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 #define DEPTH_BITS 13
 157 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
 158
 159 struct object_entry
 160 {
 161         struct object_entry *next;
 162         uint32_t offset;
 163         uint32_t type : TYPE_BITS,
 164                 pack_id : PACK_ID_BITS,
 165                 depth : DEPTH_BITS;
 166         unsigned char sha1[20];
 167 };
 168
 169 struct object_entry_pool
 170 {
 171         struct object_entry_pool *next_pool;
 172         struct object_entry *next_free;
 173         struct object_entry *end;
 174         struct object_entry entries[FLEX_ARRAY]; /* more */
 175 };
 176
 177 struct mark_set
 178 {
 179         union {
 180                 struct object_entry *marked[1024];
 181                 struct mark_set *sets[1024];
 182         } data;
 183         unsigned int shift;
 184 };
 185
 186 struct last_object
 187 {
 188         struct strbuf data;
 189         uint32_t offset;
 190         unsigned int depth;
 191         unsigned no_swap : 1;
 192 };
 193
 194 struct mem_pool
 195 {
 196         struct mem_pool *next_pool;
 197         char *next_free;
 198         char *end;
 199         uintmax_t space[FLEX_ARRAY]; /* more */
 200 };
 201
 202 struct atom_str
 203 {
 204         struct atom_str *next_atom;
 205         unsigned short str_len;
 206         char str_dat[FLEX_ARRAY]; /* more */
 207 };
 208
 209 struct tree_content;
 210 struct tree_entry
 211 {
 212         struct tree_content *tree;
 213         struct atom_str* name;
 214         struct tree_entry_ms
 215         {
 216                 uint16_t mode;
 217                 unsigned char sha1[20];
 218         } versions[2];
 219 };
 220
 221 struct tree_content
 222 {
 223         unsigned int entry_capacity; /* must match avail_tree_content */
 224         unsigned int entry_count;
 225         unsigned int delta_depth;
 226         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 227 };
 228
 229 struct avail_tree_content
 230 {
 231         unsigned int entry_capacity; /* must match tree_content */
 232         struct avail_tree_content *next_avail;
 233 };
 234
 235 struct branch
 236 {
 237         struct branch *table_next_branch;
 238         struct branch *active_next_branch;
 239         const char *name;
 240         struct tree_entry branch_tree;
 241         uintmax_t last_commit;
 242         unsigned active : 1;
 243         unsigned pack_id : PACK_ID_BITS;
 244         unsigned char sha1[20];
 245 };
 246
 247 struct tag
 248 {
 249         struct tag *next_tag;
 250         const char *name;
 251         unsigned int pack_id;
 252         unsigned char sha1[20];
 253 };
 254
 255 struct hash_list
 256 {
 257         struct hash_list *next;
 258         unsigned char sha1[20];
 259 };
 260
 261 typedef enum {
 262         WHENSPEC_RAW = 1,
 263         WHENSPEC_RFC2822,
 264         WHENSPEC_NOW,
 265 } whenspec_type;
 266
 267 struct recent_command
 268 {
 269         struct recent_command *prev;
 270         struct recent_command *next;
 271         char *buf;
 272 };
 273
 274 /* Configured limits on output */
 275 static unsigned long max_depth = 10;
 276 static off_t max_packsize = (1LL << 32) - 1;
 277 static int force_update;
 278 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
 279 static int pack_compression_seen;
 280
 281 /* Stats and misc. counters */
 282 static uintmax_t alloc_count;
 283 static uintmax_t marks_set_count;
 284 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 285 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 286 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 287 static unsigned long object_count;
 288 static unsigned long branch_count;
 289 static unsigned long branch_load_count;
 290 static int failure;
 291 static FILE *pack_edges;
 292
 293 /* Memory pools */
 294 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 295 static size_t total_allocd;
 296 static struct mem_pool *mem_pool;
 297
 298 /* Atom management */
 299 static unsigned int atom_table_sz = 4451;
 300 static unsigned int atom_cnt;
 301 static struct atom_str **atom_table;
 302
 303 /* The .pack file being generated */
 304 static unsigned int pack_id;
 305 static struct packed_git *pack_data;
 306 static struct packed_git **all_packs;
 307 static unsigned long pack_size;
 308
 309 /* Table of objects we've written. */
 310 static unsigned int object_entry_alloc = 5000;
 311 static struct object_entry_pool *blocks;
 312 static struct object_entry *object_table[1 << 16];
 313 static struct mark_set *marks;
 314 static const char* mark_file;
 315
 316 /* Our last blob */
 317 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 318
 319 /* Tree management */
 320 static unsigned int tree_entry_alloc = 1000;
 321 static void *avail_tree_entry;
 322 static unsigned int avail_tree_table_sz = 100;
 323 static struct avail_tree_content **avail_tree_table;
 324 static struct strbuf old_tree = STRBUF_INIT;
 325 static struct strbuf new_tree = STRBUF_INIT;
 326
 327 /* Branch data */
 328 static unsigned long max_active_branches = 5;
 329 static unsigned long cur_active_branches;
 330 static unsigned long branch_table_sz = 1039;
 331 static struct branch **branch_table;
 332 static struct branch *active_branches;
 333
 334 /* Tag data */
 335 static struct tag *first_tag;
 336 static struct tag *last_tag;
 337
 338 /* Input stream parsing */
 339 static whenspec_type whenspec = WHENSPEC_RAW;
 340 static struct strbuf command_buf = STRBUF_INIT;
 341 static int unread_command_buf;
 342 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 343 static struct recent_command *cmd_tail = &cmd_hist;
 344 static struct recent_command *rc_free;
 345 static unsigned int cmd_save = 100;
 346 static uintmax_t next_mark;
 347 static struct strbuf new_data = STRBUF_INIT;
 348
 349 static void write_branch_report(FILE *rpt, struct branch *b)
 350 {
 351         fprintf(rpt, "%s:\n", b->name);
 352
 353         fprintf(rpt, "  status      :");
 354         if (b->active)
 355                 fputs(" active", rpt);
 356         if (b->branch_tree.tree)
 357                 fputs(" loaded", rpt);
 358         if (is_null_sha1(b->branch_tree.versions[1].sha1))
 359                 fputs(" dirty", rpt);
 360         fputc('\n', rpt);
 361
 362         fprintf(rpt, "  tip commit  : %s\n", sha1_to_hex(b->sha1));
 363         fprintf(rpt, "  old tree    : %s\n", sha1_to_hex(b->branch_tree.versions[0].sha1));
 364         fprintf(rpt, "  cur tree    : %s\n", sha1_to_hex(b->branch_tree.versions[1].sha1));
 365         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 366
 367         fputs("  last pack   : ", rpt);
 368         if (b->pack_id < MAX_PACK_ID)
 369                 fprintf(rpt, "%u", b->pack_id);
 370         fputc('\n', rpt);
 371
 372         fputc('\n', rpt);
 373 }
 374
 375 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
 376
 377 static void write_crash_report(const char *err)
 378 {
 379         char *loc = git_path("fast_import_crash_%d", getpid());
 380         FILE *rpt = fopen(loc, "w");
 381         struct branch *b;
 382         unsigned long lu;
 383         struct recent_command *rc;
 384
 385         if (!rpt) {
 386                 error("can't write crash report %s: %s", loc, strerror(errno));
 387                 return;
 388         }
 389
 390         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 391
 392         fprintf(rpt, "fast-import crash report:\n");
 393         fprintf(rpt, "    fast-import process: %d\n", getpid());
 394         fprintf(rpt, "    parent process     : %d\n", getppid());
 395         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_LOCAL));
 396         fputc('\n', rpt);
 397
 398         fputs("fatal: ", rpt);
 399         fputs(err, rpt);
 400         fputc('\n', rpt);
 401
 402         fputc('\n', rpt);
 403         fputs("Most Recent Commands Before Crash\n", rpt);
 404         fputs("---------------------------------\n", rpt);
 405         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 406                 if (rc->next == &cmd_hist)
 407                         fputs("* ", rpt);
 408                 else
 409                         fputs("  ", rpt);
 410                 fputs(rc->buf, rpt);
 411                 fputc('\n', rpt);
 412         }
 413
 414         fputc('\n', rpt);
 415         fputs("Active Branch LRU\n", rpt);
 416         fputs("-----------------\n", rpt);
 417         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 418                 cur_active_branches,
 419                 max_active_branches);
 420         fputc('\n', rpt);
 421         fputs("  pos  clock name\n", rpt);
 422         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 423         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 424                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 425                         ++lu, b->last_commit, b->name);
 426
 427         fputc('\n', rpt);
 428         fputs("Inactive Branches\n", rpt);
 429         fputs("-----------------\n", rpt);
 430         for (lu = 0; lu < branch_table_sz; lu++) {
 431                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 432                         write_branch_report(rpt, b);
 433         }
 434
 435         if (first_tag) {
 436                 struct tag *tg;
 437                 fputc('\n', rpt);
 438                 fputs("Annotated Tags\n", rpt);
 439                 fputs("--------------\n", rpt);
 440                 for (tg = first_tag; tg; tg = tg->next_tag) {
 441                         fputs(sha1_to_hex(tg->sha1), rpt);
 442                         fputc(' ', rpt);
 443                         fputs(tg->name, rpt);
 444                         fputc('\n', rpt);
 445                 }
 446         }
 447
 448         fputc('\n', rpt);
 449         fputs("Marks\n", rpt);
 450         fputs("-----\n", rpt);
 451         if (mark_file)
 452                 fprintf(rpt, "  exported to %s\n", mark_file);
 453         else
 454                 dump_marks_helper(rpt, 0, marks);
 455
 456         fputc('\n', rpt);
 457         fputs("-------------------\n", rpt);
 458         fputs("END OF CRASH REPORT\n", rpt);
 459         fclose(rpt);
 460 }
 461
 462 static NORETURN void die_nicely(const char *err, va_list params)
 463 {
 464         static int zombie;
 465         char message[2 * PATH_MAX];
 466
 467         vsnprintf(message, sizeof(message), err, params);
 468         fputs("fatal: ", stderr);
 469         fputs(message, stderr);
 470         fputc('\n', stderr);
 471
 472         if (!zombie) {
 473                 zombie = 1;
 474                 write_crash_report(message);
 475         }
 476         exit(128);
 477 }
 478
 479 static void alloc_objects(unsigned int cnt)
 480 {
 481         struct object_entry_pool *b;
 482
 483         b = xmalloc(sizeof(struct object_entry_pool)
 484                 + cnt * sizeof(struct object_entry));
 485         b->next_pool = blocks;
 486         b->next_free = b->entries;
 487         b->end = b->entries + cnt;
 488         blocks = b;
 489         alloc_count += cnt;
 490 }
 491
 492 static struct object_entry *new_object(unsigned char *sha1)
 493 {
 494         struct object_entry *e;
 495
 496         if (blocks->next_free == blocks->end)
 497                 alloc_objects(object_entry_alloc);
 498
 499         e = blocks->next_free++;
 500         hashcpy(e->sha1, sha1);
 501         return e;
 502 }
 503
 504 static struct object_entry *find_object(unsigned char *sha1)
 505 {
 506         unsigned int h = sha1[0] << 8 | sha1[1];
 507         struct object_entry *e;
 508         for (e = object_table[h]; e; e = e->next)
 509                 if (!hashcmp(sha1, e->sha1))
