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         | progress
  12         ;
  13
  14   new_blob ::= 'blob' lf
  15     mark?
  16     file_content;
  17   file_content ::= data;
  18
  19   new_commit ::= 'commit' sp ref_str lf
  20     mark?
  21     ('author' sp name '<' email '>' when lf)?
  22     'committer' sp name '<' email '>' when lf
  23     commit_msg
  24     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  25     ('merge' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)*
  26     file_change*
  27     lf?;
  28   commit_msg ::= data;
  29
  30   file_change ::= file_clr
  31     | file_del
  32     | file_rnm
  33     | file_cpy
  34     | file_obm
  35     | file_inm;
  36   file_clr ::= 'deleteall' lf;
  37   file_del ::= 'D' sp path_str lf;
  38   file_rnm ::= 'R' sp path_str sp path_str lf;
  39   file_cpy ::= 'C' sp path_str sp path_str lf;
  40   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  41   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  42     data;
  43
  44   new_tag ::= 'tag' sp tag_str lf
  45     'from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf
  46     'tagger' sp name '<' email '>' when lf
  47     tag_msg;
  48   tag_msg ::= data;
  49
  50   reset_branch ::= 'reset' sp ref_str lf
  51     ('from' sp (ref_str | hexsha1 | sha1exp_str | idnum) lf)?
  52     lf?;
  53
  54   checkpoint ::= 'checkpoint' lf
  55     lf?;
  56
  57   progress ::= 'progress' sp not_lf* lf
  58     lf?;
  59
  60      # note: the first idnum in a stream should be 1 and subsequent
  61      # idnums should not have gaps between values as this will cause
  62      # the stream parser to reserve space for the gapped values.  An
  63      # idnum can be updated in the future to a new object by issuing
  64      # a new mark directive with the old idnum.
  65      #
  66   mark ::= 'mark' sp idnum lf;
  67   data ::= (delimited_data | exact_data)
  68     lf?;
  69
  70     # note: delim may be any string but must not contain lf.
  71     # data_line may contain any data but must not be exactly
  72     # delim.
  73   delimited_data ::= 'data' sp '<<' delim lf
  74     (data_line lf)*
  75     delim lf;
  76
  77      # note: declen indicates the length of binary_data in bytes.
  78      # declen does not include the lf preceeding the binary data.
  79      #
  80   exact_data ::= 'data' sp declen lf
  81     binary_data;
  82
  83      # note: quoted strings are C-style quoting supporting \c for
  84      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  85      # is the signed byte value in octal.  Note that the only
  86      # characters which must actually be escaped to protect the
  87      # stream formatting is: \, " and LF.  Otherwise these values
  88      # are UTF8.
  89      #
  90   ref_str     ::= ref;
  91   sha1exp_str ::= sha1exp;
  92   tag_str     ::= tag;
  93   path_str    ::= path    | '"' quoted(path)    '"' ;
  94   mode        ::= '100644' | '644'
  95                 | '100755' | '755'
  96                 | '120000'
  97                 ;
  98
  99   declen ::= # unsigned 32 bit value, ascii base10 notation;
 100   bigint ::= # unsigned integer value, ascii base10 notation;
 101   binary_data ::= # file content, not interpreted;
 102
 103   when         ::= raw_when | rfc2822_when;
 104   raw_when     ::= ts sp tz;
 105   rfc2822_when ::= # Valid RFC 2822 date and time;
 106
 107   sp ::= # ASCII space character;
 108   lf ::= # ASCII newline (LF) character;
 109
 110      # note: a colon (':') must precede the numerical value assigned to
 111      # an idnum.  This is to distinguish it from a ref or tag name as
 112      # GIT does not permit ':' in ref or tag strings.
 113      #
 114   idnum   ::= ':' bigint;
 115   path    ::= # GIT style file path, e.g. "a/b/c";
 116   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 117   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 118   sha1exp ::= # Any valid GIT SHA1 expression;
 119   hexsha1 ::= # SHA1 in hexadecimal format;
 120
 121      # note: name and email are UTF8 strings, however name must not
 122      # contain '<' or lf and email must not contain any of the
 123      # following: '<', '>', lf.
 124      #
 125   name  ::= # valid GIT author/committer name;
 126   email ::= # valid GIT author/committer email;
 127   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 128   tz    ::= # GIT style timezone;
 129
 130      # note: comments may appear anywhere in the input, except
 131      # within a data command.  Any form of the data command
 132      # always escapes the related input from comment processing.
 133      #
 134      # In case it is not clear, the '#' that starts the comment
 135      # must be the first character on that the line (an lf have
 136      # preceeded it).
 137      #
 138   comment ::= '#' not_lf* lf;
 139   not_lf  ::= # Any byte that is not ASCII newline (LF);
 140 */
 141
 142 #include "builtin.h"
 143 #include "cache.h"
 144 #include "object.h"
 145 #include "blob.h"
 146 #include "tree.h"
 147 #include "commit.h"
 148 #include "delta.h"
 149 #include "pack.h"
 150 #include "refs.h"
 151 #include "csum-file.h"
 152 #include "quote.h"
 153
 154 #define PACK_ID_BITS 16
 155 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 156
 157 struct object_entry
 158 {
 159         struct object_entry *next;
 160         uint32_t offset;
 161         unsigned type : TYPE_BITS;
 162         unsigned pack_id : PACK_ID_BITS;
 163         unsigned char sha1[20];
 164 };
 165
 166 struct object_entry_pool
 167 {
 168         struct object_entry_pool *next_pool;
 169         struct object_entry *next_free;
 170         struct object_entry *end;
 171         struct object_entry entries[FLEX_ARRAY]; /* more */
 172 };
 173
 174 struct mark_set
 175 {
 176         union {
 177                 struct object_entry *marked[1024];
 178                 struct mark_set *sets[1024];
 179         } data;
 180         unsigned int shift;
 181 };
 182
 183 struct last_object
 184 {
 185         void *data;
 186         unsigned long len;
 187         uint32_t offset;
 188         unsigned int depth;
 189         unsigned no_free:1;
 190 };
 191
 192 struct mem_pool
 193 {
 194         struct mem_pool *next_pool;
 195         char *next_free;
 196         char *end;
 197         char space[FLEX_ARRAY]; /* more */
 198 };
 199
 200 struct atom_str
 201 {
 202         struct atom_str *next_atom;
 203         unsigned short str_len;
 204         char str_dat[FLEX_ARRAY]; /* more */
 205 };
 206
 207 struct tree_content;
 208 struct tree_entry
 209 {
 210         struct tree_content *tree;
 211         struct atom_str* name;
 212         struct tree_entry_ms
 213         {
 214                 uint16_t mode;
 215                 unsigned char sha1[20];
 216         } versions[2];
 217 };
 218
 219 struct tree_content
 220 {
 221         unsigned int entry_capacity; /* must match avail_tree_content */
 222         unsigned int entry_count;
 223         unsigned int delta_depth;
 224         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 225 };
 226
 227 struct avail_tree_content
 228 {
 229         unsigned int entry_capacity; /* must match tree_content */
 230         struct avail_tree_content *next_avail;
 231 };
 232
 233 struct branch
 234 {
 235         struct branch *table_next_branch;
 236         struct branch *active_next_branch;
 237         const char *name;
 238         struct tree_entry branch_tree;
 239         uintmax_t last_commit;
 240         unsigned active : 1;
 241         unsigned pack_id : PACK_ID_BITS;
 242         unsigned char sha1[20];
 243 };
 244
 245 struct tag
 246 {
 247         struct tag *next_tag;
 248         const char *name;
 249         unsigned int pack_id;
 250         unsigned char sha1[20];
 251 };
 252
 253 struct hash_list
 254 {
 255         struct hash_list *next;
 256         unsigned char sha1[20];
 257 };
 258
 259 typedef enum {
 260         WHENSPEC_RAW = 1,
 261         WHENSPEC_RFC2822,
 262         WHENSPEC_NOW,
 263 } whenspec_type;
 264
 265 struct recent_command
 266 {
 267         struct recent_command *prev;
 268         struct recent_command *next;
 269         char *buf;
 270 };
 271
 272 /* Configured limits on output */
 273 static unsigned long max_depth = 10;
 274 static off_t max_packsize = (1LL << 32) - 1;
 275 static int force_update;
 276
 277 /* Stats and misc. counters */
 278 static uintmax_t alloc_count;
 279 static uintmax_t marks_set_count;
 280 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 281 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 282 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 283 static unsigned long object_count;
 284 static unsigned long branch_count;
 285 static unsigned long branch_load_count;
 286 static int failure;
 287 static FILE *pack_edges;
 288
 289 /* Memory pools */
 290 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 291 static size_t total_allocd;
 292 static struct mem_pool *mem_pool;
 293
 294 /* Atom management */
 295 static unsigned int atom_table_sz = 4451;
 296 static unsigned int atom_cnt;
 297 static struct atom_str **atom_table;
 298
 299 /* The .pack file being generated */
 300 static unsigned int pack_id;
 301 static struct packed_git *pack_data;
 302 static struct packed_git **all_packs;
 303 static unsigned long pack_size;
 304
 305 /* Table of objects we've written. */
 306 static unsigned int object_entry_alloc = 5000;
 307 static struct object_entry_pool *blocks;
 308 static struct object_entry *object_table[1 << 16];
 309 static struct mark_set *marks;
 310 static const char* mark_file;
 311
 312 /* Our last blob */
 313 static struct last_object last_blob;
 314
 315 /* Tree management */
 316 static unsigned int tree_entry_alloc = 1000;
 317 static void *avail_tree_entry;
 318 static unsigned int avail_tree_table_sz = 100;
 319 static struct avail_tree_content **avail_tree_table;
 320 static struct strbuf old_tree = STRBUF_INIT;
 321 static struct strbuf new_tree = STRBUF_INIT;
 322
 323 /* Branch data */
 324 static unsigned long max_active_branches = 5;
 325 static unsigned long cur_active_branches;
 326 static unsigned long branch_table_sz = 1039;
 327 static struct branch **branch_table;
 328 static struct branch *active_branches;
 329
 330 /* Tag data */
 331 static struct tag *first_tag;
 332 static struct tag *last_tag;
 333
 334 /* Input stream parsing */
 335 static whenspec_type whenspec = WHENSPEC_RAW;
 336 static struct strbuf command_buf = STRBUF_INIT;
 337 static int unread_command_buf;
 338 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 339 static struct recent_command *cmd_tail = &cmd_hist;
 340 static struct recent_command *rc_free;
 341 static unsigned int cmd_save = 100;
 342 static uintmax_t next_mark;
 343 static struct strbuf new_data = STRBUF_INIT;
 344
 345 static void write_branch_report(FILE *rpt, struct branch *b)
 346 {
 347         fprintf(rpt, "%s:\n", b->name);
 348
 349         fprintf(rpt, "  status      :");
 350         if (b->active)
 351                 fputs(" active", rpt);
