fast-import.c

   1 /*
   2 (See Documentation/git-fast-import.txt for maintained documentation.)
   3 Format of STDIN stream:
   4
   5   stream ::= cmd*;
   6
   7   cmd ::= new_blob
   8         | new_commit
   9         | new_tag
  10         | reset_branch
  11         | checkpoint
  12         | progress
  13         ;
  14
  15   new_blob ::= 'blob' lf
  16     mark?
  17     file_content;
  18   file_content ::= data;
  19
  20   new_commit ::= 'commit' sp ref_str lf
  21     mark?
  22     ('author' (sp name)? sp '<' email '>' sp when lf)?
  23     'committer' (sp name)? sp '<' email '>' sp when lf
  24     commit_msg
  25     ('from' sp commit-ish lf)?
  26     ('merge' sp commit-ish lf)*
  27     (file_change | ls)*
  28     lf?;
  29   commit_msg ::= data;
  30
  31   ls ::= 'ls' sp '"' quoted(path) '"' lf;
  32
  33   file_change ::= file_clr
  34     | file_del
  35     | file_rnm
  36     | file_cpy
  37     | file_obm
  38     | file_inm;
  39   file_clr ::= 'deleteall' lf;
  40   file_del ::= 'D' sp path_str lf;
  41   file_rnm ::= 'R' sp path_str sp path_str lf;
  42   file_cpy ::= 'C' sp path_str sp path_str lf;
  43   file_obm ::= 'M' sp mode sp (hexsha1 | idnum) sp path_str lf;
  44   file_inm ::= 'M' sp mode sp 'inline' sp path_str lf
  45     data;
  46   note_obm ::= 'N' sp (hexsha1 | idnum) sp commit-ish lf;
  47   note_inm ::= 'N' sp 'inline' sp commit-ish lf
  48     data;
  49
  50   new_tag ::= 'tag' sp tag_str lf
  51     'from' sp commit-ish lf
  52     ('tagger' (sp name)? sp '<' email '>' sp when lf)?
  53     tag_msg;
  54   tag_msg ::= data;
  55
  56   reset_branch ::= 'reset' sp ref_str lf
  57     ('from' sp commit-ish lf)?
  58     lf?;
  59
  60   checkpoint ::= 'checkpoint' lf
  61     lf?;
  62
  63   progress ::= 'progress' sp not_lf* lf
  64     lf?;
  65
  66      # note: the first idnum in a stream should be 1 and subsequent
  67      # idnums should not have gaps between values as this will cause
  68      # the stream parser to reserve space for the gapped values.  An
  69      # idnum can be updated in the future to a new object by issuing
  70      # a new mark directive with the old idnum.
  71      #
  72   mark ::= 'mark' sp idnum lf;
  73   data ::= (delimited_data | exact_data)
  74     lf?;
  75
  76     # note: delim may be any string but must not contain lf.
  77     # data_line may contain any data but must not be exactly
  78     # delim.
  79   delimited_data ::= 'data' sp '<<' delim lf
  80     (data_line lf)*
  81     delim lf;
  82
  83      # note: declen indicates the length of binary_data in bytes.
  84      # declen does not include the lf preceding the binary data.
  85      #
  86   exact_data ::= 'data' sp declen lf
  87     binary_data;
  88
  89      # note: quoted strings are C-style quoting supporting \c for
  90      # common escapes of 'c' (e..g \n, \t, \\, \") or \nnn where nnn
  91      # is the signed byte value in octal.  Note that the only
  92      # characters which must actually be escaped to protect the
  93      # stream formatting is: \, " and LF.  Otherwise these values
  94      # are UTF8.
  95      #
  96   commit-ish  ::= (ref_str | hexsha1 | sha1exp_str | idnum);
  97   ref_str     ::= ref;
  98   sha1exp_str ::= sha1exp;
  99   tag_str     ::= tag;
 100   path_str    ::= path    | '"' quoted(path)    '"' ;
 101   mode        ::= '100644' | '644'
 102                 | '100755' | '755'
 103                 | '120000'
 104                 ;
 105
 106   declen ::= # unsigned 32 bit value, ascii base10 notation;
 107   bigint ::= # unsigned integer value, ascii base10 notation;
 108   binary_data ::= # file content, not interpreted;
 109
 110   when         ::= raw_when | rfc2822_when;
 111   raw_when     ::= ts sp tz;
 112   rfc2822_when ::= # Valid RFC 2822 date and time;
 113
 114   sp ::= # ASCII space character;
 115   lf ::= # ASCII newline (LF) character;
 116
 117      # note: a colon (':') must precede the numerical value assigned to
 118      # an idnum.  This is to distinguish it from a ref or tag name as
 119      # GIT does not permit ':' in ref or tag strings.
 120      #
 121   idnum   ::= ':' bigint;
 122   path    ::= # GIT style file path, e.g. "a/b/c";
 123   ref     ::= # GIT ref name, e.g. "refs/heads/MOZ_GECKO_EXPERIMENT";
 124   tag     ::= # GIT tag name, e.g. "FIREFOX_1_5";
 125   sha1exp ::= # Any valid GIT SHA1 expression;
 126   hexsha1 ::= # SHA1 in hexadecimal format;
 127
 128      # note: name and email are UTF8 strings, however name must not
 129      # contain '<' or lf and email must not contain any of the
 130      # following: '<', '>', lf.
 131      #
 132   name  ::= # valid GIT author/committer name;
 133   email ::= # valid GIT author/committer email;
 134   ts    ::= # time since the epoch in seconds, ascii base10 notation;
 135   tz    ::= # GIT style timezone;
 136
 137      # note: comments, ls and cat requests may appear anywhere
 138      # in the input, except within a data command.  Any form
 139      # of the data command always escapes the related input
 140      # from comment processing.
 141      #
 142      # In case it is not clear, the '#' that starts the comment
 143      # must be the first character on that line (an lf
 144      # preceded it).
 145      #
 146
 147   cat_blob ::= 'cat-blob' sp (hexsha1 | idnum) lf;
 148   ls_tree  ::= 'ls' sp (hexsha1 | idnum) sp path_str lf;
 149
 150   comment ::= '#' not_lf* lf;
 151   not_lf  ::= # Any byte that is not ASCII newline (LF);
 152 */
 153
 154 #include "builtin.h"
 155 #include "cache.h"
 156 #include "object.h"
 157 #include "blob.h"
 158 #include "tree.h"
 159 #include "commit.h"
 160 #include "delta.h"
 161 #include "pack.h"
 162 #include "refs.h"
 163 #include "csum-file.h"
 164 #include "quote.h"
 165 #include "exec_cmd.h"
 166 #include "dir.h"
 167
 168 #define PACK_ID_BITS 16
 169 #define MAX_PACK_ID ((1<<PACK_ID_BITS)-1)
 170 #define DEPTH_BITS 13
 171 #define MAX_DEPTH ((1<<DEPTH_BITS)-1)
 172
 173 /*
 174  * We abuse the setuid bit on directories to mean "do not delta".
 175  */
 176 #define NO_DELTA S_ISUID
 177
 178 struct object_entry {
 179         struct pack_idx_entry idx;
 180         struct object_entry *next;
 181         uint32_t type : TYPE_BITS,
 182                 pack_id : PACK_ID_BITS,
 183                 depth : DEPTH_BITS;
 184 };
 185
 186 struct object_entry_pool {
 187         struct object_entry_pool *next_pool;
 188         struct object_entry *next_free;
 189         struct object_entry *end;
 190         struct object_entry entries[FLEX_ARRAY]; /* more */
 191 };
 192
 193 struct mark_set {
 194         union {
 195                 struct object_entry *marked[1024];
 196                 struct mark_set *sets[1024];
 197         } data;
 198         unsigned int shift;
 199 };
 200
 201 struct last_object {
 202         struct strbuf data;
 203         off_t offset;
 204         unsigned int depth;
 205         unsigned no_swap : 1;
 206 };
 207
 208 struct mem_pool {
 209         struct mem_pool *next_pool;
 210         char *next_free;
 211         char *end;
 212         uintmax_t space[FLEX_ARRAY]; /* more */
 213 };
 214
 215 struct atom_str {
 216         struct atom_str *next_atom;
 217         unsigned short str_len;
 218         char str_dat[FLEX_ARRAY]; /* more */
 219 };
 220
 221 struct tree_content;
 222 struct tree_entry {
 223         struct tree_content *tree;
 224         struct atom_str *name;
 225         struct tree_entry_ms {
 226                 uint16_t mode;
 227                 unsigned char sha1[20];
 228         } versions[2];
 229 };
 230
 231 struct tree_content {
 232         unsigned int entry_capacity; /* must match avail_tree_content */
 233         unsigned int entry_count;
 234         unsigned int delta_depth;
 235         struct tree_entry *entries[FLEX_ARRAY]; /* more */
 236 };
 237
 238 struct avail_tree_content {
 239         unsigned int entry_capacity; /* must match tree_content */
 240         struct avail_tree_content *next_avail;
 241 };
 242
 243 struct branch {
 244         struct branch *table_next_branch;
 245         struct branch *active_next_branch;
 246         const char *name;
 247         struct tree_entry branch_tree;
 248         uintmax_t last_commit;
 249         uintmax_t num_notes;
 250         unsigned active : 1;
 251         unsigned pack_id : PACK_ID_BITS;
 252         unsigned char sha1[20];
 253 };
 254
 255 struct tag {
 256         struct tag *next_tag;
 257         const char *name;
 258         unsigned int pack_id;
 259         unsigned char sha1[20];
 260 };
 261
 262 struct hash_list {
 263         struct hash_list *next;
 264         unsigned char sha1[20];
 265 };
 266
 267 typedef enum {
 268         WHENSPEC_RAW = 1,
 269         WHENSPEC_RFC2822,
 270         WHENSPEC_NOW
 271 } whenspec_type;
 272
 273 struct recent_command {
 274         struct recent_command *prev;
 275         struct recent_command *next;
 276         char *buf;
 277 };
 278
 279 /* Configured limits on output */
 280 static unsigned long max_depth = 10;
 281 static off_t max_packsize;
 282 static int force_update;
 283 static int pack_compression_level = Z_DEFAULT_COMPRESSION;
 284 static int pack_compression_seen;
 285
 286 /* Stats and misc. counters */
 287 static uintmax_t alloc_count;
 288 static uintmax_t marks_set_count;
 289 static uintmax_t object_count_by_type[1 << TYPE_BITS];
 290 static uintmax_t duplicate_count_by_type[1 << TYPE_BITS];
 291 static uintmax_t delta_count_by_type[1 << TYPE_BITS];
 292 static uintmax_t delta_count_attempts_by_type[1 << TYPE_BITS];
 293 static unsigned long object_count;
 294 static unsigned long branch_count;
 295 static unsigned long branch_load_count;
 296 static int failure;
 297 static FILE *pack_edges;
 298 static unsigned int show_stats = 1;
 299 static int global_argc;
 300 static char **global_argv;
 301
 302 /* Memory pools */
 303 static size_t mem_pool_alloc = 2*1024*1024 - sizeof(struct mem_pool);
 304 static size_t total_allocd;
 305 static struct mem_pool *mem_pool;
 306
 307 /* Atom management */
 308 static unsigned int atom_table_sz = 4451;
 309 static unsigned int atom_cnt;
 310 static struct atom_str **atom_table;
 311
 312 /* The .pack file being generated */
 313 static struct pack_idx_option pack_idx_opts;
 314 static unsigned int pack_id;
 315 static struct sha1file *pack_file;
 316 static struct packed_git *pack_data;
 317 static struct packed_git **all_packs;
 318 static off_t pack_size;
 319
 320 /* Table of objects we've written. */
 321 static unsigned int object_entry_alloc = 5000;
 322 static struct object_entry_pool *blocks;
 323 static struct object_entry *object_table[1 << 16];
 324 static struct mark_set *marks;
 325 static const char *export_marks_file;
 326 static const char *import_marks_file;
 327 static int import_marks_file_from_stream;
 328 static int import_marks_file_ignore_missing;
 329 static int relative_marks_paths;
 330
 331 /* Our last blob */
 332 static struct last_object last_blob = { STRBUF_INIT, 0, 0, 0 };
 333
 334 /* Tree management */
 335 static unsigned int tree_entry_alloc = 1000;
 336 static void *avail_tree_entry;
 337 static unsigned int avail_tree_table_sz = 100;
 338 static struct avail_tree_content **avail_tree_table;
 339 static struct strbuf old_tree = STRBUF_INIT;
 340 static struct strbuf new_tree = STRBUF_INIT;
 341
 342 /* Branch data */
 343 static unsigned long max_active_branches = 5;
 344 static unsigned long cur_active_branches;
 345 static unsigned long branch_table_sz = 1039;
 346 static struct branch **branch_table;
 347 static struct branch *active_branches;
 348
 349 /* Tag data */
 350 static struct tag *first_tag;
 351 static struct tag *last_tag;
 352
 353 /* Input stream parsing */
 354 static whenspec_type whenspec = WHENSPEC_RAW;
 355 static struct strbuf command_buf = STRBUF_INIT;
 356 static int unread_command_buf;
 357 static struct recent_command cmd_hist = {&cmd_hist, &cmd_hist, NULL};
 358 static struct recent_command *cmd_tail = &cmd_hist;
 359 static struct recent_command *rc_free;
 360 static unsigned int cmd_save = 100;
 361 static uintmax_t next_mark;
 362 static struct strbuf new_data = STRBUF_INIT;
 363 static int seen_data_command;
 364 static int require_explicit_termination;
 365
 366 /* Signal handling */
 367 static volatile sig_atomic_t checkpoint_requested;
 368
 369 /* Where to write output of cat-blob commands */
 370 static int cat_blob_fd = STDOUT_FILENO;
 371
 372 static void parse_argv(void);
 373 static void parse_cat_blob(void);
 374 static void parse_ls(struct branch *b);
 375
 376 static void write_branch_report(FILE *rpt, struct branch *b)
 377 {
 378         fprintf(rpt, "%s:\n", b->name);
 379
 380         fprintf(rpt, "  status      :");
 381         if (b->active)
 382                 fputs(" active", rpt);
 383         if (b->branch_tree.tree)
 384                 fputs(" loaded", rpt);
 385         if (is_null_sha1(b->branch_tree.versions[1].sha1))
 386                 fputs(" dirty", rpt);
 387         fputc('\n', rpt);
 388
 389         fprintf(rpt, "  tip commit  : %s\n", sha1_to_hex(b->sha1));
 390         fprintf(rpt, "  old tree    : %s\n", sha1_to_hex(b->branch_tree.versions[0].sha1));
 391         fprintf(rpt, "  cur tree    : %s\n", sha1_to_hex(b->branch_tree.versions[1].sha1));
 392         fprintf(rpt, "  commit clock: %" PRIuMAX "\n", b->last_commit);
 393
 394         fputs("  last pack   : ", rpt);
 395         if (b->pack_id < MAX_PACK_ID)
 396                 fprintf(rpt, "%u", b->pack_id);
 397         fputc('\n', rpt);
 398
 399         fputc('\n', rpt);
 400 }
 401
 402 static void dump_marks_helper(FILE *, uintmax_t, struct mark_set *);
 403
 404 static void write_crash_report(const char *err)
 405 {
 406         char *loc = git_path("fast_import_crash_%"PRIuMAX, (uintmax_t) getpid());
 407         FILE *rpt = fopen(loc, "w");
 408         struct branch *b;
 409         unsigned long lu;
 410         struct recent_command *rc;
 411
 412         if (!rpt) {
 413                 error("can't write crash report %s: %s", loc, strerror(errno));
 414                 return;
 415         }
 416
 417         fprintf(stderr, "fast-import: dumping crash report to %s\n", loc);
 418
 419         fprintf(rpt, "fast-import crash report:\n");
 420         fprintf(rpt, "    fast-import process: %"PRIuMAX"\n", (uintmax_t) getpid());
 421         fprintf(rpt, "    parent process     : %"PRIuMAX"\n", (uintmax_t) getppid());
 422         fprintf(rpt, "    at %s\n", show_date(time(NULL), 0, DATE_LOCAL));
 423         fputc('\n', rpt);
 424
 425         fputs("fatal: ", rpt);
 426         fputs(err, rpt);
 427         fputc('\n', rpt);
 428
 429         fputc('\n', rpt);
 430         fputs("Most Recent Commands Before Crash\n", rpt);
 431         fputs("---------------------------------\n", rpt);
 432         for (rc = cmd_hist.next; rc != &cmd_hist; rc = rc->next) {
 433                 if (rc->next == &cmd_hist)
 434                         fputs("* ", rpt);
 435                 else
 436                         fputs("  ", rpt);
 437                 fputs(rc->buf, rpt);
 438                 fputc('\n', rpt);
 439         }
 440
 441         fputc('\n', rpt);
 442         fputs("Active Branch LRU\n", rpt);
 443         fputs("-----------------\n", rpt);
 444         fprintf(rpt, "    active_branches = %lu cur, %lu max\n",
 445                 cur_active_branches,
 446                 max_active_branches);
 447         fputc('\n', rpt);
 448         fputs("  pos  clock name\n", rpt);
 449         fputs("  ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~\n", rpt);
 450         for (b = active_branches, lu = 0; b; b = b->active_next_branch)
 451                 fprintf(rpt, "  %2lu) %6" PRIuMAX" %s\n",
