fast-import.c

   1 /*
   2 Format of STDIN stream:
   3
   4   stream ::= cmd*;
   5
   6   cmd ::= new_blob
   7         | new_commit
   8         | new_tag
   9         | reset_branch
  10         | checkpoint
  11         ;
  12
  13   new_blob ::= 'blob' lf
  14     mark?
  15     file_content;
  16   file_content ::= data;
  17
  18   new_commit ::= 'commit' sp ref_str lf
  19     mark?
  20     ('author' sp name '<' email '>' when lf)?
  21     'committer' sp name '<' email '>' when lf
  22     commit_msg
  23     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  24     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  25     file_change*
  26     lf;
  27   commit_msg ::= data;
  28
  29   file_change ::= file_clr
  30     | file_del
  31     | file_rnm
  32     | file_cpy
  33     | file_obm
  34     | file_inm;
  35   file_clr ::= 'deleteall' lf;
  36   file_del ::= 'D' sp path_str lf;
  37   file_rnm ::= 'R' sp path_str sp path_str lf;
  38   file_cpy ::= 'C' sp path_str sp path_str lf;
  39   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  40   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  41     data;
  42
  43   new_tag ::= 'tag' sp tag_str lf
  44     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  45     'tagger' sp name '<' email '>' when lf
  46     tag_msg;
  47   tag_msg ::= data;
  48
  49   reset_branch ::= 'reset' sp ref_str lf
  50     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  51     lf;
  52
  53   checkpoint ::= 'checkpoint' lf
  54     lf;
  55
  56      # note: the first idnum in a stream should be 1 and subsequent
  57      # idnums should not have gaps between values as this will cause
  58      # the stream parser to reserve space for the gapped values.  An
  59      # idnum can be updated in the future to a new object by issuing
  60      # a new mark directive with the old idnum.
  61      #
  62   mark ::= 'mark' sp idnum lf;
  63   data ::= (delimited_data | exact_data)
  64     lf?;
  65
  66     # note: delim may be any string but must not contain lf.
  67     # data_line may contain any data but must not be exactly
  68     # delim.
  69   delimited_data ::= 'data' sp '<<' delim lf
  70     (data_line lf)*
  71     delim lf;
  72
  73      # note: declen indicates the length of binary_data in bytes.
  74      # declen does not include the lf preceeding the binary data.
  75      #
  76   exact_data ::= 'data' sp declen lf
  77     binary_data;
  78
  79      # note: quoted strings are C-style quoting supporting \c for
  80      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  81      # is the signed byte value in octal.  Note that the only
  82      # characters which must actually be escaped to protect the
  83      # stream formatting is: \, " and LF.  Otherwise these values
  84      # are UTF8.
  85      #
  86   ref_str     ::= ref;
  87   sha1exp_str ::= sha1exp;
  88   tag_str     ::= tag;
  89   path_str    ::= path    | '"' quoted(path)    '"' ;
  90   mode        ::= '100644' | '644'
  91                 | '100755' | '755'
  92                 | '120000'
  93                 ;
  94
  95   declen ::= # unsigned 32 bit value, ascii base10 notation;
  96   bigint ::= # unsigned integer value, ascii base10 notation;
  97   binary_data ::= # file content, not interpreted;
  98
  99   when         ::= raw_when | rfc2822_when;
 100   raw_when     ::= ts sp tz;
 101   rfc2822_when ::= # Valid RFC 2822 date and time;
 102
 103   sp ::= # ASCII space character;
 104   lf ::= # ASCII newline (LF) character;
 105
 106      # note: a colon (':') must precede the numerical value assigned to
 107      # an idnum.  This is to distinguish it from a ref or tag name as
 108      # GIT does not permit ':' in ref or tag strings.
 109      #
 110   idnum   ::= ':' bigint;
 111   path    ::= # GIT style file path, e.g. "a/b/c";
 112   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 113   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 114   sha1exp ::= # Any valid GIT SHA1 expression;
 115   hexsha1 ::= # SHA1 in hexadecimal format;
 116
 117      # note: name and email are UTF8 strings, however name must not
 118      # contain '<' or lf and email must not contain any of the
 119      # following: '<', '>', lf.
 120      #
 121   name  ::= # valid GIT author/committer name;
 122   email ::= # valid GIT author/committer email;
 123   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 124   tz    ::= # GIT style timezone;
 125
 126      # note: comments may appear anywhere in the input, except
 127      # within a data command.  Any form of the data command
 128      # always escapes the related input from comment processing.
 129      #
 130      # In case it is not clear, the '#' that starts the comment
 131      # must be the first character on that the line (an lf have
 132      # preceeded it).
 133      #
 134   comment ::= '#' not_lf* lf;
 135   not_lf  ::= # Any byte that is not ASCII newline (LF);
 136 */
 137
 138 #include "builtin.h"
 139 #include "cache.h"
 140 #include "object.h"
 141 #include "blob.h"
 142 #include "tree.h"
 143 #include "commit.h"
 144 #include "delta.h"
 145 #include "pack.h"
 146 #include "refs.h"
 147 #include "csum-file.h"
 148 #include "strbuf.h"
 149 #include "quote.h"
 150
 151 #define PACK_ID_BITS 16
 152 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 153
 154 struct object_entry
 155 {
 156         struct object_entry *next;
 157         uint32_t offset;
 158         unsigned type : TYPE_BITS;
 159         unsigned pack_id : PACK_ID_BITS;
 160         unsigned char sha1[20];
 161 };
 162
 163 struct object_entry_pool
 164 {
 165         struct object_entry_pool *next_pool;
 166         struct object_entry *next_free;
 167         struct object_entry *end;
 168         struct object_entry entries[FLEX_ARRAY]; /* more */
 169 };
 170
 171 struct mark_set
 172 {
 173         union {
 174                 struct object_entry *marked[1024];
 175                 struct mark_set *sets[1024];
 176         } data;
 177         unsigned int shift;
 178 };
 179
 180 struct last_object
 181 {
 182         void *data;
 183         unsigned long len;
 184         uint32_t offset;
 185         unsigned int depth;
 186         unsigned no_free:1;
 187 };
 188
 189 struct mem_pool
 190 {
 191         struct mem_pool *next_pool;
 192         char *next_free;
 193         char *end;
 194         char space[FLEX_ARRAY]; /* more */
 195 };
 196
 197 struct atom_str
 198 {
 199         struct atom_str *next_atom;
 200         unsigned short str_len;
 201         char str_dat[FLEX_ARRAY]; /* more */
 202 };
 203
 204 struct tree_content;
 205 struct tree_entry
 206 {
 207         struct tree_content *tree;
 208         struct atom_str* name;
 209         struct tree_entry_ms
 210         {
 211                 uint16_t mode;
 212                 unsigned char sha1[20];
 213         } versions[2];
 214 };
 215
 216 struct tree_content
 217 {
 218         unsigned int entry_capacity; /* must match avail_tree_content */
 219         unsigned int entry_count;
 220         unsigned int delta_depth;
 221         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 222 };
 223
 224 struct avail_tree_content
 225 {
 226         unsigned int entry_capacity; /* must match tree_content */
 227         struct avail_tree_content *next_avail;
 228 };
 229
 230 struct branch
 231 {
 232         struct branch *table_next_branch;
 233         struct branch *active_next_branch;
 234         const char *name;
 235         struct tree_entry branch_tree;
 236         uintmax_t last_commit;
 237         unsigned active : 1;
 238         unsigned pack_id : PACK_ID_BITS;
 239         unsigned char sha1[20];
 240 };
 241
 242 struct tag
 243 {
 244         struct tag *next_tag;
 245         const char *name;
 246         unsigned int pack_id;
 247         unsigned char sha1[20];
 248 };
 249
 250 struct dbuf
 251 {
 252         void *buffer;
 253         size_t capacity;
 254 };
 255
 256 struct hash_list
 257 {
 258         struct hash_list *next;
 259         unsigned char sha1[20];
 260 };
 261
 262 typedef enum {
 263         WHENSPEC_RAW = 1,
 264         WHENSPEC_RFC2822,
 265         WHENSPEC_NOW,
 266 } whenspec_type;
 267
 268 /* Configured limits on output */
 269 static unsigned long max_depth = 10;
 270 static off_t max_packsize = (1LL << 32) - 1;
 271 static int force_update;
 272
 273 /* Stats and misc. counters */
 274 static uintmax_t alloc_count;
 275 static uintmax_t marks_set_count;
 276 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 277 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 278 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 279 static unsigned long object_count;
 280 static unsigned long branch_count;
 281 static unsigned long branch_load_count;
 282 static int failure;
 283 static FILE *pack_edges;
 284
 285 /* Memory pools */
 286 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 287 static size_t total_allocd;
 288 static struct mem_pool *mem_pool;
 289
 290 /* Atom management */
 291 static unsigned int atom_table_sz = 4451;
 292 static unsigned int atom_cnt;
 293 static struct atom_str **atom_table;
 294
 295 /* The .pack file being generated */
 296 static unsigned int pack_id;
 297 static struct packed_git *pack_data;
 298 static struct packed_git **all_packs;
 299 static unsigned long pack_size;
 300
 301 /* Table of objects we've written. */
 302 static unsigned int object_entry_alloc = 5000;
 303 static struct object_entry_pool *blocks;
 304 static struct object_entry *object_table[1 << 16];
 305 static struct mark_set *marks;
 306 static const char* mark_file;
 307
 308 /* Our last blob */
 309 static struct last_object last_blob;
 310
 311 /* Tree management */
 312 static unsigned int tree_entry_alloc = 1000;
 313 static void *avail_tree_entry;
 314 static unsigned int avail_tree_table_sz = 100;
 315 static struct avail_tree_content **avail_tree_table;
 316 static struct dbuf old_tree;
 317 static struct dbuf new_tree;
 318
 319 /* Branch data */
 320 static unsigned long max_active_branches = 5;
 321 static unsigned long cur_active_branches;
 322 static unsigned long branch_table_sz = 1039;
 323 static struct branch **branch_table;
 324 static struct branch *active_branches;
 325
 326 /* Tag data */
 327 static struct tag *first_tag;
 328 static struct tag *last_tag;
 329
 330 /* Input stream parsing */
 331 static whenspec_type whenspec = WHENSPEC_RAW;
 332 static struct strbuf command_buf;
 333 static uintmax_t next_mark;