 352         if (b->branch_tree.tree)
 353                 fputs(" loaded", rpt);
 354         if (is_null_sha1(b->branch_tree.versions[1].sha1))
 355                 fputs(" dirty", rpt);
 356         fputc('\n', rpt);
 357
 358         fprintf(rpt, "  tip commit  : %s\n", sha1_to_hex(b->sha1));
 359         fprintf(rpt, "  old tree    : %s\n", sha1_to_hex(b->branch_tree.versions[0].sha1));
 360         fprintf(rpt, "  cur tree    : %s\n", sha1_to_hex(b->branch_tree.versions[1].sha1));
 361         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 362
 363         fputs("  last pack   : ", rpt);
 364         if (b->pack_id < MAX_PACK_ID)
 365                 fprintf(rpt, "%u", b->pack_id);
 366         fputc('\n', rpt);
 367
 368         fputc('\n', rpt);
 369 }
 370
 371 static void write_crash_report(const char *err)
 372 {
 373         char *loc = git_path("fast_import_crash_%d", getpid());
 374         FILE *rpt = fopen(loc, "w");
 375         struct branch *b;
 376         unsigned long lu;
 377         struct recent_command *rc;
 378
 379         if (!rpt) {
 380                 error("can't write crash report %s: %s", loc, strerror(errno));
 381                 return;
 382         }
 383
 384         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 385
 386         fprintf(rpt, "fast-import crash report:\n");
 387         fprintf(rpt, "    fast-import process: %d\n", getpid());
 388         fprintf(rpt, "    parent process     : %d\n", getppid());
 389         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_LOCAL));
 390         fputc('\n', rpt);
 391
 392         fputs("fatal: ", rpt);
 393         fputs(err, rpt);
 394         fputc('\n', rpt);
 395
 396         fputc('\n', rpt);
 397         fputs("Most Recent Commands Before Crash\n", rpt);
 398         fputs("---------------------------------\n", rpt);
 399         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 400                 if (rc->next == &cmd_hist)
 401                         fputs("* ", rpt);
 402                 else
 403                         fputs("  ", rpt);
 404                 fputs(rc->buf, rpt);
 405                 fputc('\n', rpt);
 406         }
 407
 408         fputc('\n', rpt);
 409         fputs("Active Branch LRU\n", rpt);
 410         fputs("-----------------\n", rpt);
 411         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 412                 cur_active_branches,
 413                 max_active_branches);
 414         fputc('\n', rpt);
 415         fputs("  pos  clock name\n", rpt);
 416         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 417         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 418                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 419                         ++lu, b->last_commit, b->name);
 420
 421         fputc('\n', rpt);
 422         fputs("Inactive Branches\n", rpt);
 423         fputs("-----------------\n", rpt);
 424         for (lu = 0; lu < branch_table_sz; lu++) {
 425                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 426                         write_branch_report(rpt, b);
 427         }
 428
 429         fputc('\n', rpt);
 430         fputs("-------------------\n", rpt);
 431         fputs("END OF CRASH REPORT\n", rpt);
 432         fclose(rpt);
 433 }
 434
 435 static NORETURN void die_nicely(const char *err, va_list params)
 436 {
 437         static int zombie;
 438         char message[2 * PATH_MAX];
 439
 440         vsnprintf(message, sizeof(message), err, params);
 441         fputs("fatal: ", stderr);
 442         fputs(message, stderr);
 443         fputc('\n', stderr);
 444
 445         if (!zombie) {
 446                 zombie = 1;
 447                 write_crash_report(message);
 448         }
 449         exit(128);
 450 }
 451
 452 static void alloc_objects(unsigned int cnt)
 453 {
 454         struct object_entry_pool *b;
 455
 456         b = xmalloc(sizeof(struct object_entry_pool)
 457                 + cnt * sizeof(struct object_entry));
 458         b->next_pool = blocks;
 459         b->next_free = b->entries;
 460         b->end = b->entries + cnt;
 461         blocks = b;
 462         alloc_count += cnt;
 463 }
 464
 465 static struct object_entry *new_object(unsigned char *sha1)
 466 {
 467         struct object_entry *e;
 468
 469         if (blocks->next_free == blocks->end)
 470                 alloc_objects(object_entry_alloc);
 471
 472         e = blocks->next_free++;
 473         hashcpy(e->sha1, sha1);
 474         return e;
 475 }
 476
 477 static struct object_entry *find_object(unsigned char *sha1)
 478 {
 479         unsigned int h = sha1[0] << 8 | sha1[1];
 480         struct object_entry *e;
 481         for (e = object_table[h]; e; e = e->next)
 482                 if (!hashcmp(sha1, e->sha1))
 483                         return e;
 484         return NULL;
 485 }
 486
 487 static struct object_entry *insert_object(unsigned char *sha1)
 488 {
 489         unsigned int h = sha1[0] << 8 | sha1[1];
 490         struct object_entry *e = object_table[h];
 491         struct object_entry *p = NULL;
 492
 493         while (e) {
 494                 if (!hashcmp(sha1, e->sha1))
 495                         return e;
 496                 p = e;
 497                 e = e->next;
 498         }
 499
 500         e = new_object(sha1);
 501         e->next = NULL;
 502         e->offset = 0;
 503         if (p)
 504                 p->next = e;
 505         else
 506                 object_table[h] = e;
 507         return e;
 508 }
 509
 510 static unsigned int hc_str(const char *s, size_t len)
 511 {
 512         unsigned int r = 0;
 513         while (len-- > 0)
 514                 r = r * 31 + *s++;
 515         return r;
 516 }
 517
 518 static void *pool_alloc(size_t len)
 519 {
 520         struct mem_pool *p;
 521         void *r;
 522
 523         for (p = mem_pool; p; p = p->next_pool)
 524                 if ((p->end - p->next_free >= len))
 525                         break;
 526
 527         if (!p) {
 528                 if (len >= (mem_pool_alloc/2)) {
 529                         total_allocd += len;
 530                         return xmalloc(len);
 531                 }
 532                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 533                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 534                 p->next_pool = mem_pool;
 535                 p->next_free = p->space;
 536                 p->end = p->next_free + mem_pool_alloc;
 537                 mem_pool = p;
 538         }
 539
 540         r = p->next_free;
 541         /* round out to a pointer alignment */
 542         if (len & (sizeof(void*) - 1))
 543                 len += sizeof(void*) - (len & (sizeof(void*) - 1));
 544         p->next_free += len;
 545         return r;
 546 }
 547
 548 static void *pool_calloc(size_t count, size_t size)
 549 {
 550         size_t len = count * size;
 551         void *r = pool_alloc(len);
 552         memset(r, 0, len);
 553         return r;
 554 }
 555
 556 static char *pool_strdup(const char *s)
 557 {
 558         char *r = pool_alloc(strlen(s) + 1);
 559         strcpy(r, s);
 560         return r;
 561 }
 562
 563 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 564 {
 565         struct mark_set *s = marks;
 566         while ((idnum >> s->shift) >= 1024) {
 567                 s = pool_calloc(1, sizeof(struct mark_set));
 568                 s->shift = marks->shift + 10;
 569                 s->data.sets[0] = marks;
 570                 marks = s;
 571         }
 572         while (s->shift) {
 573                 uintmax_t i = idnum >> s->shift;
 574                 idnum -= i << s->shift;
 575                 if (!s->data.sets[i]) {
 576                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 577                         s->data.sets[i]->shift = s->shift - 10;
 578                 }
 579                 s = s->data.sets[i];
 580         }
 581         if (!s->data.marked[idnum])
 582                 marks_set_count++;
 583         s->data.marked[idnum] = oe;
 584 }
 585
 586 static struct object_entry *find_mark(uintmax_t idnum)
 587 {
 588         uintmax_t orig_idnum = idnum;
 589         struct mark_set *s = marks;
 590         struct object_entry *oe = NULL;
 591         if ((idnum >> s->shift) < 1024) {
 592                 while (s && s->shift) {
 593                         uintmax_t i = idnum >> s->shift;
 594                         idnum -= i << s->shift;
 595                         s = s->data.sets[i];
 596                 }
 597                 if (s)
 598                         oe = s->data.marked[idnum];
 599         }
 600         if (!oe)
 601                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 602         return oe;
 603 }
 604
 605 static struct atom_str *to_atom(const char *s, unsigned short len)
 606 {
 607         unsigned int hc = hc_str(s, len) % atom_table_sz;
 608         struct atom_str *c;
 609
 610         for (c = atom_table[hc]; c; c = c->next_atom)
 611                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 612                         return c;
 613
 614         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 615         c->str_len = len;
 616         strncpy(c->str_dat, s, len);
 617         c->str_dat[len] = 0;
 618         c->next_atom = atom_table[hc];
 619         atom_table[hc] = c;
 620         atom_cnt++;
 621         return c;
 622 }
 623
 624 static struct branch *lookup_branch(const char *name)
 625 {
 626         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 627         struct branch *b;
 628
 629         for (b = branch_table[hc]; b; b = b->table_next_branch)
 630                 if (!strcmp(name, b->name))
 631                         return b;
 632         return NULL;
 633 }
 634
 635 static struct branch *new_branch(const char *name)
 636 {
 637         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 638         struct branch* b = lookup_branch(name);
 639
 640         if (b)
 641                 die("Invalid attempt to create duplicate branch: %s", name);
 642         switch (check_ref_format(name)) {
 643         case  0: break; /* its valid */
 644         case -2: break; /* valid, but too few '/', allow anyway */
 645         default:
 646                 die("Branch name doesn't conform to GIT standards: %s", name);
 647         }
 648
 649         b = pool_calloc(1, sizeof(struct branch));
 650         b->name = pool_strdup(name);
 651         b->table_next_branch = branch_table[hc];
 652         b->branch_tree.versions[0].mode = S_IFDIR;
 653         b->branch_tree.versions[1].mode = S_IFDIR;
 654         b->active = 0;
 655         b->pack_id = MAX_PACK_ID;
 656         branch_table[hc] = b;
 657         branch_count++;
 658         return b;
 659 }
 660
 661 static unsigned int hc_entries(unsigned int cnt)
 662 {
 663         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 664         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 665 }
 666
 667 static struct tree_content *new_tree_content(unsigned int cnt)
 668 {