 510                         return e;
 511         return NULL;
 512 }
 513
 514 static struct object_entry *insert_object(unsigned char *sha1)
 515 {
 516         unsigned int h = sha1[0] << 8 | sha1[1];
 517         struct object_entry *e = object_table[h];
 518         struct object_entry *p = NULL;
 519
 520         while (e) {
 521                 if (!hashcmp(sha1, e->sha1))
 522                         return e;
 523                 p = e;
 524                 e = e->next;
 525         }
 526
 527         e = new_object(sha1);
 528         e->next = NULL;
 529         e->offset = 0;
 530         if (p)
 531                 p->next = e;
 532         else
 533                 object_table[h] = e;
 534         return e;
 535 }
 536
 537 static unsigned int hc_str(const char *s, size_t len)
 538 {
 539         unsigned int r = 0;
 540         while (len-- > 0)
 541                 r = r * 31 + *s++;
 542         return r;
 543 }
 544
 545 static void *pool_alloc(size_t len)
 546 {
 547         struct mem_pool *p;
 548         void *r;
 549
 550         for (p = mem_pool; p; p = p->next_pool)
 551                 if ((p->end - p->next_free >= len))
 552                         break;
 553
 554         if (!p) {
 555                 if (len >= (mem_pool_alloc/2)) {
 556                         total_allocd += len;
 557                         return xmalloc(len);
 558                 }
 559                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 560                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 561                 p->next_pool = mem_pool;
 562                 p->next_free = (char *) p->space;
 563                 p->end = p->next_free + mem_pool_alloc;
 564                 mem_pool = p;
 565         }
 566
 567         r = p->next_free;
 568         /* round out to a 'uintmax_t' alignment */
 569         if (len & (sizeof(uintmax_t) - 1))
 570                 len += sizeof(uintmax_t) - (len & (sizeof(uintmax_t) - 1));
 571         p->next_free += len;
 572         return r;
 573 }
 574
 575 static void *pool_calloc(size_t count, size_t size)
 576 {
 577         size_t len = count * size;
 578         void *r = pool_alloc(len);
 579         memset(r, 0, len);
 580         return r;
 581 }
 582
 583 static char *pool_strdup(const char *s)
 584 {
 585         char *r = pool_alloc(strlen(s) + 1);
 586         strcpy(r, s);
 587         return r;
 588 }
 589
 590 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 591 {
 592         struct mark_set *s = marks;
 593         while ((idnum >> s->shift) >= 1024) {
 594                 s = pool_calloc(1, sizeof(struct mark_set));
 595                 s->shift = marks->shift + 10;
 596                 s->data.sets[0] = marks;
 597                 marks = s;
 598         }
 599         while (s->shift) {
 600                 uintmax_t i = idnum >> s->shift;
 601                 idnum -= i << s->shift;
 602                 if (!s->data.sets[i]) {
 603                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 604                         s->data.sets[i]->shift = s->shift - 10;
 605                 }
 606                 s = s->data.sets[i];
 607         }
 608         if (!s->data.marked[idnum])
 609                 marks_set_count++;
 610         s->data.marked[idnum] = oe;
 611 }
 612
 613 static struct object_entry *find_mark(uintmax_t idnum)
 614 {
 615         uintmax_t orig_idnum = idnum;
 616         struct mark_set *s = marks;
 617         struct object_entry *oe = NULL;
 618         if ((idnum >> s->shift) < 1024) {
 619                 while (s && s->shift) {
 620                         uintmax_t i = idnum >> s->shift;
 621                         idnum -= i << s->shift;
 622                         s = s->data.sets[i];
 623                 }
 624                 if (s)
 625                         oe = s->data.marked[idnum];
 626         }
 627         if (!oe)
 628                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 629         return oe;
 630 }
 631
 632 static struct atom_str *to_atom(const char *s, unsigned short len)
 633 {
 634         unsigned int hc = hc_str(s, len) % atom_table_sz;
 635         struct atom_str *c;
 636
 637         for (c = atom_table[hc]; c; c = c->next_atom)
 638                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 639                         return c;
 640
 641         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 642         c->str_len = len;
 643         strncpy(c->str_dat, s, len);
 644         c->str_dat[len] = 0;
 645         c->next_atom = atom_table[hc];
 646         atom_table[hc] = c;
 647         atom_cnt++;
 648         return c;
 649 }
 650
 651 static struct branch *lookup_branch(const char *name)
 652 {
 653         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 654         struct branch *b;
 655
 656         for (b = branch_table[hc]; b; b = b->table_next_branch)
 657                 if (!strcmp(name, b->name))
 658                         return b;
 659         return NULL;
 660 }
 661
 662 static struct branch *new_branch(const char *name)
 663 {
 664         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 665         struct branch* b = lookup_branch(name);
 666
 667         if (b)
 668                 die("Invalid attempt to create duplicate branch: %s", name);
 669         switch (check_ref_format(name)) {
 670         case 0: break; /* its valid */
 671         case CHECK_REF_FORMAT_ONELEVEL:
 672                 break; /* valid, but too few '/', allow anyway */
 673         default:
 674                 die("Branch name doesn't conform to GIT standards: %s", name);
 675         }
 676
 677         b = pool_calloc(1, sizeof(struct branch));
 678         b->name = pool_strdup(name);
 679         b->table_next_branch = branch_table[hc];
 680         b->branch_tree.versions[0].mode = S_IFDIR;
 681         b->branch_tree.versions[1].mode = S_IFDIR;
 682         b->active = 0;
 683         b->pack_id = MAX_PACK_ID;
 684         branch_table[hc] = b;
 685         branch_count++;
 686         return b;
 687 }
 688
 689 static unsigned int hc_entries(unsigned int cnt)
 690 {
 691         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 692         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 693 }
 694
 695 static struct tree_content *new_tree_content(unsigned int cnt)
 696 {
 697         struct avail_tree_content *f, *l = NULL;
 698         struct tree_content *t;
 699         unsigned int hc = hc_entries(cnt);
 700
 701         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 702                 if (f->entry_capacity >= cnt)
 703                         break;
 704
 705         if (f) {
 706                 if (l)
 707                         l->next_avail = f->next_avail;
 708                 else
 709                         avail_tree_table[hc] = f->next_avail;
 710         } else {
 711                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 712                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 713                 f->entry_capacity = cnt;
 714         }
 715
 716         t = (struct tree_content*)f;
 717         t->entry_count = 0;
 718         t->delta_depth = 0;
 719         return t;
 720 }
 721
 722 static void release_tree_entry(struct tree_entry *e);
 723 static void release_tree_content(struct tree_content *t)
 724 {
 725         struct avail_tree_content *f = (struct avail_tree_content*)t;
 726         unsigned int hc = hc_entries(f->entry_capacity);
 727         f->next_avail = avail_tree_table[hc];
 728         avail_tree_table[hc] = f;
 729 }
 730
 731 static void release_tree_content_recursive(struct tree_content *t)
 732 {
 733         unsigned int i;
 734         for (i = 0; i < t->entry_count; i++)
 735                 release_tree_entry(t->entries[i]);
 736         release_tree_content(t);
 737 }
 738
 739 static struct tree_content *grow_tree_content(
 740         struct tree_content *t,
 741         int amt)
 742 {
 743         struct tree_content *r = new_tree_content(t->entry_count + amt);
 744         r->entry_count = t->entry_count;
 745         r->delta_depth = t->delta_depth;
 746         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 747         release_tree_content(t);
 748         return r;
 749 }
 750
 751 static struct tree_entry *new_tree_entry(void)
 752 {
 753         struct tree_entry *e;
 754
 755         if (!avail_tree_entry) {
 756                 unsigned int n = tree_entry_alloc;
 757                 total_allocd += n * sizeof(struct tree_entry);
 758                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 759                 while (n-- > 1) {
 760                         *((void**)e) = e + 1;
 761                         e++;
 762                 }
 763                 *((void**)e) = NULL;
 764         }
 765
 766         e = avail_tree_entry;
 767         avail_tree_entry = *((void**)e);
 768         return e;
 769 }
 770
 771 static void release_tree_entry(struct tree_entry *e)
 772 {
 773         if (e->tree)
 774                 release_tree_content_recursive(e->tree);
 775         *((void**)e) = avail_tree_entry;
 776         avail_tree_entry = e;
 777 }
 778
 779 static struct tree_content *dup_tree_content(struct tree_content *s)
 780 {
 781         struct tree_content *d;
 782         struct tree_entry *a, *b;
 783         unsigned int i;
 784
 785         if (!s)
 786                 return NULL;
 787         d = new_tree_content(s->entry_count);
 788         for (i = 0; i < s->entry_count; i++) {
 789                 a = s->entries[i];
 790                 b = new_tree_entry();
 791                 memcpy(b, a, sizeof(*a));
 792                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 793                         b->tree = dup_tree_content(a->tree);
 794                 else
 795                         b->tree = NULL;
 796                 d->entries[i] = b;
 797         }
 798         d->entry_count = s->entry_count;
 799         d->delta_depth = s->delta_depth;
 800
 801         return d;
 802 }
 803
 804 static void start_packfile(void)
 805 {
 806         static char tmpfile[PATH_MAX];
 807         struct packed_git *p;
 808         struct pack_header hdr;
 809         int pack_fd;
 810
 811         snprintf(tmpfile, sizeof(tmpfile),
 812                 "%s/tmp_pack_XXXXXX", get_object_directory());
 813         pack_fd = xmkstemp(tmpfile);
 814         p = xcalloc(1, sizeof(*p) + strlen(tmpfile) + 2);
 815         strcpy(p->pack_name, tmpfile);
 816         p->pack_fd = pack_fd;
 817
 818         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 819         hdr.hdr_version = htonl(2);
 820         hdr.hdr_entries = 0;
 821         write_or_die(p->pack_fd, &hdr, sizeof(hdr));
 822
 823         pack_data = p;
 824         pack_size = sizeof(hdr);
 825         object_count = 0;
 826
 827         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 828         all_packs[pack_id] = p;
 829 }
 830
 831 static int oecmp (const void *a_, const void *b_)
 832 {
 833         struct object_entry *a = *((struct object_entry**)a_);
 834         struct object_entry *b = *((struct object_entry**)b_);
 835         return hashcmp(a->sha1, b->sha1);
 836 }
 837
 838 static char *create_index(void)
 839 {
 840         static char tmpfile[PATH_MAX];
 841         SHA_CTX ctx;
 842         struct sha1file *f;
 843         struct object_entry **idx, **c, **last, *e;
 844         struct object_entry_pool *o;
 845         uint32_t array[256];
 846         int i, idx_fd;
 847
 848         /* Build the sorted table of object IDs. */
 849         idx = xmalloc(object_count * sizeof(struct object_entry*));
 850         c = idx;
 851         for (o = blocks; o; o = o->next_pool)