 452                         ++lu, b->last_commit, b->name);
 453
 454         fputc('\n', rpt);
 455         fputs("Inactive Branches\n", rpt);
 456         fputs("-----------------\n", rpt);
 457         for (lu = 0; lu < branch_table_sz; lu++) {
 458                 for (b = branch_table[lu]; b; b = b->table_next_branch)
 459                         write_branch_report(rpt, b);
 460         }
 461
 462         if (first_tag) {
 463                 struct tag *tg;
 464                 fputc('\n', rpt);
 465                 fputs("Annotated Tags\n", rpt);
 466                 fputs("--------------\n", rpt);
 467                 for (tg = first_tag; tg; tg = tg->next_tag) {
 468                         fputs(sha1_to_hex(tg->sha1), rpt);
 469                         fputc(' ', rpt);
 470                         fputs(tg->name, rpt);
 471                         fputc('\n', rpt);
 472                 }
 473         }
 474
 475         fputc('\n', rpt);
 476         fputs("Marks\n", rpt);
 477         fputs("-----\n", rpt);
 478         if (export_marks_file)
 479                 fprintf(rpt, "  exported to %s\n", export_marks_file);
 480         else
 481                 dump_marks_helper(rpt, 0, marks);
 482
 483         fputc('\n', rpt);
 484         fputs("-------------------\n", rpt);
 485         fputs("END OF CRASH REPORT\n", rpt);
 486         fclose(rpt);
 487 }
 488
 489 static void end_packfile(void);
 490 static void unkeep_all_packs(void);
 491 static void dump_marks(void);
 492
 493 static NORETURN void die_nicely(const char *err, va_list params)
 494 {
 495         static int zombie;
 496         char message[2 * PATH_MAX];
 497
 498         vsnprintf(message, sizeof(message), err, params);
 499         fputs("fatal: ", stderr);
 500         fputs(message, stderr);
 501         fputc('\n', stderr);
 502
 503         if (!zombie) {
 504                 zombie = 1;
 505                 write_crash_report(message);
 506                 end_packfile();
 507                 unkeep_all_packs();
 508                 dump_marks();
 509         }
 510         exit(128);
 511 }
 512
 513 #ifndef SIGUSR1 /* Windows, for example */
 514
 515 static void set_checkpoint_signal(void)
 516 {
 517 }
 518
 519 #else
 520
 521 static void checkpoint_signal(int signo)
 522 {
 523         checkpoint_requested = 1;
 524 }
 525
 526 static void set_checkpoint_signal(void)
 527 {
 528         struct sigaction sa;
 529
 530         memset(&sa, 0, sizeof(sa));
 531         sa.sa_handler = checkpoint_signal;
 532         sigemptyset(&sa.sa_mask);
 533         sa.sa_flags = SA_RESTART;
 534         sigaction(SIGUSR1, &sa, NULL);
 535 }
 536
 537 #endif
 538
 539 static void alloc_objects(unsigned int cnt)
 540 {
 541         struct object_entry_pool *b;
 542
 543         b = xmalloc(sizeof(struct object_entry_pool)
 544                 + cnt * sizeof(struct object_entry));
 545         b->next_pool = blocks;
 546         b->next_free = b->entries;
 547         b->end = b->entries + cnt;
 548         blocks = b;
 549         alloc_count += cnt;
 550 }
 551
 552 static struct object_entry *new_object(unsigned char *sha1)
 553 {
 554         struct object_entry *e;
 555
 556         if (blocks->next_free == blocks->end)
 557                 alloc_objects(object_entry_alloc);
 558
 559         e = blocks->next_free++;
 560         hashcpy(e->idx.sha1, sha1);
 561         return e;
 562 }
 563
 564 static struct object_entry *find_object(unsigned char *sha1)
 565 {
 566         unsigned int h = sha1[0] << 8 | sha1[1];
 567         struct object_entry *e;
 568         for (e = object_table[h]; e; e = e->next)
 569                 if (!hashcmp(sha1, e->idx.sha1))
 570                         return e;
 571         return NULL;
 572 }
 573
 574 static struct object_entry *insert_object(unsigned char *sha1)
 575 {
 576         unsigned int h = sha1[0] << 8 | sha1[1];
 577         struct object_entry *e = object_table[h];
 578
 579         while (e) {
 580                 if (!hashcmp(sha1, e->idx.sha1))
 581                         return e;
 582                 e = e->next;
 583         }
 584
 585         e = new_object(sha1);
 586         e->next = object_table[h];
 587         e->idx.offset = 0;
 588         object_table[h] = e;
 589         return e;
 590 }
 591
 592 static unsigned int hc_str(const char *s, size_t len)
 593 {
 594         unsigned int r = 0;
 595         while (len-- > 0)
 596                 r = r * 31 + *s++;
 597         return r;
 598 }
 599
 600 static void *pool_alloc(size_t len)
 601 {
 602         struct mem_pool *p;
 603         void *r;
 604
 605         /* round up to a 'uintmax_t' alignment */
 606         if (len & (sizeof(uintmax_t) - 1))
 607                 len += sizeof(uintmax_t) - (len & (sizeof(uintmax_t) - 1));
 608
 609         for (p = mem_pool; p; p = p->next_pool)
 610                 if ((p->end - p->next_free >= len))
 611                         break;
 612
 613         if (!p) {
 614                 if (len >= (mem_pool_alloc/2)) {
 615                         total_allocd += len;
 616                         return xmalloc(len);
 617                 }
 618                 total_allocd += sizeof(struct mem_pool) + mem_pool_alloc;
 619                 p = xmalloc(sizeof(struct mem_pool) + mem_pool_alloc);
 620                 p->next_pool = mem_pool;
 621                 p->next_free = (char *) p->space;
 622                 p->end = p->next_free + mem_pool_alloc;
 623                 mem_pool = p;
 624         }
 625
 626         r = p->next_free;
 627         p->next_free += len;
 628         return r;
 629 }
 630
 631 static void *pool_calloc(size_t count, size_t size)
 632 {
 633         size_t len = count * size;
 634         void *r = pool_alloc(len);
 635         memset(r, 0, len);
 636         return r;
 637 }
 638
 639 static char *pool_strdup(const char *s)
 640 {
 641         char *r = pool_alloc(strlen(s) + 1);
 642         strcpy(r, s);
 643         return r;
 644 }
 645
 646 static void insert_mark(uintmax_t idnum, struct object_entry *oe)
 647 {
 648         struct mark_set *s = marks;
 649         while ((idnum >> s->shift) >= 1024) {
 650                 s = pool_calloc(1, sizeof(struct mark_set));
 651                 s->shift = marks->shift + 10;
 652                 s->data.sets[0] = marks;
 653                 marks = s;
 654         }
 655         while (s->shift) {
 656                 uintmax_t i = idnum >> s->shift;
 657                 idnum -= i << s->shift;
 658                 if (!s->data.sets[i]) {
 659                         s->data.sets[i] = pool_calloc(1, sizeof(struct mark_set));
 660                         s->data.sets[i]->shift = s->shift - 10;
 661                 }
 662                 s = s->data.sets[i];
 663         }
 664         if (!s->data.marked[idnum])
 665                 marks_set_count++;
 666         s->data.marked[idnum] = oe;
 667 }
 668
 669 static struct object_entry *find_mark(uintmax_t idnum)
 670 {
 671         uintmax_t orig_idnum = idnum;
 672         struct mark_set *s = marks;
 673         struct object_entry *oe = NULL;
 674         if ((idnum >> s->shift) < 1024) {
 675                 while (s && s->shift) {
 676                         uintmax_t i = idnum >> s->shift;
 677                         idnum -= i << s->shift;
 678                         s = s->data.sets[i];
 679                 }
 680                 if (s)
 681                         oe = s->data.marked[idnum];
 682         }
 683         if (!oe)
 684                 die("mark :%" PRIuMAX " not declared", orig_idnum);
 685         return oe;
 686 }
 687
 688 static struct atom_str *to_atom(const char *s, unsigned short len)
 689 {
 690         unsigned int hc = hc_str(s, len) % atom_table_sz;
 691         struct atom_str *c;
 692
 693         for (c = atom_table[hc]; c; c = c->next_atom)
 694                 if (c->str_len == len && !strncmp(s, c->str_dat, len))
 695                         return c;
 696
 697         c = pool_alloc(sizeof(struct atom_str) + len + 1);
 698         c->str_len = len;
 699         strncpy(c->str_dat, s, len);
 700         c->str_dat[len] = 0;
 701         c->next_atom = atom_table[hc];
 702         atom_table[hc] = c;
 703         atom_cnt++;
 704         return c;
 705 }
 706
 707 static struct branch *lookup_branch(const char *name)
 708 {
 709         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 710         struct branch *b;
 711
 712         for (b = branch_table[hc]; b; b = b->table_next_branch)
 713                 if (!strcmp(name, b->name))
 714                         return b;
 715         return NULL;
 716 }
 717
 718 static struct branch *new_branch(const char *name)
 719 {
 720         unsigned int hc = hc_str(name, strlen(name)) % branch_table_sz;
 721         struct branch *b = lookup_branch(name);
 722
 723         if (b)
 724                 die("Invalid attempt to create duplicate branch: %s", name);
 725         if (check_refname_format(name, REFNAME_ALLOW_ONELEVEL))
 726                 die("Branch name doesn't conform to GIT standards: %s", name);
 727
 728         b = pool_calloc(1, sizeof(struct branch));
 729         b->name = pool_strdup(name);
 730         b->table_next_branch = branch_table[hc];
 731         b->branch_tree.versions[0].mode = S_IFDIR;
 732         b->branch_tree.versions[1].mode = S_IFDIR;
 733         b->num_notes = 0;
 734         b->active = 0;
 735         b->pack_id = MAX_PACK_ID;
 736         branch_table[hc] = b;
 737         branch_count++;
 738         return b;
 739 }
 740
 741 static unsigned int hc_entries(unsigned int cnt)
 742 {
 743         cnt = cnt & 7 ? (cnt / 8) + 1 : cnt / 8;
 744         return cnt < avail_tree_table_sz ? cnt : avail_tree_table_sz - 1;
 745 }
 746
 747 static struct tree_content *new_tree_content(unsigned int cnt)
 748 {
 749         struct avail_tree_content *f, *l = NULL;
 750         struct tree_content *t;
 751         unsigned int hc = hc_entries(cnt);
 752
 753         for (f = avail_tree_table[hc]; f; l = f, f = f->next_avail)
 754                 if (f->entry_capacity >= cnt)
 755                         break;
 756
 757         if (f) {
 758                 if (l)
 759                         l->next_avail = f->next_avail;
 760                 else
 761                         avail_tree_table[hc] = f->next_avail;
 762         } else {
 763                 cnt = cnt & 7 ? ((cnt / 8) + 1) * 8 : cnt;
 764                 f = pool_alloc(sizeof(*t) + sizeof(t->entries[0]) * cnt);
 765                 f->entry_capacity = cnt;
 766         }
 767
 768         t = (struct tree_content*)f;
 769         t->entry_count = 0;
 770         t->delta_depth = 0;
 771         return t;
 772 }
 773
 774 static void release_tree_entry(struct tree_entry *e);
 775 static void release_tree_content(struct tree_content *t)
 776 {
 777         struct avail_tree_content *f = (struct avail_tree_content*)t;
 778         unsigned int hc = hc_entries(f->entry_capacity);
 779         f->next_avail = avail_tree_table[hc];
 780         avail_tree_table[hc] = f;
 781 }
 782
 783 static void release_tree_content_recursive(struct tree_content *t)
 784 {
 785         unsigned int i;
 786         for (i = 0; i < t->entry_count; i++)
 787                 release_tree_entry(t->entries[i]);
 788         release_tree_content(t);
 789 }
 790
 791 static struct tree_content *grow_tree_content(
 792         struct tree_content *t,
 793         int amt)
 794 {
 795         struct tree_content *r = new_tree_content(t->entry_count + amt);
 796         r->entry_count = t->entry_count;
 797         r->delta_depth = t->delta_depth;
 798         memcpy(r->entries,t->entries,t->entry_count*sizeof(t->entries[0]));
 799         release_tree_content(t);
 800         return r;
 801 }
 802
 803 static struct tree_entry *new_tree_entry(void)
 804 {
 805         struct tree_entry *e;
 806
 807         if (!avail_tree_entry) {
 808                 unsigned int n = tree_entry_alloc;
 809                 total_allocd += n * sizeof(struct tree_entry);
 810                 avail_tree_entry = e = xmalloc(n * sizeof(struct tree_entry));
 811                 while (n-- > 1) {
 812                         *((void**)e) = e + 1;
 813                         e++;
 814                 }
 815                 *((void**)e) = NULL;
 816         }
 817
 818         e = avail_tree_entry;
 819         avail_tree_entry = *((void**)e);
 820         return e;
 821 }
 822
 823 static void release_tree_entry(struct tree_entry *e)
 824 {
 825         if (e->tree)
 826                 release_tree_content_recursive(e->tree);
 827         *((void**)e) = avail_tree_entry;
 828         avail_tree_entry = e;
 829 }
 830
 831 static struct tree_content *dup_tree_content(struct tree_content *s)
 832 {
 833         struct tree_content *d;
 834         struct tree_entry *a, *b;
 835         unsigned int i;
 836
 837         if (!s)
 838                 return NULL;
 839         d = new_tree_content(s->entry_count);
 840         for (i = 0; i < s->entry_count; i++) {
 841                 a = s->entries[i];
 842                 b = new_tree_entry();
 843                 memcpy(b, a, sizeof(*a));
 844                 if (a->tree && is_null_sha1(b->versions[1].sha1))
 845                         b->tree = dup_tree_content(a->tree);
 846                 else
 847                         b->tree = NULL;
 848                 d->entries[i] = b;
 849         }
 850         d->entry_count = s->entry_count;
 851         d->delta_depth = s->delta_depth;
 852
 853         return d;
 854 }
 855
 856 static void start_packfile(void)
 857 {
 858         static char tmp_file[PATH_MAX];
 859         struct packed_git *p;
 860         struct pack_header hdr;
 861         int pack_fd;
 862
 863         pack_fd = odb_mkstemp(tmp_file, sizeof(tmp_file),
 864                               "pack/tmp_pack_XXXXXX");
 865         p = xcalloc(1, sizeof(*p) + strlen(tmp_file) + 2);
 866         strcpy(p->pack_name, tmp_file);
 867         p->pack_fd = pack_fd;
 868         p->do_not_close = 1;
 869         pack_file = sha1fd(pack_fd, p->pack_name);
 870
 871         hdr.hdr_signature = htonl(PACK_SIGNATURE);
 872         hdr.hdr_version = htonl(2);
 873         hdr.hdr_entries = 0;
 874         sha1write(pack_file, &hdr, sizeof(hdr));
 875
 876         pack_data = p;
 877         pack_size = sizeof(hdr);
 878         object_count = 0;
 879
 880         all_packs = xrealloc(all_packs, sizeof(*all_packs) * (pack_id + 1));
 881         all_packs[pack_id] = p;
 882 }
 883
 884 static const char *create_index(void)
 885 {
 886         const char *tmpfile;
 887         struct pack_idx_entry **idx, **c, **last;
 888         struct object_entry *e;
 889         struct object_entry_pool *o;
 890
 891         /* Build the table of object IDs. */
 892         idx = xmalloc(object_count * sizeof(*idx));
 893         c = idx;
 894         for (o = blocks; o; o = o->next_pool)
 895                 for (e = o->next_free; e-- != o->entries;)
 896                         if (pack_id == e->pack_id)
 897                                 *c++ = &e->idx;
 898         last = idx + object_count;
 899         if (c != last)
 900                 die("internal consistency error creating the index");
 901
 902         tmpfile = write_idx_file(NULL, idx, object_count, &pack_idx_opts, pack_data->sha1);
 903         free(idx);
 904         return tmpfile;
 905 }
 906
 907 static char *keep_pack(const char *curr_index_name)
 908 {
 909         static char name[PATH_MAX];
 910         static const char *keep_msg = "fast-import";
 911         int keep_fd;
 912
 913         keep_fd = odb_pack_keep(name, sizeof(name), pack_data->sha1);
 914         if (keep_fd < 0)
 915                 die_errno("cannot create keep file");
