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