 334 static struct dbuf new_data;
 335
 336
 337 static void alloc_objects(unsigned int cnt)
 338 {
 339         struct object_entry_pool *b;
 340
 341         b = xmalloc(sizeof(struct object_entry_pool)
 342                 + cnt * sizeof(struct object_entry));
 343         b->next_pool = blocks;
 344         b->next_free = b->entries;
 345         b->end = b->entries + cnt;
 346         blocks = b;
 347         alloc_count += cnt;
 348 }
 349
 350 static struct object_entry *new_object(unsigned char *sha1)
 351 {
 352         struct object_entry *e;
 353
 354         if (blocks->next_free == blocks->end)
 355                 alloc_objects(object_entry_alloc);
 356
 357         e = blocks->next_free++;
 358         hashcpy(e->sha1, sha1);
 359         return e;
 360 }
 361
 362 static struct object_entry *find_object(unsigned char *sha1)
 363 {
 364         unsigned int h = sha1[0] << 8 | sha1[1];
 365         struct object_entry *e;
 366         for (e = object_table[h]; e; e = e->next)
 367                 if (!hashcmp(sha1, e->sha1))
 368                         return e;
 369         return NULL;
 370 }
 371
 372 static struct object_entry *insert_object(unsigned char *sha1)
 373 {
 374         unsigned int h = sha1[0] << 8 | sha1[1];
 375         struct object_entry *e = object_table[h];
 376         struct object_entry *p = NULL;
 377
 378         while (e) {
 379                 if (!hashcmp(sha1, e->sha1))
 380                         return e;
 381                 p = e;
 382                 e = e->next;
 383         }
 384
 385         e = new_object(sha1);
 386         e->next = NULL;
 387         e->offset = 0;
 388         if (p)
 389                 p->next = e;
 390         else
 391                 object_table[h] = e;
 392         return e;
 393 }
 394
 395 static unsigned int hc_str(const char *s, size_t len)
 396 {
 397         unsigned int r = 0;
 398         while (len-- > 0)
 399                 r = r * 31 + *s++;
 400         return r;
 401 }
 402
 403 static void *pool_alloc(size_t len)
 404 {
 405         struct mem_pool *p;
 406         void *r;
 407
 408         for (p = mem_pool; p; p = p->next_pool)
 409                 if ((p->end - p->next_free >= len))
 410                         break;
 411
 412         if (!p) {
 413                 if (len >= (mem_pool_alloc/2)) {
 414                         total_allocd += len;
 415                         return xmalloc(len);
 416                 }
 417                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 418                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 419                 p->next_pool = mem_pool;
 420                 p->next_free = p->space;
 421                 p->end = p->next_free + mem_pool_alloc;
 422                 mem_pool = p;
 423         }
 424
 425         r = p->next_free;
 426         /* round out to a pointer alignment */
 427         if (len & (sizeof(void*) - 1))
 428                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 429         p->next_free += len;
 430         return r;
 431 }
 432
 433 static void *pool_calloc(size_t count, size_t size)
 434 {
 435         size_t len = count * size;
 436         void *r = pool_alloc(len);
 437         memset(r, 0, len);
 438         return r;
 439 }
 440
 441 static char *pool_strdup(const char *s)
 442 {
 443         char *r = pool_alloc(strlen(s) + 1);
 444         strcpy(r, s);
 445         return r;
 446 }
 447
 448 static void size_dbuf(struct dbuf *b, size_t maxlen)
 449 {
 450         if (b->buffer) {
 451                 if (b->capacity >= maxlen)
 452                         return;
 453                 free(b->buffer);
 454         }
 455         b->capacity = ((maxlen / 1024) + 1) * 1024;
 456         b->buffer = xmalloc(b->capacity);
 457 }
 458
 459 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 460 {
 461         struct mark_set *s = marks;
 462         while ((idnum >> s->shift) >= 1024) {
 463                 s = pool_calloc(1, sizeof(struct mark_set));
 464                 s->shift = marks->shift + 10;
 465                 s->data.sets[0] = marks;
 466                 marks = s;
 467         }
 468         while (s->shift) {
 469                 uintmax_t i = idnum >> s->shift;
 470                 idnum -= i << s->shift;
 471                 if (!s->data.sets[i]) {
 472                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 473                         s->data.sets[i]->shift = s->shift - 10;
 474                 }
 475                 s = s->data.sets[i];
 476         }
 477         if (!s->data.marked[idnum])
 478                 marks_set_count++;
 479         s->data.marked[idnum] = oe;
 480 }
 481
 482 static struct object_entry *find_mark(uintmax_t idnum)
 483 {
 484         uintmax_t orig_idnum = idnum;
 485         struct mark_set *s = marks;
 486         struct object_entry *oe = NULL;
 487         if ((idnum >> s->shift) < 1024) {
 488                 while (s && s->shift) {
 489                         uintmax_t i = idnum >> s->shift;
 490                         idnum -= i << s->shift;
 491                         s = s->data.sets[i];
 492                 }
 493                 if (s)
 494                         oe = s->data.marked[idnum];
 495         }
 496         if (!oe)
 497                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 498         return oe;
 499 }
 500
 501 static struct atom_str *to_atom(const char *s, unsigned short len)
 502 {
 503         unsigned int hc = hc_str(s, len) % atom_table_sz;
 504         struct atom_str *c;
 505
 506         for (c = atom_table[hc]; c; c = c->next_atom)
 507                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 508                         return c;
 509
 510         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 511         c->str_len = len;
 512         strncpy(c->str_dat, s, len);
 513         c->str_dat[len] = 0;
 514         c->next_atom = atom_table[hc];
 515         atom_table[hc] = c;
 516         atom_cnt++;
 517         return c;
 518 }
 519
 520 static struct branch *lookup_branch(const char *name)
 521 {
 522         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 523         struct branch *b;
 524
 525         for (b = branch_table[hc]; b; b = b->table_next_branch)
 526                 if (!strcmp(name, b->name))
 527                         return b;
 528         return NULL;
 529 }
 530
 531 static struct branch *new_branch(const char *name)
 532 {
 533         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 534         struct branch* b = lookup_branch(name);
 535
 536         if (b)
 537                 die("Invalid attempt to create duplicate branch: %s", name);
 538         switch (check_ref_format(name)) {
 539         case  0: break; /* its valid */
 540         case -2: break; /* valid, but too few '/', allow anyway */
 541         default:
 542                 die("Branch name doesn't conform to GIT standards: %s", name);
 543         }
 544
 545         b = pool_calloc(1, sizeof(struct branch));
 546         b->name = pool_strdup(name);
 547         b->table_next_branch = branch_table[hc];
 548         b->branch_tree.versions[0].mode = S_IFDIR;
 549         b->branch_tree.versions[1].mode = S_IFDIR;
 550         b->active = 0;
 551         b->pack_id = MAX_PACK_ID;
 552         branch_table[hc] = b;
 553         branch_count++;
 554         return b;
 555 }
 556
 557 static unsigned int hc_entries(unsigned int cnt)
 558 {
 559         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 560         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 561 }
 562
 563 static struct tree_content *new_tree_content(unsigned int cnt)
 564 {
 565         struct avail_tree_content *f, *l = NULL;
 566         struct tree_content *t;
 567         unsigned int hc = hc_entries(cnt);
 568
 569         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 570                 if (f->entry_capacity >= cnt)
 571                         break;
 572
 573         if (f) {
 574                 if (l)
 575                         l->next_avail = f->next_avail;
 576                 else
 577                         avail_tree_table[hc] = f->next_avail;
 578         } else {
 579                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 580                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 581                 f->entry_capacity = cnt;
 582         }
 583
 584         t = (struct tree_content*)f;
 585         t->entry_count = 0;
 586         t->delta_depth = 0;
 587         return t;
 588 }
 589
 590 static void release_tree_entry(struct tree_entry *e);
 591 static void release_tree_content(struct tree_content *t)
 592 {
 593         struct avail_tree_content *f = (struct avail_tree_content*)t;
 594         unsigned int hc = hc_entries(f->entry_capacity);
 595         f->next_avail = avail_tree_table[hc];
 596         avail_tree_table[hc] = f;
 597 }
 598
 599 static void release_tree_content_recursive(struct tree_content *t)
 600 {
 601         unsigned int i;
 602         for (i = 0; i < t->entry_count; i++)
 603                 release_tree_entry(t->entries[i]);
 604         release_tree_content(t);
 605 }
 606
 607 static struct tree_content *grow_tree_content(
 608         struct tree_content *t,
 609         int amt)
 610 {
 611         struct tree_content *r = new_tree_content(t->entry_count + amt);
 612         r->entry_count = t->entry_count;
 613         r->delta_depth = t->delta_depth;
 614         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 615         release_tree_content(t);
 616         return r;
 617 }
 618
 619 static struct tree_entry *new_tree_entry(void)
 620 {
 621         struct tree_entry *e;
 622
 623         if (!avail_tree_entry) {
 624                 unsigned int n = tree_entry_alloc;
 625                 total_allocd += n * sizeof(struct tree_entry);
 626                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 627                 while (n-- > 1) {
 628                         *((void**)e) = e + 1;
 629                         e++;
 630                 }
 631                 *((void**)e) = NULL;
 632         }
 633
 634         e = avail_tree_entry;
 635         avail_tree_entry = *((void**)e);
 636         return e;
 637 }
 638
 639 static void release_tree_entry(struct tree_entry *e)
 640 {
 641         if (e->tree)
 642                 release_tree_content_recursive(e->tree);
 643         *((void**)e) = avail_tree_entry;
 644         avail_tree_entry = e;
 645 }
 646