 916         write_or_die(keep_fd, keep_msg, strlen(keep_msg));
 917         if (close(keep_fd))
 918                 die_errno("failed to write keep file");
 919
 920         snprintf(name, sizeof(name), "%s/pack/pack-%s.pack",
 921                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 922         if (move_temp_to_file(pack_data->pack_name, name))
 923                 die("cannot store pack file");
 924
 925         snprintf(name, sizeof(name), "%s/pack/pack-%s.idx",
 926                  get_object_directory(), sha1_to_hex(pack_data->sha1));
 927         if (move_temp_to_file(curr_index_name, name))
 928                 die("cannot store index file");
 929         free((void *)curr_index_name);
 930         return name;
 931 }
 932
 933 static void unkeep_all_packs(void)
 934 {
 935         static char name[PATH_MAX];
 936         int k;
 937
 938         for (k = 0; k < pack_id; k++) {
 939                 struct packed_git *p = all_packs[k];
 940                 snprintf(name, sizeof(name), "%s/pack/pack-%s.keep",
 941                          get_object_directory(), sha1_to_hex(p->sha1));
 942                 unlink_or_warn(name);
 943         }
 944 }
 945
 946 static void end_packfile(void)
 947 {
 948         struct packed_git *old_p = pack_data, *new_p;
 949
 950         clear_delta_base_cache();
 951         if (object_count) {
 952                 unsigned char cur_pack_sha1[20];
 953                 char *idx_name;
 954                 int i;
 955                 struct branch *b;
 956                 struct tag *t;
 957
 958                 close_pack_windows(pack_data);
 959                 sha1close(pack_file, cur_pack_sha1, 0);
 960                 fixup_pack_header_footer(pack_data->pack_fd, pack_data->sha1,
 961                                     pack_data->pack_name, object_count,
 962                                     cur_pack_sha1, pack_size);
 963                 close(pack_data->pack_fd);
 964                 idx_name = keep_pack(create_index());
 965
 966                 /* Register the packfile with core git's machinery. */
 967                 new_p = add_packed_git(idx_name, strlen(idx_name), 1);
 968                 if (!new_p)
 969                         die("core git rejected index %s", idx_name);
 970                 all_packs[pack_id] = new_p;
 971                 install_packed_git(new_p);
 972
 973                 /* Print the boundary */
 974                 if (pack_edges) {
 975                         fprintf(pack_edges, "%s:", new_p->pack_name);
 976                         for (i = 0; i < branch_table_sz; i++) {
 977                                 for (b = branch_table[i]; b; b = b->table_next_branch) {
 978                                         if (b->pack_id == pack_id)
 979                                                 fprintf(pack_edges, " %s", sha1_to_hex(b->sha1));
 980                                 }
 981                         }
 982                         for (t = first_tag; t; t = t->next_tag) {
 983                                 if (t->pack_id == pack_id)
 984                                         fprintf(pack_edges, " %s", sha1_to_hex(t->sha1));
 985                         }
 986                         fputc('\n', pack_edges);
 987                         fflush(pack_edges);
 988                 }
 989
 990                 pack_id++;
 991         }
 992         else {
 993                 close(old_p->pack_fd);
 994                 unlink_or_warn(old_p->pack_name);
 995         }
 996         free(old_p);
 997
 998         /* We can't carry a delta across packfiles. */
 999         strbuf_release(&last_blob.data);
1000         last_blob.offset = 0;
1001         last_blob.depth = 0;
1002 }
1003
1004 static void cycle_packfile(void)
1005 {
1006         end_packfile();
1007         start_packfile();
1008 }
1009
1010 static int store_object(
1011         enum object_type type,
1012         struct strbuf *dat,
1013         struct last_object *last,
1014         unsigned char *sha1out,
1015         uintmax_t mark)
1016 {
1017         void *out, *delta;
1018         struct object_entry *e;
1019         unsigned char hdr[96];
1020         unsigned char sha1[20];
1021         unsigned long hdrlen, deltalen;
1022         git_SHA_CTX c;
1023         git_zstream s;
1024
1025         hdrlen = sprintf((char *)hdr,"%s %lu", typename(type),
1026                 (unsigned long)dat->len) + 1;
1027         git_SHA1_Init(&c);
1028         git_SHA1_Update(&c, hdr, hdrlen);
1029         git_SHA1_Update(&c, dat->buf, dat->len);
1030         git_SHA1_Final(sha1, &c);
1031         if (sha1out)
1032                 hashcpy(sha1out, sha1);
1033
1034         e = insert_object(sha1);
1035         if (mark)
1036                 insert_mark(mark, e);
1037         if (e->idx.offset) {
1038                 duplicate_count_by_type[type]++;
1039                 return 1;
1040         } else if (find_sha1_pack(sha1, packed_git)) {
1041                 e->type = type;
1042                 e->pack_id = MAX_PACK_ID;
1043                 e->idx.offset = 1; /* just not zero! */
1044                 duplicate_count_by_type[type]++;
1045                 return 1;
1046         }
1047
1048         if (last && last->data.buf && last->depth < max_depth && dat->len > 20) {
1049                 delta_count_attempts_by_type[type]++;
1050                 delta = diff_delta(last->data.buf, last->data.len,
1051                         dat->buf, dat->len,
1052                         &deltalen, dat->len - 20);
1053         } else
1054                 delta = NULL;
1055
1056         memset(&s, 0, sizeof(s));
1057         git_deflate_init(&s, pack_compression_level);
1058         if (delta) {
1059                 s.next_in = delta;
1060                 s.avail_in = deltalen;
1061         } else {
1062                 s.next_in = (void *)dat->buf;
1063                 s.avail_in = dat->len;
1064         }
1065         s.avail_out = git_deflate_bound(&s, s.avail_in);
1066         s.next_out = out = xmalloc(s.avail_out);
1067         while (git_deflate(&s, Z_FINISH) == Z_OK)
1068                 ; /* nothing */
1069         git_deflate_end(&s);
1070
1071         /* Determine if we should auto-checkpoint. */
1072         if ((max_packsize && (pack_size + 60 + s.total_out) > max_packsize)
1073                 || (pack_size + 60 + s.total_out) < pack_size) {
1074
1075                 /* This new object needs to *not* have the current pack_id. */
1076                 e->pack_id = pack_id + 1;
1077                 cycle_packfile();
1078
1079                 /* We cannot carry a delta into the new pack. */
1080                 if (delta) {
1081                         free(delta);
1082                         delta = NULL;
1083
1084                         memset(&s, 0, sizeof(s));
1085                         git_deflate_init(&s, pack_compression_level);
1086                         s.next_in = (void *)dat->buf;
1087                         s.avail_in = dat->len;
1088                         s.avail_out = git_deflate_bound(&s, s.avail_in);
1089                         s.next_out = out = xrealloc(out, s.avail_out);
1090                         while (git_deflate(&s, Z_FINISH) == Z_OK)
1091                                 ; /* nothing */
1092                         git_deflate_end(&s);
1093                 }
1094         }
1095
1096         e->type = type;
1097         e->pack_id = pack_id;
1098         e->idx.offset = pack_size;
1099         object_count++;
1100         object_count_by_type[type]++;
1101
1102         crc32_begin(pack_file);
1103
1104         if (delta) {
1105                 off_t ofs = e->idx.offset - last->offset;
1106                 unsigned pos = sizeof(hdr) - 1;
1107
1108                 delta_count_by_type[type]++;
1109                 e->depth = last->depth + 1;
1110
1111                 hdrlen = encode_in_pack_object_header(OBJ_OFS_DELTA, deltalen, hdr);
1112                 sha1write(pack_file, hdr, hdrlen);
1113                 pack_size += hdrlen;
1114
1115                 hdr[pos] = ofs & 127;
1116                 while (ofs >>= 7)
1117                         hdr[--pos] = 128 | (--ofs & 127);
1118                 sha1write(pack_file, hdr + pos, sizeof(hdr) - pos);
1119                 pack_size += sizeof(hdr) - pos;
1120         } else {
1121                 e->depth = 0;
1122                 hdrlen = encode_in_pack_object_header(type, dat->len, hdr);
1123                 sha1write(pack_file, hdr, hdrlen);
1124                 pack_size += hdrlen;
1125         }
1126
1127         sha1write(pack_file, out, s.total_out);
1128         pack_size += s.total_out;
1129
1130         e->idx.crc32 = crc32_end(pack_file);
1131
1132         free(out);
1133         free(delta);
1134         if (last) {
1135                 if (last->no_swap) {
1136                         last->data = *dat;
1137                 } else {
1138                         strbuf_swap(&last->data, dat);
1139                 }
1140                 last->offset = e->idx.offset;
1141                 last->depth = e->depth;
1142         }
1143         return 0;
1144 }
1145
1146 static void truncate_pack(struct sha1file_checkpoint *checkpoint)
1147 {
1148         if (sha1file_truncate(pack_file, checkpoint))
1149                 die_errno("cannot truncate pack to skip duplicate");
1150         pack_size = checkpoint->offset;
1151 }
1152
1153 static void stream_blob(uintmax_t len, unsigned char *sha1out, uintmax_t mark)
1154 {
1155         size_t in_sz = 64 * 1024, out_sz = 64 * 1024;
1156         unsigned char *in_buf = xmalloc(in_sz);
1157         unsigned char *out_buf = xmalloc(out_sz);
1158         struct object_entry *e;
1159         unsigned char sha1[20];
1160         unsigned long hdrlen;
1161         off_t offset;
1162         git_SHA_CTX c;
1163         git_zstream s;
1164         struct sha1file_checkpoint checkpoint;
1165         int status = Z_OK;
1166
1167         /* Determine if we should auto-checkpoint. */
1168         if ((max_packsize && (pack_size + 60 + len) > max_packsize)
1169                 || (pack_size + 60 + len) < pack_size)
1170                 cycle_packfile();
1171
1172         sha1file_checkpoint(pack_file, &checkpoint);
1173         offset = checkpoint.offset;
1174
1175         hdrlen = snprintf((char *)out_buf, out_sz, "blob %" PRIuMAX, len) + 1;
1176         if (out_sz <= hdrlen)
1177                 die("impossibly large object header");
1178
1179         git_SHA1_Init(&c);
1180         git_SHA1_Update(&c, out_buf, hdrlen);
1181
1182         crc32_begin(pack_file);
1183
1184         memset(&s, 0, sizeof(s));
1185         git_deflate_init(&s, pack_compression_level);
1186
1187         hdrlen = encode_in_pack_object_header(OBJ_BLOB, len, out_buf);
1188         if (out_sz <= hdrlen)
1189                 die("impossibly large object header");
1190
1191         s.next_out = out_buf + hdrlen;
1192         s.avail_out = out_sz - hdrlen;
1193
1194         while (status != Z_STREAM_END) {
1195                 if (0 < len && !s.avail_in) {
1196                         size_t cnt = in_sz < len ? in_sz : (size_t)len;
1197                         size_t n = fread(in_buf, 1, cnt, stdin);
1198                         if (!n && feof(stdin))
1199                                 die("EOF in data (%" PRIuMAX " bytes remaining)", len);
1200
1201                         git_SHA1_Update(&c, in_buf, n);
1202                         s.next_in = in_buf;
1203                         s.avail_in = n;
1204                         len -= n;
1205                 }
1206
1207                 status = git_deflate(&s, len ? 0 : Z_FINISH);
1208
1209                 if (!s.avail_out || status == Z_STREAM_END) {
1210                         size_t n = s.next_out - out_buf;
1211                         sha1write(pack_file, out_buf, n);
1212                         pack_size += n;
1213                         s.next_out = out_buf;
1214                         s.avail_out = out_sz;
1215                 }
1216
1217                 switch (status) {
1218                 case Z_OK:
1219                 case Z_BUF_ERROR:
1220                 case Z_STREAM_END:
1221                         continue;
1222                 default:
1223                         die("unexpected deflate failure: %d", status);
1224                 }
1225         }
1226         git_deflate_end(&s);
1227         git_SHA1_Final(sha1, &c);
1228
1229         if (sha1out)
1230                 hashcpy(sha1out, sha1);
1231
1232         e = insert_object(sha1);
1233
1234         if (mark)
1235                 insert_mark(mark, e);
1236
1237         if (e->idx.offset) {
1238                 duplicate_count_by_type[OBJ_BLOB]++;
1239                 truncate_pack(&checkpoint);
1240
1241         } else if (find_sha1_pack(sha1, packed_git)) {
1242                 e->type = OBJ_BLOB;
1243                 e->pack_id = MAX_PACK_ID;
1244                 e->idx.offset = 1; /* just not zero! */
1245                 duplicate_count_by_type[OBJ_BLOB]++;
1246                 truncate_pack(&checkpoint);
1247
1248         } else {
1249                 e->depth = 0;
1250                 e->type = OBJ_BLOB;
1251                 e->pack_id = pack_id;
1252                 e->idx.offset = offset;
1253                 e->idx.crc32 = crc32_end(pack_file);
1254                 object_count++;
1255                 object_count_by_type[OBJ_BLOB]++;
1256         }
1257
1258         free(in_buf);
1259         free(out_buf);
1260 }
1261
1262 /* All calls must be guarded by find_object() or find_mark() to
1263  * ensure the 'struct object_entry' passed was written by this
1264  * process instance.  We unpack the entry by the offset, avoiding
1265  * the need for the corresponding .idx file.  This unpacking rule
1266  * works because we only use OBJ_REF_DELTA within the packfiles
1267  * created by fast-import.
1268  *
1269  * oe must not be NULL.  Such an oe usually comes from giving
1270  * an unknown SHA-1 to find_object() or an undefined mark to
1271  * find_mark().  Callers must test for this condition and use
1272  * the standard read_sha1_file() when it happens.
1273  *
1274  * oe->pack_id must not be MAX_PACK_ID.  Such an oe is usually from
1275  * find_mark(), where the mark was reloaded from an existing marks
1276  * file and is referencing an object that this fast-import process
1277  * instance did not write out to a packfile.  Callers must test for
1278  * this condition and use read_sha1_file() instead.
1279  */
1280 static void *gfi_unpack_entry(
1281         struct object_entry *oe,
1282         unsigned long *sizep)
1283 {
1284         enum object_type type;
1285         struct packed_git *p = all_packs[oe->pack_id];
1286         if (p == pack_data && p->pack_size < (pack_size + 20)) {
1287                 /* The object is stored in the packfile we are writing to
1288                  * and we have modified it since the last time we scanned
1289                  * back to read a previously written object.  If an old
1290                  * window covered [p->pack_size, p->pack_size + 20) its
1291                  * data is stale and is not valid.  Closing all windows
1292                  * and updating the packfile length ensures we can read
1293                  * the newly written data.
1294                  */
1295                 close_pack_windows(p);
1296                 sha1flush(pack_file);
1297
1298                 /* We have to offer 20 bytes additional on the end of
1299                  * the packfile as the core unpacker code assumes the
1300                  * footer is present at the file end and must promise
1301                  * at least 20 bytes within any window it maps.  But
1302                  * we don't actually create the footer here.