 669         struct avail_tree_content *f, *l = NULL;
 670         struct tree_content *t;
 671         unsigned int hc = hc_entries(cnt);
 672
 673         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 674                 if (f->entry_capacity >= cnt)
 675                         break;
 676
 677         if (f) {
 678                 if (l)
 679                         l->next_avail = f->next_avail;
 680                 else
 681                         avail_tree_table[hc] = f->next_avail;
 682         } else {
 683                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 684                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 685                 f->entry_capacity = cnt;
 686         }
 687
 688         t = (struct tree_content*)f;
 689         t->entry_count = 0;
 690         t->delta_depth = 0;
 691         return t;
 692 }
 693
 694 static void release_tree_entry(struct tree_entry *e);
 695 static void release_tree_content(struct tree_content *t)
 696 {
 697         struct avail_tree_content *f = (struct avail_tree_content*)t;
 698         unsigned int hc = hc_entries(f->entry_capacity);
 699         f->next_avail = avail_tree_table[hc];
 700         avail_tree_table[hc] = f;
 701 }
 702
 703 static void release_tree_content_recursive(struct tree_content *t)
 704 {
 705         unsigned int i;
 706         for (i = 0; i < t->entry_count; i++)
 707                 release_tree_entry(t->entries[i]);
 708         release_tree_content(t);
 709 }
 710
 711 static struct tree_content *grow_tree_content(
 712         struct tree_content *t,
 713         int amt)
 714 {
 715         struct tree_content *r = new_tree_content(t->entry_count + amt);
 716         r->entry_count = t->entry_count;
 717         r->delta_depth = t->delta_depth;
 718         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 719         release_tree_content(t);
 720         return r;
 721 }
 722
 723 static struct tree_entry *new_tree_entry(void)
 724 {
 725         struct tree_entry *e;
 726
 727         if (!avail_tree_entry) {
 728                 unsigned int n = tree_entry_alloc;
 729                 total_allocd += n * sizeof(struct tree_entry);
 730                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 731                 while (n-- > 1) {
 732                         *((void**)e) = e + 1;
 733                         e++;
 734                 }
 735                 *((void**)e) = NULL;
 736         }
 737
 738         e = avail_tree_entry;
 739         avail_tree_entry = *((void**)e);
 740         return e;
 741 }
 742
 743 static void release_tree_entry(struct tree_entry *e)
 744 {
 745         if (e->tree)
 746                 release_tree_content_recursive(e->tree);
 747         *((void**)e) = avail_tree_entry;
 748         avail_tree_entry = e;
 749 }
 750
 751 static struct tree_content *dup_tree_content(struct tree_content *s)
 752 {
 753         struct tree_content *d;
 754         struct tree_entry *a, *b;
 755         unsigned int i;
 756
 757         if (!s)
 758                 return NULL;
 759         d = new_tree_content(s->entry_count);
 760         for (i = 0; i < s->entry_count; i++) {
 761                 a = s->entries[i];
 762                 b = new_tree_entry();
 763                 memcpy(b, a, sizeof(*a));
 764                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 765                         b->tree = dup_tree_content(a->tree);
 766                 else
 767                         b->tree = NULL;
 768                 d->entries[i] = b;
 769         }
 770         d->entry_count = s->entry_count;
 771         d->delta_depth = s->delta_depth;
 772
 773         return d;
 774 }
 775
 776 static void start_packfile(void)
 777 {
 778         static char tmpfile[PATH_MAX];
 779         struct packed_git *p;
 780         struct pack_header hdr;
 781         int pack_fd;
 782
 783         snprintf(tmpfile, sizeof(tmpfile),
 784                 "%s/tmp_pack_XXXXXX", get_object_directory());
 785         pack_fd = xmkstemp(tmpfile);
 786         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 787         strcpy(p->pack_name, tmpfile);
 788         p->pack_fd = pack_fd;
 789
 790         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 791         hdr.hdr_version = htonl(2);
 792         hdr.hdr_entries = 0;
 793         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 794
 795         pack_data = p;
 796         pack_size = sizeof(hdr);
 797         object_count = 0;
 798
 799         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 800         all_packs[pack_id] = p;
 801 }
 802
 803 static int oecmp (const void *a_, const void *b_)
 804 {
 805         struct object_entry *a = *((struct object_entry**)a_);
 806         struct object_entry *b = *((struct object_entry**)b_);
 807         return hashcmp(a->sha1, b->sha1);
 808 }
 809
 810 static char *create_index(void)
 811 {
 812         static char tmpfile[PATH_MAX];
 813         SHA_CTX ctx;
 814         struct sha1file *f;
 815         struct object_entry **idx, **c, **last, *e;
 816         struct object_entry_pool *o;
 817         uint32_t array[256];
 818         int i, idx_fd;
 819
 820         /* Build the sorted table of object IDs. */
 821         idx = xmalloc(object_count * sizeof(struct object_entry*));
 822         c = idx;
 823         for (o = blocks; o; o = o->next_pool)
 824                 for (e = o->next_free; e-- != o->entries;)
 825                         if (pack_id == e->pack_id)
 826                                 *c++ = e;
 827         last = idx + object_count;
 828         if (c != last)
 829                 die("internal consistency error creating the index");
 830         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 831
 832         /* Generate the fan-out array. */
 833         c = idx;
 834         for (i = 0; i < 256; i++) {
 835                 struct object_entry **next = c;;
 836                 while (next < last) {
 837                         if ((*next)->sha1[0] != i)
 838                                 break;
 839                         next++;
 840                 }
 841                 array[i] = htonl(next - idx);
 842                 c = next;
 843         }
 844
 845         snprintf(tmpfile, sizeof(tmpfile),
 846                 "%s/tmp_idx_XXXXXX", get_object_directory());
 847         idx_fd = xmkstemp(tmpfile);
 848         f = sha1fd(idx_fd, tmpfile);
 849         sha1write(f, array, 256 * sizeof(int));
 850         SHA1_Init(&ctx);
 851         for (c = idx; c != last; c++) {
 852                 uint32_t offset = htonl((*c)->offset);
 853                 sha1write(f, &offset, 4);
 854                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 855                 SHA1_Update(&ctx, (*c)->sha1, 20);
 856         }
 857         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 858         sha1close(f, NULL, 1);
 859         free(idx);
 860         SHA1_Final(pack_data->sha1, &ctx);
 861         return tmpfile;
 862 }
 863
 864 static char *keep_pack(char *curr_index_name)
 865 {
 866         static char name[PATH_MAX];
 867         static const char *keep_msg = "fast-import";
 868         int keep_fd;
 869
 870         chmod(pack_data->pack_name, 0444);
 871         chmod(curr_index_name, 0444);
 872
 873         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 874                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 875         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 876         if (keep_fd < 0)
 877                 die("cannot create keep file");
 878         write(keep_fd, keep_msg, strlen(keep_msg));
 879         close(keep_fd);
 880
 881         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 882                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 883         if (move_temp_to_file(pack_data->pack_name, name))
 884                 die("cannot store pack file");
 885
 886         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 887                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 888         if (move_temp_to_file(curr_index_name, name))
 889                 die("cannot store index file");
 890         return name;
 891 }
 892
 893 static void unkeep_all_packs(void)
 894 {
 895         static char name[PATH_MAX];
 896         int k;
 897
 898         for (k = 0; k < pack_id; k++) {
 899                 struct packed_git *p = all_packs[k];
 900                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 901                          get_object_directory(), sha1_to_hex(p->sha1));
 902                 unlink(name);
 903         }
 904 }
 905
 906 static void end_packfile(void)
 907 {
 908         struct packed_git *old_p = pack_data, *new_p;
 909
 910         if (object_count) {
 911                 char *idx_name;
 912                 int i;
 913                 struct branch *b;
 914                 struct tag *t;
 915
 916                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 917                                     pack_data->pack_name, object_count);
 918                 close(pack_data->pack_fd);
 919                 idx_name = keep_pack(create_index());
 920
 921                 /* Register the packfile with core git's machinary. */
 922                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 923                 if (!new_p)
 924                         die("core git rejected index %s", idx_name);
 925                 new_p->windows = old_p->windows;
 926                 all_packs[pack_id] = new_p;
 927                 install_packed_git(new_p);
 928
 929                 /* Print the boundary */
 930                 if (pack_edges) {
 931                         fprintf(pack_edges, "%s:", new_p->pack_name);
 932                         for (i = 0; i < branch_table_sz; i++) {
 933                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 934                                         if (b->pack_id == pack_id)
 935                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 936                                 }
 937                         }
 938                         for (t = first_tag; t; t = t->next_tag) {
 939                                 if (t->pack_id == pack_id)
 940                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 941                         }
 942                         fputc('\n', pack_edges);
 943                         fflush(pack_edges);
 944                 }
 945
 946                 pack_id++;
 947         }
 948         else
 949                 unlink(old_p->pack_name);
 950         free(old_p);
 951
 952         /* We can't carry a delta across packfiles. */
 953         free(last_blob.data);
 954         last_blob.data = NULL;
 955         last_blob.len = 0;
 956         last_blob.offset = 0;
 957         last_blob.depth = 0;
 958 }
 959
 960 static void cycle_packfile(void)
 961 {
 962         end_packfile();
 963         start_packfile();
 964 }
 965
 966 static size_t encode_header(
 967         enum object_type type,
 968         size_t size,
 969         unsigned char *hdr)