 647 static struct tree_content *dup_tree_content(struct tree_content *s)
 648 {
 649         struct tree_content *d;
 650         struct tree_entry *a, *b;
 651         unsigned int i;
 652
 653         if (!s)
 654                 return NULL;
 655         d = new_tree_content(s->entry_count);
 656         for (i = 0; i < s->entry_count; i++) {
 657                 a = s->entries[i];
 658                 b = new_tree_entry();
 659                 memcpy(b, a, sizeof(*a));
 660                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 661                         b->tree = dup_tree_content(a->tree);
 662                 else
 663                         b->tree = NULL;
 664                 d->entries[i] = b;
 665         }
 666         d->entry_count = s->entry_count;
 667         d->delta_depth = s->delta_depth;
 668
 669         return d;
 670 }
 671
 672 static void start_packfile(void)
 673 {
 674         static char tmpfile[PATH_MAX];
 675         struct packed_git *p;
 676         struct pack_header hdr;
 677         int pack_fd;
 678
 679         snprintf(tmpfile, sizeof(tmpfile),
 680                 "%s/tmp_pack_XXXXXX", get_object_directory());
 681         pack_fd = xmkstemp(tmpfile);
 682         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 683         strcpy(p->pack_name, tmpfile);
 684         p->pack_fd = pack_fd;
 685
 686         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 687         hdr.hdr_version = htonl(2);
 688         hdr.hdr_entries = 0;
 689         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 690
 691         pack_data = p;
 692         pack_size = sizeof(hdr);
 693         object_count = 0;
 694
 695         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 696         all_packs[pack_id] = p;
 697 }
 698
 699 static int oecmp (const void *a_, const void *b_)
 700 {
 701         struct object_entry *a = *((struct object_entry**)a_);
 702         struct object_entry *b = *((struct object_entry**)b_);
 703         return hashcmp(a->sha1, b->sha1);
 704 }
 705
 706 static char *create_index(void)
 707 {
 708         static char tmpfile[PATH_MAX];
 709         SHA_CTX ctx;
 710         struct sha1file *f;
 711         struct object_entry **idx, **c, **last, *e;
 712         struct object_entry_pool *o;
 713         uint32_t array[256];
 714         int i, idx_fd;
 715
 716         /* Build the sorted table of object IDs. */
 717         idx = xmalloc(object_count * sizeof(struct object_entry*));
 718         c = idx;
 719         for (o = blocks; o; o = o->next_pool)
 720                 for (e = o->next_free; e-- != o->entries;)
 721                         if (pack_id == e->pack_id)
 722                                 *c++ = e;
 723         last = idx + object_count;
 724         if (c != last)
 725                 die("internal consistency error creating the index");
 726         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 727
 728         /* Generate the fan-out array. */
 729         c = idx;
 730         for (i = 0; i < 256; i++) {
 731                 struct object_entry **next = c;;
 732                 while (next < last) {
 733                         if ((*next)->sha1[0] != i)
 734                                 break;
 735                         next++;
 736                 }
 737                 array[i] = htonl(next - idx);
 738                 c = next;
 739         }
 740
 741         snprintf(tmpfile, sizeof(tmpfile),
 742                 "%s/tmp_idx_XXXXXX", get_object_directory());
 743         idx_fd = xmkstemp(tmpfile);
 744         f = sha1fd(idx_fd, tmpfile);
 745         sha1write(f, array, 256 * sizeof(int));
 746         SHA1_Init(&ctx);
 747         for (c = idx; c != last; c++) {
 748                 uint32_t offset = htonl((*c)->offset);
 749                 sha1write(f, &offset, 4);
 750                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 751                 SHA1_Update(&ctx, (*c)->sha1, 20);
 752         }
 753         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 754         sha1close(f, NULL, 1);
 755         free(idx);
 756         SHA1_Final(pack_data->sha1, &ctx);
 757         return tmpfile;
 758 }
 759
 760 static char *keep_pack(char *curr_index_name)
 761 {
 762         static char name[PATH_MAX];
 763         static const char *keep_msg = "fast-import";
 764         int keep_fd;
 765
 766         chmod(pack_data->pack_name, 0444);
 767         chmod(curr_index_name, 0444);
 768
 769         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 770                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 771         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 772         if (keep_fd < 0)
 773                 die("cannot create keep file");
 774         write(keep_fd, keep_msg, strlen(keep_msg));
 775         close(keep_fd);
 776
 777         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 778                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 779         if (move_temp_to_file(pack_data->pack_name, name))
 780                 die("cannot store pack file");
 781
 782         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 783                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 784         if (move_temp_to_file(curr_index_name, name))
 785                 die("cannot store index file");
 786         return name;
 787 }
 788
 789 static void unkeep_all_packs(void)
 790 {
 791         static char name[PATH_MAX];
 792         int k;
 793
 794         for (k = 0; k < pack_id; k++) {
 795                 struct packed_git *p = all_packs[k];
 796                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 797                          get_object_directory(), sha1_to_hex(p->sha1));
 798                 unlink(name);
 799         }
 800 }
 801
 802 static void end_packfile(void)
 803 {
 804         struct packed_git *old_p = pack_data, *new_p;
 805
 806         if (object_count) {
 807                 char *idx_name;
 808                 int i;
 809                 struct branch *b;
 810                 struct tag *t;
 811
 812                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 813                                     pack_data->pack_name, object_count);
 814                 close(pack_data->pack_fd);
 815                 idx_name = keep_pack(create_index());
 816
 817                 /* Register the packfile with core git's machinary. */
 818                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 819                 if (!new_p)
 820                         die("core git rejected index %s", idx_name);
 821                 new_p->windows = old_p->windows;
 822                 all_packs[pack_id] = new_p;
 823                 install_packed_git(new_p);
 824
 825                 /* Print the boundary */
 826                 if (pack_edges) {
 827                         fprintf(pack_edges, "%s:", new_p->pack_name);
 828                         for (i = 0; i < branch_table_sz; i++) {
 829                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 830                                         if (b->pack_id == pack_id)
 831                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 832                                 }
 833                         }
 834                         for (t = first_tag; t; t = t->next_tag) {
 835                                 if (t->pack_id == pack_id)
 836                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 837                         }
 838                         fputc('\n', pack_edges);
 839                         fflush(pack_edges);
 840                 }
 841
 842                 pack_id++;
 843         }
 844         else
 845                 unlink(old_p->pack_name);
 846         free(old_p);
 847
 848         /* We can't carry a delta across packfiles. */
 849         free(last_blob.data);
 850         last_blob.data = NULL;
 851         last_blob.len = 0;
 852         last_blob.offset = 0;
 853         last_blob.depth = 0;
 854 }
 855
 856 static void cycle_packfile(void)
 857 {
 858         end_packfile();
 859         start_packfile();
 860 }
 861
 862 static size_t encode_header(
 863         enum object_type type,
 864         size_t size,
 865         unsigned char *hdr)
 866 {
 867         int n = 1;
 868         unsigned char c;
 869
 870         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 871                 die("bad type %d", type);
 872
 873         c = (type << 4) | (size & 15);
 874         size >>= 4;
 875         while (size) {
 876                 *hdr++ = c | 0x80;
 877                 c = size & 0x7f;
 878                 size >>= 7;
 879                 n++;
 880         }
 881         *hdr = c;
 882         return n;
 883 }
 884
 885 static int store_object(
 886         enum object_type type,
 887         void *dat,
 888         size_t datlen,
 889         struct last_object *last,
 890         unsigned char *sha1out,
 891         uintmax_t mark)
 892 {
 893         void *out, *delta;
 894         struct object_entry *e;
 895         unsigned char hdr[96];
 896         unsigned char sha1[20];
 897         unsigned long hdrlen, deltalen;
 898         SHA_CTX c;
 899         z_stream s;
 900
 901         hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
 902                 (unsigned long)datlen) + 1;
 903         SHA1_Init(&c);
 904         SHA1_Update(&c, hdr, hdrlen);
 905         SHA1_Update(&c, dat, datlen);
 906         SHA1_Final(sha1, &c);
 907         if (sha1out)
 908                 hashcpy(sha1out, sha1);
 909
 910         e = insert_object(sha1);
 911         if (mark)
 912                 insert_mark(mark, e);
 913         if (e->offset) {
 914                 duplicate_count_by_type[type]++;
 915                 return 1;
 916         } else if (find_sha1_pack(sha1, packed_git)) {
 917                 e->type = type;
 918                 e->pack_id = MAX_PACK_ID;
 919                 e->offset = 1; /* just not zero! */
 920                 duplicate_count_by_type[type]++;
 921                 return 1;
 922         }
 923
 924         if (last && last->data && last->depth < max_depth) {
 925                 delta = diff_delta(last->data, last->len,
 926                         dat, datlen,
 927                         &deltalen, 0);
 928                 if (delta && deltalen >= datlen) {
 929                         free(delta);
 930                         delta = NULL;
 931                 }
 932         } else
 933                 delta = NULL;