1303                  */
1304                 p->pack_size = pack_size + 20;
1305         }
1306         return unpack_entry(p, oe->idx.offset, &type, sizep);
1307 }
1308
1309 static const char *get_mode(const char *str, uint16_t *modep)
1310 {
1311         unsigned char c;
1312         uint16_t mode = 0;
1313
1314         while ((c = *str++) != ' ') {
1315                 if (c < '0' || c > '7')
1316                         return NULL;
1317                 mode = (mode << 3) + (c - '0');
1318         }
1319         *modep = mode;
1320         return str;
1321 }
1322
1323 static void load_tree(struct tree_entry *root)
1324 {
1325         unsigned char *sha1 = root->versions[1].sha1;
1326         struct object_entry *myoe;
1327         struct tree_content *t;
1328         unsigned long size;
1329         char *buf;
1330         const char *c;
1331
1332         root->tree = t = new_tree_content(8);
1333         if (is_null_sha1(sha1))
1334                 return;
1335
1336         myoe = find_object(sha1);
1337         if (myoe && myoe->pack_id != MAX_PACK_ID) {
1338                 if (myoe->type != OBJ_TREE)
1339                         die("Not a tree: %s", sha1_to_hex(sha1));
1340                 t->delta_depth = myoe->depth;
1341                 buf = gfi_unpack_entry(myoe, &size);
1342                 if (!buf)
1343                         die("Can't load tree %s", sha1_to_hex(sha1));
1344         } else {
1345                 enum object_type type;
1346                 buf = read_sha1_file(sha1, &type, &size);
1347                 if (!buf || type != OBJ_TREE)
1348                         die("Can't load tree %s", sha1_to_hex(sha1));
1349         }
1350
1351         c = buf;
1352         while (c != (buf + size)) {
1353                 struct tree_entry *e = new_tree_entry();
1354
1355                 if (t->entry_count == t->entry_capacity)
1356                         root->tree = t = grow_tree_content(t, t->entry_count);
1357                 t->entries[t->entry_count++] = e;
1358
1359                 e->tree = NULL;
1360                 c = get_mode(c, &e->versions[1].mode);
1361                 if (!c)
1362                         die("Corrupt mode in %s", sha1_to_hex(sha1));
1363                 e->versions[0].mode = e->versions[1].mode;
1364                 e->name = to_atom(c, strlen(c));
1365                 c += e->name->str_len + 1;
1366                 hashcpy(e->versions[0].sha1, (unsigned char *)c);
1367                 hashcpy(e->versions[1].sha1, (unsigned char *)c);
1368                 c += 20;
1369         }
1370         free(buf);
1371 }
1372
1373 static int tecmp0 (const void *_a, const void *_b)
1374 {
1375         struct tree_entry *a = *((struct tree_entry**)_a);
1376         struct tree_entry *b = *((struct tree_entry**)_b);
1377         return base_name_compare(
1378                 a->name->str_dat, a->name->str_len, a->versions[0].mode,
1379                 b->name->str_dat, b->name->str_len, b->versions[0].mode);
1380 }
1381
1382 static int tecmp1 (const void *_a, const void *_b)
1383 {
1384         struct tree_entry *a = *((struct tree_entry**)_a);
1385         struct tree_entry *b = *((struct tree_entry**)_b);
1386         return base_name_compare(
1387                 a->name->str_dat, a->name->str_len, a->versions[1].mode,
1388                 b->name->str_dat, b->name->str_len, b->versions[1].mode);
1389 }
1390
1391 static void mktree(struct tree_content *t, int v, struct strbuf *b)
1392 {
1393         size_t maxlen = 0;
1394         unsigned int i;
1395
1396         if (!v)
1397                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp0);
1398         else
1399                 qsort(t->entries,t->entry_count,sizeof(t->entries[0]),tecmp1);
1400
1401         for (i = 0; i < t->entry_count; i++) {
1402                 if (t->entries[i]->versions[v].mode)
1403                         maxlen += t->entries[i]->name->str_len + 34;
1404         }
1405
1406         strbuf_reset(b);
1407         strbuf_grow(b, maxlen);
1408         for (i = 0; i < t->entry_count; i++) {
1409                 struct tree_entry *e = t->entries[i];
1410                 if (!e->versions[v].mode)
1411                         continue;
1412                 strbuf_addf(b, "%o %s%c",
1413                         (unsigned int)(e->versions[v].mode & ~NO_DELTA),
1414                         e->name->str_dat, '\0');
1415                 strbuf_add(b, e->versions[v].sha1, 20);
1416         }
1417 }
1418
1419 static void store_tree(struct tree_entry *root)
1420 {
1421         struct tree_content *t = root->tree;
1422         unsigned int i, j, del;
1423         struct last_object lo = { STRBUF_INIT, 0, 0, /* no_swap */ 1 };
1424         struct object_entry *le = NULL;
1425
1426         if (!is_null_sha1(root->versions[1].sha1))
1427                 return;
1428
1429         for (i = 0; i < t->entry_count; i++) {
1430                 if (t->entries[i]->tree)
1431                         store_tree(t->entries[i]);
1432         }
1433
1434         if (!(root->versions[0].mode & NO_DELTA))
1435                 le = find_object(root->versions[0].sha1);
1436         if (S_ISDIR(root->versions[0].mode) && le && le->pack_id == pack_id) {
1437                 mktree(t, 0, &old_tree);
1438                 lo.data = old_tree;
1439                 lo.offset = le->idx.offset;
1440                 lo.depth = t->delta_depth;
1441         }
1442
1443         mktree(t, 1, &new_tree);
1444         store_object(OBJ_TREE, &new_tree, &lo, root->versions[1].sha1, 0);
1445
1446         t->delta_depth = lo.depth;
1447         for (i = 0, j = 0, del = 0; i < t->entry_count; i++) {
1448                 struct tree_entry *e = t->entries[i];
1449                 if (e->versions[1].mode) {
1450                         e->versions[0].mode = e->versions[1].mode;
1451                         hashcpy(e->versions[0].sha1, e->versions[1].sha1);
1452                         t->entries[j++] = e;
1453                 } else {
1454                         release_tree_entry(e);
1455                         del++;
1456                 }
1457         }
1458         t->entry_count -= del;
1459 }
1460
1461 static void tree_content_replace(
1462         struct tree_entry *root,
1463         const unsigned char *sha1,
1464         const uint16_t mode,
1465         struct tree_content *newtree)
1466 {
1467         if (!S_ISDIR(mode))
1468                 die("Root cannot be a non-directory");
1469         hashclr(root->versions[0].sha1);
1470         hashcpy(root->versions[1].sha1, sha1);
1471         if (root->tree)
1472                 release_tree_content_recursive(root->tree);
1473         root->tree = newtree;
1474 }
1475
1476 static int tree_content_set(
1477         struct tree_entry *root,
1478         const char *p,
1479         const unsigned char *sha1,
1480         const uint16_t mode,
1481         struct tree_content *subtree)
1482 {
1483         struct tree_content *t;
1484         const char *slash1;
1485         unsigned int i, n;
1486         struct tree_entry *e;
1487
1488         slash1 = strchrnul(p, '/');
1489         n = slash1 - p;
1490         if (!n)
1491                 die("Empty path component found in input");
1492         if (!*slash1 && !S_ISDIR(mode) && subtree)
1493                 die("Non-directories cannot have subtrees");
1494
1495         if (!root->tree)
1496                 load_tree(root);
1497         t = root->tree;
1498         for (i = 0; i < t->entry_count; i++) {
1499                 e = t->entries[i];
1500                 if (e->name->str_len == n && !strncmp_icase(p, e->name->str_dat, n)) {
1501                         if (!*slash1) {
1502                                 if (!S_ISDIR(mode)
1503                                                 && e->versions[1].mode == mode
1504                                                 && !hashcmp(e->versions[1].sha1, sha1))
1505                                         return 0;
1506                                 e->versions[1].mode = mode;
1507                                 hashcpy(e->versions[1].sha1, sha1);
1508                                 if (e->tree)
1509                                         release_tree_content_recursive(e->tree);
1510                                 e->tree = subtree;
1511
1512                                 /*
1513                                  * We need to leave e->versions[0].sha1 alone
1514                                  * to avoid modifying the preimage tree used
1515                                  * when writing out the parent directory.
1516                                  * But after replacing the subdir with a
1517                                  * completely different one, it's not a good
1518                                  * delta base any more, and besides, we've
1519                                  * thrown away the tree entries needed to
1520                                  * make a delta against it.
1521                                  *
1522                                  * So let's just explicitly disable deltas
1523                                  * for the subtree.
1524                                  */
1525                                 if (S_ISDIR(e->versions[0].mode))
1526                                         e->versions[0].mode |= NO_DELTA;
1527
1528                                 hashclr(root->versions[1].sha1);
1529                                 return 1;
1530                         }
1531                         if (!S_ISDIR(e->versions[1].mode)) {
1532                                 e->tree = new_tree_content(8);
1533                                 e->versions[1].mode = S_IFDIR;
1534                         }
1535                         if (!e->tree)
1536                                 load_tree(e);
1537                         if (tree_content_set(e, slash1 + 1, sha1, mode, subtree)) {
1538                                 hashclr(root->versions[1].sha1);
1539                                 return 1;
1540                         }
1541                         return 0;
1542                 }
1543         }
1544
1545         if (t->entry_count == t->entry_capacity)
1546                 root->tree = t = grow_tree_content(t, t->entry_count);
1547         e = new_tree_entry();
1548         e->name = to_atom(p, n);
1549         e->versions[0].mode = 0;
1550         hashclr(e->versions[0].sha1);
1551         t->entries[t->entry_count++] = e;
1552         if (*slash1) {
1553                 e->tree = new_tree_content(8);
1554                 e->versions[1].mode = S_IFDIR;
1555                 tree_content_set(e, slash1 + 1, sha1, mode, subtree);
1556         } else {
1557                 e->tree = subtree;
1558                 e->versions[1].mode = mode;
1559                 hashcpy(e->versions[1].sha1, sha1);
1560         }
1561         hashclr(root->versions[1].sha1);
1562         return 1;
1563 }
1564
1565 static int tree_content_remove(
1566         struct tree_entry *root,
1567         const char *p,
1568         struct tree_entry *backup_leaf,
1569         int allow_root)
1570 {
1571         struct tree_content *t;
1572         const char *slash1;
1573         unsigned int i, n;
1574         struct tree_entry *e;
1575
1576         slash1 = strchrnul(p, '/');
1577         n = slash1 - p;
1578
1579         if (!root->tree)
1580                 load_tree(root);
1581
1582         if (!*p && allow_root) {
1583                 e = root;
1584                 goto del_entry;
1585         }
1586
1587         t = root->tree;
1588         for (i = 0; i < t->entry_count; i++) {
1589                 e = t->entries[i];
1590                 if (e->name->str_len == n && !strncmp_icase(p, e->name->str_dat, n)) {
1591                         if (*slash1 && !S_ISDIR(e->versions[1].mode))
1592                                 /*
1593                                  * If p names a file in some subdirectory, and a
1594                                  * file or symlink matching the name of the
1595                                  * parent directory of p exists, then p cannot
1596                                  * exist and need not be deleted.
1597                                  */
1598                                 return 1;
1599                         if (!*slash1 || !S_ISDIR(e->versions[1].mode))
1600                                 goto del_entry;
1601                         if (!e->tree)
1602                                 load_tree(e);
1603                         if (tree_content_remove(e, slash1 + 1, backup_leaf, 0)) {
1604                                 for (n = 0; n < e->tree->entry_count; n++) {
1605                                         if (e->tree->entries[n]->versions[1].mode) {
1606                                                 hashclr(root->versions[1].sha1);
1607                                                 return 1;
1608                                         }
1609                                 }
1610                                 backup_leaf = NULL;
1611                                 goto del_entry;
1612                         }
1613                         return 0;
1614                 }
1615         }
1616         return 0;
1617
1618 del_entry:
1619         if (backup_leaf)
1620                 memcpy(backup_leaf, e, sizeof(*backup_leaf));
1621         else if (e->tree)
1622                 release_tree_content_recursive(e->tree);
1623         e->tree = NULL;
1624         e->versions[1].mode = 0;
1625         hashclr(e->versions[1].sha1);
1626         hashclr(root->versions[1].sha1);
1627         return 1;
1628 }
1629
1630 static int tree_content_get(
1631         struct tree_entry *root,
1632         const char *p,
1633         struct tree_entry *leaf,
1634         int allow_root)
1635 {
1636         struct tree_content *t;
1637         const char *slash1;
1638         unsigned int i, n;
1639         struct tree_entry *e;
1640
1641         slash1 = strchrnul(p, '/');
1642         n = slash1 - p;
1643         if (!n && !allow_root)
1644                 die("Empty path component found in input");
1645
1646         if (!root->tree)
1647                 load_tree(root);
1648
1649         if (!n) {
1650                 e = root;
1651                 goto found_entry;
1652         }
1653
1654         t = root->tree;
1655         for (i = 0; i < t->entry_count; i++) {
1656                 e = t->entries[i];
1657                 if (e->name->str_len == n && !strncmp_icase(p, e->name->str_dat, n)) {
1658                         if (!*slash1)
1659                                 goto found_entry;
1660                         if (!S_ISDIR(e->versions[1].mode))
1661                                 return 0;
1662                         if (!e->tree)
1663                                 load_tree(e);
1664                         return tree_content_get(e, slash1 + 1, leaf, 0);
1665                 }
1666         }
1667         return 0;
1668
1669 found_entry:
1670         memcpy(leaf, e, sizeof(*leaf));
1671         if (e->tree && is_null_sha1(e->versions[1].sha1))
1672                 leaf->tree = dup_tree_content(e->tree);
1673         else
1674                 leaf->tree = NULL;
1675         return 1;
1676 }
1677
1678 static int update_branch(struct branch *b)
1679 {
1680         static const char *msg = "fast-import";
1681         struct ref_lock *lock;
1682         unsigned char old_sha1[20];
1683
1684         if (is_null_sha1(b->sha1))
1685                 return 0;
1686         if (read_ref(b->name, old_sha1))
1687                 hashclr(old_sha1);
1688         lock = lock_any_ref_for_update(b->name, old_sha1, 0, NULL);
1689         if (!lock)
1690                 return error("Unable to lock %s", b->name);
1691         if (!force_update && !is_null_sha1(old_sha1)) {
1692                 struct commit *old_cmit, *new_cmit;
1693
1694                 old_cmit = lookup_commit_reference_gently(old_sha1, 0);
1695                 new_cmit = lookup_commit_reference_gently(b->sha1, 0);
1696                 if (!old_cmit || !new_cmit) {
1697                         unlock_ref(lock);
1698                         return error("Branch %s is missing commits.", b->name);
1699                 }
1700
1701                 if (!in_merge_bases(old_cmit, new_cmit)) {
1702                         unlock_ref(lock);
1703                         warning("Not updating %s"
1704                                 " (new tip %s does not contain %s)",
1705                                 b->name, sha1_to_hex(b->sha1), sha1_to_hex(old_sha1));
1706                         return -1;
1707                 }
1708         }
1709         if (write_ref_sha1(lock, b->sha1, msg) < 0)
1710                 return error("Unable to update %s", b->name);
1711         return 0;
1712 }
1713
1714 static void dump_branches(void)
1715 {
1716         unsigned int i;
1717         struct branch *b;
1718
1719         for (i = 0; i < branch_table_sz; i++) {
1720                 for (b = branch_table[i]; b; b = b->table_next_branch)
1721                         failure |= update_branch(b);
1722         }
1723 }
1724
1725 static void dump_tags(void)
1726 {
1727         static const char *msg = "fast-import";
1728         struct tag *t;
1729         struct ref_lock *lock;
1730         char ref_name[PATH_MAX];
1731
1732         for (t = first_tag; t; t = t->next_tag) {
1733                 sprintf(ref_name, "tags/%s", t->name);
1734                 lock = lock_ref_sha1(ref_name, NULL);
1735                 if (!lock || write_ref_sha1(lock, t->sha1, msg) < 0)
1736                         failure |= error("Unable to update %s", ref_name);
1737         }
1738 }
1739
1740 static void dump_marks_helper(FILE *f,
1741         uintmax_t base,
1742         struct mark_set *m)
1743 {
1744         uintmax_t k;
1745         if (m->shift) {
1746                 for (k = 0; k < 1024; k++) {
1747                         if (m->data.sets[k])
1748                                 dump_marks_helper(f, base + (k << m->shift),
1749                                         m->data.sets[k]);
1750                 }
1751         } else {
1752                 for (k = 0; k < 1024; k++) {
1753                         if (m->data.marked[k])
1754                                 fprintf(f, ":%" PRIuMAX " %s\n", base + k,
1755                                         sha1_to_hex(m->data.marked[k]->idx.sha1));
1756                 }
1757         }
1758 }
1759
1760 static void dump_marks(void)
1761 {
1762         static struct lock_file mark_lock;
1763         int mark_fd;
1764         FILE *f;
1765
1766         if (!export_marks_file)
1767                 return;
1768
1769         mark_fd = hold_lock_file_for_update(&mark_lock, export_marks_file, 0);
1770         if (mark_fd < 0) {
1771                 failure |= error("Unable to write marks file %s: %s",
1772                         export_marks_file, strerror(errno));
1773                 return;
1774         }
1775
1776         f = fdopen(mark_fd, "w");
1777         if (!f) {
1778                 int saved_errno = errno;
1779                 rollback_lock_file(&mark_lock);
1780                 failure |= error("Unable to write marks file %s: %s",
1781                         export_marks_file, strerror(saved_errno));
1782                 return;
1783         }
1784
1785         /*
1786          * Since the lock file was fdopen()'ed, it should not be close()'ed.