 852                 for (e = o->next_free; e-- != o->entries;)
 853                         if (pack_id == e->pack_id)
 854                                 *c++ = e;
 855         last = idx + object_count;
 856         if (c != last)
 857                 die("internal consistency error creating the index");
 858         qsort(idx, object_count, sizeof(struct object_entry*), oecmp);
 859
 860         /* Generate the fan-out array. */
 861         c = idx;
 862         for (i = 0; i < 256; i++) {
 863                 struct object_entry **next = c;;
 864                 while (next < last) {
 865                         if ((*next)->sha1[0] != i)
 866                                 break;
 867                         next++;
 868                 }
 869                 array[i] = htonl(next - idx);
 870                 c = next;
 871         }
 872
 873         snprintf(tmpfile, sizeof(tmpfile),
 874                 "%s/tmp_idx_XXXXXX", get_object_directory());
 875         idx_fd = xmkstemp(tmpfile);
 876         f = sha1fd(idx_fd, tmpfile);
 877         sha1write(f, array, 256 * sizeof(int));
 878         SHA1_Init(&ctx);
 879         for (c = idx; c != last; c++) {
 880                 uint32_t offset = htonl((*c)->offset);
 881                 sha1write(f, &offset, 4);
 882                 sha1write(f, (*c)->sha1, sizeof((*c)->sha1));
 883                 SHA1_Update(&ctx, (*c)->sha1, 20);
 884         }
 885         sha1write(f, pack_data->sha1, sizeof(pack_data->sha1));
 886         sha1close(f, NULL, 1);
 887         free(idx);
 888         SHA1_Final(pack_data->sha1, &ctx);
 889         return tmpfile;
 890 }
 891
 892 static char *keep_pack(char *curr_index_name)
 893 {
 894         static char name[PATH_MAX];
 895         static const char *keep_msg = "fast-import";
 896         int keep_fd;
 897
 898         chmod(pack_data->pack_name, 0444);
 899         chmod(curr_index_name, 0444);
 900
 901         snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 902                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 903         keep_fd = open(name, O_RDWR|O_CREAT|O_EXCL, 0600);
 904         if (keep_fd < 0)
 905                 die("cannot create keep file");
 906         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 907         if (close(keep_fd))
 908                 die("failed to write keep file");
 909
 910         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 911                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 912         if (move_temp_to_file(pack_data->pack_name, name))
 913                 die("cannot store pack file");
 914
 915         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 916                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 917         if (move_temp_to_file(curr_index_name, name))
 918                 die("cannot store index file");
 919         return name;
 920 }
 921
 922 static void unkeep_all_packs(void)
 923 {
 924         static char name[PATH_MAX];
 925         int k;
 926
 927         for (k = 0; k < pack_id; k++) {
 928                 struct packed_git *p = all_packs[k];
 929                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 930                          get_object_directory(), sha1_to_hex(p->sha1));
 931                 unlink(name);
 932         }
 933 }
 934
 935 static void end_packfile(void)
 936 {
 937         struct packed_git *old_p = pack_data, *new_p;
 938
 939         if (object_count) {
 940                 char *idx_name;
 941                 int i;
 942                 struct branch *b;
 943                 struct tag *t;
 944
 945                 close_pack_windows(pack_data);
 946                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 947                                     pack_data->pack_name, object_count);
 948                 close(pack_data->pack_fd);
 949                 idx_name = keep_pack(create_index());
 950
 951                 /* Register the packfile with core git's machinary. */
 952                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 953                 if (!new_p)
 954                         die("core git rejected index %s", idx_name);
 955                 all_packs[pack_id] = new_p;
 956                 install_packed_git(new_p);
 957
 958                 /* Print the boundary */
 959                 if (pack_edges) {
 960                         fprintf(pack_edges, "%s:", new_p->pack_name);
 961                         for (i = 0; i < branch_table_sz; i++) {
 962                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 963                                         if (b->pack_id == pack_id)
 964                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 965                                 }
 966                         }
 967                         for (t = first_tag; t; t = t->next_tag) {
 968                                 if (t->pack_id == pack_id)
 969                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 970                         }
 971                         fputc('\n', pack_edges);
 972                         fflush(pack_edges);
 973                 }
 974
 975                 pack_id++;
 976         }
 977         else
 978                 unlink(old_p->pack_name);
 979         free(old_p);
 980
 981         /* We can't carry a delta across packfiles. */
 982         strbuf_release(&last_blob.data);
 983         last_blob.offset = 0;
 984         last_blob.depth = 0;
 985 }
 986
 987 static void cycle_packfile(void)
 988 {
 989         end_packfile();
 990         start_packfile();
 991 }
 992
 993 static size_t encode_header(
 994         enum object_type type,
 995         size_t size,
 996         unsigned char *hdr)
 997 {
 998         int n = 1;
 999         unsigned char c;
1000
1001         if (type < OBJ_COMMIT || type > OBJ_REF_DELTA)
1002                 die("bad type %d", type);
1003
1004         c = (type << 4) | (size & 15);
1005         size >>= 4;
1006         while (size) {
1007                 *hdr++ = c | 0x80;
1008                 c = size & 0x7f;
1009                 size >>= 7;
1010                 n++;
1011         }
1012         *hdr = c;
1013         return n;
1014 }
1015
1016 static int store_object(
1017         enum object_type type,
1018         struct strbuf *dat,
1019         struct last_object *last,
1020         unsigned char *sha1out,
1021         uintmax_t mark)
1022 {
1023         void *out, *delta;
1024         struct object_entry *e;
1025         unsigned char hdr[96];
1026         unsigned char sha1[20];
1027         unsigned long hdrlen, deltalen;
1028         SHA_CTX c;
1029         z_stream s;
1030
1031         hdrlen = sprintf((char*)hdr,"%s %lu", typename(type),
1032                 (unsigned long)dat->len) + 1;
1033         SHA1_Init(&c);
1034         SHA1_Update(&c, hdr, hdrlen);
1035         SHA1_Update(&c, dat->buf, dat->len);
1036         SHA1_Final(sha1, &c);
1037         if (sha1out)
1038                 hashcpy(sha1out, sha1);
1039
1040         e = insert_object(sha1);
1041         if (mark)
1042                 insert_mark(mark, e);
1043         if (e->offset) {
1044                 duplicate_count_by_type[type]++;
1045                 return 1;
1046         } else if (find_sha1_pack(sha1, packed_git)) {
1047                 e->type = type;
1048                 e->pack_id = MAX_PACK_ID;
1049                 e->offset = 1; /* just not zero! */
1050                 duplicate_count_by_type[type]++;
1051                 return 1;
1052         }
1053
1054         if (last && last->data.buf && last->depth < max_depth) {
1055                 delta = diff_delta(last->data.buf, last->data.len,
1056                         dat->buf, dat->len,
1057                         &deltalen, 0);
1058                 if (delta && deltalen >= dat->len) {
1059                         free(delta);
1060                         delta = NULL;
1061                 }
1062         } else
1063                 delta = NULL;
1064
1065         memset(&s, 0, sizeof(s));
1066         deflateInit(&s, pack_compression_level);
1067         if (delta) {
1068                 s.next_in = delta;
1069                 s.avail_in = deltalen;
1070         } else {
1071                 s.next_in = (void *)dat->buf;
1072                 s.avail_in = dat->len;
1073         }
1074         s.avail_out = deflateBound(&s, s.avail_in);
1075         s.next_out = out = xmalloc(s.avail_out);
1076         while (deflate(&s, Z_FINISH) == Z_OK)
1077                 /* nothing */;
1078         deflateEnd(&s);
1079
1080         /* Determine if we should auto-checkpoint. */
1081         if ((pack_size + 60 + s.total_out) > max_packsize
1082                 || (pack_size + 60 + s.total_out) < pack_size) {
1083
1084                 /* This new object needs to *not* have the current pack_id. */
1085                 e->pack_id = pack_id + 1;
1086                 cycle_packfile();
1087
1088                 /* We cannot carry a delta into the new pack. */
1089                 if (delta) {
1090                         free(delta);
1091                         delta = NULL;
1092
1093                         memset(&s, 0, sizeof(s));
1094                         deflateInit(&s, pack_compression_level);
1095                         s.next_in = (void *)dat->buf;
1096                         s.avail_in = dat->len;
1097                         s.avail_out = deflateBound(&s, s.avail_in);
1098                         s.next_out = out = xrealloc(out, s.avail_out);
1099                         while (deflate(&s, Z_FINISH) == Z_OK)
1100                                 /* nothing */;
1101                         deflateEnd(&s);
1102                 }
1103         }
1104
1105         e->type = type;
1106         e->pack_id = pack_id;
1107         e->offset = pack_size;
1108         object_count++;
1109         object_count_by_type[type]++;
1110
1111         if (delta) {
1112                 unsigned long ofs = e->offset - last->offset;
1113                 unsigned pos = sizeof(hdr) - 1;
1114
1115                 delta_count_by_type[type]++;
1116                 e->depth = last->depth + 1;
1117
1118                 hdrlen = encode_header(OBJ_OFS_DELTA, deltalen, hdr);
1119                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1120                 pack_size += hdrlen;
1121
1122                 hdr[pos] = ofs & 127;
1123                 while (ofs >>= 7)
1124                         hdr[--pos] = 128 | (--ofs & 127);
1125                 write_or_die(pack_data->pack_fd, hdr + pos, sizeof(hdr) - pos);
1126                 pack_size += sizeof(hdr) - pos;
1127         } else {
1128                 e->depth = 0;
1129                 hdrlen = encode_header(type, dat->len, hdr);
1130                 write_or_die(pack_data->pack_fd, hdr, hdrlen);
1131                 pack_size += hdrlen;
1132         }
1133
1134         write_or_die(pack_data->pack_fd, out, s.total_out);
1135         pack_size += s.total_out;
1136
1137         free(out);
1138         free(delta);
1139         if (last) {
1140                 if (last->no_swap) {
1141                         last->data = *dat;
1142                 } else {
1143                         strbuf_swap(&last->data, dat);
1144                 }
1145                 last->offset = e->offset;
1146                 last->depth = e->depth;
1147         }
1148         return 0;
1149 }
1150
1151 /* All calls must be guarded by find_object() or find_mark() to
1152  * ensure the 'struct object_entry' passed was written by this
1153  * process instance.  We unpack the entry by the offset, avoiding
1154  * the need for the corresponding .idx file.  This unpacking rule
1155  * works because we only use OBJ_REF_DELTA within the packfiles
1156  * created by fast-import.