 934
 935         memset(&s, 0, sizeof(s));
 936         deflateInit(&s, zlib_compression_level);
 937         if (delta) {
 938                 s.next_in = delta;
 939                 s.avail_in = deltalen;
 940         } else {
 941                 s.next_in = dat;
 942                 s.avail_in = datlen;
 943         }
 944         s.avail_out = deflateBound(&s, s.avail_in);
 945         s.next_out = out = xmalloc(s.avail_out);
 946         while (deflate(&s, Z_FINISH) == Z_OK)
 947                 /* nothing */;
 948         deflateEnd(&s);
 949
 950         /* Determine if we should auto-checkpoint. */
 951         if ((pack_size + 60 + s.total_out) > max_packsize
 952                 || (pack_size + 60 + s.total_out) < pack_size) {
 953
 954                 /* This new object needs to *not* have the current pack_id. */
 955                 e->pack_id = pack_id + 1;
 956                 cycle_packfile();
 957
 958                 /* We cannot carry a delta into the new pack. */
 959                 if (delta) {
 960                         free(delta);
 961                         delta = NULL;
 962
 963                         memset(&s, 0, sizeof(s));
 964                         deflateInit(&s, zlib_compression_level);
 965                         s.next_in = dat;
 966                         s.avail_in = datlen;
 967                         s.avail_out = deflateBound(&s, s.avail_in);
 968                         s.next_out = out = xrealloc(out, s.avail_out);
 969                         while (deflate(&s, Z_FINISH) == Z_OK)
 970                                 /* nothing */;
 971                         deflateEnd(&s);
 972                 }
 973         }
 974
 975         e->type = type;
 976         e->pack_id = pack_id;
 977         e->offset = pack_size;
 978         object_count++;
 979         object_count_by_type[type]++;
 980
 981         if (delta) {
 982                 unsigned long ofs = e->offset - last->offset;
 983                 unsigned pos = sizeof(hdr) - 1;
 984
 985                 delta_count_by_type[type]++;
 986                 last->depth++;
 987
 988                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
 989                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
 990                 pack_size += hdrlen;
 991
 992                 hdr[pos] = ofs & 127;
 993                 while (ofs >>= 7)
 994                         hdr[--pos] = 128 | (--ofs & 127);
 995                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
 996                 pack_size += sizeof(hdr) - pos;
 997         } else {
 998                 if (last)
 999                         last->depth = 0;
1000                 hdrlen = encode_header(type, datlen, hdr);
1001                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1002                 pack_size += hdrlen;
1003         }
1004
1005         write_or_die(pack_data->pack_fd, out, s.total_out);
1006         pack_size += s.total_out;
1007
1008         free(out);
1009         free(delta);
1010         if (last) {
1011                 if (!last->no_free)
1012                         free(last->data);
1013                 last->data = dat;
1014                 last->offset = e->offset;
1015                 last->len = datlen;
1016         }
1017         return 0;
1018 }
1019
1020 static void *gfi_unpack_entry(
1021         struct object_entry *oe,
1022         unsigned long *sizep)
1023 {
1024         enum object_type type;
1025         struct packed_git *p = all_packs[oe->pack_id];
1026         if (p == pack_data)
1027                 p->pack_size = pack_size + 20;
1028         return unpack_entry(p, oe->offset, &type, sizep);
1029 }
1030
1031 static const char *get_mode(const char *str, uint16_t *modep)
1032 {
1033         unsigned char c;
1034         uint16_t mode = 0;
1035
1036         while ((c = *str++) != ' ') {
1037                 if (c < '0' || c > '7')
1038                         return NULL;
1039                 mode = (mode << 3) + (c - '0');
1040         }
1041         *modep = mode;
1042         return str;
1043 }
1044
1045 static void load_tree(struct tree_entry *root)
1046 {
1047         unsigned char* sha1 = root->versions[1].sha1;
1048         struct object_entry *myoe;
1049         struct tree_content *t;
1050         unsigned long size;
1051         char *buf;
1052         const char *c;
1053
1054         root->tree = t = new_tree_content(8);
1055         if (is_null_sha1(sha1))
1056                 return;
1057
1058         myoe = find_object(sha1);
1059         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1060                 if (myoe->type != OBJ_TREE)
1061                         die("Not a tree: %s", sha1_to_hex(sha1));
1062                 t->delta_depth = 0;
1063                 buf = gfi_unpack_entry(myoe, &size);
1064         } else {
1065                 enum object_type type;
1066                 buf = read_sha1_file(sha1, &type, &size);
1067                 if (!buf || type != OBJ_TREE)
1068                         die("Can't load tree %s", sha1_to_hex(sha1));
1069         }
1070
1071         c = buf;
1072         while (c != (buf + size)) {
1073                 struct tree_entry *e = new_tree_entry();
1074
1075                 if (t->entry_count == t->entry_capacity)
1076                         root->tree = t = grow_tree_content(t, t->entry_count);
1077                 t->entries[t->entry_count++] = e;
1078
1079                 e->tree = NULL;
1080                 c = get_mode(c, &e->versions[1].mode);
1081                 if (!c)
1082                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1083                 e->versions[0].mode = e->versions[1].mode;
1084                 e->name = to_atom(c, strlen(c));
1085                 c += e->name->str_len + 1;
1086                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1087                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1088                 c += 20;
1089         }
1090         free(buf);
1091 }
1092
1093 static int tecmp0 (const void *_a, const void *_b)
1094 {
1095         struct tree_entry *a = *((struct tree_entry**)_a);
1096         struct tree_entry *b = *((struct tree_entry**)_b);
1097         return base_name_compare(
1098                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1099                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1100 }
1101
1102 static int tecmp1 (const void *_a, const void *_b)
1103 {
1104         struct tree_entry *a = *((struct tree_entry**)_a);
1105         struct tree_entry *b = *((struct tree_entry**)_b);
1106         return base_name_compare(
1107                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1108                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1109 }
1110
1111 static void mktree(struct tree_content *t,
1112         int v,
1113         unsigned long *szp,
1114         struct dbuf *b)
1115 {
1116         size_t maxlen = 0;
1117         unsigned int i;
1118         char *c;
1119
1120         if (!v)
1121                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1122         else
1123                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1124
1125         for (i = 0; i < t->entry_count; i++) {
1126                 if (t->entries[i]->versions[v].mode)
1127                         maxlen += t->entries[i]->name->str_len + 34;
1128         }
1129
1130         size_dbuf(b, maxlen);
1131         c = b->buffer;
1132         for (i = 0; i < t->entry_count; i++) {
1133                 struct tree_entry *e = t->entries[i];
1134                 if (!e->versions[v].mode)
1135                         continue;
1136                 c += sprintf(c, "%o", (unsigned int)e->versions[v].mode);
1137                 *c++ = ' ';
1138                 strcpy(c, e->name->str_dat);
1139                 c += e->name->str_len + 1;
1140                 hashcpy((unsigned char*)c, e->versions[v].sha1);
1141                 c += 20;
1142         }
1143         *szp = c - (char*)b->buffer;
1144 }
1145
1146 static void store_tree(struct tree_entry *root)
1147 {
1148         struct tree_content *t = root->tree;
1149         unsigned int i, j, del;
1150         unsigned long new_len;
1151         struct last_object lo;
1152         struct object_entry *le;
1153
1154         if (!is_null_sha1(root->versions[1].sha1))
1155                 return;
1156
1157         for (i = 0; i < t->entry_count; i++) {
1158                 if (t->entries[i]->tree)
1159                         store_tree(t->entries[i]);
1160         }
1161
1162         le = find_object(root->versions[0].sha1);
1163         if (!S_ISDIR(root->versions[0].mode)
1164                 || !le
1165                 || le->pack_id != pack_id) {
1166                 lo.data = NULL;
1167                 lo.depth = 0;
1168                 lo.no_free = 0;
1169         } else {
1170                 mktree(t, 0, &lo.len, &old_tree);
1171                 lo.data = old_tree.buffer;
1172                 lo.offset = le->offset;
1173                 lo.depth = t->delta_depth;
1174                 lo.no_free = 1;
1175         }
1176
1177         mktree(t, 1, &new_len, &new_tree);
1178         store_object(OBJ_TREE, new_tree.buffer, new_len,
1179                 &lo, root->versions[1].sha1, 0);
1180
1181         t->delta_depth = lo.depth;
1182         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1183                 struct tree_entry *e = t->entries[i];
1184                 if (e->versions[1].mode) {
1185                         e->versions[0].mode = e->versions[1].mode;
1186                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1187                         t->entries[j++] = e;
1188                 } else {
1189                         release_tree_entry(e);
1190                         del++;
1191                 }
1192         }
1193         t->entry_count -= del;
1194 }
1195
1196 static int tree_content_set(
1197         struct tree_entry *root,
1198         const char *p,
1199         const unsigned char *sha1,
1200         const uint16_t mode,
1201         struct tree_content *subtree)
1202 {
1203         struct tree_content *t = root->tree;
1204         const char *slash1;
1205         unsigned int i, n;
1206         struct tree_entry *e;
1207
1208         slash1 = strchr(p, '/');
1209         if (slash1)
1210                 n = slash1 - p;
1211         else
1212                 n = strlen(p);
1213         if (!n)
1214                 die("Empty path component found in input");