1787          * Assign -1 to the lock file descriptor so that commit_lock_file()
1788          * won't try to close() it.
1789          */
1790         mark_lock.fd = -1;
1791
1792         dump_marks_helper(f, 0, marks);
1793         if (ferror(f) || fclose(f)) {
1794                 int saved_errno = errno;
1795                 rollback_lock_file(&mark_lock);
1796                 failure |= error("Unable to write marks file %s: %s",
1797                         export_marks_file, strerror(saved_errno));
1798                 return;
1799         }
1800
1801         if (commit_lock_file(&mark_lock)) {
1802                 int saved_errno = errno;
1803                 rollback_lock_file(&mark_lock);
1804                 failure |= error("Unable to commit marks file %s: %s",
1805                         export_marks_file, strerror(saved_errno));
1806                 return;
1807         }
1808 }
1809
1810 static void read_marks(void)
1811 {
1812         char line[512];
1813         FILE *f = fopen(import_marks_file, "r");
1814         if (f)
1815                 ;
1816         else if (import_marks_file_ignore_missing && errno == ENOENT)
1817                 return; /* Marks file does not exist */
1818         else
1819                 die_errno("cannot read '%s'", import_marks_file);
1820         while (fgets(line, sizeof(line), f)) {
1821                 uintmax_t mark;
1822                 char *end;
1823                 unsigned char sha1[20];
1824                 struct object_entry *e;
1825
1826                 end = strchr(line, '\n');
1827                 if (line[0] != ':' || !end)
1828                         die("corrupt mark line: %s", line);
1829                 *end = 0;
1830                 mark = strtoumax(line + 1, &end, 10);
1831                 if (!mark || end == line + 1
1832                         || *end != ' ' || get_sha1_hex(end + 1, sha1))
1833                         die("corrupt mark line: %s", line);
1834                 e = find_object(sha1);
1835                 if (!e) {
1836                         enum object_type type = sha1_object_info(sha1, NULL);
1837                         if (type < 0)
1838                                 die("object not found: %s", sha1_to_hex(sha1));
1839                         e = insert_object(sha1);
1840                         e->type = type;
1841                         e->pack_id = MAX_PACK_ID;
1842                         e->idx.offset = 1; /* just not zero! */
1843                 }
1844                 insert_mark(mark, e);
1845         }
1846         fclose(f);
1847 }
1848
1849
1850 static int read_next_command(void)
1851 {
1852         static int stdin_eof = 0;
1853
1854         if (stdin_eof) {
1855                 unread_command_buf = 0;
1856                 return EOF;
1857         }
1858
1859         for (;;) {
1860                 if (unread_command_buf) {
1861                         unread_command_buf = 0;
1862                 } else {
1863                         struct recent_command *rc;
1864
1865                         strbuf_detach(&command_buf, NULL);
1866                         stdin_eof = strbuf_getline(&command_buf, stdin, '\n');
1867                         if (stdin_eof)
1868                                 return EOF;
1869
1870                         if (!seen_data_command
1871                                 && !starts_with(command_buf.buf, "feature ")
1872                                 && !starts_with(command_buf.buf, "option ")) {
1873                                 parse_argv();
1874                         }
1875
1876                         rc = rc_free;
1877                         if (rc)
1878                                 rc_free = rc->next;
1879                         else {
1880                                 rc = cmd_hist.next;
1881                                 cmd_hist.next = rc->next;
1882                                 cmd_hist.next->prev = &cmd_hist;
1883                                 free(rc->buf);
1884                         }
1885
1886                         rc->buf = command_buf.buf;
1887                         rc->prev = cmd_tail;
1888                         rc->next = cmd_hist.prev;
1889                         rc->prev->next = rc;
1890                         cmd_tail = rc;
1891                 }
1892                 if (starts_with(command_buf.buf, "cat-blob ")) {
1893                         parse_cat_blob();
1894                         continue;
1895                 }
1896                 if (command_buf.buf[0] == '#')
1897                         continue;
1898                 return 0;
1899         }
1900 }
1901
1902 static void skip_optional_lf(void)
1903 {
1904         int term_char = fgetc(stdin);
1905         if (term_char != '\n' && term_char != EOF)
1906                 ungetc(term_char, stdin);
1907 }
1908
1909 static void parse_mark(void)
1910 {
1911         if (starts_with(command_buf.buf, "mark :")) {
1912                 next_mark = strtoumax(command_buf.buf + 6, NULL, 10);
1913                 read_next_command();
1914         }
1915         else
1916                 next_mark = 0;
1917 }
1918
1919 static int parse_data(struct strbuf *sb, uintmax_t limit, uintmax_t *len_res)
1920 {
1921         strbuf_reset(sb);
1922
1923         if (!starts_with(command_buf.buf, "data "))
1924                 die("Expected 'data n' command, found: %s", command_buf.buf);
1925
1926         if (starts_with(command_buf.buf + 5, "<<")) {
1927                 char *term = xstrdup(command_buf.buf + 5 + 2);
1928                 size_t term_len = command_buf.len - 5 - 2;
1929
1930                 strbuf_detach(&command_buf, NULL);
1931                 for (;;) {
1932                         if (strbuf_getline(&command_buf, stdin, '\n') == EOF)
1933                                 die("EOF in data (terminator '%s' not found)", term);
1934                         if (term_len == command_buf.len
1935                                 && !strcmp(term, command_buf.buf))
1936                                 break;
1937                         strbuf_addbuf(sb, &command_buf);
1938                         strbuf_addch(sb, '\n');
1939                 }
1940                 free(term);
1941         }
1942         else {
1943                 uintmax_t len = strtoumax(command_buf.buf + 5, NULL, 10);
1944                 size_t n = 0, length = (size_t)len;
1945
1946                 if (limit && limit < len) {
1947                         *len_res = len;
1948                         return 0;
1949                 }
1950                 if (length < len)
1951                         die("data is too large to use in this context");
1952
1953                 while (n < length) {
1954                         size_t s = strbuf_fread(sb, length - n, stdin);
1955                         if (!s && feof(stdin))
1956                                 die("EOF in data (%lu bytes remaining)",
1957                                         (unsigned long)(length - n));
1958                         n += s;
1959                 }
1960         }
1961
1962         skip_optional_lf();
1963         return 1;
1964 }
1965
1966 static int validate_raw_date(const char *src, char *result, int maxlen)
1967 {
1968         const char *orig_src = src;
1969         char *endp;
1970         unsigned long num;
1971
1972         errno = 0;
1973
1974         num = strtoul(src, &endp, 10);
1975         /* NEEDSWORK: perhaps check for reasonable values? */
1976         if (errno || endp == src || *endp != ' ')
1977                 return -1;
1978
1979         src = endp + 1;
1980         if (*src != '-' && *src != '+')
1981                 return -1;
1982
1983         num = strtoul(src + 1, &endp, 10);
1984         if (errno || endp == src + 1 || *endp || (endp - orig_src) >= maxlen ||
1985             1400 < num)
1986                 return -1;
1987
1988         strcpy(result, orig_src);
1989         return 0;
1990 }
1991
1992 static char *parse_ident(const char *buf)
1993 {
1994         const char *ltgt;
1995         size_t name_len;
1996         char *ident;
1997
1998         /* ensure there is a space delimiter even if there is no name */
1999         if (*buf == '<')
2000                 --buf;
2001
2002         ltgt = buf + strcspn(buf, "<>");
2003         if (*ltgt != '<')
2004                 die("Missing < in ident string: %s", buf);
2005         if (ltgt != buf && ltgt[-1] != ' ')
2006                 die("Missing space before < in ident string: %s", buf);
2007         ltgt = ltgt + 1 + strcspn(ltgt + 1, "<>");
2008         if (*ltgt != '>')
2009                 die("Missing > in ident string: %s", buf);
2010         ltgt++;
2011         if (*ltgt != ' ')
2012                 die("Missing space after > in ident string: %s", buf);
2013         ltgt++;
2014         name_len = ltgt - buf;
2015         ident = xmalloc(name_len + 24);
2016         strncpy(ident, buf, name_len);
2017
2018         switch (whenspec) {
2019         case WHENSPEC_RAW:
2020                 if (validate_raw_date(ltgt, ident + name_len, 24) < 0)
2021                         die("Invalid raw date \"%s\" in ident: %s", ltgt, buf);
2022                 break;
2023         case WHENSPEC_RFC2822:
2024                 if (parse_date(ltgt, ident + name_len, 24) < 0)
2025                         die("Invalid rfc2822 date \"%s\" in ident: %s", ltgt, buf);
2026                 break;
2027         case WHENSPEC_NOW:
2028                 if (strcmp("now", ltgt))
2029                         die("Date in ident must be 'now': %s", buf);
2030                 datestamp(ident + name_len, 24);
2031                 break;
2032         }
2033
2034         return ident;
2035 }
2036
2037 static void parse_and_store_blob(
2038         struct last_object *last,
2039         unsigned char *sha1out,
2040         uintmax_t mark)
2041 {
2042         static struct strbuf buf = STRBUF_INIT;
2043         uintmax_t len;
2044
2045         if (parse_data(&buf, big_file_threshold, &len))
2046                 store_object(OBJ_BLOB, &buf, last, sha1out, mark);
2047         else {
2048                 if (last) {
2049                         strbuf_release(&last->data);
2050                         last->offset = 0;
2051                         last->depth = 0;
2052                 }
2053                 stream_blob(len, sha1out, mark);
2054                 skip_optional_lf();
2055         }
2056 }
2057
2058 static void parse_new_blob(void)
2059 {
2060         read_next_command();
2061         parse_mark();
2062         parse_and_store_blob(&last_blob, NULL, next_mark);
2063 }
2064
2065 static void unload_one_branch(void)
2066 {
2067         while (cur_active_branches
2068                 && cur_active_branches >= max_active_branches) {
2069                 uintmax_t min_commit = ULONG_MAX;
2070                 struct branch *e, *l = NULL, *p = NULL;
2071
2072                 for (e = active_branches; e; e = e->active_next_branch) {
2073                         if (e->last_commit < min_commit) {
2074                                 p = l;
2075                                 min_commit = e->last_commit;
2076                         }
2077                         l = e;
2078                 }
2079
2080                 if (p) {
2081                         e = p->active_next_branch;
2082                         p->active_next_branch = e->active_next_branch;
2083                 } else {
2084                         e = active_branches;
2085                         active_branches = e->active_next_branch;
2086                 }
2087                 e->active = 0;
2088                 e->active_next_branch = NULL;
2089                 if (e->branch_tree.tree) {
2090                         release_tree_content_recursive(e->branch_tree.tree);
2091                         e->branch_tree.tree = NULL;
2092                 }
2093                 cur_active_branches--;
2094         }
2095 }
2096
2097 static void load_branch(struct branch *b)
2098 {
2099         load_tree(&b->branch_tree);
2100         if (!b->active) {
2101                 b->active = 1;
2102                 b->active_next_branch = active_branches;
2103                 active_branches = b;
2104                 cur_active_branches++;
2105                 branch_load_count++;
2106         }
2107 }
2108
2109 static unsigned char convert_num_notes_to_fanout(uintmax_t num_notes)
2110 {
2111         unsigned char fanout = 0;
2112         while ((num_notes >>= 8))
2113                 fanout++;
2114         return fanout;
2115 }
2116
2117 static void construct_path_with_fanout(const char *hex_sha1,
2118                 unsigned char fanout, char *path)
2119 {
2120         unsigned int i = 0, j = 0;
2121         if (fanout >= 20)
2122                 die("Too large fanout (%u)", fanout);
2123         while (fanout) {
2124                 path[i++] = hex_sha1[j++];
2125                 path[i++] = hex_sha1[j++];
2126                 path[i++] = '/';
2127                 fanout--;
2128         }
2129         memcpy(path + i, hex_sha1 + j, 40 - j);
2130         path[i + 40 - j] = '\0';
2131 }
2132
2133 static uintmax_t do_change_note_fanout(
2134                 struct tree_entry *orig_root, struct tree_entry *root,
2135                 char *hex_sha1, unsigned int hex_sha1_len,
2136                 char *fullpath, unsigned int fullpath_len,
2137                 unsigned char fanout)
2138 {
2139         struct tree_content *t = root->tree;
2140         struct tree_entry *e, leaf;
2141         unsigned int i, tmp_hex_sha1_len, tmp_fullpath_len;
2142         uintmax_t num_notes = 0;
2143         unsigned char sha1[20];
2144         char realpath[60];
2145
2146         for (i = 0; t && i < t->entry_count; i++) {
2147                 e = t->entries[i];
2148                 tmp_hex_sha1_len = hex_sha1_len + e->name->str_len;
2149                 tmp_fullpath_len = fullpath_len;
2150
2151                 /*
2152                  * We're interested in EITHER existing note entries (entries
2153                  * with exactly 40 hex chars in path, not including directory
2154                  * separators), OR directory entries that may contain note
2155                  * entries (with < 40 hex chars in path).
2156                  * Also, each path component in a note entry must be a multiple
2157                  * of 2 chars.
2158                  */
2159                 if (!e->versions[1].mode ||
2160                     tmp_hex_sha1_len > 40 ||
2161                     e->name->str_len % 2)
2162                         continue;
2163
2164                 /* This _may_ be a note entry, or a subdir containing notes */
2165                 memcpy(hex_sha1 + hex_sha1_len, e->name->str_dat,
2166                        e->name->str_len);
2167                 if (tmp_fullpath_len)
2168                         fullpath[tmp_fullpath_len++] = '/';
2169                 memcpy(fullpath + tmp_fullpath_len, e->name->str_dat,
2170                        e->name->str_len);
2171                 tmp_fullpath_len += e->name->str_len;
2172                 fullpath[tmp_fullpath_len] = '\0';
2173
2174                 if (tmp_hex_sha1_len == 40 && !get_sha1_hex(hex_sha1, sha1)) {
2175                         /* This is a note entry */
2176                         if (fanout == 0xff) {
2177                                 /* Counting mode, no rename */
2178                                 num_notes++;
2179                                 continue;
2180                         }
2181                         construct_path_with_fanout(hex_sha1, fanout, realpath);
2182                         if (!strcmp(fullpath, realpath)) {
2183                                 /* Note entry is in correct location */
2184                                 num_notes++;
2185                                 continue;
2186                         }
2187
2188                         /* Rename fullpath to realpath */
2189                         if (!tree_content_remove(orig_root, fullpath, &leaf, 0))
2190                                 die("Failed to remove path %s", fullpath);
2191                         tree_content_set(orig_root, realpath,
2192                                 leaf.versions[1].sha1,
2193                                 leaf.versions[1].mode,
2194                                 leaf.tree);
2195                 } else if (S_ISDIR(e->versions[1].mode)) {
2196                         /* This is a subdir that may contain note entries */
2197                         if (!e->tree)
2198                                 load_tree(e);
2199                         num_notes += do_change_note_fanout(orig_root, e,
2200                                 hex_sha1, tmp_hex_sha1_len,
2201                                 fullpath, tmp_fullpath_len, fanout);
2202                 }
2203
2204                 /* The above may have reallocated the current tree_content */
2205                 t = root->tree;
2206         }
2207         return num_notes;
2208 }
2209
2210 static uintmax_t change_note_fanout(struct tree_entry *root,
2211                 unsigned char fanout)
2212 {
2213         char hex_sha1[40], path[60];
2214         return do_change_note_fanout(root, root, hex_sha1, 0, path, 0, fanout);
2215 }
2216
2217 /*
2218  * Given a pointer into a string, parse a mark reference:
2219  *
2220  *   idnum ::= ':' bigint;
2221  *
2222  * Return the first character after the value in *endptr.