 970 {
 971         int n = 1;
 972         unsigned char c;
 973
 974         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
 975                 die("bad type %d", type);
 976
 977         c = (type << 4) | (size & 15);
 978         size >>= 4;
 979         while (size) {
 980                 *hdr++ = c | 0x80;
 981                 c = size & 0x7f;
 982                 size >>= 7;
 983                 n++;
 984         }
 985         *hdr = c;
 986         return n;
 987 }
 988
 989 static int store_object(
 990         enum object_type type,
 991         struct strbuf *dat,
 992         struct last_object *last,
 993         unsigned char *sha1out,
 994         uintmax_t mark)
 995 {
 996         void *out, *delta;
 997         struct object_entry *e;
 998         unsigned char hdr[96];
 999         unsigned char sha1[20];
1000         unsigned long hdrlen, deltalen;
1001         SHA_CTX c;
1002         z_stream s;
1003
1004         hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
1005                 (unsigned long)dat->len) + 1;
1006         SHA1_Init(&c);
1007         SHA1_Update(&c, hdr, hdrlen);
1008         SHA1_Update(&c, dat->buf, dat->len);
1009         SHA1_Final(sha1, &c);
1010         if (sha1out)
1011                 hashcpy(sha1out, sha1);
1012
1013         e = insert_object(sha1);
1014         if (mark)
1015                 insert_mark(mark, e);
1016         if (e->offset) {
1017                 duplicate_count_by_type[type]++;
1018                 return 1;
1019         } else if (find_sha1_pack(sha1, packed_git)) {
1020                 e->type = type;
1021                 e->pack_id = MAX_PACK_ID;
1022                 e->offset = 1; /* just not zero! */
1023                 duplicate_count_by_type[type]++;
1024                 return 1;
1025         }
1026
1027         if (last && last->data && last->depth < max_depth) {
1028                 delta = diff_delta(last->data, last->len,
1029                         dat->buf, dat->len,
1030                         &deltalen, 0);
1031                 if (delta && deltalen >= dat->len) {
1032                         free(delta);
1033                         delta = NULL;
1034                 }
1035         } else
1036                 delta = NULL;
1037
1038         memset(&s, 0, sizeof(s));
1039         deflateInit(&s, zlib_compression_level);
1040         if (delta) {
1041                 s.next_in = delta;
1042                 s.avail_in = deltalen;
1043         } else {
1044                 s.next_in = (void *)dat->buf;
1045                 s.avail_in = dat->len;
1046         }
1047         s.avail_out = deflateBound(&s, s.avail_in);
1048         s.next_out = out = xmalloc(s.avail_out);
1049         while (deflate(&s, Z_FINISH) == Z_OK)
1050                 /* nothing */;
1051         deflateEnd(&s);
1052
1053         /* Determine if we should auto-checkpoint. */
1054         if ((pack_size + 60 + s.total_out) > max_packsize
1055                 || (pack_size + 60 + s.total_out) < pack_size) {
1056
1057                 /* This new object needs to *not* have the current pack_id. */
1058                 e->pack_id = pack_id + 1;
1059                 cycle_packfile();
1060
1061                 /* We cannot carry a delta into the new pack. */
1062                 if (delta) {
1063                         free(delta);
1064                         delta = NULL;
1065
1066                         memset(&s, 0, sizeof(s));
1067                         deflateInit(&s, zlib_compression_level);
1068                         s.next_in = (void *)dat->buf;
1069                         s.avail_in = dat->len;
1070                         s.avail_out = deflateBound(&s, s.avail_in);
1071                         s.next_out = out = xrealloc(out, s.avail_out);
1072                         while (deflate(&s, Z_FINISH) == Z_OK)
1073                                 /* nothing */;
1074                         deflateEnd(&s);
1075                 }
1076         }
1077
1078         e->type = type;
1079         e->pack_id = pack_id;
1080         e->offset = pack_size;
1081         object_count++;
1082         object_count_by_type[type]++;
1083
1084         if (delta) {
1085                 unsigned long ofs = e->offset - last->offset;
1086                 unsigned pos = sizeof(hdr) - 1;
1087
1088                 delta_count_by_type[type]++;
1089                 last->depth++;
1090
1091                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
1092                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1093                 pack_size += hdrlen;
1094
1095                 hdr[pos] = ofs & 127;
1096                 while (ofs >>= 7)
1097                         hdr[--pos] = 128 | (--ofs & 127);
1098                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
1099                 pack_size += sizeof(hdr) - pos;
1100         } else {
1101                 if (last)
1102                         last->depth = 0;
1103                 hdrlen = encode_header(type, dat->len, hdr);
1104                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1105                 pack_size += hdrlen;
1106         }
1107
1108         write_or_die(pack_data->pack_fd, out, s.total_out);
1109         pack_size += s.total_out;
1110
1111         free(out);
1112         free(delta);
1113         if (last) {
1114                 if (!last->no_free)
1115                         free(last->data);
1116                 last->offset = e->offset;
1117                 last->data = dat->buf;
1118                 last->len = dat->len;
1119         }
1120         return 0;
1121 }
1122
1123 static void *gfi_unpack_entry(
1124         struct object_entry *oe,
1125         unsigned long *sizep)
1126 {
1127         enum object_type type;
1128         struct packed_git *p = all_packs[oe->pack_id];
1129         if (p == pack_data)
1130                 p->pack_size = pack_size + 20;
1131         return unpack_entry(p, oe->offset, &type, sizep);
1132 }
1133
1134 static const char *get_mode(const char *str, uint16_t *modep)
1135 {
1136         unsigned char c;
1137         uint16_t mode = 0;
1138
1139         while ((c = *str++) != ' ') {
1140                 if (c < '0' || c > '7')
1141                         return NULL;
1142                 mode = (mode << 3) + (c - '0');
1143         }
1144         *modep = mode;
1145         return str;
1146 }
1147
1148 static void load_tree(struct tree_entry *root)
1149 {
1150         unsigned char* sha1 = root->versions[1].sha1;
1151         struct object_entry *myoe;
1152         struct tree_content *t;
1153         unsigned long size;
1154         char *buf;
1155         const char *c;
1156
1157         root->tree = t = new_tree_content(8);
1158         if (is_null_sha1(sha1))
1159                 return;
1160
1161         myoe = find_object(sha1);
1162         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1163                 if (myoe->type != OBJ_TREE)
1164                         die("Not a tree: %s", sha1_to_hex(sha1));
1165                 t->delta_depth = 0;
1166                 buf = gfi_unpack_entry(myoe, &size);
1167         } else {
1168                 enum object_type type;
1169                 buf = read_sha1_file(sha1, &type, &size);
1170                 if (!buf || type != OBJ_TREE)
1171                         die("Can't load tree %s", sha1_to_hex(sha1));
1172         }
1173
1174         c = buf;
1175         while (c != (buf + size)) {
1176                 struct tree_entry *e = new_tree_entry();
1177
1178                 if (t->entry_count == t->entry_capacity)
1179                         root->tree = t = grow_tree_content(t, t->entry_count);
1180                 t->entries[t->entry_count++] = e;
1181
1182                 e->tree = NULL;
1183                 c = get_mode(c, &e->versions[1].mode);
1184                 if (!c)
1185                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1186                 e->versions[0].mode = e->versions[1].mode;
1187                 e->name = to_atom(c, strlen(c));
1188                 c += e->name->str_len + 1;
1189                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1190                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1191                 c += 20;
1192         }
1193         free(buf);
1194 }
1195
1196 static int tecmp0 (const void *_a, const void *_b)
1197 {
1198         struct tree_entry *a = *((struct tree_entry**)_a);
1199         struct tree_entry *b = *((struct tree_entry**)_b);
1200         return base_name_compare(
1201                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1202                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1203 }
1204
1205 static int tecmp1 (const void *_a, const void *_b)
1206 {
1207         struct tree_entry *a = *((struct tree_entry**)_a);
1208         struct tree_entry *b = *((struct tree_entry**)_b);
1209         return base_name_compare(
1210                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1211                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1212 }
1213
1214 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1215 {
1216         size_t maxlen = 0;
1217         unsigned int i;
1218
1219         if (!v)
1220                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1221         else
1222                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1223
1224         for (i = 0; i < t->entry_count; i++) {
1225                 if (t->entries[i]->versions[v].mode)
1226                         maxlen += t->entries[i]->name->str_len + 34;
1227         }
1228
1229         strbuf_reset(b);
1230         strbuf_grow(b, maxlen);
1231         for (i = 0; i < t->entry_count; i++) {
1232                 struct tree_entry *e = t->entries[i];
1233                 if (!e->versions[v].mode)
1234                         continue;
1235                 strbuf_addf(b, "%o %s%c", (unsigned int)e->versions[v].mode,
1236                                         e->name->str_dat, '\0');
1237                 strbuf_add(b, e->versions[v].sha1, 20);
1238         }
1239 }
1240
1241 static void store_tree(struct tree_entry *root)
1242 {
1243         struct tree_content *t = root->tree;
1244         unsigned int i, j, del;
1245         struct last_object lo;
1246         struct object_entry *le;
1247
1248         if (!is_null_sha1(root->versions[1].sha1))
1249                 return;
1250
1251         for (i = 0; i < t->entry_count; i++) {
1252                 if (t->entries[i]->tree)
1253                         store_tree(t->entries[i]);
1254         }
1255
1256         le = find_object(root->versions[0].sha1);
1257         if (!S_ISDIR(root->versions[0].mode)
1258                 || !le
1259                 || le->pack_id != pack_id) {
1260                 lo.data = NULL;
1261                 lo.depth = 0;
1262                 lo.no_free = 0;
1263         } else {
1264                 mktree(t, 0, &old_tree);
1265                 lo.len  = old_tree.len;
1266                 lo.data = old_tree.buf;
1267                 lo.offset = le->offset;