1215         if (!slash1 && !S_ISDIR(mode) && subtree)
1216                 die("Non-directories cannot have subtrees");
1217
1218         for (i = 0; i < t->entry_count; i++) {
1219                 e = t->entries[i];
1220                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1221                         if (!slash1) {
1222                                 if (!S_ISDIR(mode)
1223                                                 && e->versions[1].mode == mode
1224                                                 && !hashcmp(e->versions[1].sha1, sha1))
1225                                         return 0;
1226                                 e->versions[1].mode = mode;
1227                                 hashcpy(e->versions[1].sha1, sha1);
1228                                 if (e->tree)
1229                                         release_tree_content_recursive(e->tree);
1230                                 e->tree = subtree;
1231                                 hashclr(root->versions[1].sha1);
1232                                 return 1;
1233                         }
1234                         if (!S_ISDIR(e->versions[1].mode)) {
1235                                 e->tree = new_tree_content(8);
1236                                 e->versions[1].mode = S_IFDIR;
1237                         }
1238                         if (!e->tree)
1239                                 load_tree(e);
1240                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1241                                 hashclr(root->versions[1].sha1);
1242                                 return 1;
1243                         }
1244                         return 0;
1245                 }
1246         }
1247
1248         if (t->entry_count == t->entry_capacity)
1249                 root->tree = t = grow_tree_content(t, t->entry_count);
1250         e = new_tree_entry();
1251         e->name = to_atom(p, n);
1252         e->versions[0].mode = 0;
1253         hashclr(e->versions[0].sha1);
1254         t->entries[t->entry_count++] = e;
1255         if (slash1) {
1256                 e->tree = new_tree_content(8);
1257                 e->versions[1].mode = S_IFDIR;
1258                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1259         } else {
1260                 e->tree = subtree;
1261                 e->versions[1].mode = mode;
1262                 hashcpy(e->versions[1].sha1, sha1);
1263         }
1264         hashclr(root->versions[1].sha1);
1265         return 1;
1266 }
1267
1268 static int tree_content_remove(
1269         struct tree_entry *root,
1270         const char *p,
1271         struct tree_entry *backup_leaf)
1272 {
1273         struct tree_content *t = root->tree;
1274         const char *slash1;
1275         unsigned int i, n;
1276         struct tree_entry *e;
1277
1278         slash1 = strchr(p, '/');
1279         if (slash1)
1280                 n = slash1 - p;
1281         else
1282                 n = strlen(p);
1283
1284         for (i = 0; i < t->entry_count; i++) {
1285                 e = t->entries[i];
1286                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1287                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1288                                 goto del_entry;
1289                         if (!e->tree)
1290                                 load_tree(e);
1291                         if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1292                                 for (n = 0; n < e->tree->entry_count; n++) {
1293                                         if (e->tree->entries[n]->versions[1].mode) {
1294                                                 hashclr(root->versions[1].sha1);
1295                                                 return 1;
1296                                         }
1297                                 }
1298                                 backup_leaf = NULL;
1299                                 goto del_entry;
1300                         }
1301                         return 0;
1302                 }
1303         }
1304         return 0;
1305
1306 del_entry:
1307         if (backup_leaf)
1308                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1309         else if (e->tree)
1310                 release_tree_content_recursive(e->tree);
1311         e->tree = NULL;
1312         e->versions[1].mode = 0;
1313         hashclr(e->versions[1].sha1);
1314         hashclr(root->versions[1].sha1);
1315         return 1;
1316 }
1317
1318 static int tree_content_get(
1319         struct tree_entry *root,
1320         const char *p,
1321         struct tree_entry *leaf)
1322 {
1323         struct tree_content *t = root->tree;
1324         const char *slash1;
1325         unsigned int i, n;
1326         struct tree_entry *e;
1327
1328         slash1 = strchr(p, '/');
1329         if (slash1)
1330                 n = slash1 - p;
1331         else
1332                 n = strlen(p);
1333
1334         for (i = 0; i < t->entry_count; i++) {
1335                 e = t->entries[i];
1336                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1337                         if (!slash1) {
1338                                 memcpy(leaf, e, sizeof(*leaf));
1339                                 if (e->tree && is_null_sha1(e->versions[1].sha1))
1340                                         leaf->tree = dup_tree_content(e->tree);
1341                                 else
1342                                         leaf->tree = NULL;
1343                                 return 1;
1344                         }
1345                         if (!S_ISDIR(e->versions[1].mode))
1346                                 return 0;
1347                         if (!e->tree)
1348                                 load_tree(e);
1349                         return tree_content_get(e, slash1 + 1, leaf);
1350                 }
1351         }
1352         return 0;
1353 }
1354
1355 static int update_branch(struct branch *b)
1356 {
1357         static const char *msg = "fast-import";
1358         struct ref_lock *lock;
1359         unsigned char old_sha1[20];
1360
1361         if (read_ref(b->name, old_sha1))
1362                 hashclr(old_sha1);
1363         lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1364         if (!lock)
1365                 return error("Unable to lock %s", b->name);
1366         if (!force_update && !is_null_sha1(old_sha1)) {
1367                 struct commit *old_cmit, *new_cmit;
1368
1369                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1370                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1371                 if (!old_cmit || !new_cmit) {
1372                         unlock_ref(lock);
1373                         return error("Branch %s is missing commits.", b->name);
1374                 }
1375
1376                 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1377                         unlock_ref(lock);
1378                         warning("Not updating %s"
1379                                 " (new tip %s does not contain %s)",
1380                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1381                         return -1;
1382                 }
1383         }
1384         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1385                 return error("Unable to update %s", b->name);
1386         return 0;
1387 }
1388
1389 static void dump_branches(void)
1390 {
1391         unsigned int i;
1392         struct branch *b;
1393
1394         for (i = 0; i < branch_table_sz; i++) {
1395                 for (b = branch_table[i]; b; b = b->table_next_branch)
1396                         failure |= update_branch(b);
1397         }
1398 }
1399
1400 static void dump_tags(void)
1401 {
1402         static const char *msg = "fast-import";
1403         struct tag *t;
1404         struct ref_lock *lock;
1405         char ref_name[PATH_MAX];
1406
1407         for (t = first_tag; t; t = t->next_tag) {
1408                 sprintf(ref_name, "tags/%s", t->name);
1409                 lock = lock_ref_sha1(ref_name, NULL);
1410                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1411                         failure |= error("Unable to update %s", ref_name);
1412         }
1413 }
1414
1415 static void dump_marks_helper(FILE *f,
1416         uintmax_t base,
1417         struct mark_set *m)
1418 {
1419         uintmax_t k;
1420         if (m->shift) {
1421                 for (k = 0; k < 1024; k++) {
1422                         if (m->data.sets[k])
1423                                 dump_marks_helper(f, (base + k) << m->shift,
1424                                         m->data.sets[k]);
1425                 }
1426         } else {
1427                 for (k = 0; k < 1024; k++) {
1428                         if (m->data.marked[k])
1429                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1430                                         sha1_to_hex(m->data.marked[k]->sha1));
1431                 }
1432         }
1433 }
1434
1435 static void dump_marks(void)
1436 {
1437         static struct lock_file mark_lock;
1438         int mark_fd;
1439         FILE *f;
1440
1441         if (!mark_file)
1442                 return;
1443
1444         mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1445         if (mark_fd < 0) {
1446                 failure |= error("Unable to write marks file %s: %s",
1447                         mark_file, strerror(errno));
1448                 return;
1449         }
1450
1451         f = fdopen(mark_fd, "w");
1452         if (!f) {
1453                 rollback_lock_file(&mark_lock);
1454                 failure |= error("Unable to write marks file %s: %s",
1455                         mark_file, strerror(errno));
1456                 return;
1457         }
1458
1459         dump_marks_helper(f, 0, marks);
1460         fclose(f);
1461         if (commit_lock_file(&mark_lock))
1462                 failure |= error("Unable to write marks file %s: %s",
1463                         mark_file, strerror(errno));
1464 }
1465
1466 static void read_next_command(void)
1467 {
1468         do {
1469                 read_line(&command_buf, stdin, '\n');
1470         } while (!command_buf.eof && command_buf.buf[0] == '#');
1471 }
1472
1473 static void skip_optional_lf()
1474 {
1475         int term_char = fgetc(stdin);
1476         if (term_char != '\n' && term_char != EOF)
1477                 ungetc(term_char, stdin);
1478 }
1479
1480 static void cmd_mark(void)
1481 {