2223  *
2224  * Complain if the following character is not what is expected,
2225  * either a space or end of the string.
2226  */
2227 static uintmax_t parse_mark_ref(const char *p, char **endptr)
2228 {
2229         uintmax_t mark;
2230
2231         assert(*p == ':');
2232         p++;
2233         mark = strtoumax(p, endptr, 10);
2234         if (*endptr == p)
2235                 die("No value after ':' in mark: %s", command_buf.buf);
2236         return mark;
2237 }
2238
2239 /*
2240  * Parse the mark reference, and complain if this is not the end of
2241  * the string.
2242  */
2243 static uintmax_t parse_mark_ref_eol(const char *p)
2244 {
2245         char *end;
2246         uintmax_t mark;
2247
2248         mark = parse_mark_ref(p, &end);
2249         if (*end != '\0')
2250                 die("Garbage after mark: %s", command_buf.buf);
2251         return mark;
2252 }
2253
2254 /*
2255  * Parse the mark reference, demanding a trailing space.  Return a
2256  * pointer to the space.
2257  */
2258 static uintmax_t parse_mark_ref_space(const char **p)
2259 {
2260         uintmax_t mark;
2261         char *end;
2262
2263         mark = parse_mark_ref(*p, &end);
2264         if (*end != ' ')
2265                 die("Missing space after mark: %s", command_buf.buf);
2266         *p = end;
2267         return mark;
2268 }
2269
2270 static void file_change_m(struct branch *b)
2271 {
2272         const char *p = command_buf.buf + 2;
2273         static struct strbuf uq = STRBUF_INIT;
2274         const char *endp;
2275         struct object_entry *oe;
2276         unsigned char sha1[20];
2277         uint16_t mode, inline_data = 0;
2278
2279         p = get_mode(p, &mode);
2280         if (!p)
2281                 die("Corrupt mode: %s", command_buf.buf);
2282         switch (mode) {
2283         case 0644:
2284         case 0755:
2285                 mode |= S_IFREG;
2286         case S_IFREG | 0644:
2287         case S_IFREG | 0755:
2288         case S_IFLNK:
2289         case S_IFDIR:
2290         case S_IFGITLINK:
2291                 /* ok */
2292                 break;
2293         default:
2294                 die("Corrupt mode: %s", command_buf.buf);
2295         }
2296
2297         if (*p == ':') {
2298                 oe = find_mark(parse_mark_ref_space(&p));
2299                 hashcpy(sha1, oe->idx.sha1);
2300         } else if (starts_with(p, "inline ")) {
2301                 inline_data = 1;
2302                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2303                 p += strlen("inline");  /* advance to space */
2304         } else {
2305                 if (get_sha1_hex(p, sha1))
2306                         die("Invalid dataref: %s", command_buf.buf);
2307                 oe = find_object(sha1);
2308                 p += 40;
2309                 if (*p != ' ')
2310                         die("Missing space after SHA1: %s", command_buf.buf);
2311         }
2312         assert(*p == ' ');
2313         p++;  /* skip space */
2314
2315         strbuf_reset(&uq);
2316         if (!unquote_c_style(&uq, p, &endp)) {
2317                 if (*endp)
2318                         die("Garbage after path in: %s", command_buf.buf);
2319                 p = uq.buf;
2320         }
2321
2322         /* Git does not track empty, non-toplevel directories. */
2323         if (S_ISDIR(mode) && !memcmp(sha1, EMPTY_TREE_SHA1_BIN, 20) && *p) {
2324                 tree_content_remove(&b->branch_tree, p, NULL, 0);
2325                 return;
2326         }
2327
2328         if (S_ISGITLINK(mode)) {
2329                 if (inline_data)
2330                         die("Git links cannot be specified 'inline': %s",
2331                                 command_buf.buf);
2332                 else if (oe) {
2333                         if (oe->type != OBJ_COMMIT)
2334                                 die("Not a commit (actually a %s): %s",
2335                                         typename(oe->type), command_buf.buf);
2336                 }
2337                 /*
2338                  * Accept the sha1 without checking; it expected to be in
2339                  * another repository.
2340                  */
2341         } else if (inline_data) {
2342                 if (S_ISDIR(mode))
2343                         die("Directories cannot be specified 'inline': %s",
2344                                 command_buf.buf);
2345                 if (p != uq.buf) {
2346                         strbuf_addstr(&uq, p);
2347                         p = uq.buf;
2348                 }
2349                 read_next_command();
2350                 parse_and_store_blob(&last_blob, sha1, 0);
2351         } else {
2352                 enum object_type expected = S_ISDIR(mode) ?
2353                                                 OBJ_TREE: OBJ_BLOB;
2354                 enum object_type type = oe ? oe->type :
2355                                         sha1_object_info(sha1, NULL);
2356                 if (type < 0)
2357                         die("%s not found: %s",
2358                                         S_ISDIR(mode) ?  "Tree" : "Blob",
2359                                         command_buf.buf);
2360                 if (type != expected)
2361                         die("Not a %s (actually a %s): %s",
2362                                 typename(expected), typename(type),
2363                                 command_buf.buf);
2364         }
2365
2366         if (!*p) {
2367                 tree_content_replace(&b->branch_tree, sha1, mode, NULL);
2368                 return;
2369         }
2370         tree_content_set(&b->branch_tree, p, sha1, mode, NULL);
2371 }
2372
2373 static void file_change_d(struct branch *b)
2374 {
2375         const char *p = command_buf.buf + 2;
2376         static struct strbuf uq = STRBUF_INIT;
2377         const char *endp;
2378
2379         strbuf_reset(&uq);
2380         if (!unquote_c_style(&uq, p, &endp)) {
2381                 if (*endp)
2382                         die("Garbage after path in: %s", command_buf.buf);
2383                 p = uq.buf;
2384         }
2385         tree_content_remove(&b->branch_tree, p, NULL, 1);
2386 }
2387
2388 static void file_change_cr(struct branch *b, int rename)
2389 {
2390         const char *s, *d;
2391         static struct strbuf s_uq = STRBUF_INIT;
2392         static struct strbuf d_uq = STRBUF_INIT;
2393         const char *endp;
2394         struct tree_entry leaf;
2395
2396         s = command_buf.buf + 2;
2397         strbuf_reset(&s_uq);
2398         if (!unquote_c_style(&s_uq, s, &endp)) {
2399                 if (*endp != ' ')
2400                         die("Missing space after source: %s", command_buf.buf);
2401         } else {
2402                 endp = strchr(s, ' ');
2403                 if (!endp)
2404                         die("Missing space after source: %s", command_buf.buf);
2405                 strbuf_add(&s_uq, s, endp - s);
2406         }
2407         s = s_uq.buf;
2408
2409         endp++;
2410         if (!*endp)
2411                 die("Missing dest: %s", command_buf.buf);
2412
2413         d = endp;
2414         strbuf_reset(&d_uq);
2415         if (!unquote_c_style(&d_uq, d, &endp)) {
2416                 if (*endp)
2417                         die("Garbage after dest in: %s", command_buf.buf);
2418                 d = d_uq.buf;
2419         }
2420
2421         memset(&leaf, 0, sizeof(leaf));
2422         if (rename)
2423                 tree_content_remove(&b->branch_tree, s, &leaf, 1);
2424         else
2425                 tree_content_get(&b->branch_tree, s, &leaf, 1);
2426         if (!leaf.versions[1].mode)
2427                 die("Path %s not in branch", s);
2428         if (!*d) {      /* C "path/to/subdir" "" */
2429                 tree_content_replace(&b->branch_tree,
2430                         leaf.versions[1].sha1,
2431                         leaf.versions[1].mode,
2432                         leaf.tree);
2433                 return;
2434         }
2435         tree_content_set(&b->branch_tree, d,
2436                 leaf.versions[1].sha1,
2437                 leaf.versions[1].mode,
2438                 leaf.tree);
2439 }
2440
2441 static void note_change_n(struct branch *b, unsigned char *old_fanout)
2442 {
2443         const char *p = command_buf.buf + 2;
2444         static struct strbuf uq = STRBUF_INIT;
2445         struct object_entry *oe;
2446         struct branch *s;
2447         unsigned char sha1[20], commit_sha1[20];
2448         char path[60];
2449         uint16_t inline_data = 0;
2450         unsigned char new_fanout;
2451
2452         /*
2453          * When loading a branch, we don't traverse its tree to count the real
2454          * number of notes (too expensive to do this for all non-note refs).
2455          * This means that recently loaded notes refs might incorrectly have
2456          * b->num_notes == 0, and consequently, old_fanout might be wrong.
2457          *
2458          * Fix this by traversing the tree and counting the number of notes
2459          * when b->num_notes == 0. If the notes tree is truly empty, the
2460          * calculation should not take long.
2461          */
2462         if (b->num_notes == 0 && *old_fanout == 0) {
2463                 /* Invoke change_note_fanout() in "counting mode". */
2464                 b->num_notes = change_note_fanout(&b->branch_tree, 0xff);
2465                 *old_fanout = convert_num_notes_to_fanout(b->num_notes);
2466         }
2467
2468         /* Now parse the notemodify command. */
2469         /* <dataref> or 'inline' */
2470         if (*p == ':') {
2471                 oe = find_mark(parse_mark_ref_space(&p));
2472                 hashcpy(sha1, oe->idx.sha1);
2473         } else if (starts_with(p, "inline ")) {
2474                 inline_data = 1;
2475                 oe = NULL; /* not used with inline_data, but makes gcc happy */
2476                 p += strlen("inline");  /* advance to space */
2477         } else {
2478                 if (get_sha1_hex(p, sha1))
2479                         die("Invalid dataref: %s", command_buf.buf);
2480                 oe = find_object(sha1);
2481                 p += 40;
2482                 if (*p != ' ')
2483                         die("Missing space after SHA1: %s", command_buf.buf);
2484         }
2485         assert(*p == ' ');
2486         p++;  /* skip space */
2487
2488         /* <commit-ish> */
2489         s = lookup_branch(p);
2490         if (s) {
2491                 if (is_null_sha1(s->sha1))
2492                         die("Can't add a note on empty branch.");
2493                 hashcpy(commit_sha1, s->sha1);
2494         } else if (*p == ':') {
2495                 uintmax_t commit_mark = parse_mark_ref_eol(p);
2496                 struct object_entry *commit_oe = find_mark(commit_mark);
2497                 if (commit_oe->type != OBJ_COMMIT)
2498                         die("Mark :%" PRIuMAX " not a commit", commit_mark);
2499                 hashcpy(commit_sha1, commit_oe->idx.sha1);
2500         } else if (!get_sha1(p, commit_sha1)) {
2501                 unsigned long size;
2502                 char *buf = read_object_with_reference(commit_sha1,
2503                         commit_type, &size, commit_sha1);
2504                 if (!buf || size < 46)
2505                         die("Not a valid commit: %s", p);
2506                 free(buf);
2507         } else
2508                 die("Invalid ref name or SHA1 expression: %s", p);
2509
2510         if (inline_data) {
2511                 if (p != uq.buf) {
2512                         strbuf_addstr(&uq, p);
2513                         p = uq.buf;
2514                 }
2515                 read_next_command();
2516                 parse_and_store_blob(&last_blob, sha1, 0);
2517         } else if (oe) {
2518                 if (oe->type != OBJ_BLOB)
2519                         die("Not a blob (actually a %s): %s",
2520                                 typename(oe->type), command_buf.buf);
2521         } else if (!is_null_sha1(sha1)) {
2522                 enum object_type type = sha1_object_info(sha1, NULL);
2523                 if (type < 0)
2524                         die("Blob not found: %s", command_buf.buf);
2525                 if (type != OBJ_BLOB)
2526                         die("Not a blob (actually a %s): %s",
2527                             typename(type), command_buf.buf);
2528         }
2529
2530         construct_path_with_fanout(sha1_to_hex(commit_sha1), *old_fanout, path);
2531         if (tree_content_remove(&b->branch_tree, path, NULL, 0))
2532                 b->num_notes--;
2533
2534         if (is_null_sha1(sha1))
2535                 return; /* nothing to insert */
2536
2537         b->num_notes++;
2538         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2539         construct_path_with_fanout(sha1_to_hex(commit_sha1), new_fanout, path);
2540         tree_content_set(&b->branch_tree, path, sha1, S_IFREG | 0644, NULL);
2541 }
2542
2543 static void file_change_deleteall(struct branch *b)
2544 {
2545         release_tree_content_recursive(b->branch_tree.tree);
2546         hashclr(b->branch_tree.versions[0].sha1);
2547         hashclr(b->branch_tree.versions[1].sha1);
2548         load_tree(&b->branch_tree);
2549         b->num_notes = 0;
2550 }
2551
2552 static void parse_from_commit(struct branch *b, char *buf, unsigned long size)
2553 {
2554         if (!buf || size < 46)
2555                 die("Not a valid commit: %s", sha1_to_hex(b->sha1));
2556         if (memcmp("tree ", buf, 5)
2557                 || get_sha1_hex(buf + 5, b->branch_tree.versions[1].sha1))
2558                 die("The commit %s is corrupt", sha1_to_hex(b->sha1));
2559         hashcpy(b->branch_tree.versions[0].sha1,
2560                 b->branch_tree.versions[1].sha1);
2561 }
2562
2563 static void parse_from_existing(struct branch *b)
2564 {
2565         if (is_null_sha1(b->sha1)) {
2566                 hashclr(b->branch_tree.versions[0].sha1);
2567                 hashclr(b->branch_tree.versions[1].sha1);
2568         } else {
2569                 unsigned long size;
2570                 char *buf;
2571
2572                 buf = read_object_with_reference(b->sha1,
2573                         commit_type, &size, b->sha1);
2574                 parse_from_commit(b, buf, size);
2575                 free(buf);
2576         }
2577 }
2578
2579 static int parse_from(struct branch *b)
2580 {
2581         const char *from;
2582         struct branch *s;
2583
2584         if (!starts_with(command_buf.buf, "from "))
2585                 return 0;
2586
2587         if (b->branch_tree.tree) {
2588                 release_tree_content_recursive(b->branch_tree.tree);
2589                 b->branch_tree.tree = NULL;
2590         }
2591
2592         from = strchr(command_buf.buf, ' ') + 1;
2593         s = lookup_branch(from);
2594         if (b == s)
2595                 die("Can't create a branch from itself: %s", b->name);
2596         else if (s) {
2597                 unsigned char *t = s->branch_tree.versions[1].sha1;
2598                 hashcpy(b->sha1, s->sha1);
2599                 hashcpy(b->branch_tree.versions[0].sha1, t);
2600                 hashcpy(b->branch_tree.versions[1].sha1, t);
2601         } else if (*from == ':') {
2602                 uintmax_t idnum = parse_mark_ref_eol(from);
2603                 struct object_entry *oe = find_mark(idnum);
2604                 if (oe->type != OBJ_COMMIT)
2605                         die("Mark :%" PRIuMAX " not a commit", idnum);
2606                 hashcpy(b->sha1, oe->idx.sha1);
2607                 if (oe->pack_id != MAX_PACK_ID) {
2608                         unsigned long size;
2609                         char *buf = gfi_unpack_entry(oe, &size);
2610                         parse_from_commit(b, buf, size);
2611                         free(buf);
2612                 } else
2613                         parse_from_existing(b);
2614         } else if (!get_sha1(from, b->sha1))
2615                 parse_from_existing(b);
2616         else
2617                 die("Invalid ref name or SHA1 expression: %s", from);
2618
2619         read_next_command();
2620         return 1;
2621 }
2622
2623 static struct hash_list *parse_merge(unsigned int *count)
2624 {
2625         struct hash_list *list = NULL, **tail = &list, *n;
2626         const char *from;
2627         struct branch *s;
2628
2629         *count = 0;
2630         while (starts_with(command_buf.buf, "merge ")) {
2631                 from = strchr(command_buf.buf, ' ') + 1;
2632                 n = xmalloc(sizeof(*n));
2633                 s = lookup_branch(from);
2634                 if (s)
2635                         hashcpy(n->sha1, s->sha1);
2636                 else if (*from == ':') {
2637                         uintmax_t idnum = parse_mark_ref_eol(from);
2638                         struct object_entry *oe = find_mark(idnum);
2639                         if (oe->type != OBJ_COMMIT)
2640                                 die("Mark :%" PRIuMAX " not a commit", idnum);
2641                         hashcpy(n->sha1, oe->idx.sha1);
2642                 } else if (!get_sha1(from, n->sha1)) {
2643                         unsigned long size;