1268                 lo.depth = t->delta_depth;
1269                 lo.no_free = 1;
1270         }
1271
1272         mktree(t, 1, &new_tree);
1273         store_object(OBJ_TREE, &new_tree, &lo, root->versions[1].sha1, 0);
1274
1275         t->delta_depth = lo.depth;
1276         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1277                 struct tree_entry *e = t->entries[i];
1278                 if (e->versions[1].mode) {
1279                         e->versions[0].mode = e->versions[1].mode;
1280                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1281                         t->entries[j++] = e;
1282                 } else {
1283                         release_tree_entry(e);
1284                         del++;
1285                 }
1286         }
1287         t->entry_count -= del;
1288 }
1289
1290 static int tree_content_set(
1291         struct tree_entry *root,
1292         const char *p,
1293         const unsigned char *sha1,
1294         const uint16_t mode,
1295         struct tree_content *subtree)
1296 {
1297         struct tree_content *t = root->tree;
1298         const char *slash1;
1299         unsigned int i, n;
1300         struct tree_entry *e;
1301
1302         slash1 = strchr(p, '/');
1303         if (slash1)
1304                 n = slash1 - p;
1305         else
1306                 n = strlen(p);
1307         if (!n)
1308                 die("Empty path component found in input");
1309         if (!slash1 && !S_ISDIR(mode) && subtree)
1310                 die("Non-directories cannot have subtrees");
1311
1312         for (i = 0; i < t->entry_count; i++) {
1313                 e = t->entries[i];
1314                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1315                         if (!slash1) {
1316                                 if (!S_ISDIR(mode)
1317                                                 && e->versions[1].mode == mode
1318                                                 && !hashcmp(e->versions[1].sha1, sha1))
1319                                         return 0;
1320                                 e->versions[1].mode = mode;
1321                                 hashcpy(e->versions[1].sha1, sha1);
1322                                 if (e->tree)
1323                                         release_tree_content_recursive(e->tree);
1324                                 e->tree = subtree;
1325                                 hashclr(root->versions[1].sha1);
1326                                 return 1;
1327                         }
1328                         if (!S_ISDIR(e->versions[1].mode)) {
1329                                 e->tree = new_tree_content(8);
1330                                 e->versions[1].mode = S_IFDIR;
1331                         }
1332                         if (!e->tree)
1333                                 load_tree(e);
1334                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1335                                 hashclr(root->versions[1].sha1);
1336                                 return 1;
1337                         }
1338                         return 0;
1339                 }
1340         }
1341
1342         if (t->entry_count == t->entry_capacity)
1343                 root->tree = t = grow_tree_content(t, t->entry_count);
1344         e = new_tree_entry();
1345         e->name = to_atom(p, n);
1346         e->versions[0].mode = 0;
1347         hashclr(e->versions[0].sha1);
1348         t->entries[t->entry_count++] = e;
1349         if (slash1) {
1350                 e->tree = new_tree_content(8);
1351                 e->versions[1].mode = S_IFDIR;
1352                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1353         } else {
1354                 e->tree = subtree;
1355                 e->versions[1].mode = mode;
1356                 hashcpy(e->versions[1].sha1, sha1);
1357         }
1358         hashclr(root->versions[1].sha1);
1359         return 1;
1360 }
1361
1362 static int tree_content_remove(
1363         struct tree_entry *root,
1364         const char *p,
1365         struct tree_entry *backup_leaf)
1366 {
1367         struct tree_content *t = root->tree;
1368         const char *slash1;
1369         unsigned int i, n;
1370         struct tree_entry *e;
1371
1372         slash1 = strchr(p, '/');
1373         if (slash1)
1374                 n = slash1 - p;
1375         else
1376                 n = strlen(p);
1377
1378         for (i = 0; i < t->entry_count; i++) {
1379                 e = t->entries[i];
1380                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1381                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1382                                 goto del_entry;
1383                         if (!e->tree)
1384                                 load_tree(e);
1385                         if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1386                                 for (n = 0; n < e->tree->entry_count; n++) {
1387                                         if (e->tree->entries[n]->versions[1].mode) {
1388                                                 hashclr(root->versions[1].sha1);
1389                                                 return 1;
1390                                         }
1391                                 }
1392                                 backup_leaf = NULL;
1393                                 goto del_entry;
1394                         }
1395                         return 0;
1396                 }
1397         }
1398         return 0;
1399
1400 del_entry:
1401         if (backup_leaf)
1402                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1403         else if (e->tree)
1404                 release_tree_content_recursive(e->tree);
1405         e->tree = NULL;
1406         e->versions[1].mode = 0;
1407         hashclr(e->versions[1].sha1);
1408         hashclr(root->versions[1].sha1);
1409         return 1;
1410 }
1411
1412 static int tree_content_get(
1413         struct tree_entry *root,
1414         const char *p,
1415         struct tree_entry *leaf)
1416 {
1417         struct tree_content *t = root->tree;
1418         const char *slash1;
1419         unsigned int i, n;
1420         struct tree_entry *e;
1421
1422         slash1 = strchr(p, '/');
1423         if (slash1)
1424                 n = slash1 - p;
1425         else
1426                 n = strlen(p);
1427
1428         for (i = 0; i < t->entry_count; i++) {
1429                 e = t->entries[i];
1430                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1431                         if (!slash1) {
1432                                 memcpy(leaf, e, sizeof(*leaf));
1433                                 if (e->tree && is_null_sha1(e->versions[1].sha1))
1434                                         leaf->tree = dup_tree_content(e->tree);
1435                                 else
1436                                         leaf->tree = NULL;
1437                                 return 1;
1438                         }
1439                         if (!S_ISDIR(e->versions[1].mode))
1440                                 return 0;
1441                         if (!e->tree)
1442                                 load_tree(e);
1443                         return tree_content_get(e, slash1 + 1, leaf);
1444                 }
1445         }
1446         return 0;
1447 }
1448
1449 static int update_branch(struct branch *b)
1450 {
1451         static const char *msg = "fast-import";
1452         struct ref_lock *lock;
1453         unsigned char old_sha1[20];
1454
1455         if (read_ref(b->name, old_sha1))
1456                 hashclr(old_sha1);
1457         lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1458         if (!lock)
1459                 return error("Unable to lock %s", b->name);
1460         if (!force_update && !is_null_sha1(old_sha1)) {
1461                 struct commit *old_cmit, *new_cmit;
1462
1463                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1464                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1465                 if (!old_cmit || !new_cmit) {
1466                         unlock_ref(lock);
1467                         return error("Branch %s is missing commits.", b->name);
1468                 }
1469
1470                 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1471                         unlock_ref(lock);
1472                         warning("Not updating %s"
1473                                 " (new tip %s does not contain %s)",
1474                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1475                         return -1;
1476                 }
1477         }
1478         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1479                 return error("Unable to update %s", b->name);
1480         return 0;
1481 }
1482
1483 static void dump_branches(void)
1484 {
1485         unsigned int i;
1486         struct branch *b;
1487
1488         for (i = 0; i < branch_table_sz; i++) {
1489                 for (b = branch_table[i]; b; b = b->table_next_branch)
1490                         failure |= update_branch(b);
1491         }
1492 }
1493
1494 static void dump_tags(void)
1495 {
1496         static const char *msg = "fast-import";
1497         struct tag *t;
1498         struct ref_lock *lock;
1499         char ref_name[PATH_MAX];
1500
1501         for (t = first_tag; t; t = t->next_tag) {
1502                 sprintf(ref_name, "tags/%s", t->name);
1503                 lock = lock_ref_sha1(ref_name, NULL);
1504                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1505                         failure |= error("Unable to update %s", ref_name);
1506         }
1507 }
1508
1509 static void dump_marks_helper(FILE *f,
1510         uintmax_t base,
1511         struct mark_set *m)
1512 {
1513         uintmax_t k;
1514         if (m->shift) {
1515                 for (k = 0; k < 1024; k++) {
1516                         if (m->data.sets[k])
1517                                 dump_marks_helper(f, (base + k) << m->shift,
1518                                         m->data.sets[k]);
1519                 }
1520         } else {
1521                 for (k = 0; k < 1024; k++) {
1522                         if (m->data.marked[k])
1523                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1524                                         sha1_to_hex(m->data.marked[k]->sha1));
1525                 }
1526         }
1527 }
1528
1529 static void dump_marks(void)
1530 {
1531         static struct lock_file mark_lock;
1532         int mark_fd;
1533         FILE *f;
1534
1535         if (!mark_file)
1536                 return;
1537
1538         mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1539         if (mark_fd < 0) {
1540                 failure |= error("Unable to write marks file %s: %s",
1541                         mark_file, strerror(errno));
1542                 return;
1543         }
1544
1545         f = fdopen(mark_fd, "w");
1546         if (!f) {
1547                 rollback_lock_file(&mark_lock);