1482         if (!prefixcmp(command_buf.buf, "mark :")) {
1483                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1484                 read_next_command();
1485         }
1486         else
1487                 next_mark = 0;
1488 }
1489
1490 static void *cmd_data (size_t *size)
1491 {
1492         size_t length;
1493         char *buffer;
1494
1495         if (prefixcmp(command_buf.buf, "data "))
1496                 die("Expected 'data n' command, found: %s", command_buf.buf);
1497
1498         if (!prefixcmp(command_buf.buf + 5, "<<")) {
1499                 char *term = xstrdup(command_buf.buf + 5 + 2);
1500                 size_t sz = 8192, term_len = command_buf.len - 5 - 2;
1501                 length = 0;
1502                 buffer = xmalloc(sz);
1503                 for (;;) {
1504                         read_line(&command_buf, stdin, '\n');
1505                         if (command_buf.eof)
1506                                 die("EOF in data (terminator '%s' not found)", term);
1507                         if (term_len == command_buf.len
1508                                 && !strcmp(term, command_buf.buf))
1509                                 break;
1510                         ALLOC_GROW(buffer, length + command_buf.len, sz);
1511                         memcpy(buffer + length,
1512                                 command_buf.buf,
1513                                 command_buf.len - 1);
1514                         length += command_buf.len - 1;
1515                         buffer[length++] = '\n';
1516                 }
1517                 free(term);
1518         }
1519         else {
1520                 size_t n = 0;
1521                 length = strtoul(command_buf.buf + 5, NULL, 10);
1522                 buffer = xmalloc(length);
1523                 while (n < length) {
1524                         size_t s = fread(buffer + n, 1, length - n, stdin);
1525                         if (!s && feof(stdin))
1526                                 die("EOF in data (%lu bytes remaining)",
1527                                         (unsigned long)(length - n));
1528                         n += s;
1529                 }
1530         }
1531
1532         skip_optional_lf();
1533         *size = length;
1534         return buffer;
1535 }
1536
1537 static int validate_raw_date(const char *src, char *result, int maxlen)
1538 {
1539         const char *orig_src = src;
1540         char *endp, sign;
1541
1542         strtoul(src, &endp, 10);
1543         if (endp == src || *endp != ' ')
1544                 return -1;
1545
1546         src = endp + 1;
1547         if (*src != '-' && *src != '+')
1548                 return -1;
1549         sign = *src;
1550
1551         strtoul(src + 1, &endp, 10);
1552         if (endp == src || *endp || (endp - orig_src) >= maxlen)
1553                 return -1;
1554
1555         strcpy(result, orig_src);
1556         return 0;
1557 }
1558
1559 static char *parse_ident(const char *buf)
1560 {
1561         const char *gt;
1562         size_t name_len;
1563         char *ident;
1564
1565         gt = strrchr(buf, '>');
1566         if (!gt)
1567                 die("Missing > in ident string: %s", buf);
1568         gt++;
1569         if (*gt != ' ')
1570                 die("Missing space after > in ident string: %s", buf);
1571         gt++;
1572         name_len = gt - buf;
1573         ident = xmalloc(name_len + 24);
1574         strncpy(ident, buf, name_len);
1575
1576         switch (whenspec) {
1577         case WHENSPEC_RAW:
1578                 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1579                         die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1580                 break;
1581         case WHENSPEC_RFC2822:
1582                 if (parse_date(gt, ident + name_len, 24) < 0)
1583                         die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1584                 break;
1585         case WHENSPEC_NOW:
1586                 if (strcmp("now", gt))
1587                         die("Date in ident must be 'now': %s", buf);
1588                 datestamp(ident + name_len, 24);
1589                 break;
1590         }
1591
1592         return ident;
1593 }
1594
1595 static void cmd_new_blob(void)
1596 {
1597         size_t l;
1598         void *d;
1599
1600         read_next_command();
1601         cmd_mark();
1602         d = cmd_data(&l);
1603
1604         if (store_object(OBJ_BLOB, d, l, &last_blob, NULL, next_mark))
1605                 free(d);
1606 }
1607
1608 static void unload_one_branch(void)
1609 {
1610         while (cur_active_branches
1611                 && cur_active_branches >= max_active_branches) {
1612                 uintmax_t min_commit = ULONG_MAX;
1613                 struct branch *e, *l = NULL, *p = NULL;
1614
1615                 for (e = active_branches; e; e = e->active_next_branch) {
1616                         if (e->last_commit < min_commit) {
1617                                 p = l;
1618                                 min_commit = e->last_commit;
1619                         }
1620                         l = e;
1621                 }
1622
1623                 if (p) {
1624                         e = p->active_next_branch;
1625                         p->active_next_branch = e->active_next_branch;
1626                 } else {
1627                         e = active_branches;
1628                         active_branches = e->active_next_branch;
1629                 }
1630                 e->active = 0;
1631                 e->active_next_branch = NULL;
1632                 if (e->branch_tree.tree) {
1633                         release_tree_content_recursive(e->branch_tree.tree);
1634                         e->branch_tree.tree = NULL;
1635                 }
1636                 cur_active_branches--;
1637         }
1638 }
1639
1640 static void load_branch(struct branch *b)
1641 {
1642         load_tree(&b->branch_tree);
1643         if (!b->active) {
1644                 b->active = 1;
1645                 b->active_next_branch = active_branches;
1646                 active_branches = b;
1647                 cur_active_branches++;
1648                 branch_load_count++;
1649         }
1650 }
1651
1652 static void file_change_m(struct branch *b)
1653 {
1654         const char *p = command_buf.buf + 2;
1655         char *p_uq;
1656         const char *endp;
1657         struct object_entry *oe = oe;
1658         unsigned char sha1[20];
1659         uint16_t mode, inline_data = 0;
1660
1661         p = get_mode(p, &mode);
1662         if (!p)
1663                 die("Corrupt mode: %s", command_buf.buf);
1664         switch (mode) {
1665         case S_IFREG | 0644:
1666         case S_IFREG | 0755:
1667         case S_IFLNK:
1668         case 0644:
1669         case 0755:
1670                 /* ok */
1671                 break;
1672         default:
1673                 die("Corrupt mode: %s", command_buf.buf);
1674         }
1675
1676         if (*p == ':') {
1677                 char *x;
1678                 oe = find_mark(strtoumax(p + 1, &x, 10));
1679                 hashcpy(sha1, oe->sha1);
1680                 p = x;
1681         } else if (!prefixcmp(p, "inline")) {
1682                 inline_data = 1;
1683                 p += 6;
1684         } else {
1685                 if (get_sha1_hex(p, sha1))
1686                         die("Invalid SHA1: %s", command_buf.buf);
1687                 oe = find_object(sha1);
1688                 p += 40;
1689         }
1690         if (*p++ != ' ')
1691                 die("Missing space after SHA1: %s", command_buf.buf);
1692
1693         p_uq = unquote_c_style(p, &endp);
1694         if (p_uq) {
1695                 if (*endp)
1696                         die("Garbage after path in: %s", command_buf.buf);
1697                 p = p_uq;
1698         }
1699
1700         if (inline_data) {
1701                 size_t l;
1702                 void *d;
1703                 if (!p_uq)
1704                         p = p_uq = xstrdup(p);
1705                 read_next_command();
1706                 d = cmd_data(&l);
1707                 if (store_object(OBJ_BLOB, d, l, &last_blob, sha1, 0))
1708                         free(d);
1709         } else if (oe) {
1710                 if (oe->type != OBJ_BLOB)
1711                         die("Not a blob (actually a %s): %s",
1712                                 command_buf.buf, typename(oe->type));
1713         } else {
1714                 enum object_type type = sha1_object_info(sha1, NULL);
1715                 if (type < 0)
1716                         die("Blob not found: %s", command_buf.buf);
1717                 if (type != OBJ_BLOB)
1718                         die("Not a blob (actually a %s): %s",
1719                             typename(type), command_buf.buf);
1720         }
1721
1722         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1723         free(p_uq);
1724 }
1725
1726 static void file_change_d(struct branch *b)
1727 {
1728         const char *p = command_buf.buf + 2;
1729         char *p_uq;
1730         const char *endp;
1731
1732         p_uq = unquote_c_style(p, &endp);
1733         if (p_uq) {
1734                 if (*endp)
1735                         die("Garbage after path in: %s", command_buf.buf);
1736                 p = p_uq;
1737         }
1738         tree_content_remove(&b->branch_tree, p, NULL);
1739         free(p_uq);
1740 }
1741
1742 static void file_change_cr(struct branch *b, int rename)
1743 {
1744         const char *s, *d;
1745         char *s_uq, *d_uq;
1746         const char *endp;
1747         struct tree_entry leaf;
1748
1749         s = command_buf.buf + 2;
1750         s_uq = unquote_c_style(s, &endp);
1751         if (s_uq) {
1752                 if (*endp != ' ')
1753                         die("Missing space after source: %s", command_buf.buf);
1754         }
1755         else {
1756                 endp = strchr(s, ' ');
1757                 if (!endp)
1758                         die("Missing space after source: %s", command_buf.buf);
1759                 s_uq = xmalloc(endp - s + 1);
1760                 memcpy(s_uq, s, endp - s);
1761                 s_uq[endp - s] = 0;
1762         }
1763         s = s_uq;
1764
1765         endp++;
1766         if (!*endp)
1767                 die("Missing dest: %s", command_buf.buf);
1768
1769         d = endp;
1770         d_uq = unquote_c_style(d, &endp);
1771         if (d_uq) {
1772                 if (*endp)