2644                         char *buf = read_object_with_reference(n->sha1,
2645                                 commit_type, &size, n->sha1);
2646                         if (!buf || size < 46)
2647                                 die("Not a valid commit: %s", from);
2648                         free(buf);
2649                 } else
2650                         die("Invalid ref name or SHA1 expression: %s", from);
2651
2652                 n->next = NULL;
2653                 *tail = n;
2654                 tail = &n->next;
2655
2656                 (*count)++;
2657                 read_next_command();
2658         }
2659         return list;
2660 }
2661
2662 static void parse_new_commit(void)
2663 {
2664         static struct strbuf msg = STRBUF_INIT;
2665         struct branch *b;
2666         char *sp;
2667         char *author = NULL;
2668         char *committer = NULL;
2669         struct hash_list *merge_list = NULL;
2670         unsigned int merge_count;
2671         unsigned char prev_fanout, new_fanout;
2672
2673         /* Obtain the branch name from the rest of our command */
2674         sp = strchr(command_buf.buf, ' ') + 1;
2675         b = lookup_branch(sp);
2676         if (!b)
2677                 b = new_branch(sp);
2678
2679         read_next_command();
2680         parse_mark();
2681         if (starts_with(command_buf.buf, "author ")) {
2682                 author = parse_ident(command_buf.buf + 7);
2683                 read_next_command();
2684         }
2685         if (starts_with(command_buf.buf, "committer ")) {
2686                 committer = parse_ident(command_buf.buf + 10);
2687                 read_next_command();
2688         }
2689         if (!committer)
2690                 die("Expected committer but didn't get one");
2691         parse_data(&msg, 0, NULL);
2692         read_next_command();
2693         parse_from(b);
2694         merge_list = parse_merge(&merge_count);
2695
2696         /* ensure the branch is active/loaded */
2697         if (!b->branch_tree.tree || !max_active_branches) {
2698                 unload_one_branch();
2699                 load_branch(b);
2700         }
2701
2702         prev_fanout = convert_num_notes_to_fanout(b->num_notes);
2703
2704         /* file_change* */
2705         while (command_buf.len > 0) {
2706                 if (starts_with(command_buf.buf, "M "))
2707                         file_change_m(b);
2708                 else if (starts_with(command_buf.buf, "D "))
2709                         file_change_d(b);
2710                 else if (starts_with(command_buf.buf, "R "))
2711                         file_change_cr(b, 1);
2712                 else if (starts_with(command_buf.buf, "C "))
2713                         file_change_cr(b, 0);
2714                 else if (starts_with(command_buf.buf, "N "))
2715                         note_change_n(b, &prev_fanout);
2716                 else if (!strcmp("deleteall", command_buf.buf))
2717                         file_change_deleteall(b);
2718                 else if (starts_with(command_buf.buf, "ls "))
2719                         parse_ls(b);
2720                 else {
2721                         unread_command_buf = 1;
2722                         break;
2723                 }
2724                 if (read_next_command() == EOF)
2725                         break;
2726         }
2727
2728         new_fanout = convert_num_notes_to_fanout(b->num_notes);
2729         if (new_fanout != prev_fanout)
2730                 b->num_notes = change_note_fanout(&b->branch_tree, new_fanout);
2731
2732         /* build the tree and the commit */
2733         store_tree(&b->branch_tree);
2734         hashcpy(b->branch_tree.versions[0].sha1,
2735                 b->branch_tree.versions[1].sha1);
2736
2737         strbuf_reset(&new_data);
2738         strbuf_addf(&new_data, "tree %s\n",
2739                 sha1_to_hex(b->branch_tree.versions[1].sha1));
2740         if (!is_null_sha1(b->sha1))
2741                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(b->sha1));
2742         while (merge_list) {
2743                 struct hash_list *next = merge_list->next;
2744                 strbuf_addf(&new_data, "parent %s\n", sha1_to_hex(merge_list->sha1));
2745                 free(merge_list);
2746                 merge_list = next;
2747         }
2748         strbuf_addf(&new_data,
2749                 "author %s\n"
2750                 "committer %s\n"
2751                 "\n",
2752                 author ? author : committer, committer);
2753         strbuf_addbuf(&new_data, &msg);
2754         free(author);
2755         free(committer);
2756
2757         if (!store_object(OBJ_COMMIT, &new_data, NULL, b->sha1, next_mark))
2758                 b->pack_id = pack_id;
2759         b->last_commit = object_count_by_type[OBJ_COMMIT];
2760 }
2761
2762 static void parse_new_tag(void)
2763 {
2764         static struct strbuf msg = STRBUF_INIT;
2765         char *sp;
2766         const char *from;
2767         char *tagger;
2768         struct branch *s;
2769         struct tag *t;
2770         uintmax_t from_mark = 0;
2771         unsigned char sha1[20];
2772         enum object_type type;
2773
2774         /* Obtain the new tag name from the rest of our command */
2775         sp = strchr(command_buf.buf, ' ') + 1;
2776         t = pool_alloc(sizeof(struct tag));
2777         memset(t, 0, sizeof(struct tag));
2778         t->name = pool_strdup(sp);
2779         if (last_tag)
2780                 last_tag->next_tag = t;
2781         else
2782                 first_tag = t;
2783         last_tag = t;
2784         read_next_command();
2785
2786         /* from ... */
2787         if (!starts_with(command_buf.buf, "from "))
2788                 die("Expected from command, got %s", command_buf.buf);
2789         from = strchr(command_buf.buf, ' ') + 1;
2790         s = lookup_branch(from);
2791         if (s) {
2792                 if (is_null_sha1(s->sha1))
2793                         die("Can't tag an empty branch.");
2794                 hashcpy(sha1, s->sha1);
2795                 type = OBJ_COMMIT;
2796         } else if (*from == ':') {
2797                 struct object_entry *oe;
2798                 from_mark = parse_mark_ref_eol(from);
2799                 oe = find_mark(from_mark);
2800                 type = oe->type;
2801                 hashcpy(sha1, oe->idx.sha1);
2802         } else if (!get_sha1(from, sha1)) {
2803                 struct object_entry *oe = find_object(sha1);
2804                 if (!oe) {
2805                         type = sha1_object_info(sha1, NULL);
2806                         if (type < 0)
2807                                 die("Not a valid object: %s", from);
2808                 } else
2809                         type = oe->type;
2810         } else
2811                 die("Invalid ref name or SHA1 expression: %s", from);
2812         read_next_command();
2813
2814         /* tagger ... */
2815         if (starts_with(command_buf.buf, "tagger ")) {
2816                 tagger = parse_ident(command_buf.buf + 7);
2817                 read_next_command();
2818         } else
2819                 tagger = NULL;
2820
2821         /* tag payload/message */
2822         parse_data(&msg, 0, NULL);
2823
2824         /* build the tag object */
2825         strbuf_reset(&new_data);
2826
2827         strbuf_addf(&new_data,
2828                     "object %s\n"
2829                     "type %s\n"
2830                     "tag %s\n",
2831                     sha1_to_hex(sha1), typename(type), t->name);
2832         if (tagger)
2833                 strbuf_addf(&new_data,
2834                             "tagger %s\n", tagger);
2835         strbuf_addch(&new_data, '\n');
2836         strbuf_addbuf(&new_data, &msg);
2837         free(tagger);
2838
2839         if (store_object(OBJ_TAG, &new_data, NULL, t->sha1, 0))
2840                 t->pack_id = MAX_PACK_ID;
2841         else
2842                 t->pack_id = pack_id;
2843 }
2844
2845 static void parse_reset_branch(void)
2846 {
2847         struct branch *b;
2848         char *sp;
2849
2850         /* Obtain the branch name from the rest of our command */
2851         sp = strchr(command_buf.buf, ' ') + 1;
2852         b = lookup_branch(sp);
2853         if (b) {
2854                 hashclr(b->sha1);
2855                 hashclr(b->branch_tree.versions[0].sha1);
2856                 hashclr(b->branch_tree.versions[1].sha1);
2857                 if (b->branch_tree.tree) {
2858                         release_tree_content_recursive(b->branch_tree.tree);
2859                         b->branch_tree.tree = NULL;
2860                 }
2861         }
2862         else
2863                 b = new_branch(sp);
2864         read_next_command();
2865         parse_from(b);
2866         if (command_buf.len > 0)
2867                 unread_command_buf = 1;
2868 }
2869
2870 static void cat_blob_write(const char *buf, unsigned long size)
2871 {
2872         if (write_in_full(cat_blob_fd, buf, size) != size)
2873                 die_errno("Write to frontend failed");
2874 }
2875
2876 static void cat_blob(struct object_entry *oe, unsigned char sha1[20])
2877 {
2878         struct strbuf line = STRBUF_INIT;
2879         unsigned long size;
2880         enum object_type type = 0;
2881         char *buf;
2882
2883         if (!oe || oe->pack_id == MAX_PACK_ID) {
2884                 buf = read_sha1_file(sha1, &type, &size);
2885         } else {
2886                 type = oe->type;
2887                 buf = gfi_unpack_entry(oe, &size);
2888         }
2889
2890         /*
2891          * Output based on batch_one_object() from cat-file.c.
2892          */
2893         if (type <= 0) {
2894                 strbuf_reset(&line);
2895                 strbuf_addf(&line, "%s missing\n", sha1_to_hex(sha1));
2896                 cat_blob_write(line.buf, line.len);
2897                 strbuf_release(&line);
2898                 free(buf);
2899                 return;
2900         }
2901         if (!buf)
2902                 die("Can't read object %s", sha1_to_hex(sha1));
2903         if (type != OBJ_BLOB)
2904                 die("Object %s is a %s but a blob was expected.",
2905                     sha1_to_hex(sha1), typename(type));
2906         strbuf_reset(&line);
2907         strbuf_addf(&line, "%s %s %lu\n", sha1_to_hex(sha1),
2908                                                 typename(type), size);
2909         cat_blob_write(line.buf, line.len);
2910         strbuf_release(&line);
2911         cat_blob_write(buf, size);
2912         cat_blob_write("\n", 1);
2913         if (oe && oe->pack_id == pack_id) {
2914                 last_blob.offset = oe->idx.offset;
2915                 strbuf_attach(&last_blob.data, buf, size, size);
2916                 last_blob.depth = oe->depth;
2917         } else
2918                 free(buf);
2919 }
2920
2921 static void parse_cat_blob(void)
2922 {
2923         const char *p;
2924         struct object_entry *oe = oe;
2925         unsigned char sha1[20];
2926
2927         /* cat-blob SP <object> LF */
2928         p = command_buf.buf + strlen("cat-blob ");
2929         if (*p == ':') {
2930                 oe = find_mark(parse_mark_ref_eol(p));
2931                 if (!oe)
2932                         die("Unknown mark: %s", command_buf.buf);
2933                 hashcpy(sha1, oe->idx.sha1);
2934         } else {
2935                 if (get_sha1_hex(p, sha1))
2936                         die("Invalid dataref: %s", command_buf.buf);
2937                 if (p[40])
2938                         die("Garbage after SHA1: %s", command_buf.buf);
2939                 oe = find_object(sha1);
2940         }
2941
2942         cat_blob(oe, sha1);
2943 }
2944
2945 static struct object_entry *dereference(struct object_entry *oe,
2946                                         unsigned char sha1[20])
2947 {
2948         unsigned long size;
2949         char *buf = NULL;
2950         if (!oe) {
2951                 enum object_type type = sha1_object_info(sha1, NULL);
2952                 if (type < 0)
2953                         die("object not found: %s", sha1_to_hex(sha1));
2954                 /* cache it! */
2955                 oe = insert_object(sha1);
2956                 oe->type = type;
2957                 oe->pack_id = MAX_PACK_ID;
2958                 oe->idx.offset = 1;
2959         }
2960         switch (oe->type) {
2961         case OBJ_TREE:  /* easy case. */
2962                 return oe;
2963         case OBJ_COMMIT:
2964         case OBJ_TAG:
2965                 break;
2966         default:
2967                 die("Not a tree-ish: %s", command_buf.buf);
2968         }
2969
2970         if (oe->pack_id != MAX_PACK_ID) {       /* in a pack being written */
2971                 buf = gfi_unpack_entry(oe, &size);
2972         } else {
2973                 enum object_type unused;
2974                 buf = read_sha1_file(sha1, &unused, &size);
2975         }
2976         if (!buf)
2977                 die("Can't load object %s", sha1_to_hex(sha1));
2978
2979         /* Peel one layer. */
2980         switch (oe->type) {
2981         case OBJ_TAG:
2982                 if (size < 40 + strlen("object ") ||
2983                     get_sha1_hex(buf + strlen("object "), sha1))
2984                         die("Invalid SHA1 in tag: %s", command_buf.buf);
2985                 break;
2986         case OBJ_COMMIT:
2987                 if (size < 40 + strlen("tree ") ||
2988                     get_sha1_hex(buf + strlen("tree "), sha1))
2989                         die("Invalid SHA1 in commit: %s", command_buf.buf);
2990         }
2991
2992         free(buf);
2993         return find_object(sha1);
2994 }
2995
2996 static struct object_entry *parse_treeish_dataref(const char **p)
2997 {
2998         unsigned char sha1[20];
2999         struct object_entry *e;
3000
3001         if (**p == ':') {       /* <mark> */
3002                 e = find_mark(parse_mark_ref_space(p));
3003                 if (!e)
3004                         die("Unknown mark: %s", command_buf.buf);
3005                 hashcpy(sha1, e->idx.sha1);
3006         } else {        /* <sha1> */
3007                 if (get_sha1_hex(*p, sha1))
3008                         die("Invalid dataref: %s", command_buf.buf);
3009                 e = find_object(sha1);
3010                 *p += 40;
3011         }
3012
3013         while (!e || e->type != OBJ_TREE)
3014                 e = dereference(e, sha1);
3015         return e;
3016 }
3017
3018 static void print_ls(int mode, const unsigned char *sha1, const char *path)
3019 {
3020         static struct strbuf line = STRBUF_INIT;
3021
3022         /* See show_tree(). */
3023         const char *type =
3024                 S_ISGITLINK(mode) ? commit_type :
3025                 S_ISDIR(mode) ? tree_type :
3026                 blob_type;
3027
3028         if (!mode) {
3029                 /* missing SP path LF */
3030                 strbuf_reset(&line);
3031                 strbuf_addstr(&line, "missing ");
3032                 quote_c_style(path, &line, NULL, 0);
3033                 strbuf_addch(&line, '\n');
3034         } else {
3035                 /* mode SP type SP object_name TAB path LF */
3036                 strbuf_reset(&line);
3037                 strbuf_addf(&line, "%06o %s %s\t",
3038                                 mode & ~NO_DELTA, type, sha1_to_hex(sha1));
3039                 quote_c_style(path, &line, NULL, 0);
3040                 strbuf_addch(&line, '\n');
3041         }
3042         cat_blob_write(line.buf, line.len);
3043 }
3044
3045 static void parse_ls(struct branch *b)
3046 {
3047         const char *p;
3048         struct tree_entry *root = NULL;
3049         struct tree_entry leaf = {NULL};
3050
3051         /* ls SP (<tree-ish> SP)? <path> */
3052         p = command_buf.buf + strlen("ls ");
3053         if (*p == '"') {
3054                 if (!b)
3055                         die("Not in a commit: %s", command_buf.buf);
3056                 root = &b->branch_tree;
3057         } else {
3058                 struct object_entry *e = parse_treeish_dataref(&p);
3059                 root = new_tree_entry();
3060                 hashcpy(root->versions[1].sha1, e->idx.sha1);
3061                 if (!is_null_sha1(root->versions[1].sha1))
3062                         root->versions[1].mode = S_IFDIR;
3063                 load_tree(root);
3064                 if (*p++ != ' ')
3065                         die("Missing space after tree-ish: %s", command_buf.buf);
3066         }
3067         if (*p == '"') {
3068                 static struct strbuf uq = STRBUF_INIT;
3069                 const char *endp;
3070                 strbuf_reset(&uq);
3071                 if (unquote_c_style(&uq, p, &endp))
3072                         die("Invalid path: %s", command_buf.buf);
3073                 if (*endp)
3074                         die("Garbage after path in: %s", command_buf.buf);
3075                 p = uq.buf;
3076         }
3077         tree_content_get(root, p, &leaf, 1);
3078         /*
3079          * A directory in preparation would have a sha1 of zero
3080          * until it is saved.  Save, for simplicity.