1548                 failure |= error("Unable to write marks file %s: %s",
1549                         mark_file, strerror(errno));
1550                 return;
1551         }
1552
1553         dump_marks_helper(f, 0, marks);
1554         fclose(f);
1555         if (commit_lock_file(&mark_lock))
1556                 failure |= error("Unable to write marks file %s: %s",
1557                         mark_file, strerror(errno));
1558 }
1559
1560 static int read_next_command(void)
1561 {
1562         static int stdin_eof = 0;
1563
1564         if (stdin_eof) {
1565                 unread_command_buf = 0;
1566                 return EOF;
1567         }
1568
1569         do {
1570                 if (unread_command_buf) {
1571                         unread_command_buf = 0;
1572                 } else {
1573                         struct recent_command *rc;
1574
1575                         strbuf_detach(&command_buf);
1576                         stdin_eof = strbuf_getline(&command_buf, stdin, '\n');
1577                         if (stdin_eof)
1578                                 return EOF;
1579
1580                         rc = rc_free;
1581                         if (rc)
1582                                 rc_free = rc->next;
1583                         else {
1584                                 rc = cmd_hist.next;
1585                                 cmd_hist.next = rc->next;
1586                                 cmd_hist.next->prev = &cmd_hist;
1587                                 free(rc->buf);
1588                         }
1589
1590                         rc->buf = command_buf.buf;
1591                         rc->prev = cmd_tail;
1592                         rc->next = cmd_hist.prev;
1593                         rc->prev->next = rc;
1594                         cmd_tail = rc;
1595                 }
1596         } while (command_buf.buf[0] == '#');
1597
1598         return 0;
1599 }
1600
1601 static void skip_optional_lf(void)
1602 {
1603         int term_char = fgetc(stdin);
1604         if (term_char != '\n' && term_char != EOF)
1605                 ungetc(term_char, stdin);
1606 }
1607
1608 static void cmd_mark(void)
1609 {
1610         if (!prefixcmp(command_buf.buf, "mark :")) {
1611                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1612                 read_next_command();
1613         }
1614         else
1615                 next_mark = 0;
1616 }
1617
1618 static void cmd_data(struct strbuf *sb)
1619 {
1620         strbuf_reset(sb);
1621
1622         if (prefixcmp(command_buf.buf, "data "))
1623                 die("Expected 'data n' command, found: %s", command_buf.buf);
1624
1625         if (!prefixcmp(command_buf.buf + 5, "<<")) {
1626                 char *term = xstrdup(command_buf.buf + 5 + 2);
1627                 size_t term_len = command_buf.len - 5 - 2;
1628
1629                 for (;;) {
1630                         if (strbuf_getline(&command_buf, stdin, '\n') == EOF)
1631                                 die("EOF in data (terminator '%s' not found)", term);
1632                         if (term_len == command_buf.len
1633                                 && !strcmp(term, command_buf.buf))
1634                                 break;
1635                         strbuf_addbuf(sb, &command_buf);
1636                         strbuf_addch(sb, '\n');
1637                 }
1638                 free(term);
1639         }
1640         else {
1641                 size_t n = 0, length;
1642
1643                 length = strtoul(command_buf.buf + 5, NULL, 10);
1644
1645                 while (n < length) {
1646                         size_t s = strbuf_fread(sb, length - n, stdin);
1647                         if (!s && feof(stdin))
1648                                 die("EOF in data (%lu bytes remaining)",
1649                                         (unsigned long)(length - n));
1650                         n += s;
1651                 }
1652         }
1653
1654         skip_optional_lf();
1655 }
1656
1657 static int validate_raw_date(const char *src, char *result, int maxlen)
1658 {
1659         const char *orig_src = src;
1660         char *endp, sign;
1661
1662         strtoul(src, &endp, 10);
1663         if (endp == src || *endp != ' ')
1664                 return -1;
1665
1666         src = endp + 1;
1667         if (*src != '-' && *src != '+')
1668                 return -1;
1669         sign = *src;
1670
1671         strtoul(src + 1, &endp, 10);
1672         if (endp == src || *endp || (endp - orig_src) >= maxlen)
1673                 return -1;
1674
1675         strcpy(result, orig_src);
1676         return 0;
1677 }
1678
1679 static char *parse_ident(const char *buf)
1680 {
1681         const char *gt;
1682         size_t name_len;
1683         char *ident;
1684
1685         gt = strrchr(buf, '>');
1686         if (!gt)
1687                 die("Missing > in ident string: %s", buf);
1688         gt++;
1689         if (*gt != ' ')
1690                 die("Missing space after > in ident string: %s", buf);
1691         gt++;
1692         name_len = gt - buf;
1693         ident = xmalloc(name_len + 24);
1694         strncpy(ident, buf, name_len);
1695
1696         switch (whenspec) {
1697         case WHENSPEC_RAW:
1698                 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1699                         die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1700                 break;
1701         case WHENSPEC_RFC2822:
1702                 if (parse_date(gt, ident + name_len, 24) < 0)
1703                         die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1704                 break;
1705         case WHENSPEC_NOW:
1706                 if (strcmp("now", gt))
1707                         die("Date in ident must be 'now': %s", buf);
1708                 datestamp(ident + name_len, 24);
1709                 break;
1710         }
1711
1712         return ident;
1713 }
1714
1715 static void cmd_new_blob(void)
1716 {
1717         struct strbuf buf;
1718
1719         read_next_command();
1720         cmd_mark();
1721         strbuf_init(&buf, 0);
1722         cmd_data(&buf);
1723         if (store_object(OBJ_BLOB, &buf, &last_blob, NULL, next_mark))
1724                 strbuf_release(&buf);
1725 }
1726
1727 static void unload_one_branch(void)
1728 {
1729         while (cur_active_branches
1730                 && cur_active_branches >= max_active_branches) {
1731                 uintmax_t min_commit = ULONG_MAX;
1732                 struct branch *e, *l = NULL, *p = NULL;
1733
1734                 for (e = active_branches; e; e = e->active_next_branch) {
1735                         if (e->last_commit < min_commit) {
1736                                 p = l;
1737                                 min_commit = e->last_commit;
1738                         }
1739                         l = e;
1740                 }
1741
1742                 if (p) {
1743                         e = p->active_next_branch;
1744                         p->active_next_branch = e->active_next_branch;
1745                 } else {
1746                         e = active_branches;
1747                         active_branches = e->active_next_branch;
1748                 }
1749                 e->active = 0;
1750                 e->active_next_branch = NULL;
1751                 if (e->branch_tree.tree) {
1752                         release_tree_content_recursive(e->branch_tree.tree);
1753                         e->branch_tree.tree = NULL;
1754                 }
1755                 cur_active_branches--;
1756         }
1757 }
1758
1759 static void load_branch(struct branch *b)
1760 {
1761         load_tree(&b->branch_tree);
1762         if (!b->active) {
1763                 b->active = 1;
1764                 b->active_next_branch = active_branches;
1765                 active_branches = b;
1766                 cur_active_branches++;
1767                 branch_load_count++;
1768         }
1769 }
1770
1771 static void file_change_m(struct branch *b)
1772 {
1773         const char *p = command_buf.buf + 2;
1774         char *p_uq;
1775         const char *endp;
1776         struct object_entry *oe = oe;
1777         unsigned char sha1[20];
1778         uint16_t mode, inline_data = 0;
1779
1780         p = get_mode(p, &mode);
1781         if (!p)
1782                 die("Corrupt mode: %s", command_buf.buf);
1783         switch (mode) {
1784         case S_IFREG | 0644:
1785         case S_IFREG | 0755:
1786         case S_IFLNK:
1787         case 0644:
1788         case 0755:
1789                 /* ok */
1790                 break;
1791         default:
1792                 die("Corrupt mode: %s", command_buf.buf);
1793         }
1794
1795         if (*p == ':') {
1796                 char *x;
1797                 oe = find_mark(strtoumax(p + 1, &x, 10));
1798                 hashcpy(sha1, oe->sha1);
1799                 p = x;
1800         } else if (!prefixcmp(p, "inline")) {
1801                 inline_data = 1;
1802                 p += 6;
1803         } else {
1804                 if (get_sha1_hex(p, sha1))
1805                         die("Invalid SHA1: %s", command_buf.buf);
1806                 oe = find_object(sha1);
1807                 p += 40;
1808         }
1809         if (*p++ != ' ')
1810                 die("Missing space after SHA1: %s", command_buf.buf);
1811
1812         p_uq = unquote_c_style(p, &endp);
1813         if (p_uq) {
1814                 if (*endp)
1815                         die("Garbage after path in: %s", command_buf.buf);
1816                 p = p_uq;
1817         }
1818
1819         if (inline_data) {
1820                 struct strbuf buf;
1821
1822                 if (!p_uq)
1823                         p = p_uq = xstrdup(p);
1824                 read_next_command();
1825                 strbuf_init(&buf, 0);
1826                 cmd_data(&buf);
1827                 if (store_object(OBJ_BLOB, &buf, &last_blob, sha1, 0))
1828                         strbuf_release(&buf);
1829         } else if (oe) {
1830                 if (oe->type != OBJ_BLOB)
1831                         die("Not a blob (actually a %s): %s",
1832                                 command_buf.buf, typename(oe->type));
1833         } else {
1834                 enum object_type type = sha1_object_info(sha1, NULL);
1835                 if (type < 0)
1836                         die("Blob not found: %s", command_buf.buf);
1837                 if (type != OBJ_BLOB)
1838                         die("Not a blob (actually a %s): %s",
1839                             typename(type), command_buf.buf);
1840         }
1841
1842         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1843         free(p_uq);
1844 }
1845
1846 static void file_change_d(struct branch *b)
1847 {
1848         const char *p = command_buf.buf + 2;
1849         char *p_uq;
1850         const char *endp;
1851
1852         p_uq = unquote_c_style(p, &endp);
1853         if (p_uq) {
1854                 if (*endp)
1855                         die("Garbage after path in: %s", command_buf.buf);
1856                 p = p_uq;
1857         }
