeeded.
1549          */
1550         ret = errno;
1551         dir = strrchr(filename, '/');
1552         if (dir) {
1553                 *dir = 0;
1554                 if (!mkdir(filename, 0777) && adjust_shared_perm(filename)) {
1555                         *dir = '/';
1556                         return -2;
1557                 }
1558                 *dir = '/';
1559                 if (!link(tmpfile, filename))
1560                         return 0;
1561                 ret = errno;
1562         }
1563         return ret;
1564 }
1565
1566 /*
1567  * Move the just written object into its final resting place
1568  */
1569 int move_temp_to_file(const char *tmpfile, const char *filename)
1570 {
1571         int ret = link_temp_to_file(tmpfile, filename);
1572
1573         /*
1574          * Coda hack - coda doesn't like cross-directory links,
1575          * so we fall back to a rename, which will mean that it
1576          * won't be able to check collisions, but that's not a
1577          * big deal.
1578          *
1579          * The same holds for FAT formatted media.
1580          *
1581          * When this succeeds, we just return 0. We have nothing
1582          * left to unlink.
1583          */
1584         if (ret && ret != EEXIST) {
1585                 if (!rename(tmpfile, filename))
1586                         return 0;
1587                 ret = errno;
1588         }
1589         unlink(tmpfile);
1590         if (ret) {
1591                 if (ret != EEXIST) {
1592                         return error("unable to write sha1 filename %s: %s\n", filename, strerror(ret));
1593                 }
1594                 /* FIXME!!! Collision check here ? */
1595         }
1596
1597         return 0;
1598 }
1599
1600 static int write_buffer(int fd, const void *buf, size_t len)
1601 {
1602         while (len) {
1603                 ssize_t size;
1604
1605                 size = write(fd, buf, len);
1606                 if (!size)
1607                         return error("file write: disk full");
1608                 if (size < 0) {
1609                         if (errno == EINTR || errno == EAGAIN)
1610                                 continue;
1611                         return error("file write error (%s)", strerror(errno));
1612                 }
1613                 len -= size;
1614                 buf = (char *) buf + size;
1615         }
1616         return 0;
1617 }
1618
1619 static int write_binary_header(unsigned char *hdr, enum object_type type, unsigned long len)
1620 {
1621         int hdr_len;
1622         unsigned char c;
1623
1624         c = (type << 4) | (len & 15);
1625         len >>= 4;
1626         hdr_len = 1;
1627         while (len) {
1628                 *hdr++ = c | 0x80;
1629                 hdr_len++;
1630                 c = (len & 0x7f);
1631                 len >>= 7;
1632         }
1633         *hdr = c;
1634         return hdr_len;
1635 }
1636
1637 static void setup_object_header(z_stream *stream, const char *type, unsigned long len)
1638 {
1639         int obj_type, hdr;
1640
1641         if (use_legacy_headers) {
1642                 while (deflate(stream, 0) == Z_OK)
1643                         /* nothing */;
1644                 return;
1645         }
1646         if (!strcmp(type, blob_type))
1647                 obj_type = OBJ_BLOB;
1648         else if (!strcmp(type, tree_type))
1649                 obj_type = OBJ_TREE;
1650         else if (!strcmp(type, commit_type))
1651                 obj_type = OBJ_COMMIT;
1652         else if (!strcmp(type, tag_type))
1653                 obj_type = OBJ_TAG;
1654         else
1655                 die("trying to generate bogus object of type '%s'", type);
1656         hdr = write_binary_header(stream->next_out, obj_type, len);
1657         stream->total_out = hdr;
1658         stream->next_out += hdr;
1659         stream->avail_out -= hdr;
1660 }
1661
1662 int hash_sha1_file(void *buf, unsigned long len, const char *type,
1663                    unsigned char *sha1)
1664 {
1665         unsigned char hdr[50];
1666         int hdrlen;
1667         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1668         return 0;
1669 }
1670
1671 int write_sha1_file(void *buf, unsigned long len, const char *type, unsigned char *returnsha1)
1672 {
1673         int size;
1674         unsigned char *compressed;
1675         z_stream stream;
1676         unsigned char sha1[20];
1677         char *filename;
1678         static char tmpfile[PATH_MAX];
1679         unsigned char hdr[50];
1680         int fd, hdrlen;
1681
1682         /* Normally if we have it in the pack then we do not bother writing
1683          * it out into .git/objects/??/?{38} file.