1157  *
1158  * oe must not be NULL.  Such an oe usually comes from giving
1159  * an unknown SHA-1 to find_object() or an undefined mark to
1160  * find_mark().  Callers must test for this condition and use
1161  * the standard read_sha1_file() when it happens.
1162  *
1163  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1164  * find_mark(), where the mark was reloaded from an existing marks
1165  * file and is referencing an object that this fast-import process
1166  * instance did not write out to a packfile.  Callers must test for
1167  * this condition and use read_sha1_file() instead.
1168  */
1169 static void *gfi_unpack_entry(
1170         struct object_entry *oe,
1171         unsigned long *sizep)
1172 {
1173         enum object_type type;
1174         struct packed_git *p = all_packs[oe->pack_id];
1175         if (p == pack_data && p->pack_size < (pack_size + 20)) {
1176                 /* The object is stored in the packfile we are writing to
1177                  * and we have modified it since the last time we scanned
1178                  * back to read a previously written object.  If an old
1179                  * window covered [p->pack_size, p->pack_size + 20) its
1180                  * data is stale and is not valid.  Closing all windows
1181                  * and updating the packfile length ensures we can read
1182                  * the newly written data.
1183                  */
1184                 close_pack_windows(p);
1185
1186                 /* We have to offer 20 bytes additional on the end of
1187                  * the packfile as the core unpacker code assumes the
1188                  * footer is present at the file end and must promise
1189                  * at least 20 bytes within any window it maps.  But
1190                  * we don't actually create the footer here.
1191                  */
1192                 p->pack_size = pack_size + 20;
1193         }
1194         return unpack_entry(p, oe->offset, &type, sizep);
1195 }
1196
1197 static const char *get_mode(const char *str, uint16_t *modep)
1198 {
1199         unsigned char c;
1200         uint16_t mode = 0;
1201
1202         while ((c = *str++) != ' ') {
1203                 if (c < '0' || c > '7')
1204                         return NULL;
1205                 mode = (mode << 3) + (c - '0');
1206         }
1207         *modep = mode;
1208         return str;
1209 }
1210
1211 static void load_tree(struct tree_entry *root)
1212 {
1213         unsigned char* sha1 = root->versions[1].sha1;
1214         struct object_entry *myoe;
1215         struct tree_content *t;
1216         unsigned long size;
1217         char *buf;
1218         const char *c;
1219
1220         root->tree = t = new_tree_content(8);
1221         if (is_null_sha1(sha1))
1222                 return;
1223
1224         myoe = find_object(sha1);
1225         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1226                 if (myoe->type != OBJ_TREE)
1227                         die("Not a tree: %s", sha1_to_hex(sha1));
1228                 t->delta_depth = myoe->depth;
1229                 buf = gfi_unpack_entry(myoe, &size);
1230                 if (!buf)
1231                         die("Can't load tree %s", sha1_to_hex(sha1));
1232         } else {
1233                 enum object_type type;
1234                 buf = read_sha1_file(sha1, &type, &size);
1235                 if (!buf || type != OBJ_TREE)
1236                         die("Can't load tree %s", sha1_to_hex(sha1));
1237         }
1238
1239         c = buf;
1240         while (c != (buf + size)) {
1241                 struct tree_entry *e = new_tree_entry();
1242
1243                 if (t->entry_count == t->entry_capacity)
1244                         root->tree = t = grow_tree_content(t, t->entry_count);
1245                 t->entries[t->entry_count++] = e;
1246
1247                 e->tree = NULL;
1248                 c = get_mode(c, &e->versions[1].mode);
1249                 if (!c)
1250                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1251                 e->versions[0].mode = e->versions[1].mode;
1252                 e->name = to_atom(c, strlen(c));
1253                 c += e->name->str_len + 1;
1254                 hashcpy(e->versions[0].sha1, (unsigned char*)c);
1255                 hashcpy(e->versions[1].sha1, (unsigned char*)c);
1256                 c += 20;
1257         }
1258         free(buf);
1259 }
1260
1261 static int tecmp0 (const void *_a, const void *_b)
1262 {
1263         struct tree_entry *a = *((struct tree_entry**)_a);
1264         struct tree_entry *b = *((struct tree_entry**)_b);
1265         return base_name_compare(
1266                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1267                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1268 }
1269
1270 static int tecmp1 (const void *_a, const void *_b)
1271 {
1272         struct tree_entry *a = *((struct tree_entry**)_a);
1273         struct tree_entry *b = *((struct tree_entry**)_b);
1274         return base_name_compare(
1275                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1276                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1277 }
1278
1279 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1280 {
1281         size_t maxlen = 0;
1282         unsigned int i;
1283
1284         if (!v)
1285                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1286         else
1287                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1288
1289         for (i = 0; i < t->entry_count; i++) {
1290                 if (t->entries[i]->versions[v].mode)
1291                         maxlen += t->entries[i]->name->str_len + 34;
1292         }
1293
1294         strbuf_reset(b);
1295         strbuf_grow(b, maxlen);
1296         for (i = 0; i < t->entry_count; i++) {
1297                 struct tree_entry *e = t->entries[i];
1298                 if (!e->versions[v].mode)
1299                         continue;
1300                 strbuf_addf(b, "%o %s%c", (unsigned int)e->versions[v].mode,
1301                                         e->name->str_dat, '\0');
1302                 strbuf_add(b, e->versions[v].sha1, 20);
1303         }
1304 }
1305
1306 static void store_tree(struct tree_entry *root)
1307 {
1308         struct tree_content *t = root->tree;
1309         unsigned int i, j, del;
1310         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1311         struct object_entry *le;
1312
1313         if (!is_null_sha1(root->versions[1].sha1))
1314                 return;
1315
1316         for (i = 0; i < t->entry_count; i++) {
1317                 if (t->entries[i]->tree)
1318                         store_tree(t->entries[i]);
1319         }
1320
1321         le = find_object(root->versions[0].sha1);
1322         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1323                 mktree(t, 0, &old_tree);
1324                 lo.data = old_tree;
1325                 lo.offset = le->offset;
1326                 lo.depth = t->delta_depth;
1327         }
1328
1329         mktree(t, 1, &new_tree);
1330         store_object(OBJ_TREE, &new_tree, &lo, root->versions[1].sha1, 0);
1331
1332         t->delta_depth = lo.depth;
1333         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1334                 struct tree_entry *e = t->entries[i];
1335                 if (e->versions[1].mode) {
1336                         e->versions[0].mode = e->versions[1].mode;
1337                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1338                         t->entries[j++] = e;
1339                 } else {
1340                         release_tree_entry(e);
1341                         del++;
1342                 }
1343         }
1344         t->entry_count -= del;
1345 }
1346
1347 static int tree_content_set(
1348         struct tree_entry *root,
1349         const char *p,
1350         const unsigned char *sha1,
1351         const uint16_t mode,
1352         struct tree_content *subtree)
1353 {
1354         struct tree_content *t = root->tree;
1355         const char *slash1;
1356         unsigned int i, n;
1357         struct tree_entry *e;
1358
1359         slash1 = strchr(p, '/');
1360         if (slash1)
1361                 n = slash1 - p;
1362         else
1363                 n = strlen(p);
1364         if (!n)
1365                 die("Empty path component found in input");
1366         if (!slash1 && !S_ISDIR(mode) && subtree)
1367                 die("Non-directories cannot have subtrees");
1368
1369         for (i = 0; i < t->entry_count; i++) {
1370                 e = t->entries[i];
1371                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1372                         if (!slash1) {
1373                                 if (!S_ISDIR(mode)
1374                                                 && e->versions[1].mode == mode
1375                                                 && !hashcmp(e->versions[1].sha1, sha1))
1376                                         return 0;
1377                                 e->versions[1].mode = mode;
1378                                 hashcpy(e->versions[1].sha1, sha1);
1379                                 if (e->tree)