1858         tree_content_remove(&b->branch_tree, p, NULL);
1859         free(p_uq);
1860 }
1861
1862 static void file_change_cr(struct branch *b, int rename)
1863 {
1864         const char *s, *d;
1865         char *s_uq, *d_uq;
1866         const char *endp;
1867         struct tree_entry leaf;
1868
1869         s = command_buf.buf + 2;
1870         s_uq = unquote_c_style(s, &endp);
1871         if (s_uq) {
1872                 if (*endp != ' ')
1873                         die("Missing space after source: %s", command_buf.buf);
1874         }
1875         else {
1876                 endp = strchr(s, ' ');
1877                 if (!endp)
1878                         die("Missing space after source: %s", command_buf.buf);
1879                 s_uq = xmalloc(endp - s + 1);
1880                 memcpy(s_uq, s, endp - s);
1881                 s_uq[endp - s] = 0;
1882         }
1883         s = s_uq;
1884
1885         endp++;
1886         if (!*endp)
1887                 die("Missing dest: %s", command_buf.buf);
1888
1889         d = endp;
1890         d_uq = unquote_c_style(d, &endp);
1891         if (d_uq) {
1892                 if (*endp)
1893                         die("Garbage after dest in: %s", command_buf.buf);
1894                 d = d_uq;
1895         }
1896
1897         memset(&leaf, 0, sizeof(leaf));
1898         if (rename)
1899                 tree_content_remove(&b->branch_tree, s, &leaf);
1900         else
1901                 tree_content_get(&b->branch_tree, s, &leaf);
1902         if (!leaf.versions[1].mode)
1903                 die("Path %s not in branch", s);
1904         tree_content_set(&b->branch_tree, d,
1905                 leaf.versions[1].sha1,
1906                 leaf.versions[1].mode,
1907                 leaf.tree);
1908
1909         free(s_uq);
1910         free(d_uq);
1911 }
1912
1913 static void file_change_deleteall(struct branch *b)
1914 {
1915         release_tree_content_recursive(b->branch_tree.tree);
1916         hashclr(b->branch_tree.versions[0].sha1);
1917         hashclr(b->branch_tree.versions[1].sha1);
1918         load_tree(&b->branch_tree);
1919 }
1920
1921 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1922 {
1923         if (!buf || size < 46)
1924                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1925         if (memcmp("tree ", buf, 5)
1926                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1927                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1928         hashcpy(b->branch_tree.versions[0].sha1,
1929                 b->branch_tree.versions[1].sha1);
1930 }
1931
1932 static void cmd_from_existing(struct branch *b)
1933 {
1934         if (is_null_sha1(b->sha1)) {
1935                 hashclr(b->branch_tree.versions[0].sha1);
1936                 hashclr(b->branch_tree.versions[1].sha1);
1937         } else {
1938                 unsigned long size;
1939                 char *buf;
1940
1941                 buf = read_object_with_reference(b->sha1,
1942                         commit_type, &size, b->sha1);
1943                 cmd_from_commit(b, buf, size);
1944                 free(buf);
1945         }
1946 }
1947
1948 static int cmd_from(struct branch *b)
1949 {
1950         const char *from;
1951         struct branch *s;
1952
1953         if (prefixcmp(command_buf.buf, "from "))
1954                 return 0;
1955
1956         if (b->branch_tree.tree) {
1957                 release_tree_content_recursive(b->branch_tree.tree);
1958                 b->branch_tree.tree = NULL;
1959         }
1960
1961         from = strchr(command_buf.buf, ' ') + 1;
1962         s = lookup_branch(from);
1963         if (b == s)
1964                 die("Can't create a branch from itself: %s", b->name);
1965         else if (s) {
1966                 unsigned char *t = s->branch_tree.versions[1].sha1;
1967                 hashcpy(b->sha1, s->sha1);
1968                 hashcpy(b->branch_tree.versions[0].sha1, t);
1969                 hashcpy(b->branch_tree.versions[1].sha1, t);
1970         } else if (*from == ':') {
1971                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
1972                 struct object_entry *oe = find_mark(idnum);
1973                 if (oe->type != OBJ_COMMIT)
1974                         die("Mark :%" PRIuMAX " not a commit", idnum);
1975                 hashcpy(b->sha1, oe->sha1);
1976                 if (oe->pack_id != MAX_PACK_ID) {
1977                         unsigned long size;
1978                         char *buf = gfi_unpack_entry(oe, &size);
1979                         cmd_from_commit(b, buf, size);
1980                         free(buf);
1981                 } else
1982                         cmd_from_existing(b);
1983         } else if (!get_sha1(from, b->sha1))
1984                 cmd_from_existing(b);
1985         else
1986                 die("Invalid ref name or SHA1 expression: %s", from);
1987
1988         read_next_command();
1989         return 1;
1990 }
1991
1992 static struct hash_list *cmd_merge(unsigned int *count)
1993 {
1994         struct hash_list *list = NULL, *n, *e = e;
1995         const char *from;
1996         struct branch *s;
1997
1998         *count = 0;
1999         while (!prefixcmp(command_buf.buf, "merge ")) {
2000                 from = strchr(command_buf.buf, ' ') + 1;
2001                 n = xmalloc(sizeof(*n));
2002                 s = lookup_branch(from);
2003                 if (s)
2004                         hashcpy(n->sha1, s->sha1);
2005                 else if (*from == ':') {
2006                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
2007                         struct object_entry *oe = find_mark(idnum);
2008                         if (oe->type != OBJ_COMMIT)
2009                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2010                         hashcpy(n->sha1, oe->sha1);
2011                 } else if (!get_sha1(from, n->sha1)) {
2012                         unsigned long size;
2013                         char *buf = read_object_with_reference(n->sha1,
2014                                 commit_type, &size, n->sha1);
2015                         if (!buf || size < 46)
2016                                 die("Not a valid commit: %s", from);
2017                         free(buf);
2018                 } else
2019                         die("Invalid ref name or SHA1 expression: %s", from);
2020
2021                 n->next = NULL;
2022                 if (list)
2023                         e->next = n;
2024                 else
2025                         list = n;
2026                 e = n;
2027                 (*count)++;
2028                 read_next_command();
2029         }
2030         return list;
2031 }
2032
2033 static void cmd_new_commit(void)
2034 {
2035         static struct strbuf msg = STRBUF_INIT;
2036         struct branch *b;
2037         char *sp;
2038         char *author = NULL;
2039         char *committer = NULL;
2040         struct hash_list *merge_list = NULL;
2041         unsigned int merge_count;
2042
2043         /* Obtain the branch name from the rest of our command */
2044         sp = strchr(command_buf.buf, ' ') + 1;
2045         b = lookup_branch(sp);
2046         if (!b)
2047                 b = new_branch(sp);
2048
2049         read_next_command();
2050         cmd_mark();
2051         if (!prefixcmp(command_buf.buf, "author ")) {
2052                 author = parse_ident(command_buf.buf + 7);
2053                 read_next_command();
2054         }
2055         if (!prefixcmp(command_buf.buf, "committer ")) {
2056                 committer = parse_ident(command_buf.buf + 10);
2057                 read_next_command();
2058         }
2059         if (!committer)
2060                 die("Expected committer but didn't get one");
2061         cmd_data(&msg);
2062         read_next_command();
2063         cmd_from(b);
2064         merge_list = cmd_merge(&merge_count);
2065
2066         /* ensure the branch is active/loaded */
2067         if (!b->branch_tree.tree || !max_active_branches) {
2068                 unload_one_branch();
2069                 load_branch(b);
2070         }
2071
2072         /* file_change* */
2073         while (command_buf.len > 0) {
2074                 if (!prefixcmp(command_buf.buf, "M "))
2075                         file_change_m(b);
2076                 else if (!prefixcmp(command_buf.buf, "D "))
2077                         file_change_d(b);
2078                 else if (!prefixcmp(command_buf.buf, "R "))
2079                         file_change_cr(b, 1);
2080                 else if (!prefixcmp(command_buf.buf, "C "))
2081                         file_change_cr(b, 0);
2082                 else if (!strcmp("deleteall", command_buf.buf))
2083                         file_change_deleteall(b);
2084                 else {
2085                         unread_command_buf = 1;
2086                         break;
2087                 }
2088                 if (read_next_command() == EOF)
2089                         break;
2090         }
2091
2092         /* build the tree and the commit */
2093         store_tree(&b->branch_tree);
2094         hashcpy(b->branch_tree.versions[0].sha1,
2095                 b->branch_tree.versions[1].sha1);
2096
2097         strbuf_reset(&new_data);
2098         strbuf_addf(&new_data, "tree %s\n",
2099                 sha1_to_hex(b->branch_tree.versions[1].sha1));
2100         if (!is_null_sha1(b->sha1))
2101                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(b->sha1));
2102         while (merge_list) {
2103                 struct hash_list *next = merge_list->next;
2104                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(merge_list->sha1));
2105                 free(merge_list);
2106                 merge_list = next;
2107         }
2108         strbuf_addf(&new_data,
2109                 "author %s\n"
2110                 "committer %s\n"
2111                 "\n",
2112                 author ? author : committer, committer);
2113         strbuf_addbuf(&new_data, &msg);
2114         free(author);
2115         free(committer);
2116
2117         if (!store_object(OBJ_COMMIT, &new_data, NULL, b->sha1, next_mark))
2118                 b->pack_id = pack_id;
2119         b->last_commit = object_count_by_type[OBJ_COMMIT];
2120 }
2121
2122 static void cmd_new_tag(void)
2123 {
2124         static struct strbuf msg = STRBUF_INIT;
2125         char *sp;
2126         const char *from;
2127         char *tagger;
2128         struct branch *s;
2129         struct tag *t;
2130         uintmax_t from_mark = 0;
2131         unsigned char sha1[20];
2132
2133         /* Obtain the new tag name from the rest of our command */
2134         sp = strchr(command_buf.buf, ' ') + 1;
2135         t = pool_alloc(sizeof(struct tag));
2136         t->next_tag = NULL;
2137         t->name = pool_strdup(sp);
2138         if (last_tag)
2139                 last_tag->next_tag = t;
2140         else
2141                 first_tag = t;
2142         last_tag = t;
2143         read_next_command();
2144
2145         /* from ... */