1684          */
1685         write_sha1_file_prepare(buf, len, type, sha1, hdr, &hdrlen);
1686         filename = sha1_file_name(sha1);
1687         if (returnsha1)
1688                 hashcpy(returnsha1, sha1);
1689         if (has_sha1_file(sha1))
1690                 return 0;
1691         fd = open(filename, O_RDONLY);
1692         if (fd >= 0) {
1693                 /*
1694                  * FIXME!!! We might do collision checking here, but we'd
1695                  * need to uncompress the old file and check it. Later.
1696                  */
1697                 close(fd);
1698                 return 0;
1699         }
1700
1701         if (errno != ENOENT) {
1702                 return error("sha1 file %s: %s\n", filename, strerror(errno));
1703         }
1704
1705         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1706
1707         fd = mkstemp(tmpfile);
1708         if (fd < 0) {
1709                 if (errno == EPERM)
1710                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1711                 else
1712                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1713         }
1714
1715         /* Set it up */
1716         memset(&stream, 0, sizeof(stream));
1717         deflateInit(&stream, zlib_compression_level);
1718         size = 8 + deflateBound(&stream, len+hdrlen);
1719         compressed = xmalloc(size);
1720
1721         /* Compress it */
1722         stream.next_out = compressed;
1723         stream.avail_out = size;
1724
1725         /* First header.. */
1726         stream.next_in = hdr;
1727         stream.avail_in = hdrlen;
1728         setup_object_header(&stream, type, len);
1729
1730         /* Then the data itself.. */
1731         stream.next_in = buf;
1732         stream.avail_in = len;
1733         while (deflate(&stream, Z_FINISH) == Z_OK)
1734                 /* nothing */;
1735         deflateEnd(&stream);
1736         size = stream.total_out;
1737
1738         if (write_buffer(fd, compressed, size) < 0)
1739                 die("unable to write sha1 file");
1740         fchmod(fd, 0444);
1741         close(fd);
1742         free(compressed);
1743
1744         return move_temp_to_file(tmpfile, filename);
1745 }
1746
1747 /*
1748  * We need to unpack and recompress the object for writing
1749  * it out to a different file.
1750  */
1751 static void *repack_object(const unsigned char *sha1, unsigned long *objsize)
1752 {
1753         size_t size;
1754         z_stream stream;
1755         unsigned char *unpacked;
1756         unsigned long len;
1757         char type[20];
1758         char hdr[50];
1759         int hdrlen;
1760         void *buf;
1761
1762         /* need to unpack and recompress it by itself */
1763         unpacked = read_packed_sha1(sha1, type, &len);
1764
1765         hdrlen = sprintf(hdr, "%s %lu", type, len) + 1;
1766
1767         /* Set it up */
1768         memset(&stream, 0, sizeof(stream));
1769         deflateInit(&stream, zlib_compression_level);
1770         size = deflateBound(&stream, len + hdrlen);
1771         buf = xmalloc(size);
1772
1773         /* Compress it */
1774         stream.next_out = buf;
1775         stream.avail_out = size;
1776
1777         /* First header.. */
1778         stream.next_in = (void *)hdr;
1779         stream.avail_in = hdrlen;
1780         while (deflate(&stream, 0) == Z_OK)
1781                 /* nothing */;
1782
1783         /* Then the data itself.. */
1784         stream.next_in = unpacked;
1785         stream.avail_in = len;
1786         while (deflate(&stream, Z_FINISH) == Z_OK)
1787                 /* nothing */;
1788         deflateEnd(&stream);
1789         free(unpacked);
1790
1791         *objsize = stream.total_out;
1792         return buf;
1793 }
1794
1795 int write_sha1_to_fd(int fd, const unsigned char *sha1)
1796 {
1797         int retval;
1798         unsigned long objsize;
1799         void *buf = map_sha1_file(sha1, &objsize);
1800
1801         if (buf) {
1802                 retval = write_buffer(fd, buf, objsize);
1803                 munmap(buf, objsize);
1804                 return retval;
1805         }
1806
1807         buf = repack_object(sha1, &objsize);
1808         retval = write_buffer(fd, buf, objsize);