3081          */
3082         if (S_ISDIR(leaf.versions[1].mode))
3083                 store_tree(&leaf);
3084
3085         print_ls(leaf.versions[1].mode, leaf.versions[1].sha1, p);
3086         if (leaf.tree)
3087                 release_tree_content_recursive(leaf.tree);
3088         if (!b || root != &b->branch_tree)
3089                 release_tree_entry(root);
3090 }
3091
3092 static void checkpoint(void)
3093 {
3094         checkpoint_requested = 0;
3095         if (object_count) {
3096                 cycle_packfile();
3097                 dump_branches();
3098                 dump_tags();
3099                 dump_marks();
3100         }
3101 }
3102
3103 static void parse_checkpoint(void)
3104 {
3105         checkpoint_requested = 1;
3106         skip_optional_lf();
3107 }
3108
3109 static void parse_progress(void)
3110 {
3111         fwrite(command_buf.buf, 1, command_buf.len, stdout);
3112         fputc('\n', stdout);
3113         fflush(stdout);
3114         skip_optional_lf();
3115 }
3116
3117 static char* make_fast_import_path(const char *path)
3118 {
3119         struct strbuf abs_path = STRBUF_INIT;
3120
3121         if (!relative_marks_paths || is_absolute_path(path))
3122                 return xstrdup(path);
3123         strbuf_addf(&abs_path, "%s/info/fast-import/%s", get_git_dir(), path);
3124         return strbuf_detach(&abs_path, NULL);
3125 }
3126
3127 static void option_import_marks(const char *marks,
3128                                         int from_stream, int ignore_missing)
3129 {
3130         if (import_marks_file) {
3131                 if (from_stream)
3132                         die("Only one import-marks command allowed per stream");
3133
3134                 /* read previous mark file */
3135                 if(!import_marks_file_from_stream)
3136                         read_marks();
3137         }
3138
3139         import_marks_file = make_fast_import_path(marks);
3140         safe_create_leading_directories_const(import_marks_file);
3141         import_marks_file_from_stream = from_stream;
3142         import_marks_file_ignore_missing = ignore_missing;
3143 }
3144
3145 static void option_date_format(const char *fmt)
3146 {
3147         if (!strcmp(fmt, "raw"))
3148                 whenspec = WHENSPEC_RAW;
3149         else if (!strcmp(fmt, "rfc2822"))
3150                 whenspec = WHENSPEC_RFC2822;
3151         else if (!strcmp(fmt, "now"))
3152                 whenspec = WHENSPEC_NOW;
3153         else
3154                 die("unknown --date-format argument %s", fmt);
3155 }
3156
3157 static unsigned long ulong_arg(const char *option, const char *arg)
3158 {
3159         char *endptr;
3160         unsigned long rv = strtoul(arg, &endptr, 0);
3161         if (strchr(arg, '-') || endptr == arg || *endptr)
3162                 die("%s: argument must be a non-negative integer", option);
3163         return rv;
3164 }
3165
3166 static void option_depth(const char *depth)
3167 {
3168         max_depth = ulong_arg("--depth", depth);
3169         if (max_depth > MAX_DEPTH)
3170                 die("--depth cannot exceed %u", MAX_DEPTH);
3171 }
3172
3173 static void option_active_branches(const char *branches)
3174 {
3175         max_active_branches = ulong_arg("--active-branches", branches);
3176 }
3177
3178 static void option_export_marks(const char *marks)
3179 {
3180         export_marks_file = make_fast_import_path(marks);
3181         safe_create_leading_directories_const(export_marks_file);
3182 }
3183
3184 static void option_cat_blob_fd(const char *fd)
3185 {
3186         unsigned long n = ulong_arg("--cat-blob-fd", fd);
3187         if (n > (unsigned long) INT_MAX)
3188                 die("--cat-blob-fd cannot exceed %d", INT_MAX);
3189         cat_blob_fd = (int) n;
3190 }
3191
3192 static void option_export_pack_edges(const char *edges)
3193 {
3194         if (pack_edges)
3195                 fclose(pack_edges);
3196         pack_edges = fopen(edges, "a");
3197         if (!pack_edges)
3198                 die_errno("Cannot open '%s'", edges);
3199 }
3200
3201 static int parse_one_option(const char *option)
3202 {
3203         if (starts_with(option, "max-pack-size=")) {
3204                 unsigned long v;
3205                 if (!git_parse_ulong(option + 14, &v))
3206                         return 0;
3207                 if (v < 8192) {
3208                         warning("max-pack-size is now in bytes, assuming --max-pack-size=%lum", v);
3209                         v *= 1024 * 1024;
3210                 } else if (v < 1024 * 1024) {
3211                         warning("minimum max-pack-size is 1 MiB");
3212                         v = 1024 * 1024;
3213                 }
3214                 max_packsize = v;
3215         } else if (starts_with(option, "big-file-threshold=")) {
3216                 unsigned long v;
3217                 if (!git_parse_ulong(option + 19, &v))
3218                         return 0;
3219                 big_file_threshold = v;
3220         } else if (starts_with(option, "depth=")) {
3221                 option_depth(option + 6);
3222         } else if (starts_with(option, "active-branches=")) {
3223                 option_active_branches(option + 16);
3224         } else if (starts_with(option, "export-pack-edges=")) {
3225                 option_export_pack_edges(option + 18);
3226         } else if (starts_with(option, "quiet")) {
3227                 show_stats = 0;
3228         } else if (starts_with(option, "stats")) {
3229                 show_stats = 1;
3230         } else {
3231                 return 0;
3232         }
3233
3234         return 1;
3235 }
3236
3237 static int parse_one_feature(const char *feature, int from_stream)
3238 {
3239         if (starts_with(feature, "date-format=")) {
3240                 option_date_format(feature + 12);
3241         } else if (starts_with(feature, "import-marks=")) {
3242                 option_import_marks(feature + 13, from_stream, 0);
3243         } else if (starts_with(feature, "import-marks-if-exists=")) {
3244                 option_import_marks(feature + strlen("import-marks-if-exists="),
3245                                         from_stream, 1);
3246         } else if (starts_with(feature, "export-marks=")) {
3247                 option_export_marks(feature + 13);
3248         } else if (!strcmp(feature, "cat-blob")) {
3249                 ; /* Don't die - this feature is supported */
3250         } else if (!strcmp(feature, "relative-marks")) {
3251                 relative_marks_paths = 1;
3252         } else if (!strcmp(feature, "no-relative-marks")) {
3253                 relative_marks_paths = 0;
3254         } else if (!strcmp(feature, "done")) {
3255                 require_explicit_termination = 1;
3256         } else if (!strcmp(feature, "force")) {
3257                 force_update = 1;
3258         } else if (!strcmp(feature, "notes") || !strcmp(feature, "ls")) {
3259                 ; /* do nothing; we have the feature */
3260         } else {
3261                 return 0;
3262         }
3263
3264         return 1;
3265 }
3266
3267 static void parse_feature(void)
3268 {
3269         char *feature = command_buf.buf + 8;
3270
3271         if (seen_data_command)
3272                 die("Got feature command '%s' after data command", feature);
3273
3274         if (parse_one_feature(feature, 1))
3275                 return;
3276
3277         die("This version of fast-import does not support feature %s.", feature);
3278 }
3279
3280 static void parse_option(void)
3281 {
3282         char *option = command_buf.buf + 11;
3283
3284         if (seen_data_command)
3285                 die("Got option command '%s' after data command", option);
3286
3287         if (parse_one_option(option))
3288                 return;
3289
3290         die("This version of fast-import does not support option: %s", option);
3291 }
3292
3293 static int git_pack_config(const char *k, const char *v, void *cb)
3294 {
3295         if (!strcmp(k, "pack.depth")) {
3296                 max_depth = git_config_int(k, v);
3297                 if (max_depth > MAX_DEPTH)
3298                         max_depth = MAX_DEPTH;
3299                 return 0;
3300         }
3301         if (!strcmp(k, "pack.compression")) {
3302                 int level = git_config_int(k, v);
3303                 if (level == -1)
3304                         level = Z_DEFAULT_COMPRESSION;
3305                 else if (level < 0 || level > Z_BEST_COMPRESSION)
3306                         die("bad pack compression level %d", level);
3307                 pack_compression_level = level;
3308                 pack_compression_seen = 1;
3309                 return 0;
3310         }
3311         if (!strcmp(k, "pack.indexversion")) {
3312                 pack_idx_opts.version = git_config_int(k, v);
3313                 if (pack_idx_opts.version > 2)
3314                         die("bad pack.indexversion=%"PRIu32,
3315                             pack_idx_opts.version);
3316                 return 0;
3317         }
3318         if (!strcmp(k, "pack.packsizelimit")) {
3319                 max_packsize = git_config_ulong(k, v);
3320                 return 0;
3321         }
3322         return git_default_config(k, v, cb);
3323 }
3324
3325 static const char fast_import_usage[] =
3326 "git fast-import [--date-format=<f>] [--max-pack-size=<n>] [--big-file-threshold=<n>] [--depth=<n>] [--active-branches=<n>] [--export-marks=<marks.file>]";
3327
3328 static void parse_argv(void)
3329 {
3330         unsigned int i;
3331
3332         for (i = 1; i < global_argc; i++) {
3333                 const char *a = global_argv[i];
3334
3335                 if (*a != '-' || !strcmp(a, "--"))
3336                         break;
3337
3338                 if (parse_one_option(a + 2))
3339                         continue;
3340
3341                 if (parse_one_feature(a + 2, 0))
3342                         continue;
3343
3344                 if (starts_with(a + 2, "cat-blob-fd=")) {
3345                         option_cat_blob_fd(a + 2 + strlen("cat-blob-fd="));
3346                         continue;
3347                 }
3348
3349                 die("unknown option %s", a);
3350         }
3351         if (i != global_argc)
3352                 usage(fast_import_usage);
3353
3354         seen_data_command = 1;
3355         if (import_marks_file)
3356                 read_marks();
3357 }
3358
3359 int main(int argc, char **argv)
3360 {
3361         unsigned int i;
3362
3363         git_extract_argv0_path(argv[0]);
3364
3365         git_setup_gettext();
3366
3367         if (argc == 2 && !strcmp(argv[1], "-h"))
3368                 usage(fast_import_usage);
3369
3370         setup_git_directory();
3371         reset_pack_idx_option(&pack_idx_opts);
3372         git_config(git_pack_config, NULL);
3373         if (!pack_compression_seen && core_compression_seen)
3374                 pack_compression_level = core_compression_level;
3375
3376         alloc_objects(object_entry_alloc);
3377         strbuf_init(&command_buf, 0);
3378         atom_table = xcalloc(atom_table_sz, sizeof(struct atom_str*));
3379         branch_table = xcalloc(branch_table_sz, sizeof(struct branch*));
3380         avail_tree_table = xcalloc(avail_tree_table_sz, sizeof(struct avail_tree_content*));
3381         marks = pool_calloc(1, sizeof(struct mark_set));
3382
3383         global_argc = argc;
3384         global_argv = argv;
3385
3386         rc_free = pool_alloc(cmd_save * sizeof(*rc_free));
3387         for (i = 0; i < (cmd_save - 1); i++)
3388                 rc_free[i].next = &rc_free[i + 1];
3389         rc_free[cmd_save - 1].next = NULL;
3390
3391         prepare_packed_git();
3392         start_packfile();
3393         set_die_routine(die_nicely);
3394         set_checkpoint_signal();
3395         while (read_next_command() != EOF) {
3396                 if (!strcmp("blob", command_buf.buf))
3397                         parse_new_blob();
3398                 else if (starts_with(command_buf.buf, "ls "))
3399                         parse_ls(NULL);
3400                 else if (starts_with(command_buf.buf, "commit "))
3401                         parse_new_commit();
3402                 else if (starts_with(command_buf.buf, "tag "))
3403                         parse_new_tag();
3404                 else if (starts_with(command_buf.buf, "reset "))
3405                         parse_reset_branch();
3406                 else if (!strcmp("checkpoint", command_buf.buf))
3407                         parse_checkpoint();
3408                 else if (!strcmp("done", command_buf.buf))
3409                         break;
3410                 else if (starts_with(command_buf.buf, "progress "))
3411                         parse_progress();
3412                 else if (starts_with(command_buf.buf, "feature "))
3413                         parse_feature();
3414                 else if (starts_with(command_buf.buf, "option git "))
3415                         parse_option();
3416                 else if (starts_with(command_buf.buf, "option "))
3417                         /* ignore non-git options*/;
3418                 else
3419                         die("Unsupported command: %s", command_buf.buf);
3420
3421                 if (checkpoint_requested)
3422                         checkpoint();
3423         }
3424
3425         /* argv hasn't been parsed yet, do so */
3426         if (!seen_data_command)
3427                 parse_argv();
3428
3429         if (require_explicit_termination && feof(stdin))
3430                 die("stream ends early");
3431
3432         end_packfile();
3433
3434         dump_branches();
3435         dump_tags();
3436         unkeep_all_packs();
3437         dump_marks();
3438
3439         if (pack_edges)
3440                 fclose(pack_edges);
3441
3442         if (show_stats) {
3443                 uintmax_t total_count = 0, duplicate_count = 0;
3444                 for (i = 0; i < ARRAY_SIZE(object_count_by_type); i++)
3445                         total_count += object_count_by_type[i];
3446                 for (i = 0; i < ARRAY_SIZE(duplicate_count_by_type); i++)
3447                         duplicate_count += duplicate_count_by_type[i];
3448
3449                 fprintf(stderr, "%s statistics:\n", argv[0]);
3450                 fprintf(stderr, "---------------------------------------------------------------------\n");
3451                 fprintf(stderr, "Alloc'd objects: %10" PRIuMAX "\n", alloc_count);
3452                 fprintf(stderr, "Total objects:   %10" PRIuMAX " (%10" PRIuMAX " duplicates                  )\n", total_count, duplicate_count);
3453                 fprintf(stderr, "      blobs  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_BLOB], duplicate_count_by_type[OBJ_BLOB], delta_count_by_type[OBJ_BLOB], delta_count_attempts_by_type[OBJ_BLOB]);
3454                 fprintf(stderr, "      trees  :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TREE], duplicate_count_by_type[OBJ_TREE], delta_count_by_type[OBJ_TREE], delta_count_attempts_by_type[OBJ_TREE]);
3455                 fprintf(stderr, "      commits:   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_COMMIT], duplicate_count_by_type[OBJ_COMMIT], delta_count_by_type[OBJ_COMMIT], delta_count_attempts_by_type[OBJ_COMMIT]);
3456                 fprintf(stderr, "      tags   :   %10" PRIuMAX " (%10" PRIuMAX " duplicates %10" PRIuMAX " deltas of %10" PRIuMAX" attempts)\n", object_count_by_type[OBJ_TAG], duplicate_count_by_type[OBJ_TAG], delta_count_by_type[OBJ_TAG], delta_count_attempts_by_type[OBJ_TAG]);
3457                 fprintf(stderr, "Total branches:  %10lu (%10lu loads     )\n", branch_count, branch_load_count);
3458                 fprintf(stderr, "      marks:     %10" PRIuMAX " (%10" PRIuMAX " unique    )\n", (((uintmax_t)1) << marks->shift) * 1024, marks_set_count);
3459                 fprintf(stderr, "      atoms:     %10u\n", atom_cnt);
3460                 fprintf(stderr, "Memory total:    %10" PRIuMAX " KiB\n", (total_allocd + alloc_count*sizeof(struct object_entry))/1024);
3461                 fprintf(stderr, "       pools:    %10lu KiB\n", (unsigned long)(total_allocd/1024));
3462                 fprintf(stderr, "     objects:    %10" PRIuMAX " KiB\n", (alloc_count*sizeof(struct object_entry))/1024);
3463                 fprintf(stderr, "---------------------------------------------------------------------\n");
3464                 pack_report();
3465                 fprintf(stderr, "---------------------------------------------------------------------\n");
3466                 fprintf(stderr, "\n");
3467         }
3468
3469         return failure ? 1 : 0;
3470 }