1773                         die("Garbage after dest in: %s", command_buf.buf);
1774                 d = d_uq;
1775         }
1776
1777         memset(&leaf, 0, sizeof(leaf));
1778         if (rename)
1779                 tree_content_remove(&b->branch_tree, s, &leaf);
1780         else
1781                 tree_content_get(&b->branch_tree, s, &leaf);
1782         if (!leaf.versions[1].mode)
1783                 die("Path %s not in branch", s);
1784         tree_content_set(&b->branch_tree, d,
1785                 leaf.versions[1].sha1,
1786                 leaf.versions[1].mode,
1787                 leaf.tree);
1788
1789         free(s_uq);
1790         free(d_uq);
1791 }
1792
1793 static void file_change_deleteall(struct branch *b)
1794 {
1795         release_tree_content_recursive(b->branch_tree.tree);
1796         hashclr(b->branch_tree.versions[0].sha1);
1797         hashclr(b->branch_tree.versions[1].sha1);
1798         load_tree(&b->branch_tree);
1799 }
1800
1801 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1802 {
1803         if (!buf || size < 46)
1804                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1805         if (memcmp("tree ", buf, 5)
1806                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1807                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1808         hashcpy(b->branch_tree.versions[0].sha1,
1809                 b->branch_tree.versions[1].sha1);
1810 }
1811
1812 static void cmd_from_existing(struct branch *b)
1813 {
1814         if (is_null_sha1(b->sha1)) {
1815                 hashclr(b->branch_tree.versions[0].sha1);
1816                 hashclr(b->branch_tree.versions[1].sha1);
1817         } else {
1818                 unsigned long size;
1819                 char *buf;
1820
1821                 buf = read_object_with_reference(b->sha1,
1822                         commit_type, &size, b->sha1);
1823                 cmd_from_commit(b, buf, size);
1824                 free(buf);
1825         }
1826 }
1827
1828 static void cmd_from(struct branch *b)
1829 {
1830         const char *from;
1831         struct branch *s;
1832
1833         if (prefixcmp(command_buf.buf, "from "))
1834                 return;
1835
1836         if (b->branch_tree.tree) {
1837                 release_tree_content_recursive(b->branch_tree.tree);
1838                 b->branch_tree.tree = NULL;
1839         }
1840
1841         from = strchr(command_buf.buf, ' ') + 1;
1842         s = lookup_branch(from);
1843         if (b == s)
1844                 die("Can't create a branch from itself: %s", b->name);
1845         else if (s) {
1846                 unsigned char *t = s->branch_tree.versions[1].sha1;
1847                 hashcpy(b->sha1, s->sha1);
1848                 hashcpy(b->branch_tree.versions[0].sha1, t);
1849                 hashcpy(b->branch_tree.versions[1].sha1, t);
1850         } else if (*from == ':') {
1851                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1852                 struct object_entry *oe = find_mark(idnum);
1853                 if (oe->type != OBJ_COMMIT)
1854                         die("Mark :%" PRIuMAX " not a commit", idnum);
1855                 hashcpy(b->sha1, oe->sha1);
1856                 if (oe->pack_id != MAX_PACK_ID) {
1857                         unsigned long size;
1858                         char *buf = gfi_unpack_entry(oe, &size);
1859                         cmd_from_commit(b, buf, size);
1860                         free(buf);
1861                 } else
1862                         cmd_from_existing(b);
1863         } else if (!get_sha1(from, b->sha1))
1864                 cmd_from_existing(b);
1865         else
1866                 die("Invalid ref name or SHA1 expression: %s", from);
1867
1868         read_next_command();
1869 }
1870
1871 static struct hash_list *cmd_merge(unsigned int *count)
1872 {
1873         struct hash_list *list = NULL, *n, *e = e;
1874         const char *from;
1875         struct branch *s;
1876
1877         *count = 0;
1878         while (!prefixcmp(command_buf.buf, "merge ")) {
1879                 from = strchr(command_buf.buf, ' ') + 1;
1880                 n = xmalloc(sizeof(*n));
1881                 s = lookup_branch(from);
1882                 if (s)
1883                         hashcpy(n->sha1, s->sha1);
1884                 else if (*from == ':') {
1885                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1886                         struct object_entry *oe = find_mark(idnum);
1887                         if (oe->type != OBJ_COMMIT)
1888                                 die("Mark :%" PRIuMAX " not a commit", idnum);
1889                         hashcpy(n->sha1, oe->sha1);
1890                 } else if (!get_sha1(from, n->sha1)) {
1891                         unsigned long size;
1892                         char *buf = read_object_with_reference(n->sha1,
1893                                 commit_type, &size, n->sha1);
1894                         if (!buf || size < 46)
1895                                 die("Not a valid commit: %s", from);
1896                         free(buf);
1897                 } else
1898                         die("Invalid ref name or SHA1 expression: %s", from);
1899
1900                 n->next = NULL;
1901                 if (list)
1902                         e->next = n;
1903                 else
1904                         list = n;
1905                 e = n;
1906                 (*count)++;
1907                 read_next_command();
1908         }
1909         return list;
1910 }
1911
1912 static void cmd_new_commit(void)
1913 {
1914         struct branch *b;
1915         void *msg;
1916         size_t msglen;
1917         char *sp;
1918         char *author = NULL;
1919         char *committer = NULL;
1920         struct hash_list *merge_list = NULL;
1921         unsigned int merge_count;
1922
1923         /* Obtain the branch name from the rest of our command */
1924         sp = strchr(command_buf.buf, ' ') + 1;
1925         b = lookup_branch(sp);
1926         if (!b)
1927                 b = new_branch(sp);
1928
1929         read_next_command();
1930         cmd_mark();
1931         if (!prefixcmp(command_buf.buf, "author ")) {
1932                 author = parse_ident(command_buf.buf + 7);
1933                 read_next_command();
1934         }
1935         if (!prefixcmp(command_buf.buf, "committer ")) {
1936                 committer = parse_ident(command_buf.buf + 10);
1937                 read_next_command();
1938         }
1939         if (!committer)
1940                 die("Expected committer but didn't get one");
1941         msg = cmd_data(&msglen);
1942         read_next_command();
1943         cmd_from(b);
1944         merge_list = cmd_merge(&merge_count);
1945
1946         /* ensure the branch is active/loaded */
1947         if (!b->branch_tree.tree || !max_active_branches) {
1948                 unload_one_branch();
1949                 load_branch(b);
1950         }
1951
1952         /* file_change* */
1953         for (;;) {
1954                 if (1 == command_buf.len)
1955                         break;
1956                 else if (!prefixcmp(command_buf.buf, "M "))
1957                         file_change_m(b);
1958                 else if (!prefixcmp(command_buf.buf, "D "))
1959                         file_change_d(b);
1960                 else if (!prefixcmp(command_buf.buf, "R "))
1961                         file_change_cr(b, 1);
1962                 else if (!prefixcmp(command_buf.buf, "C "))
1963                         file_change_cr(b, 0);
1964                 else if (!strcmp("deleteall", command_buf.buf))
1965                         file_change_deleteall(b);
1966                 else
1967                         die("Unsupported file_change: %s", command_buf.buf);
1968                 read_next_command();
1969         }
1970
1971         /* build the tree and the commit */
1972         store_tree(&b->branch_tree);
1973         hashcpy(b->branch_tree.versions[0].sha1,
1974                 b->branch_tree.versions[1].sha1);
1975         size_dbuf(&new_data, 114 + msglen
1976                 + merge_count * 49
1977                 + (author
1978                         ? strlen(author) + strlen(committer)
1979                         : 2 * strlen(committer)));
1980         sp = new_data.buffer;
1981         sp += sprintf(sp, "tree %s\n",
1982                 sha1_to_hex(b->branch_tree.versions[1].sha1));
1983         if (!is_null_sha1(b->sha1))
1984                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(b->sha1));
1985         while (merge_list) {
1986                 struct hash_list *next = merge_list->next;
1987                 sp += sprintf(sp, "parent %s\n", sha1_to_hex(merge_list->sha1));
1988                 free(merge_list);
1989                 merge_list = next;
1990         }
1991         sp += sprintf(sp, "author %s\n", author ? author : committer);
1992         sp += sprintf(sp, "committer %s\n", committer);
1993         *sp++ = '\n';
1994         memcpy(sp, msg, msglen);
1995         sp += msglen;
1996         free(author);
1997         free(committer);
1998         free(msg);
1999
2000         if (!store_object(OBJ_COMMIT,
2001                 new_data.buffer, sp - (char*)new_data.buffer,
2002                 NULL, b->sha1, next_mark))
2003                 b->pack_id = pack_id;
2004         b->last_commit = object_count_by_type[OBJ_COMMIT];
2005 }
2006
2007 static void cmd_new_tag(void)
2008 {
2009         char *sp;
2010         const char *from;
2011         char *tagger;
2012         struct branch *s;
2013         void *msg;
2014         size_t msglen;
2015         struct tag *t;
2016         uintmax_t from_mark = 0;
2017         unsigned char sha1[20];
2018
2019         /* Obtain the new tag name from the rest of our command */
2020         sp = strchr(command_buf.buf, ' ') + 1;
2021         t = pool_alloc(sizeof(struct tag));
2022         t->next_tag = NULL;
2023         t->name = pool_strdup(sp);
2024         if (last_tag)
2025                 last_tag->next_tag = t;
2026         else
2027                 first_tag = t;
2028         last_tag = t;
2029         read_next_command();
2030
2031         /* from ... */
2032         if (prefixcmp(command_buf.buf, "from "))
2033                 die("Expected from command, got %s", command_buf.buf);
2034         from = strchr(command_buf.buf, ' ') + 1;
2035         s = lookup_branch(from);
2036         if (s) {
2037                 hashcpy(sha1, s->sha1);
2038         } else if (*from == ':') {