1380                                         release_tree_content_recursive(e->tree);
1381                                 e->tree = subtree;
1382                                 hashclr(root->versions[1].sha1);
1383                                 return 1;
1384                         }
1385                         if (!S_ISDIR(e->versions[1].mode)) {
1386                                 e->tree = new_tree_content(8);
1387                                 e->versions[1].mode = S_IFDIR;
1388                         }
1389                         if (!e->tree)
1390                                 load_tree(e);
1391                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1392                                 hashclr(root->versions[1].sha1);
1393                                 return 1;
1394                         }
1395                         return 0;
1396                 }
1397         }
1398
1399         if (t->entry_count == t->entry_capacity)
1400                 root->tree = t = grow_tree_content(t, t->entry_count);
1401         e = new_tree_entry();
1402         e->name = to_atom(p, n);
1403         e->versions[0].mode = 0;
1404         hashclr(e->versions[0].sha1);
1405         t->entries[t->entry_count++] = e;
1406         if (slash1) {
1407                 e->tree = new_tree_content(8);
1408                 e->versions[1].mode = S_IFDIR;
1409                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1410         } else {
1411                 e->tree = subtree;
1412                 e->versions[1].mode = mode;
1413                 hashcpy(e->versions[1].sha1, sha1);
1414         }
1415         hashclr(root->versions[1].sha1);
1416         return 1;
1417 }
1418
1419 static int tree_content_remove(
1420         struct tree_entry *root,
1421         const char *p,
1422         struct tree_entry *backup_leaf)
1423 {
1424         struct tree_content *t = root->tree;
1425         const char *slash1;
1426         unsigned int i, n;
1427         struct tree_entry *e;
1428
1429         slash1 = strchr(p, '/');
1430         if (slash1)
1431                 n = slash1 - p;
1432         else
1433                 n = strlen(p);
1434
1435         for (i = 0; i < t->entry_count; i++) {
1436                 e = t->entries[i];
1437                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1438                         if (!slash1 || !S_ISDIR(e->versions[1].mode))
1439                                 goto del_entry;
1440                         if (!e->tree)
1441                                 load_tree(e);
1442                         if (tree_content_remove(e, slash1 + 1, backup_leaf)) {
1443                                 for (n = 0; n < e->tree->entry_count; n++) {
1444                                         if (e->tree->entries[n]->versions[1].mode) {
1445                                                 hashclr(root->versions[1].sha1);
1446                                                 return 1;
1447                                         }
1448                                 }
1449                                 backup_leaf = NULL;
1450                                 goto del_entry;
1451                         }
1452                         return 0;
1453                 }
1454         }
1455         return 0;
1456
1457 del_entry:
1458         if (backup_leaf)
1459                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1460         else if (e->tree)
1461                 release_tree_content_recursive(e->tree);
1462         e->tree = NULL;
1463         e->versions[1].mode = 0;
1464         hashclr(e->versions[1].sha1);
1465         hashclr(root->versions[1].sha1);
1466         return 1;
1467 }
1468
1469 static int tree_content_get(
1470         struct tree_entry *root,
1471         const char *p,
1472         struct tree_entry *leaf)
1473 {
1474         struct tree_content *t = root->tree;
1475         const char *slash1;
1476         unsigned int i, n;
1477         struct tree_entry *e;
1478
1479         slash1 = strchr(p, '/');
1480         if (slash1)
1481                 n = slash1 - p;
1482         else
1483                 n = strlen(p);
1484
1485         for (i = 0; i < t->entry_count; i++) {
1486                 e = t->entries[i];
1487                 if (e->name->str_len == n && !strncmp(p, e->name->str_dat, n)) {
1488                         if (!slash1) {
1489                                 memcpy(leaf, e, sizeof(*leaf));
1490                                 if (e->tree && is_null_sha1(e->versions[1].sha1))
1491                                         leaf->tree = dup_tree_content(e->tree);
1492                                 else
1493                                         leaf->tree = NULL;
1494                                 return 1;
1495                         }
1496                         if (!S_ISDIR(e->versions[1].mode))
1497                                 return 0;
1498                         if (!e->tree)
1499                                 load_tree(e);
1500                         return tree_content_get(e, slash1 + 1, leaf);
1501                 }
1502         }
1503         return 0;
1504 }
1505
1506 static int update_branch(struct branch *b)
1507 {
1508         static const char *msg = "fast-import";
1509         struct ref_lock *lock;
1510         unsigned char old_sha1[20];
1511
1512         if (read_ref(b->name, old_sha1))
1513                 hashclr(old_sha1);
1514         lock = lock_any_ref_for_update(b->name, old_sha1, 0);
1515         if (!lock)
1516                 return error("Unable to lock %s", b->name);
1517         if (!force_update && !is_null_sha1(old_sha1)) {
1518                 struct commit *old_cmit, *new_cmit;
1519
1520                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1521                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1522                 if (!old_cmit || !new_cmit) {
1523                         unlock_ref(lock);
1524                         return error("Branch %s is missing commits.", b->name);
1525                 }
1526
1527                 if (!in_merge_bases(old_cmit, &new_cmit, 1)) {
1528                         unlock_ref(lock);
1529                         warning("Not updating %s"
1530                                 " (new tip %s does not contain %s)",
1531                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1532                         return -1;
1533                 }
1534         }
1535         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1536                 return error("Unable to update %s", b->name);
1537         return 0;
1538 }
1539
1540 static void dump_branches(void)
1541 {
1542         unsigned int i;
1543         struct branch *b;
1544
1545         for (i = 0; i < branch_table_sz; i++) {
1546                 for (b = branch_table[i]; b; b = b->table_next_branch)
1547                         failure |= update_branch(b);
1548         }
1549 }
1550
1551 static void dump_tags(void)
1552 {
1553         static const char *msg = "fast-import";
1554         struct tag *t;
1555         struct ref_lock *lock;
1556         char ref_name[PATH_MAX];
1557
1558         for (t = first_tag; t; t = t->next_tag) {
1559                 sprintf(ref_name, "tags/%s", t->name);
1560                 lock = lock_ref_sha1(ref_name, NULL);
1561                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1562                         failure |= error("Unable to update %s", ref_name);
1563         }
1564 }
1565
1566 static void dump_marks_helper(FILE *f,
1567         uintmax_t base,
1568         struct mark_set *m)
1569 {
1570         uintmax_t k;
1571         if (m->shift) {
1572                 for (k = 0; k < 1024; k++) {
1573                         if (m->data.sets[k])
1574                                 dump_marks_helper(f, (base + k) << m->shift,
1575                                         m->data.sets[k]);
1576                 }
1577         } else {
1578                 for (k = 0; k < 1024; k++) {
1579                         if (m->data.marked[k])
1580                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1581                                         sha1_to_hex(m->data.marked[k]->sha1));
1582                 }
1583         }
1584 }
1585
1586 static void dump_marks(void)
1587 {
1588         static struct lock_file mark_lock;
1589         int mark_fd;
1590         FILE *f;
1591
1592         if (!mark_file)
1593                 return;
1594
1595         mark_fd = hold_lock_file_for_update(&mark_lock, mark_file, 0);
1596         if (mark_fd < 0) {
1597                 failure |= error("Unable to write marks file %s: %s",
1598                         mark_file, strerror(errno));
1599                 return;
1600         }
1601
1602         f = fdopen(mark_fd, "w");
1603         if (!f) {
1604                 int saved_errno = errno;
1605                 rollback_lock_file(&mark_lock);
1606                 failure |= error("Unable to write marks file %s: %s",
1607                         mark_file, strerror(saved_errno));
1608                 return;
1609         }
1610
1611         /*
1612          * Since the lock file was fdopen()'ed, it should not be close()'ed.
1613          * Assign -1 to the lock file descriptor so that commit_lock_file()
1614          * won't try to close() it.