1809         free(buf);
1810         return retval;
1811 }
1812
1813 int write_sha1_from_fd(const unsigned char *sha1, int fd, char *buffer,
1814                        size_t bufsize, size_t *bufposn)
1815 {
1816         char tmpfile[PATH_MAX];
1817         int local;
1818         z_stream stream;
1819         unsigned char real_sha1[20];
1820         unsigned char discard[4096];
1821         int ret;
1822         SHA_CTX c;
1823
1824         snprintf(tmpfile, sizeof(tmpfile), "%s/obj_XXXXXX", get_object_directory());
1825
1826         local = mkstemp(tmpfile);
1827         if (local < 0) {
1828                 if (errno == EPERM)
1829                         return error("insufficient permission for adding an object to repository database %s\n", get_object_directory());
1830                 else
1831                         return error("unable to create temporary sha1 filename %s: %s\n", tmpfile, strerror(errno));
1832         }
1833
1834         memset(&stream, 0, sizeof(stream));
1835
1836         inflateInit(&stream);
1837
1838         SHA1_Init(&c);
1839
1840         do {
1841                 ssize_t size;
1842                 if (*bufposn) {
1843                         stream.avail_in = *bufposn;
1844                         stream.next_in = (unsigned char *) buffer;
1845                         do {
1846                                 stream.next_out = discard;
1847                                 stream.avail_out = sizeof(discard);
1848                                 ret = inflate(&stream, Z_SYNC_FLUSH);
1849                                 SHA1_Update(&c, discard, sizeof(discard) -
1850                                             stream.avail_out);
1851                         } while (stream.avail_in && ret == Z_OK);
1852                         if (write_buffer(local, buffer, *bufposn - stream.avail_in) < 0)
1853                                 die("unable to write sha1 file");
1854                         memmove(buffer, buffer + *bufposn - stream.avail_in,
1855                                 stream.avail_in);
1856                         *bufposn = stream.avail_in;
1857                         if (ret != Z_OK)
1858                                 break;
1859                 }
1860                 size = read(fd, buffer + *bufposn, bufsize - *bufposn);
1861                 if (size <= 0) {
1862                         close(local);
1863                         unlink(tmpfile);
1864                         if (!size)
1865                                 return error("Connection closed?");
1866                         perror("Reading from connection");
1867                         return -1;
1868                 }
1869                 *bufposn += size;
1870         } while (1);
1871         inflateEnd(&stream);
1872
1873         close(local);
1874         SHA1_Final(real_sha1, &c);
1875         if (ret != Z_STREAM_END) {
1876                 unlink(tmpfile);
1877                 return error("File %s corrupted", sha1_to_hex(sha1));
1878         }
1879         if (hashcmp(sha1, real_sha1)) {
1880                 unlink(tmpfile);
1881                 return error("File %s has bad hash", sha1_to_hex(sha1));
1882         }
1883
1884         return move_temp_to_file(tmpfile, sha1_file_name(sha1));
1885 }
1886
1887 int has_pack_index(const unsigned char *sha1)
1888 {
1889         struct stat st;
1890         if (stat(sha1_pack_index_name(sha1), &st))
1891                 return 0;
1892         return 1;
1893 }
1894
1895 int has_pack_file(const unsigned char *sha1)
1896 {
1897         struct stat st;
1898         if (stat(sha1_pack_name(sha1), &st))
1899                 return 0;
1900         return 1;
1901 }
1902
1903 int has_sha1_pack(const unsigned char *sha1, const char **ignore_packed)
1904 {
1905         struct pack_entry e;
1906         return find_pack_entry(sha1, &e, ignore_packed);
1907 }
1908
1909 int has_sha1_file(const unsigned char *sha1)
1910 {
1911         struct stat st;
1912         struct pack_entry e;
1913
1914         if (find_pack_entry(sha1, &e, NULL))
1915                 return 1;
1916         return find_sha1_file(sha1, &st) ? 1 : 0;
1917 }
1918
1919 /*
1920  * reads from fd as long as possible into a supplied buffer of size bytes.