2039                 struct object_entry *oe;
2040                 from_mark = strtoumax(from + 1, NULL, 10);
2041                 oe = find_mark(from_mark);
2042                 if (oe->type != OBJ_COMMIT)
2043                         die("Mark :%" PRIuMAX " not a commit", from_mark);
2044                 hashcpy(sha1, oe->sha1);
2045         } else if (!get_sha1(from, sha1)) {
2046                 unsigned long size;
2047                 char *buf;
2048
2049                 buf = read_object_with_reference(sha1,
2050                         commit_type, &size, sha1);
2051                 if (!buf || size < 46)
2052                         die("Not a valid commit: %s", from);
2053                 free(buf);
2054         } else
2055                 die("Invalid ref name or SHA1 expression: %s", from);
2056         read_next_command();
2057
2058         /* tagger ... */
2059         if (prefixcmp(command_buf.buf, "tagger "))
2060                 die("Expected tagger command, got %s", command_buf.buf);
2061         tagger = parse_ident(command_buf.buf + 7);
2062
2063         /* tag payload/message */
2064         read_next_command();
2065         msg = cmd_data(&msglen);
2066
2067         /* build the tag object */
2068         size_dbuf(&new_data, 67+strlen(t->name)+strlen(tagger)+msglen);
2069         sp = new_data.buffer;
2070         sp += sprintf(sp, "object %s\n", sha1_to_hex(sha1));
2071         sp += sprintf(sp, "type %s\n", commit_type);
2072         sp += sprintf(sp, "tag %s\n", t->name);
2073         sp += sprintf(sp, "tagger %s\n", tagger);
2074         *sp++ = '\n';
2075         memcpy(sp, msg, msglen);
2076         sp += msglen;
2077         free(tagger);
2078         free(msg);
2079
2080         if (store_object(OBJ_TAG, new_data.buffer,
2081                 sp - (char*)new_data.buffer,
2082                 NULL, t->sha1, 0))
2083                 t->pack_id = MAX_PACK_ID;
2084         else
2085                 t->pack_id = pack_id;
2086 }
2087
2088 static void cmd_reset_branch(void)
2089 {
2090         struct branch *b;
2091         char *sp;
2092
2093         /* Obtain the branch name from the rest of our command */
2094         sp = strchr(command_buf.buf, ' ') + 1;
2095         b = lookup_branch(sp);
2096         if (b) {
2097                 hashclr(b->sha1);
2098                 hashclr(b->branch_tree.versions[0].sha1);
2099                 hashclr(b->branch_tree.versions[1].sha1);
2100                 if (b->branch_tree.tree) {
2101                         release_tree_content_recursive(b->branch_tree.tree);
2102                         b->branch_tree.tree = NULL;
2103                 }
2104         }
2105         else
2106                 b = new_branch(sp);
2107         read_next_command();
2108         cmd_from(b);
2109 }
2110
2111 static void cmd_checkpoint(void)
2112 {
2113         if (object_count) {
2114                 cycle_packfile();
2115                 dump_branches();
2116                 dump_tags();
2117                 dump_marks();
2118         }
2119         read_next_command();
2120 }
2121
2122 static void import_marks(const char *input_file)
2123 {
2124         char line[512];
2125         FILE *f = fopen(input_file, "r");
2126         if (!f)
2127                 die("cannot read %s: %s", input_file, strerror(errno));
2128         while (fgets(line, sizeof(line), f)) {
2129                 uintmax_t mark;
2130                 char *end;
2131                 unsigned char sha1[20];
2132                 struct object_entry *e;
2133
2134                 end = strchr(line, '\n');
2135                 if (line[0] != ':' || !end)
2136                         die("corrupt mark line: %s", line);
2137                 *end = 0;
2138                 mark = strtoumax(line + 1, &end, 10);
2139                 if (!mark || end == line + 1
2140                         || *end != ' ' || get_sha1(end + 1, sha1))
2141                         die("corrupt mark line: %s", line);
2142                 e = find_object(sha1);
2143                 if (!e) {
2144                         enum object_type type = sha1_object_info(sha1, NULL);
2145                         if (type < 0)
2146                                 die("object not found: %s", sha1_to_hex(sha1));
2147                         e = insert_object(sha1);
2148                         e->type = type;
2149                         e->pack_id = MAX_PACK_ID;
2150                         e->offset = 1; /* just not zero! */
2151                 }
2152                 insert_mark(mark, e);
2153         }
2154         fclose(f);
2155 }
2156
2157 static const char fast_import_usage[] =
2158 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2159
2160 int main(int argc, const char **argv)
2161 {
2162         int i, show_stats = 1;
2163
2164         git_config(git_default_config);
2165         alloc_objects(object_entry_alloc);
2166         strbuf_init(&command_buf);
2167         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2168         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2169         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2170         marks = pool_calloc(1, sizeof(struct mark_set));
2171
2172         for (i = 1; i < argc; i++) {
2173                 const char *a = argv[i];
2174
2175                 if (*a != '-' || !strcmp(a, "--"))
2176                         break;
2177                 else if (!prefixcmp(a, "--date-format=")) {
2178                         const char *fmt = a + 14;
2179                         if (!strcmp(fmt, "raw"))
2180                                 whenspec = WHENSPEC_RAW;
2181                         else if (!strcmp(fmt, "rfc2822"))
2182                                 whenspec = WHENSPEC_RFC2822;
2183                         else if (!strcmp(fmt, "now"))
2184                                 whenspec = WHENSPEC_NOW;
2185                         else
2186                                 die("unknown --date-format argument %s", fmt);
2187                 }
2188                 else if (!prefixcmp(a, "--max-pack-size="))
2189                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2190                 else if (!prefixcmp(a, "--depth="))
2191                         max_depth = strtoul(a + 8, NULL, 0);
2192                 else if (!prefixcmp(a, "--active-branches="))
2193                         max_active_branches = strtoul(a + 18, NULL, 0);
2194                 else if (!prefixcmp(a, "--import-marks="))
2195                         import_marks(a + 15);
2196                 else if (!prefixcmp(a, "--export-marks="))
2197                         mark_file = a + 15;
2198                 else if (!prefixcmp(a, "--export-pack-edges=")) {
2199                         if (pack_edges)
2200                                 fclose(pack_edges);
2201                         pack_edges = fopen(a + 20, "a");
2202                         if (!pack_edges)
2203                                 die("Cannot open %s: %s", a + 20, strerror(errno));
2204                 } else if (!strcmp(a, "--force"))
2205                         force_update = 1;
2206                 else if (!strcmp(a, "--quiet"))
2207                         show_stats = 0;
2208                 else if (!strcmp(a, "--stats"))
2209                         show_stats = 1;
2210                 else
2211                         die("unknown option %s", a);
2212         }
2213         if (i != argc)
2214                 usage(fast_import_usage);
2215
2216         prepare_packed_git();
2217         start_packfile();
2218         for (;;) {
2219                 read_next_command();
2220                 if (command_buf.eof)
2221                         break;
2222                 else if (!strcmp("blob", command_buf.buf))
2223                         cmd_new_blob();
2224                 else if (!prefixcmp(command_buf.buf, "commit "))
2225                         cmd_new_commit();
2226                 else if (!prefixcmp(command_buf.buf, "tag "))
2227                         cmd_new_tag();
2228                 else if (!prefixcmp(command_buf.buf, "reset "))
2229                         cmd_reset_branch();
2230                 else if (!strcmp("checkpoint", command_buf.buf))
2231                         cmd_checkpoint();
2232                 else
2233                         die("Unsupported command: %s", command_buf.buf);
2234         }
2235         end_packfile();
2236
2237         dump_branches();
2238         dump_tags();
2239         unkeep_all_packs();
2240         dump_marks();
2241
2242         if (pack_edges)
2243                 fclose(pack_edges);
2244
2245         if (show_stats) {
2246                 uintmax_t total_count = 0, duplicate_count = 0;
2247                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2248                         total_count += object_count_by_type[i];
2249                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2250                         duplicate_count += duplicate_count_by_type[i];
2251
2252                 fprintf(stderr, "%s statistics:\n", argv[0]);
2253                 fprintf(stderr, "---------------------------------------------------------------------\n");
2254                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2255                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
2256                 fprintf(stderr, "      blobs  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB]);
2257                 fprintf(stderr, "      trees  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE]);
2258                 fprintf(stderr, "      commits:   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT]);
2259                 fprintf(stderr, "      tags   :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG]);
2260                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2261                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2262                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2263                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2264                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
2265                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2266                 fprintf(stderr, "---------------------------------------------------------------------\n");
2267                 pack_report();
2268                 fprintf(stderr, "---------------------------------------------------------------------\n");
2269                 fprintf(stderr, "\n");
2270         }
2271
2272         return failure ? 1 : 0;
2273 }