1615          */
1616         mark_lock.fd = -1;
1617
1618         dump_marks_helper(f, 0, marks);
1619         if (ferror(f) || fclose(f)) {
1620                 int saved_errno = errno;
1621                 rollback_lock_file(&mark_lock);
1622                 failure |= error("Unable to write marks file %s: %s",
1623                         mark_file, strerror(saved_errno));
1624                 return;
1625         }
1626
1627         if (commit_lock_file(&mark_lock)) {
1628                 int saved_errno = errno;
1629                 rollback_lock_file(&mark_lock);
1630                 failure |= error("Unable to commit marks file %s: %s",
1631                         mark_file, strerror(saved_errno));
1632                 return;
1633         }
1634 }
1635
1636 static int read_next_command(void)
1637 {
1638         static int stdin_eof = 0;
1639
1640         if (stdin_eof) {
1641                 unread_command_buf = 0;
1642                 return EOF;
1643         }
1644
1645         do {
1646                 if (unread_command_buf) {
1647                         unread_command_buf = 0;
1648                 } else {
1649                         struct recent_command *rc;
1650
1651                         strbuf_detach(&command_buf, NULL);
1652                         stdin_eof = strbuf_getline(&command_buf, stdin, '\n');
1653                         if (stdin_eof)
1654                                 return EOF;
1655
1656                         rc = rc_free;
1657                         if (rc)
1658                                 rc_free = rc->next;
1659                         else {
1660                                 rc = cmd_hist.next;
1661                                 cmd_hist.next = rc->next;
1662                                 cmd_hist.next->prev = &cmd_hist;
1663                                 free(rc->buf);
1664                         }
1665
1666                         rc->buf = command_buf.buf;
1667                         rc->prev = cmd_tail;
1668                         rc->next = cmd_hist.prev;
1669                         rc->prev->next = rc;
1670                         cmd_tail = rc;
1671                 }
1672         } while (command_buf.buf[0] == '#');
1673
1674         return 0;
1675 }
1676
1677 static void skip_optional_lf(void)
1678 {
1679         int term_char = fgetc(stdin);
1680         if (term_char != '\n' && term_char != EOF)
1681                 ungetc(term_char, stdin);
1682 }
1683
1684 static void cmd_mark(void)
1685 {
1686         if (!prefixcmp(command_buf.buf, "mark :")) {
1687                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1688                 read_next_command();
1689         }
1690         else
1691                 next_mark = 0;
1692 }
1693
1694 static void cmd_data(struct strbuf *sb)
1695 {
1696         strbuf_reset(sb);
1697
1698         if (prefixcmp(command_buf.buf, "data "))
1699                 die("Expected 'data n' command, found: %s", command_buf.buf);
1700
1701         if (!prefixcmp(command_buf.buf + 5, "<<")) {
1702                 char *term = xstrdup(command_buf.buf + 5 + 2);
1703                 size_t term_len = command_buf.len - 5 - 2;
1704
1705                 strbuf_detach(&command_buf, NULL);
1706                 for (;;) {
1707                         if (strbuf_getline(&command_buf, stdin, '\n') == EOF)
1708                                 die("EOF in data (terminator '%s' not found)", term);
1709                         if (term_len == command_buf.len
1710                                 && !strcmp(term, command_buf.buf))
1711                                 break;
1712                         strbuf_addbuf(sb, &command_buf);
1713                         strbuf_addch(sb, '\n');
1714                 }
1715                 free(term);
1716         }
1717         else {
1718                 size_t n = 0, length;
1719
1720                 length = strtoul(command_buf.buf + 5, NULL, 10);
1721
1722                 while (n < length) {
1723                         size_t s = strbuf_fread(sb, length - n, stdin);
1724                         if (!s && feof(stdin))
1725                                 die("EOF in data (%lu bytes remaining)",
1726                                         (unsigned long)(length - n));
1727                         n += s;
1728                 }
1729         }
1730
1731         skip_optional_lf();
1732 }
1733
1734 static int validate_raw_date(const char *src, char *result, int maxlen)
1735 {
1736         const char *orig_src = src;
1737         char *endp, sign;
1738
1739         strtoul(src, &endp, 10);
1740         if (endp == src || *endp != ' ')
1741                 return -1;
1742
1743         src = endp + 1;
1744         if (*src != '-' && *src != '+')
1745                 return -1;
1746         sign = *src;
1747
1748         strtoul(src + 1, &endp, 10);
1749         if (endp == src || *endp || (endp - orig_src) >= maxlen)
1750                 return -1;
1751
1752         strcpy(result, orig_src);
1753         return 0;
1754 }
1755
1756 static char *parse_ident(const char *buf)
1757 {
1758         const char *gt;
1759         size_t name_len;
1760         char *ident;
1761
1762         gt = strrchr(buf, '>');
1763         if (!gt)
1764                 die("Missing > in ident string: %s", buf);
1765         gt++;
1766         if (*gt != ' ')
1767                 die("Missing space after > in ident string: %s", buf);
1768         gt++;
1769         name_len = gt - buf;
1770         ident = xmalloc(name_len + 24);
1771         strncpy(ident, buf, name_len);
1772
1773         switch (whenspec) {
1774         case WHENSPEC_RAW:
1775                 if (validate_raw_date(gt, ident + name_len, 24) < 0)
1776                         die("Invalid raw date \"%s\" in ident: %s", gt, buf);
1777                 break;
1778         case WHENSPEC_RFC2822:
1779                 if (parse_date(gt, ident + name_len, 24) < 0)
1780                         die("Invalid rfc2822 date \"%s\" in ident: %s", gt, buf);
1781                 break;
1782         case WHENSPEC_NOW:
1783                 if (strcmp("now", gt))
1784                         die("Date in ident must be 'now': %s", buf);
1785                 datestamp(ident + name_len, 24);
1786                 break;
1787         }
1788
1789         return ident;
1790 }
1791
1792 static void cmd_new_blob(void)
1793 {
1794         static struct strbuf buf = STRBUF_INIT;
1795
1796         read_next_command();
1797         cmd_mark();
1798         cmd_data(&buf);
1799         store_object(OBJ_BLOB, &buf, &last_blob, NULL, next_mark);
1800 }
1801
1802 static void unload_one_branch(void)
1803 {
1804         while (cur_active_branches
1805                 && cur_active_branches >= max_active_branches) {
1806                 uintmax_t min_commit = ULONG_MAX;
1807                 struct branch *e, *l = NULL, *p = NULL;
1808
1809                 for (e = active_branches; e; e = e->active_next_branch) {
1810                         if (e->last_commit < min_commit) {
1811                                 p = l;
1812                                 min_commit = e->last_commit;
1813                         }
1814                         l = e;
1815                 }
1816
1817                 if (p) {
1818                         e = p->active_next_branch;
1819                         p->active_next_branch = e->active_next_branch;
1820                 } else {
1821                         e = active_branches;
1822                         active_branches = e->active_next_branch;
1823                 }
1824                 e->active = 0;
1825                 e->active_next_branch = NULL;
1826                 if (e->branch_tree.tree) {
1827                         release_tree_content_recursive(e->branch_tree.tree);
1828                         e->branch_tree.tree = NULL;
1829                 }
1830                 cur_active_branches--;
1831         }
1832 }
1833
1834 static void load_branch(struct branch *b)
1835 {
1836         load_tree(&b->branch_tree);
1837         if (!b->active) {
1838                 b->active = 1;
1839                 b->active_next_branch = active_branches;
1840                 active_branches = b;
1841                 cur_active_branches++;
1842                 branch_load_count++;
1843         }
1844 }
1845
1846 static void file_change_m(struct branch *b)
1847 {
1848         const char *p = command_buf.buf + 2;
1849         static struct strbuf uq = STRBUF_INIT;
1850         const char *endp;
1851         struct object_entry *oe = oe;
1852         unsigned char sha1[20];
1853         uint16_t mode, inline_data = 0;
1854
1855         p = get_mode(p, &mode);
1856         if (!p)
1857                 die("Corrupt mode: %s", command_buf.buf);
1858         switch (mode) {
1859         case S_IFREG | 0644:
1860         case S_IFREG | 0755:
1861         case S_IFLNK:
1862         case 0644:
1863         case 0755:
1864                 /* ok */
1865                 break;
1866         default:
1867                 die("Corrupt mode: %s", command_buf.buf);
1868         }
1869
1870         if (*p == ':') {
1871                 char *x;
1872                 oe = find_mark(strtoumax(p + 1, &x, 10));
1873                 hashcpy(sha1, oe->sha1);
1874                 p = x;
1875         } else if (!prefixcmp(p, "inline")) {
1876                 inline_data = 1;
1877                 p += 6;
1878         } else {
1879                 if (get_sha1_hex(p, sha1))
1880                         die("Invalid SHA1: %s", command_buf.buf);
1881                 oe = find_object(sha1);
1882                 p += 40;
1883         }
1884         if (*p++ != ' ')
1885                 die("Missing space after SHA1: %s", command_buf.buf);
1886
1887         strbuf_reset(&uq);
1888         if (!unquote_c_style(&uq, p, &endp)) {
1889                 if (*endp)
1890                         die("Garbage after path in: %s", command_buf.buf);
1891                 p = uq.buf;
1892         }
1893
1894         if (inline_data) {
1895                 static struct strbuf buf = STRBUF_INIT;
1896
1897                 if (p != uq.buf) {
1898                         strbuf_addstr(&uq, p);
1899                         p = uq.buf;
1900                 }
1901                 read_next_command();
1902                 cmd_data(&buf);
1903                 store_object(OBJ_BLOB, &buf, &last_blob, sha1, 0);
1904         } else if (oe) {
1905                 if (oe->type != OBJ_BLOB)
1906                         die("Not a blob (actually a %s): %s",
1907                                 typename(oe->type), command_buf.buf);
1908         } else {
1909                 enum object_type type = sha1_object_info(sha1, NULL);
1910                 if (type < 0)
1911                         die("Blob not found: %s", command_buf.buf);
1912                 if (type != OBJ_BLOB)
1913                         die("Not a blob (actually a %s): %s",
1914                             typename(type), command_buf.buf);
1915         }
1916
1917         tree_content_set(&b->branch_tree, p, sha1, S_IFREG | mode, NULL);
1918 }
1919
1920 static void file_change_d(struct branch *b)
1921 {
1922         const char *p = command_buf.buf + 2;
1923         static struct strbuf uq = STRBUF_INIT;
1924         const char *endp;
1925
1926         strbuf_reset(&uq);
1927         if (!unquote_c_style(&uq, p, &endp)) {
1928                 if (*endp)
1929                         die("Garbage after path in: %s", command_buf.buf);
1930                 p = uq.buf;
1931         }
1932         tree_content_remove(&b->branch_tree, p, NULL);
1933 }
1934
1935 static void file_change_cr(struct branch *b, int rename)
1936 {