1921  * If necessary the buffer's size is increased using realloc()
1922  *
1923  * returns 0 if anything went fine and -1 otherwise
1924  *
1925  * NOTE: both buf and size may change, but even when -1 is returned
1926  * you still have to free() it yourself.
1927  */
1928 int read_pipe(int fd, char** return_buf, unsigned long* return_size)
1929 {
1930         char* buf = *return_buf;
1931         unsigned long size = *return_size;
1932         int iret;
1933         unsigned long off = 0;
1934
1935         do {
1936                 iret = xread(fd, buf + off, size - off);
1937                 if (iret > 0) {
1938                         off += iret;
1939                         if (off == size) {
1940                                 size *= 2;
1941                                 buf = xrealloc(buf, size);
1942                         }
1943                 }
1944         } while (iret > 0);
1945
1946         *return_buf = buf;
1947         *return_size = off;
1948
1949         if (iret < 0)
1950                 return -1;
1951         return 0;
1952 }
1953
1954 int index_pipe(unsigned char *sha1, int fd, const char *type, int write_object)
1955 {
1956         unsigned long size = 4096;
1957         char *buf = xmalloc(size);
1958         int ret;
1959
1960         if (read_pipe(fd, &buf, &size)) {
1961                 free(buf);
1962                 return -1;
1963         }
1964
1965         if (!type)
1966                 type = blob_type;
1967         if (write_object)
1968                 ret = write_sha1_file(buf, size, type, sha1);
1969         else
1970                 ret = hash_sha1_file(buf, size, type, sha1);
1971         free(buf);
1972         return ret;
1973 }
1974
1975 int index_fd(unsigned char *sha1, int fd, struct stat *st, int write_object, const char *type)
1976 {
1977         unsigned long size = st->st_size;
1978         void *buf;
1979         int ret;
1980
1981         buf = "";
1982         if (size)
1983                 buf = mmap(NULL, size, PROT_READ, MAP_PRIVATE, fd, 0);
1984         close(fd);
1985         if (buf == MAP_FAILED)
1986                 return -1;
1987
1988         if (!type)
1989                 type = blob_type;
1990         if (write_object)
1991                 ret = write_sha1_file(buf, size, type, sha1);
1992         else
1993                 ret = hash_sha1_file(buf, size, type, sha1);
1994         if (size)
1995                 munmap(buf, size);
1996         return ret;
1997 }
1998
1999 int index_path(unsigned char *sha1, const char *path, struct stat *st, int write_object)
2000 {
2001         int fd;
2002         char *target;
2003
2004         switch (st->st_mode & S_IFMT) {
2005         case S_IFREG:
2006                 fd = open(path, O_RDONLY);
2007                 if (fd < 0)
2008                         return error("open(\"%s\"): %s", path,
2009                                      strerror(errno));
2010                 if (index_fd(sha1, fd, st, write_object, NULL) < 0)
2011                         return error("%s: failed to insert into database",
2012                                      path);
2013                 break;
2014         case S_IFLNK:
2015                 target = xmalloc(st->st_size+1);
2016                 if (readlink(path, target, st->st_size+1) != st->st_size) {
2017                         char *errstr = strerror(errno);
2018                         free(target);
2019                         return error("readlink(\"%s\"): %s", path,
2020                                      errstr);
2021                 }
2022                 if (!write_object)
2023                         hash_sha1_file(target, st->st_size, blob_type, sha1);
2024                 else if (write_sha1_file(target, st->st_size, blob_type, sha1))
2025                         return error("%s: failed to insert into database",
2026                                      path);
2027                 free(target);
2028                 break;
2029         default:
2030                 return error("%s: unsupported file type", path);
2031         }
2032         return 0;
2033 }