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