1937         const char *s, *d;
1938         static struct strbuf s_uq = STRBUF_INIT;
1939         static struct strbuf d_uq = STRBUF_INIT;
1940         const char *endp;
1941         struct tree_entry leaf;
1942
1943         s = command_buf.buf + 2;
1944         strbuf_reset(&s_uq);
1945         if (!unquote_c_style(&s_uq, s, &endp)) {
1946                 if (*endp != ' ')
1947                         die("Missing space after source: %s", command_buf.buf);
1948         } else {
1949                 endp = strchr(s, ' ');
1950                 if (!endp)
1951                         die("Missing space after source: %s", command_buf.buf);
1952                 strbuf_add(&s_uq, s, endp - s);
1953         }
1954         s = s_uq.buf;
1955
1956         endp++;
1957         if (!*endp)
1958                 die("Missing dest: %s", command_buf.buf);
1959
1960         d = endp;
1961         strbuf_reset(&d_uq);
1962         if (!unquote_c_style(&d_uq, d, &endp)) {
1963                 if (*endp)
1964                         die("Garbage after dest in: %s", command_buf.buf);
1965                 d = d_uq.buf;
1966         }
1967
1968         memset(&leaf, 0, sizeof(leaf));
1969         if (rename)
1970                 tree_content_remove(&b->branch_tree, s, &leaf);
1971         else
1972                 tree_content_get(&b->branch_tree, s, &leaf);
1973         if (!leaf.versions[1].mode)
1974                 die("Path %s not in branch", s);
1975         tree_content_set(&b->branch_tree, d,
1976                 leaf.versions[1].sha1,
1977                 leaf.versions[1].mode,
1978                 leaf.tree);
1979 }
1980
1981 static void file_change_deleteall(struct branch *b)
1982 {
1983         release_tree_content_recursive(b->branch_tree.tree);
1984         hashclr(b->branch_tree.versions[0].sha1);
1985         hashclr(b->branch_tree.versions[1].sha1);
1986         load_tree(&b->branch_tree);
1987 }
1988
1989 static void cmd_from_commit(struct branch *b, char *buf, unsigned long size)
1990 {
1991         if (!buf || size < 46)
1992                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
1993         if (memcmp("tree ", buf, 5)
1994                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
1995                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
1996         hashcpy(b->branch_tree.versions[0].sha1,
1997                 b->branch_tree.versions[1].sha1);
1998 }
1999
2000 static void cmd_from_existing(struct branch *b)
2001 {
2002         if (is_null_sha1(b->sha1)) {
2003                 hashclr(b->branch_tree.versions[0].sha1);
2004                 hashclr(b->branch_tree.versions[1].sha1);
2005         } else {
2006                 unsigned long size;
2007                 char *buf;
2008
2009                 buf = read_object_with_reference(b->sha1,
2010                         commit_type, &size, b->sha1);
2011                 cmd_from_commit(b, buf, size);
2012                 free(buf);
2013         }
2014 }
2015
2016 static int cmd_from(struct branch *b)
2017 {
2018         const char *from;
2019         struct branch *s;
2020
2021         if (prefixcmp(command_buf.buf, "from "))
2022                 return 0;
2023
2024         if (b->branch_tree.tree) {
2025                 release_tree_content_recursive(b->branch_tree.tree);
2026                 b->branch_tree.tree = NULL;
2027         }
2028
2029         from = strchr(command_buf.buf, ' ') + 1;
2030         s = lookup_branch(from);
2031         if (b == s)
2032                 die("Can't create a branch from itself: %s", b->name);
2033         else if (s) {
2034                 unsigned char *t = s->branch_tree.versions[1].sha1;
2035                 hashcpy(b->sha1, s->sha1);
2036                 hashcpy(b->branch_tree.versions[0].sha1, t);
2037                 hashcpy(b->branch_tree.versions[1].sha1, t);
2038         } else if (*from == ':') {
2039                 uintmax_t idnum = strtoumax(from + 1, NULL, 10);
2040                 struct object_entry *oe = find_mark(idnum);
2041                 if (oe->type != OBJ_COMMIT)
2042                         die("Mark :%" PRIuMAX " not a commit", idnum);
2043                 hashcpy(b->sha1, oe->sha1);
2044                 if (oe->pack_id != MAX_PACK_ID) {
2045                         unsigned long size;
2046                         char *buf = gfi_unpack_entry(oe, &size);
2047                         cmd_from_commit(b, buf, size);
2048                         free(buf);
2049                 } else
2050                         cmd_from_existing(b);
2051         } else if (!get_sha1(from, b->sha1))
2052                 cmd_from_existing(b);
2053         else
2054                 die("Invalid ref name or SHA1 expression: %s", from);
2055
2056         read_next_command();
2057         return 1;
2058 }
2059
2060 static struct hash_list *cmd_merge(unsigned int *count)
2061 {
2062         struct hash_list *list = NULL, *n, *e = e;
2063         const char *from;
2064         struct branch *s;
2065
2066         *count = 0;
2067         while (!prefixcmp(command_buf.buf, "merge ")) {
2068                 from = strchr(command_buf.buf, ' ') + 1;
2069                 n = xmalloc(sizeof(*n));
2070                 s = lookup_branch(from);
2071                 if (s)
2072                         hashcpy(n->sha1, s->sha1);
2073                 else if (*from == ':') {
2074                         uintmax_t idnum = strtoumax(from + 1, NULL, 10);
2075                         struct object_entry *oe = find_mark(idnum);
2076                         if (oe->type != OBJ_COMMIT)
2077                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2078                         hashcpy(n->sha1, oe->sha1);
2079                 } else if (!get_sha1(from, n->sha1)) {
2080                         unsigned long size;
2081                         char *buf = read_object_with_reference(n->sha1,
2082                                 commit_type, &size, n->sha1);
2083                         if (!buf || size < 46)
2084                                 die("Not a valid commit: %s", from);
2085                         free(buf);
2086                 } else
2087                         die("Invalid ref name or SHA1 expression: %s", from);
2088
2089                 n->next = NULL;
2090                 if (list)
2091                         e->next = n;
2092                 else
2093                         list = n;
2094                 e = n;
2095                 (*count)++;
2096                 read_next_command();
2097         }
2098         return list;
2099 }
2100
2101 static void cmd_new_commit(void)
2102 {
2103         static struct strbuf msg = STRBUF_INIT;
2104         struct branch *b;
2105         char *sp;
2106         char *author = NULL;
2107         char *committer = NULL;
2108         struct hash_list *merge_list = NULL;
2109         unsigned int merge_count;
2110
2111         /* Obtain the branch name from the rest of our command */
2112         sp = strchr(command_buf.buf, ' ') + 1;
2113         b = lookup_branch(sp);
2114         if (!b)
2115                 b = new_branch(sp);
2116
2117         read_next_command();
2118         cmd_mark();
2119         if (!prefixcmp(command_buf.buf, "author ")) {
2120                 author = parse_ident(command_buf.buf + 7);
2121                 read_next_command();
2122         }
2123         if (!prefixcmp(command_buf.buf, "committer ")) {
2124                 committer = parse_ident(command_buf.buf + 10);
2125                 read_next_command();
2126         }
2127         if (!committer)
2128                 die("Expected committer but didn't get one");
2129         cmd_data(&msg);
2130         read_next_command();
2131         cmd_from(b);
2132         merge_list = cmd_merge(&merge_count);
2133
2134         /* ensure the branch is active/loaded */
2135         if (!b->branch_tree.tree || !max_active_branches) {
2136                 unload_one_branch();
2137                 load_branch(b);
2138         }
2139
2140         /* file_change* */
2141         while (command_buf.len > 0) {
2142                 if (!prefixcmp(command_buf.buf, "M "))
2143                         file_change_m(b);
2144                 else if (!prefixcmp(command_buf.buf, "D "))
2145                         file_change_d(b);
2146                 else if (!prefixcmp(command_buf.buf, "R "))
2147                         file_change_cr(b, 1);
2148                 else if (!prefixcmp(command_buf.buf, "C "))
2149                         file_change_cr(b, 0);
2150                 else if (!strcmp("deleteall", command_buf.buf))
2151                         file_change_deleteall(b);
2152                 else {
2153                         unread_command_buf = 1;
2154                         break;
2155                 }
2156                 if (read_next_command() == EOF)
2157                         break;
2158         }
2159
2160         /* build the tree and the commit */
2161         store_tree(&b->branch_tree);
2162         hashcpy(b->branch_tree.versions[0].sha1,
2163                 b->branch_tree.versions[1].sha1);
2164
2165         strbuf_reset(&new_data);
2166         strbuf_addf(&new_data, "tree %s\n",
2167                 sha1_to_hex(b->branch_tree.versions[1].sha1));
2168         if (!is_null_sha1(b->sha1))
2169                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(b->sha1));
2170         while (merge_list) {
2171                 struct hash_list *next = merge_list->next;
2172                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(merge_list->sha1));
2173                 free(merge_list);
2174                 merge_list = next;
2175         }
2176         strbuf_addf(&new_data,
2177                 "author %s\n"
2178                 "committer %s\n"
2179                 "\n",
2180                 author ? author : committer, committer);
2181         strbuf_addbuf(&new_data, &msg);
2182         free(author);
2183         free(committer);
2184
2185         if (!store_object(OBJ_COMMIT, &new_data, NULL, b->sha1, next_mark))
2186                 b->pack_id = pack_id;
2187         b->last_commit = object_count_by_type[OBJ_COMMIT];
2188 }
2189
2190 static void cmd_new_tag(void)
2191 {
2192         static struct strbuf msg = STRBUF_INIT;
2193         char *sp;
2194         const char *from;
2195         char *tagger;
2196         struct branch *s;
2197         struct tag *t;
2198         uintmax_t from_mark = 0;
2199         unsigned char sha1[20];
2200
2201         /* Obtain the new tag name from the rest of our command */
2202         sp = strchr(command_buf.buf, ' ') + 1;
2203         t = pool_alloc(sizeof(struct tag));
2204         t->next_tag = NULL;
2205         t->name = pool_strdup(sp);
2206         if (last_tag)
2207                 last_tag->next_tag = t;
2208         else
2209                 first_tag = t;
2210         last_tag = t;
2211         read_next_command();
2212
2213         /* from ... */
2214         if (prefixcmp(command_buf.buf, "from "))
2215                 die("Expected from command, got %s", command_buf.buf);
2216         from = strchr(command_buf.buf, ' ') + 1;
2217         s = lookup_branch(from);
2218         if (s) {
2219                 hashcpy(sha1, s->sha1);
2220         } else if (*from == ':') {
2221                 struct object_entry *oe;
2222                 from_mark = strtoumax(from + 1, NULL, 10);
2223                 oe = find_mark(from_mark);
2224                 if (oe->type != OBJ_COMMIT)
2225                         die("Mark :%" PRIuMAX " not a commit", from_mark);
2226                 hashcpy(sha1, oe->sha1);
2227         } else if (!get_sha1(from, sha1)) {
2228                 unsigned long size;