2146         if (prefixcmp(command_buf.buf, "from "))
2147                 die("Expected from command, got %s", command_buf.buf);
2148         from = strchr(command_buf.buf, ' ') + 1;
2149         s = lookup_branch(from);
2150         if (s) {
2151                 hashcpy(sha1, s->sha1);
2152         } else if (*from == ':') {
2153                 struct object_entry *oe;
2154                 from_mark = strtoumax(from + 1, NULL, 10);
2155                 oe = find_mark(from_mark);
2156                 if (oe->type != OBJ_COMMIT)
2157                         die("Mark :%" PRIuMAX " not a commit", from_mark);
2158                 hashcpy(sha1, oe->sha1);
2159         } else if (!get_sha1(from, sha1)) {
2160                 unsigned long size;
2161                 char *buf;
2162
2163                 buf = read_object_with_reference(sha1,
2164                         commit_type, &size, sha1);
2165                 if (!buf || size < 46)
2166                         die("Not a valid commit: %s", from);
2167                 free(buf);
2168         } else
2169                 die("Invalid ref name or SHA1 expression: %s", from);
2170         read_next_command();
2171
2172         /* tagger ... */
2173         if (prefixcmp(command_buf.buf, "tagger "))
2174                 die("Expected tagger command, got %s", command_buf.buf);
2175         tagger = parse_ident(command_buf.buf + 7);
2176
2177         /* tag payload/message */
2178         read_next_command();
2179         cmd_data(&msg);
2180
2181         /* build the tag object */
2182         strbuf_reset(&new_data);
2183         strbuf_addf(&new_data,
2184                 "object %s\n"
2185                 "type %s\n"
2186                 "tag %s\n"
2187                 "tagger %s\n"
2188                 "\n",
2189                 sha1_to_hex(sha1), commit_type, t->name, tagger);
2190         strbuf_addbuf(&new_data, &msg);
2191         free(tagger);
2192
2193         if (store_object(OBJ_TAG, &new_data, NULL, t->sha1, 0))
2194                 t->pack_id = MAX_PACK_ID;
2195         else
2196                 t->pack_id = pack_id;
2197 }
2198
2199 static void cmd_reset_branch(void)
2200 {
2201         struct branch *b;
2202         char *sp;
2203
2204         /* Obtain the branch name from the rest of our command */
2205         sp = strchr(command_buf.buf, ' ') + 1;
2206         b = lookup_branch(sp);
2207         if (b) {
2208                 hashclr(b->sha1);
2209                 hashclr(b->branch_tree.versions[0].sha1);
2210                 hashclr(b->branch_tree.versions[1].sha1);
2211                 if (b->branch_tree.tree) {
2212                         release_tree_content_recursive(b->branch_tree.tree);
2213                         b->branch_tree.tree = NULL;
2214                 }
2215         }
2216         else
2217                 b = new_branch(sp);
2218         read_next_command();
2219         if (!cmd_from(b) && command_buf.len > 0)
2220                 unread_command_buf = 1;
2221 }
2222
2223 static void cmd_checkpoint(void)
2224 {
2225         if (object_count) {
2226                 cycle_packfile();
2227                 dump_branches();
2228                 dump_tags();
2229                 dump_marks();
2230         }
2231         skip_optional_lf();
2232 }
2233
2234 static void cmd_progress(void)
2235 {
2236         fwrite(command_buf.buf, 1, command_buf.len, stdout);
2237         fputc('\n', stdout);
2238         fflush(stdout);
2239         skip_optional_lf();
2240 }
2241
2242 static void import_marks(const char *input_file)
2243 {
2244         char line[512];
2245         FILE *f = fopen(input_file, "r");
2246         if (!f)
2247                 die("cannot read %s: %s", input_file, strerror(errno));
2248         while (fgets(line, sizeof(line), f)) {
2249                 uintmax_t mark;
2250                 char *end;
2251                 unsigned char sha1[20];
2252                 struct object_entry *e;
2253
2254                 end = strchr(line, '\n');
2255                 if (line[0] != ':' || !end)
2256                         die("corrupt mark line: %s", line);
2257                 *end = 0;
2258                 mark = strtoumax(line + 1, &end, 10);
2259                 if (!mark || end == line + 1
2260                         || *end != ' ' || get_sha1(end + 1, sha1))
2261                         die("corrupt mark line: %s", line);
2262                 e = find_object(sha1);
2263                 if (!e) {
2264                         enum object_type type = sha1_object_info(sha1, NULL);
2265                         if (type < 0)
2266                                 die("object not found: %s", sha1_to_hex(sha1));
2267                         e = insert_object(sha1);
2268                         e->type = type;
2269                         e->pack_id = MAX_PACK_ID;
2270                         e->offset = 1; /* just not zero! */
2271                 }
2272                 insert_mark(mark, e);
2273         }
2274         fclose(f);
2275 }
2276
2277 static const char fast_import_usage[] =
2278 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2279
2280 int main(int argc, const char **argv)
2281 {
2282         unsigned int i, show_stats = 1;
2283
2284         git_config(git_default_config);
2285         alloc_objects(object_entry_alloc);
2286         strbuf_init(&command_buf, 0);
2287         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2288         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2289         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2290         marks = pool_calloc(1, sizeof(struct mark_set));
2291
2292         for (i = 1; i < argc; i++) {
2293                 const char *a = argv[i];
2294
2295                 if (*a != '-' || !strcmp(a, "--"))
2296                         break;
2297                 else if (!prefixcmp(a, "--date-format=")) {
2298                         const char *fmt = a + 14;
2299                         if (!strcmp(fmt, "raw"))
2300                                 whenspec = WHENSPEC_RAW;
2301                         else if (!strcmp(fmt, "rfc2822"))
2302                                 whenspec = WHENSPEC_RFC2822;
2303                         else if (!strcmp(fmt, "now"))
2304                                 whenspec = WHENSPEC_NOW;
2305                         else
2306                                 die("unknown --date-format argument %s", fmt);
2307                 }
2308                 else if (!prefixcmp(a, "--max-pack-size="))
2309                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2310                 else if (!prefixcmp(a, "--depth="))
2311                         max_depth = strtoul(a + 8, NULL, 0);
2312                 else if (!prefixcmp(a, "--active-branches="))
2313                         max_active_branches = strtoul(a + 18, NULL, 0);
2314                 else if (!prefixcmp(a, "--import-marks="))
2315                         import_marks(a + 15);
2316                 else if (!prefixcmp(a, "--export-marks="))
2317                         mark_file = a + 15;
2318                 else if (!prefixcmp(a, "--export-pack-edges=")) {
2319                         if (pack_edges)
2320                                 fclose(pack_edges);
2321                         pack_edges = fopen(a + 20, "a");
2322                         if (!pack_edges)
2323                                 die("Cannot open %s: %s", a + 20, strerror(errno));
2324                 } else if (!strcmp(a, "--force"))
2325                         force_update = 1;
2326                 else if (!strcmp(a, "--quiet"))
2327                         show_stats = 0;
2328                 else if (!strcmp(a, "--stats"))
2329                         show_stats = 1;
2330                 else
2331                         die("unknown option %s", a);
2332         }
2333         if (i != argc)
2334                 usage(fast_import_usage);
2335
2336         rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
2337         for (i = 0; i < (cmd_save - 1); i++)
2338                 rc_free[i].next = &rc_free[i + 1];
2339         rc_free[cmd_save - 1].next = NULL;
2340
2341         prepare_packed_git();
2342         start_packfile();
2343         set_die_routine(die_nicely);
2344         while (read_next_command() != EOF) {
2345                 if (!strcmp("blob", command_buf.buf))
2346                         cmd_new_blob();
2347                 else if (!prefixcmp(command_buf.buf, "commit "))
2348                         cmd_new_commit();
2349                 else if (!prefixcmp(command_buf.buf, "tag "))
2350                         cmd_new_tag();
2351                 else if (!prefixcmp(command_buf.buf, "reset "))
2352                         cmd_reset_branch();
2353                 else if (!strcmp("checkpoint", command_buf.buf))
2354                         cmd_checkpoint();
2355                 else if (!prefixcmp(command_buf.buf, "progress "))
2356                         cmd_progress();
2357                 else
2358                         die("Unsupported command: %s", command_buf.buf);
2359         }
2360         end_packfile();
2361
2362         dump_branches();
2363         dump_tags();
2364         unkeep_all_packs();
2365         dump_marks();
2366
2367         if (pack_edges)
2368                 fclose(pack_edges);
2369
2370         if (show_stats) {
2371                 uintmax_t total_count = 0, duplicate_count = 0;
2372                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2373                         total_count += object_count_by_type[i];
2374                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2375                         duplicate_count += duplicate_count_by_type[i];
2376
2377                 fprintf(stderr, "%s statistics:\n", argv[0]);
2378                 fprintf(stderr, "---------------------------------------------------------------------\n");
2379                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2380                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
2381                 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]);
2382                 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]);
2383                 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]);
2384                 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]);
2385                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2386                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2387                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2388                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2389                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
2390                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2391                 fprintf(stderr, "---------------------------------------------------------------------\n");
2392                 pack_report();
2393                 fprintf(stderr, "---------------------------------------------------------------------\n");
2394                 fprintf(stderr, "\n");
2395         }
2396
2397         return failure ? 1 : 0;
2398 }