2229                 char *buf;
2230
2231                 buf = read_object_with_reference(sha1,
2232                         commit_type, &size, sha1);
2233                 if (!buf || size < 46)
2234                         die("Not a valid commit: %s", from);
2235                 free(buf);
2236         } else
2237                 die("Invalid ref name or SHA1 expression: %s", from);
2238         read_next_command();
2239
2240         /* tagger ... */
2241         if (prefixcmp(command_buf.buf, "tagger "))
2242                 die("Expected tagger command, got %s", command_buf.buf);
2243         tagger = parse_ident(command_buf.buf + 7);
2244
2245         /* tag payload/message */
2246         read_next_command();
2247         cmd_data(&msg);
2248
2249         /* build the tag object */
2250         strbuf_reset(&new_data);
2251         strbuf_addf(&new_data,
2252                 "object %s\n"
2253                 "type %s\n"
2254                 "tag %s\n"
2255                 "tagger %s\n"
2256                 "\n",
2257                 sha1_to_hex(sha1), commit_type, t->name, tagger);
2258         strbuf_addbuf(&new_data, &msg);
2259         free(tagger);
2260
2261         if (store_object(OBJ_TAG, &new_data, NULL, t->sha1, 0))
2262                 t->pack_id = MAX_PACK_ID;
2263         else
2264                 t->pack_id = pack_id;
2265 }
2266
2267 static void cmd_reset_branch(void)
2268 {
2269         struct branch *b;
2270         char *sp;
2271
2272         /* Obtain the branch name from the rest of our command */
2273         sp = strchr(command_buf.buf, ' ') + 1;
2274         b = lookup_branch(sp);
2275         if (b) {
2276                 hashclr(b->sha1);
2277                 hashclr(b->branch_tree.versions[0].sha1);
2278                 hashclr(b->branch_tree.versions[1].sha1);
2279                 if (b->branch_tree.tree) {
2280                         release_tree_content_recursive(b->branch_tree.tree);
2281                         b->branch_tree.tree = NULL;
2282                 }
2283         }
2284         else
2285                 b = new_branch(sp);
2286         read_next_command();
2287         if (!cmd_from(b) && command_buf.len > 0)
2288                 unread_command_buf = 1;
2289 }
2290
2291 static void cmd_checkpoint(void)
2292 {
2293         if (object_count) {
2294                 cycle_packfile();
2295                 dump_branches();
2296                 dump_tags();
2297                 dump_marks();
2298         }
2299         skip_optional_lf();
2300 }
2301
2302 static void cmd_progress(void)
2303 {
2304         fwrite(command_buf.buf, 1, command_buf.len, stdout);
2305         fputc('\n', stdout);
2306         fflush(stdout);
2307         skip_optional_lf();
2308 }
2309
2310 static void import_marks(const char *input_file)
2311 {
2312         char line[512];
2313         FILE *f = fopen(input_file, "r");
2314         if (!f)
2315                 die("cannot read %s: %s", input_file, strerror(errno));
2316         while (fgets(line, sizeof(line), f)) {
2317                 uintmax_t mark;
2318                 char *end;
2319                 unsigned char sha1[20];
2320                 struct object_entry *e;
2321
2322                 end = strchr(line, '\n');
2323                 if (line[0] != ':' || !end)
2324                         die("corrupt mark line: %s", line);
2325                 *end = 0;
2326                 mark = strtoumax(line + 1, &end, 10);
2327                 if (!mark || end == line + 1
2328                         || *end != ' ' || get_sha1(end + 1, sha1))
2329                         die("corrupt mark line: %s", line);
2330                 e = find_object(sha1);
2331                 if (!e) {
2332                         enum object_type type = sha1_object_info(sha1, NULL);
2333                         if (type < 0)
2334                                 die("object not found: %s", sha1_to_hex(sha1));
2335                         e = insert_object(sha1);
2336                         e->type = type;
2337                         e->pack_id = MAX_PACK_ID;
2338                         e->offset = 1; /* just not zero! */
2339                 }
2340                 insert_mark(mark, e);
2341         }
2342         fclose(f);
2343 }
2344
2345 static int git_pack_config(const char *k, const char *v)
2346 {
2347         if (!strcmp(k, "pack.depth")) {
2348                 max_depth = git_config_int(k, v);
2349                 if (max_depth > MAX_DEPTH)
2350                         max_depth = MAX_DEPTH;
2351                 return 0;
2352         }
2353         if (!strcmp(k, "pack.compression")) {
2354                 int level = git_config_int(k, v);
2355                 if (level == -1)
2356                         level = Z_DEFAULT_COMPRESSION;
2357                 else if (level < 0 || level > Z_BEST_COMPRESSION)
2358                         die("bad pack compression level %d", level);
2359                 pack_compression_level = level;
2360                 pack_compression_seen = 1;
2361                 return 0;
2362         }
2363         return git_default_config(k, v);
2364 }
2365
2366 static const char fast_import_usage[] =
2367 "git-fast-import [--date-format=f] [--max-pack-size=n] [--depth=n] [--active-branches=n] [--export-marks=marks.file]";
2368
2369 int main(int argc, const char **argv)
2370 {
2371         unsigned int i, show_stats = 1;
2372
2373         git_config(git_pack_config);
2374         if (!pack_compression_seen && core_compression_seen)
2375                 pack_compression_level = core_compression_level;
2376
2377         alloc_objects(object_entry_alloc);
2378         strbuf_init(&command_buf, 0);
2379         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
2380         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
2381         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
2382         marks = pool_calloc(1, sizeof(struct mark_set));
2383
2384         for (i = 1; i < argc; i++) {
2385                 const char *a = argv[i];
2386
2387                 if (*a != '-' || !strcmp(a, "--"))
2388                         break;
2389                 else if (!prefixcmp(a, "--date-format=")) {
2390                         const char *fmt = a + 14;
2391                         if (!strcmp(fmt, "raw"))
2392                                 whenspec = WHENSPEC_RAW;
2393                         else if (!strcmp(fmt, "rfc2822"))
2394                                 whenspec = WHENSPEC_RFC2822;
2395                         else if (!strcmp(fmt, "now"))
2396                                 whenspec = WHENSPEC_NOW;
2397                         else
2398                                 die("unknown --date-format argument %s", fmt);
2399                 }
2400                 else if (!prefixcmp(a, "--max-pack-size="))
2401                         max_packsize = strtoumax(a + 16, NULL, 0) * 1024 * 1024;
2402                 else if (!prefixcmp(a, "--depth=")) {
2403                         max_depth = strtoul(a + 8, NULL, 0);
2404                         if (max_depth > MAX_DEPTH)
2405                                 die("--depth cannot exceed %u", MAX_DEPTH);
2406                 }
2407                 else if (!prefixcmp(a, "--active-branches="))
2408                         max_active_branches = strtoul(a + 18, NULL, 0);
2409                 else if (!prefixcmp(a, "--import-marks="))
2410                         import_marks(a + 15);
2411                 else if (!prefixcmp(a, "--export-marks="))
2412                         mark_file = a + 15;
2413                 else if (!prefixcmp(a, "--export-pack-edges=")) {
2414                         if (pack_edges)
2415                                 fclose(pack_edges);
2416                         pack_edges = fopen(a + 20, "a");
2417                         if (!pack_edges)
2418                                 die("Cannot open %s: %s", a + 20, strerror(errno));
2419                 } else if (!strcmp(a, "--force"))
2420                         force_update = 1;
2421                 else if (!strcmp(a, "--quiet"))
2422                         show_stats = 0;
2423                 else if (!strcmp(a, "--stats"))
2424                         show_stats = 1;
2425                 else
2426                         die("unknown option %s", a);
2427         }
2428         if (i != argc)
2429                 usage(fast_import_usage);
2430
2431         rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
2432         for (i = 0; i < (cmd_save - 1); i++)
2433                 rc_free[i].next = &rc_free[i + 1];
2434         rc_free[cmd_save - 1].next = NULL;
2435
2436         prepare_packed_git();
2437         start_packfile();
2438         set_die_routine(die_nicely);
2439         while (read_next_command() != EOF) {
2440                 if (!strcmp("blob", command_buf.buf))
2441                         cmd_new_blob();
2442                 else if (!prefixcmp(command_buf.buf, "commit "))
2443                         cmd_new_commit();
2444                 else if (!prefixcmp(command_buf.buf, "tag "))
2445                         cmd_new_tag();
2446                 else if (!prefixcmp(command_buf.buf, "reset "))
2447                         cmd_reset_branch();
2448                 else if (!strcmp("checkpoint", command_buf.buf))
2449                         cmd_checkpoint();
2450                 else if (!prefixcmp(command_buf.buf, "progress "))
2451                         cmd_progress();
2452                 else
2453                         die("Unsupported command: %s", command_buf.buf);
2454         }
2455         end_packfile();
2456
2457         dump_branches();
2458         dump_tags();
2459         unkeep_all_packs();
2460         dump_marks();
2461
2462         if (pack_edges)
2463                 fclose(pack_edges);
2464
2465         if (show_stats) {
2466                 uintmax_t total_count = 0, duplicate_count = 0;
2467                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
2468                         total_count += object_count_by_type[i];
2469                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
2470                         duplicate_count += duplicate_count_by_type[i];
2471
2472                 fprintf(stderr, "%s statistics:\n", argv[0]);
2473                 fprintf(stderr, "---------------------------------------------------------------------\n");
2474                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
2475                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
2476                 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]);
2477                 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]);
2478                 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]);
2479                 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]);
2480                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
2481                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
2482                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
2483                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
2484                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
2485                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
2486                 fprintf(stderr, "---------------------------------------------------------------------\n");
2487                 pack_report();
2488                 fprintf(stderr, "---------------------------------------------------------------------\n");
2489                 fprintf(stderr, "\n");
2490         }
2491
2492         return failure ? 1 : 0;
2493 }